mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Swapnil Sapkal <swapnil.sapkal@amd.com>
To: Namhyung Kim <namhyung@kernel.org>
Cc: <peterz@infradead.org>, <mingo@redhat.com>, <acme@kernel.org>,
	<irogers@google.com>, <james.clark@arm.com>,
	<ravi.bangoria@amd.com>, <yu.c.chen@intel.com>,
	<mark.rutland@arm.com>, <alexander.shishkin@linux.intel.com>,
	<jolsa@kernel.org>, <rostedt@goodmis.org>,
	<vincent.guittot@linaro.org>, <adrian.hunter@intel.com>,
	<kan.liang@linux.intel.com>, <gautham.shenoy@amd.com>,
	<kprateek.nayak@amd.com>, <juri.lelli@redhat.com>,
	<yangjihong@bytedance.com>, <void@manifault.com>, <tj@kernel.org>,
	<sshegde@linux.ibm.com>, <ctshao@google.com>,
	<quic_zhonhan@quicinc.com>, <thomas.falcon@intel.com>,
	<blakejones@google.com>, <ashelat@redhat.com>, <leo.yan@arm.com>,
	<dvyukov@google.com>, <ak@linux.intel.com>, <yujie.liu@intel.com>,
	<graham.woodward@arm.com>, <ben.gainey@arm.com>,
	<vineethr@linux.ibm.com>, <tim.c.chen@linux.intel.com>,
	<linux@treblig.org>, <linux-kernel@vger.kernel.org>,
	<linux-perf-users@vger.kernel.org>, <santosh.shukla@amd.com>,
	<sandipan.das@amd.com>
Subject: Re: [PATCH RESEND v4 03/11] perf header: Support CPU DOMAIN relation info
Date: Mon, 19 Jan 2026 23:41:51 +0530	[thread overview]
Message-ID: <3dffd1ec-54a8-4763-88db-c615bb361f95@amd.com> (raw)
In-Reply-To: <bef26ca0-6600-431e-8a23-4d612a24c87c@amd.com>

Hi Namhyung,

On 09-01-2026 16:54, Swapnil Sapkal wrote:
> Hello Namhyung,
> 
> On 03-01-2026 03:56, Namhyung Kim wrote:
>> On Tue, Sep 09, 2025 at 11:42:19AM +0000, Swapnil Sapkal wrote:
>>> '/proc/schedstat' gives the info about load balancing statistics within
>>> a given domain. It also contains the cpu_mask giving information about
>>> the sibling cpus and domain names after schedstat version 17. Storing
>>> this information in perf header will help tools like `perf sched stats`
>>> for better analysis.
>>>
>>> Signed-off-by: Swapnil Sapkal <swapnil.sapkal@amd.com>
>>> ---
>>>   .../Documentation/perf.data-file-format.txt   |  17 +
>>>   tools/perf/builtin-inject.c                   |   1 +
>>>   tools/perf/util/env.h                         |  16 +
>>>   tools/perf/util/header.c                      | 304 ++++++++++++++++++
>>>   tools/perf/util/header.h                      |   1 +
>>>   tools/perf/util/util.c                        |  42 +++
>>>   tools/perf/util/util.h                        |   3 +
>>>   7 files changed, 384 insertions(+)
>>>
>>> diff --git a/tools/perf/Documentation/perf.data-file-format.txt b/ 
>>> tools/perf/Documentation/perf.data-file-format.txt
>>> index cd95ba09f727..92dbba1003cf 100644
>>> --- a/tools/perf/Documentation/perf.data-file-format.txt
>>> +++ b/tools/perf/Documentation/perf.data-file-format.txt
>>> @@ -437,6 +437,23 @@ struct {
>>>       } [nr_pmu];
>>>   };
>>> +    HEADER_CPU_DOMAIN_INFO = 32,
>>> +
>>> +List of cpu-domain relation info. The format of the data is as below.
>>> +
>>> +struct domain_info {
>>> +    int domain;
>>> +    char dname[];
>>> +    char cpumask[];
>>> +    char cpulist[];
>>> +};
>>> +
>>> +struct cpu_domain_info {
>>> +    int cpu;
>>> +    int nr_domains;
>>> +    struct domain_info domains[];
>>> +};
>>> +
>>>       other bits are reserved and should ignored for now
>>>       HEADER_FEAT_BITS    = 256,
>>> diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
>>> index a114b3fa1bea..f43a7ec44b5f 100644
>>> --- a/tools/perf/builtin-inject.c
>>> +++ b/tools/perf/builtin-inject.c
>>> @@ -2058,6 +2058,7 @@ static bool keep_feat(int feat)
>>>       case HEADER_CLOCK_DATA:
>>>       case HEADER_HYBRID_TOPOLOGY:
>>>       case HEADER_PMU_CAPS:
>>> +    case HEADER_CPU_DOMAIN_INFO:
>>>           return true;
>>>       /* Information that can be updated */
>>>       case HEADER_BUILD_ID:
>>> diff --git a/tools/perf/util/env.h b/tools/perf/util/env.h
>>> index e00179787a34..71034c4b4488 100644
>>> --- a/tools/perf/util/env.h
>>> +++ b/tools/perf/util/env.h
>>> @@ -54,6 +54,19 @@ struct pmu_caps {
>>>       char            *pmu_name;
>>>   };
>>> +struct domain_info {
>>> +    u32    domain;
>>> +    char    *dname;
>>> +    char    *cpumask;
>>> +    char    *cpulist;
>>> +};
>>> +
>>> +struct cpu_domain_map {
>>> +    u32            cpu;
>>> +    u32            nr_domains;
>>> +    struct domain_info    **domains;
>>> +};
>>> +
>>>   typedef const char *(arch_syscalls__strerrno_t)(int err);
>>>   struct perf_env {
>>> @@ -70,6 +83,8 @@ struct perf_env {
>>>       unsigned int        max_branches;
>>>       unsigned int        br_cntr_nr;
>>>       unsigned int        br_cntr_width;
>>> +    unsigned int        schedstat_version;
>>> +    unsigned int        max_sched_domains;
>>>       int            kernel_is_64_bit;
>>>       int            nr_cmdline;
>>> @@ -92,6 +107,7 @@ struct perf_env {
>>>       char            **cpu_pmu_caps;
>>>       struct cpu_topology_map    *cpu;
>>>       struct cpu_cache_level    *caches;
>>> +    struct cpu_domain_map    **cpu_domain;
>>>       int             caches_cnt;
>>>       u32            comp_ratio;
>>>       u32            comp_ver;
>>> diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
>>> index 4f2a6e10ed5c..7ff7434bac2c 100644
>>> --- a/tools/perf/util/header.c
>>> +++ b/tools/perf/util/header.c
>>> @@ -1621,6 +1621,184 @@ static int write_pmu_caps(struct feat_fd *ff,
>>>       return 0;
>>>   }
>>> +static void free_cpu_domain_info(struct cpu_domain_map **cd_map, u32 
>>> schedstat_version, u32 nr)
>>> +{
>>> +    for (u32 i = 0; i < nr; i++) {
>>> +        if (cd_map[i]->domains) {
>>> +            for (u32 j = 0; j < cd_map[i]->nr_domains; j++) {
>>> +                struct domain_info *d_info = cd_map[i]->domains[j];
>>> +
>>
>> I'm not sure if it needs a NULL check for d_info before access.
>>
> 
> Thanks for catching this. Yes we need a NULL check here.
> 
> If some of the cpus are offline then it is possible that some sched 
> domains can be degenerated and for some cpus.
> 
> for e.g. cpu0, cpu128 are smt siblings. If cpu 128 is offline then SMT 
> domain will be degenerated for cpu0 alone. All the other cpus will have 
> a valid SMT domain. Thus SMT for the d_info can be NULL.
> 
>>
>>> +                if (schedstat_version >= 17)
>>> +                    free(d_info->dname);
>>> +
>>> +                free(d_info->cpumask);
>>> +                free(d_info->cpulist);
>>> +            }
>>> +            free(cd_map[i]->domains);
>>> +        }
>>> +    }
>>> +
>>> +    free(cd_map);
>>> +}
>>> +
>>> +static struct cpu_domain_map  **build_cpu_domain_map(u32 
>>> *schedstat_version, u32 *max_sched_domains,
>>> +                             u32 nr)
>>> +{
>>> +    struct domain_info *domain_info;
>>> +    struct cpu_domain_map **cd_map;
>>> +    char dname[16], cpumask[256];
>>> +    char cpulist[1024];
>>> +    char *line = NULL;
>>> +    u32 cpu, domain;
>>> +    u32 dcount = 0;
>>> +    size_t len;
>>> +    FILE *fp;
>>> +
>>> +    fp = fopen("/proc/schedstat", "r");
>>> +    if (!fp) {
>>> +        pr_err("Failed to open /proc/schedstat\n");
>>> +        return NULL;
>>> +    }
>>> +
>>> +    cd_map = calloc(nr, sizeof(*cd_map));
>>> +    if (!cd_map)
>>> +        goto out;
>>> +
>>> +    while (getline(&line, &len, fp) > 0) {
>>> +        int retval;
>>> +
>>> +        if (strncmp(line, "version", 7) == 0) {
>>> +            retval = sscanf(line, "version %d\n", schedstat_version);
>>> +            if (retval != 1)
>>> +                continue;
>>> +
>>> +        } else if (strncmp(line, "cpu", 3) == 0) {
>>> +            retval = sscanf(line, "cpu%u %*s", &cpu);
>>> +            if (retval == 1) {
>>> +                cd_map[cpu] = calloc(1, sizeof(*cd_map[cpu]));
>>> +                if (!cd_map[cpu])
>>> +                    goto out_free_line;
>>> +                cd_map[cpu]->cpu = cpu;
>>> +            } else
>>> +                continue;
>>> +
>>> +            dcount = 0;
>>> +        } else if (strncmp(line, "domain", 6) == 0) {
>>> +            dcount++;
>>> +
>>> +            cd_map[cpu]->domains = realloc(cd_map[cpu]->domains,
>>> +                               dcount * sizeof(domain_info));
>>> +            if (!cd_map[cpu]->domains)
>>> +                goto out_free_line;
>>
>> Please use a temporary variable to save the result in order to not lose
>> the original pointer in case of failure.
>>
> 
> Sure, will do.
> 
>>> +
>>> +            domain_info = calloc(1, sizeof(*domain_info));
>>> +            if (!domain_info)
>>> +                goto out_free_line;
>>> +
>>> +            cd_map[cpu]->domains[dcount - 1] = domain_info;
>>> +
>>> +            if (*schedstat_version >= 17) {
>>> +                retval = sscanf(line, "domain%u %s %s %*s", &domain, 
>>> dname,
>>> +                        cpumask);
>>> +                if (retval != 3)
>>> +                    continue;
>>> +
>>> +                domain_info->dname = calloc(strlen(dname) + 1, 
>>> sizeof(char));
>>> +                if (!domain_info->dname)
>>> +                    goto out_free_line;
>>> +
>>> +                strcpy(domain_info->dname, dname);
>>
>> This can be simply:
>>                 domain_info->dname = strdup(dname);
>>
> 
> Sure, will do.
> 
>>
>>> +            } else {
>>> +                retval = sscanf(line, "domain%u %s %*s", &domain, 
>>> cpumask);
>>> +                if (retval != 2)
>>> +                    continue;
>>> +            }
>>> +
>>> +            domain_info->domain = domain;
>>> +            if (domain > *max_sched_domains)
>>> +                *max_sched_domains = domain;
>>> +
>>> +            domain_info->cpumask = calloc(strlen(cpumask) + 1, 
>>> sizeof(char));
>>> +            if (!domain_info->cpumask)
>>> +                goto out_free_line;
>>> +
>>> +            strcpy(domain_info->cpumask, cpumask);
>>> +
>>> +            cpumask_to_cpulist(cpumask, cpulist);
>>> +            domain_info->cpulist = calloc(strlen(cpulist) + 1, 
>>> sizeof(char));
>>> +            if (!domain_info->cpulist)
>>> +                goto out_free_line;
>>
>> All error paths should call free_cpu_domain_info() at some point and
>> free the intermediate domain_info properly.
>>
> 
> Sure, will do.

Sorry, I missed to mention here that the parent function is taking care 
of freeing the memory by calling free_cpu_domain_info() in case of error 
path also. Because of which I have not addressed this in the next version.

--
Thanks and Regards,
Swapnil



  reply	other threads:[~2026-01-19 18:12 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-09-09 11:42 [PATCH RESEND v4 00/11] perf sched: Introduce stats tool Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 01/11] perf: Add print_separator to util Swapnil Sapkal
2026-01-02 22:12   ` Namhyung Kim
2026-01-09 11:23     ` Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 02/11] tools/lib: Add list_is_first() Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 03/11] perf header: Support CPU DOMAIN relation info Swapnil Sapkal
2025-09-17  6:20   ` kernel test robot
2026-01-02 22:26   ` Namhyung Kim
2026-01-09 11:24     ` Swapnil Sapkal
2026-01-19 18:11       ` Swapnil Sapkal [this message]
2025-09-09 11:42 ` [PATCH RESEND v4 04/11] perf sched stats: Add record and rawdump support Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 05/11] perf sched stats: Add schedstat v16 support Swapnil Sapkal
2026-01-02 22:28   ` Namhyung Kim
2026-01-09 11:24     ` Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 06/11] perf sched stats: Add schedstat v17 support Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 07/11] perf sched stats: Add support for report subcommand Swapnil Sapkal
2026-01-02 22:35   ` Namhyung Kim
2026-01-09 11:25     ` Swapnil Sapkal
2026-01-19 18:35       ` Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 08/11] perf sched stats: Add support for live mode Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 09/11] perf sched stats: Add support for diff subcommand Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 10/11] perf sched stats: Add basic perf sched stats test Swapnil Sapkal
2025-09-09 11:42 ` [PATCH RESEND v4 11/11] perf sched stats: Add details in man page Swapnil Sapkal

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=3dffd1ec-54a8-4763-88db-c615bb361f95@amd.com \
    --to=swapnil.sapkal@amd.com \
    --cc=acme@kernel.org \
    --cc=adrian.hunter@intel.com \
    --cc=ak@linux.intel.com \
    --cc=alexander.shishkin@linux.intel.com \
    --cc=ashelat@redhat.com \
    --cc=ben.gainey@arm.com \
    --cc=blakejones@google.com \
    --cc=ctshao@google.com \
    --cc=dvyukov@google.com \
    --cc=gautham.shenoy@amd.com \
    --cc=graham.woodward@arm.com \
    --cc=irogers@google.com \
    --cc=james.clark@arm.com \
    --cc=jolsa@kernel.org \
    --cc=juri.lelli@redhat.com \
    --cc=kan.liang@linux.intel.com \
    --cc=kprateek.nayak@amd.com \
    --cc=leo.yan@arm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-perf-users@vger.kernel.org \
    --cc=linux@treblig.org \
    --cc=mark.rutland@arm.com \
    --cc=mingo@redhat.com \
    --cc=namhyung@kernel.org \
    --cc=peterz@infradead.org \
    --cc=quic_zhonhan@quicinc.com \
    --cc=ravi.bangoria@amd.com \
    --cc=rostedt@goodmis.org \
    --cc=sandipan.das@amd.com \
    --cc=santosh.shukla@amd.com \
    --cc=sshegde@linux.ibm.com \
    --cc=thomas.falcon@intel.com \
    --cc=tim.c.chen@linux.intel.com \
    --cc=tj@kernel.org \
    --cc=vincent.guittot@linaro.org \
    --cc=vineethr@linux.ibm.com \
    --cc=void@manifault.com \
    --cc=yangjihong@bytedance.com \
    --cc=yu.c.chen@intel.com \
    --cc=yujie.liu@intel.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®