sugar/linux
1
0
Fork 0
Code Activity

tools/power turbostat: Avoid probing the same perf counters

Browse Source 
For the RAPL package energy status counter, Intel and AMD share the same
perf_subsys and perf_name, but with different MSR addresses.

Both rapl_counter_arch_infos[0] and rapl_counter_arch_infos[1] are
introduced to describe this counter for different Vendors.

As a result, the perf counter is probed twice, and causes a failure in
in get_rapl_counters() because expected_read_size and actual_read_size
don't match.

Fix the problem by skipping the already probed counter.

Note, this is not a perfect fix. For example, if different
vendors/platforms use the same MSR value for different purpose, the code
can be fooled when it probes a rapl_counter_arch_infos[] entry that does
not belong to the running Vendor/Platform.

In a long run, better to put rapl_counter_arch_infos[] into the
platform_features so that this becomes Vendor/Platform specific.

Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This commit is contained in:
Zhang Rui
2025-05-30 08:09:28 +08:00
committed by Len Brown
parent ff3d019e98
commit 69078520fd
1 changed files with 15 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
tools/power/x86/turbostat/turbostat.c
View File

@@ -7991,6 +7991,21 @@ void rapl_perf_init(void)
struct rapl_counter_info_t *rci = &rapl_counter_info_perdomain[next_domain];
/*
* rapl_counter_arch_infos[] can have multiple entries describing the same
* counter, due to the difference from different platforms/Vendors.
* E.g. rapl_counter_arch_infos[0] and rapl_counter_arch_infos[1] share the
* same perf_subsys and perf_name, but with different MSR address.
* rapl_counter_arch_infos[0] is for Intel and rapl_counter_arch_infos[1]
* is for AMD.
* In this case, it is possible that multiple rapl_counter_arch_infos[]
* entries are probed just because their perf/msr is duplicate and valid.
*
* Thus need a check to avoid re-probe the same counters.
*/
if (rci->source[cai->rci_index] != COUNTER_SOURCE_NONE)
break;
/* Use perf API for this counter */
if (add_rapl_perf_counter(cpu, rci, cai, &scale, &unit) != -1) {
rci->source[cai->rci_index] = COUNTER_SOURCE_PERF;