Re: [PATCH v3] perf/x86/intel/uncore: Fix CAS_COUNT_WRITE issue for ICX

From: Xing Zhengjun
Date: Wed Dec 22 2021 - 20:33:28 EST




On 12/23/2021 12:30 AM, Liang, Kan wrote:


On 11/18/2021 11:02 AM, zhengjun.xing@xxxxxxxxxxxxxxx wrote:
From: Zhengjun Xing <zhengjun.xing@xxxxxxxxxxxxxxx>

The user recently report a perf issue in the ICX platform, when test by

If you have the user's name, you may want to add a Reported-by tag to give them credit. If you don't have, it doesn't matter either.

Unfortunately, I haven't the exactly user's name.

perf event “uncore_imc_x/cas_count_write”,the write bandwidth is always
very small (only 0.38MB/s), it is caused by the wrong "umask" for the
"cas_count_write" event. When double-checking, find "cas_count_read"
also is wrong.

The public document for ICX uncore:

https://www.intel.com/content/www/us/en/develop/download/3rd-gen-intel-xeon-processor-scalable-uncore-pm.html

On page 142, Table 2-143, defines Unit Masks for CAS_COUNT:
RD b00001111
WR b00110000


I think we usually want a permanent reference in the change log. The document may be updated later. The page number or the table number may not be accurate anymore.

I guess you may want to give the exact document name and the version number here. So people can still easily locate the information several years later.
E.g., "3rd Gen Intel® Xeon® Processor Scalable Family, Codename Ice Lake, Uncore Performance Monitoring Reference Manual, Revision 1.00, May 2021"


Thanks, I will update it in the new version patch.

So Corrected both "cas_count_read" and "cas_count_write" for ICX.

Old settings:
  hswep_uncore_imc_events
    INTEL_UNCORE_EVENT_DESC(cas_count_read,  "event=0x04,umask=0x03")
      INTEL_UNCORE_EVENT_DESC(cas_count_write, "event=0x04,umask=0x0c")

New settings:
  snr_uncore_imc_events
    INTEL_UNCORE_EVENT_DESC(cas_count_read,  "event=0x04,umask=0x0f")
    INTEL_UNCORE_EVENT_DESC(cas_count_write, "event=0x04,umask=0x30"),

Fixes: 2b3b76b5ec67 ("perf/x86/intel/uncore: Add Ice Lake server uncore support")
Reviewed-by: Adrian Hunter <adrian.hunter@xxxxxxxxx>

Other than the above comments, the patch looks good to me.

Reviewed-by: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>


Thanks,
Kan

Signed-off-by: Zhengjun Xing <zhengjun.xing@xxxxxxxxxxxxxxx>
Cc: stable@xxxxxxxxxxxxxxx
---
Change log:

   v3:
     * Add change log

   v2:
     * Add stable tag

  arch/x86/events/intel/uncore_snbep.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/events/intel/uncore_snbep.c b/arch/x86/events/intel/uncore_snbep.c
index 5ddc0f30db6f..a6fd8eb410a9 100644
--- a/arch/x86/events/intel/uncore_snbep.c
+++ b/arch/x86/events/intel/uncore_snbep.c
@@ -5468,7 +5468,7 @@ static struct intel_uncore_type icx_uncore_imc = {
      .fixed_ctr_bits    = 48,
      .fixed_ctr    = SNR_IMC_MMIO_PMON_FIXED_CTR,
      .fixed_ctl    = SNR_IMC_MMIO_PMON_FIXED_CTL,
-    .event_descs    = hswep_uncore_imc_events,
+    .event_descs    = snr_uncore_imc_events,
      .perf_ctr    = SNR_IMC_MMIO_PMON_CTR0,
      .event_ctl    = SNR_IMC_MMIO_PMON_CTL0,
      .event_mask    = SNBEP_PMON_RAW_EVENT_MASK,

--
Zhengjun Xing