[PATCH 0/5] target/hppa: Clean up conversion from/to MMU index and priviledge level

[PATCH 0/5] target/hppa: Clean up conversion from/to MMU index and priviledge level

[deller]

From: Helge Deller <deller@gmx.de>

Make the conversion between priviledge level and QEMU MMU index
consitent, and afterwards switch to MMU indices 11-15.

Signed-off-by: Helge Deller <deller@gmx.de>

Helge Deller (5):
  target/hppa: Add missing PL1 and PL2 priviledge levels
  target/hppa: Add priviledge to MMU index conversion helpers
  target/hppa: Do not use hardcoded value for tlb_flush_*()
  target/hppa: Use privilege helper in hppa_get_physical_address()
  target/hppa: Switch to use MMU indices 11-15

 target/hppa/cpu.h        | 19 +++++++++++++++----
 target/hppa/helper.c     |  2 +-
 target/hppa/mem_helper.c | 16 ++++++++--------
 target/hppa/translate.c  |  9 +++++----
 4 files changed, 29 insertions(+), 17 deletions(-)

-- 
2.41.0

[PATCH 1/5] target/hppa: Add missing PL1 and PL2 priviledge levels

[deller]

From: Helge Deller <deller@gmx.de>

The hppa CPU has 4 priviledge levels (0-3).
Mention the missing PL1 and PL2 levels, although the Linux kernel
uses only 0 (KERNEL) and 3 (USER). Not sure about HP-UX.

Signed-off-by: Helge Deller <deller@gmx.de>
---
 target/hppa/cpu.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 75c5c0ccf7..6c5b0e67c8 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -31,8 +31,11 @@
 #define TCG_GUEST_DEFAULT_MO        TCG_MO_ALL
 
 #define MMU_KERNEL_IDX   0
+#define MMU_PL1_IDX      1
+#define MMU_PL2_IDX      2
 #define MMU_USER_IDX     3
 #define MMU_PHYS_IDX     4
+
 #define TARGET_INSN_START_EXTRA_WORDS 1
 
 /* Hardware exceptions, interrupts, faults, and traps.  */
-- 
2.41.0

[PATCH 2/5] target/hppa: Add priviledge to MMU index conversion helpers

[deller]

From: Helge Deller <deller@gmx.de>

Add two macros which convert priviledge level to/from MMU index:

- PRIV_TO_MMU_IDX(priv)
    returns the MMU index for the given priviledge level

- MMU_IDX_TO_PRIV(mmu_idx)
    returns the corresponding priviledge level for this MMU index

The introduction of those macros make the code easier to read and
will help to improve performance in follow-up patch.

Signed-off-by: Helge Deller <deller@gmx.de>
---
 target/hppa/cpu.h       | 5 ++++-
 target/hppa/translate.c | 9 +++++----
 2 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 6c5b0e67c8..50b513f0ea 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -36,6 +36,9 @@
 #define MMU_USER_IDX     3
 #define MMU_PHYS_IDX     4
 
+#define PRIV_TO_MMU_IDX(priv)    (priv)
+#define MMU_IDX_TO_PRIV(mmu_idx) (mmu_idx)
+
 #define TARGET_INSN_START_EXTRA_WORDS 1
 
 /* Hardware exceptions, interrupts, faults, and traps.  */
@@ -236,7 +239,7 @@ static inline int cpu_mmu_index(CPUHPPAState *env, bool ifetch)
     return MMU_USER_IDX;
 #else
     if (env->psw & (ifetch ? PSW_C : PSW_D)) {
-        return env->iaoq_f & 3;
+        return PRIV_TO_MMU_IDX(env->iaoq_f & 3);
     }
     return MMU_PHYS_IDX;  /* mmu disabled */
 #endif
diff --git a/target/hppa/translate.c b/target/hppa/translate.c
index d66fcb3e6a..e3af668252 100644
--- a/target/hppa/translate.c
+++ b/target/hppa/translate.c
@@ -4057,14 +4057,15 @@ static void hppa_tr_init_disas_context(DisasContextBase *dcbase, CPUState *cs)
     ctx->tb_flags = ctx->base.tb->flags;
 
 #ifdef CONFIG_USER_ONLY
-    ctx->privilege = MMU_USER_IDX;
+    ctx->privilege = MMU_IDX_TO_PRIV(MMU_USER_IDX);
     ctx->mmu_idx = MMU_USER_IDX;
-    ctx->iaoq_f = ctx->base.pc_first | MMU_USER_IDX;
-    ctx->iaoq_b = ctx->base.tb->cs_base | MMU_USER_IDX;
+    ctx->iaoq_f = ctx->base.pc_first | ctx->privilege;
+    ctx->iaoq_b = ctx->base.tb->cs_base | ctx->privilege;
     ctx->unalign = (ctx->tb_flags & TB_FLAG_UNALIGN ? MO_UNALN : MO_ALIGN);
 #else
     ctx->privilege = (ctx->tb_flags >> TB_FLAG_PRIV_SHIFT) & 3;
-    ctx->mmu_idx = (ctx->tb_flags & PSW_D ? ctx->privilege : MMU_PHYS_IDX);
+    ctx->mmu_idx = (ctx->tb_flags & PSW_D ?
+                    PRIV_TO_MMU_IDX(ctx->privilege) : MMU_PHYS_IDX);
 
     /* Recover the IAOQ values from the GVA + PRIV.  */
     uint64_t cs_base = ctx->base.tb->cs_base;
-- 
2.41.0

[PATCH 3/5] target/hppa: Do not use hardcoded value for tlb_flush_*()

[deller]

From: Helge Deller <deller@gmx.de>

Avoid using hardcoded values when calling the tlb_flush*() functions.
Instead define the correct mask (HPPA_MMU_FLUSH_MASK) and use it.
Skip flushing the MMU for physical addresses.

Signed-off-by: Helge Deller <deller@gmx.de>
---
 target/hppa/cpu.h        | 5 +++++
 target/hppa/helper.c     | 2 +-
 target/hppa/mem_helper.c | 7 +++----
 3 files changed, 9 insertions(+), 5 deletions(-)

diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 50b513f0ea..6623712644 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -41,6 +41,11 @@
 
 #define TARGET_INSN_START_EXTRA_WORDS 1
 
+/* No need to flush MMU_PHYS_IDX  */
+#define HPPA_MMU_FLUSH_MASK                             \
+        (1 << MMU_KERNEL_IDX | 1 << MMU_PL1_IDX |       \
+         1 << MMU_PL2_IDX    | 1 << MMU_USER_IDX)
+
 /* Hardware exceptions, interrupts, faults, and traps.  */
 #define EXCP_HPMC                1  /* high priority machine check */
 #define EXCP_POWER_FAIL          2
diff --git a/target/hppa/helper.c b/target/hppa/helper.c
index 74b8747083..a8d3f456ee 100644
--- a/target/hppa/helper.c
+++ b/target/hppa/helper.c
@@ -71,7 +71,7 @@ void cpu_hppa_put_psw(CPUHPPAState *env, target_ureg psw)
     /* If PSW_P changes, it affects how we translate addresses.  */
     if ((psw ^ old_psw) & PSW_P) {
 #ifndef CONFIG_USER_ONLY
-        tlb_flush_by_mmuidx(env_cpu(env), 0xf);
+        tlb_flush_by_mmuidx(env_cpu(env), HPPA_MMU_FLUSH_MASK);
 #endif
     }
 }
diff --git a/target/hppa/mem_helper.c b/target/hppa/mem_helper.c
index 5046cc8f9d..6f04c101dd 100644
--- a/target/hppa/mem_helper.c
+++ b/target/hppa/mem_helper.c
@@ -50,8 +50,7 @@ static void hppa_flush_tlb_ent(CPUHPPAState *env, hppa_tlb_entry *ent)
     trace_hppa_tlb_flush_ent(env, ent, ent->va_b, ent->va_e, ent->pa);
 
     for (i = 0; i < n; ++i, addr += TARGET_PAGE_SIZE) {
-        /* Do not flush MMU_PHYS_IDX.  */
-        tlb_flush_page_by_mmuidx(cs, addr, 0xf);
+        tlb_flush_page_by_mmuidx(cs, addr, HPPA_MMU_FLUSH_MASK);
     }
 
     memset(ent, 0, sizeof(*ent));
@@ -335,13 +334,13 @@ void HELPER(ptlbe)(CPUHPPAState *env)
 {
     trace_hppa_tlb_ptlbe(env);
     memset(env->tlb, 0, sizeof(env->tlb));
-    tlb_flush_by_mmuidx(env_cpu(env), 0xf);
+    tlb_flush_by_mmuidx(env_cpu(env), HPPA_MMU_FLUSH_MASK);
 }
 
 void cpu_hppa_change_prot_id(CPUHPPAState *env)
 {
     if (env->psw & PSW_P) {
-        tlb_flush_by_mmuidx(env_cpu(env), 0xf);
+        tlb_flush_by_mmuidx(env_cpu(env), HPPA_MMU_FLUSH_MASK);
     }
 }
 
-- 
2.41.0

[PATCH 4/5] target/hppa: Use privilege helper in hppa_get_physical_address()

[deller]

From: Helge Deller <deller@gmx.de>

Convert hppa_get_physical_address() to use the privilege helper macro.

Signed-off-by: Helge Deller <deller@gmx.de>
---
 target/hppa/mem_helper.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/target/hppa/mem_helper.c b/target/hppa/mem_helper.c
index 6f04c101dd..66cf84a0d9 100644
--- a/target/hppa/mem_helper.c
+++ b/target/hppa/mem_helper.c
@@ -73,7 +73,7 @@ int hppa_get_physical_address(CPUHPPAState *env, vaddr addr, int mmu_idx,
                               int type, hwaddr *pphys, int *pprot)
 {
     hwaddr phys;
-    int prot, r_prot, w_prot, x_prot;
+    int prot, r_prot, w_prot, x_prot, mmu_priv;
     hppa_tlb_entry *ent;
     int ret = -1;
 
@@ -97,9 +97,10 @@ int hppa_get_physical_address(CPUHPPAState *env, vaddr addr, int mmu_idx,
     phys = ent->pa + (addr & ~TARGET_PAGE_MASK);
 
     /* Map TLB access_rights field to QEMU protection.  */
-    r_prot = (mmu_idx <= ent->ar_pl1) * PAGE_READ;
-    w_prot = (mmu_idx <= ent->ar_pl2) * PAGE_WRITE;
-    x_prot = (ent->ar_pl2 <= mmu_idx && mmu_idx <= ent->ar_pl1) * PAGE_EXEC;
+    mmu_priv = MMU_IDX_TO_PRIV(mmu_idx);
+    r_prot = (mmu_priv <= ent->ar_pl1) * PAGE_READ;
+    w_prot = (mmu_priv <= ent->ar_pl2) * PAGE_WRITE;
+    x_prot = (ent->ar_pl2 <= mmu_priv && mmu_priv <= ent->ar_pl1) * PAGE_EXEC;
     switch (ent->ar_type) {
     case 0: /* read-only: data page */
         prot = r_prot;
-- 
2.41.0

[PATCH 5/5] target/hppa: Switch to use MMU indices 11-15

[deller]

From: Helge Deller <deller@gmx.de>

The MMU indices 9-15 will use shorter assembler instructions
when run on a x86-64 host. So, switch over to those to get
smaller code and maybe minimally faster emulation.

Signed-off-by: Helge Deller <deller@gmx.de>
---
 target/hppa/cpu.h | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/target/hppa/cpu.h b/target/hppa/cpu.h
index 6623712644..fa13694dab 100644
--- a/target/hppa/cpu.h
+++ b/target/hppa/cpu.h
@@ -30,14 +30,14 @@
    basis.  It's probably easier to fall back to a strong memory model.  */
 #define TCG_GUEST_DEFAULT_MO        TCG_MO_ALL
 
-#define MMU_KERNEL_IDX   0
-#define MMU_PL1_IDX      1
-#define MMU_PL2_IDX      2
-#define MMU_USER_IDX     3
-#define MMU_PHYS_IDX     4
-
-#define PRIV_TO_MMU_IDX(priv)    (priv)
-#define MMU_IDX_TO_PRIV(mmu_idx) (mmu_idx)
+#define MMU_KERNEL_IDX   11
+#define MMU_PL1_IDX      12
+#define MMU_PL2_IDX      13
+#define MMU_USER_IDX     14
+#define MMU_PHYS_IDX     15
+
+#define PRIV_TO_MMU_IDX(priv)    (MMU_KERNEL_IDX + (priv))
+#define MMU_IDX_TO_PRIV(mmu_idx) ((mmu_idx) - MMU_KERNEL_IDX)
 
 #define TARGET_INSN_START_EXTRA_WORDS 1
 
-- 
2.41.0

Re: [PATCH 2/5] target/hppa: Add priviledge to MMU index conversion helpers

[Philippe Mathieu-Daudé]

On 24/8/23 23:04, deller@kernel.org wrote:
> From: Helge Deller <deller@gmx.de>
> 
> Add two macros which convert priviledge level to/from MMU index:
> 
> - PRIV_TO_MMU_IDX(priv)
>      returns the MMU index for the given priviledge level
> 
> - MMU_IDX_TO_PRIV(mmu_idx)
>      returns the corresponding priviledge level for this MMU index
> 
> The introduction of those macros make the code easier to read and
> will help to improve performance in follow-up patch.
> 
> Signed-off-by: Helge Deller <deller@gmx.de>
> ---
>   target/hppa/cpu.h       | 5 ++++-
>   target/hppa/translate.c | 9 +++++----
>   2 files changed, 9 insertions(+), 5 deletions(-)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

Re: [PATCH 1/5] target/hppa: Add missing PL1 and PL2 priviledge levels

[Philippe Mathieu-Daudé]

On 24/8/23 23:04, deller@kernel.org wrote:
> From: Helge Deller <deller@gmx.de>
> 
> The hppa CPU has 4 priviledge levels (0-3).
> Mention the missing PL1 and PL2 levels, although the Linux kernel
> uses only 0 (KERNEL) and 3 (USER). Not sure about HP-UX.
> 
> Signed-off-by: Helge Deller <deller@gmx.de>
> ---
>   target/hppa/cpu.h | 3 +++
>   1 file changed, 3 insertions(+)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

Re: [PATCH 3/5] target/hppa: Do not use hardcoded value for tlb_flush_*()

[Philippe Mathieu-Daudé]

On 24/8/23 23:04, deller@kernel.org wrote:
> From: Helge Deller <deller@gmx.de>
> 
> Avoid using hardcoded values when calling the tlb_flush*() functions.
> Instead define the correct mask (HPPA_MMU_FLUSH_MASK) and use it.
> Skip flushing the MMU for physical addresses.

Alternatively:

   Instead, define and use HPPA_MMU_FLUSH_MASK (keeping the current
   behavior, which doesn't flush the physical address MMU).

> 
> Signed-off-by: Helge Deller <deller@gmx.de>
> ---
>   target/hppa/cpu.h        | 5 +++++
>   target/hppa/helper.c     | 2 +-
>   target/hppa/mem_helper.c | 7 +++----
>   3 files changed, 9 insertions(+), 5 deletions(-)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>

Re: [PATCH 4/5] target/hppa: Use privilege helper in hppa_get_physical_address()

[Philippe Mathieu-Daudé]

On 24/8/23 23:04, deller@kernel.org wrote:
> From: Helge Deller <deller@gmx.de>
> 
> Convert hppa_get_physical_address() to use the privilege helper macro.
> 
> Signed-off-by: Helge Deller <deller@gmx.de>
> ---
>   target/hppa/mem_helper.c | 9 +++++----
>   1 file changed, 5 insertions(+), 4 deletions(-)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>