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:   Tue, 18 Jul 2023 12:56:48 +0300
From:   Adrian Hunter <adrian.hunter@...el.com>
To:     Yang Jihong <yangjihong1@...wei.com>, peterz@...radead.org,
        mingo@...hat.com, acme@...nel.org, mark.rutland@....com,
        alexander.shishkin@...ux.intel.com, jolsa@...nel.org,
        namhyung@...nel.org, irogers@...gle.com, kan.liang@...ux.intel.com,
        james.clark@....com, tmricht@...ux.ibm.com, ak@...ux.intel.com,
        anshuman.khandual@....com, linux-kernel@...r.kernel.org,
        linux-perf-users@...r.kernel.org
Subject: Re: [PATCH v2 5/7] perf evlist: Skip dummy event sample_type check
 for evlist_config

On 18/07/23 12:30, Yang Jihong wrote:
> Hello,
> 
> On 2023/7/17 22:41, Adrian Hunter wrote:
>> On 15/07/23 06:29, Yang Jihong wrote:
>>> The dummp event does not contain sampls data. Therefore, sample_type does
>>> not need to be checked.
>>>
>>> Currently, the sample id format of the actual sampling event may be changed
>>> after the dummy event is added.
>>>
>>> Signed-off-by: Yang Jihong <yangjihong1@...wei.com>
>>> ---
>>>   tools/perf/util/record.c | 7 +++++++
>>>   1 file changed, 7 insertions(+)
>>>
>>> diff --git a/tools/perf/util/record.c b/tools/perf/util/record.c
>>> index 9eb5c6a08999..0240be3b340f 100644
>>> --- a/tools/perf/util/record.c
>>> +++ b/tools/perf/util/record.c
>>> @@ -128,6 +128,13 @@ void evlist__config(struct evlist *evlist, struct record_opts *opts, struct call
>>>           evlist__for_each_entry(evlist, evsel) {
>>>               if (evsel->core.attr.sample_type == first->core.attr.sample_type)
>>>                   continue;
>>> +
>>> +            /*
>>> +             * Skip the sample_type check for the dummy event
>>> +             * because it does not have any samples anyway.
>>> +             */
>>> +            if (evsel__is_dummy_event(evsel))
>>> +                continue;
>>
>> Sideband event records have "ID samples" so the sample type still matters.
>>
> Okay, will remove this patch in next version.
> 
> Can I ask a little more about this?
> 
> Use PERF_SAMPLE_IDENTIFICATION instead of PERF_SAMPLE_ID because for samples of type PERF_RECORD_SAMPLE, there may be different record formats due to different *sample_type* settings, so the fixed SAMPLE_ID  location mode PERF_SAMPLE_NAME is required here.
> 
> However, for the sideband event, the samples of the PERF_RECORD_SAMPLE type is not recorded (only PERF_RECORD_MMAP, PERF_RECORD_COMM, and so on). Therefore, the "use sample identifier "check can be skipped here.
> 
> That's my understanding of PERF_SAMPLE_IDENTIFICATION . If there is any error, please help to correct it.
> 
> *Sideband event records have "ID samples" so the sample type still matters.*
> 
> Does this mean that sideband will also record samples of type PERF_RECORD_SAMPLE? What exactly is the sampling data?

No.  There are additional members as defined by struct sample_id for PERF_RECORD_MMAP:

https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/uapi/linux/perf_event.h?h=v6.4#n872

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ