* [PATCH v4.4 backport 2/3] powerpc/64s: Simple RFI macro conversions
2018-02-22 12:35 ` [PATCH v4.9 backport 1/3] powerpc/64s: Fix conversion of slb_miss_common to use RFI_TO_USER/KERNEL Michael Ellerman
@ 2018-02-22 12:35 ` Michael Ellerman
2018-02-23 12:11 ` Michael Ellerman
2018-02-22 12:35 ` [PATCH v4.4 backport 3/3] powerpc/64s: Improve RFI L1-D cache flush fallback Michael Ellerman
1 sibling, 1 reply; 7+ messages in thread
From: Michael Ellerman @ 2018-02-22 12:35 UTC (permalink / raw)
To: stable, greg; +Cc: npiggin, linuxppc-dev, linux, corsac
From: Nicholas Piggin <npiggin@gmail.com>
commit 222f20f140623ef6033491d0103ee0875fe87d35 upstream.
This commit does simple conversions of rfi/rfid to the new macros that
include the expected destination context. By simple we mean cases
where there is a single well known destination context, and it's
simply a matter of substituting the instruction for the appropriate
macro.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Backport to 4.9, use RFI_TO_KERNEL in idle_book3s.S]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
arch/powerpc/include/asm/exception-64s.h | 2 +-
arch/powerpc/kernel/entry_64.S | 14 +++++++++-----
arch/powerpc/kernel/exceptions-64s.S | 22 +++++++++++-----------
arch/powerpc/kernel/idle_book3s.S | 7 ++++---
arch/powerpc/kvm/book3s_hv_rmhandlers.S | 7 +++----
arch/powerpc/kvm/book3s_rmhandlers.S | 7 +++++--
arch/powerpc/kvm/book3s_segment.S | 4 ++--
7 files changed, 35 insertions(+), 28 deletions(-)
diff --git a/arch/powerpc/include/asm/exception-64s.h b/arch/powerpc/include/asm/exception-64s.h
index cab6d2a46c41..903e76a9f158 100644
--- a/arch/powerpc/include/asm/exception-64s.h
+++ b/arch/powerpc/include/asm/exception-64s.h
@@ -242,7 +242,7 @@ END_FTR_SECTION_NESTED(ftr,ftr,943)
mtspr SPRN_##h##SRR0,r12; \
mfspr r12,SPRN_##h##SRR1; /* and SRR1 */ \
mtspr SPRN_##h##SRR1,r10; \
- h##rfid; \
+ h##RFI_TO_KERNEL; \
b . /* prevent speculative execution */
#define EXCEPTION_PROLOG_PSERIES_1(label, h) \
__EXCEPTION_PROLOG_PSERIES_1(label, h)
diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S
index c33b69d10919..2dc52e6d2af4 100644
--- a/arch/powerpc/kernel/entry_64.S
+++ b/arch/powerpc/kernel/entry_64.S
@@ -39,6 +39,11 @@
#include <asm/tm.h>
#include <asm/ppc-opcode.h>
#include <asm/export.h>
+#ifdef CONFIG_PPC_BOOK3S
+#include <asm/exception-64s.h>
+#else
+#include <asm/exception-64e.h>
+#endif
/*
* System calls.
@@ -396,8 +401,7 @@ tabort_syscall:
mtmsrd r10, 1
mtspr SPRN_SRR0, r11
mtspr SPRN_SRR1, r12
-
- rfid
+ RFI_TO_USER
b . /* prevent speculative execution */
#endif
@@ -1073,7 +1077,7 @@ _GLOBAL(enter_rtas)
mtspr SPRN_SRR0,r5
mtspr SPRN_SRR1,r6
- rfid
+ RFI_TO_KERNEL
b . /* prevent speculative execution */
rtas_return_loc:
@@ -1098,7 +1102,7 @@ rtas_return_loc:
mtspr SPRN_SRR0,r3
mtspr SPRN_SRR1,r4
- rfid
+ RFI_TO_KERNEL
b . /* prevent speculative execution */
.align 3
@@ -1169,7 +1173,7 @@ _GLOBAL(enter_prom)
LOAD_REG_IMMEDIATE(r12, MSR_SF | MSR_ISF | MSR_LE)
andc r11,r11,r12
mtsrr1 r11
- rfid
+ RFI_TO_KERNEL
#endif /* CONFIG_PPC_BOOK3E */
1: /* Return from OF */
diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
index d0be752ea86c..29892500e646 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -244,7 +244,7 @@ BEGIN_FTR_SECTION
LOAD_HANDLER(r12, machine_check_handle_early)
1: mtspr SPRN_SRR0,r12
mtspr SPRN_SRR1,r11
- rfid
+ RFI_TO_KERNEL
b . /* prevent speculative execution */
2:
/* Stack overflow. Stay on emergency stack and panic.
@@ -280,7 +280,7 @@ machine_check_pSeries_0:
mtspr SPRN_SRR0,r12
mfspr r12,SPRN_SRR1
mtspr SPRN_SRR1,r10
- rfid
+ RFI_TO_KERNEL
b . /* prevent speculative execution */
TRAMP_KVM_SKIP(PACA_EXMC, 0x200)
@@ -446,7 +446,7 @@ EXC_COMMON_BEGIN(machine_check_handle_early)
li r3,MSR_ME
andc r10,r10,r3 /* Turn off MSR_ME */
mtspr SPRN_SRR1,r10
- rfid
+ RFI_TO_KERNEL
b .
2:
/*
@@ -464,7 +464,7 @@ EXC_COMMON_BEGIN(machine_check_handle_early)
*/
bl machine_check_queue_event
MACHINE_CHECK_HANDLER_WINDUP
- rfid
+ RFI_TO_USER_OR_KERNEL
9:
/* Deliver the machine check to host kernel in V mode. */
MACHINE_CHECK_HANDLER_WINDUP
@@ -706,7 +706,7 @@ END_MMU_FTR_SECTION_IFCLR(MMU_FTR_TYPE_RADIX)
mtspr SPRN_SRR0,r10
ld r10,PACAKMSR(r13)
mtspr SPRN_SRR1,r10
- rfid
+ RFI_TO_KERNEL
b .
EXC_COMMON_BEGIN(unrecov_slb)
@@ -893,7 +893,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE) \
mtspr SPRN_SRR0,r10 ; \
ld r10,PACAKMSR(r13) ; \
mtspr SPRN_SRR1,r10 ; \
- rfid ; \
+ RFI_TO_KERNEL ; \
b . ; /* prevent speculative execution */
#define SYSCALL_PSERIES_3 \
@@ -901,7 +901,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_REAL_LE) \
1: mfspr r12,SPRN_SRR1 ; \
xori r12,r12,MSR_LE ; \
mtspr SPRN_SRR1,r12 ; \
- rfid ; /* return to userspace */ \
+ RFI_TO_USER ; /* return to userspace */ \
b . ; /* prevent speculative execution */
#if defined(CONFIG_RELOCATABLE)
@@ -1276,7 +1276,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_CFAR)
ld r11,PACA_EXGEN+EX_R11(r13)
ld r12,PACA_EXGEN+EX_R12(r13)
ld r13,PACA_EXGEN+EX_R13(r13)
- HRFID
+ HRFI_TO_UNKNOWN
b .
#endif
@@ -1350,7 +1350,7 @@ masked_##_H##interrupt: \
ld r10,PACA_EXGEN+EX_R10(r13); \
ld r11,PACA_EXGEN+EX_R11(r13); \
GET_SCRATCH0(r13); \
- ##_H##rfid; \
+ ##_H##RFI_TO_KERNEL; \
b .
/*
@@ -1372,7 +1372,7 @@ TRAMP_REAL_BEGIN(kvmppc_skip_interrupt)
addi r13, r13, 4
mtspr SPRN_SRR0, r13
GET_SCRATCH0(r13)
- rfid
+ RFI_TO_KERNEL
b .
TRAMP_REAL_BEGIN(kvmppc_skip_Hinterrupt)
@@ -1384,7 +1384,7 @@ TRAMP_REAL_BEGIN(kvmppc_skip_Hinterrupt)
addi r13, r13, 4
mtspr SPRN_HSRR0, r13
GET_SCRATCH0(r13)
- hrfid
+ HRFI_TO_KERNEL
b .
#endif
diff --git a/arch/powerpc/kernel/idle_book3s.S b/arch/powerpc/kernel/idle_book3s.S
index b350ac5e3111..d92c95333435 100644
--- a/arch/powerpc/kernel/idle_book3s.S
+++ b/arch/powerpc/kernel/idle_book3s.S
@@ -9,6 +9,7 @@
*/
#include <linux/threads.h>
+#include <asm/exception-64s.h>
#include <asm/processor.h>
#include <asm/page.h>
#include <asm/cputable.h>
@@ -178,7 +179,7 @@ _GLOBAL(pnv_powersave_common)
mtmsrd r6, 1 /* clear RI before setting SRR0/1 */
mtspr SPRN_SRR0, r5
mtspr SPRN_SRR1, r7
- rfid
+ RFI_TO_KERNEL
.globl pnv_enter_arch207_idle_mode
pnv_enter_arch207_idle_mode:
@@ -668,7 +669,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_HVMODE)
mtcr r6
mtspr SPRN_SRR1,r4
mtspr SPRN_SRR0,r5
- rfid
+ RFI_TO_KERNEL
/*
* R3 here contains the value that will be returned to the caller
@@ -689,4 +690,4 @@ END_FTR_SECTION_IFSET(CPU_FTR_HVMODE)
mtcr r6
mtspr SPRN_SRR1,r4
mtspr SPRN_SRR0,r5
- rfid
+ RFI_TO_KERNEL
diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index 0447a22a4df6..55fbc0c78721 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
@@ -65,7 +65,7 @@ _GLOBAL_TOC(kvmppc_hv_entry_trampoline)
mtmsrd r0,1 /* clear RI in MSR */
mtsrr0 r5
mtsrr1 r6
- RFI
+ RFI_TO_KERNEL
kvmppc_call_hv_entry:
ld r4, HSTATE_KVM_VCPU(r13)
@@ -171,7 +171,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
mtsrr0 r8
mtsrr1 r7
beq cr1, 13f /* machine check */
- RFI
+ RFI_TO_KERNEL
/* On POWER7, we have external interrupts set to use HSRR0/1 */
11: mtspr SPRN_HSRR0, r8
@@ -1018,8 +1018,7 @@ BEGIN_FTR_SECTION
END_FTR_SECTION_IFSET(CPU_FTR_HAS_PPR)
ld r0, VCPU_GPR(R0)(r4)
ld r4, VCPU_GPR(R4)(r4)
-
- hrfid
+ HRFI_TO_GUEST
b .
secondary_too_late:
diff --git a/arch/powerpc/kvm/book3s_rmhandlers.S b/arch/powerpc/kvm/book3s_rmhandlers.S
index 42a4b237df5f..34a5adeff084 100644
--- a/arch/powerpc/kvm/book3s_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_rmhandlers.S
@@ -46,6 +46,9 @@
#define FUNC(name) name
+#define RFI_TO_KERNEL RFI
+#define RFI_TO_GUEST RFI
+
.macro INTERRUPT_TRAMPOLINE intno
.global kvmppc_trampoline_\intno
@@ -141,7 +144,7 @@ kvmppc_handler_skip_ins:
GET_SCRATCH0(r13)
/* And get back into the code */
- RFI
+ RFI_TO_KERNEL
#endif
/*
@@ -164,6 +167,6 @@ _GLOBAL_TOC(kvmppc_entry_trampoline)
ori r5, r5, MSR_EE
mtsrr0 r7
mtsrr1 r6
- RFI
+ RFI_TO_KERNEL
#include "book3s_segment.S"
diff --git a/arch/powerpc/kvm/book3s_segment.S b/arch/powerpc/kvm/book3s_segment.S
index ca8f174289bb..7c982956d709 100644
--- a/arch/powerpc/kvm/book3s_segment.S
+++ b/arch/powerpc/kvm/book3s_segment.S
@@ -156,7 +156,7 @@ no_dcbz32_on:
PPC_LL r9, SVCPU_R9(r3)
PPC_LL r3, (SVCPU_R3)(r3)
- RFI
+ RFI_TO_GUEST
kvmppc_handler_trampoline_enter_end:
@@ -389,5 +389,5 @@ END_FTR_SECTION_IFSET(CPU_FTR_HVMODE)
cmpwi r12, BOOK3S_INTERRUPT_DOORBELL
beqa BOOK3S_INTERRUPT_DOORBELL
- RFI
+ RFI_TO_KERNEL
kvmppc_handler_trampoline_exit_end:
--
2.14.1
^ permalink raw reply related [flat|nested] 7+ messages in thread
* [PATCH v4.4 backport 3/3] powerpc/64s: Improve RFI L1-D cache flush fallback
2018-02-22 12:35 ` [PATCH v4.9 backport 1/3] powerpc/64s: Fix conversion of slb_miss_common to use RFI_TO_USER/KERNEL Michael Ellerman
2018-02-22 12:35 ` [PATCH v4.4 backport 2/3] powerpc/64s: Simple RFI macro conversions Michael Ellerman
@ 2018-02-22 12:35 ` Michael Ellerman
2018-02-22 14:26 ` Nicholas Piggin
1 sibling, 1 reply; 7+ messages in thread
From: Michael Ellerman @ 2018-02-22 12:35 UTC (permalink / raw)
To: stable, greg; +Cc: npiggin, linuxppc-dev, linux, corsac
From: Nicholas Piggin <npiggin@gmail.com>
commit bdcb1aefc5b3f7d0f1dc8b02673602bca2ff7a4b upstream.
The fallback RFI flush is used when firmware does not provide a way
to flush the cache. It's a "displacement flush" that evicts useful
data by displacing it with an uninteresting buffer.
The flush has to take care to work with implementation specific cache
replacment policies, so the recipe has been in flux. The initial
slow but conservative approach is to touch all lines of a congruence
class, with dependencies between each load. It has since been
determined that a linear pattern of loads without dependencies is
sufficient, and is significantly faster.
Measuring the speed of a null syscall with RFI fallback flush enabled
gives the relative improvement:
P8 - 1.83x
P9 - 1.75x
The flush also becomes simpler and more adaptable to different cache
geometries.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Backport to 4.9]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
---
arch/powerpc/include/asm/paca.h | 3 +-
arch/powerpc/kernel/asm-offsets.c | 3 +-
arch/powerpc/kernel/exceptions-64s.S | 76 +++++++++++++++++-------------------
arch/powerpc/kernel/setup_64.c | 13 +-----
4 files changed, 39 insertions(+), 56 deletions(-)
diff --git a/arch/powerpc/include/asm/paca.h b/arch/powerpc/include/asm/paca.h
index ea43897183fd..c75ee2d886fc 100644
--- a/arch/powerpc/include/asm/paca.h
+++ b/arch/powerpc/include/asm/paca.h
@@ -212,8 +212,7 @@ struct paca_struct {
*/
u64 exrfi[13] __aligned(0x80);
void *rfi_flush_fallback_area;
- u64 l1d_flush_congruence;
- u64 l1d_flush_sets;
+ u64 l1d_flush_size;
#endif
};
diff --git a/arch/powerpc/kernel/asm-offsets.c b/arch/powerpc/kernel/asm-offsets.c
index 64bcbd580495..14fbbd9035ca 100644
--- a/arch/powerpc/kernel/asm-offsets.c
+++ b/arch/powerpc/kernel/asm-offsets.c
@@ -242,8 +242,7 @@ int main(void)
DEFINE(PACA_IN_MCE, offsetof(struct paca_struct, in_mce));
DEFINE(PACA_RFI_FLUSH_FALLBACK_AREA, offsetof(struct paca_struct, rfi_flush_fallback_area));
DEFINE(PACA_EXRFI, offsetof(struct paca_struct, exrfi));
- DEFINE(PACA_L1D_FLUSH_CONGRUENCE, offsetof(struct paca_struct, l1d_flush_congruence));
- DEFINE(PACA_L1D_FLUSH_SETS, offsetof(struct paca_struct, l1d_flush_sets));
+ DEFINE(PACA_L1D_FLUSH_SIZE, offsetof(struct paca_struct, l1d_flush_size));
#endif
DEFINE(PACAHWCPUID, offsetof(struct paca_struct, hw_cpu_id));
DEFINE(PACAKEXECSTATE, offsetof(struct paca_struct, kexec_state));
diff --git a/arch/powerpc/kernel/exceptions-64s.S b/arch/powerpc/kernel/exceptions-64s.S
index 29892500e646..7614d1dd2c0b 100644
--- a/arch/powerpc/kernel/exceptions-64s.S
+++ b/arch/powerpc/kernel/exceptions-64s.S
@@ -1602,39 +1602,37 @@ rfi_flush_fallback:
std r9,PACA_EXRFI+EX_R9(r13)
std r10,PACA_EXRFI+EX_R10(r13)
std r11,PACA_EXRFI+EX_R11(r13)
- std r12,PACA_EXRFI+EX_R12(r13)
- std r8,PACA_EXRFI+EX_R13(r13)
mfctr r9
ld r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
- ld r11,PACA_L1D_FLUSH_SETS(r13)
- ld r12,PACA_L1D_FLUSH_CONGRUENCE(r13)
- /*
- * The load adresses are at staggered offsets within cachelines,
- * which suits some pipelines better (on others it should not
- * hurt).
- */
- addi r12,r12,8
+ ld r11,PACA_L1D_FLUSH_SIZE(r13)
+ srdi r11,r11,(7 + 3) /* 128 byte lines, unrolled 8x */
mtctr r11
DCBT_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
/* order ld/st prior to dcbt stop all streams with flushing */
sync
-1: li r8,0
- .rept 8 /* 8-way set associative */
- ldx r11,r10,r8
- add r8,r8,r12
- xor r11,r11,r11 // Ensure r11 is 0 even if fallback area is not
- add r8,r8,r11 // Add 0, this creates a dependency on the ldx
- .endr
- addi r10,r10,128 /* 128 byte cache line */
+
+ /*
+ * The load adresses are at staggered offsets within cachelines,
+ * which suits some pipelines better (on others it should not
+ * hurt).
+ */
+1:
+ ld r11,(0x80 + 8)*0(r10)
+ ld r11,(0x80 + 8)*1(r10)
+ ld r11,(0x80 + 8)*2(r10)
+ ld r11,(0x80 + 8)*3(r10)
+ ld r11,(0x80 + 8)*4(r10)
+ ld r11,(0x80 + 8)*5(r10)
+ ld r11,(0x80 + 8)*6(r10)
+ ld r11,(0x80 + 8)*7(r10)
+ addi r10,r10,0x80*8
bdnz 1b
mtctr r9
ld r9,PACA_EXRFI+EX_R9(r13)
ld r10,PACA_EXRFI+EX_R10(r13)
ld r11,PACA_EXRFI+EX_R11(r13)
- ld r12,PACA_EXRFI+EX_R12(r13)
- ld r8,PACA_EXRFI+EX_R13(r13)
GET_SCRATCH0(r13);
rfid
@@ -1645,39 +1643,37 @@ hrfi_flush_fallback:
std r9,PACA_EXRFI+EX_R9(r13)
std r10,PACA_EXRFI+EX_R10(r13)
std r11,PACA_EXRFI+EX_R11(r13)
- std r12,PACA_EXRFI+EX_R12(r13)
- std r8,PACA_EXRFI+EX_R13(r13)
mfctr r9
ld r10,PACA_RFI_FLUSH_FALLBACK_AREA(r13)
- ld r11,PACA_L1D_FLUSH_SETS(r13)
- ld r12,PACA_L1D_FLUSH_CONGRUENCE(r13)
- /*
- * The load adresses are at staggered offsets within cachelines,
- * which suits some pipelines better (on others it should not
- * hurt).
- */
- addi r12,r12,8
+ ld r11,PACA_L1D_FLUSH_SIZE(r13)
+ srdi r11,r11,(7 + 3) /* 128 byte lines, unrolled 8x */
mtctr r11
DCBT_STOP_ALL_STREAM_IDS(r11) /* Stop prefetch streams */
/* order ld/st prior to dcbt stop all streams with flushing */
sync
-1: li r8,0
- .rept 8 /* 8-way set associative */
- ldx r11,r10,r8
- add r8,r8,r12
- xor r11,r11,r11 // Ensure r11 is 0 even if fallback area is not
- add r8,r8,r11 // Add 0, this creates a dependency on the ldx
- .endr
- addi r10,r10,128 /* 128 byte cache line */
+
+ /*
+ * The load adresses are at staggered offsets within cachelines,
+ * which suits some pipelines better (on others it should not
+ * hurt).
+ */
+1:
+ ld r11,(0x80 + 8)*0(r10)
+ ld r11,(0x80 + 8)*1(r10)
+ ld r11,(0x80 + 8)*2(r10)
+ ld r11,(0x80 + 8)*3(r10)
+ ld r11,(0x80 + 8)*4(r10)
+ ld r11,(0x80 + 8)*5(r10)
+ ld r11,(0x80 + 8)*6(r10)
+ ld r11,(0x80 + 8)*7(r10)
+ addi r10,r10,0x80*8
bdnz 1b
mtctr r9
ld r9,PACA_EXRFI+EX_R9(r13)
ld r10,PACA_EXRFI+EX_R10(r13)
ld r11,PACA_EXRFI+EX_R11(r13)
- ld r12,PACA_EXRFI+EX_R12(r13)
- ld r8,PACA_EXRFI+EX_R13(r13)
GET_SCRATCH0(r13);
hrfid
diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c
index 7c30a91c1f86..5243501d95ef 100644
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -745,19 +745,8 @@ static void init_fallback_flush(void)
memset(l1d_flush_fallback_area, 0, l1d_size * 2);
for_each_possible_cpu(cpu) {
- /*
- * The fallback flush is currently coded for 8-way
- * associativity. Different associativity is possible, but it
- * will be treated as 8-way and may not evict the lines as
- * effectively.
- *
- * 128 byte lines are mandatory.
- */
- u64 c = l1d_size / 8;
-
paca[cpu].rfi_flush_fallback_area = l1d_flush_fallback_area;
- paca[cpu].l1d_flush_congruence = c;
- paca[cpu].l1d_flush_sets = c / 128;
+ paca[cpu].l1d_flush_size = l1d_size;
}
}
--
2.14.1
^ permalink raw reply related [flat|nested] 7+ messages in thread