linux-kernel - Re: [PATCH v1] perf tools: Fix pattern matching for same substring used in different pmu type

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 11 Jun 2021 10:54:38 +0800
From:   "Jin, Yao" <yao.jin@...ux.intel.com>
To:     acme@...nel.org, jolsa@...nel.org, peterz@...radead.org,
        mingo@...hat.com, alexander.shishkin@...ux.intel.com
Cc:     Linux-kernel@...r.kernel.org, ak@...ux.intel.com,
        kan.liang@...el.com, yao.jin@...el.com
Subject: Re: [PATCH v1] perf tools: Fix pattern matching for same substring
 used in different pmu type

And, though this patch is to fix the uncore_imc/uncore_imc_free_running mismatching issue. We are 
also surprised to see it can solve another hybrid PMU issue on Alderlake.

On Alderlake,

# ./perf stat -e cpu/event=0xc2,umask=0x2/ true

Performance counter stats for 'true':

            702,246      cpu_core/event=0xc2,umask=0x2/
      <not counted>      cpu_atom/event=0xc2,umask=0x2/

It should error out with the wrong PMU rather than using 'cpu_core' and'cpu_atom' instead.

This is still the pattern matching issue. The pattern is "cpu*". Both "cpu_core" and "cpu_atom" can 
match the pattern "cpu*", so the parser wrongly thinks they are the same PMU type.

Now with this patch,

# ./perf stat -e cpu/cpu-cycles/ true
event syntax error: 'cpu/cpu-cycles/'
                      \___ Cannot find PMU `cpu'. Missing kernel support?
Run 'perf list' for a list of valid events

Thanks
Jin Yao

On 6/9/2021 12:57 PM, Jin Yao wrote:
> Some different pmu types may have same substring. For example,
> on Icelake server, we have pmu types "uncore_imc" and
> "uncore_imc_free_running". Both pmu types have substring "uncore_imc".
> But the parser would wrongly think they are the same pmu type.
> 
> We enable an imc event,
> perf stat -e uncore_imc/event=0xe3/ -a -- sleep 1
> 
> Perf actually expands the event to:
> uncore_imc_0/event=0xe3/
> uncore_imc_1/event=0xe3/
> uncore_imc_2/event=0xe3/
> uncore_imc_3/event=0xe3/
> uncore_imc_4/event=0xe3/
> uncore_imc_5/event=0xe3/
> uncore_imc_6/event=0xe3/
> uncore_imc_7/event=0xe3/
> uncore_imc_free_running_0/event=0xe3/
> uncore_imc_free_running_1/event=0xe3/
> uncore_imc_free_running_3/event=0xe3/
> uncore_imc_free_running_4/event=0xe3/
> 
> That's because the "uncore_imc_free_running" matches the
> pattern "uncore_imc*".
> 
> Now we check that the last characters of pmu name is
> '_<digit>'.
> 
> Fixes: b2b9d3a3f021 ("perf pmu: Support wildcards on pmu name in dynamic pmu events")
> Signed-off-by: Jin Yao <yao.jin@...ux.intel.com>
> ---
>   tools/perf/util/parse-events.y |  2 ++
>   tools/perf/util/pmu.c          | 25 ++++++++++++++++++++++++-
>   tools/perf/util/pmu.h          |  1 +
>   3 files changed, 27 insertions(+), 1 deletion(-)
> 
> diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y
> index aba12a4d488e..7a694c7f7f1a 100644
> --- a/tools/perf/util/parse-events.y
> +++ b/tools/perf/util/parse-events.y
> @@ -317,6 +317,8 @@ event_pmu_name opt_pmu_config
>   			    strncmp($1, "uncore_", 7))
>   				name += 7;
>   			if (!fnmatch(pattern, name, 0)) {
> +				if (!perf_pmu__valid_suffix($1, name))
> +					continue;
>   				if (parse_events_copy_term_list(orig_terms, &terms))
>   					CLEANUP_YYABORT;
>   				if (!parse_events_add_pmu(_parse_state, list, pmu->name, terms, true, false))
> diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
> index 88c8ecdc60b0..78af01959830 100644
> --- a/tools/perf/util/pmu.c
> +++ b/tools/perf/util/pmu.c
> @@ -3,6 +3,7 @@
>   #include <linux/compiler.h>
>   #include <linux/string.h>
>   #include <linux/zalloc.h>
> +#include <linux/ctype.h>
>   #include <subcmd/pager.h>
>   #include <sys/types.h>
>   #include <errno.h>
> @@ -768,7 +769,7 @@ bool pmu_uncore_alias_match(const char *pmu_name, const char *name)
>   	 */
>   	for (; tok; name += strlen(tok), tok = strtok_r(NULL, ",", &tmp)) {
>   		name = strstr(name, tok);
> -		if (!name) {
> +		if (!name || !perf_pmu__valid_suffix(tok, (char *)name)) {
>   			res = false;
>   			goto out;
>   		}
> @@ -1872,3 +1873,25 @@ bool perf_pmu__has_hybrid(void)
>   
>   	return !list_empty(&perf_pmu__hybrid_pmus);
>   }
> +
> +bool perf_pmu__valid_suffix(char *tok, char *pmu_name)
> +{
> +	char *p;
> +
> +	/*
> +	 * The pmu_name has substring tok. If the format of
> +	 * pmu_name is <tok> or <tok>_<digit>, return true.
> +	 */
> +	p = pmu_name + strlen(tok);
> +	if (*p == 0)
> +		return true;
> +
> +	if (*p != '_')
> +		return false;
> +
> +	++p;
> +	if (*p == 0 || !isdigit(*p))
> +		return false;
> +
> +	return true;
> +}
> diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h
> index a790ef758171..ebfd2b71532b 100644
> --- a/tools/perf/util/pmu.h
> +++ b/tools/perf/util/pmu.h
> @@ -133,5 +133,6 @@ void perf_pmu__warn_invalid_config(struct perf_pmu *pmu, __u64 config,
>   				   char *name);
>   
>   bool perf_pmu__has_hybrid(void);
> +bool perf_pmu__valid_suffix(char *tok, char *pmu_name);
>   
>   #endif /* __PMU_H */
>