From: Anjali K <anjalik@linux.ibm.com>
To: mpe@ellerman.id.au, linuxppc-dev@lists.ozlabs.org
Cc: anjalik@linux.ibm.com, kjain@linux.ibm.com,
atrajeev@linux.vnet.ibm.com, maddy@linux.ibm.com
Subject: [PATCH v2] powerpc/perf: Set cpumode flags using sample address
Date: Fri, 17 May 2024 15:16:07 +0530 [thread overview]
Message-ID: <20240517094607.422166-1-anjalik@linux.ibm.com> (raw)
Currently in some cases, when the sampled instruction address register
latches to a specific address during sampling, there is an inconsistency
in the privilege bits captured in the sampled event register.
For example, a snippet from the perf report on a power10 system is:
Overhead Address Command Shared Object Symbol
........ .................. ............ ................. .......................
2.41% 0x7fff9f94a02c null_syscall [unknown] [k] 0x00007fff9f94a02c
2.20% 0x7fff9f94a02c null_syscall libc.so.6 [.] syscall
perf_get_misc_flags() function looks at the privilege bits to return
the corresponding flags to be used for the address symbol and these
privilege bit details are read from the sampled event register. In the
above snippet, address "0x00007fff9f94a02c" is shown as "k" (kernel) due
to the inconsistent privilege bits captured in the sampled event register.
To address this case, the proposed fix is to additionally check whether the
sampled address is in the kernel area. Since this is specific to the latest
platform, a new pmu flag is added called "PPMU_P10" and is used to
contain the proposed fix.
Signed-off-by: Anjali K <anjalik@linux.ibm.com>
---
Changelog:
V1->V2:
Fixed the build warning reported by the kernel test bot
Added a new flag PPMU_P10 and used it instead of PPMU_ARCH_31 to restrict
the changes to the current platform (Power10)
arch/powerpc/include/asm/perf_event_server.h | 1 +
arch/powerpc/perf/core-book3s.c | 43 ++++++++------------
arch/powerpc/perf/power10-pmu.c | 3 +-
3 files changed, 20 insertions(+), 27 deletions(-)
diff --git a/arch/powerpc/include/asm/perf_event_server.h b/arch/powerpc/include/asm/perf_event_server.h
index e2221d29fdf9..12f7bfb4cab1 100644
--- a/arch/powerpc/include/asm/perf_event_server.h
+++ b/arch/powerpc/include/asm/perf_event_server.h
@@ -90,6 +90,7 @@ struct power_pmu {
#define PPMU_ARCH_31 0x00000200 /* Has MMCR3, SIER2 and SIER3 */
#define PPMU_P10_DD1 0x00000400 /* Is power10 DD1 processor version */
#define PPMU_HAS_ATTR_CONFIG1 0x00000800 /* Using config1 attribute */
+#define PPMU_P10 0x00001000 /* For power10 pmu */
/*
* Values for flags to get_alternatives()
diff --git a/arch/powerpc/perf/core-book3s.c b/arch/powerpc/perf/core-book3s.c
index 6b5f8a94e7d8..8a2677463a73 100644
--- a/arch/powerpc/perf/core-book3s.c
+++ b/arch/powerpc/perf/core-book3s.c
@@ -266,31 +266,12 @@ static inline u32 perf_flags_from_msr(struct pt_regs *regs)
static inline u32 perf_get_misc_flags(struct pt_regs *regs)
{
bool use_siar = regs_use_siar(regs);
- unsigned long mmcra = regs->dsisr;
- int marked = mmcra & MMCRA_SAMPLE_ENABLE;
+ unsigned long siar = mfspr(SPRN_SIAR);
+ unsigned long addr;
if (!use_siar)
return perf_flags_from_msr(regs);
- /*
- * Check the address in SIAR to identify the
- * privilege levels since the SIER[MSR_HV, MSR_PR]
- * bits are not set for marked events in power10
- * DD1.
- */
- if (marked && (ppmu->flags & PPMU_P10_DD1)) {
- unsigned long siar = mfspr(SPRN_SIAR);
- if (siar) {
- if (is_kernel_addr(siar))
- return PERF_RECORD_MISC_KERNEL;
- return PERF_RECORD_MISC_USER;
- } else {
- if (is_kernel_addr(regs->nip))
- return PERF_RECORD_MISC_KERNEL;
- return PERF_RECORD_MISC_USER;
- }
- }
-
/*
* If we don't have flags in MMCRA, rather than using
* the MSR, we intuit the flags from the address in
@@ -298,19 +279,29 @@ static inline u32 perf_get_misc_flags(struct pt_regs *regs)
* results
*/
if (ppmu->flags & PPMU_NO_SIPR) {
- unsigned long siar = mfspr(SPRN_SIAR);
if (is_kernel_addr(siar))
return PERF_RECORD_MISC_KERNEL;
return PERF_RECORD_MISC_USER;
}
/* PR has priority over HV, so order below is important */
- if (regs_sipr(regs))
- return PERF_RECORD_MISC_USER;
-
- if (regs_sihv(regs) && (freeze_events_kernel != MMCR0_FCHV))
+ if (regs_sipr(regs)) {
+ if (!(ppmu->flags & PPMU_P10))
+ return PERF_RECORD_MISC_USER;
+ } else if (regs_sihv(regs) && (freeze_events_kernel != MMCR0_FCHV))
return PERF_RECORD_MISC_HYPERVISOR;
+ /*
+ * Check the address in SIAR to identify the
+ * privilege levels since the SIER[MSR_HV, MSR_PR]
+ * bits are not set correctly in power10 sometimes
+ */
+ if (ppmu->flags & PPMU_P10) {
+ addr = siar ? siar : regs->nip;
+ if (!is_kernel_addr(addr))
+ return PERF_RECORD_MISC_USER;
+ }
+
return PERF_RECORD_MISC_KERNEL;
}
diff --git a/arch/powerpc/perf/power10-pmu.c b/arch/powerpc/perf/power10-pmu.c
index 62a68b6b2d4b..bb57b7cfe640 100644
--- a/arch/powerpc/perf/power10-pmu.c
+++ b/arch/powerpc/perf/power10-pmu.c
@@ -593,7 +593,8 @@ static struct power_pmu power10_pmu = {
.get_mem_weight = isa207_get_mem_weight,
.disable_pmc = isa207_disable_pmc,
.flags = PPMU_HAS_SIER | PPMU_ARCH_207S |
- PPMU_ARCH_31 | PPMU_HAS_ATTR_CONFIG1,
+ PPMU_ARCH_31 | PPMU_HAS_ATTR_CONFIG1 |
+ PPMU_P10,
.n_generic = ARRAY_SIZE(power10_generic_events),
.generic_events = power10_generic_events,
.cache_events = &power10_cache_events,
base-commit: dd5a440a31fae6e459c0d6271dddd62825505361
--
2.39.3
next reply other threads:[~2024-05-17 9:47 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2024-05-17 9:46 Anjali K [this message]
2024-05-24 7:25 ` [PATCH v2] powerpc/perf: Set cpumode flags using sample address Michael Ellerman
2024-05-27 9:06 ` Anjali K
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20240517094607.422166-1-anjalik@linux.ibm.com \
--to=anjalik@linux.ibm.com \
--cc=atrajeev@linux.vnet.ibm.com \
--cc=kjain@linux.ibm.com \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=maddy@linux.ibm.com \
--cc=mpe@ellerman.id.au \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.