[PATCH v3 00/41] Mirror map JIT memory for TCG

qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed

* [PATCH v3 00/41] Mirror map JIT memory for TCG
@ 2020-11-06  3:28 Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer Richard Henderson
                   ` (43 more replies)
  0 siblings, 44 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

This is my take on Joelle's patch set:
https://lists.nongnu.org/archive/html/qemu-devel/2020-10/msg07837.html

Changes for v3:
  * Even more patches -- all tcg backends converted.
  * Fixups for darwin/ios merged (Joelle).
  * Feature renamed to splitwx (Paolo).


r~


Richard Henderson (41):
  tcg: Enhance flush_icache_range with separate data pointer
  tcg: Move tcg prologue pointer out of TCGContext
  tcg: Move tcg epilogue pointer out of TCGContext
  tcg: Add in_code_gen_buffer
  tcg: Introduce tcg_splitwx_to_{rx,rw}
  tcg: Adjust TCGLabel for const
  tcg: Adjust tcg_out_call for const
  tcg: Adjust tcg_out_label for const
  tcg: Adjust tcg_register_jit for const
  tcg: Adjust tb_target_set_jmp_target for split-wx
  tcg: Make DisasContextBase.tb const
  tcg: Make tb arg to synchronize_from_tb const
  tcg: Use Error with alloc_code_gen_buffer
  tcg: Add --accel tcg,split-wx property
  accel/tcg: Support split-wx for linux with memfd
  accel/tcg: Support split-wx for darwin/iOS with vm_remap
  tcg: Return the TB pointer from the rx region from exit_tb
  tcg/i386: Support split-wx code generation
  tcg/aarch64: Use B not BL for tcg_out_goto_long
  tcg/aarch64: Implement flush_idcache_range manually
  tcg/aarch64: Support split-wx code generation
  disas: Push const down through host disasassembly
  tcg/tci: Push const down through bytecode reading
  tcg: Introduce tcg_tbrel_diff
  tcg/ppc: Use tcg_tbrel_diff
  tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB
  tcg/ppc: Support split-wx code generation
  tcg/sparc: Use tcg_tbrel_diff
  tcg/sparc: Support split-wx code generation
  tcg/s390: Use tcg_tbrel_diff
  tcg/s390: Support split-wx code generation
  tcg/riscv: Fix branch range checks
  tcg/riscv: Remove branch-over-branch fallback
  tcg/riscv: Support split-wx code generation
  accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd
  tcg/mips: Do not assert on relocation overflow
  tcg/mips: Support split-wx code generation
  tcg/arm: Support split-wx code generation
  tcg: Remove TCG_TARGET_SUPPORT_MIRROR
  tcg: Constify tcg_code_gen_epilogue
  tcg: Constify TCGLabelQemuLdst.raddr

 accel/tcg/tcg-runtime.h      |   2 +-
 include/disas/dis-asm.h      |   4 +-
 include/disas/disas.h        |   2 +-
 include/exec/exec-all.h      |   2 +-
 include/exec/gen-icount.h    |   4 +-
 include/exec/log.h           |   2 +-
 include/exec/translator.h    |   2 +-
 include/hw/core/cpu.h        |   3 +-
 include/sysemu/tcg.h         |   3 +-
 include/tcg/tcg-op.h         |   2 +-
 include/tcg/tcg.h            |  56 +++++--
 tcg/aarch64/tcg-target.h     |   8 +-
 tcg/arm/tcg-target.h         |  10 +-
 tcg/i386/tcg-target.h        |   9 +-
 tcg/mips/tcg-target.h        |  10 +-
 tcg/ppc/tcg-target.h         |   4 +-
 tcg/riscv/tcg-target.h       |  10 +-
 tcg/s390/tcg-target.h        |  11 +-
 tcg/sparc/tcg-target.h       |  10 +-
 tcg/tci/tcg-target.h         |  11 +-
 accel/tcg/cpu-exec.c         |  41 +++--
 accel/tcg/tcg-all.c          |  26 ++-
 accel/tcg/tcg-runtime.c      |   4 +-
 accel/tcg/translate-all.c    | 307 +++++++++++++++++++++++++++--------
 accel/tcg/translator.c       |   4 +-
 bsd-user/main.c              |   2 +-
 disas.c                      |   2 +-
 disas/capstone.c             |   2 +-
 linux-user/main.c            |   2 +-
 softmmu/physmem.c            |   9 +-
 target/arm/cpu.c             |   3 +-
 target/arm/translate-a64.c   |   2 +-
 target/avr/cpu.c             |   3 +-
 target/hppa/cpu.c            |   3 +-
 target/i386/cpu.c            |   3 +-
 target/microblaze/cpu.c      |   3 +-
 target/mips/cpu.c            |   3 +-
 target/riscv/cpu.c           |   3 +-
 target/rx/cpu.c              |   3 +-
 target/sh4/cpu.c             |   3 +-
 target/sparc/cpu.c           |   3 +-
 target/tricore/cpu.c         |   2 +-
 tcg/tcg-op.c                 |  15 +-
 tcg/tcg.c                    |  86 ++++++++--
 tcg/tci.c                    |  60 ++++---
 accel/tcg/trace-events       |   2 +-
 tcg/aarch64/tcg-target.c.inc | 139 ++++++++++++----
 tcg/arm/tcg-target.c.inc     |  41 ++---
 tcg/i386/tcg-target.c.inc    |  36 ++--
 tcg/mips/tcg-target.c.inc    |  97 +++++------
 tcg/ppc/tcg-target.c.inc     | 105 ++++++------
 tcg/riscv/tcg-target.c.inc   | 125 +++++---------
 tcg/s390/tcg-target.c.inc    |  91 +++++------
 tcg/sparc/tcg-target.c.inc   |  58 +++----
 tcg/tcg-ldst.c.inc           |   2 +-
 tcg/tcg-pool.c.inc           |   6 +-
 tcg/tci/tcg-target.c.inc     |   2 +-
 57 files changed, 917 insertions(+), 546 deletions(-)

-- 
2.25.1



^ permalink raw reply	[flat|nested] 57+ messages in thread

* [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06 20:31   ` Alex Bennée
  2020-11-06  3:28 ` [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext Richard Henderson
                   ` (42 subsequent siblings)
  43 siblings, 1 reply; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

We are shortly going to have a split rw/rx jit buffer.  Depending
on the host, we need to flush the dcache at the rw data pointer and
flush the icache at the rx code pointer.

For now, the two passed pointers are identical, so there is no
effective change in behaviour.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.h     |  9 +++++++--
 tcg/arm/tcg-target.h         |  8 ++++++--
 tcg/i386/tcg-target.h        |  3 ++-
 tcg/mips/tcg-target.h        |  8 ++++++--
 tcg/ppc/tcg-target.h         |  2 +-
 tcg/riscv/tcg-target.h       |  8 ++++++--
 tcg/s390/tcg-target.h        |  3 ++-
 tcg/sparc/tcg-target.h       |  8 +++++---
 tcg/tci/tcg-target.h         |  3 ++-
 softmmu/physmem.c            |  9 ++++++++-
 tcg/tcg.c                    |  6 ++++--
 tcg/aarch64/tcg-target.c.inc |  2 +-
 tcg/mips/tcg-target.c.inc    |  2 +-
 tcg/ppc/tcg-target.c.inc     | 21 +++++++++++----------
 tcg/sparc/tcg-target.c.inc   |  4 ++--
 15 files changed, 64 insertions(+), 32 deletions(-)

diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index 663dd0b95e..d0a6a059b7 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -148,9 +148,14 @@ typedef enum {
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
-    __builtin___clear_cache((char *)start, (char *)stop);
+    /* TODO: Copy this from gcc to avoid 4 loops instead of 2. */
+    if (rw != rx) {
+        __builtin___clear_cache((char *)rw, (char *)(rw + len));
+    }
+    __builtin___clear_cache((char *)rx, (char *)(rx + len));
 }
 
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index 17e771374d..fa88b24e43 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -134,9 +134,13 @@ enum {
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
-    __builtin___clear_cache((char *) start, (char *) stop);
+    if (rw != rx) {
+        __builtin___clear_cache((char *)rw, (char *)(rw + len));
+    }
+    __builtin___clear_cache((char *)rx, (char *)(rx + len));
 }
 
 /* not defined -- call should be eliminated at compile time */
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index abd4ac7fc0..8323e72639 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -206,7 +206,8 @@ extern bool have_avx2;
 #define TCG_TARGET_extract_i64_valid(ofs, len) \
     (((ofs) == 8 && (len) == 8) || ((ofs) + (len)) == 32)
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
 }
 
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index c6b091d849..47b1226ee9 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -207,9 +207,13 @@ extern bool use_mips32r2_instructions;
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
-    cacheflush ((void *)start, stop-start, ICACHE);
+    if (rx != rw) {
+        cacheflush((void *)rw, len, DCACHE);
+    }
+    cacheflush((void *)rx, len, ICACHE);
 }
 
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index be10363956..fbb6dc1b47 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -175,7 +175,7 @@ extern bool have_vsx;
 #define TCG_TARGET_HAS_bitsel_vec       have_vsx
 #define TCG_TARGET_HAS_cmpsel_vec       0
 
-void flush_icache_range(uintptr_t start, uintptr_t stop);
+void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len);
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_DEFAULT_MO (0)
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index 032439d806..0fa6ae358e 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -159,9 +159,13 @@ typedef enum {
 #define TCG_TARGET_HAS_mulsh_i64        1
 #endif
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
-    __builtin___clear_cache((char *)start, (char *)stop);
+    if (rx != rw) {
+        __builtin___clear_cache((char *)rw, (char *)(rw + len));
+    }
+    __builtin___clear_cache((char *)rx, (char *)(rx + len));
 }
 
 /* not defined -- call should be eliminated at compile time */
diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h
index 63c8797bd3..c3dc2e8938 100644
--- a/tcg/s390/tcg-target.h
+++ b/tcg/s390/tcg-target.h
@@ -145,7 +145,8 @@ enum {
     TCG_AREG0 = TCG_REG_R10,
 };
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
 }
 
diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index 633841ebf2..c27c40231e 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -168,10 +168,12 @@ extern bool use_vis3_instructions;
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
-    uintptr_t p;
-    for (p = start & -8; p < ((stop + 7) & -8); p += 8) {
+    /* No additional data flush to the RW virtual address required. */
+    uintptr_t p, end = (rx + len + 7) & -8;
+    for (p = rx & -8; p < end; p += 8) {
         __asm__ __volatile__("flush\t%0" : : "r" (p));
     }
 }
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 8c1c1d265d..6460449719 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -191,7 +191,8 @@ void tci_disas(uint8_t opc);
 
 #define HAVE_TCG_QEMU_TB_EXEC
 
-static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
 }
 
diff --git a/softmmu/physmem.c b/softmmu/physmem.c
index 0b31be2928..d1e2da5502 100644
--- a/softmmu/physmem.c
+++ b/softmmu/physmem.c
@@ -2946,7 +2946,14 @@ static inline MemTxResult address_space_write_rom_internal(AddressSpace *as,
                 invalidate_and_set_dirty(mr, addr1, l);
                 break;
             case FLUSH_CACHE:
-                flush_icache_range((uintptr_t)ram_ptr, (uintptr_t)ram_ptr + l);
+                /*
+                 * FIXME: This function is currently located in tcg/host/,
+                 * but we never come here when tcg is enabled; only for
+                 * real hardware acceleration.  This can actively fail
+                 * when TCI is configured, since that function is a nop.
+                 * We should move this to util/ or something.
+                 */
+                flush_idcache_range((uintptr_t)ram_ptr, (uintptr_t)ram_ptr, l);
                 break;
             }
         }
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 43c6cf8f52..d5a72c226f 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -1077,7 +1077,8 @@ void tcg_prologue_init(TCGContext *s)
 #endif
 
     buf1 = s->code_ptr;
-    flush_icache_range((uintptr_t)buf0, (uintptr_t)buf1);
+    flush_idcache_range((uintptr_t)buf0, (uintptr_t)buf0,
+                        tcg_ptr_byte_diff(buf1, buf0));
 
     /* Deduct the prologue from the buffer.  */
     prologue_size = tcg_current_code_size(s);
@@ -4320,7 +4321,8 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb)
     }
 
     /* flush instruction cache */
-    flush_icache_range((uintptr_t)s->code_buf, (uintptr_t)s->code_ptr);
+    flush_idcache_range((uintptr_t)s->code_buf, (uintptr_t)s->code_buf,
+                        tcg_ptr_byte_diff(s->code_ptr, s->code_buf));
 
     return tcg_current_code_size(s);
 }
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 26f71cb599..83af3108a4 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1363,7 +1363,7 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
     }
     pair = (uint64_t)i2 << 32 | i1;
     qatomic_set((uint64_t *)jmp_addr, pair);
-    flush_icache_range(jmp_addr, jmp_addr + 8);
+    flush_idcache_range(jmp_addr, jmp_addr, 8);
 }
 
 static inline void tcg_out_goto_label(TCGContext *s, TCGLabel *l)
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 41be574e89..c255ecb444 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -2660,7 +2660,7 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
                               uintptr_t addr)
 {
     qatomic_set((uint32_t *)jmp_addr, deposit32(OPC_J, 0, 26, addr >> 2));
-    flush_icache_range(jmp_addr, jmp_addr + 4);
+    flush_idcache_range(jmp_addr, jmp_addr, 4);
 }
 
 typedef struct {
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 18ee989f95..a848e98383 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -1753,12 +1753,12 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
         /* As per the enclosing if, this is ppc64.  Avoid the _Static_assert
            within qatomic_set that would fail to build a ppc32 host.  */
         qatomic_set__nocheck((uint64_t *)jmp_addr, pair);
-        flush_icache_range(jmp_addr, jmp_addr + 8);
+        flush_idcache_range(jmp_addr, jmp_addr, 8);
     } else {
         intptr_t diff = addr - jmp_addr;
         tcg_debug_assert(in_range_b(diff));
         qatomic_set((uint32_t *)jmp_addr, B | (diff & 0x3fffffc));
-        flush_icache_range(jmp_addr, jmp_addr + 4);
+        flush_idcache_range(jmp_addr, jmp_addr, 4);
     }
 }
 
@@ -3864,22 +3864,23 @@ void tcg_register_jit(void *buf, size_t buf_size)
 }
 #endif /* __ELF__ */
 
-void flush_icache_range(uintptr_t start, uintptr_t stop)
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+void flush_idcache_range(uintptr_t rx, uintptr_t rw, uintptr_t len)
 {
-    uintptr_t p, start1, stop1;
+    uintptr_t p, start, stop;
     size_t dsize = qemu_dcache_linesize;
     size_t isize = qemu_icache_linesize;
 
-    start1 = start & ~(dsize - 1);
-    stop1 = (stop + dsize - 1) & ~(dsize - 1);
-    for (p = start1; p < stop1; p += dsize) {
+    start = rw & ~(dsize - 1);
+    stop = (rw + len + dsize - 1) & ~(dsize - 1);
+    for (p = start; p < stop; p += dsize) {
         asm volatile ("dcbst 0,%0" : : "r"(p) : "memory");
     }
     asm volatile ("sync" : : : "memory");
 
-    start &= start & ~(isize - 1);
-    stop1 = (stop + isize - 1) & ~(isize - 1);
-    for (p = start1; p < stop1; p += isize) {
+    start = rx & ~(isize - 1);
+    stop = (rx + len + isize - 1) & ~(isize - 1);
+    for (p = start; p < stop; p += isize) {
         asm volatile ("icbi 0,%0" : : "r"(p) : "memory");
     }
     asm volatile ("sync" : : : "memory");
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 6775bd30fc..6e2d755f6a 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -1836,7 +1836,7 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
     if (!USE_REG_TB) {
         qatomic_set((uint32_t *)jmp_addr,
 		    deposit32(CALL, 0, 30, br_disp >> 2));
-        flush_icache_range(jmp_addr, jmp_addr + 4);
+        flush_idcache_range(jmp_addr, jmp_addr, 4);
         return;
     }
 
@@ -1860,5 +1860,5 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
     }
 
     qatomic_set((uint64_t *)jmp_addr, deposit64(i2, 32, 32, i1));
-    flush_icache_range(jmp_addr, jmp_addr + 8);
+    flush_idcache_range(jmp_addr, jmp_addr, 8);
 }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer
  2020-11-06  3:28 ` [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer Richard Henderson
@ 2020-11-06 20:31   ` Alex Bennée
  2020-11-06 22:51     ` Richard Henderson
  0 siblings, 1 reply; 57+ messages in thread
From: Alex Bennée @ 2020-11-06 20:31 UTC (permalink / raw)
  To: Richard Henderson; +Cc: j, qemu-devel


Richard Henderson <richard.henderson@linaro.org> writes:

> We are shortly going to have a split rw/rx jit buffer.  Depending
> on the host, we need to flush the dcache at the rw data pointer and
> flush the icache at the rx code pointer.
>
> For now, the two passed pointers are identical, so there is no
> effective change in behaviour.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  tcg/aarch64/tcg-target.h     |  9 +++++++--
>  tcg/arm/tcg-target.h         |  8 ++++++--
>  tcg/i386/tcg-target.h        |  3 ++-
>  tcg/mips/tcg-target.h        |  8 ++++++--
>  tcg/ppc/tcg-target.h         |  2 +-
>  tcg/riscv/tcg-target.h       |  8 ++++++--
>  tcg/s390/tcg-target.h        |  3 ++-
>  tcg/sparc/tcg-target.h       |  8 +++++---
>  tcg/tci/tcg-target.h         |  3 ++-
>  softmmu/physmem.c            |  9 ++++++++-
>  tcg/tcg.c                    |  6 ++++--
>  tcg/aarch64/tcg-target.c.inc |  2 +-
>  tcg/mips/tcg-target.c.inc    |  2 +-
>  tcg/ppc/tcg-target.c.inc     | 21 +++++++++++----------
>  tcg/sparc/tcg-target.c.inc   |  4 ++--
>  15 files changed, 64 insertions(+), 32 deletions(-)
>
> diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
> index 663dd0b95e..d0a6a059b7 100644
> --- a/tcg/aarch64/tcg-target.h
> +++ b/tcg/aarch64/tcg-target.h
> @@ -148,9 +148,14 @@ typedef enum {
>  #define TCG_TARGET_DEFAULT_MO (0)
>  #define TCG_TARGET_HAS_MEMORY_BSWAP     1
>  
> -static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
> +/* Flush the dcache at RW, and the icache at RX, as necessary. */
> +static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
>  {
> -    __builtin___clear_cache((char *)start, (char *)stop);
> +    /* TODO: Copy this from gcc to avoid 4 loops instead of 2. */

Why not do it now?

> +    if (rw != rx) {
> +        __builtin___clear_cache((char *)rw, (char *)(rw + len));
> +    }
> +    __builtin___clear_cache((char *)rx, (char *)(rx + len));
>  }
>  
<snip>
>  
> -static inline void flush_icache_range(uintptr_t start, uintptr_t stop)
> +/* Flush the dcache at RW, and the icache at RX, as necessary. */
> +static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
>  {
>  }

I take it i386 is just too primitive to care about flushing things?

Anyway:

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer
  2020-11-06 20:31   ` Alex Bennée
@ 2020-11-06 22:51     ` Richard Henderson
  2020-11-07 21:08       ` Alex Bennée
  0 siblings, 1 reply; 57+ messages in thread
From: Richard Henderson @ 2020-11-06 22:51 UTC (permalink / raw)
  To: Alex Bennée; +Cc: j, qemu-devel

On 11/6/20 12:31 PM, Alex Bennée wrote:
>> +/* Flush the dcache at RW, and the icache at RX, as necessary. */
>> +static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
>>  {
>> -    __builtin___clear_cache((char *)start, (char *)stop);
>> +    /* TODO: Copy this from gcc to avoid 4 loops instead of 2. */
> 
> Why not do it now?

Do you really want to be including that into a patch that touches all host
backends?  I do it later, in a change by itself.

> I take it i386 is just too primitive to care about flushing things?

Or the reverse.  ;-)


r~


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer
  2020-11-06 22:51     ` Richard Henderson
@ 2020-11-07 21:08       ` Alex Bennée
  0 siblings, 0 replies; 57+ messages in thread
From: Alex Bennée @ 2020-11-07 21:08 UTC (permalink / raw)
  To: Richard Henderson; +Cc: j, qemu-devel


Richard Henderson <richard.henderson@linaro.org> writes:

> On 11/6/20 12:31 PM, Alex Bennée wrote:
>>> +/* Flush the dcache at RW, and the icache at RX, as necessary. */
>>> +static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
>>>  {
>>> -    __builtin___clear_cache((char *)start, (char *)stop);
>>> +    /* TODO: Copy this from gcc to avoid 4 loops instead of 2. */
>> 
>> Why not do it now?
>
> Do you really want to be including that into a patch that touches all host
> backends?  I do it later, in a change by itself.

Ahh I hadn't noticed. I was just thrown by the TODO and wondering what
it was.

>
>> I take it i386 is just too primitive to care about flushing things?
>
> Or the reverse.  ;-)
>
>
> r~


-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-07 21:10   ` Alex Bennée
  2020-11-06  3:28 ` [PATCH v3 03/41] tcg: Move tcg epilogue " Richard Henderson
                   ` (41 subsequent siblings)
  43 siblings, 1 reply; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

This value is constant across all thread-local copies of TCGContext,
so we might as well move it out of thread-local storage.

Use the correct function pointer type, and name the variable
tcg_qemu_tb_exec, which means that we are able to remove the
macro that does the casting.

Replace HAVE_TCG_QEMU_TB_EXEC with CONFIG_TCG_INTERPRETER,
as this is somewhat clearer in intent.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h    | 9 ++++-----
 tcg/tci/tcg-target.h | 2 --
 tcg/tcg.c            | 9 ++++++++-
 tcg/tci.c            | 3 ++-
 4 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 8ff9dad4ef..9cc412f90c 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -621,7 +621,6 @@ struct TCGContext {
        here, because there's too much arithmetic throughout that relies
        on addition and subtraction working on bytes.  Rely on the GCC
        extension that allows arithmetic on void*.  */
-    void *code_gen_prologue;
     void *code_gen_epilogue;
     void *code_gen_buffer;
     size_t code_gen_buffer_size;
@@ -1222,11 +1221,11 @@ static inline unsigned get_mmuidx(TCGMemOpIdx oi)
 #define TB_EXIT_IDXMAX    1
 #define TB_EXIT_REQUESTED 3
 
-#ifdef HAVE_TCG_QEMU_TB_EXEC
-uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr);
+#ifdef CONFIG_TCG_INTERPRETER
+uintptr_t tcg_qemu_tb_exec(CPUArchState *env, void *tb_ptr);
 #else
-# define tcg_qemu_tb_exec(env, tb_ptr) \
-    ((uintptr_t (*)(void *, void *))tcg_ctx->code_gen_prologue)(env, tb_ptr)
+typedef uintptr_t tcg_prologue_fn(CPUArchState *env, void *tb_ptr);
+extern tcg_prologue_fn *tcg_qemu_tb_exec;
 #endif
 
 void tcg_register_jit(void *buf, size_t buf_size);
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 6460449719..49f3291f8a 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -189,8 +189,6 @@ typedef enum {
 
 void tci_disas(uint8_t opc);
 
-#define HAVE_TCG_QEMU_TB_EXEC
-
 /* Flush the dcache at RW, and the icache at RX, as necessary. */
 static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
diff --git a/tcg/tcg.c b/tcg/tcg.c
index d5a72c226f..a6f47b033c 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -161,6 +161,10 @@ static TCGContext **tcg_ctxs;
 static unsigned int n_tcg_ctxs;
 TCGv_env cpu_env = 0;
 
+#ifndef CONFIG_TCG_INTERPRETER
+tcg_prologue_fn *tcg_qemu_tb_exec;
+#endif
+
 struct tcg_region_tree {
     QemuMutex lock;
     GTree *tree;
@@ -1054,7 +1058,10 @@ void tcg_prologue_init(TCGContext *s)
     s->code_ptr = buf0;
     s->code_buf = buf0;
     s->data_gen_ptr = NULL;
-    s->code_gen_prologue = buf0;
+
+#ifndef CONFIG_TCG_INTERPRETER
+    tcg_qemu_tb_exec = (tcg_prologue_fn *)buf0;
+#endif
 
     /* Compute a high-water mark, at which we voluntarily flush the buffer
        and start over.  The size here is arbitrary, significantly larger
diff --git a/tcg/tci.c b/tcg/tci.c
index 82039fd163..d996eb7cf8 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -475,8 +475,9 @@ static bool tci_compare64(uint64_t u0, uint64_t u1, TCGCond condition)
 #endif
 
 /* Interpret pseudo code in tb. */
-uintptr_t tcg_qemu_tb_exec(CPUArchState *env, uint8_t *tb_ptr)
+uintptr_t tcg_qemu_tb_exec(CPUArchState *env, void *v_tb_ptr)
 {
+    uint8_t *tb_ptr = v_tb_ptr;
     tcg_target_ulong regs[TCG_TARGET_NB_REGS];
     long tcg_temps[CPU_TEMP_BUF_NLONGS];
     uintptr_t sp_value = (uintptr_t)(tcg_temps + CPU_TEMP_BUF_NLONGS);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext
  2020-11-06  3:28 ` [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext Richard Henderson
@ 2020-11-07 21:10   ` Alex Bennée
  0 siblings, 0 replies; 57+ messages in thread
From: Alex Bennée @ 2020-11-07 21:10 UTC (permalink / raw)
  To: Richard Henderson; +Cc: j, qemu-devel


Richard Henderson <richard.henderson@linaro.org> writes:

> This value is constant across all thread-local copies of TCGContext,
> so we might as well move it out of thread-local storage.
>
> Use the correct function pointer type, and name the variable
> tcg_qemu_tb_exec, which means that we are able to remove the
> macro that does the casting.
>
> Replace HAVE_TCG_QEMU_TB_EXEC with CONFIG_TCG_INTERPRETER,
> as this is somewhat clearer in intent.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>

-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* [PATCH v3 03/41] tcg: Move tcg epilogue pointer out of TCGContext
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 04/41] tcg: Add in_code_gen_buffer Richard Henderson
                   ` (40 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

This value is constant across all thread-local copies of TCGContext,
so we might as well move it out of thread-local storage.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h            | 2 +-
 accel/tcg/tcg-runtime.c      | 2 +-
 tcg/tcg.c                    | 3 ++-
 tcg/aarch64/tcg-target.c.inc | 4 ++--
 tcg/arm/tcg-target.c.inc     | 2 +-
 tcg/i386/tcg-target.c.inc    | 4 ++--
 tcg/mips/tcg-target.c.inc    | 2 +-
 tcg/ppc/tcg-target.c.inc     | 2 +-
 tcg/riscv/tcg-target.c.inc   | 4 ++--
 tcg/s390/tcg-target.c.inc    | 4 ++--
 tcg/sparc/tcg-target.c.inc   | 2 +-
 11 files changed, 16 insertions(+), 15 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 9cc412f90c..bb1e97b13b 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -621,7 +621,6 @@ struct TCGContext {
        here, because there's too much arithmetic throughout that relies
        on addition and subtraction working on bytes.  Rely on the GCC
        extension that allows arithmetic on void*.  */
-    void *code_gen_epilogue;
     void *code_gen_buffer;
     size_t code_gen_buffer_size;
     void *code_gen_ptr;
@@ -678,6 +677,7 @@ struct TCGContext {
 
 extern TCGContext tcg_init_ctx;
 extern __thread TCGContext *tcg_ctx;
+extern void *tcg_code_gen_epilogue;
 extern TCGv_env cpu_env;
 
 static inline size_t temp_idx(TCGTemp *ts)
diff --git a/accel/tcg/tcg-runtime.c b/accel/tcg/tcg-runtime.c
index 446465a09a..f85dfefeab 100644
--- a/accel/tcg/tcg-runtime.c
+++ b/accel/tcg/tcg-runtime.c
@@ -154,7 +154,7 @@ void *HELPER(lookup_tb_ptr)(CPUArchState *env)
 
     tb = tb_lookup__cpu_state(cpu, &pc, &cs_base, &flags, curr_cflags());
     if (tb == NULL) {
-        return tcg_ctx->code_gen_epilogue;
+        return tcg_code_gen_epilogue;
     }
     qemu_log_mask_and_addr(CPU_LOG_EXEC, pc,
                            "Chain %d: %p ["
diff --git a/tcg/tcg.c b/tcg/tcg.c
index a6f47b033c..1e83eed83f 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -160,6 +160,7 @@ static int tcg_out_ldst_finalize(TCGContext *s);
 static TCGContext **tcg_ctxs;
 static unsigned int n_tcg_ctxs;
 TCGv_env cpu_env = 0;
+void *tcg_code_gen_epilogue;
 
 #ifndef CONFIG_TCG_INTERPRETER
 tcg_prologue_fn *tcg_qemu_tb_exec;
@@ -1130,7 +1131,7 @@ void tcg_prologue_init(TCGContext *s)
 
     /* Assert that goto_ptr is implemented completely.  */
     if (TCG_TARGET_HAS_goto_ptr) {
-        tcg_debug_assert(s->code_gen_epilogue != NULL);
+        tcg_debug_assert(tcg_code_gen_epilogue != NULL);
     }
 }
 
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 83af3108a4..76f8ae48ad 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1873,7 +1873,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
     case INDEX_op_exit_tb:
         /* Reuse the zeroing that exists for goto_ptr.  */
         if (a0 == 0) {
-            tcg_out_goto_long(s, s->code_gen_epilogue);
+            tcg_out_goto_long(s, tcg_code_gen_epilogue);
         } else {
             tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_X0, a0);
             tcg_out_goto_long(s, tb_ret_addr);
@@ -2894,7 +2894,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_X0, 0);
 
     /* TB epilogue */
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 62c37a954b..1e32bf42b8 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -2297,7 +2297,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R0, 0);
     tcg_out_epilogue(s);
 }
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index d8797ed398..424dd1cdcf 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -2267,7 +2267,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
     case INDEX_op_exit_tb:
         /* Reuse the zeroing that exists for goto_ptr.  */
         if (a0 == 0) {
-            tcg_out_jmp(s, s->code_gen_epilogue);
+            tcg_out_jmp(s, tcg_code_gen_epilogue);
         } else {
             tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_EAX, a0);
             tcg_out_jmp(s, tb_ret_addr);
@@ -3825,7 +3825,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_EAX, 0);
 
     /* TB epilogue */
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index c255ecb444..f641105f9a 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -2483,7 +2483,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_V0, TCG_REG_ZERO);
 
     /* TB epilogue */
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index a848e98383..be116c6164 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -2341,7 +2341,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out32(s, BCCTR | BO_ALWAYS);
 
     /* Epilogue */
-    s->code_gen_epilogue = tb_ret_addr = s->code_ptr;
+    tcg_code_gen_epilogue = tb_ret_addr = s->code_ptr;
 
     tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_R0, TCG_REG_R1, FRAME_SIZE+LR_OFFSET);
     for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); ++i) {
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index d536f3ccc1..ab08af7457 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -1288,7 +1288,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
     case INDEX_op_exit_tb:
         /* Reuse the zeroing that exists for goto_ptr.  */
         if (a0 == 0) {
-            tcg_out_call_int(s, s->code_gen_epilogue, true);
+            tcg_out_call_int(s, tcg_code_gen_epilogue, true);
         } else {
             tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_A0, a0);
             tcg_out_call_int(s, tb_ret_addr, true);
@@ -1822,7 +1822,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, tcg_target_call_iarg_regs[1], 0);
 
     /* Return path for goto_ptr. Set return value to 0 */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_A0, TCG_REG_ZERO);
 
     /* TB epilogue */
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index c5e096449b..ac99ccea73 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -1756,7 +1756,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
         /* Reuse the zeroing that exists for goto_ptr.  */
         a0 = args[0];
         if (a0 == 0) {
-            tgen_gotoi(s, S390_CC_ALWAYS, s->code_gen_epilogue);
+            tgen_gotoi(s, S390_CC_ALWAYS, tcg_code_gen_epilogue);
         } else {
             tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, a0);
             tgen_gotoi(s, S390_CC_ALWAYS, tb_ret_addr);
@@ -2561,7 +2561,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, 0);
 
     /* TB epilogue */
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 6e2d755f6a..5b3bc91b05 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -1038,7 +1038,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_nop(s);
 
     /* Epilogue for goto_ptr.  */
-    s->code_gen_epilogue = s->code_ptr;
+    tcg_code_gen_epilogue = s->code_ptr;
     tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN);
     /* delay slot */
     tcg_out_movi_imm13(s, TCG_REG_O0, 0);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 04/41] tcg: Add in_code_gen_buffer
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (2 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 03/41] tcg: Move tcg epilogue " Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 05/41] tcg: Introduce tcg_splitwx_to_{rx,rw} Richard Henderson
                   ` (39 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Create a function to determine if a pointer is within the buffer.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h         |  6 ++++++
 accel/tcg/translate-all.c | 26 ++++++++------------------
 2 files changed, 14 insertions(+), 18 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index bb1e97b13b..e4d0ace44b 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -680,6 +680,12 @@ extern __thread TCGContext *tcg_ctx;
 extern void *tcg_code_gen_epilogue;
 extern TCGv_env cpu_env;
 
+static inline bool in_code_gen_buffer(const void *p)
+{
+    const TCGContext *s = &tcg_init_ctx;
+    return (size_t)(p - s->code_gen_buffer) <= s->code_gen_buffer_size;
+}
+
 static inline size_t temp_idx(TCGTemp *ts)
 {
     ptrdiff_t n = ts - tcg_ctx->temps;
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 4572b4901f..744f97a717 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -392,27 +392,18 @@ void tb_destroy(TranslationBlock *tb)
 
 bool cpu_restore_state(CPUState *cpu, uintptr_t host_pc, bool will_exit)
 {
-    TranslationBlock *tb;
-    bool r = false;
-    uintptr_t check_offset;
-
-    /* The host_pc has to be in the region of current code buffer. If
-     * it is not we will not be able to resolve it here. The two cases
-     * where host_pc will not be correct are:
+    /*
+     * The host_pc has to be in the region of the code buffer.
+     * If it is not we will not be able to resolve it here.
+     * The two cases where host_pc will not be correct are:
      *
      *  - fault during translation (instruction fetch)
      *  - fault from helper (not using GETPC() macro)
      *
      * Either way we need return early as we can't resolve it here.
-     *
-     * We are using unsigned arithmetic so if host_pc <
-     * tcg_init_ctx.code_gen_buffer check_offset will wrap to way
-     * above the code_gen_buffer_size
      */
-    check_offset = host_pc - (uintptr_t) tcg_init_ctx.code_gen_buffer;
-
-    if (check_offset < tcg_init_ctx.code_gen_buffer_size) {
-        tb = tcg_tb_lookup(host_pc);
+    if (in_code_gen_buffer((const void *)host_pc)) {
+        TranslationBlock *tb = tcg_tb_lookup(host_pc);
         if (tb) {
             cpu_restore_state_from_tb(cpu, tb, host_pc, will_exit);
             if (tb_cflags(tb) & CF_NOCACHE) {
@@ -421,11 +412,10 @@ bool cpu_restore_state(CPUState *cpu, uintptr_t host_pc, bool will_exit)
                 tcg_tb_remove(tb);
                 tb_destroy(tb);
             }
-            r = true;
+            return true;
         }
     }
-
-    return r;
+    return false;
 }
 
 static void page_init(void)
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 05/41] tcg: Introduce tcg_splitwx_to_{rx,rw}
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (3 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 04/41] tcg: Add in_code_gen_buffer Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 06/41] tcg: Adjust TCGLabel for const Richard Henderson
                   ` (38 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Add two helper functions, using a global variable to hold
the displacement.  The displacement is currently always 0,
so no change in behaviour.

Begin using the functions in tcg common code only.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/tcg-runtime.h   |  2 +-
 include/disas/disas.h     |  2 +-
 include/exec/exec-all.h   |  2 +-
 include/exec/log.h        |  2 +-
 include/tcg/tcg.h         | 26 ++++++++++++++----
 accel/tcg/cpu-exec.c      |  2 +-
 accel/tcg/tcg-runtime.c   |  2 +-
 accel/tcg/translate-all.c | 33 +++++++++++------------
 disas.c                   |  4 ++-
 tcg/tcg.c                 | 56 ++++++++++++++++++++++++++++++++++-----
 tcg/tci.c                 |  5 ++--
 accel/tcg/trace-events    |  2 +-
 tcg/tcg-pool.c.inc        |  6 ++++-
 13 files changed, 104 insertions(+), 40 deletions(-)

diff --git a/accel/tcg/tcg-runtime.h b/accel/tcg/tcg-runtime.h
index 4eda24e63a..c276c8beb5 100644
--- a/accel/tcg/tcg-runtime.h
+++ b/accel/tcg/tcg-runtime.h
@@ -24,7 +24,7 @@ DEF_HELPER_FLAGS_1(clrsb_i64, TCG_CALL_NO_RWG_SE, i64, i64)
 DEF_HELPER_FLAGS_1(ctpop_i32, TCG_CALL_NO_RWG_SE, i32, i32)
 DEF_HELPER_FLAGS_1(ctpop_i64, TCG_CALL_NO_RWG_SE, i64, i64)
 
-DEF_HELPER_FLAGS_1(lookup_tb_ptr, TCG_CALL_NO_WG_SE, ptr, env)
+DEF_HELPER_FLAGS_1(lookup_tb_ptr, TCG_CALL_NO_WG_SE, cptr, env)
 
 DEF_HELPER_FLAGS_1(exit_atomic, TCG_CALL_NO_WG, noreturn, env)
 
diff --git a/include/disas/disas.h b/include/disas/disas.h
index 36c33f6f19..d363e95ede 100644
--- a/include/disas/disas.h
+++ b/include/disas/disas.h
@@ -7,7 +7,7 @@
 #include "cpu.h"
 
 /* Disassemble this for me please... (debugging). */
-void disas(FILE *out, void *code, unsigned long size);
+void disas(FILE *out, const void *code, unsigned long size);
 void target_disas(FILE *out, CPUState *cpu, target_ulong code,
                   target_ulong size);
 
diff --git a/include/exec/exec-all.h b/include/exec/exec-all.h
index 4707ac140c..aa65103702 100644
--- a/include/exec/exec-all.h
+++ b/include/exec/exec-all.h
@@ -448,7 +448,7 @@ int probe_access_flags(CPUArchState *env, target_ulong addr,
  * Note: the address of search data can be obtained by adding @size to @ptr.
  */
 struct tb_tc {
-    void *ptr;    /* pointer to the translated code */
+    const void *ptr;    /* pointer to the translated code */
     size_t size;
 };
 
diff --git a/include/exec/log.h b/include/exec/log.h
index e02fff5de1..3c7fa65ead 100644
--- a/include/exec/log.h
+++ b/include/exec/log.h
@@ -56,7 +56,7 @@ static inline void log_target_disas(CPUState *cpu, target_ulong start,
     rcu_read_unlock();
 }
 
-static inline void log_disas(void *code, unsigned long size)
+static inline void log_disas(const void *code, unsigned long size)
 {
     QemuLogFile *logfile;
     rcu_read_lock();
diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index e4d0ace44b..249f83be72 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -678,6 +678,7 @@ struct TCGContext {
 extern TCGContext tcg_init_ctx;
 extern __thread TCGContext *tcg_ctx;
 extern void *tcg_code_gen_epilogue;
+extern uintptr_t tcg_splitwx_diff;
 extern TCGv_env cpu_env;
 
 static inline bool in_code_gen_buffer(const void *p)
@@ -686,6 +687,21 @@ static inline bool in_code_gen_buffer(const void *p)
     return (size_t)(p - s->code_gen_buffer) <= s->code_gen_buffer_size;
 }
 
+#ifdef CONFIG_DEBUG_TCG
+const void *tcg_splitwx_to_rx(void *rw);
+void *tcg_splitwx_to_rw(const void *rx);
+#else
+static inline const void *tcg_splitwx_to_rx(void *rw)
+{
+    return rw ? rw + tcg_splitwx_diff : NULL;
+}
+
+static inline void *tcg_splitwx_to_rw(const void *rx)
+{
+    return rx ? (void *)rx - tcg_splitwx_diff : NULL;
+}
+#endif
+
 static inline size_t temp_idx(TCGTemp *ts)
 {
     ptrdiff_t n = ts - tcg_ctx->temps;
@@ -1106,7 +1122,7 @@ static inline TCGLabel *arg_label(TCGArg i)
  * correct result.
  */
 
-static inline ptrdiff_t tcg_ptr_byte_diff(void *a, void *b)
+static inline ptrdiff_t tcg_ptr_byte_diff(const void *a, const void *b)
 {
     return a - b;
 }
@@ -1120,9 +1136,9 @@ static inline ptrdiff_t tcg_ptr_byte_diff(void *a, void *b)
  * to the destination address.
  */
 
-static inline ptrdiff_t tcg_pcrel_diff(TCGContext *s, void *target)
+static inline ptrdiff_t tcg_pcrel_diff(TCGContext *s, const void *target)
 {
-    return tcg_ptr_byte_diff(target, s->code_ptr);
+    return tcg_ptr_byte_diff(target, tcg_splitwx_to_rx(s->code_ptr));
 }
 
 /**
@@ -1228,9 +1244,9 @@ static inline unsigned get_mmuidx(TCGMemOpIdx oi)
 #define TB_EXIT_REQUESTED 3
 
 #ifdef CONFIG_TCG_INTERPRETER
-uintptr_t tcg_qemu_tb_exec(CPUArchState *env, void *tb_ptr);
+uintptr_t tcg_qemu_tb_exec(CPUArchState *env, const void *tb_ptr);
 #else
-typedef uintptr_t tcg_prologue_fn(CPUArchState *env, void *tb_ptr);
+typedef uintptr_t tcg_prologue_fn(CPUArchState *env, const void *tb_ptr);
 extern tcg_prologue_fn *tcg_qemu_tb_exec;
 #endif
 
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 58aea605d8..1e3cb570f6 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -150,7 +150,7 @@ static inline tcg_target_ulong cpu_tb_exec(CPUState *cpu, TranslationBlock *itb)
     uintptr_t ret;
     TranslationBlock *last_tb;
     int tb_exit;
-    uint8_t *tb_ptr = itb->tc.ptr;
+    const void *tb_ptr = itb->tc.ptr;
 
     qemu_log_mask_and_addr(CPU_LOG_EXEC, itb->pc,
                            "Trace %d: %p ["
diff --git a/accel/tcg/tcg-runtime.c b/accel/tcg/tcg-runtime.c
index f85dfefeab..d736f4ff55 100644
--- a/accel/tcg/tcg-runtime.c
+++ b/accel/tcg/tcg-runtime.c
@@ -145,7 +145,7 @@ uint64_t HELPER(ctpop_i64)(uint64_t arg)
     return ctpop64(arg);
 }
 
-void *HELPER(lookup_tb_ptr)(CPUArchState *env)
+const void *HELPER(lookup_tb_ptr)(CPUArchState *env)
 {
     CPUState *cpu = env_cpu(env);
     TranslationBlock *tb;
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 744f97a717..7b85ddacd2 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -269,9 +269,9 @@ static uint8_t *encode_sleb128(uint8_t *p, target_long val)
 
 /* Decode a signed leb128 sequence at *PP; increment *PP past the
    decoded value.  Return the decoded value.  */
-static target_long decode_sleb128(uint8_t **pp)
+static target_long decode_sleb128(const uint8_t **pp)
 {
-    uint8_t *p = *pp;
+    const uint8_t *p = *pp;
     target_long val = 0;
     int byte, shift = 0;
 
@@ -342,7 +342,7 @@ static int cpu_restore_state_from_tb(CPUState *cpu, TranslationBlock *tb,
     target_ulong data[TARGET_INSN_START_WORDS] = { tb->pc };
     uintptr_t host_pc = (uintptr_t)tb->tc.ptr;
     CPUArchState *env = cpu->env_ptr;
-    uint8_t *p = tb->tc.ptr + tb->tc.size;
+    const uint8_t *p = tb->tc.ptr + tb->tc.size;
     int i, j, num_insns = tb->icount;
 #ifdef CONFIG_PROFILER
     TCGProfile *prof = &tcg_ctx->prof;
@@ -393,7 +393,7 @@ void tb_destroy(TranslationBlock *tb)
 bool cpu_restore_state(CPUState *cpu, uintptr_t host_pc, bool will_exit)
 {
     /*
-     * The host_pc has to be in the region of the code buffer.
+     * The host_pc has to be in the rx region of the code buffer.
      * If it is not we will not be able to resolve it here.
      * The two cases where host_pc will not be correct are:
      *
@@ -402,7 +402,7 @@ bool cpu_restore_state(CPUState *cpu, uintptr_t host_pc, bool will_exit)
      *
      * Either way we need return early as we can't resolve it here.
      */
-    if (in_code_gen_buffer((const void *)host_pc)) {
+    if (in_code_gen_buffer((const void *)(host_pc - tcg_splitwx_diff))) {
         TranslationBlock *tb = tcg_tb_lookup(host_pc);
         if (tb) {
             cpu_restore_state_from_tb(cpu, tb, host_pc, will_exit);
@@ -1712,7 +1712,7 @@ TranslationBlock *tb_gen_code(CPUState *cpu,
     }
 
     gen_code_buf = tcg_ctx->code_gen_ptr;
-    tb->tc.ptr = gen_code_buf;
+    tb->tc.ptr = tcg_splitwx_to_rx(gen_code_buf);
     tb->pc = pc;
     tb->cs_base = cs_base;
     tb->flags = flags;
@@ -1806,15 +1806,19 @@ TranslationBlock *tb_gen_code(CPUState *cpu,
     if (qemu_loglevel_mask(CPU_LOG_TB_OUT_ASM) &&
         qemu_log_in_addr_range(tb->pc)) {
         FILE *logfile = qemu_log_lock();
-        int code_size, data_size = 0;
+        int code_size, data_size;
+        const tcg_target_ulong *rx_data_gen_ptr;
         size_t chunk_start;
         int insn = 0;
 
         if (tcg_ctx->data_gen_ptr) {
-            code_size = tcg_ctx->data_gen_ptr - tb->tc.ptr;
+            rx_data_gen_ptr = tcg_splitwx_to_rx(tcg_ctx->data_gen_ptr);
+            code_size = (const void *)rx_data_gen_ptr - tb->tc.ptr;
             data_size = gen_code_size - code_size;
         } else {
+            rx_data_gen_ptr = 0;
             code_size = gen_code_size;
+            data_size = 0;
         }
 
         /* Dump header and the first instruction */
@@ -1849,16 +1853,9 @@ TranslationBlock *tb_gen_code(CPUState *cpu,
         if (data_size) {
             int i;
             qemu_log("  data: [size=%d]\n", data_size);
-            for (i = 0; i < data_size; i += sizeof(tcg_target_ulong)) {
-                if (sizeof(tcg_target_ulong) == 8) {
-                    qemu_log("0x%08" PRIxPTR ":  .quad  0x%016" PRIx64 "\n",
-                             (uintptr_t)tcg_ctx->data_gen_ptr + i,
-                             *(uint64_t *)(tcg_ctx->data_gen_ptr + i));
-                } else {
-                    qemu_log("0x%08" PRIxPTR ":  .long  0x%08x\n",
-                             (uintptr_t)tcg_ctx->data_gen_ptr + i,
-                             *(uint32_t *)(tcg_ctx->data_gen_ptr + i));
-                }
+            for (i = 0; i < data_size / sizeof(tcg_target_ulong); i++) {
+                qemu_log("0x%08" PRIxPTR ":  .quad  0x%" TCG_PRIlx "\n",
+                         (uintptr_t)&rx_data_gen_ptr[i], rx_data_gen_ptr[i]);
             }
         }
         qemu_log("\n");
diff --git a/disas.c b/disas.c
index 7c18d7d2a7..de1de7be94 100644
--- a/disas.c
+++ b/disas.c
@@ -299,8 +299,10 @@ char *plugin_disas(CPUState *cpu, uint64_t addr, size_t size)
 }
 
 /* Disassemble this for me please... (debugging). */
-void disas(FILE *out, void *code, unsigned long size)
+void disas(FILE *out, const void *ccode, unsigned long size)
 {
+    /* TODO: Push constness through the disas backends. */
+    void *code = (void *)ccode;
     uintptr_t pc;
     int count;
     CPUDebug s;
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 1e83eed83f..cea3a4e4f2 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -161,6 +161,7 @@ static TCGContext **tcg_ctxs;
 static unsigned int n_tcg_ctxs;
 TCGv_env cpu_env = 0;
 void *tcg_code_gen_epilogue;
+uintptr_t tcg_splitwx_diff;
 
 #ifndef CONFIG_TCG_INTERPRETER
 tcg_prologue_fn *tcg_qemu_tb_exec;
@@ -405,8 +406,9 @@ static void tcg_region_trees_init(void)
     }
 }
 
-static struct tcg_region_tree *tc_ptr_to_region_tree(void *p)
+static struct tcg_region_tree *tc_ptr_to_region_tree(const void *cp)
 {
+    void *p = tcg_splitwx_to_rw(cp);
     size_t region_idx;
 
     if (p < region.start_aligned) {
@@ -700,6 +702,7 @@ void tcg_region_init(void)
     size_t region_size;
     size_t n_regions;
     size_t i;
+    uintptr_t splitwx_diff;
 
     n_regions = tcg_n_regions();
 
@@ -730,6 +733,7 @@ void tcg_region_init(void)
     region.end -= page_size;
 
     /* set guard pages */
+    splitwx_diff = tcg_splitwx_diff;
     for (i = 0; i < region.n; i++) {
         void *start, *end;
         int rc;
@@ -737,6 +741,10 @@ void tcg_region_init(void)
         tcg_region_bounds(i, &start, &end);
         rc = qemu_mprotect_none(end, page_size);
         g_assert(!rc);
+        if (splitwx_diff) {
+            rc = qemu_mprotect_none(end + splitwx_diff, page_size);
+            g_assert(!rc);
+        }
     }
 
     tcg_region_trees_init();
@@ -751,6 +759,29 @@ void tcg_region_init(void)
 #endif
 }
 
+#ifdef CONFIG_DEBUG_TCG
+const void *tcg_splitwx_to_rx(void *rw)
+{
+    /* Pass NULL pointers unchanged. */
+    if (rw) {
+        g_assert(in_code_gen_buffer(rw));
+        rw += tcg_splitwx_diff;
+    }
+    return rw;
+}
+
+void *tcg_splitwx_to_rw(const void *rx)
+{
+    /* Pass NULL pointers unchanged. */
+    if (rx) {
+        rx -= tcg_splitwx_diff;
+        /* Assert that we end with a pointer in the rw region. */
+        g_assert(in_code_gen_buffer(rx));
+    }
+    return (void *)rx;
+}
+#endif /* CONFIG_DEBUG_TCG */
+
 static void alloc_tcg_plugin_context(TCGContext *s)
 {
 #ifdef CONFIG_PLUGIN
@@ -1060,8 +1091,15 @@ void tcg_prologue_init(TCGContext *s)
     s->code_buf = buf0;
     s->data_gen_ptr = NULL;
 
+    /*
+     * The region trees are not yet configured, but tcg_splitwx_to_rx
+     * needs the bounds for an assert.
+     */
+    region.start = buf0;
+    region.end = buf0 + total_size;
+
 #ifndef CONFIG_TCG_INTERPRETER
-    tcg_qemu_tb_exec = (tcg_prologue_fn *)buf0;
+    tcg_qemu_tb_exec = (tcg_prologue_fn *)tcg_splitwx_to_rx(buf0);
 #endif
 
     /* Compute a high-water mark, at which we voluntarily flush the buffer
@@ -1085,7 +1123,7 @@ void tcg_prologue_init(TCGContext *s)
 #endif
 
     buf1 = s->code_ptr;
-    flush_idcache_range((uintptr_t)buf0, (uintptr_t)buf0,
+    flush_idcache_range((uintptr_t)tcg_splitwx_to_rx(buf0), (uintptr_t)buf0,
                         tcg_ptr_byte_diff(buf1, buf0));
 
     /* Deduct the prologue from the buffer.  */
@@ -4224,8 +4262,13 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb)
 
     tcg_reg_alloc_start(s);
 
-    s->code_buf = tb->tc.ptr;
-    s->code_ptr = tb->tc.ptr;
+    /*
+     * Reset the buffer pointers when restarting after overflow.
+     * TODO: Move this into translate-all.c with the rest of the
+     * buffer management.  Having only this done here is confusing.
+     */
+    s->code_buf = tcg_splitwx_to_rw(tb->tc.ptr);
+    s->code_ptr = s->code_buf;
 
 #ifdef TCG_TARGET_NEED_LDST_LABELS
     QSIMPLEQ_INIT(&s->ldst_labels);
@@ -4329,7 +4372,8 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb)
     }
 
     /* flush instruction cache */
-    flush_idcache_range((uintptr_t)s->code_buf, (uintptr_t)s->code_buf,
+    flush_idcache_range((uintptr_t)tcg_splitwx_to_rx(s->code_buf),
+                        (uintptr_t)s->code_buf,
                         tcg_ptr_byte_diff(s->code_ptr, s->code_buf));
 
     return tcg_current_code_size(s);
diff --git a/tcg/tci.c b/tcg/tci.c
index d996eb7cf8..262a2b39ce 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -475,9 +475,10 @@ static bool tci_compare64(uint64_t u0, uint64_t u1, TCGCond condition)
 #endif
 
 /* Interpret pseudo code in tb. */
-uintptr_t tcg_qemu_tb_exec(CPUArchState *env, void *v_tb_ptr)
+uintptr_t tcg_qemu_tb_exec(CPUArchState *env, const void *v_tb_ptr)
 {
-    uint8_t *tb_ptr = v_tb_ptr;
+    /* TODO: Propagate const through this file. */
+    uint8_t *tb_ptr = (uint8_t *)v_tb_ptr;
     tcg_target_ulong regs[TCG_TARGET_NB_REGS];
     long tcg_temps[CPU_TEMP_BUF_NLONGS];
     uintptr_t sp_value = (uintptr_t)(tcg_temps + CPU_TEMP_BUF_NLONGS);
diff --git a/accel/tcg/trace-events b/accel/tcg/trace-events
index 385b9f749b..6eefb37f5d 100644
--- a/accel/tcg/trace-events
+++ b/accel/tcg/trace-events
@@ -7,4 +7,4 @@ exec_tb_nocache(void *tb, uintptr_t pc) "tb:%p pc=0x%"PRIxPTR
 exec_tb_exit(void *last_tb, unsigned int flags) "tb:%p flags=0x%x"
 
 # translate-all.c
-translate_block(void *tb, uintptr_t pc, uint8_t *tb_code) "tb:%p, pc:0x%"PRIxPTR", tb_code:%p"
+translate_block(void *tb, uintptr_t pc, const void *tb_code) "tb:%p, pc:0x%"PRIxPTR", tb_code:%p"
diff --git a/tcg/tcg-pool.c.inc b/tcg/tcg-pool.c.inc
index 82cbcc89bd..90c2e63b7f 100644
--- a/tcg/tcg-pool.c.inc
+++ b/tcg/tcg-pool.c.inc
@@ -140,6 +140,8 @@ static int tcg_out_pool_finalize(TCGContext *s)
 
     for (; p != NULL; p = p->next) {
         size_t size = sizeof(tcg_target_ulong) * p->nlong;
+        uintptr_t value;
+
         if (!l || l->nlong != p->nlong || memcmp(l->data, p->data, size)) {
             if (unlikely(a > s->code_gen_highwater)) {
                 return -1;
@@ -148,7 +150,9 @@ static int tcg_out_pool_finalize(TCGContext *s)
             a += size;
             l = p;
         }
-        if (!patch_reloc(p->label, p->rtype, (intptr_t)a - size, p->addend)) {
+
+        value = (uintptr_t)tcg_splitwx_to_rx(a) - size;
+        if (!patch_reloc(p->label, p->rtype, value, p->addend)) {
             return -2;
         }
     }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 06/41] tcg: Adjust TCGLabel for const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (4 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 05/41] tcg: Introduce tcg_splitwx_to_{rx,rw} Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 07/41] tcg: Adjust tcg_out_call " Richard Henderson
                   ` (37 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Change TCGLabel.u.value_ptr to const, and initialize it with
tcg_splitwx_to_rx.  Propagate const through tcg/host/ only
as far as needed to avoid errors from the value_ptr change.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h            | 2 +-
 tcg/tcg.c                    | 2 +-
 tcg/aarch64/tcg-target.c.inc | 2 +-
 tcg/arm/tcg-target.c.inc     | 2 +-
 tcg/mips/tcg-target.c.inc    | 5 +++--
 tcg/ppc/tcg-target.c.inc     | 4 ++--
 tcg/s390/tcg-target.c.inc    | 2 +-
 7 files changed, 10 insertions(+), 9 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 249f83be72..b2ba16ea8c 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -261,7 +261,7 @@ struct TCGLabel {
     unsigned refs : 16;
     union {
         uintptr_t value;
-        tcg_insn_unit *value_ptr;
+        const tcg_insn_unit *value_ptr;
     } u;
     QSIMPLEQ_HEAD(, TCGRelocation) relocs;
     QSIMPLEQ_ENTRY(TCGLabel) next;
diff --git a/tcg/tcg.c b/tcg/tcg.c
index cea3a4e4f2..a88b314e97 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -305,7 +305,7 @@ static void tcg_out_label(TCGContext *s, TCGLabel *l, tcg_insn_unit *ptr)
 {
     tcg_debug_assert(!l->has_value);
     l->has_value = 1;
-    l->u.value_ptr = ptr;
+    l->u.value_ptr = tcg_splitwx_to_rx(ptr);
 }
 
 TCGLabel *gen_new_label(void)
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 76f8ae48ad..96dc9f4d0b 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1306,7 +1306,7 @@ static void tcg_out_cmp(TCGContext *s, TCGType ext, TCGReg a,
     }
 }
 
-static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
+static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target)
 {
     ptrdiff_t offset = target - s->code_ptr;
     tcg_debug_assert(offset == sextract64(offset, 0, 26));
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 1e32bf42b8..f8f485d807 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -1019,7 +1019,7 @@ static inline void tcg_out_st8(TCGContext *s, int cond,
  * with the code buffer limited to 16MB we wouldn't need the long case.
  * But we also use it for the tail-call to the qemu_ld/st helpers, which does.
  */
-static void tcg_out_goto(TCGContext *s, int cond, tcg_insn_unit *addr)
+static void tcg_out_goto(TCGContext *s, int cond, const tcg_insn_unit *addr)
 {
     intptr_t addri = (intptr_t)addr;
     ptrdiff_t disp = tcg_pcrel_diff(s, addr);
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index f641105f9a..a3f838fa51 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -144,7 +144,8 @@ static tcg_insn_unit *bswap32_addr;
 static tcg_insn_unit *bswap32u_addr;
 static tcg_insn_unit *bswap64_addr;
 
-static inline uint32_t reloc_pc16_val(tcg_insn_unit *pc, tcg_insn_unit *target)
+static inline uint32_t reloc_pc16_val(tcg_insn_unit *pc,
+                                      const tcg_insn_unit *target)
 {
     /* Let the compiler perform the right-shift as part of the arithmetic.  */
     ptrdiff_t disp = target - (pc + 1);
@@ -152,7 +153,7 @@ static inline uint32_t reloc_pc16_val(tcg_insn_unit *pc, tcg_insn_unit *target)
     return disp & 0xffff;
 }
 
-static inline void reloc_pc16(tcg_insn_unit *pc, tcg_insn_unit *target)
+static inline void reloc_pc16(tcg_insn_unit *pc, const tcg_insn_unit *target)
 {
     *pc = deposit32(*pc, 0, 16, reloc_pc16_val(pc, target));
 }
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index be116c6164..8a0b20a86b 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -184,7 +184,7 @@ static inline bool in_range_b(tcg_target_long target)
     return target == sextract64(target, 0, 26);
 }
 
-static uint32_t reloc_pc24_val(tcg_insn_unit *pc, tcg_insn_unit *target)
+static uint32_t reloc_pc24_val(tcg_insn_unit *pc, const tcg_insn_unit *target)
 {
     ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
     tcg_debug_assert(in_range_b(disp));
@@ -201,7 +201,7 @@ static bool reloc_pc24(tcg_insn_unit *pc, tcg_insn_unit *target)
     return false;
 }
 
-static uint16_t reloc_pc14_val(tcg_insn_unit *pc, tcg_insn_unit *target)
+static uint16_t reloc_pc14_val(tcg_insn_unit *pc, const tcg_insn_unit *target)
 {
     ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
     tcg_debug_assert(disp == (int16_t) disp);
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index ac99ccea73..1b5c4f0ab0 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -1302,7 +1302,7 @@ static void tgen_extract(TCGContext *s, TCGReg dest, TCGReg src,
     tcg_out_risbg(s, dest, src, 64 - len, 63, 64 - ofs, 1);
 }
 
-static void tgen_gotoi(TCGContext *s, int cc, tcg_insn_unit *dest)
+static void tgen_gotoi(TCGContext *s, int cc, const tcg_insn_unit *dest)
 {
     ptrdiff_t off = dest - s->code_ptr;
     if (off == (int16_t)off) {
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 07/41] tcg: Adjust tcg_out_call for const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (5 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 06/41] tcg: Adjust TCGLabel for const Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 08/41] tcg: Adjust tcg_out_label " Richard Henderson
                   ` (36 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

We must change all targets at once, since all must match
the declaration in tcg.c.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tcg.c                    | 2 +-
 tcg/aarch64/tcg-target.c.inc | 2 +-
 tcg/arm/tcg-target.c.inc     | 2 +-
 tcg/i386/tcg-target.c.inc    | 4 ++--
 tcg/mips/tcg-target.c.inc    | 6 +++---
 tcg/ppc/tcg-target.c.inc     | 8 ++++----
 tcg/riscv/tcg-target.c.inc   | 6 +++---
 tcg/s390/tcg-target.c.inc    | 2 +-
 tcg/sparc/tcg-target.c.inc   | 4 ++--
 tcg/tci/tcg-target.c.inc     | 2 +-
 10 files changed, 19 insertions(+), 19 deletions(-)

diff --git a/tcg/tcg.c b/tcg/tcg.c
index a88b314e97..d3eeea355c 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -148,7 +148,7 @@ static void tcg_out_st(TCGContext *s, TCGType type, TCGReg arg, TCGReg arg1,
                        intptr_t arg2);
 static bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val,
                         TCGReg base, intptr_t ofs);
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *target);
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *target);
 static int tcg_target_const_match(tcg_target_long val, TCGType type,
                                   const TCGArgConstraint *arg_ct);
 #ifdef TCG_TARGET_NEED_LDST_LABELS
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 96dc9f4d0b..6d8152c468 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1329,7 +1329,7 @@ static inline void tcg_out_callr(TCGContext *s, TCGReg reg)
     tcg_out_insn(s, 3207, BLR, reg);
 }
 
-static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *target)
+static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *target)
 {
     ptrdiff_t offset = target - s->code_ptr;
     if (offset == sextract64(offset, 0, 26)) {
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index f8f485d807..a197e6bc45 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -1033,7 +1033,7 @@ static void tcg_out_goto(TCGContext *s, int cond, const tcg_insn_unit *addr)
 
 /* The call case is mostly used for helpers - so it's not unreasonable
  * for them to be beyond branch range */
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *addr)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *addr)
 {
     intptr_t addri = (intptr_t)addr;
     ptrdiff_t disp = tcg_pcrel_diff(s, addr);
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 424dd1cdcf..095553ce28 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -1591,7 +1591,7 @@ static void tcg_out_clz(TCGContext *s, int rexw, TCGReg dest, TCGReg arg1,
     }
 }
 
-static void tcg_out_branch(TCGContext *s, int call, tcg_insn_unit *dest)
+static void tcg_out_branch(TCGContext *s, int call, const tcg_insn_unit *dest)
 {
     intptr_t disp = tcg_pcrel_diff(s, dest) - 5;
 
@@ -1610,7 +1610,7 @@ static void tcg_out_branch(TCGContext *s, int call, tcg_insn_unit *dest)
     }
 }
 
-static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *dest)
+static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest)
 {
     tcg_out_branch(s, 1, dest);
 }
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index a3f838fa51..6d2c369a85 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -517,7 +517,7 @@ static void tcg_out_opc_sa64(TCGContext *s, MIPSInsn opc1, MIPSInsn opc2,
  * Type jump.
  * Returns true if the branch was in range and the insn was emitted.
  */
-static bool tcg_out_opc_jmp(TCGContext *s, MIPSInsn opc, void *target)
+static bool tcg_out_opc_jmp(TCGContext *s, MIPSInsn opc, const void *target)
 {
     uintptr_t dest = (uintptr_t)target;
     uintptr_t from = (uintptr_t)s->code_ptr + 4;
@@ -1080,7 +1080,7 @@ static void tcg_out_movcond(TCGContext *s, TCGCond cond, TCGReg ret,
     }
 }
 
-static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
+static void tcg_out_call_int(TCGContext *s, const tcg_insn_unit *arg, bool tail)
 {
     /* Note that the ABI requires the called function's address to be
        loaded into T9, even if a direct branch is in range.  */
@@ -1098,7 +1098,7 @@ static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
     }
 }
 
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *arg)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *arg)
 {
     tcg_out_call_int(s, arg, false);
     tcg_out_nop(s);
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 8a0b20a86b..36129b976f 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -1106,7 +1106,7 @@ static void tcg_out_xori32(TCGContext *s, TCGReg dst, TCGReg src, uint32_t c)
     tcg_out_zori32(s, dst, src, c, XORI, XORIS);
 }
 
-static void tcg_out_b(TCGContext *s, int mask, tcg_insn_unit *target)
+static void tcg_out_b(TCGContext *s, int mask, const tcg_insn_unit *target)
 {
     ptrdiff_t disp = tcg_pcrel_diff(s, target);
     if (in_range_b(disp)) {
@@ -1762,13 +1762,13 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
     }
 }
 
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *target)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *target)
 {
 #ifdef _CALL_AIX
     /* Look through the descriptor.  If the branch is in range, and we
        don't have to spend too much effort on building the toc.  */
-    void *tgt = ((void **)target)[0];
-    uintptr_t toc = ((uintptr_t *)target)[1];
+    const void *tgt = ((const void * const *)target)[0];
+    uintptr_t toc = ((const uintptr_t *)target)[1];
     intptr_t diff = tcg_pcrel_diff(s, tgt);
 
     if (in_range_b(diff) && toc == (uint32_t)toc) {
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index ab08af7457..4416a93e1f 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -449,7 +449,7 @@ static bool reloc_jimm20(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
     return false;
 }
 
-static bool reloc_call(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_call(tcg_insn_unit *code_ptr, const tcg_insn_unit *target)
 {
     intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
     int32_t lo = sextreg(offset, 0, 12);
@@ -861,7 +861,7 @@ static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
     tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
 }
 
-static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
+static void tcg_out_call_int(TCGContext *s, const tcg_insn_unit *arg, bool tail)
 {
     TCGReg link = tail ? TCG_REG_ZERO : TCG_REG_RA;
     ptrdiff_t offset = tcg_pcrel_diff(s, arg);
@@ -888,7 +888,7 @@ static void tcg_out_call_int(TCGContext *s, tcg_insn_unit *arg, bool tail)
     }
 }
 
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *arg)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *arg)
 {
     tcg_out_call_int(s, arg, false);
 }
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index 1b5c4f0ab0..37f4dc4cfd 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -1415,7 +1415,7 @@ static void tgen_brcond(TCGContext *s, TCGType type, TCGCond c,
     tgen_branch(s, cc, l);
 }
 
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *dest)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest)
 {
     ptrdiff_t off = dest - s->code_ptr;
     if (off == (int32_t)off) {
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 5b3bc91b05..1a40911660 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -840,7 +840,7 @@ static void tcg_out_addsub2_i64(TCGContext *s, TCGReg rl, TCGReg rh,
     tcg_out_mov(s, TCG_TYPE_I64, rl, tmp);
 }
 
-static void tcg_out_call_nodelay(TCGContext *s, tcg_insn_unit *dest,
+static void tcg_out_call_nodelay(TCGContext *s, const tcg_insn_unit *dest,
                                  bool in_prologue)
 {
     ptrdiff_t disp = tcg_pcrel_diff(s, dest);
@@ -855,7 +855,7 @@ static void tcg_out_call_nodelay(TCGContext *s, tcg_insn_unit *dest,
     }
 }
 
-static void tcg_out_call(TCGContext *s, tcg_insn_unit *dest)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest)
 {
     tcg_out_call_nodelay(s, dest, false);
     tcg_out_nop(s);
diff --git a/tcg/tci/tcg-target.c.inc b/tcg/tci/tcg-target.c.inc
index 231b9b1775..d5a4d9d37c 100644
--- a/tcg/tci/tcg-target.c.inc
+++ b/tcg/tci/tcg-target.c.inc
@@ -545,7 +545,7 @@ static void tcg_out_movi(TCGContext *s, TCGType type,
     old_code_ptr[1] = s->code_ptr - old_code_ptr;
 }
 
-static inline void tcg_out_call(TCGContext *s, tcg_insn_unit *arg)
+static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *arg)
 {
     uint8_t *old_code_ptr = s->code_ptr;
     tcg_out_op_t(s, INDEX_op_call);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 08/41] tcg: Adjust tcg_out_label for const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (6 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 07/41] tcg: Adjust tcg_out_call " Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 09/41] tcg: Adjust tcg_register_jit " Richard Henderson
                   ` (35 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Simplify the arguments to always use s->code_ptr instead of
take it as an argument.  That makes it easy to ensure that
the value_ptr is always the rx version.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tcg.c                 |  6 +++---
 tcg/i386/tcg-target.c.inc | 10 +++++-----
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/tcg/tcg.c b/tcg/tcg.c
index d3eeea355c..e5d2208e88 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -301,11 +301,11 @@ static void tcg_out_reloc(TCGContext *s, tcg_insn_unit *code_ptr, int type,
     QSIMPLEQ_INSERT_TAIL(&l->relocs, r, next);
 }
 
-static void tcg_out_label(TCGContext *s, TCGLabel *l, tcg_insn_unit *ptr)
+static void tcg_out_label(TCGContext *s, TCGLabel *l)
 {
     tcg_debug_assert(!l->has_value);
     l->has_value = 1;
-    l->u.value_ptr = tcg_splitwx_to_rx(ptr);
+    l->u.value_ptr = tcg_splitwx_to_rx(s->code_ptr);
 }
 
 TCGLabel *gen_new_label(void)
@@ -4322,7 +4322,7 @@ int tcg_gen_code(TCGContext *s, TranslationBlock *tb)
             break;
         case INDEX_op_set_label:
             tcg_reg_alloc_bb_end(s, s->reserved_regs);
-            tcg_out_label(s, arg_label(op->args[0]), s->code_ptr);
+            tcg_out_label(s, arg_label(op->args[0]));
             break;
         case INDEX_op_call:
             tcg_reg_alloc_call(s, op);
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 095553ce28..0ac1ef3d82 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -1452,7 +1452,7 @@ static void tcg_out_brcond2(TCGContext *s, const TCGArg *args,
     default:
         tcg_abort();
     }
-    tcg_out_label(s, label_next, s->code_ptr);
+    tcg_out_label(s, label_next);
 }
 #endif
 
@@ -1494,10 +1494,10 @@ static void tcg_out_setcond2(TCGContext *s, const TCGArg *args,
 
         tcg_out_movi(s, TCG_TYPE_I32, args[0], 0);
         tcg_out_jxx(s, JCC_JMP, label_over, 1);
-        tcg_out_label(s, label_true, s->code_ptr);
+        tcg_out_label(s, label_true);
 
         tcg_out_movi(s, TCG_TYPE_I32, args[0], 1);
-        tcg_out_label(s, label_over, s->code_ptr);
+        tcg_out_label(s, label_over);
     } else {
         /* When the destination does not overlap one of the arguments,
            clear the destination first, jump if cond false, and emit an
@@ -1511,7 +1511,7 @@ static void tcg_out_setcond2(TCGContext *s, const TCGArg *args,
         tcg_out_brcond2(s, new_args, const_args+1, 1);
 
         tgen_arithi(s, ARITH_ADD, args[0], 1, 0);
-        tcg_out_label(s, label_over, s->code_ptr);
+        tcg_out_label(s, label_over);
     }
 }
 #endif
@@ -1525,7 +1525,7 @@ static void tcg_out_cmov(TCGContext *s, TCGCond cond, int rexw,
         TCGLabel *over = gen_new_label();
         tcg_out_jxx(s, tcg_cond_to_jcc[tcg_invert_cond(cond)], over, 1);
         tcg_out_mov(s, TCG_TYPE_I32, dest, v1);
-        tcg_out_label(s, over, s->code_ptr);
+        tcg_out_label(s, over);
     }
 }
 
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 09/41] tcg: Adjust tcg_register_jit for const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (7 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 08/41] tcg: Adjust tcg_out_label " Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 10/41] tcg: Adjust tb_target_set_jmp_target for split-wx Richard Henderson
                   ` (34 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

We must change all targets at once, since all must match
the declaration in tcg.c.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h            |  2 +-
 tcg/tcg.c                    | 10 +++++-----
 tcg/aarch64/tcg-target.c.inc |  2 +-
 tcg/arm/tcg-target.c.inc     |  2 +-
 tcg/i386/tcg-target.c.inc    |  2 +-
 tcg/mips/tcg-target.c.inc    |  2 +-
 tcg/ppc/tcg-target.c.inc     |  2 +-
 tcg/riscv/tcg-target.c.inc   |  2 +-
 tcg/s390/tcg-target.c.inc    |  2 +-
 tcg/sparc/tcg-target.c.inc   |  2 +-
 10 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index b2ba16ea8c..67d57695c2 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -1250,7 +1250,7 @@ typedef uintptr_t tcg_prologue_fn(CPUArchState *env, const void *tb_ptr);
 extern tcg_prologue_fn *tcg_qemu_tb_exec;
 #endif
 
-void tcg_register_jit(void *buf, size_t buf_size);
+void tcg_register_jit(const void *buf, size_t buf_size);
 
 #if TCG_TARGET_MAYBE_vec
 /* Return zero if the tuple (opc, type, vece) is unsupportable;
diff --git a/tcg/tcg.c b/tcg/tcg.c
index e5d2208e88..07a4bd2c57 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -96,7 +96,7 @@ typedef struct QEMU_PACKED {
     DebugFrameFDEHeader fde;
 } DebugFrameHeader;
 
-static void tcg_register_jit_int(void *buf, size_t size,
+static void tcg_register_jit_int(const void *buf, size_t size,
                                  const void *debug_frame,
                                  size_t debug_frame_size)
     __attribute__((unused));
@@ -1134,7 +1134,7 @@ void tcg_prologue_init(TCGContext *s)
     total_size -= prologue_size;
     s->code_gen_buffer_size = total_size;
 
-    tcg_register_jit(s->code_gen_buffer, total_size);
+    tcg_register_jit(tcg_splitwx_to_rx(s->code_gen_buffer), total_size);
 
 #ifdef DEBUG_DISAS
     if (qemu_loglevel_mask(CPU_LOG_TB_OUT_ASM)) {
@@ -4502,7 +4502,7 @@ static int find_string(const char *strtab, const char *str)
     }
 }
 
-static void tcg_register_jit_int(void *buf_ptr, size_t buf_size,
+static void tcg_register_jit_int(const void *buf_ptr, size_t buf_size,
                                  const void *debug_frame,
                                  size_t debug_frame_size)
 {
@@ -4704,13 +4704,13 @@ static void tcg_register_jit_int(void *buf_ptr, size_t buf_size,
 /* No support for the feature.  Provide the entry point expected by exec.c,
    and implement the internal function we declared earlier.  */
 
-static void tcg_register_jit_int(void *buf, size_t size,
+static void tcg_register_jit_int(const void *buf, size_t size,
                                  const void *debug_frame,
                                  size_t debug_frame_size)
 {
 }
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
 }
 #endif /* ELF_HOST_MACHINE */
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 6d8152c468..9ace859db3 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -2964,7 +2964,7 @@ static const DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index a197e6bc45..9b9400f164 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -2353,7 +2353,7 @@ static const DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 0ac1ef3d82..7f74c77d7f 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -3998,7 +3998,7 @@ static const DebugFrame debug_frame = {
 #endif
 
 #if defined(ELF_HOST_MACHINE)
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 6d2c369a85..e9c8c24741 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -2703,7 +2703,7 @@ static const DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 36129b976f..ff667b1531 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -3847,7 +3847,7 @@ static DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     uint8_t *p = &debug_frame.fde_reg_ofs[3];
     int i;
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 4416a93e1f..025e3cd0bb 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -1907,7 +1907,7 @@ static const DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index 37f4dc4cfd..1444914428 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -2620,7 +2620,7 @@ static const DebugFrame debug_frame = {
     }
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 1a40911660..4c81d5f1c2 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -1816,7 +1816,7 @@ static const DebugFrame debug_frame = {
     .fde_ret_save = { 9, 15, 31 },      /* DW_CFA_register o7, i7 */
 };
 
-void tcg_register_jit(void *buf, size_t buf_size)
+void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 10/41] tcg: Adjust tb_target_set_jmp_target for split-wx
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (8 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 09/41] tcg: Adjust tcg_register_jit " Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 11/41] tcg: Make DisasContextBase.tb const Richard Henderson
                   ` (33 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Pass both rx and rw addresses to tb_target_set_jmp_target.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.h     |  2 +-
 tcg/arm/tcg-target.h         |  2 +-
 tcg/i386/tcg-target.h        |  6 +++---
 tcg/mips/tcg-target.h        |  2 +-
 tcg/ppc/tcg-target.h         |  2 +-
 tcg/riscv/tcg-target.h       |  2 +-
 tcg/s390/tcg-target.h        |  8 ++++----
 tcg/sparc/tcg-target.h       |  2 +-
 tcg/tci/tcg-target.h         |  6 +++---
 accel/tcg/cpu-exec.c         |  4 +++-
 tcg/aarch64/tcg-target.c.inc | 12 ++++++------
 tcg/mips/tcg-target.c.inc    |  8 ++++----
 tcg/ppc/tcg-target.c.inc     | 16 ++++++++--------
 tcg/sparc/tcg-target.c.inc   | 14 +++++++-------
 14 files changed, 44 insertions(+), 42 deletions(-)

diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index d0a6a059b7..91313d93be 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -158,7 +158,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
     __builtin___clear_cache((char *)rx, (char *)(rx + len));
 }
 
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #ifdef CONFIG_SOFTMMU
 #define TCG_TARGET_NEED_LDST_LABELS
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index fa88b24e43..b21a2fb6a1 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -144,7 +144,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 }
 
 /* not defined -- call should be eliminated at compile time */
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #ifdef CONFIG_SOFTMMU
 #define TCG_TARGET_NEED_LDST_LABELS
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index 8323e72639..f52ba0ffec 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -211,11 +211,11 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
 }
 
-static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
-                                            uintptr_t jmp_addr, uintptr_t addr)
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                                            uintptr_t jmp_rw, uintptr_t addr)
 {
     /* patch the branch destination */
-    qatomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
+    qatomic_set((int32_t *)jmp_rw, addr - (jmp_rx + 4));
     /* no need to flush icache explicitly */
 }
 
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index 47b1226ee9..cd548dacec 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -216,7 +216,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
     cacheflush((void *)rx, len, ICACHE);
 }
 
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #ifdef CONFIG_SOFTMMU
 #define TCG_TARGET_NEED_LDST_LABELS
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index fbb6dc1b47..8f3e4c924a 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -176,7 +176,7 @@ extern bool have_vsx;
 #define TCG_TARGET_HAS_cmpsel_vec       0
 
 void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len);
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index 0fa6ae358e..e03fd17427 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -169,7 +169,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 }
 
 /* not defined -- call should be eliminated at compile time */
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_DEFAULT_MO (0)
 
diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h
index c3dc2e8938..c5a749e425 100644
--- a/tcg/s390/tcg-target.h
+++ b/tcg/s390/tcg-target.h
@@ -150,12 +150,12 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 {
 }
 
-static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
-                                            uintptr_t jmp_addr, uintptr_t addr)
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                                            uintptr_t jmp_rw, uintptr_t addr)
 {
     /* patch the branch destination */
-    intptr_t disp = addr - (jmp_addr - 2);
-    qatomic_set((int32_t *)jmp_addr, disp / 2);
+    intptr_t disp = addr - (jmp_rx - 2);
+    qatomic_set((int32_t *)jmp_rw, disp / 2);
     /* no need to flush icache explicitly */
 }
 
diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index c27c40231e..87e2be61e6 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -178,7 +178,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
     }
 }
 
-void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t);
+void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_NEED_POOL_LABELS
 
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 49f3291f8a..a19a6b06e5 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -201,11 +201,11 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-static inline void tb_target_set_jmp_target(uintptr_t tc_ptr,
-                                            uintptr_t jmp_addr, uintptr_t addr)
+static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                                            uintptr_t jmp_rw, uintptr_t addr)
 {
     /* patch the branch destination */
-    qatomic_set((int32_t *)jmp_addr, addr - (jmp_addr + 4));
+    qatomic_set((int32_t *)jmp_rw, addr - (jmp_rx + 4));
     /* no need to flush icache explicitly */
 }
 
diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 1e3cb570f6..272d596e0c 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -354,7 +354,9 @@ void tb_set_jmp_target(TranslationBlock *tb, int n, uintptr_t addr)
     if (TCG_TARGET_HAS_direct_jump) {
         uintptr_t offset = tb->jmp_target_arg[n];
         uintptr_t tc_ptr = (uintptr_t)tb->tc.ptr;
-        tb_target_set_jmp_target(tc_ptr, tc_ptr + offset, addr);
+        uintptr_t jmp_rx = tc_ptr + offset;
+        uintptr_t jmp_rw = jmp_rx - tcg_splitwx_diff;
+        tb_target_set_jmp_target(tc_ptr, jmp_rx, jmp_rw, addr);
     } else {
         tb->jmp_target_arg[n] = addr;
     }
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 9ace859db3..fea784cf75 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1340,21 +1340,21 @@ static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *target)
     }
 }
 
-void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
-                              uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                              uintptr_t jmp_rw, uintptr_t addr)
 {
     tcg_insn_unit i1, i2;
     TCGType rt = TCG_TYPE_I64;
     TCGReg  rd = TCG_REG_TMP;
     uint64_t pair;
 
-    ptrdiff_t offset = addr - jmp_addr;
+    ptrdiff_t offset = addr - jmp_rx;
 
     if (offset == sextract64(offset, 0, 26)) {
         i1 = I3206_B | ((offset >> 2) & 0x3ffffff);
         i2 = NOP;
     } else {
-        offset = (addr >> 12) - (jmp_addr >> 12);
+        offset = (addr >> 12) - (jmp_rx >> 12);
 
         /* patch ADRP */
         i1 = I3406_ADRP | (offset & 3) << 29 | (offset & 0x1ffffc) << (5 - 2) | rd;
@@ -1362,8 +1362,8 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
         i2 = I3401_ADDI | rt << 31 | (addr & 0xfff) << 10 | rd << 5 | rd;
     }
     pair = (uint64_t)i2 << 32 | i1;
-    qatomic_set((uint64_t *)jmp_addr, pair);
-    flush_idcache_range(jmp_addr, jmp_addr, 8);
+    qatomic_set((uint64_t *)jmp_rw, pair);
+    flush_idcache_range(jmp_rx, jmp_rw, 8);
 }
 
 static inline void tcg_out_goto_label(TCGContext *s, TCGLabel *l)
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index e9c8c24741..52638e920c 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -2657,11 +2657,11 @@ static void tcg_target_init(TCGContext *s)
     tcg_regset_set_reg(s->reserved_regs, TCG_REG_GP);   /* global pointer */
 }
 
-void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
-                              uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                              uintptr_t jmp_rw, uintptr_t addr)
 {
-    qatomic_set((uint32_t *)jmp_addr, deposit32(OPC_J, 0, 26, addr >> 2));
-    flush_idcache_range(jmp_addr, jmp_addr, 4);
+    qatomic_set((uint32_t *)jmp_rw, deposit32(OPC_J, 0, 26, addr >> 2));
+    flush_idcache_range(jmp_rx, jmp_rw, 4);
 }
 
 typedef struct {
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index ff667b1531..bc0057eedf 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -1722,13 +1722,13 @@ static void tcg_out_mb(TCGContext *s, TCGArg a0)
     tcg_out32(s, insn);
 }
 
-void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
-                              uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                              uintptr_t jmp_rw, uintptr_t addr)
 {
     if (TCG_TARGET_REG_BITS == 64) {
         tcg_insn_unit i1, i2;
         intptr_t tb_diff = addr - tc_ptr;
-        intptr_t br_diff = addr - (jmp_addr + 4);
+        intptr_t br_diff = addr - (jmp_rx + 4);
         uint64_t pair;
 
         /* This does not exercise the range of the branch, but we do
@@ -1752,13 +1752,13 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
 
         /* As per the enclosing if, this is ppc64.  Avoid the _Static_assert
            within qatomic_set that would fail to build a ppc32 host.  */
-        qatomic_set__nocheck((uint64_t *)jmp_addr, pair);
-        flush_idcache_range(jmp_addr, jmp_addr, 8);
+        qatomic_set__nocheck((uint64_t *)jmp_rw, pair);
+        flush_idcache_range(jmp_rx, jmp_rw, 8);
     } else {
-        intptr_t diff = addr - jmp_addr;
+        intptr_t diff = addr - jmp_rx;
         tcg_debug_assert(in_range_b(diff));
-        qatomic_set((uint32_t *)jmp_addr, B | (diff & 0x3fffffc));
-        flush_idcache_range(jmp_addr, jmp_addr, 4);
+        qatomic_set((uint32_t *)jmp_rw, B | (diff & 0x3fffffc));
+        flush_idcache_range(jmp_rx, jmp_rw, 4);
     }
 }
 
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 4c81d5f1c2..d599ae27b5 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -1821,11 +1821,11 @@ void tcg_register_jit(const void *buf, size_t buf_size)
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
 
-void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
-                              uintptr_t addr)
+void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
+                              uintptr_t jmp_rw, uintptr_t addr)
 {
     intptr_t tb_disp = addr - tc_ptr;
-    intptr_t br_disp = addr - jmp_addr;
+    intptr_t br_disp = addr - jmp_rx;
     tcg_insn_unit i1, i2;
 
     /* We can reach the entire address space for ILP32.
@@ -1834,9 +1834,9 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
     tcg_debug_assert(br_disp == (int32_t)br_disp);
 
     if (!USE_REG_TB) {
-        qatomic_set((uint32_t *)jmp_addr,
+        qatomic_set((uint32_t *)jmp_rw,
 		    deposit32(CALL, 0, 30, br_disp >> 2));
-        flush_idcache_range(jmp_addr, jmp_addr, 4);
+        flush_idcache_range(jmp_rx, jmp_rw, 4);
         return;
     }
 
@@ -1859,6 +1859,6 @@ void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_addr,
               | INSN_IMM13((tb_disp & 0x3ff) | -0x400));
     }
 
-    qatomic_set((uint64_t *)jmp_addr, deposit64(i2, 32, 32, i1));
-    flush_idcache_range(jmp_addr, jmp_addr, 8);
+    qatomic_set((uint64_t *)jmp_rw, deposit64(i2, 32, 32, i1));
+    flush_idcache_range(jmp_rx, jmp_rw, 8);
 }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 11/41] tcg: Make DisasContextBase.tb const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (9 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 10/41] tcg: Adjust tb_target_set_jmp_target for split-wx Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 12/41] tcg: Make tb arg to synchronize_from_tb const Richard Henderson
                   ` (32 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

There is nothing within the translators that ought to be
changing the TranslationBlock data, so make it const.

This does not actually use the read-only copy of the
data structure that exists within the rx region.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/exec/gen-icount.h  | 4 ++--
 include/exec/translator.h  | 2 +-
 include/tcg/tcg-op.h       | 2 +-
 accel/tcg/translator.c     | 4 ++--
 target/arm/translate-a64.c | 2 +-
 tcg/tcg-op.c               | 2 +-
 6 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/include/exec/gen-icount.h b/include/exec/gen-icount.h
index 822c43cfd3..aa4b44354a 100644
--- a/include/exec/gen-icount.h
+++ b/include/exec/gen-icount.h
@@ -32,7 +32,7 @@ static inline void gen_io_end(void)
     tcg_temp_free_i32(tmp);
 }
 
-static inline void gen_tb_start(TranslationBlock *tb)
+static inline void gen_tb_start(const TranslationBlock *tb)
 {
     TCGv_i32 count, imm;
 
@@ -71,7 +71,7 @@ static inline void gen_tb_start(TranslationBlock *tb)
     tcg_temp_free_i32(count);
 }
 
-static inline void gen_tb_end(TranslationBlock *tb, int num_insns)
+static inline void gen_tb_end(const TranslationBlock *tb, int num_insns)
 {
     if (tb_cflags(tb) & CF_USE_ICOUNT) {
         /* Update the num_insn immediate parameter now that we know
diff --git a/include/exec/translator.h b/include/exec/translator.h
index 638e1529c5..24232ead41 100644
--- a/include/exec/translator.h
+++ b/include/exec/translator.h
@@ -67,7 +67,7 @@ typedef enum DisasJumpType {
  * Architecture-agnostic disassembly context.
  */
 typedef struct DisasContextBase {
-    TranslationBlock *tb;
+    const TranslationBlock *tb;
     target_ulong pc_first;
     target_ulong pc_next;
     DisasJumpType is_jmp;
diff --git a/include/tcg/tcg-op.h b/include/tcg/tcg-op.h
index 5abf17fecc..cbe39a3b95 100644
--- a/include/tcg/tcg-op.h
+++ b/include/tcg/tcg-op.h
@@ -805,7 +805,7 @@ static inline void tcg_gen_insn_start(target_ulong pc, target_ulong a1,
  * be NULL and @idx should be 0.  Otherwise, @tb should be valid and
  * @idx should be one of the TB_EXIT_ values.
  */
-void tcg_gen_exit_tb(TranslationBlock *tb, unsigned idx);
+void tcg_gen_exit_tb(const TranslationBlock *tb, unsigned idx);
 
 /**
  * tcg_gen_goto_tb() - output goto_tb TCG operation
diff --git a/accel/tcg/translator.c b/accel/tcg/translator.c
index fb1e19c585..a49a794065 100644
--- a/accel/tcg/translator.c
+++ b/accel/tcg/translator.c
@@ -133,8 +133,8 @@ void translator_loop(const TranslatorOps *ops, DisasContextBase *db,
     }
 
     /* The disas_log hook may use these values rather than recompute.  */
-    db->tb->size = db->pc_next - db->pc_first;
-    db->tb->icount = db->num_insns;
+    tb->size = db->pc_next - db->pc_first;
+    tb->icount = db->num_insns;
 
 #ifdef DEBUG_DISAS
     if (qemu_loglevel_mask(CPU_LOG_TB_IN_ASM)
diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 072754fa24..297782e6ef 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -410,7 +410,7 @@ static inline bool use_goto_tb(DisasContext *s, int n, uint64_t dest)
 
 static inline void gen_goto_tb(DisasContext *s, int n, uint64_t dest)
 {
-    TranslationBlock *tb;
+    const TranslationBlock *tb;
 
     tb = s->base.tb;
     if (use_goto_tb(s, n, dest)) {
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index 4b8a473fad..e3dc0cb4cb 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -2664,7 +2664,7 @@ void tcg_gen_extr32_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i64 arg)
 
 /* QEMU specific operations.  */
 
-void tcg_gen_exit_tb(TranslationBlock *tb, unsigned idx)
+void tcg_gen_exit_tb(const TranslationBlock *tb, unsigned idx)
 {
     uintptr_t val = (uintptr_t)tb + idx;
 
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 12/41] tcg: Make tb arg to synchronize_from_tb const
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (10 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 11/41] tcg: Make DisasContextBase.tb const Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 13/41] tcg: Use Error with alloc_code_gen_buffer Richard Henderson
                   ` (31 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

There is nothing within the translators that ought to be
changing the TranslationBlock data, so make it const.

This does not actually use the read-only copy of the
data structure that exists within the rx region.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/hw/core/cpu.h   | 3 ++-
 target/arm/cpu.c        | 3 ++-
 target/avr/cpu.c        | 3 ++-
 target/hppa/cpu.c       | 3 ++-
 target/i386/cpu.c       | 3 ++-
 target/microblaze/cpu.c | 3 ++-
 target/mips/cpu.c       | 3 ++-
 target/riscv/cpu.c      | 3 ++-
 target/rx/cpu.c         | 3 ++-
 target/sh4/cpu.c        | 3 ++-
 target/sparc/cpu.c      | 3 ++-
 target/tricore/cpu.c    | 2 +-
 12 files changed, 23 insertions(+), 12 deletions(-)

diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
index 3d92c967ff..44c336a96a 100644
--- a/include/hw/core/cpu.h
+++ b/include/hw/core/cpu.h
@@ -189,7 +189,8 @@ struct CPUClass {
     void (*get_memory_mapping)(CPUState *cpu, MemoryMappingList *list,
                                Error **errp);
     void (*set_pc)(CPUState *cpu, vaddr value);
-    void (*synchronize_from_tb)(CPUState *cpu, struct TranslationBlock *tb);
+    void (*synchronize_from_tb)(CPUState *cpu,
+                                const struct TranslationBlock *tb);
     bool (*tlb_fill)(CPUState *cpu, vaddr address, int size,
                      MMUAccessType access_type, int mmu_idx,
                      bool probe, uintptr_t retaddr);
diff --git a/target/arm/cpu.c b/target/arm/cpu.c
index 07492e9f9a..2f9be1c0ee 100644
--- a/target/arm/cpu.c
+++ b/target/arm/cpu.c
@@ -54,7 +54,8 @@ static void arm_cpu_set_pc(CPUState *cs, vaddr value)
     }
 }
 
-static void arm_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void arm_cpu_synchronize_from_tb(CPUState *cs,
+                                        const TranslationBlock *tb)
 {
     ARMCPU *cpu = ARM_CPU(cs);
     CPUARMState *env = &cpu->env;
diff --git a/target/avr/cpu.c b/target/avr/cpu.c
index 5d9c4ad5bf..6f3d5a9e4a 100644
--- a/target/avr/cpu.c
+++ b/target/avr/cpu.c
@@ -41,7 +41,8 @@ static bool avr_cpu_has_work(CPUState *cs)
             && cpu_interrupts_enabled(env);
 }
 
-static void avr_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void avr_cpu_synchronize_from_tb(CPUState *cs,
+                                        const TranslationBlock *tb)
 {
     AVRCPU *cpu = AVR_CPU(cs);
     CPUAVRState *env = &cpu->env;
diff --git a/target/hppa/cpu.c b/target/hppa/cpu.c
index 71b6aca45d..e28f047d10 100644
--- a/target/hppa/cpu.c
+++ b/target/hppa/cpu.c
@@ -35,7 +35,8 @@ static void hppa_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.iaoq_b = value + 4;
 }
 
-static void hppa_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void hppa_cpu_synchronize_from_tb(CPUState *cs,
+                                         const TranslationBlock *tb)
 {
     HPPACPU *cpu = HPPA_CPU(cs);
 
diff --git a/target/i386/cpu.c b/target/i386/cpu.c
index 0d8606958e..01a8acafe3 100644
--- a/target/i386/cpu.c
+++ b/target/i386/cpu.c
@@ -7012,7 +7012,8 @@ static void x86_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.eip = value;
 }
 
-static void x86_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void x86_cpu_synchronize_from_tb(CPUState *cs,
+                                        const TranslationBlock *tb)
 {
     X86CPU *cpu = X86_CPU(cs);
 
diff --git a/target/microblaze/cpu.c b/target/microblaze/cpu.c
index 9b2482159d..c8e754cfb1 100644
--- a/target/microblaze/cpu.c
+++ b/target/microblaze/cpu.c
@@ -83,7 +83,8 @@ static void mb_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.iflags = 0;
 }
 
-static void mb_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void mb_cpu_synchronize_from_tb(CPUState *cs,
+                                       const TranslationBlock *tb)
 {
     MicroBlazeCPU *cpu = MICROBLAZE_CPU(cs);
 
diff --git a/target/mips/cpu.c b/target/mips/cpu.c
index 76d50b00b4..79eee215cf 100644
--- a/target/mips/cpu.c
+++ b/target/mips/cpu.c
@@ -44,7 +44,8 @@ static void mips_cpu_set_pc(CPUState *cs, vaddr value)
     }
 }
 
-static void mips_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void mips_cpu_synchronize_from_tb(CPUState *cs,
+                                         const TranslationBlock *tb)
 {
     MIPSCPU *cpu = MIPS_CPU(cs);
     CPUMIPSState *env = &cpu->env;
diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 6a0264fc6b..1b2f40de47 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -285,7 +285,8 @@ static void riscv_cpu_set_pc(CPUState *cs, vaddr value)
     env->pc = value;
 }
 
-static void riscv_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void riscv_cpu_synchronize_from_tb(CPUState *cs,
+                                          const TranslationBlock *tb)
 {
     RISCVCPU *cpu = RISCV_CPU(cs);
     CPURISCVState *env = &cpu->env;
diff --git a/target/rx/cpu.c b/target/rx/cpu.c
index 23ee17a701..2bb14144a7 100644
--- a/target/rx/cpu.c
+++ b/target/rx/cpu.c
@@ -33,7 +33,8 @@ static void rx_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.pc = value;
 }
 
-static void rx_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void rx_cpu_synchronize_from_tb(CPUState *cs,
+                                       const TranslationBlock *tb)
 {
     RXCPU *cpu = RX_CPU(cs);
 
diff --git a/target/sh4/cpu.c b/target/sh4/cpu.c
index 3c68021c56..1e0f05a15b 100644
--- a/target/sh4/cpu.c
+++ b/target/sh4/cpu.c
@@ -34,7 +34,8 @@ static void superh_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.pc = value;
 }
 
-static void superh_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void superh_cpu_synchronize_from_tb(CPUState *cs,
+                                           const TranslationBlock *tb)
 {
     SuperHCPU *cpu = SUPERH_CPU(cs);
 
diff --git a/target/sparc/cpu.c b/target/sparc/cpu.c
index cf21efd85f..b9241208b1 100644
--- a/target/sparc/cpu.c
+++ b/target/sparc/cpu.c
@@ -691,7 +691,8 @@ static void sparc_cpu_set_pc(CPUState *cs, vaddr value)
     cpu->env.npc = value + 4;
 }
 
-static void sparc_cpu_synchronize_from_tb(CPUState *cs, TranslationBlock *tb)
+static void sparc_cpu_synchronize_from_tb(CPUState *cs,
+                                          const TranslationBlock *tb)
 {
     SPARCCPU *cpu = SPARC_CPU(cs);
 
diff --git a/target/tricore/cpu.c b/target/tricore/cpu.c
index 2f2e5b029f..4bff1d4718 100644
--- a/target/tricore/cpu.c
+++ b/target/tricore/cpu.c
@@ -42,7 +42,7 @@ static void tricore_cpu_set_pc(CPUState *cs, vaddr value)
 }
 
 static void tricore_cpu_synchronize_from_tb(CPUState *cs,
-                                            TranslationBlock *tb)
+                                            const TranslationBlock *tb)
 {
     TriCoreCPU *cpu = TRICORE_CPU(cs);
     CPUTriCoreState *env = &cpu->env;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 13/41] tcg: Use Error with alloc_code_gen_buffer
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (11 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 12/41] tcg: Make tb arg to synchronize_from_tb const Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 14/41] tcg: Add --accel tcg,split-wx property Richard Henderson
                   ` (30 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Report better error messages than just "could not allocate".
Let alloc_code_gen_buffer set ctx->code_gen_buffer_size
and ctx->code_gen_buffer, and simply return bool.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 60 ++++++++++++++++++++++-----------------
 1 file changed, 34 insertions(+), 26 deletions(-)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 7b85ddacd2..2824b3e387 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -59,6 +59,7 @@
 #include "sysemu/cpus.h"
 #include "sysemu/cpu-timers.h"
 #include "sysemu/tcg.h"
+#include "qapi/error.h"
 
 /* #define DEBUG_TB_INVALIDATE */
 /* #define DEBUG_TB_FLUSH */
@@ -963,7 +964,7 @@ static void page_lock_pair(PageDesc **ret_p1, tb_page_addr_t phys1,
   (DEFAULT_CODE_GEN_BUFFER_SIZE_1 < MAX_CODE_GEN_BUFFER_SIZE \
    ? DEFAULT_CODE_GEN_BUFFER_SIZE_1 : MAX_CODE_GEN_BUFFER_SIZE)
 
-static inline size_t size_code_gen_buffer(size_t tb_size)
+static size_t size_code_gen_buffer(size_t tb_size)
 {
     /* Size the buffer.  */
     if (tb_size == 0) {
@@ -1014,7 +1015,7 @@ static inline void *split_cross_256mb(void *buf1, size_t size1)
 static uint8_t static_code_gen_buffer[DEFAULT_CODE_GEN_BUFFER_SIZE]
     __attribute__((aligned(CODE_GEN_ALIGN)));
 
-static inline void *alloc_code_gen_buffer(void)
+static bool alloc_code_gen_buffer(size_t tb_size, Error **errp)
 {
     void *buf = static_code_gen_buffer;
     void *end = static_code_gen_buffer + sizeof(static_code_gen_buffer);
@@ -1027,9 +1028,8 @@ static inline void *alloc_code_gen_buffer(void)
     size = end - buf;
 
     /* Honor a command-line option limiting the size of the buffer.  */
-    if (size > tcg_ctx->code_gen_buffer_size) {
-        size = QEMU_ALIGN_DOWN(tcg_ctx->code_gen_buffer_size,
-                               qemu_real_host_page_size);
+    if (size > tb_size) {
+        size = QEMU_ALIGN_DOWN(tb_size, qemu_real_host_page_size);
     }
     tcg_ctx->code_gen_buffer_size = size;
 
@@ -1041,31 +1041,43 @@ static inline void *alloc_code_gen_buffer(void)
 #endif
 
     if (qemu_mprotect_rwx(buf, size)) {
-        abort();
+        error_setg_errno(errp, errno, "mprotect of jit buffer");
+        return false;
     }
     qemu_madvise(buf, size, QEMU_MADV_HUGEPAGE);
 
-    return buf;
+    tcg_ctx->code_gen_buffer = buf;
+    return true;
 }
 #elif defined(_WIN32)
-static inline void *alloc_code_gen_buffer(void)
+static bool alloc_code_gen_buffer(size_t size, Error **errp)
 {
-    size_t size = tcg_ctx->code_gen_buffer_size;
-    return VirtualAlloc(NULL, size, MEM_RESERVE | MEM_COMMIT,
-                        PAGE_EXECUTE_READWRITE);
+    void *buf = VirtualAlloc(NULL, size, MEM_RESERVE | MEM_COMMIT,
+                             PAGE_EXECUTE_READWRITE);
+    if (buf == NULL) {
+        error_setg_win32(errp, GetLastError(),
+                         "allocate %zu bytes for jit buffer", size);
+        return false;
+    }
+
+    tcg_ctx->code_gen_buffer = buf;
+    tcg_ctx->code_gen_buffer_size = size;
+    return true;
 }
 #else
-static inline void *alloc_code_gen_buffer(void)
+static bool alloc_code_gen_buffer(size_t size, Error **errp)
 {
     int prot = PROT_WRITE | PROT_READ | PROT_EXEC;
     int flags = MAP_PRIVATE | MAP_ANONYMOUS;
-    size_t size = tcg_ctx->code_gen_buffer_size;
     void *buf;
 
     buf = mmap(NULL, size, prot, flags, -1, 0);
     if (buf == MAP_FAILED) {
-        return NULL;
+        error_setg_errno(errp, errno,
+                         "allocate %zu bytes for jit buffer", size);
+        return false;
     }
+    tcg_ctx->code_gen_buffer_size = size;
 
 #ifdef __mips__
     if (cross_256mb(buf, size)) {
@@ -1104,20 +1116,11 @@ static inline void *alloc_code_gen_buffer(void)
     /* Request large pages for the buffer.  */
     qemu_madvise(buf, size, QEMU_MADV_HUGEPAGE);
 
-    return buf;
+    tcg_ctx->code_gen_buffer = buf;
+    return true;
 }
 #endif /* USE_STATIC_CODE_GEN_BUFFER, WIN32, POSIX */
 
-static inline void code_gen_alloc(size_t tb_size)
-{
-    tcg_ctx->code_gen_buffer_size = size_code_gen_buffer(tb_size);
-    tcg_ctx->code_gen_buffer = alloc_code_gen_buffer();
-    if (tcg_ctx->code_gen_buffer == NULL) {
-        fprintf(stderr, "Could not allocate dynamic translator buffer\n");
-        exit(1);
-    }
-}
-
 static bool tb_cmp(const void *ap, const void *bp)
 {
     const TranslationBlock *a = ap;
@@ -1144,11 +1147,16 @@ static void tb_htable_init(void)
    size. */
 void tcg_exec_init(unsigned long tb_size)
 {
+    bool ok;
+
     tcg_allowed = true;
     cpu_gen_init();
     page_init();
     tb_htable_init();
-    code_gen_alloc(tb_size);
+
+    ok = alloc_code_gen_buffer(size_code_gen_buffer(tb_size), &error_fatal);
+    assert(ok);
+
 #if defined(CONFIG_SOFTMMU)
     /* There's no guest base to take into account, so go ahead and
        initialize the prologue now.  */
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 14/41] tcg: Add --accel tcg,split-wx property
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (12 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 13/41] tcg: Use Error with alloc_code_gen_buffer Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd Richard Henderson
                   ` (29 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Plumb the value through to alloc_code_gen_buffer.  This is not
supported by any os or tcg backend, so for now enabling it will
result in an error.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/sysemu/tcg.h      |  3 ++-
 tcg/aarch64/tcg-target.h  |  1 +
 tcg/arm/tcg-target.h      |  1 +
 tcg/i386/tcg-target.h     |  1 +
 tcg/mips/tcg-target.h     |  1 +
 tcg/ppc/tcg-target.h      |  1 +
 tcg/riscv/tcg-target.h    |  1 +
 tcg/s390/tcg-target.h     |  1 +
 tcg/sparc/tcg-target.h    |  1 +
 tcg/tci/tcg-target.h      |  1 +
 accel/tcg/tcg-all.c       | 26 +++++++++++++++++++++++++-
 accel/tcg/translate-all.c | 35 +++++++++++++++++++++++++++--------
 bsd-user/main.c           |  2 +-
 linux-user/main.c         |  2 +-
 14 files changed, 65 insertions(+), 12 deletions(-)

diff --git a/include/sysemu/tcg.h b/include/sysemu/tcg.h
index d9d3ca8559..00349fb18a 100644
--- a/include/sysemu/tcg.h
+++ b/include/sysemu/tcg.h
@@ -8,7 +8,8 @@
 #ifndef SYSEMU_TCG_H
 #define SYSEMU_TCG_H
 
-void tcg_exec_init(unsigned long tb_size);
+void tcg_exec_init(unsigned long tb_size, int splitwx);
+
 #ifdef CONFIG_TCG
 extern bool tcg_allowed;
 #define tcg_enabled() (tcg_allowed)
diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index 91313d93be..fa64058d43 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -164,5 +164,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif /* AARCH64_TCG_TARGET_H */
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index b21a2fb6a1..e355d6a4b2 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -150,5 +150,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index f52ba0ffec..1b9d41bd56 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -236,5 +236,6 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index cd548dacec..d231522dc9 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -206,6 +206,7 @@ extern bool use_mips32r2_instructions;
 
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 /* Flush the dcache at RW, and the icache at RX, as necessary. */
 static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index 8f3e4c924a..78d6a5e96f 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -185,5 +185,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index e03fd17427..3c2e8305b0 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -179,5 +179,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_POOL_LABELS
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP 0
+#define TCG_TARGET_SUPPORT_MIRROR   0
 
 #endif
diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h
index c5a749e425..8324197127 100644
--- a/tcg/s390/tcg-target.h
+++ b/tcg/s390/tcg-target.h
@@ -163,5 +163,6 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif
diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index 87e2be61e6..517840705f 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -181,5 +181,6 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_NEED_POOL_LABELS
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 #endif
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index a19a6b06e5..3653fef947 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -200,6 +200,7 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 #define TCG_TARGET_DEFAULT_MO  (0)
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
+#define TCG_TARGET_SUPPORT_MIRROR       0
 
 static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
                                             uintptr_t jmp_rw, uintptr_t addr)
diff --git a/accel/tcg/tcg-all.c b/accel/tcg/tcg-all.c
index fa1208158f..5f07ab09a9 100644
--- a/accel/tcg/tcg-all.c
+++ b/accel/tcg/tcg-all.c
@@ -38,6 +38,7 @@ struct TCGState {
     AccelState parent_obj;
 
     bool mttcg_enabled;
+    int splitwx_enabled;
     unsigned long tb_size;
 };
 typedef struct TCGState TCGState;
@@ -94,6 +95,13 @@ static void tcg_accel_instance_init(Object *obj)
     TCGState *s = TCG_STATE(obj);
 
     s->mttcg_enabled = default_mttcg_enabled();
+
+    /* If debugging enabled, default "auto on", otherwise off. */
+#ifdef CONFIG_DEBUG_TCG
+    s->splitwx_enabled = -1;
+#else
+    s->splitwx_enabled = 0;
+#endif
 }
 
 bool mttcg_enabled;
@@ -102,7 +110,7 @@ static int tcg_init(MachineState *ms)
 {
     TCGState *s = TCG_STATE(current_accel());
 
-    tcg_exec_init(s->tb_size * 1024 * 1024);
+    tcg_exec_init(s->tb_size * 1024 * 1024, s->splitwx_enabled);
     mttcg_enabled = s->mttcg_enabled;
     cpus_register_accel(&tcg_cpus);
 
@@ -168,6 +176,18 @@ static void tcg_set_tb_size(Object *obj, Visitor *v,
     s->tb_size = value;
 }
 
+static bool tcg_get_splitwx(Object *obj, Error **errp)
+{
+    TCGState *s = TCG_STATE(obj);
+    return s->splitwx_enabled;
+}
+
+static void tcg_set_splitwx(Object *obj, bool value, Error **errp)
+{
+    TCGState *s = TCG_STATE(obj);
+    s->splitwx_enabled = value;
+}
+
 static void tcg_accel_class_init(ObjectClass *oc, void *data)
 {
     AccelClass *ac = ACCEL_CLASS(oc);
@@ -185,6 +205,10 @@ static void tcg_accel_class_init(ObjectClass *oc, void *data)
     object_class_property_set_description(oc, "tb-size",
         "TCG translation block cache size");
 
+    object_class_property_add_bool(oc, "split-wx",
+        tcg_get_splitwx, tcg_set_splitwx);
+    object_class_property_set_description(oc, "split-wx",
+        "Map jit pages into separate RW and RX regions");
 }
 
 static const TypeInfo tcg_accel_type = {
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 2824b3e387..a29cb4a42e 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1015,13 +1015,19 @@ static inline void *split_cross_256mb(void *buf1, size_t size1)
 static uint8_t static_code_gen_buffer[DEFAULT_CODE_GEN_BUFFER_SIZE]
     __attribute__((aligned(CODE_GEN_ALIGN)));
 
-static bool alloc_code_gen_buffer(size_t tb_size, Error **errp)
+static bool alloc_code_gen_buffer(size_t tb_size, int splitwx, Error **errp)
 {
-    void *buf = static_code_gen_buffer;
-    void *end = static_code_gen_buffer + sizeof(static_code_gen_buffer);
+    void *buf, *end;
     size_t size;
 
+    if (splitwx > 0) {
+        error_setg(errp, "jit split-wx not supported");
+        return false;
+    }
+
     /* page-align the beginning and end of the buffer */
+    buf = static_code_gen_buffer;
+    end = static_code_gen_buffer + sizeof(static_code_gen_buffer);
     buf = QEMU_ALIGN_PTR_UP(buf, qemu_real_host_page_size);
     end = QEMU_ALIGN_PTR_DOWN(end, qemu_real_host_page_size);
 
@@ -1050,9 +1056,16 @@ static bool alloc_code_gen_buffer(size_t tb_size, Error **errp)
     return true;
 }
 #elif defined(_WIN32)
-static bool alloc_code_gen_buffer(size_t size, Error **errp)
+static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
 {
-    void *buf = VirtualAlloc(NULL, size, MEM_RESERVE | MEM_COMMIT,
+    void *buf;
+
+    if (splitwx > 0) {
+        error_setg(errp, "jit split-wx not supported");
+        return false;
+    }
+
+    buf = VirtualAlloc(NULL, size, MEM_RESERVE | MEM_COMMIT,
                              PAGE_EXECUTE_READWRITE);
     if (buf == NULL) {
         error_setg_win32(errp, GetLastError(),
@@ -1065,12 +1078,17 @@ static bool alloc_code_gen_buffer(size_t size, Error **errp)
     return true;
 }
 #else
-static bool alloc_code_gen_buffer(size_t size, Error **errp)
+static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
 {
     int prot = PROT_WRITE | PROT_READ | PROT_EXEC;
     int flags = MAP_PRIVATE | MAP_ANONYMOUS;
     void *buf;
 
+    if (splitwx > 0) {
+        error_setg(errp, "jit split-wx not supported");
+        return false;
+    }
+
     buf = mmap(NULL, size, prot, flags, -1, 0);
     if (buf == MAP_FAILED) {
         error_setg_errno(errp, errno,
@@ -1145,7 +1163,7 @@ static void tb_htable_init(void)
 /* Must be called before using the QEMU cpus. 'tb_size' is the size
    (in bytes) allocated to the translation buffer. Zero means default
    size. */
-void tcg_exec_init(unsigned long tb_size)
+void tcg_exec_init(unsigned long tb_size, int splitwx)
 {
     bool ok;
 
@@ -1154,7 +1172,8 @@ void tcg_exec_init(unsigned long tb_size)
     page_init();
     tb_htable_init();
 
-    ok = alloc_code_gen_buffer(size_code_gen_buffer(tb_size), &error_fatal);
+    ok = alloc_code_gen_buffer(size_code_gen_buffer(tb_size),
+                               splitwx, &error_fatal);
     assert(ok);
 
 #if defined(CONFIG_SOFTMMU)
diff --git a/bsd-user/main.c b/bsd-user/main.c
index ac40d79bfa..ffd4888a26 100644
--- a/bsd-user/main.c
+++ b/bsd-user/main.c
@@ -910,7 +910,7 @@ int main(int argc, char **argv)
     }
 
     /* init tcg before creating CPUs and to get qemu_host_page_size */
-    tcg_exec_init(0);
+    tcg_exec_init(0, false);
 
     cpu_type = parse_cpu_option(cpu_model);
     cpu = cpu_create(cpu_type);
diff --git a/linux-user/main.c b/linux-user/main.c
index 75c9785157..3856b2611d 100644
--- a/linux-user/main.c
+++ b/linux-user/main.c
@@ -705,7 +705,7 @@ int main(int argc, char **argv, char **envp)
     cpu_type = parse_cpu_option(cpu_model);
 
     /* init tcg before creating CPUs and to get qemu_host_page_size */
-    tcg_exec_init(0);
+    tcg_exec_init(0, false);
 
     cpu = cpu_create(cpu_type);
     env = cpu->env_ptr;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (13 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 14/41] tcg: Add --accel tcg,split-wx property Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-10 17:03   ` Alex Bennée
  2020-11-06  3:28 ` [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap Richard Henderson
                   ` (28 subsequent siblings)
  43 siblings, 1 reply; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

We cannot use a real temp file, because we would need to find
a filesystem that does not have noexec enabled.  However, a
memfd is not associated with any filesystem.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 84 +++++++++++++++++++++++++++++++++++----
 1 file changed, 76 insertions(+), 8 deletions(-)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index a29cb4a42e..1931e65365 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1078,17 +1078,11 @@ static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
     return true;
 }
 #else
-static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
+static bool alloc_code_gen_buffer_anon(size_t size, int prot,
+                                       int flags, Error **errp)
 {
-    int prot = PROT_WRITE | PROT_READ | PROT_EXEC;
-    int flags = MAP_PRIVATE | MAP_ANONYMOUS;
     void *buf;
 
-    if (splitwx > 0) {
-        error_setg(errp, "jit split-wx not supported");
-        return false;
-    }
-
     buf = mmap(NULL, size, prot, flags, -1, 0);
     if (buf == MAP_FAILED) {
         error_setg_errno(errp, errno,
@@ -1137,6 +1131,80 @@ static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
     tcg_ctx->code_gen_buffer = buf;
     return true;
 }
+
+#ifdef CONFIG_POSIX
+#include "qemu/memfd.h"
+
+static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
+{
+    void *buf_rw, *buf_rx;
+    int fd = -1;
+
+    buf_rw = qemu_memfd_alloc("tcg-jit", size, 0, &fd, errp);
+    if (buf_rw == NULL) {
+        return false;
+    }
+
+    buf_rx = mmap(NULL, size, PROT_READ | PROT_EXEC, MAP_SHARED, fd, 0);
+    if (buf_rx == MAP_FAILED) {
+        error_setg_errno(errp, errno,
+                         "failed to map shared memory for execute");
+        munmap(buf_rw, size);
+        close(fd);
+        return false;
+    }
+    close(fd);
+
+    tcg_ctx->code_gen_buffer = buf_rw;
+    tcg_ctx->code_gen_buffer_size = size;
+    tcg_splitwx_diff = buf_rx - buf_rw;
+
+    /* Request large pages for the buffer and the splitwx.  */
+    qemu_madvise(buf_rw, size, QEMU_MADV_HUGEPAGE);
+    qemu_madvise(buf_rx, size, QEMU_MADV_HUGEPAGE);
+    return true;
+}
+#endif /* CONFIG_POSIX */
+
+static bool alloc_code_gen_buffer_splitwx(size_t size, Error **errp)
+{
+    if (TCG_TARGET_SUPPORT_MIRROR) {
+#ifdef CONFIG_POSIX
+        return alloc_code_gen_buffer_splitwx_memfd(size, errp);
+#endif
+    }
+    error_setg(errp, "jit split-wx not supported");
+    return false;
+}
+
+static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
+{
+    ERRP_GUARD();
+    int prot, flags;
+
+    if (splitwx) {
+        if (alloc_code_gen_buffer_splitwx(size, errp)) {
+            return true;
+        }
+        /*
+         * If splitwx force-on (1), fail;
+         * if splitwx default-on (-1), fall through to splitwx off.
+         */
+        if (splitwx > 0) {
+            return false;
+        }
+        error_free_or_abort(errp);
+    }
+
+    prot = PROT_READ | PROT_WRITE | PROT_EXEC;
+    flags = MAP_PRIVATE | MAP_ANONYMOUS;
+#ifdef CONFIG_TCG_INTERPRETER
+    /* The tcg interpreter does not need execute permission. */
+    prot = PROT_READ | PROT_WRITE;
+#endif
+
+    return alloc_code_gen_buffer_anon(size, prot, flags, errp);
+}
 #endif /* USE_STATIC_CODE_GEN_BUFFER, WIN32, POSIX */
 
 static bool tb_cmp(const void *ap, const void *bp)
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd
  2020-11-06  3:28 ` [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd Richard Henderson
@ 2020-11-10 17:03   ` Alex Bennée
  2020-11-10 17:26     ` Richard Henderson
  0 siblings, 1 reply; 57+ messages in thread
From: Alex Bennée @ 2020-11-10 17:03 UTC (permalink / raw)
  To: Richard Henderson; +Cc: j, qemu-devel


Richard Henderson <richard.henderson@linaro.org> writes:

> We cannot use a real temp file, because we would need to find
> a filesystem that does not have noexec enabled.  However, a
> memfd is not associated with any filesystem.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

It looks like this breaks --enable-tcg-interpreter:

  FAILED: libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o
  cc -Ilibqemu-ppc64-softmmu.fa.p -I. -I../.. -Itarget/ppc -I../../target/ppc -I../../dtc/libfdt -I../../capstone/include/capstone -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/spice-server -I/usr/include/spice-1 -I/usr/include/pixman-1 -I/usr/include/libdrm -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include -fdiagnostics-color=auto -pipe -Wall -Winvalid-pch -Werror -std=gnu99 -O2 -g -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -m64 -mcx16 -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -Wstrict-prototypes -Wredundant-decls -Wundef -Wwrite-strings -Wmissing-prototypes -fno-strict-aliasing -fno-common -fwrapv -Wold-style-declaration -Wold-style-definition -Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wno-missing-include-dirs -Wno-shift-negative-value -Wno-psabi -fstack-protector-strong -DLEGACY_RDMA_REG_MR -isystem /home/alex/lsrc/qemu.git/linux-headers -isystem linux-headers -iquote /home/alex/lsrc/qemu.git/tcg/tci -iquote . -iquote /home/alex/lsrc/qemu.git -iquote /home/alex/lsrc/qemu.git/accel/tcg -iquote /home/alex/lsrc/qemu.git/include -iquote /home/alex/lsrc/qemu.git/disas/libvixl -pthread -fPIC -isystem../../linux-headers -isystemlinux-headers -DNEED_CPU_H '-DCONFIG_TARGET="ppc64-softmmu-config-target.h"' '-DCONFIG_DEVICES="ppc64-softmmu-config-devices.h"' -MD -MQ libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o -MF libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o.d -o libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o -c ../../accel/tcg/translate-all.c
  ../../accel/tcg/translate-all.c:1138:13: error: ‘alloc_code_gen_buffer_splitwx_memfd’ defined but not used [-Werror=unused-function]
   static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
               ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  cc1: all warnings being treated as errors


> ---
>  accel/tcg/translate-all.c | 84 +++++++++++++++++++++++++++++++++++----
>  1 file changed, 76 insertions(+), 8 deletions(-)
>
> diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
> index a29cb4a42e..1931e65365 100644
> --- a/accel/tcg/translate-all.c
> +++ b/accel/tcg/translate-all.c
> @@ -1078,17 +1078,11 @@ static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
>      return true;
>  }
>  #else
> -static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
> +static bool alloc_code_gen_buffer_anon(size_t size, int prot,
> +                                       int flags, Error **errp)
>  {
> -    int prot = PROT_WRITE | PROT_READ | PROT_EXEC;
> -    int flags = MAP_PRIVATE | MAP_ANONYMOUS;
>      void *buf;
>  
> -    if (splitwx > 0) {
> -        error_setg(errp, "jit split-wx not supported");
> -        return false;
> -    }
> -
>      buf = mmap(NULL, size, prot, flags, -1, 0);
>      if (buf == MAP_FAILED) {
>          error_setg_errno(errp, errno,
> @@ -1137,6 +1131,80 @@ static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
>      tcg_ctx->code_gen_buffer = buf;
>      return true;
>  }
> +
> +#ifdef CONFIG_POSIX
> +#include "qemu/memfd.h"
> +
> +static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
> +{
> +    void *buf_rw, *buf_rx;
> +    int fd = -1;
> +
> +    buf_rw = qemu_memfd_alloc("tcg-jit", size, 0, &fd, errp);
> +    if (buf_rw == NULL) {
> +        return false;
> +    }
> +
> +    buf_rx = mmap(NULL, size, PROT_READ | PROT_EXEC, MAP_SHARED, fd, 0);
> +    if (buf_rx == MAP_FAILED) {
> +        error_setg_errno(errp, errno,
> +                         "failed to map shared memory for execute");
> +        munmap(buf_rw, size);
> +        close(fd);
> +        return false;
> +    }
> +    close(fd);
> +
> +    tcg_ctx->code_gen_buffer = buf_rw;
> +    tcg_ctx->code_gen_buffer_size = size;
> +    tcg_splitwx_diff = buf_rx - buf_rw;
> +
> +    /* Request large pages for the buffer and the splitwx.  */
> +    qemu_madvise(buf_rw, size, QEMU_MADV_HUGEPAGE);
> +    qemu_madvise(buf_rx, size, QEMU_MADV_HUGEPAGE);
> +    return true;
> +}
> +#endif /* CONFIG_POSIX */
> +
> +static bool alloc_code_gen_buffer_splitwx(size_t size, Error **errp)
> +{
> +    if (TCG_TARGET_SUPPORT_MIRROR) {
> +#ifdef CONFIG_POSIX
> +        return alloc_code_gen_buffer_splitwx_memfd(size, errp);
> +#endif
> +    }
> +    error_setg(errp, "jit split-wx not supported");
> +    return false;
> +}
> +
> +static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
> +{
> +    ERRP_GUARD();
> +    int prot, flags;
> +
> +    if (splitwx) {
> +        if (alloc_code_gen_buffer_splitwx(size, errp)) {
> +            return true;
> +        }
> +        /*
> +         * If splitwx force-on (1), fail;
> +         * if splitwx default-on (-1), fall through to splitwx off.
> +         */
> +        if (splitwx > 0) {
> +            return false;
> +        }
> +        error_free_or_abort(errp);
> +    }
> +
> +    prot = PROT_READ | PROT_WRITE | PROT_EXEC;
> +    flags = MAP_PRIVATE | MAP_ANONYMOUS;
> +#ifdef CONFIG_TCG_INTERPRETER
> +    /* The tcg interpreter does not need execute permission. */
> +    prot = PROT_READ | PROT_WRITE;
> +#endif
> +
> +    return alloc_code_gen_buffer_anon(size, prot, flags, errp);
> +}
>  #endif /* USE_STATIC_CODE_GEN_BUFFER, WIN32, POSIX */
>  
>  static bool tb_cmp(const void *ap, const void *bp)


-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd
  2020-11-10 17:03   ` Alex Bennée
@ 2020-11-10 17:26     ` Richard Henderson
  0 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-10 17:26 UTC (permalink / raw)
  To: Alex Bennée; +Cc: j, qemu-devel

On 11/10/20 9:03 AM, Alex Bennée wrote:
> 
> Richard Henderson <richard.henderson@linaro.org> writes:
> 
>> We cannot use a real temp file, because we would need to find
>> a filesystem that does not have noexec enabled.  However, a
>> memfd is not associated with any filesystem.
>>
>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> 
> It looks like this breaks --enable-tcg-interpreter:
> 
>   FAILED: libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o
>   cc -Ilibqemu-ppc64-softmmu.fa.p -I. -I../.. -Itarget/ppc -I../../target/ppc -I../../dtc/libfdt -I../../capstone/include/capstone -Iqapi -Itrace -Iui -Iui/shader -I/usr/include/spice-server -I/usr/include/spice-1 -I/usr/include/pixman-1 -I/usr/include/libdrm -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include -fdiagnostics-color=auto -pipe -Wall -Winvalid-pch -Werror -std=gnu99 -O2 -g -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -m64 -mcx16 -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -Wstrict-prototypes -Wredundant-decls -Wundef -Wwrite-strings -Wmissing-prototypes -fno-strict-aliasing -fno-common -fwrapv -Wold-style-declaration -Wold-style-definition -Wtype-limits -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wempty-body -Wnested-externs -Wendif-labels -Wexpansion-to-defined -Wno-missing-include-dirs -Wno-shift-negative-value -Wno-psabi -fstack-protector-strong -DLEGACY_RDMA_REG_MR -isystem /home/alex/lsrc/qemu.git/linux-headers -isystem linux-headers -iquote /home/alex/lsrc/qemu.git/tcg/tci -iquote . -iquote /home/alex/lsrc/qemu.git -iquote /home/alex/lsrc/qemu.git/accel/tcg -iquote /home/alex/lsrc/qemu.git/include -iquote /home/alex/lsrc/qemu.git/disas/libvixl -pthread -fPIC -isystem../../linux-headers -isystemlinux-headers -DNEED_CPU_H '-DCONFIG_TARGET="ppc64-softmmu-config-target.h"' '-DCONFIG_DEVICES="ppc64-softmmu-config-devices.h"' -MD -MQ libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o -MF libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o.d -o libqemu-ppc64-softmmu.fa.p/accel_tcg_translate-all.c.o -c ../../accel/tcg/translate-all.c
>   ../../accel/tcg/translate-all.c:1138:13: error: ‘alloc_code_gen_buffer_splitwx_memfd’ defined but not used [-Werror=unused-function]
>    static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
>                ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>   cc1: all warnings being treated as errors

Bah, this will actually be patch 39, where I switch from if to ifdef.  I'll
have to rearrange those.


r~

^ permalink raw reply	[flat|nested] 57+ messages in thread

* [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (14 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-08  3:37   ` Joelle van Dyne
  2020-11-10 17:37   ` Alex Bennée
  2020-11-06  3:28 ` [PATCH v3 17/41] tcg: Return the TB pointer from the rx region from exit_tb Richard Henderson
                   ` (27 subsequent siblings)
  43 siblings, 2 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Cribbed from code posted by Joelle van Dyne <j@getutm.app>,
and rearranged to a cleaner structure.  Completely untested.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 65 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 65 insertions(+)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 1931e65365..17df6c94fa 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1166,9 +1166,71 @@ static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
 }
 #endif /* CONFIG_POSIX */
 
+#ifdef CONFIG_DARWIN
+#include <mach/mach.h>
+
+extern kern_return_t mach_vm_remap(vm_map_t target_task,
+                                   mach_vm_address_t *target_address,
+                                   mach_vm_size_t size,
+                                   mach_vm_offset_t mask,
+                                   int flags,
+                                   vm_map_t src_task,
+                                   mach_vm_address_t src_address,
+                                   boolean_t copy,
+                                   vm_prot_t *cur_protection,
+                                   vm_prot_t *max_protection,
+                                   vm_inherit_t inheritance);
+
+static bool alloc_code_gen_buffer_splitwx_vmremap(size_t size, Error **errp)
+{
+    kern_return_t ret;
+    mach_vm_address_t buf_rw, buf_rx;
+    vm_prot_t cur_prot, max_prot;
+
+    /* Map the read-write portion via normal anon memory. */
+    if (!alloc_code_gen_buffer_anon(size, PROT_READ | PROT_WRITE,
+                                    MAP_PRIVATE | MAP_ANONYMOUS, errp)) {
+        return false;
+    }
+
+    buf_rw = (mach_vm_address_t)tcg_ctx->code_gen_buffer;
+    buf_rx = 0;
+    ret = mach_vm_remap(mach_task_self(),
+                        &buf_rx,
+                        size,
+                        0,
+                        VM_FLAGS_ANYWHERE,
+                        mach_task_self(),
+                        buf_rw,
+                        false,
+                        &cur_prot,
+                        &max_prot,
+                        VM_INHERIT_NONE);
+    if (ret != KERN_SUCCESS) {
+        /* TODO: Convert "ret" to a human readable error message. */
+        error_setg(errp, "vm_remap for jit splitwx failed");
+        munmap((void *)buf_rw, size);
+        return false;
+    }
+
+    if (mprotect((void *)buf_rx, size, PROT_READ | PROT_EXEC) != 0) {
+        error_setg_errno(errp, errno, "mprotect for jit splitwx");
+        munmap((void *)buf_rx, size);
+        munmap((void *)buf_rw, size);
+        return false;
+    }
+
+    tcg_splitwx_diff = buf_rx - buf_rw;
+    return true;
+}
+#endif /* CONFIG_DARWIN */
+
 static bool alloc_code_gen_buffer_splitwx(size_t size, Error **errp)
 {
     if (TCG_TARGET_SUPPORT_MIRROR) {
+#ifdef CONFIG_DARWIN
+        return alloc_code_gen_buffer_splitwx_vmremap(size, errp);
+#endif
 #ifdef CONFIG_POSIX
         return alloc_code_gen_buffer_splitwx_memfd(size, errp);
 #endif
@@ -1201,6 +1263,9 @@ static bool alloc_code_gen_buffer(size_t size, int splitwx, Error **errp)
 #ifdef CONFIG_TCG_INTERPRETER
     /* The tcg interpreter does not need execute permission. */
     prot = PROT_READ | PROT_WRITE;
+#elif defined(CONFIG_DARWIN)
+    /* Applicable to both iOS and macOS (Apple Silicon). */
+    flags |= MAP_JIT;
 #endif
 
     return alloc_code_gen_buffer_anon(size, prot, flags, errp);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap
  2020-11-06  3:28 ` [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap Richard Henderson
@ 2020-11-08  3:37   ` Joelle van Dyne
  2020-11-10 17:37   ` Alex Bennée
  1 sibling, 0 replies; 57+ messages in thread
From: Joelle van Dyne @ 2020-11-08  3:37 UTC (permalink / raw)
  To: Richard Henderson; +Cc: QEMU Developers

On Thu, Nov 5, 2020 at 7:29 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> Cribbed from code posted by Joelle van Dyne <j@getutm.app>,
> and rearranged to a cleaner structure.  Completely untested.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

Reviewed-by: Joelle van Dyne <j@getutm.app>


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap
  2020-11-06  3:28 ` [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap Richard Henderson
  2020-11-08  3:37   ` Joelle van Dyne
@ 2020-11-10 17:37   ` Alex Bennée
  2020-11-10 17:57     ` Joelle van Dyne
  1 sibling, 1 reply; 57+ messages in thread
From: Alex Bennée @ 2020-11-10 17:37 UTC (permalink / raw)
  To: Richard Henderson; +Cc: j, qemu-devel


Richard Henderson <richard.henderson@linaro.org> writes:

> Cribbed from code posted by Joelle van Dyne <j@getutm.app>,
> and rearranged to a cleaner structure.  Completely untested.
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
>  accel/tcg/translate-all.c | 65 +++++++++++++++++++++++++++++++++++++++
>  1 file changed, 65 insertions(+)
>
> diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
> index 1931e65365..17df6c94fa 100644
> --- a/accel/tcg/translate-all.c
> +++ b/accel/tcg/translate-all.c
> @@ -1166,9 +1166,71 @@ static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
>  }
>  #endif /* CONFIG_POSIX */
>  
> +#ifdef CONFIG_DARWIN
> +#include <mach/mach.h>
> +
> +extern kern_return_t mach_vm_remap(vm_map_t target_task,
> +                                   mach_vm_address_t *target_address,
> +                                   mach_vm_size_t size,
> +                                   mach_vm_offset_t mask,
> +                                   int flags,
> +                                   vm_map_t src_task,
> +                                   mach_vm_address_t src_address,
> +                                   boolean_t copy,
> +                                   vm_prot_t *cur_protection,
> +                                   vm_prot_t *max_protection,
> +                                   vm_inherit_t inheritance);

Our checkpatch really doesn't like the extern being dropped in here but
having grepped the xnu source I'm not sure we have a choice. I'm curious
how stable the function might be given it's not in a published header.

-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap
  2020-11-10 17:37   ` Alex Bennée
@ 2020-11-10 17:57     ` Joelle van Dyne
  0 siblings, 0 replies; 57+ messages in thread
From: Joelle van Dyne @ 2020-11-10 17:57 UTC (permalink / raw)
  To: Alex Bennée; +Cc: Richard Henderson, Joelle van Dyne, QEMU Developers

FWIW, it's a syscall that's been around for as long as I can remember.
In macOS 11 they added a new mach_vm_remap but kept the old one for
compatibility so I don't think it's going away any time soon.

-j

On Tue, Nov 10, 2020 at 9:37 AM Alex Bennée <alex.bennee@linaro.org> wrote:
>
>
> Richard Henderson <richard.henderson@linaro.org> writes:
>
> > Cribbed from code posted by Joelle van Dyne <j@getutm.app>,
> > and rearranged to a cleaner structure.  Completely untested.
> >
> > Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> > ---
> >  accel/tcg/translate-all.c | 65 +++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 65 insertions(+)
> >
> > diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
> > index 1931e65365..17df6c94fa 100644
> > --- a/accel/tcg/translate-all.c
> > +++ b/accel/tcg/translate-all.c
> > @@ -1166,9 +1166,71 @@ static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
> >  }
> >  #endif /* CONFIG_POSIX */
> >
> > +#ifdef CONFIG_DARWIN
> > +#include <mach/mach.h>
> > +
> > +extern kern_return_t mach_vm_remap(vm_map_t target_task,
> > +                                   mach_vm_address_t *target_address,
> > +                                   mach_vm_size_t size,
> > +                                   mach_vm_offset_t mask,
> > +                                   int flags,
> > +                                   vm_map_t src_task,
> > +                                   mach_vm_address_t src_address,
> > +                                   boolean_t copy,
> > +                                   vm_prot_t *cur_protection,
> > +                                   vm_prot_t *max_protection,
> > +                                   vm_inherit_t inheritance);
>
> Our checkpatch really doesn't like the extern being dropped in here but
> having grepped the xnu source I'm not sure we have a choice. I'm curious
> how stable the function might be given it's not in a published header.
>
> --
> Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

* [PATCH v3 17/41] tcg: Return the TB pointer from the rx region from exit_tb
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (15 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 18/41] tcg/i386: Support split-wx code generation Richard Henderson
                   ` (26 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

This produces a small pc-relative displacement within the
generated code to the TB structure that preceeds it.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/cpu-exec.c | 35 ++++++++++++++++++++++-------------
 tcg/tcg-op.c         | 13 ++++++++++++-
 2 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
index 272d596e0c..8df0a1782e 100644
--- a/accel/tcg/cpu-exec.c
+++ b/accel/tcg/cpu-exec.c
@@ -144,12 +144,13 @@ static void init_delay_params(SyncClocks *sc, const CPUState *cpu)
 #endif /* CONFIG USER ONLY */
 
 /* Execute a TB, and fix up the CPU state afterwards if necessary */
-static inline tcg_target_ulong cpu_tb_exec(CPUState *cpu, TranslationBlock *itb)
+static inline TranslationBlock *cpu_tb_exec(CPUState *cpu,
+                                            TranslationBlock *itb,
+                                            int *tb_exit)
 {
     CPUArchState *env = cpu->env_ptr;
     uintptr_t ret;
     TranslationBlock *last_tb;
-    int tb_exit;
     const void *tb_ptr = itb->tc.ptr;
 
     qemu_log_mask_and_addr(CPU_LOG_EXEC, itb->pc,
@@ -177,11 +178,20 @@ static inline tcg_target_ulong cpu_tb_exec(CPUState *cpu, TranslationBlock *itb)
 
     ret = tcg_qemu_tb_exec(env, tb_ptr);
     cpu->can_do_io = 1;
-    last_tb = (TranslationBlock *)(ret & ~TB_EXIT_MASK);
-    tb_exit = ret & TB_EXIT_MASK;
-    trace_exec_tb_exit(last_tb, tb_exit);
+    /*
+     * TODO: Delay swapping back to the read-write region of the TB
+     * until we actually need to modify the TB.  The read-only copy,
+     * coming from the rx region, shares the same host TLB entry as
+     * the code that executed the exit_tb opcode that arrived here.
+     * If we insist on touching both the RX and the RW pages, we
+     * double the host TLB pressure.
+     */
+    last_tb = tcg_splitwx_to_rw((void *)(ret & ~TB_EXIT_MASK));
+    *tb_exit = ret & TB_EXIT_MASK;
 
-    if (tb_exit > TB_EXIT_IDX1) {
+    trace_exec_tb_exit(last_tb, *tb_exit);
+
+    if (*tb_exit > TB_EXIT_IDX1) {
         /* We didn't start executing this TB (eg because the instruction
          * counter hit zero); we must restore the guest PC to the address
          * of the start of the TB.
@@ -199,7 +209,7 @@ static inline tcg_target_ulong cpu_tb_exec(CPUState *cpu, TranslationBlock *itb)
             cc->set_pc(cpu, last_tb->pc);
         }
     }
-    return ret;
+    return last_tb;
 }
 
 #ifndef CONFIG_USER_ONLY
@@ -210,6 +220,7 @@ static void cpu_exec_nocache(CPUState *cpu, int max_cycles,
 {
     TranslationBlock *tb;
     uint32_t cflags = curr_cflags() | CF_NOCACHE;
+    int tb_exit;
 
     if (ignore_icount) {
         cflags &= ~CF_USE_ICOUNT;
@@ -227,7 +238,7 @@ static void cpu_exec_nocache(CPUState *cpu, int max_cycles,
 
     /* execute the generated code */
     trace_exec_tb_nocache(tb, tb->pc);
-    cpu_tb_exec(cpu, tb);
+    cpu_tb_exec(cpu, tb, &tb_exit);
 
     mmap_lock();
     tb_phys_invalidate(tb, -1);
@@ -244,6 +255,7 @@ void cpu_exec_step_atomic(CPUState *cpu)
     uint32_t flags;
     uint32_t cflags = 1;
     uint32_t cf_mask = cflags & CF_HASH_MASK;
+    int tb_exit;
 
     if (sigsetjmp(cpu->jmp_env, 0) == 0) {
         start_exclusive();
@@ -260,7 +272,7 @@ void cpu_exec_step_atomic(CPUState *cpu)
         cc->cpu_exec_enter(cpu);
         /* execute the generated code */
         trace_exec_tb(tb, pc);
-        cpu_tb_exec(cpu, tb);
+        cpu_tb_exec(cpu, tb, &tb_exit);
         cc->cpu_exec_exit(cpu);
     } else {
         /*
@@ -653,13 +665,10 @@ static inline bool cpu_handle_interrupt(CPUState *cpu,
 static inline void cpu_loop_exec_tb(CPUState *cpu, TranslationBlock *tb,
                                     TranslationBlock **last_tb, int *tb_exit)
 {
-    uintptr_t ret;
     int32_t insns_left;
 
     trace_exec_tb(tb, tb->pc);
-    ret = cpu_tb_exec(cpu, tb);
-    tb = (TranslationBlock *)(ret & ~TB_EXIT_MASK);
-    *tb_exit = ret & TB_EXIT_MASK;
+    tb = cpu_tb_exec(cpu, tb, tb_exit);
     if (*tb_exit != TB_EXIT_REQUESTED) {
         *last_tb = tb;
         return;
diff --git a/tcg/tcg-op.c b/tcg/tcg-op.c
index e3dc0cb4cb..56bb8db040 100644
--- a/tcg/tcg-op.c
+++ b/tcg/tcg-op.c
@@ -2666,7 +2666,18 @@ void tcg_gen_extr32_i64(TCGv_i64 lo, TCGv_i64 hi, TCGv_i64 arg)
 
 void tcg_gen_exit_tb(const TranslationBlock *tb, unsigned idx)
 {
-    uintptr_t val = (uintptr_t)tb + idx;
+    /*
+     * Let the jit code return the read-only version of the
+     * TranslationBlock, so that we minimize the pc-relative
+     * distance of the address of the exit_tb code to TB.
+     * This will improve utilization of pc-relative address loads.
+     *
+     * TODO: Move this to translator_loop, so that all const
+     * TranslationBlock pointers refer to read-only memory.
+     * This requires coordination with targets that do not use
+     * the translator_loop.
+     */
+    uintptr_t val = (uintptr_t)tcg_splitwx_to_rx((void *)tb) + idx;
 
     if (tb == NULL) {
         tcg_debug_assert(idx == 0);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 18/41] tcg/i386: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (16 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 17/41] tcg: Return the TB pointer from the rx region from exit_tb Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:28 ` [PATCH v3 19/41] tcg/aarch64: Use B not BL for tcg_out_goto_long Richard Henderson
                   ` (25 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/i386/tcg-target.h     |  2 +-
 tcg/i386/tcg-target.c.inc | 20 +++++++++++---------
 2 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index 1b9d41bd56..bbbd1c2d4a 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -236,6 +236,6 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 7f74c77d7f..23c7a8a383 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -165,7 +165,7 @@ static bool have_lzcnt;
 # define have_lzcnt 0
 #endif
 
-static tcg_insn_unit *tb_ret_addr;
+static const tcg_insn_unit *tb_ret_addr;
 
 static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
                         intptr_t value, intptr_t addend)
@@ -173,7 +173,7 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
     value += addend;
     switch(type) {
     case R_386_PC32:
-        value -= (uintptr_t)code_ptr;
+        value -= (uintptr_t)tcg_splitwx_to_rx(code_ptr);
         if (value != (int32_t)value) {
             return false;
         }
@@ -182,7 +182,7 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
         tcg_patch32(code_ptr, value);
         break;
     case R_386_PC8:
-        value -= (uintptr_t)code_ptr;
+        value -= (uintptr_t)tcg_splitwx_to_rx(code_ptr);
         if (value != (int8_t)value) {
             return false;
         }
@@ -1006,7 +1006,7 @@ static void tcg_out_movi(TCGContext *s, TCGType type,
     }
 
     /* Try a 7 byte pc-relative lea before the 10 byte movq.  */
-    diff = arg - ((uintptr_t)s->code_ptr + 7);
+    diff = tcg_pcrel_diff(s, (const void *)arg) - 7;
     if (diff == (int32_t)diff) {
         tcg_out_opc(s, OPC_LEA | P_REXW, ret, 0, 0);
         tcg_out8(s, (LOWREGMASK(ret) << 3) | 5);
@@ -1615,7 +1615,7 @@ static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest)
     tcg_out_branch(s, 1, dest);
 }
 
-static void tcg_out_jmp(TCGContext *s, tcg_insn_unit *dest)
+static void tcg_out_jmp(TCGContext *s, const tcg_insn_unit *dest)
 {
     tcg_out_branch(s, 0, dest);
 }
@@ -1786,7 +1786,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, bool is_64,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
     if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) {
         label->label_ptr[1] = label_ptr[1];
@@ -2280,7 +2281,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
             /* jump displacement must be aligned for atomic patching;
              * see if we need to add extra nops before jump
              */
-            gap = tcg_pcrel_diff(s, QEMU_ALIGN_PTR_UP(s->code_ptr + 1, 4));
+            gap = QEMU_ALIGN_PTR_UP(s->code_ptr + 1, 4) - s->code_ptr;
             if (gap != 1) {
                 tcg_out_nopn(s, gap - 1);
             }
@@ -3825,11 +3826,12 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_EAX, 0);
 
     /* TB epilogue */
-    tb_ret_addr = s->code_ptr;
+    tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr);
 
     tcg_out_addi(s, TCG_REG_CALL_STACK, stack_addend);
 
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 19/41] tcg/aarch64: Use B not BL for tcg_out_goto_long
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (17 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 18/41] tcg/i386: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:28 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 20/41] tcg/aarch64: Implement flush_idcache_range manually Richard Henderson
                   ` (24 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:28 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

A typo generated a branch-and-link insn instead of plain branch.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.c.inc | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index fea784cf75..bd888bc66d 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1317,7 +1317,7 @@ static inline void tcg_out_goto_long(TCGContext *s, tcg_insn_unit *target)
 {
     ptrdiff_t offset = target - s->code_ptr;
     if (offset == sextract64(offset, 0, 26)) {
-        tcg_out_insn(s, 3206, BL, offset);
+        tcg_out_insn(s, 3206, B, offset);
     } else {
         tcg_out_movi(s, TCG_TYPE_I64, TCG_REG_TMP, (intptr_t)target);
         tcg_out_insn(s, 3207, BR, TCG_REG_TMP);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 20/41] tcg/aarch64: Implement flush_idcache_range manually
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (18 preceding siblings ...)
  2020-11-06  3:28 ` [PATCH v3 19/41] tcg/aarch64: Use B not BL for tcg_out_goto_long Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 21/41] tcg/aarch64: Support split-wx code generation Richard Henderson
                   ` (23 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Copy the single pointer implementation from libgcc and modify it to
support the double pointer interface we require.  This halves the
number of cache operations required when split-rwx is enabled.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.h     | 11 +------
 tcg/aarch64/tcg-target.c.inc | 64 ++++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+), 10 deletions(-)

diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index fa64058d43..e62d38ba55 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -148,16 +148,7 @@ typedef enum {
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
 
-/* Flush the dcache at RW, and the icache at RX, as necessary. */
-static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
-{
-    /* TODO: Copy this from gcc to avoid 4 loops instead of 2. */
-    if (rw != rx) {
-        __builtin___clear_cache((char *)rw, (char *)(rw + len));
-    }
-    __builtin___clear_cache((char *)rx, (char *)(rx + len));
-}
-
+void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len);
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #ifdef CONFIG_SOFTMMU
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index bd888bc66d..8aa1fafd91 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -2968,3 +2968,67 @@ void tcg_register_jit(const void *buf, size_t buf_size)
 {
     tcg_register_jit_int(buf, buf_size, &debug_frame, sizeof(debug_frame));
 }
+
+/* Flush the dcache at RW, and the icache at RX, as necessary. */
+#ifdef CONFIG_DARWIN
+/* Apple does not expose CTR_EL0, so we must use system interfaces. */
+extern void sys_icache_invalidate(void *start, size_t len);
+extern void sys_dcache_flush(void *start, size_t len);
+void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
+{
+    sys_dcache_flush((void *)rw, len);
+    sys_icache_invalidate((void *)rx, len);
+}
+#else
+/*
+ * This is a copy of gcc's __aarch64_sync_cache_range, modified
+ * to fit this three-operand interface.
+ */
+void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
+{
+    const unsigned CTR_IDC = 1u << 28;
+    const unsigned CTR_DIC = 1u << 29;
+    static unsigned int cache_info;
+    uintptr_t icache_lsize, dcache_lsize, p;
+
+    if (!cache_info) {
+        /*
+         * CTR_EL0 [3:0] contains log2 of icache line size in words.
+         * CTR_EL0 [19:16] contains log2 of dcache line size in words.
+         */
+        asm volatile("mrs\t%0, ctr_el0" : "=r"(cache_info));
+    }
+
+    icache_lsize = 4 << extract32(cache_info, 0, 4);
+    dcache_lsize = 4 << extract32(cache_info, 16, 4);
+
+    /*
+     * If CTR_EL0.IDC is enabled, Data cache clean to the Point of Unification
+     * is not required for instruction to data coherence.
+     */
+    if (!(cache_info & CTR_IDC)) {
+        /*
+         * Loop over the address range, clearing one cache line at once.
+         * Data cache must be flushed to unification first to make sure
+         * the instruction cache fetches the updated data.
+         */
+        for (p = rw & -dcache_lsize; p < rw + len; p += dcache_lsize) {
+            asm volatile("dc\tcvau, %0" : : "r" (p) : "memory");
+        }
+        asm volatile("dsb\tish" : : : "memory");
+    }
+
+    /*
+     * If CTR_EL0.DIC is enabled, Instruction cache cleaning to the Point
+     * of Unification is not required for instruction to data coherence.
+     */
+    if (!(cache_info & CTR_DIC)) {
+        for (p = rx & -icache_lsize; p < rx + len; p += icache_lsize) {
+            asm volatile("ic\tivau, %0" : : "r"(p) : "memory");
+        }
+        asm volatile ("dsb\tish" : : : "memory");
+    }
+
+    asm volatile("isb" : : : "memory");
+}
+#endif /* CONFIG_DARWIN */
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 21/41] tcg/aarch64: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (19 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 20/41] tcg/aarch64: Implement flush_idcache_range manually Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 22/41] disas: Push const down through host disasassembly Richard Henderson
                   ` (22 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.h     |  2 +-
 tcg/aarch64/tcg-target.c.inc | 57 ++++++++++++++++++++----------------
 2 files changed, 33 insertions(+), 26 deletions(-)

diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index e62d38ba55..abb94f9458 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -155,6 +155,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif /* AARCH64_TCG_TARGET_H */
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 8aa1fafd91..e398913c0c 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -78,38 +78,42 @@ static const int tcg_target_call_oarg_regs[1] = {
 #define TCG_REG_GUEST_BASE TCG_REG_X28
 #endif
 
-static inline bool reloc_pc26(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_pc26(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = target - code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t offset = target - src_rx;
+
     if (offset == sextract64(offset, 0, 26)) {
         /* read instruction, mask away previous PC_REL26 parameter contents,
            set the proper offset, then write back the instruction. */
-        *code_ptr = deposit32(*code_ptr, 0, 26, offset);
+        *src_rw = deposit32(*src_rw, 0, 26, offset);
         return true;
     }
     return false;
 }
 
-static inline bool reloc_pc19(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_pc19(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = target - code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t offset = target - src_rx;
+
     if (offset == sextract64(offset, 0, 19)) {
-        *code_ptr = deposit32(*code_ptr, 5, 19, offset);
+        *src_rw = deposit32(*src_rw, 5, 19, offset);
         return true;
     }
     return false;
 }
 
-static inline bool patch_reloc(tcg_insn_unit *code_ptr, int type,
-                               intptr_t value, intptr_t addend)
+static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
+                        intptr_t value, intptr_t addend)
 {
     tcg_debug_assert(addend == 0);
     switch (type) {
     case R_AARCH64_JUMP26:
     case R_AARCH64_CALL26:
-        return reloc_pc26(code_ptr, (tcg_insn_unit *)value);
+        return reloc_pc26(code_ptr, (const tcg_insn_unit *)value);
     case R_AARCH64_CONDBR19:
-        return reloc_pc19(code_ptr, (tcg_insn_unit *)value);
+        return reloc_pc19(code_ptr, (const tcg_insn_unit *)value);
     default:
         g_assert_not_reached();
     }
@@ -1050,12 +1054,13 @@ static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
     /* Look for host pointer values within 4G of the PC.  This happens
        often when loading pointers to QEMU's own data structures.  */
     if (type == TCG_TYPE_I64) {
-        tcg_target_long disp = value - (intptr_t)s->code_ptr;
+        intptr_t src_rx = (intptr_t)tcg_splitwx_to_rx(s->code_ptr);
+        tcg_target_long disp = value - src_rx;
         if (disp == sextract64(disp, 0, 21)) {
             tcg_out_insn(s, 3406, ADR, rd, disp);
             return;
         }
-        disp = (value >> 12) - ((intptr_t)s->code_ptr >> 12);
+        disp = (value >> 12) - (src_rx >> 12);
         if (disp == sextract64(disp, 0, 21)) {
             tcg_out_insn(s, 3406, ADRP, rd, disp);
             if (value & 0xfff) {
@@ -1308,14 +1313,14 @@ static void tcg_out_cmp(TCGContext *s, TCGType ext, TCGReg a,
 
 static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = target - s->code_ptr;
+    ptrdiff_t offset = tcg_pcrel_diff(s, target) >> 2;
     tcg_debug_assert(offset == sextract64(offset, 0, 26));
     tcg_out_insn(s, 3206, B, offset);
 }
 
-static inline void tcg_out_goto_long(TCGContext *s, tcg_insn_unit *target)
+static void tcg_out_goto_long(TCGContext *s, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = target - s->code_ptr;
+    ptrdiff_t offset = tcg_pcrel_diff(s, target) >> 2;
     if (offset == sextract64(offset, 0, 26)) {
         tcg_out_insn(s, 3206, B, offset);
     } else {
@@ -1329,9 +1334,9 @@ static inline void tcg_out_callr(TCGContext *s, TCGReg reg)
     tcg_out_insn(s, 3207, BLR, reg);
 }
 
-static inline void tcg_out_call(TCGContext *s, const tcg_insn_unit *target)
+static void tcg_out_call(TCGContext *s, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = target - s->code_ptr;
+    ptrdiff_t offset = tcg_pcrel_diff(s, target) >> 2;
     if (offset == sextract64(offset, 0, 26)) {
         tcg_out_insn(s, 3206, BL, offset);
     } else {
@@ -1393,7 +1398,7 @@ static void tcg_out_brcond(TCGContext *s, TCGType ext, TCGCond c, TCGArg a,
         tcg_out_reloc(s, s->code_ptr, R_AARCH64_CONDBR19, l, 0);
         offset = tcg_in32(s) >> 5;
     } else {
-        offset = l->u.value_ptr - s->code_ptr;
+        offset = tcg_pcrel_diff(s, l->u.value_ptr) >> 2;
         tcg_debug_assert(offset == sextract64(offset, 0, 19));
     }
 
@@ -1568,7 +1573,7 @@ static void * const qemu_st_helpers[16] = {
     [MO_BEQ]  = helper_be_stq_mmu,
 };
 
-static inline void tcg_out_adr(TCGContext *s, TCGReg rd, void *target)
+static inline void tcg_out_adr(TCGContext *s, TCGReg rd, const void *target)
 {
     ptrdiff_t offset = tcg_pcrel_diff(s, target);
     tcg_debug_assert(offset == sextract64(offset, 0, 21));
@@ -1581,7 +1586,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
     MemOp size = opc & MO_SIZE;
 
-    if (!reloc_pc19(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc19(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1606,7 +1611,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
     MemOp size = opc & MO_SIZE;
 
-    if (!reloc_pc19(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc19(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1631,7 +1636,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->type = ext;
     label->datalo_reg = data_reg;
     label->addrlo_reg = addr_reg;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
@@ -1849,7 +1855,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data_reg, TCGReg addr_reg,
 #endif /* CONFIG_SOFTMMU */
 }
 
-static tcg_insn_unit *tb_ret_addr;
+static const tcg_insn_unit *tb_ret_addr;
 
 static void tcg_out_op(TCGContext *s, TCGOpcode opc,
                        const TCGArg args[TCG_MAX_OP_ARGS],
@@ -2894,11 +2900,12 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_X0, 0);
 
     /* TB epilogue */
-    tb_ret_addr = s->code_ptr;
+    tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr);
 
     /* Remove TCG locals stack space.  */
     tcg_out_insn(s, 3401, ADDI, TCG_TYPE_I64, TCG_REG_SP, TCG_REG_SP,
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 22/41] disas: Push const down through host disasassembly
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (20 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 21/41] tcg/aarch64: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 23/41] tcg/tci: Push const down through bytecode reading Richard Henderson
                   ` (21 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/disas/dis-asm.h | 4 ++--
 disas.c                 | 4 +---
 disas/capstone.c        | 2 +-
 3 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/include/disas/dis-asm.h b/include/disas/dis-asm.h
index 2164762b46..d1133a4e04 100644
--- a/include/disas/dis-asm.h
+++ b/include/disas/dis-asm.h
@@ -358,7 +358,7 @@ typedef struct disassemble_info {
     (bfd_vma addr, struct disassemble_info * info);
 
   /* These are for buffer_read_memory.  */
-  bfd_byte *buffer;
+  const bfd_byte *buffer;
   bfd_vma buffer_vma;
   int buffer_length;
 
@@ -462,7 +462,7 @@ int print_insn_rx(bfd_vma, disassemble_info *);
 
 #ifdef CONFIG_CAPSTONE
 bool cap_disas_target(disassemble_info *info, uint64_t pc, size_t size);
-bool cap_disas_host(disassemble_info *info, void *code, size_t size);
+bool cap_disas_host(disassemble_info *info, const void *code, size_t size);
 bool cap_disas_monitor(disassemble_info *info, uint64_t pc, int count);
 bool cap_disas_plugin(disassemble_info *info, uint64_t pc, size_t size);
 #else
diff --git a/disas.c b/disas.c
index de1de7be94..a61f95b580 100644
--- a/disas.c
+++ b/disas.c
@@ -299,10 +299,8 @@ char *plugin_disas(CPUState *cpu, uint64_t addr, size_t size)
 }
 
 /* Disassemble this for me please... (debugging). */
-void disas(FILE *out, const void *ccode, unsigned long size)
+void disas(FILE *out, const void *code, unsigned long size)
 {
-    /* TODO: Push constness through the disas backends. */
-    void *code = (void *)ccode;
     uintptr_t pc;
     int count;
     CPUDebug s;
diff --git a/disas/capstone.c b/disas/capstone.c
index 7462c0e305..20bc8f9669 100644
--- a/disas/capstone.c
+++ b/disas/capstone.c
@@ -229,7 +229,7 @@ bool cap_disas_target(disassemble_info *info, uint64_t pc, size_t size)
 }
 
 /* Disassemble SIZE bytes at CODE for the host.  */
-bool cap_disas_host(disassemble_info *info, void *code, size_t size)
+bool cap_disas_host(disassemble_info *info, const void *code, size_t size)
 {
     csh handle;
     const uint8_t *cbuf;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 23/41] tcg/tci: Push const down through bytecode reading
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (21 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 22/41] disas: Push const down through host disasassembly Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 24/41] tcg: Introduce tcg_tbrel_diff Richard Henderson
                   ` (20 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/tci.c | 60 +++++++++++++++++++++++++++++++------------------------
 1 file changed, 34 insertions(+), 26 deletions(-)

diff --git a/tcg/tci.c b/tcg/tci.c
index 262a2b39ce..388c1dbee8 100644
--- a/tcg/tci.c
+++ b/tcg/tci.c
@@ -163,34 +163,34 @@ static uint64_t tci_uint64(uint32_t high, uint32_t low)
 #endif
 
 /* Read constant (native size) from bytecode. */
-static tcg_target_ulong tci_read_i(uint8_t **tb_ptr)
+static tcg_target_ulong tci_read_i(const uint8_t **tb_ptr)
 {
-    tcg_target_ulong value = *(tcg_target_ulong *)(*tb_ptr);
+    tcg_target_ulong value = *(const tcg_target_ulong *)(*tb_ptr);
     *tb_ptr += sizeof(value);
     return value;
 }
 
 /* Read unsigned constant (32 bit) from bytecode. */
-static uint32_t tci_read_i32(uint8_t **tb_ptr)
+static uint32_t tci_read_i32(const uint8_t **tb_ptr)
 {
-    uint32_t value = *(uint32_t *)(*tb_ptr);
+    uint32_t value = *(const uint32_t *)(*tb_ptr);
     *tb_ptr += sizeof(value);
     return value;
 }
 
 /* Read signed constant (32 bit) from bytecode. */
-static int32_t tci_read_s32(uint8_t **tb_ptr)
+static int32_t tci_read_s32(const uint8_t **tb_ptr)
 {
-    int32_t value = *(int32_t *)(*tb_ptr);
+    int32_t value = *(const int32_t *)(*tb_ptr);
     *tb_ptr += sizeof(value);
     return value;
 }
 
 #if TCG_TARGET_REG_BITS == 64
 /* Read constant (64 bit) from bytecode. */
-static uint64_t tci_read_i64(uint8_t **tb_ptr)
+static uint64_t tci_read_i64(const uint8_t **tb_ptr)
 {
-    uint64_t value = *(uint64_t *)(*tb_ptr);
+    uint64_t value = *(const uint64_t *)(*tb_ptr);
     *tb_ptr += sizeof(value);
     return value;
 }
@@ -198,7 +198,7 @@ static uint64_t tci_read_i64(uint8_t **tb_ptr)
 
 /* Read indexed register (native size) from bytecode. */
 static tcg_target_ulong
-tci_read_r(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+tci_read_r(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
 {
     tcg_target_ulong value = tci_read_reg(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -206,7 +206,7 @@ tci_read_r(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 }
 
 /* Read indexed register (8 bit) from bytecode. */
-static uint8_t tci_read_r8(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint8_t tci_read_r8(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
 {
     uint8_t value = tci_read_reg8(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -215,7 +215,7 @@ static uint8_t tci_read_r8(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 #if TCG_TARGET_HAS_ext8s_i32 || TCG_TARGET_HAS_ext8s_i64
 /* Read indexed register (8 bit signed) from bytecode. */
-static int8_t tci_read_r8s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static int8_t tci_read_r8s(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
 {
     int8_t value = tci_read_reg8s(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -224,7 +224,8 @@ static int8_t tci_read_r8s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 #endif
 
 /* Read indexed register (16 bit) from bytecode. */
-static uint16_t tci_read_r16(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint16_t tci_read_r16(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     uint16_t value = tci_read_reg16(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -233,7 +234,8 @@ static uint16_t tci_read_r16(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 #if TCG_TARGET_HAS_ext16s_i32 || TCG_TARGET_HAS_ext16s_i64
 /* Read indexed register (16 bit signed) from bytecode. */
-static int16_t tci_read_r16s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static int16_t tci_read_r16s(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     int16_t value = tci_read_reg16s(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -242,7 +244,8 @@ static int16_t tci_read_r16s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 #endif
 
 /* Read indexed register (32 bit) from bytecode. */
-static uint32_t tci_read_r32(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint32_t tci_read_r32(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     uint32_t value = tci_read_reg32(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -251,14 +254,16 @@ static uint32_t tci_read_r32(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 #if TCG_TARGET_REG_BITS == 32
 /* Read two indexed registers (2 * 32 bit) from bytecode. */
-static uint64_t tci_read_r64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint64_t tci_read_r64(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     uint32_t low = tci_read_r32(regs, tb_ptr);
     return tci_uint64(tci_read_r32(regs, tb_ptr), low);
 }
 #elif TCG_TARGET_REG_BITS == 64
 /* Read indexed register (32 bit signed) from bytecode. */
-static int32_t tci_read_r32s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static int32_t tci_read_r32s(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     int32_t value = tci_read_reg32s(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -266,7 +271,8 @@ static int32_t tci_read_r32s(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 }
 
 /* Read indexed register (64 bit) from bytecode. */
-static uint64_t tci_read_r64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint64_t tci_read_r64(const tcg_target_ulong *regs,
+                             const uint8_t **tb_ptr)
 {
     uint64_t value = tci_read_reg64(regs, **tb_ptr);
     *tb_ptr += 1;
@@ -276,7 +282,7 @@ static uint64_t tci_read_r64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 /* Read indexed register(s) with target address from bytecode. */
 static target_ulong
-tci_read_ulong(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+tci_read_ulong(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
 {
     target_ulong taddr = tci_read_r(regs, tb_ptr);
 #if TARGET_LONG_BITS > TCG_TARGET_REG_BITS
@@ -287,7 +293,7 @@ tci_read_ulong(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 /* Read indexed register or constant (native size) from bytecode. */
 static tcg_target_ulong
-tci_read_ri(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+tci_read_ri(const tcg_target_ulong *regs, const uint8_t **tb_ptr)
 {
     tcg_target_ulong value;
     TCGReg r = **tb_ptr;
@@ -301,7 +307,8 @@ tci_read_ri(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 }
 
 /* Read indexed register or constant (32 bit) from bytecode. */
-static uint32_t tci_read_ri32(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint32_t tci_read_ri32(const tcg_target_ulong *regs,
+                              const uint8_t **tb_ptr)
 {
     uint32_t value;
     TCGReg r = **tb_ptr;
@@ -316,14 +323,16 @@ static uint32_t tci_read_ri32(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 
 #if TCG_TARGET_REG_BITS == 32
 /* Read two indexed registers or constants (2 * 32 bit) from bytecode. */
-static uint64_t tci_read_ri64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint64_t tci_read_ri64(const tcg_target_ulong *regs,
+                              const uint8_t **tb_ptr)
 {
     uint32_t low = tci_read_ri32(regs, tb_ptr);
     return tci_uint64(tci_read_ri32(regs, tb_ptr), low);
 }
 #elif TCG_TARGET_REG_BITS == 64
 /* Read indexed register or constant (64 bit) from bytecode. */
-static uint64_t tci_read_ri64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
+static uint64_t tci_read_ri64(const tcg_target_ulong *regs,
+                              const uint8_t **tb_ptr)
 {
     uint64_t value;
     TCGReg r = **tb_ptr;
@@ -337,7 +346,7 @@ static uint64_t tci_read_ri64(const tcg_target_ulong *regs, uint8_t **tb_ptr)
 }
 #endif
 
-static tcg_target_ulong tci_read_label(uint8_t **tb_ptr)
+static tcg_target_ulong tci_read_label(const uint8_t **tb_ptr)
 {
     tcg_target_ulong label = tci_read_i(tb_ptr);
     tci_assert(label != 0);
@@ -477,8 +486,7 @@ static bool tci_compare64(uint64_t u0, uint64_t u1, TCGCond condition)
 /* Interpret pseudo code in tb. */
 uintptr_t tcg_qemu_tb_exec(CPUArchState *env, const void *v_tb_ptr)
 {
-    /* TODO: Propagate const through this file. */
-    uint8_t *tb_ptr = (uint8_t *)v_tb_ptr;
+    const uint8_t *tb_ptr = v_tb_ptr;
     tcg_target_ulong regs[TCG_TARGET_NB_REGS];
     long tcg_temps[CPU_TEMP_BUF_NLONGS];
     uintptr_t sp_value = (uintptr_t)(tcg_temps + CPU_TEMP_BUF_NLONGS);
@@ -492,7 +500,7 @@ uintptr_t tcg_qemu_tb_exec(CPUArchState *env, const void *v_tb_ptr)
         TCGOpcode opc = tb_ptr[0];
 #if defined(CONFIG_DEBUG_TCG) && !defined(NDEBUG)
         uint8_t op_size = tb_ptr[1];
-        uint8_t *old_code_ptr = tb_ptr;
+        const uint8_t *old_code_ptr = tb_ptr;
 #endif
         tcg_target_ulong t0;
         tcg_target_ulong t1;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 24/41] tcg: Introduce tcg_tbrel_diff
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (22 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 23/41] tcg/tci: Push const down through bytecode reading Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 25/41] tcg/ppc: Use tcg_tbrel_diff Richard Henderson
                   ` (19 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 67d57695c2..90ec7c1445 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -1141,6 +1141,19 @@ static inline ptrdiff_t tcg_pcrel_diff(TCGContext *s, const void *target)
     return tcg_ptr_byte_diff(target, tcg_splitwx_to_rx(s->code_ptr));
 }
 
+/**
+ * tcg_tbrel_diff
+ * @s: the tcg context
+ * @target: address of the target
+ *
+ * Produce a difference, from the beginning of the current TB code
+ * to the destination address.
+ */
+static inline ptrdiff_t tcg_tbrel_diff(TCGContext *s, const void *target)
+{
+    return tcg_ptr_byte_diff(target, tcg_splitwx_to_rx(s->code_buf));
+}
+
 /**
  * tcg_current_code_size
  * @s: the tcg context
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 25/41] tcg/ppc: Use tcg_tbrel_diff
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (23 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 24/41] tcg: Introduce tcg_tbrel_diff Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 26/41] tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB Richard Henderson
                   ` (18 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Use tcg_tbrel_diff when we need a displacement to a label,
and with a NULL argument when we need the normalizing addend.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/ppc/tcg-target.c.inc | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index bc0057eedf..dc90705d02 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -837,7 +837,7 @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret,
     }
 
     /* Load addresses within the TB with one insn.  */
-    tb_diff = arg - (intptr_t)s->code_gen_ptr;
+    tb_diff = tcg_tbrel_diff(s, (void *)arg);
     if (!in_prologue && USE_REG_TB && tb_diff == (int16_t)tb_diff) {
         tcg_out32(s, ADDI | TAI(ret, TCG_REG_TB, tb_diff));
         return;
@@ -890,7 +890,7 @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret,
     /* Use the constant pool, if possible.  */
     if (!in_prologue && USE_REG_TB) {
         new_pool_label(s, arg, R_PPC_ADDR16, s->code_ptr,
-                       -(intptr_t)s->code_gen_ptr);
+                       tcg_tbrel_diff(s, NULL));
         tcg_out32(s, LD | TAI(ret, TCG_REG_TB, 0));
         return;
     }
@@ -940,7 +940,7 @@ static void tcg_out_dupi_vec(TCGContext *s, TCGType type, TCGReg ret,
      */
     if (USE_REG_TB) {
         rel = R_PPC_ADDR16;
-        add = -(intptr_t)s->code_gen_ptr;
+        add = tcg_tbrel_diff(s, NULL);
     } else {
         rel = R_PPC_ADDR32;
         add = 0;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 26/41] tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (24 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 25/41] tcg/ppc: Use tcg_tbrel_diff Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 27/41] tcg/ppc: Support split-wx code generation Richard Henderson
                   ` (17 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

The maximum TB code gen size is UINT16_MAX, which the current
code does not support.  Use our utility function to optimally
add an arbitrary constant.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/ppc/tcg-target.c.inc | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index dc90705d02..91d5d95ddf 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -2392,9 +2392,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args,
         set_jmp_reset_offset(s, args[0]);
         if (USE_REG_TB) {
             /* For the unlinked case, need to reset TCG_REG_TB.  */
-            c = -tcg_current_code_size(s);
-            assert(c == (int16_t)c);
-            tcg_out32(s, ADDI | TAI(TCG_REG_TB, TCG_REG_TB, c));
+            tcg_out_mem_long(s, ADDI, ADD, TCG_REG_TB, TCG_REG_TB,
+                             -tcg_current_code_size(s));
         }
         break;
     case INDEX_op_goto_ptr:
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 27/41] tcg/ppc: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (25 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 26/41] tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 28/41] tcg/sparc: Use tcg_tbrel_diff Richard Henderson
                   ` (16 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/ppc/tcg-target.h     |  2 +-
 tcg/ppc/tcg-target.c.inc | 53 +++++++++++++++++++++++-----------------
 2 files changed, 31 insertions(+), 24 deletions(-)

diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index 78d6a5e96f..a8628b6cad 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -185,6 +185,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 91d5d95ddf..fe33687787 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -62,8 +62,6 @@
 #define TCG_CT_CONST_MONE 0x2000
 #define TCG_CT_CONST_WSZ  0x4000
 
-static tcg_insn_unit *tb_ret_addr;
-
 TCGPowerISA have_isa;
 static bool have_isel;
 bool have_altivec;
@@ -184,35 +182,41 @@ static inline bool in_range_b(tcg_target_long target)
     return target == sextract64(target, 0, 26);
 }
 
-static uint32_t reloc_pc24_val(tcg_insn_unit *pc, const tcg_insn_unit *target)
+static uint32_t reloc_pc24_val(const tcg_insn_unit *pc,
+			       const tcg_insn_unit *target)
 {
     ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
     tcg_debug_assert(in_range_b(disp));
     return disp & 0x3fffffc;
 }
 
-static bool reloc_pc24(tcg_insn_unit *pc, tcg_insn_unit *target)
+static bool reloc_pc24(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t disp = tcg_ptr_byte_diff(target, src_rx);
+
     if (in_range_b(disp)) {
-        *pc = (*pc & ~0x3fffffc) | (disp & 0x3fffffc);
+        *src_rw = (*src_rw & ~0x3fffffc) | (disp & 0x3fffffc);
         return true;
     }
     return false;
 }
 
-static uint16_t reloc_pc14_val(tcg_insn_unit *pc, const tcg_insn_unit *target)
+static uint16_t reloc_pc14_val(const tcg_insn_unit *pc,
+			       const tcg_insn_unit *target)
 {
     ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
     tcg_debug_assert(disp == (int16_t) disp);
     return disp & 0xfffc;
 }
 
-static bool reloc_pc14(tcg_insn_unit *pc, tcg_insn_unit *target)
+static bool reloc_pc14(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t disp = tcg_ptr_byte_diff(target, pc);
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t disp = tcg_ptr_byte_diff(target, src_rx);
+
     if (disp == (int16_t) disp) {
-        *pc = (*pc & ~0xfffc) | (disp & 0xfffc);
+        *src_rw = (*src_rw & ~0xfffc) | (disp & 0xfffc);
         return true;
     }
     return false;
@@ -673,12 +677,12 @@ static const uint32_t tcg_to_isel[] = {
 static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
                         intptr_t value, intptr_t addend)
 {
-    tcg_insn_unit *target;
+    const tcg_insn_unit *target;
     int16_t lo;
     int32_t hi;
 
     value += addend;
-    target = (tcg_insn_unit *)value;
+    target = (const tcg_insn_unit *)value;
 
     switch (type) {
     case R_PPC_REL14:
@@ -1544,7 +1548,7 @@ static void tcg_out_setcond(TCGContext *s, TCGType type, TCGCond cond,
 static void tcg_out_bc(TCGContext *s, int bc, TCGLabel *l)
 {
     if (l->has_value) {
-        bc |= reloc_pc14_val(s->code_ptr, l->u.value_ptr);
+        bc |= reloc_pc14_val(tcg_splitwx_to_rx(s->code_ptr), l->u.value_ptr);
     } else {
         tcg_out_reloc(s, s->code_ptr, R_PPC_REL14, l, 0);
     }
@@ -1997,7 +2001,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi_reg;
     label->addrlo_reg = addrlo_reg;
     label->addrhi_reg = addrhi_reg;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = lptr;
 }
 
@@ -2007,7 +2012,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
     TCGReg hi, lo, arg = TCG_REG_R3;
 
-    if (!reloc_pc14(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc14(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -2055,7 +2060,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp s_bits = opc & MO_SIZE;
     TCGReg hi, lo, arg = TCG_REG_R3;
 
-    if (!reloc_pc14(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc14(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -2306,10 +2311,10 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     int i;
 
 #ifdef _CALL_AIX
-    void **desc = (void **)s->code_ptr;
-    desc[0] = desc + 2;                   /* entry point */
-    desc[1] = 0;                          /* environment pointer */
-    s->code_ptr = (void *)(desc + 2);     /* skip over descriptor */
+    const void **desc = (const void **)s->code_ptr;
+    desc[0] = tcg_splitwx_to_rx(desc + 2);  /* entry point */
+    desc[1] = 0;                            /* environment pointer */
+    s->code_ptr = (void *)(desc + 2);       /* skip over descriptor */
 #endif
 
     tcg_set_frame(s, TCG_REG_CALL_STACK, REG_SAVE_BOT - CPU_TEMP_BUF_SIZE,
@@ -2341,7 +2346,8 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out32(s, BCCTR | BO_ALWAYS);
 
     /* Epilogue */
-    tcg_code_gen_epilogue = tb_ret_addr = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
 
     tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_R0, TCG_REG_R1, FRAME_SIZE+LR_OFFSET);
     for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); ++i) {
@@ -2362,7 +2368,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args,
     switch (opc) {
     case INDEX_op_exit_tb:
         tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R3, args[0]);
-        tcg_out_b(s, 0, tb_ret_addr);
+        tcg_out_b(s, 0, tcg_code_gen_epilogue);
         break;
     case INDEX_op_goto_tb:
         if (s->tb_jmp_insn_offset) {
@@ -2410,7 +2416,8 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc, const TCGArg *args,
             uint32_t insn = B;
 
             if (l->has_value) {
-                insn |= reloc_pc24_val(s->code_ptr, l->u.value_ptr);
+                insn |= reloc_pc24_val(tcg_splitwx_to_rx(s->code_ptr),
+                                       l->u.value_ptr);
             } else {
                 tcg_out_reloc(s, s->code_ptr, R_PPC_REL24, l, 0);
             }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 28/41] tcg/sparc: Use tcg_tbrel_diff
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (26 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 27/41] tcg/ppc: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 29/41] tcg/sparc: Support split-wx code generation Richard Henderson
                   ` (15 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/sparc/tcg-target.c.inc | 15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)

diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index d599ae27b5..8f04fdf981 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -440,7 +440,7 @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret,
 
     /* A 13-bit constant relative to the TB.  */
     if (!in_prologue && USE_REG_TB) {
-        test = arg - (uintptr_t)s->code_gen_ptr;
+        test = tcg_tbrel_diff(s, (void *)arg);
         if (check_fit_ptr(test, 13)) {
             tcg_out_arithi(s, ret, TCG_REG_TB, test, ARITH_ADD);
             return;
@@ -537,15 +537,15 @@ static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val,
     return false;
 }
 
-static void tcg_out_ld_ptr(TCGContext *s, TCGReg ret, uintptr_t arg)
+static void tcg_out_ld_ptr(TCGContext *s, TCGReg ret, const void *arg)
 {
-    intptr_t diff = arg - (uintptr_t)s->code_gen_ptr;
+    intptr_t diff = tcg_tbrel_diff(s, arg);
     if (USE_REG_TB && check_fit_ptr(diff, 13)) {
         tcg_out_ld(s, TCG_TYPE_PTR, ret, TCG_REG_TB, diff);
         return;
     }
-    tcg_out_movi(s, TCG_TYPE_PTR, ret, arg & ~0x3ff);
-    tcg_out_ld(s, TCG_TYPE_PTR, ret, ret, arg & 0x3ff);
+    tcg_out_movi(s, TCG_TYPE_PTR, ret, (uintptr_t)arg & ~0x3ff);
+    tcg_out_ld(s, TCG_TYPE_PTR, ret, ret, (uintptr_t)arg & 0x3ff);
 }
 
 static inline void tcg_out_sety(TCGContext *s, TCGReg rs)
@@ -1313,7 +1313,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
             tcg_out_movi_imm13(s, TCG_REG_O0, a0);
             break;
         } else if (USE_REG_TB) {
-            intptr_t tb_diff = a0 - (uintptr_t)s->code_gen_ptr;
+            intptr_t tb_diff = tcg_tbrel_diff(s, (void *)a0);
             if (check_fit_ptr(tb_diff, 13)) {
                 tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN);
                 /* Note that TCG_REG_TB has been unwound to O1.  */
@@ -1345,8 +1345,7 @@ static void tcg_out_op(TCGContext *s, TCGOpcode opc,
             }
         } else {
             /* indirect jump method */
-            tcg_out_ld_ptr(s, TCG_REG_TB,
-                           (uintptr_t)(s->tb_jmp_target_addr + a0));
+            tcg_out_ld_ptr(s, TCG_REG_TB, s->tb_jmp_target_addr + a0);
             tcg_out_arithi(s, TCG_REG_G0, TCG_REG_TB, 0, JMPL);
             tcg_out_nop(s);
         }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 29/41] tcg/sparc: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (27 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 28/41] tcg/sparc: Use tcg_tbrel_diff Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 30/41] tcg/s390: Use tcg_tbrel_diff Richard Henderson
                   ` (14 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/sparc/tcg-target.h     |  2 +-
 tcg/sparc/tcg-target.c.inc | 24 +++++++++++++-----------
 2 files changed, 14 insertions(+), 12 deletions(-)

diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index 517840705f..bb2505bfc7 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -181,6 +181,6 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 8f04fdf981..182124b96c 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -291,14 +291,15 @@ static inline int check_fit_i32(int32_t val, unsigned int bits)
 # define check_fit_ptr  check_fit_i32
 #endif
 
-static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
+static bool patch_reloc(tcg_insn_unit *src_rw, int type,
                         intptr_t value, intptr_t addend)
 {
-    uint32_t insn = *code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    uint32_t insn = *src_rw;
     intptr_t pcrel;
 
     value += addend;
-    pcrel = tcg_ptr_byte_diff((tcg_insn_unit *)value, code_ptr);
+    pcrel = tcg_ptr_byte_diff((tcg_insn_unit *)value, src_rx);
 
     switch (type) {
     case R_SPARC_WDISP16:
@@ -315,7 +316,7 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
         g_assert_not_reached();
     }
 
-    *code_ptr = insn;
+    *src_rw = insn;
     return true;
 }
 
@@ -868,8 +869,8 @@ static void tcg_out_mb(TCGContext *s, TCGArg a0)
 }
 
 #ifdef CONFIG_SOFTMMU
-static tcg_insn_unit *qemu_ld_trampoline[16];
-static tcg_insn_unit *qemu_st_trampoline[16];
+static const tcg_insn_unit *qemu_ld_trampoline[16];
+static const tcg_insn_unit *qemu_st_trampoline[16];
 
 static void emit_extend(TCGContext *s, TCGReg r, int op)
 {
@@ -930,7 +931,7 @@ static void build_trampolines(TCGContext *s)
         while ((uintptr_t)s->code_ptr & 15) {
             tcg_out_nop(s);
         }
-        qemu_ld_trampoline[i] = s->code_ptr;
+        qemu_ld_trampoline[i] = tcg_splitwx_to_rx(s->code_ptr);
 
         if (SPARC64 || TARGET_LONG_BITS == 32) {
             ra = TCG_REG_O3;
@@ -958,7 +959,7 @@ static void build_trampolines(TCGContext *s)
         while ((uintptr_t)s->code_ptr & 15) {
             tcg_out_nop(s);
         }
-        qemu_st_trampoline[i] = s->code_ptr;
+        qemu_st_trampoline[i] = tcg_splitwx_to_rx(s->code_ptr);
 
         if (SPARC64) {
             emit_extend(s, TCG_REG_O2, i);
@@ -1038,7 +1039,8 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_nop(s);
 
     /* Epilogue for goto_ptr.  */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN);
     /* delay slot */
     tcg_out_movi_imm13(s, TCG_REG_O0, 0);
@@ -1163,7 +1165,7 @@ static void tcg_out_qemu_ld(TCGContext *s, TCGReg data, TCGReg addr,
 #ifdef CONFIG_SOFTMMU
     unsigned memi = get_mmuidx(oi);
     TCGReg addrz, param;
-    tcg_insn_unit *func;
+    const tcg_insn_unit *func;
     tcg_insn_unit *label_ptr;
 
     addrz = tcg_out_tlb_load(s, addr, memi, memop,
@@ -1245,7 +1247,7 @@ static void tcg_out_qemu_st(TCGContext *s, TCGReg data, TCGReg addr,
 #ifdef CONFIG_SOFTMMU
     unsigned memi = get_mmuidx(oi);
     TCGReg addrz, param;
-    tcg_insn_unit *func;
+    const tcg_insn_unit *func;
     tcg_insn_unit *label_ptr;
 
     addrz = tcg_out_tlb_load(s, addr, memi, memop,
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 30/41] tcg/s390: Use tcg_tbrel_diff
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (28 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 29/41] tcg/sparc: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 31/41] tcg/s390: Support split-wx code generation Richard Henderson
                   ` (13 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Use tcg_tbrel_diff when we need a displacement to a label,
and with a NULL argument when we need the normalizing addend.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/s390/tcg-target.c.inc | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index 1444914428..e4c61fc014 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -630,7 +630,7 @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret,
             return;
         }
     } else if (USE_REG_TB && !in_prologue) {
-        ptrdiff_t off = sval - (uintptr_t)s->code_gen_ptr;
+        ptrdiff_t off = tcg_tbrel_diff(s, (void *)sval);
         if (off == sextract64(off, 0, 20)) {
             /* This is certain to be an address within TB, and therefore
                OFF will be negative; don't try RX_LA.  */
@@ -655,7 +655,7 @@ static void tcg_out_movi_int(TCGContext *s, TCGType type, TCGReg ret,
     } else if (USE_REG_TB && !in_prologue) {
         tcg_out_insn(s, RXY, LG, ret, TCG_REG_TB, TCG_REG_NONE, 0);
         new_pool_label(s, sval, R_390_20, s->code_ptr - 2,
-                       -(intptr_t)s->code_gen_ptr);
+                       tcg_tbrel_diff(s, NULL));
     } else {
         TCGReg base = ret ? ret : TCG_TMP0;
         tcg_out_insn(s, RIL, LARL, base, 0);
@@ -746,7 +746,7 @@ static void tcg_out_ld_abs(TCGContext *s, TCGType type, TCGReg dest, void *abs)
         }
     }
     if (USE_REG_TB) {
-        ptrdiff_t disp = abs - (void *)s->code_gen_ptr;
+        ptrdiff_t disp = tcg_tbrel_diff(s, abs);
         if (disp == sextract64(disp, 0, 20)) {
             tcg_out_ld(s, type, dest, TCG_REG_TB, disp);
             return;
@@ -956,7 +956,7 @@ static void tgen_andi(TCGContext *s, TCGType type, TCGReg dest, uint64_t val)
         if (!maybe_out_small_movi(s, type, TCG_TMP0, val)) {
             tcg_out_insn(s, RXY, NG, dest, TCG_REG_TB, TCG_REG_NONE, 0);
             new_pool_label(s, val & valid, R_390_20, s->code_ptr - 2,
-                           -(intptr_t)s->code_gen_ptr);
+                           tcg_tbrel_diff(s, NULL));
             return;
         }
     } else {
@@ -1015,7 +1015,7 @@ static void tgen_ori(TCGContext *s, TCGType type, TCGReg dest, uint64_t val)
     } else if (USE_REG_TB) {
         tcg_out_insn(s, RXY, OG, dest, TCG_REG_TB, TCG_REG_NONE, 0);
         new_pool_label(s, val, R_390_20, s->code_ptr - 2,
-                       -(intptr_t)s->code_gen_ptr);
+                       tcg_tbrel_diff(s, NULL));
     } else {
         /* Perform the OR via sequential modifications to the high and
            low parts.  Do this via recursion to handle 16-bit vs 32-bit
@@ -1050,7 +1050,7 @@ static void tgen_xori(TCGContext *s, TCGType type, TCGReg dest, uint64_t val)
     } else if (USE_REG_TB) {
         tcg_out_insn(s, RXY, XG, dest, TCG_REG_TB, TCG_REG_NONE, 0);
         new_pool_label(s, val, R_390_20, s->code_ptr - 2,
-                       -(intptr_t)s->code_gen_ptr);
+                       tcg_tbrel_diff(s, NULL));
     } else {
         /* Perform the xor by parts.  */
         tcg_debug_assert(s390_facilities & FACILITY_EXT_IMM);
@@ -1108,12 +1108,12 @@ static int tgen_cmp(TCGContext *s, TCGType type, TCGCond c, TCGReg r1,
                 op = (is_unsigned ? RXY_CLY : RXY_CY);
                 tcg_out_insn_RXY(s, op, r1, TCG_REG_TB, TCG_REG_NONE, 0);
                 new_pool_label(s, (uint32_t)c2, R_390_20, s->code_ptr - 2,
-                               4 - (intptr_t)s->code_gen_ptr);
+                               4 - tcg_tbrel_diff(s, NULL));
             } else {
                 op = (is_unsigned ? RXY_CLG : RXY_CG);
                 tcg_out_insn_RXY(s, op, r1, TCG_REG_TB, TCG_REG_NONE, 0);
                 new_pool_label(s, c2, R_390_20, s->code_ptr - 2,
-                               -(intptr_t)s->code_gen_ptr);
+                               tcg_tbrel_diff(s, NULL));
             }
             goto exit;
         } else {
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 31/41] tcg/s390: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (29 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 30/41] tcg/s390: Use tcg_tbrel_diff Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 32/41] tcg/riscv: Fix branch range checks Richard Henderson
                   ` (12 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/s390/tcg-target.h     |  2 +-
 tcg/s390/tcg-target.c.inc | 69 +++++++++++++++++----------------------
 2 files changed, 31 insertions(+), 40 deletions(-)

diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h
index 8324197127..1fd8b3858e 100644
--- a/tcg/s390/tcg-target.h
+++ b/tcg/s390/tcg-target.h
@@ -163,6 +163,6 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index e4c61fc014..582a8ef941 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -363,36 +363,37 @@ static void * const qemu_st_helpers[16] = {
 };
 #endif
 
-static tcg_insn_unit *tb_ret_addr;
+static const tcg_insn_unit *tb_ret_addr;
 uint64_t s390_facilities;
 
-static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
+static bool patch_reloc(tcg_insn_unit *src_rw, int type,
                         intptr_t value, intptr_t addend)
 {
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
     intptr_t pcrel2;
     uint32_t old;
 
     value += addend;
-    pcrel2 = (tcg_insn_unit *)value - code_ptr;
+    pcrel2 = (tcg_insn_unit *)value - src_rx;
 
     switch (type) {
     case R_390_PC16DBL:
         if (pcrel2 == (int16_t)pcrel2) {
-            tcg_patch16(code_ptr, pcrel2);
+            tcg_patch16(src_rw, pcrel2);
             return true;
         }
         break;
     case R_390_PC32DBL:
         if (pcrel2 == (int32_t)pcrel2) {
-            tcg_patch32(code_ptr, pcrel2);
+            tcg_patch32(src_rw, pcrel2);
             return true;
         }
         break;
     case R_390_20:
         if (value == sextract64(value, 0, 20)) {
-            old = *(uint32_t *)code_ptr & 0xf00000ff;
+            old = *(uint32_t *)src_rw & 0xf00000ff;
             old |= ((value & 0xfff) << 16) | ((value & 0xff000) >> 4);
-            tcg_patch32(code_ptr, old);
+            tcg_patch32(src_rw, old);
             return true;
         }
         break;
@@ -730,7 +731,8 @@ static inline bool tcg_out_sti(TCGContext *s, TCGType type, TCGArg val,
 }
 
 /* load data from an absolute host address */
-static void tcg_out_ld_abs(TCGContext *s, TCGType type, TCGReg dest, void *abs)
+static void tcg_out_ld_abs(TCGContext *s, TCGType type,
+                           TCGReg dest, const void *abs)
 {
     intptr_t addr = (intptr_t)abs;
 
@@ -1304,7 +1306,7 @@ static void tgen_extract(TCGContext *s, TCGReg dest, TCGReg src,
 
 static void tgen_gotoi(TCGContext *s, int cc, const tcg_insn_unit *dest)
 {
-    ptrdiff_t off = dest - s->code_ptr;
+    ptrdiff_t off = tcg_pcrel_diff(s, dest) >> 1;
     if (off == (int16_t)off) {
         tcg_out_insn(s, RI, BRC, cc, off);
     } else if (off == (int32_t)off) {
@@ -1333,34 +1335,18 @@ static void tgen_branch(TCGContext *s, int cc, TCGLabel *l)
 static void tgen_compare_branch(TCGContext *s, S390Opcode opc, int cc,
                                 TCGReg r1, TCGReg r2, TCGLabel *l)
 {
-    intptr_t off = 0;
-
-    if (l->has_value) {
-        off = l->u.value_ptr - s->code_ptr;
-        tcg_debug_assert(off == (int16_t)off);
-    } else {
-        tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2);
-    }
-
+    tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2);
     tcg_out16(s, (opc & 0xff00) | (r1 << 4) | r2);
-    tcg_out16(s, off);
+    tcg_out16(s, 0);
     tcg_out16(s, cc << 12 | (opc & 0xff));
 }
 
 static void tgen_compare_imm_branch(TCGContext *s, S390Opcode opc, int cc,
                                     TCGReg r1, int i2, TCGLabel *l)
 {
-    tcg_target_long off = 0;
-
-    if (l->has_value) {
-        off = l->u.value_ptr - s->code_ptr;
-        tcg_debug_assert(off == (int16_t)off);
-    } else {
-        tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2);
-    }
-
+    tcg_out_reloc(s, s->code_ptr + 1, R_390_PC16DBL, l, 2);
     tcg_out16(s, (opc & 0xff00) | (r1 << 4) | cc);
-    tcg_out16(s, off);
+    tcg_out16(s, 0);
     tcg_out16(s, (i2 << 8) | (opc & 0xff));
 }
 
@@ -1417,7 +1403,7 @@ static void tgen_brcond(TCGContext *s, TCGType type, TCGCond c,
 
 static void tcg_out_call(TCGContext *s, const tcg_insn_unit *dest)
 {
-    ptrdiff_t off = dest - s->code_ptr;
+    ptrdiff_t off = tcg_pcrel_diff(s, dest) >> 1;
     if (off == (int32_t)off) {
         tcg_out_insn(s, RIL, BRASL, TCG_REG_R14, off);
     } else {
@@ -1601,7 +1587,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->oi = oi;
     label->datalo_reg = data;
     label->addrlo_reg = addr;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
@@ -1613,7 +1600,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
 
     if (!patch_reloc(lb->label_ptr[0], R_390_PC16DBL,
-                     (intptr_t)s->code_ptr, 2)) {
+                     (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 2)) {
         return false;
     }
 
@@ -1638,7 +1625,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
 
     if (!patch_reloc(lb->label_ptr[0], R_390_PC16DBL,
-                     (intptr_t)s->code_ptr, 2)) {
+                     (intptr_t)tcg_splitwx_to_rx(s->code_ptr), 2)) {
         return false;
     }
 
@@ -1766,7 +1753,8 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
     case INDEX_op_goto_tb:
         a0 = args[0];
         if (s->tb_jmp_insn_offset) {
-            /* branch displacement must be aligned for atomic patching;
+            /*
+             * branch displacement must be aligned for atomic patching;
              * see if we need to add extra nop before branch
              */
             if (!QEMU_PTR_IS_ALIGNED(s->code_ptr + 1, 4)) {
@@ -1779,7 +1767,7 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
         } else {
             /* load address stored at s->tb_jmp_target_addr + a0 */
             tcg_out_ld_abs(s, TCG_TYPE_PTR, TCG_REG_TB,
-                           s->tb_jmp_target_addr + a0);
+                           tcg_splitwx_to_rx(s->tb_jmp_target_addr + a0));
             /* and go there */
             tcg_out_insn(s, RR, BCR, S390_CC_ALWAYS, TCG_REG_TB);
         }
@@ -1789,8 +1777,10 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
            TCG_REG_TB to the beginning of this TB.  */
         if (USE_REG_TB) {
             int ofs = -tcg_current_code_size(s);
-            assert(ofs == (int16_t)ofs);
-            tcg_out_insn(s, RI, AGHI, TCG_REG_TB, ofs);
+            /* All TB are restricted to 64KiB by unwind info. */
+            tcg_debug_assert(ofs == sextract64(ofs, 0, 20));
+            tcg_out_insn(s, RXY, LAY, TCG_REG_TB,
+                         TCG_REG_TB, TCG_REG_NONE, ofs);
         }
         break;
 
@@ -2561,11 +2551,12 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, 0);
 
     /* TB epilogue */
-    tb_ret_addr = s->code_ptr;
+    tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr);
 
     /* lmg %r6,%r15,fs+48(%r15) (restore registers) */
     tcg_out_insn(s, RXY, LMG, TCG_REG_R6, TCG_REG_R15, TCG_REG_R15,
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 32/41] tcg/riscv: Fix branch range checks
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (30 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 31/41] tcg/s390: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 33/41] tcg/riscv: Remove branch-over-branch fallback Richard Henderson
                   ` (11 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

The offset even checks were folded into the range check incorrectly.
By offsetting by 1, and not decrementing the width, we silently
allowed out of range branches.

Assert that the offset is always even instead.  Move tcg_out_goto
down into the CONFIG_SOFTMMU block so that it is not unused.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/riscv/tcg-target.c.inc | 28 +++++++++++++++-------------
 1 file changed, 15 insertions(+), 13 deletions(-)

diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 025e3cd0bb..195c3eff03 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -429,7 +429,8 @@ static bool reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
 {
     intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
 
-    if (offset == sextreg(offset, 1, 12) << 1) {
+    tcg_debug_assert((offset & 1) == 0);
+    if (offset == sextreg(offset, 0, 12)) {
         code_ptr[0] |= encode_sbimm12(offset);
         return true;
     }
@@ -441,7 +442,8 @@ static bool reloc_jimm20(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
 {
     intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
 
-    if (offset == sextreg(offset, 1, 20) << 1) {
+    tcg_debug_assert((offset & 1) == 0);
+    if (offset == sextreg(offset, 0, 20)) {
         code_ptr[0] |= encode_ujimm20(offset);
         return true;
     }
@@ -854,28 +856,21 @@ static void tcg_out_setcond2(TCGContext *s, TCGCond cond, TCGReg ret,
     g_assert_not_reached();
 }
 
-static inline void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
-{
-    ptrdiff_t offset = tcg_pcrel_diff(s, target);
-    tcg_debug_assert(offset == sextreg(offset, 1, 20) << 1);
-    tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, offset);
-}
-
 static void tcg_out_call_int(TCGContext *s, const tcg_insn_unit *arg, bool tail)
 {
     TCGReg link = tail ? TCG_REG_ZERO : TCG_REG_RA;
     ptrdiff_t offset = tcg_pcrel_diff(s, arg);
     int ret;
 
-    if (offset == sextreg(offset, 1, 20) << 1) {
+    tcg_debug_assert((offset & 1) == 0);
+    if (offset == sextreg(offset, 0, 20)) {
         /* short jump: -2097150 to 2097152 */
         tcg_out_opc_jump(s, OPC_JAL, link, offset);
-    } else if (TCG_TARGET_REG_BITS == 32 ||
-        offset == sextreg(offset, 1, 31) << 1) {
+    } else if (TCG_TARGET_REG_BITS == 32 || offset == (int32_t)offset) {
         /* long jump: -2147483646 to 2147483648 */
         tcg_out_opc_upper(s, OPC_AUIPC, TCG_REG_TMP0, 0);
         tcg_out_opc_imm(s, OPC_JALR, link, TCG_REG_TMP0, 0);
-        ret = reloc_call(s->code_ptr - 2, arg);\
+        ret = reloc_call(s->code_ptr - 2, arg);
         tcg_debug_assert(ret == true);
     } else if (TCG_TARGET_REG_BITS == 64) {
         /* far jump: 64-bit */
@@ -962,6 +957,13 @@ QEMU_BUILD_BUG_ON(TCG_TARGET_REG_BITS < TARGET_LONG_BITS);
 QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0);
 QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 11));
 
+static void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
+{
+    tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, 0);
+    bool ok = reloc_jimm20(s->code_ptr - 1, target);
+    tcg_debug_assert(ok);
+}
+
 static void tcg_out_tlb_load(TCGContext *s, TCGReg addrl,
                              TCGReg addrh, TCGMemOpIdx oi,
                              tcg_insn_unit **label_ptr, bool is_load)
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 33/41] tcg/riscv: Remove branch-over-branch fallback
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (31 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 32/41] tcg/riscv: Fix branch range checks Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 34/41] tcg/riscv: Support split-wx code generation Richard Henderson
                   ` (10 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Since 7ecd02a06f8, we are prepared to re-start code generation
with a smaller TB if a relocation is out of range.  We no longer
need to leave a nop in the stream Just In Case.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/riscv/tcg-target.c.inc | 56 ++++----------------------------------
 1 file changed, 6 insertions(+), 50 deletions(-)

diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 195c3eff03..02beb86977 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -469,43 +469,16 @@ static bool reloc_call(tcg_insn_unit *code_ptr, const tcg_insn_unit *target)
 static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
                         intptr_t value, intptr_t addend)
 {
-    uint32_t insn = *code_ptr;
-    intptr_t diff;
-    bool short_jmp;
-
     tcg_debug_assert(addend == 0);
-
     switch (type) {
     case R_RISCV_BRANCH:
-        diff = value - (uintptr_t)code_ptr;
-        short_jmp = diff == sextreg(diff, 0, 12);
-        if (short_jmp) {
-            return reloc_sbimm12(code_ptr, (tcg_insn_unit *)value);
-        } else {
-            /* Invert the condition */
-            insn = insn ^ (1 << 12);
-            /* Clear the offset */
-            insn &= 0x01fff07f;
-            /* Set the offset to the PC + 8 */
-            insn |= encode_sbimm12(8);
-
-            /* Move forward */
-            code_ptr[0] = insn;
-
-            /* Overwrite the NOP with jal x0,value */
-            diff = value - (uintptr_t)(code_ptr + 1);
-            insn = encode_uj(OPC_JAL, TCG_REG_ZERO, diff);
-            code_ptr[1] = insn;
-
-            return true;
-        }
-        break;
+        return reloc_sbimm12(code_ptr, (tcg_insn_unit *)value);
     case R_RISCV_JAL:
         return reloc_jimm20(code_ptr, (tcg_insn_unit *)value);
     case R_RISCV_CALL:
         return reloc_call(code_ptr, (tcg_insn_unit *)value);
     default:
-        tcg_abort();
+        g_assert_not_reached();
     }
 }
 
@@ -779,21 +752,8 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
         arg2 = t;
     }
 
-    if (l->has_value) {
-        intptr_t diff = tcg_pcrel_diff(s, l->u.value_ptr);
-        if (diff == sextreg(diff, 0, 12)) {
-            tcg_out_opc_branch(s, op, arg1, arg2, diff);
-        } else {
-            /* Invert the conditional branch.  */
-            tcg_out_opc_branch(s, op ^ (1 << 12), arg1, arg2, 8);
-            tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, diff - 4);
-        }
-    } else {
-        tcg_out_reloc(s, s->code_ptr, R_RISCV_BRANCH, l, 0);
-        tcg_out_opc_branch(s, op, arg1, arg2, 0);
-        /* NOP to allow patching later */
-        tcg_out_opc_imm(s, OPC_ADDI, TCG_REG_ZERO, TCG_REG_ZERO, 0);
-    }
+    tcg_out_reloc(s, s->code_ptr, R_RISCV_BRANCH, l, 0);
+    tcg_out_opc_branch(s, op, arg1, arg2, 0);
 }
 
 static void tcg_out_setcond(TCGContext *s, TCGCond cond, TCGReg ret,
@@ -1009,8 +969,6 @@ static void tcg_out_tlb_load(TCGContext *s, TCGReg addrl,
     /* Compare masked address with the TLB entry. */
     label_ptr[0] = s->code_ptr;
     tcg_out_opc_branch(s, OPC_BNE, TCG_REG_TMP0, TCG_REG_TMP1, 0);
-    /* NOP to allow patching later */
-    tcg_out_opc_imm(s, OPC_ADDI, TCG_REG_ZERO, TCG_REG_ZERO, 0);
 
     /* TLB Hit - translate address using addend.  */
     if (TCG_TARGET_REG_BITS > TARGET_LONG_BITS) {
@@ -1054,8 +1012,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     }
 
     /* resolve label address */
-    if (!patch_reloc(l->label_ptr[0], R_RISCV_BRANCH,
-                     (intptr_t) s->code_ptr, 0)) {
+    if (!reloc_sbimm12(l->label_ptr[0], s->code_ptr)) {
         return false;
     }
 
@@ -1089,8 +1046,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     }
 
     /* resolve label address */
-    if (!patch_reloc(l->label_ptr[0], R_RISCV_BRANCH,
-                     (intptr_t) s->code_ptr, 0)) {
+    if (!reloc_sbimm12(l->label_ptr[0], s->code_ptr)) {
         return false;
     }
 
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 34/41] tcg/riscv: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (32 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 33/41] tcg/riscv: Remove branch-over-branch fallback Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 35/41] accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd Richard Henderson
                   ` (9 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/riscv/tcg-target.h     |  2 +-
 tcg/riscv/tcg-target.c.inc | 41 +++++++++++++++++++++-----------------
 2 files changed, 24 insertions(+), 19 deletions(-)

diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index 3c2e8305b0..0eb19f2b11 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -179,6 +179,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_POOL_LABELS
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP 0
-#define TCG_TARGET_SUPPORT_MIRROR   0
+#define TCG_TARGET_SUPPORT_MIRROR   1
 
 #endif
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 02beb86977..5c1e0f8fc1 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -425,41 +425,44 @@ static void tcg_out_nop_fill(tcg_insn_unit *p, int count)
  * Relocations
  */
 
-static bool reloc_sbimm12(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_sbimm12(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    intptr_t offset = (intptr_t)target - (intptr_t)src_rx;
 
     tcg_debug_assert((offset & 1) == 0);
     if (offset == sextreg(offset, 0, 12)) {
-        code_ptr[0] |= encode_sbimm12(offset);
+        *src_rw |= encode_sbimm12(offset);
         return true;
     }
 
     return false;
 }
 
-static bool reloc_jimm20(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_jimm20(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    intptr_t offset = (intptr_t)target - (intptr_t)src_rx;
 
     tcg_debug_assert((offset & 1) == 0);
     if (offset == sextreg(offset, 0, 20)) {
-        code_ptr[0] |= encode_ujimm20(offset);
+        *src_rw |= encode_ujimm20(offset);
         return true;
     }
 
     return false;
 }
 
-static bool reloc_call(tcg_insn_unit *code_ptr, const tcg_insn_unit *target)
+static bool reloc_call(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    intptr_t offset = (intptr_t)target - (intptr_t)code_ptr;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    intptr_t offset = (intptr_t)target - (intptr_t)src_rx;
     int32_t lo = sextreg(offset, 0, 12);
     int32_t hi = offset - lo;
 
     if (offset == hi + lo) {
-        code_ptr[0] |= encode_uimm20(hi);
-        code_ptr[1] |= encode_imm12(lo);
+        src_rw[0] |= encode_uimm20(hi);
+        src_rw[1] |= encode_imm12(lo);
         return true;
     }
 
@@ -532,7 +535,7 @@ static void tcg_out_movi(TCGContext *s, TCGType type, TCGReg rd,
     if (tmp == (int32_t)tmp) {
         tcg_out_opc_upper(s, OPC_AUIPC, rd, 0);
         tcg_out_opc_imm(s, OPC_ADDI, rd, rd, 0);
-        ret = reloc_call(s->code_ptr - 2, (tcg_insn_unit *)val);
+        ret = reloc_call(s->code_ptr - 2, (const tcg_insn_unit *)val);
         tcg_debug_assert(ret == true);
         return;
     }
@@ -917,7 +920,7 @@ QEMU_BUILD_BUG_ON(TCG_TARGET_REG_BITS < TARGET_LONG_BITS);
 QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) > 0);
 QEMU_BUILD_BUG_ON(TLB_MASK_TABLE_OFS(0) < -(1 << 11));
 
-static void tcg_out_goto(TCGContext *s, tcg_insn_unit *target)
+static void tcg_out_goto(TCGContext *s, const tcg_insn_unit *target)
 {
     tcg_out_opc_jump(s, OPC_JAL, TCG_REG_ZERO, 0);
     bool ok = reloc_jimm20(s->code_ptr - 1, target);
@@ -993,7 +996,8 @@ static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
 }
 
@@ -1012,7 +1016,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     }
 
     /* resolve label address */
-    if (!reloc_sbimm12(l->label_ptr[0], s->code_ptr)) {
+    if (!reloc_sbimm12(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1046,7 +1050,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     }
 
     /* resolve label address */
-    if (!reloc_sbimm12(l->label_ptr[0], s->code_ptr)) {
+    if (!reloc_sbimm12(l->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1232,7 +1236,7 @@ static void tcg_out_qemu_st(TCGContext *s, const TCGArg *args, bool is_64)
 #endif
 }
 
-static tcg_insn_unit *tb_ret_addr;
+static const tcg_insn_unit *tb_ret_addr;
 
 static void tcg_out_op(TCGContext *s, TCGOpcode opc,
                        const TCGArg *args, const int *const_args)
@@ -1780,11 +1784,12 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, tcg_target_call_iarg_regs[1], 0);
 
     /* Return path for goto_ptr. Set return value to 0 */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_A0, TCG_REG_ZERO);
 
     /* TB epilogue */
-    tb_ret_addr = s->code_ptr;
+    tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr);
     for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); i++) {
         tcg_out_ld(s, TCG_TYPE_REG, tcg_target_callee_save_regs[i],
                    TCG_REG_SP, SAVE_OFS + i * REG_SIZE);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 35/41] accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (33 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 34/41] tcg/riscv: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 36/41] tcg/mips: Do not assert on relocation overflow Richard Henderson
                   ` (8 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Re-use the 256MiB region handling from alloc_code_gen_buffer_anon,
and replace that with the shared file mapping.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 accel/tcg/translate-all.c | 46 ++++++++++++++++++++++++++++++++-------
 1 file changed, 38 insertions(+), 8 deletions(-)

diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index 17df6c94fa..b49aaf1026 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1137,24 +1137,40 @@ static bool alloc_code_gen_buffer_anon(size_t size, int prot,
 
 static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
 {
-    void *buf_rw, *buf_rx;
+    void *buf_rw = NULL, *buf_rx = MAP_FAILED;
     int fd = -1;
 
+#ifdef __mips__
+    /* Find space for the RX mapping, vs the 256MiB regions. */
+    if (!alloc_code_gen_buffer_anon(size, PROT_NONE,
+                                    MAP_PRIVATE | MAP_ANONYMOUS |
+                                    MAP_NORESERVE, errp)) {
+        return false;
+    }
+    /* The size of the mapping may have been adjusted. */
+    size = tcg_ctx->code_gen_buffer_size;
+    buf_rx = tcg_ctx->code_gen_buffer;
+#endif
+
     buf_rw = qemu_memfd_alloc("tcg-jit", size, 0, &fd, errp);
     if (buf_rw == NULL) {
-        return false;
+        goto fail;
     }
 
+#ifdef __mips__
+    void *tmp = mmap(buf_rx, size, PROT_READ | PROT_EXEC,
+                     MAP_SHARED | MAP_FIXED, fd, 0);
+    if (tmp != buf_rx) {
+        goto fail_rx;
+    }
+#else
     buf_rx = mmap(NULL, size, PROT_READ | PROT_EXEC, MAP_SHARED, fd, 0);
     if (buf_rx == MAP_FAILED) {
-        error_setg_errno(errp, errno,
-                         "failed to map shared memory for execute");
-        munmap(buf_rw, size);
-        close(fd);
-        return false;
+        goto fail_rx;
     }
-    close(fd);
+#endif
 
+    close(fd);
     tcg_ctx->code_gen_buffer = buf_rw;
     tcg_ctx->code_gen_buffer_size = size;
     tcg_splitwx_diff = buf_rx - buf_rw;
@@ -1163,6 +1179,20 @@ static bool alloc_code_gen_buffer_splitwx_memfd(size_t size, Error **errp)
     qemu_madvise(buf_rw, size, QEMU_MADV_HUGEPAGE);
     qemu_madvise(buf_rx, size, QEMU_MADV_HUGEPAGE);
     return true;
+
+ fail_rx:
+    error_setg_errno(errp, errno, "failed to map shared memory for execute");
+ fail:
+    if (buf_rx != MAP_FAILED) {
+        munmap(buf_rx, size);
+    }
+    if (buf_rw) {
+        munmap(buf_rw, size);
+    }
+    if (fd >= 0) {
+        close(fd);
+    }
+    return false;
 }
 #endif /* CONFIG_POSIX */
 
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 36/41] tcg/mips: Do not assert on relocation overflow
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (34 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 35/41] accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 37/41] tcg/mips: Support split-wx code generation Richard Henderson
                   ` (7 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

This target was not updated with 7ecd02a06f8, and so did
not allow re-compilation with relocation overflow.

Remove reloc_26 and reloc_26_val as unused.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/mips/tcg-target.c.inc | 53 ++++++++++++++-------------------------
 1 file changed, 19 insertions(+), 34 deletions(-)

diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 52638e920c..37faf1356c 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -144,29 +144,15 @@ static tcg_insn_unit *bswap32_addr;
 static tcg_insn_unit *bswap32u_addr;
 static tcg_insn_unit *bswap64_addr;
 
-static inline uint32_t reloc_pc16_val(tcg_insn_unit *pc,
-                                      const tcg_insn_unit *target)
+static bool reloc_pc16(tcg_insn_unit *pc, const tcg_insn_unit *target)
 {
     /* Let the compiler perform the right-shift as part of the arithmetic.  */
     ptrdiff_t disp = target - (pc + 1);
-    tcg_debug_assert(disp == (int16_t)disp);
-    return disp & 0xffff;
-}
-
-static inline void reloc_pc16(tcg_insn_unit *pc, const tcg_insn_unit *target)
-{
-    *pc = deposit32(*pc, 0, 16, reloc_pc16_val(pc, target));
-}
-
-static inline uint32_t reloc_26_val(tcg_insn_unit *pc, tcg_insn_unit *target)
-{
-    tcg_debug_assert((((uintptr_t)pc ^ (uintptr_t)target) & 0xf0000000) == 0);
-    return ((uintptr_t)target >> 2) & 0x3ffffff;
-}
-
-static inline void reloc_26(tcg_insn_unit *pc, tcg_insn_unit *target)
-{
-    *pc = deposit32(*pc, 0, 26, reloc_26_val(pc, target));
+    if (disp == (int16_t)disp) {
+        *pc = deposit32(*pc, 0, 16, disp);
+        return true;
+    }
+    return false;
 }
 
 static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
@@ -174,8 +160,7 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
 {
     tcg_debug_assert(type == R_MIPS_PC16);
     tcg_debug_assert(addend == 0);
-    reloc_pc16(code_ptr, (tcg_insn_unit *)value);
-    return true;
+    return reloc_pc16(code_ptr, (const tcg_insn_unit *)value);
 }
 
 #define TCG_CT_CONST_ZERO 0x100
@@ -925,11 +910,7 @@ static void tcg_out_brcond(TCGContext *s, TCGCond cond, TCGReg arg1,
     }
 
     tcg_out_opc_br(s, b_opc, arg1, arg2);
-    if (l->has_value) {
-        reloc_pc16(s->code_ptr - 1, l->u.value_ptr);
-    } else {
-        tcg_out_reloc(s, s->code_ptr - 1, R_MIPS_PC16, l, 0);
-    }
+    tcg_out_reloc(s, s->code_ptr - 1, R_MIPS_PC16, l, 0);
     tcg_out_nop(s);
 }
 
@@ -1316,9 +1297,10 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     int i;
 
     /* resolve label address */
-    reloc_pc16(l->label_ptr[0], s->code_ptr);
-    if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
-        reloc_pc16(l->label_ptr[1], s->code_ptr);
+    if (!reloc_pc16(l->label_ptr[0], s->code_ptr)
+        || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS
+            && !reloc_pc16(l->label_ptr[1], s->code_ptr))) {
+        return false;
     }
 
     i = 1;
@@ -1346,7 +1328,9 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     }
 
     tcg_out_opc_br(s, OPC_BEQ, TCG_REG_ZERO, TCG_REG_ZERO);
-    reloc_pc16(s->code_ptr - 1, l->raddr);
+    if (!reloc_pc16(s->code_ptr - 1, l->raddr)) {
+        return false;
+    }
 
     /* delay slot */
     if (TCG_TARGET_REG_BITS == 64 && l->type == TCG_TYPE_I32) {
@@ -1366,9 +1350,10 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
     int i;
 
     /* resolve label address */
-    reloc_pc16(l->label_ptr[0], s->code_ptr);
-    if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
-        reloc_pc16(l->label_ptr[1], s->code_ptr);
+    if (!reloc_pc16(l->label_ptr[0], s->code_ptr)
+        || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS
+            && !reloc_pc16(l->label_ptr[1], s->code_ptr))) {
+        return false;
     }
 
     i = 1;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 37/41] tcg/mips: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (35 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 36/41] tcg/mips: Do not assert on relocation overflow Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 38/41] tcg/arm: " Richard Henderson
                   ` (6 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/mips/tcg-target.h     |  2 +-
 tcg/mips/tcg-target.c.inc | 43 ++++++++++++++++++++++-----------------
 2 files changed, 25 insertions(+), 20 deletions(-)

diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index d231522dc9..d7d8e6ea1c 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -206,7 +206,7 @@ extern bool use_mips32r2_instructions;
 
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 /* Flush the dcache at RW, and the icache at RX, as necessary. */
 static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 37faf1356c..a2201bd1dd 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -139,17 +139,18 @@ static const TCGReg tcg_target_call_oarg_regs[2] = {
     TCG_REG_V1
 };
 
-static tcg_insn_unit *tb_ret_addr;
-static tcg_insn_unit *bswap32_addr;
-static tcg_insn_unit *bswap32u_addr;
-static tcg_insn_unit *bswap64_addr;
+static const tcg_insn_unit *tb_ret_addr;
+static const tcg_insn_unit *bswap32_addr;
+static const tcg_insn_unit *bswap32u_addr;
+static const tcg_insn_unit *bswap64_addr;
 
-static bool reloc_pc16(tcg_insn_unit *pc, const tcg_insn_unit *target)
+static bool reloc_pc16(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
     /* Let the compiler perform the right-shift as part of the arithmetic.  */
-    ptrdiff_t disp = target - (pc + 1);
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t disp = target - (src_rx + 1);
     if (disp == (int16_t)disp) {
-        *pc = deposit32(*pc, 0, 16, disp);
+        *src_rw = deposit32(*src_rw, 0, 16, disp);
         return true;
     }
     return false;
@@ -505,7 +506,7 @@ static void tcg_out_opc_sa64(TCGContext *s, MIPSInsn opc1, MIPSInsn opc2,
 static bool tcg_out_opc_jmp(TCGContext *s, MIPSInsn opc, const void *target)
 {
     uintptr_t dest = (uintptr_t)target;
-    uintptr_t from = (uintptr_t)s->code_ptr + 4;
+    uintptr_t from = (uintptr_t)tcg_splitwx_to_rx(s->code_ptr) + 4;
     int32_t inst;
 
     /* The pc-region branch happens within the 256MB region of
@@ -617,7 +618,7 @@ static inline void tcg_out_bswap16s(TCGContext *s, TCGReg ret, TCGReg arg)
     }
 }
 
-static void tcg_out_bswap_subr(TCGContext *s, tcg_insn_unit *sub)
+static void tcg_out_bswap_subr(TCGContext *s, const tcg_insn_unit *sub)
 {
     bool ok = tcg_out_opc_jmp(s, OPC_JAL, sub);
     tcg_debug_assert(ok);
@@ -1282,7 +1283,8 @@ static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
     if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
         label->label_ptr[1] = label_ptr[1];
@@ -1291,15 +1293,16 @@ static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
 
 static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
 {
+    const tcg_insn_unit *tgt_rx = tcg_splitwx_to_rx(s->code_ptr);
     TCGMemOpIdx oi = l->oi;
     MemOp opc = get_memop(oi);
     TCGReg v0;
     int i;
 
     /* resolve label address */
-    if (!reloc_pc16(l->label_ptr[0], s->code_ptr)
+    if (!reloc_pc16(l->label_ptr[0], tgt_rx)
         || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS
-            && !reloc_pc16(l->label_ptr[1], s->code_ptr))) {
+            && !reloc_pc16(l->label_ptr[1], tgt_rx))) {
         return false;
     }
 
@@ -1344,15 +1347,16 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
 
 static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *l)
 {
+    const tcg_insn_unit *tgt_rx = tcg_splitwx_to_rx(s->code_ptr);
     TCGMemOpIdx oi = l->oi;
     MemOp opc = get_memop(oi);
     MemOp s_bits = opc & MO_SIZE;
     int i;
 
     /* resolve label address */
-    if (!reloc_pc16(l->label_ptr[0], s->code_ptr)
+    if (!reloc_pc16(l->label_ptr[0], tgt_rx)
         || (TCG_TARGET_REG_BITS < TARGET_LONG_BITS
-            && !reloc_pc16(l->label_ptr[1], s->code_ptr))) {
+            && !reloc_pc16(l->label_ptr[1], tgt_rx))) {
         return false;
     }
 
@@ -2469,11 +2473,12 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_V0, TCG_REG_ZERO);
 
     /* TB epilogue */
-    tb_ret_addr = s->code_ptr;
+    tb_ret_addr = tcg_splitwx_to_rx(s->code_ptr);
     for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); i++) {
         tcg_out_ld(s, TCG_TYPE_REG, tcg_target_callee_save_regs[i],
                    TCG_REG_SP, SAVE_OFS + i * REG_SIZE);
@@ -2493,7 +2498,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     /*
      * bswap32 -- 32-bit swap (signed result for mips64).  a0 = abcd.
      */
-    bswap32_addr = align_code_ptr(s);
+    bswap32_addr = tcg_splitwx_to_rx(align_code_ptr(s));
     /* t3 = (ssss)d000 */
     tcg_out_opc_sa(s, OPC_SLL, TCG_TMP3, TCG_TMP0, 24);
     /* t1 = 000a */
@@ -2521,7 +2526,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     /*
      * bswap32u -- unsigned 32-bit swap.  a0 = ....abcd.
      */
-    bswap32u_addr = align_code_ptr(s);
+    bswap32u_addr = tcg_splitwx_to_rx(align_code_ptr(s));
     /* t1 = (0000)000d */
     tcg_out_opc_imm(s, OPC_ANDI, TCG_TMP1, TCG_TMP0, 0xff);
     /* t3 = 000a */
@@ -2547,7 +2552,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     /*
      * bswap64 -- 64-bit swap.  a0 = abcdefgh
      */
-    bswap64_addr = align_code_ptr(s);
+    bswap64_addr = tcg_splitwx_to_rx(align_code_ptr(s));
     /* t3 = h0000000 */
     tcg_out_dsll(s, TCG_TMP3, TCG_TMP0, 56);
     /* t1 = 0000000a */
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 38/41] tcg/arm: Support split-wx code generation
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (36 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 37/41] tcg/mips: Support split-wx code generation Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 39/41] tcg: Remove TCG_TARGET_SUPPORT_MIRROR Richard Henderson
                   ` (5 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/arm/tcg-target.h     |  2 +-
 tcg/arm/tcg-target.c.inc | 37 +++++++++++++++++++++----------------
 2 files changed, 22 insertions(+), 17 deletions(-)

diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index e355d6a4b2..17f6be9cfc 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -150,6 +150,6 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       0
+#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 9b9400f164..3d2717aeb0 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -187,29 +187,32 @@ static const uint8_t tcg_cond_to_arm_cond[] = {
     [TCG_COND_GTU] = COND_HI,
 };
 
-static inline bool reloc_pc24(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_pc24(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = (tcg_ptr_byte_diff(target, code_ptr) - 8) >> 2;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t offset = (tcg_ptr_byte_diff(target, src_rx) - 8) >> 2;
+
     if (offset == sextract32(offset, 0, 24)) {
-        *code_ptr = (*code_ptr & ~0xffffff) | (offset & 0xffffff);
+        *src_rw = deposit32(*src_rw, 0, 24, offset);
         return true;
     }
     return false;
 }
 
-static inline bool reloc_pc13(tcg_insn_unit *code_ptr, tcg_insn_unit *target)
+static bool reloc_pc13(tcg_insn_unit *src_rw, const tcg_insn_unit *target)
 {
-    ptrdiff_t offset = tcg_ptr_byte_diff(target, code_ptr) - 8;
+    const tcg_insn_unit *src_rx = tcg_splitwx_to_rx(src_rw);
+    ptrdiff_t offset = tcg_ptr_byte_diff(target, src_rx) - 8;
 
     if (offset >= -0xfff && offset <= 0xfff) {
-        tcg_insn_unit insn = *code_ptr;
+        tcg_insn_unit insn = *src_rw;
         bool u = (offset >= 0);
         if (!u) {
             offset = -offset;
         }
         insn = deposit32(insn, 23, 1, u);
         insn = deposit32(insn, 0, 12, offset);
-        *code_ptr = insn;
+        *src_rw = insn;
         return true;
     }
     return false;
@@ -221,9 +224,9 @@ static bool patch_reloc(tcg_insn_unit *code_ptr, int type,
     tcg_debug_assert(addend == 0);
 
     if (type == R_ARM_PC24) {
-        return reloc_pc24(code_ptr, (tcg_insn_unit *)value);
+        return reloc_pc24(code_ptr, (const tcg_insn_unit *)value);
     } else if (type == R_ARM_PC13) {
-        return reloc_pc13(code_ptr, (tcg_insn_unit *)value);
+        return reloc_pc13(code_ptr, (const tcg_insn_unit *)value);
     } else {
         g_assert_not_reached();
     }
@@ -617,7 +620,7 @@ static void tcg_out_movi32(TCGContext *s, int cond, int rd, uint32_t arg)
 
     /* Check for a pc-relative address.  This will usually be the TB,
        or within the TB, which is immediately before the code block.  */
-    diff = arg - ((intptr_t)s->code_ptr + 8);
+    diff = tcg_pcrel_diff(s, (void *)arg) - 8;
     if (diff >= 0) {
         rot = encode_imm(diff);
         if (rot >= 0) {
@@ -1337,7 +1340,8 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    label->raddr = raddr;
+    /* TODO: Cast goes away when all hosts converted */
+    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
@@ -1348,7 +1352,7 @@ static bool tcg_out_qemu_ld_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     MemOp opc = get_memop(oi);
     void *func;
 
-    if (!reloc_pc24(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc24(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1411,7 +1415,7 @@ static bool tcg_out_qemu_st_slow_path(TCGContext *s, TCGLabelQemuLdst *lb)
     TCGMemOpIdx oi = lb->oi;
     MemOp opc = get_memop(oi);
 
-    if (!reloc_pc24(lb->label_ptr[0], s->code_ptr)) {
+    if (!reloc_pc24(lb->label_ptr[0], tcg_splitwx_to_rx(s->code_ptr))) {
         return false;
     }
 
@@ -1762,8 +1766,8 @@ static inline void tcg_out_op(TCGContext *s, TCGOpcode opc,
             TCGReg base = TCG_REG_PC;
 
             tcg_debug_assert(s->tb_jmp_insn_offset == 0);
-            ptr = (intptr_t)(s->tb_jmp_target_addr + args[0]);
-            dif = ptr - ((intptr_t)s->code_ptr + 8);
+            ptr = (intptr_t)tcg_splitwx_to_rx(s->tb_jmp_target_addr + args[0]);
+            dif = tcg_pcrel_diff(s, (void *)ptr) - 8;
             dil = sextract32(dif, 0, 12);
             if (dif != dil) {
                 /* The TB is close, but outside the 12 bits addressable by
@@ -2297,7 +2301,8 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    tcg_code_gen_epilogue = s->code_ptr;
+    /* TODO: Cast goes away when all hosts converted */
+    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R0, 0);
     tcg_out_epilogue(s);
 }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 39/41] tcg: Remove TCG_TARGET_SUPPORT_MIRROR
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (37 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 38/41] tcg/arm: " Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 40/41] tcg: Constify tcg_code_gen_epilogue Richard Henderson
                   ` (4 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Now that all native tcg hosts support splitwx,
remove the define.  Replace the one use with a
test for CONFIG_TCG_INTERPRETER.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.h  |  1 -
 tcg/arm/tcg-target.h      |  1 -
 tcg/i386/tcg-target.h     |  1 -
 tcg/mips/tcg-target.h     |  1 -
 tcg/ppc/tcg-target.h      |  1 -
 tcg/riscv/tcg-target.h    |  1 -
 tcg/s390/tcg-target.h     |  1 -
 tcg/sparc/tcg-target.h    |  1 -
 tcg/tci/tcg-target.h      |  1 -
 accel/tcg/translate-all.c | 14 +++++++-------
 10 files changed, 7 insertions(+), 16 deletions(-)

diff --git a/tcg/aarch64/tcg-target.h b/tcg/aarch64/tcg-target.h
index abb94f9458..fedd88f6fb 100644
--- a/tcg/aarch64/tcg-target.h
+++ b/tcg/aarch64/tcg-target.h
@@ -155,6 +155,5 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif /* AARCH64_TCG_TARGET_H */
diff --git a/tcg/arm/tcg-target.h b/tcg/arm/tcg-target.h
index 17f6be9cfc..b21a2fb6a1 100644
--- a/tcg/arm/tcg-target.h
+++ b/tcg/arm/tcg-target.h
@@ -150,6 +150,5 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/i386/tcg-target.h b/tcg/i386/tcg-target.h
index bbbd1c2d4a..f52ba0ffec 100644
--- a/tcg/i386/tcg-target.h
+++ b/tcg/i386/tcg-target.h
@@ -236,6 +236,5 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/mips/tcg-target.h b/tcg/mips/tcg-target.h
index d7d8e6ea1c..cd548dacec 100644
--- a/tcg/mips/tcg-target.h
+++ b/tcg/mips/tcg-target.h
@@ -206,7 +206,6 @@ extern bool use_mips32r2_instructions;
 
 #define TCG_TARGET_DEFAULT_MO (0)
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 /* Flush the dcache at RW, and the icache at RX, as necessary. */
 static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
diff --git a/tcg/ppc/tcg-target.h b/tcg/ppc/tcg-target.h
index a8628b6cad..8f3e4c924a 100644
--- a/tcg/ppc/tcg-target.h
+++ b/tcg/ppc/tcg-target.h
@@ -185,6 +185,5 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/riscv/tcg-target.h b/tcg/riscv/tcg-target.h
index 0eb19f2b11..e03fd17427 100644
--- a/tcg/riscv/tcg-target.h
+++ b/tcg/riscv/tcg-target.h
@@ -179,6 +179,5 @@ void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 #define TCG_TARGET_NEED_POOL_LABELS
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP 0
-#define TCG_TARGET_SUPPORT_MIRROR   1
 
 #endif
diff --git a/tcg/s390/tcg-target.h b/tcg/s390/tcg-target.h
index 1fd8b3858e..c5a749e425 100644
--- a/tcg/s390/tcg-target.h
+++ b/tcg/s390/tcg-target.h
@@ -163,6 +163,5 @@ static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
 #define TCG_TARGET_NEED_LDST_LABELS
 #endif
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/sparc/tcg-target.h b/tcg/sparc/tcg-target.h
index bb2505bfc7..87e2be61e6 100644
--- a/tcg/sparc/tcg-target.h
+++ b/tcg/sparc/tcg-target.h
@@ -181,6 +181,5 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 void tb_target_set_jmp_target(uintptr_t, uintptr_t, uintptr_t, uintptr_t);
 
 #define TCG_TARGET_NEED_POOL_LABELS
-#define TCG_TARGET_SUPPORT_MIRROR       1
 
 #endif
diff --git a/tcg/tci/tcg-target.h b/tcg/tci/tcg-target.h
index 3653fef947..a19a6b06e5 100644
--- a/tcg/tci/tcg-target.h
+++ b/tcg/tci/tcg-target.h
@@ -200,7 +200,6 @@ static inline void flush_idcache_range(uintptr_t rx, uintptr_t rw, size_t len)
 #define TCG_TARGET_DEFAULT_MO  (0)
 
 #define TCG_TARGET_HAS_MEMORY_BSWAP     1
-#define TCG_TARGET_SUPPORT_MIRROR       0
 
 static inline void tb_target_set_jmp_target(uintptr_t tc_ptr, uintptr_t jmp_rx,
                                             uintptr_t jmp_rw, uintptr_t addr)
diff --git a/accel/tcg/translate-all.c b/accel/tcg/translate-all.c
index b49aaf1026..06102871e7 100644
--- a/accel/tcg/translate-all.c
+++ b/accel/tcg/translate-all.c
@@ -1257,14 +1257,14 @@ static bool alloc_code_gen_buffer_splitwx_vmremap(size_t size, Error **errp)
 
 static bool alloc_code_gen_buffer_splitwx(size_t size, Error **errp)
 {
-    if (TCG_TARGET_SUPPORT_MIRROR) {
-#ifdef CONFIG_DARWIN
-        return alloc_code_gen_buffer_splitwx_vmremap(size, errp);
+#ifndef CONFIG_TCG_INTERPRETER
+# ifdef CONFIG_DARWIN
+    return alloc_code_gen_buffer_splitwx_vmremap(size, errp);
+# endif
+# ifdef CONFIG_POSIX
+    return alloc_code_gen_buffer_splitwx_memfd(size, errp);
+# endif
 #endif
-#ifdef CONFIG_POSIX
-        return alloc_code_gen_buffer_splitwx_memfd(size, errp);
-#endif
-    }
     error_setg(errp, "jit split-wx not supported");
     return false;
 }
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 40/41] tcg: Constify tcg_code_gen_epilogue
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (38 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 39/41] tcg: Remove TCG_TARGET_SUPPORT_MIRROR Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  3:29 ` [PATCH v3 41/41] tcg: Constify TCGLabelQemuLdst.raddr Richard Henderson
                   ` (3 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Now that all native tcg hosts support splitwx,
make this pointer const.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 include/tcg/tcg.h            | 2 +-
 tcg/tcg.c                    | 2 +-
 tcg/aarch64/tcg-target.c.inc | 3 +--
 tcg/arm/tcg-target.c.inc     | 3 +--
 tcg/i386/tcg-target.c.inc    | 3 +--
 tcg/mips/tcg-target.c.inc    | 3 +--
 tcg/ppc/tcg-target.c.inc     | 3 +--
 tcg/riscv/tcg-target.c.inc   | 3 +--
 tcg/s390/tcg-target.c.inc    | 3 +--
 tcg/sparc/tcg-target.c.inc   | 3 +--
 10 files changed, 10 insertions(+), 18 deletions(-)

diff --git a/include/tcg/tcg.h b/include/tcg/tcg.h
index 90ec7c1445..477919aeb6 100644
--- a/include/tcg/tcg.h
+++ b/include/tcg/tcg.h
@@ -677,7 +677,7 @@ struct TCGContext {
 
 extern TCGContext tcg_init_ctx;
 extern __thread TCGContext *tcg_ctx;
-extern void *tcg_code_gen_epilogue;
+extern const void *tcg_code_gen_epilogue;
 extern uintptr_t tcg_splitwx_diff;
 extern TCGv_env cpu_env;
 
diff --git a/tcg/tcg.c b/tcg/tcg.c
index 07a4bd2c57..5c19b1e6b3 100644
--- a/tcg/tcg.c
+++ b/tcg/tcg.c
@@ -160,7 +160,7 @@ static int tcg_out_ldst_finalize(TCGContext *s);
 static TCGContext **tcg_ctxs;
 static unsigned int n_tcg_ctxs;
 TCGv_env cpu_env = 0;
-void *tcg_code_gen_epilogue;
+const void *tcg_code_gen_epilogue;
 uintptr_t tcg_splitwx_diff;
 
 #ifndef CONFIG_TCG_INTERPRETER
diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index e398913c0c..32a60eba5e 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -2900,8 +2900,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_X0, 0);
 
     /* TB epilogue */
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index 3d2717aeb0..d6cb19ca9f 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -2301,8 +2301,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R0, 0);
     tcg_out_epilogue(s);
 }
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index 23c7a8a383..be57d2330a 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -3826,8 +3826,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_REG, TCG_REG_EAX, 0);
 
     /* TB epilogue */
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index a2201bd1dd..18fd474593 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -2473,8 +2473,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_V0, TCG_REG_ZERO);
 
     /* TB epilogue */
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index fe33687787..3580bbabc1 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -2346,8 +2346,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out32(s, BCCTR | BO_ALWAYS);
 
     /* Epilogue */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
 
     tcg_out_ld(s, TCG_TYPE_PTR, TCG_REG_R0, TCG_REG_R1, FRAME_SIZE+LR_OFFSET);
     for (i = 0; i < ARRAY_SIZE(tcg_target_callee_save_regs); ++i) {
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 5c1e0f8fc1..7b4ee4a084 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -1784,8 +1784,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_opc_imm(s, OPC_JALR, TCG_REG_ZERO, tcg_target_call_iarg_regs[1], 0);
 
     /* Return path for goto_ptr. Set return value to 0 */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_mov(s, TCG_TYPE_REG, TCG_REG_A0, TCG_REG_ZERO);
 
     /* TB epilogue */
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index 582a8ef941..b3660ffedf 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -2551,8 +2551,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
      * Return path for goto_ptr. Set return value to 0, a-la exit_tb,
      * and fall through to the rest of the epilogue.
      */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_movi(s, TCG_TYPE_PTR, TCG_REG_R2, 0);
 
     /* TB epilogue */
diff --git a/tcg/sparc/tcg-target.c.inc b/tcg/sparc/tcg-target.c.inc
index 182124b96c..922ae96481 100644
--- a/tcg/sparc/tcg-target.c.inc
+++ b/tcg/sparc/tcg-target.c.inc
@@ -1039,8 +1039,7 @@ static void tcg_target_qemu_prologue(TCGContext *s)
     tcg_out_nop(s);
 
     /* Epilogue for goto_ptr.  */
-    /* TODO: Cast goes away when all hosts converted */
-    tcg_code_gen_epilogue = (void *)tcg_splitwx_to_rx(s->code_ptr);
+    tcg_code_gen_epilogue = tcg_splitwx_to_rx(s->code_ptr);
     tcg_out_arithi(s, TCG_REG_G0, TCG_REG_I7, 8, RETURN);
     /* delay slot */
     tcg_out_movi_imm13(s, TCG_REG_O0, 0);
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* [PATCH v3 41/41] tcg: Constify TCGLabelQemuLdst.raddr
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (39 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 40/41] tcg: Constify tcg_code_gen_epilogue Richard Henderson
@ 2020-11-06  3:29 ` Richard Henderson
  2020-11-06  4:00 ` [PATCH v3 00/41] Mirror map JIT memory for TCG no-reply
                   ` (2 subsequent siblings)
  43 siblings, 0 replies; 57+ messages in thread
From: Richard Henderson @ 2020-11-06  3:29 UTC (permalink / raw)
  To: qemu-devel; +Cc: j

Now that all native tcg hosts support splitwx,
make this pointer const.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 tcg/aarch64/tcg-target.c.inc | 3 +--
 tcg/arm/tcg-target.c.inc     | 3 +--
 tcg/i386/tcg-target.c.inc    | 3 +--
 tcg/mips/tcg-target.c.inc    | 3 +--
 tcg/ppc/tcg-target.c.inc     | 3 +--
 tcg/riscv/tcg-target.c.inc   | 3 +--
 tcg/s390/tcg-target.c.inc    | 3 +--
 tcg/tcg-ldst.c.inc           | 2 +-
 8 files changed, 8 insertions(+), 15 deletions(-)

diff --git a/tcg/aarch64/tcg-target.c.inc b/tcg/aarch64/tcg-target.c.inc
index 32a60eba5e..9e15128f31 100644
--- a/tcg/aarch64/tcg-target.c.inc
+++ b/tcg/aarch64/tcg-target.c.inc
@@ -1636,8 +1636,7 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->type = ext;
     label->datalo_reg = data_reg;
     label->addrlo_reg = addr_reg;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
diff --git a/tcg/arm/tcg-target.c.inc b/tcg/arm/tcg-target.c.inc
index d6cb19ca9f..0fd1126454 100644
--- a/tcg/arm/tcg-target.c.inc
+++ b/tcg/arm/tcg-target.c.inc
@@ -1340,8 +1340,7 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
diff --git a/tcg/i386/tcg-target.c.inc b/tcg/i386/tcg-target.c.inc
index be57d2330a..18a3af53bb 100644
--- a/tcg/i386/tcg-target.c.inc
+++ b/tcg/i386/tcg-target.c.inc
@@ -1786,8 +1786,7 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, bool is_64,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
     if (TARGET_LONG_BITS > TCG_TARGET_REG_BITS) {
         label->label_ptr[1] = label_ptr[1];
diff --git a/tcg/mips/tcg-target.c.inc b/tcg/mips/tcg-target.c.inc
index 18fd474593..add157f6c3 100644
--- a/tcg/mips/tcg-target.c.inc
+++ b/tcg/mips/tcg-target.c.inc
@@ -1283,8 +1283,7 @@ static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
     if (TCG_TARGET_REG_BITS < TARGET_LONG_BITS) {
         label->label_ptr[1] = label_ptr[1];
diff --git a/tcg/ppc/tcg-target.c.inc b/tcg/ppc/tcg-target.c.inc
index 3580bbabc1..71b40e7490 100644
--- a/tcg/ppc/tcg-target.c.inc
+++ b/tcg/ppc/tcg-target.c.inc
@@ -2001,8 +2001,7 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi_reg;
     label->addrlo_reg = addrlo_reg;
     label->addrhi_reg = addrhi_reg;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = lptr;
 }
 
diff --git a/tcg/riscv/tcg-target.c.inc b/tcg/riscv/tcg-target.c.inc
index 7b4ee4a084..f48a028dac 100644
--- a/tcg/riscv/tcg-target.c.inc
+++ b/tcg/riscv/tcg-target.c.inc
@@ -996,8 +996,7 @@ static void add_qemu_ldst_label(TCGContext *s, int is_ld, TCGMemOpIdx oi,
     label->datahi_reg = datahi;
     label->addrlo_reg = addrlo;
     label->addrhi_reg = addrhi;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr[0];
 }
 
diff --git a/tcg/s390/tcg-target.c.inc b/tcg/s390/tcg-target.c.inc
index b3660ffedf..d7ef079055 100644
--- a/tcg/s390/tcg-target.c.inc
+++ b/tcg/s390/tcg-target.c.inc
@@ -1587,8 +1587,7 @@ static void add_qemu_ldst_label(TCGContext *s, bool is_ld, TCGMemOpIdx oi,
     label->oi = oi;
     label->datalo_reg = data;
     label->addrlo_reg = addr;
-    /* TODO: Cast goes away when all hosts converted */
-    label->raddr = (void *)tcg_splitwx_to_rx(raddr);
+    label->raddr = tcg_splitwx_to_rx(raddr);
     label->label_ptr[0] = label_ptr;
 }
 
diff --git a/tcg/tcg-ldst.c.inc b/tcg/tcg-ldst.c.inc
index 05f9b3ccd6..c3ce88e69d 100644
--- a/tcg/tcg-ldst.c.inc
+++ b/tcg/tcg-ldst.c.inc
@@ -28,7 +28,7 @@ typedef struct TCGLabelQemuLdst {
     TCGReg addrhi_reg;      /* reg index for high word of guest virtual addr */
     TCGReg datalo_reg;      /* reg index for low word to be loaded or stored */
     TCGReg datahi_reg;      /* reg index for high word to be loaded or stored */
-    tcg_insn_unit *raddr;   /* gen code addr of the next IR of qemu_ld/st IR */
+    const tcg_insn_unit *raddr;   /* addr of the next IR of qemu_ld/st IR */
     tcg_insn_unit *label_ptr[2]; /* label pointers to be updated */
     QSIMPLEQ_ENTRY(TCGLabelQemuLdst) next;
 } TCGLabelQemuLdst;
-- 
2.25.1



^ permalink raw reply related	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (40 preceding siblings ...)
  2020-11-06  3:29 ` [PATCH v3 41/41] tcg: Constify TCGLabelQemuLdst.raddr Richard Henderson
@ 2020-11-06  4:00 ` no-reply
  2020-11-08  3:38 ` Joelle van Dyne
  2020-11-17  3:47 ` Joelle van Dyne
  43 siblings, 0 replies; 57+ messages in thread
From: no-reply @ 2020-11-06  4:00 UTC (permalink / raw)
  To: richard.henderson; +Cc: qemu-devel, j

Patchew URL: https://patchew.org/QEMU/20201106032921.600200-1-richard.henderson@linaro.org/



Hi,

This series seems to have some coding style problems. See output below for
more information:

Type: series
Message-id: 20201106032921.600200-1-richard.henderson@linaro.org
Subject: [PATCH v3 00/41] Mirror map JIT memory for TCG

=== TEST SCRIPT BEGIN ===
#!/bin/bash
git rev-parse base > /dev/null || exit 0
git config --local diff.renamelimit 0
git config --local diff.renames True
git config --local diff.algorithm histogram
./scripts/checkpatch.pl --mailback base..
=== TEST SCRIPT END ===

Updating 3c8cf5a9c21ff8782164d1def7f44bd888713384
From https://github.com/patchew-project/qemu
 * [new tag]         patchew/20201106032921.600200-1-richard.henderson@linaro.org -> patchew/20201106032921.600200-1-richard.henderson@linaro.org
Switched to a new branch 'test'
170f310 tcg: Constify TCGLabelQemuLdst.raddr
c336494 tcg: Constify tcg_code_gen_epilogue
a009e99 tcg: Remove TCG_TARGET_SUPPORT_MIRROR
545feb7 tcg/arm: Support split-wx code generation
a873c61 tcg/mips: Support split-wx code generation
7f12d40 tcg/mips: Do not assert on relocation overflow
200ecb3 accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd
edd72db tcg/riscv: Support split-wx code generation
1c6764d tcg/riscv: Remove branch-over-branch fallback
63883fc tcg/riscv: Fix branch range checks
dff34e0 tcg/s390: Support split-wx code generation
8d88879 tcg/s390: Use tcg_tbrel_diff
f0fea63 tcg/sparc: Support split-wx code generation
5a837a3 tcg/sparc: Use tcg_tbrel_diff
684c281 tcg/ppc: Support split-wx code generation
a254bfd tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB
4c4f647 tcg/ppc: Use tcg_tbrel_diff
5f81f0e tcg: Introduce tcg_tbrel_diff
1cab418 tcg/tci: Push const down through bytecode reading
c55a8c0 disas: Push const down through host disasassembly
55b926c tcg/aarch64: Support split-wx code generation
aef71b4 tcg/aarch64: Implement flush_idcache_range manually
548fc79 tcg/aarch64: Use B not BL for tcg_out_goto_long
6cbd22a tcg/i386: Support split-wx code generation
fe36cad tcg: Return the TB pointer from the rx region from exit_tb
ef96a10 accel/tcg: Support split-wx for darwin/iOS with vm_remap
fed5e19 accel/tcg: Support split-wx for linux with memfd
caaf645 tcg: Add --accel tcg,split-wx property
f93ae22 tcg: Use Error with alloc_code_gen_buffer
b6992b5 tcg: Make tb arg to synchronize_from_tb const
043973b tcg: Make DisasContextBase.tb const
1d83486 tcg: Adjust tb_target_set_jmp_target for split-wx
eec18a6 tcg: Adjust tcg_register_jit for const
65e76b9 tcg: Adjust tcg_out_label for const
44975a9 tcg: Adjust tcg_out_call for const
c3e1e5d tcg: Adjust TCGLabel for const
72ac21e tcg: Introduce tcg_splitwx_to_{rx,rw}
3e322da tcg: Add in_code_gen_buffer
ccb0c48 tcg: Move tcg epilogue pointer out of TCGContext
09ef808 tcg: Move tcg prologue pointer out of TCGContext
e488e58 tcg: Enhance flush_icache_range with separate data pointer

=== OUTPUT BEGIN ===
1/41 Checking commit e488e58096f9 (tcg: Enhance flush_icache_range with separate data pointer)
2/41 Checking commit 09ef8082ce7f (tcg: Move tcg prologue pointer out of TCGContext)
3/41 Checking commit ccb0c482bf3e (tcg: Move tcg epilogue pointer out of TCGContext)
4/41 Checking commit 3e322da5de89 (tcg: Add in_code_gen_buffer)
5/41 Checking commit 72ac21e27103 (tcg: Introduce tcg_splitwx_to_{rx,rw})
6/41 Checking commit c3e1e5d3a470 (tcg: Adjust TCGLabel for const)
7/41 Checking commit 44975a9cbb02 (tcg: Adjust tcg_out_call for const)
8/41 Checking commit 65e76b95a029 (tcg: Adjust tcg_out_label for const)
9/41 Checking commit eec18a607903 (tcg: Adjust tcg_register_jit for const)
10/41 Checking commit 1d83486ee180 (tcg: Adjust tb_target_set_jmp_target for split-wx)
11/41 Checking commit 043973b272b1 (tcg: Make DisasContextBase.tb const)
12/41 Checking commit b6992b579570 (tcg: Make tb arg to synchronize_from_tb const)
13/41 Checking commit f93ae2267092 (tcg: Use Error with alloc_code_gen_buffer)
14/41 Checking commit caaf645ec574 (tcg: Add --accel tcg,split-wx property)
15/41 Checking commit fed5e19b3c0c (accel/tcg: Support split-wx for linux with memfd)
16/41 Checking commit ef96a10480c7 (accel/tcg: Support split-wx for darwin/iOS with vm_remap)
ERROR: externs should be avoided in .c files
#24: FILE: accel/tcg/translate-all.c:1172:
+extern kern_return_t mach_vm_remap(vm_map_t target_task,

total: 1 errors, 0 warnings, 80 lines checked

Patch 16/41 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.

17/41 Checking commit fe36cad8bf7d (tcg: Return the TB pointer from the rx region from exit_tb)
18/41 Checking commit 6cbd22af41aa (tcg/i386: Support split-wx code generation)
19/41 Checking commit 548fc7975cc4 (tcg/aarch64: Use B not BL for tcg_out_goto_long)
20/41 Checking commit aef71b442d41 (tcg/aarch64: Implement flush_idcache_range manually)
21/41 Checking commit 55b926c57a1b (tcg/aarch64: Support split-wx code generation)
22/41 Checking commit c55a8c0c489d (disas: Push const down through host disasassembly)
23/41 Checking commit 1cab4182fb0e (tcg/tci: Push const down through bytecode reading)
24/41 Checking commit 5f81f0e007d9 (tcg: Introduce tcg_tbrel_diff)
25/41 Checking commit 4c4f647cce7b (tcg/ppc: Use tcg_tbrel_diff)
26/41 Checking commit a254bfdddaf2 (tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB)
27/41 Checking commit 684c281aa228 (tcg/ppc: Support split-wx code generation)
28/41 Checking commit 5a837a3879fe (tcg/sparc: Use tcg_tbrel_diff)
29/41 Checking commit f0fea6377b00 (tcg/sparc: Support split-wx code generation)
30/41 Checking commit 8d88879495b6 (tcg/s390: Use tcg_tbrel_diff)
31/41 Checking commit dff34e06dd91 (tcg/s390: Support split-wx code generation)
32/41 Checking commit 63883fcc4015 (tcg/riscv: Fix branch range checks)
33/41 Checking commit 1c6764de6cda (tcg/riscv: Remove branch-over-branch fallback)
34/41 Checking commit edd72db707ca (tcg/riscv: Support split-wx code generation)
35/41 Checking commit 200ecb3d750f (accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd)
WARNING: architecture specific defines should be avoided
#25: FILE: accel/tcg/translate-all.c:1143:
+#ifdef __mips__

WARNING: architecture specific defines should be avoided
#43: FILE: accel/tcg/translate-all.c:1160:
+#ifdef __mips__

total: 0 errors, 2 warnings, 68 lines checked

Patch 35/41 has style problems, please review.  If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
36/41 Checking commit 7f12d40b3470 (tcg/mips: Do not assert on relocation overflow)
37/41 Checking commit a873c6132612 (tcg/mips: Support split-wx code generation)
38/41 Checking commit 545feb70c9a8 (tcg/arm: Support split-wx code generation)
39/41 Checking commit a009e993d945 (tcg: Remove TCG_TARGET_SUPPORT_MIRROR)
40/41 Checking commit c33649448a69 (tcg: Constify tcg_code_gen_epilogue)
41/41 Checking commit 170f3100faa4 (tcg: Constify TCGLabelQemuLdst.raddr)
=== OUTPUT END ===

Test command exited with code: 1


The full log is available at
http://patchew.org/logs/20201106032921.600200-1-richard.henderson@linaro.org/testing.checkpatch/?type=message.
---
Email generated automatically by Patchew [https://patchew.org/].
Please send your feedback to patchew-devel@redhat.com

^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (41 preceding siblings ...)
  2020-11-06  4:00 ` [PATCH v3 00/41] Mirror map JIT memory for TCG no-reply
@ 2020-11-08  3:38 ` Joelle van Dyne
  2020-11-17  3:47 ` Joelle van Dyne
  43 siblings, 0 replies; 57+ messages in thread
From: Joelle van Dyne @ 2020-11-08  3:38 UTC (permalink / raw)
  To: Richard Henderson; +Cc: QEMU Developers

LGTM, tested and working on iOS.

-j

On Thu, Nov 5, 2020 at 7:29 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> This is my take on Joelle's patch set:
> https://lists.nongnu.org/archive/html/qemu-devel/2020-10/msg07837.html
>
> Changes for v3:
>   * Even more patches -- all tcg backends converted.
>   * Fixups for darwin/ios merged (Joelle).
>   * Feature renamed to splitwx (Paolo).
>
>
> r~
>
>
> Richard Henderson (41):
>   tcg: Enhance flush_icache_range with separate data pointer
>   tcg: Move tcg prologue pointer out of TCGContext
>   tcg: Move tcg epilogue pointer out of TCGContext
>   tcg: Add in_code_gen_buffer
>   tcg: Introduce tcg_splitwx_to_{rx,rw}
>   tcg: Adjust TCGLabel for const
>   tcg: Adjust tcg_out_call for const
>   tcg: Adjust tcg_out_label for const
>   tcg: Adjust tcg_register_jit for const
>   tcg: Adjust tb_target_set_jmp_target for split-wx
>   tcg: Make DisasContextBase.tb const
>   tcg: Make tb arg to synchronize_from_tb const
>   tcg: Use Error with alloc_code_gen_buffer
>   tcg: Add --accel tcg,split-wx property
>   accel/tcg: Support split-wx for linux with memfd
>   accel/tcg: Support split-wx for darwin/iOS with vm_remap
>   tcg: Return the TB pointer from the rx region from exit_tb
>   tcg/i386: Support split-wx code generation
>   tcg/aarch64: Use B not BL for tcg_out_goto_long
>   tcg/aarch64: Implement flush_idcache_range manually
>   tcg/aarch64: Support split-wx code generation
>   disas: Push const down through host disasassembly
>   tcg/tci: Push const down through bytecode reading
>   tcg: Introduce tcg_tbrel_diff
>   tcg/ppc: Use tcg_tbrel_diff
>   tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB
>   tcg/ppc: Support split-wx code generation
>   tcg/sparc: Use tcg_tbrel_diff
>   tcg/sparc: Support split-wx code generation
>   tcg/s390: Use tcg_tbrel_diff
>   tcg/s390: Support split-wx code generation
>   tcg/riscv: Fix branch range checks
>   tcg/riscv: Remove branch-over-branch fallback
>   tcg/riscv: Support split-wx code generation
>   accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd
>   tcg/mips: Do not assert on relocation overflow
>   tcg/mips: Support split-wx code generation
>   tcg/arm: Support split-wx code generation
>   tcg: Remove TCG_TARGET_SUPPORT_MIRROR
>   tcg: Constify tcg_code_gen_epilogue
>   tcg: Constify TCGLabelQemuLdst.raddr
>
>  accel/tcg/tcg-runtime.h      |   2 +-
>  include/disas/dis-asm.h      |   4 +-
>  include/disas/disas.h        |   2 +-
>  include/exec/exec-all.h      |   2 +-
>  include/exec/gen-icount.h    |   4 +-
>  include/exec/log.h           |   2 +-
>  include/exec/translator.h    |   2 +-
>  include/hw/core/cpu.h        |   3 +-
>  include/sysemu/tcg.h         |   3 +-
>  include/tcg/tcg-op.h         |   2 +-
>  include/tcg/tcg.h            |  56 +++++--
>  tcg/aarch64/tcg-target.h     |   8 +-
>  tcg/arm/tcg-target.h         |  10 +-
>  tcg/i386/tcg-target.h        |   9 +-
>  tcg/mips/tcg-target.h        |  10 +-
>  tcg/ppc/tcg-target.h         |   4 +-
>  tcg/riscv/tcg-target.h       |  10 +-
>  tcg/s390/tcg-target.h        |  11 +-
>  tcg/sparc/tcg-target.h       |  10 +-
>  tcg/tci/tcg-target.h         |  11 +-
>  accel/tcg/cpu-exec.c         |  41 +++--
>  accel/tcg/tcg-all.c          |  26 ++-
>  accel/tcg/tcg-runtime.c      |   4 +-
>  accel/tcg/translate-all.c    | 307 +++++++++++++++++++++++++++--------
>  accel/tcg/translator.c       |   4 +-
>  bsd-user/main.c              |   2 +-
>  disas.c                      |   2 +-
>  disas/capstone.c             |   2 +-
>  linux-user/main.c            |   2 +-
>  softmmu/physmem.c            |   9 +-
>  target/arm/cpu.c             |   3 +-
>  target/arm/translate-a64.c   |   2 +-
>  target/avr/cpu.c             |   3 +-
>  target/hppa/cpu.c            |   3 +-
>  target/i386/cpu.c            |   3 +-
>  target/microblaze/cpu.c      |   3 +-
>  target/mips/cpu.c            |   3 +-
>  target/riscv/cpu.c           |   3 +-
>  target/rx/cpu.c              |   3 +-
>  target/sh4/cpu.c             |   3 +-
>  target/sparc/cpu.c           |   3 +-
>  target/tricore/cpu.c         |   2 +-
>  tcg/tcg-op.c                 |  15 +-
>  tcg/tcg.c                    |  86 ++++++++--
>  tcg/tci.c                    |  60 ++++---
>  accel/tcg/trace-events       |   2 +-
>  tcg/aarch64/tcg-target.c.inc | 139 ++++++++++++----
>  tcg/arm/tcg-target.c.inc     |  41 ++---
>  tcg/i386/tcg-target.c.inc    |  36 ++--
>  tcg/mips/tcg-target.c.inc    |  97 +++++------
>  tcg/ppc/tcg-target.c.inc     | 105 ++++++------
>  tcg/riscv/tcg-target.c.inc   | 125 +++++---------
>  tcg/s390/tcg-target.c.inc    |  91 +++++------
>  tcg/sparc/tcg-target.c.inc   |  58 +++----
>  tcg/tcg-ldst.c.inc           |   2 +-
>  tcg/tcg-pool.c.inc           |   6 +-
>  tcg/tci/tcg-target.c.inc     |   2 +-
>  57 files changed, 917 insertions(+), 546 deletions(-)
>
> --
> 2.25.1
>


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
                   ` (42 preceding siblings ...)
  2020-11-08  3:38 ` Joelle van Dyne
@ 2020-11-17  3:47 ` Joelle van Dyne
  2020-11-17 15:20   ` Richard Henderson
  43 siblings, 1 reply; 57+ messages in thread
From: Joelle van Dyne @ 2020-11-17  3:47 UTC (permalink / raw)
  To: Richard Henderson; +Cc: QEMU Developers

Hi, I'm wondering what the progress is for this patch set and the iOS
support one? I know 5.2 is frozen, so will this be considered for 6.0?
Apple Silicon Macs are out now and a few people are asking about QEMU
support :)

-j

On Thu, Nov 5, 2020 at 9:29 PM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> This is my take on Joelle's patch set:
> https://lists.nongnu.org/archive/html/qemu-devel/2020-10/msg07837.html
>
> Changes for v3:
>   * Even more patches -- all tcg backends converted.
>   * Fixups for darwin/ios merged (Joelle).
>   * Feature renamed to splitwx (Paolo).
>
>
> r~
>
>
> Richard Henderson (41):
>   tcg: Enhance flush_icache_range with separate data pointer
>   tcg: Move tcg prologue pointer out of TCGContext
>   tcg: Move tcg epilogue pointer out of TCGContext
>   tcg: Add in_code_gen_buffer
>   tcg: Introduce tcg_splitwx_to_{rx,rw}
>   tcg: Adjust TCGLabel for const
>   tcg: Adjust tcg_out_call for const
>   tcg: Adjust tcg_out_label for const
>   tcg: Adjust tcg_register_jit for const
>   tcg: Adjust tb_target_set_jmp_target for split-wx
>   tcg: Make DisasContextBase.tb const
>   tcg: Make tb arg to synchronize_from_tb const
>   tcg: Use Error with alloc_code_gen_buffer
>   tcg: Add --accel tcg,split-wx property
>   accel/tcg: Support split-wx for linux with memfd
>   accel/tcg: Support split-wx for darwin/iOS with vm_remap
>   tcg: Return the TB pointer from the rx region from exit_tb
>   tcg/i386: Support split-wx code generation
>   tcg/aarch64: Use B not BL for tcg_out_goto_long
>   tcg/aarch64: Implement flush_idcache_range manually
>   tcg/aarch64: Support split-wx code generation
>   disas: Push const down through host disasassembly
>   tcg/tci: Push const down through bytecode reading
>   tcg: Introduce tcg_tbrel_diff
>   tcg/ppc: Use tcg_tbrel_diff
>   tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB
>   tcg/ppc: Support split-wx code generation
>   tcg/sparc: Use tcg_tbrel_diff
>   tcg/sparc: Support split-wx code generation
>   tcg/s390: Use tcg_tbrel_diff
>   tcg/s390: Support split-wx code generation
>   tcg/riscv: Fix branch range checks
>   tcg/riscv: Remove branch-over-branch fallback
>   tcg/riscv: Support split-wx code generation
>   accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd
>   tcg/mips: Do not assert on relocation overflow
>   tcg/mips: Support split-wx code generation
>   tcg/arm: Support split-wx code generation
>   tcg: Remove TCG_TARGET_SUPPORT_MIRROR
>   tcg: Constify tcg_code_gen_epilogue
>   tcg: Constify TCGLabelQemuLdst.raddr
>
>  accel/tcg/tcg-runtime.h      |   2 +-
>  include/disas/dis-asm.h      |   4 +-
>  include/disas/disas.h        |   2 +-
>  include/exec/exec-all.h      |   2 +-
>  include/exec/gen-icount.h    |   4 +-
>  include/exec/log.h           |   2 +-
>  include/exec/translator.h    |   2 +-
>  include/hw/core/cpu.h        |   3 +-
>  include/sysemu/tcg.h         |   3 +-
>  include/tcg/tcg-op.h         |   2 +-
>  include/tcg/tcg.h            |  56 +++++--
>  tcg/aarch64/tcg-target.h     |   8 +-
>  tcg/arm/tcg-target.h         |  10 +-
>  tcg/i386/tcg-target.h        |   9 +-
>  tcg/mips/tcg-target.h        |  10 +-
>  tcg/ppc/tcg-target.h         |   4 +-
>  tcg/riscv/tcg-target.h       |  10 +-
>  tcg/s390/tcg-target.h        |  11 +-
>  tcg/sparc/tcg-target.h       |  10 +-
>  tcg/tci/tcg-target.h         |  11 +-
>  accel/tcg/cpu-exec.c         |  41 +++--
>  accel/tcg/tcg-all.c          |  26 ++-
>  accel/tcg/tcg-runtime.c      |   4 +-
>  accel/tcg/translate-all.c    | 307 +++++++++++++++++++++++++++--------
>  accel/tcg/translator.c       |   4 +-
>  bsd-user/main.c              |   2 +-
>  disas.c                      |   2 +-
>  disas/capstone.c             |   2 +-
>  linux-user/main.c            |   2 +-
>  softmmu/physmem.c            |   9 +-
>  target/arm/cpu.c             |   3 +-
>  target/arm/translate-a64.c   |   2 +-
>  target/avr/cpu.c             |   3 +-
>  target/hppa/cpu.c            |   3 +-
>  target/i386/cpu.c            |   3 +-
>  target/microblaze/cpu.c      |   3 +-
>  target/mips/cpu.c            |   3 +-
>  target/riscv/cpu.c           |   3 +-
>  target/rx/cpu.c              |   3 +-
>  target/sh4/cpu.c             |   3 +-
>  target/sparc/cpu.c           |   3 +-
>  target/tricore/cpu.c         |   2 +-
>  tcg/tcg-op.c                 |  15 +-
>  tcg/tcg.c                    |  86 ++++++++--
>  tcg/tci.c                    |  60 ++++---
>  accel/tcg/trace-events       |   2 +-
>  tcg/aarch64/tcg-target.c.inc | 139 ++++++++++++----
>  tcg/arm/tcg-target.c.inc     |  41 ++---
>  tcg/i386/tcg-target.c.inc    |  36 ++--
>  tcg/mips/tcg-target.c.inc    |  97 +++++------
>  tcg/ppc/tcg-target.c.inc     | 105 ++++++------
>  tcg/riscv/tcg-target.c.inc   | 125 +++++---------
>  tcg/s390/tcg-target.c.inc    |  91 +++++------
>  tcg/sparc/tcg-target.c.inc   |  58 +++----
>  tcg/tcg-ldst.c.inc           |   2 +-
>  tcg/tcg-pool.c.inc           |   6 +-
>  tcg/tci/tcg-target.c.inc     |   2 +-
>  57 files changed, 917 insertions(+), 546 deletions(-)
>
> --
> 2.25.1
>


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-17  3:47 ` Joelle van Dyne
@ 2020-11-17 15:20   ` Richard Henderson
  2020-11-17 15:31     ` Joelle van Dyne
  0 siblings, 1 reply; 57+ messages in thread
From: Richard Henderson @ 2020-11-17 15:20 UTC (permalink / raw)
  To: Joelle van Dyne; +Cc: QEMU Developers

On 11/16/20 7:47 PM, Joelle van Dyne wrote:
> Hi, I'm wondering what the progress is for this patch set and the iOS
> support one? I know 5.2 is frozen, so will this be considered for 6.0?
> Apple Silicon Macs are out now and a few people are asking about QEMU
> support :)

Yes, this will be considered for 6.0.

It does need to be reviewed more completely than a "LGTM", but there's time for
that.


r~


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-17 15:20   ` Richard Henderson
@ 2020-11-17 15:31     ` Joelle van Dyne
  2020-11-17 17:26       ` Alex Bennée
  0 siblings, 1 reply; 57+ messages in thread
From: Joelle van Dyne @ 2020-11-17 15:31 UTC (permalink / raw)
  To: Richard Henderson; +Cc: QEMU Developers

Sorry, are you asking for a review from me? I don’t know if I’m
qualified to review the other patches but I did review the iOS patch.

-j

On Tue, Nov 17, 2020 at 9:20 AM Richard Henderson
<richard.henderson@linaro.org> wrote:
>
> On 11/16/20 7:47 PM, Joelle van Dyne wrote:
> > Hi, I'm wondering what the progress is for this patch set and the iOS
> > support one? I know 5.2 is frozen, so will this be considered for 6.0?
> > Apple Silicon Macs are out now and a few people are asking about QEMU
> > support :)
>
> Yes, this will be considered for 6.0.
>
> It does need to be reviewed more completely than a "LGTM", but there's time for
> that.
>
>
> r~


^ permalink raw reply	[flat|nested] 57+ messages in thread

* Re: [PATCH v3 00/41] Mirror map JIT memory for TCG
  2020-11-17 15:31     ` Joelle van Dyne
@ 2020-11-17 17:26       ` Alex Bennée
  0 siblings, 0 replies; 57+ messages in thread
From: Alex Bennée @ 2020-11-17 17:26 UTC (permalink / raw)
  To: Joelle van Dyne; +Cc: Richard Henderson, qemu-devel


Joelle van Dyne <j@getutm.app> writes:

> Sorry, are you asking for a review from me? I don’t know if I’m
> qualified to review the other patches but I did review the iOS patch.

Anyone can review code and given you wrote the original patches you
certainly know enough about the flow to give some opinion. If things
aren't clear then please do ask questions. The pool of TCG backend
reviewers is small enough and helping out does help.

Failing that you can always send Tested-by: tags once you've tested a
series on the HW.

>
> -j
>
> On Tue, Nov 17, 2020 at 9:20 AM Richard Henderson
> <richard.henderson@linaro.org> wrote:
>>
>> On 11/16/20 7:47 PM, Joelle van Dyne wrote:
>> > Hi, I'm wondering what the progress is for this patch set and the iOS
>> > support one? I know 5.2 is frozen, so will this be considered for 6.0?
>> > Apple Silicon Macs are out now and a few people are asking about QEMU
>> > support :)
>>
>> Yes, this will be considered for 6.0.
>>
>> It does need to be reviewed more completely than a "LGTM", but there's time for
>> that.
>>
>>
>> r~


-- 
Alex Bennée


^ permalink raw reply	[flat|nested] 57+ messages in thread

end of thread, other threads:[~2020-11-17 17:28 UTC | newest]

Thread overview: 57+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-11-06  3:28 [PATCH v3 00/41] Mirror map JIT memory for TCG Richard Henderson
2020-11-06  3:28 ` [PATCH v3 01/41] tcg: Enhance flush_icache_range with separate data pointer Richard Henderson
2020-11-06 20:31   ` Alex Bennée
2020-11-06 22:51     ` Richard Henderson
2020-11-07 21:08       ` Alex Bennée
2020-11-06  3:28 ` [PATCH v3 02/41] tcg: Move tcg prologue pointer out of TCGContext Richard Henderson
2020-11-07 21:10   ` Alex Bennée
2020-11-06  3:28 ` [PATCH v3 03/41] tcg: Move tcg epilogue " Richard Henderson
2020-11-06  3:28 ` [PATCH v3 04/41] tcg: Add in_code_gen_buffer Richard Henderson
2020-11-06  3:28 ` [PATCH v3 05/41] tcg: Introduce tcg_splitwx_to_{rx,rw} Richard Henderson
2020-11-06  3:28 ` [PATCH v3 06/41] tcg: Adjust TCGLabel for const Richard Henderson
2020-11-06  3:28 ` [PATCH v3 07/41] tcg: Adjust tcg_out_call " Richard Henderson
2020-11-06  3:28 ` [PATCH v3 08/41] tcg: Adjust tcg_out_label " Richard Henderson
2020-11-06  3:28 ` [PATCH v3 09/41] tcg: Adjust tcg_register_jit " Richard Henderson
2020-11-06  3:28 ` [PATCH v3 10/41] tcg: Adjust tb_target_set_jmp_target for split-wx Richard Henderson
2020-11-06  3:28 ` [PATCH v3 11/41] tcg: Make DisasContextBase.tb const Richard Henderson
2020-11-06  3:28 ` [PATCH v3 12/41] tcg: Make tb arg to synchronize_from_tb const Richard Henderson
2020-11-06  3:28 ` [PATCH v3 13/41] tcg: Use Error with alloc_code_gen_buffer Richard Henderson
2020-11-06  3:28 ` [PATCH v3 14/41] tcg: Add --accel tcg,split-wx property Richard Henderson
2020-11-06  3:28 ` [PATCH v3 15/41] accel/tcg: Support split-wx for linux with memfd Richard Henderson
2020-11-10 17:03   ` Alex Bennée
2020-11-10 17:26     ` Richard Henderson
2020-11-06  3:28 ` [PATCH v3 16/41] accel/tcg: Support split-wx for darwin/iOS with vm_remap Richard Henderson
2020-11-08  3:37   ` Joelle van Dyne
2020-11-10 17:37   ` Alex Bennée
2020-11-10 17:57     ` Joelle van Dyne
2020-11-06  3:28 ` [PATCH v3 17/41] tcg: Return the TB pointer from the rx region from exit_tb Richard Henderson
2020-11-06  3:28 ` [PATCH v3 18/41] tcg/i386: Support split-wx code generation Richard Henderson
2020-11-06  3:28 ` [PATCH v3 19/41] tcg/aarch64: Use B not BL for tcg_out_goto_long Richard Henderson
2020-11-06  3:29 ` [PATCH v3 20/41] tcg/aarch64: Implement flush_idcache_range manually Richard Henderson
2020-11-06  3:29 ` [PATCH v3 21/41] tcg/aarch64: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 22/41] disas: Push const down through host disasassembly Richard Henderson
2020-11-06  3:29 ` [PATCH v3 23/41] tcg/tci: Push const down through bytecode reading Richard Henderson
2020-11-06  3:29 ` [PATCH v3 24/41] tcg: Introduce tcg_tbrel_diff Richard Henderson
2020-11-06  3:29 ` [PATCH v3 25/41] tcg/ppc: Use tcg_tbrel_diff Richard Henderson
2020-11-06  3:29 ` [PATCH v3 26/41] tcg/ppc: Use tcg_out_mem_long to reset TCG_REG_TB Richard Henderson
2020-11-06  3:29 ` [PATCH v3 27/41] tcg/ppc: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 28/41] tcg/sparc: Use tcg_tbrel_diff Richard Henderson
2020-11-06  3:29 ` [PATCH v3 29/41] tcg/sparc: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 30/41] tcg/s390: Use tcg_tbrel_diff Richard Henderson
2020-11-06  3:29 ` [PATCH v3 31/41] tcg/s390: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 32/41] tcg/riscv: Fix branch range checks Richard Henderson
2020-11-06  3:29 ` [PATCH v3 33/41] tcg/riscv: Remove branch-over-branch fallback Richard Henderson
2020-11-06  3:29 ` [PATCH v3 34/41] tcg/riscv: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 35/41] accel/tcg: Add mips support to alloc_code_gen_buffer_splitwx_memfd Richard Henderson
2020-11-06  3:29 ` [PATCH v3 36/41] tcg/mips: Do not assert on relocation overflow Richard Henderson
2020-11-06  3:29 ` [PATCH v3 37/41] tcg/mips: Support split-wx code generation Richard Henderson
2020-11-06  3:29 ` [PATCH v3 38/41] tcg/arm: " Richard Henderson
2020-11-06  3:29 ` [PATCH v3 39/41] tcg: Remove TCG_TARGET_SUPPORT_MIRROR Richard Henderson
2020-11-06  3:29 ` [PATCH v3 40/41] tcg: Constify tcg_code_gen_epilogue Richard Henderson
2020-11-06  3:29 ` [PATCH v3 41/41] tcg: Constify TCGLabelQemuLdst.raddr Richard Henderson
2020-11-06  4:00 ` [PATCH v3 00/41] Mirror map JIT memory for TCG no-reply
2020-11-08  3:38 ` Joelle van Dyne
2020-11-17  3:47 ` Joelle van Dyne
2020-11-17 15:20   ` Richard Henderson
2020-11-17 15:31     ` Joelle van Dyne
2020-11-17 17:26       ` Alex Bennée

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).