[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <1d10fab5-d375-79b8-122d-f3cb66f2d94a@intel.com>
Date: Mon, 31 Jul 2023 12:26:00 +0300
From: Adrian Hunter <adrian.hunter@...el.com>
To: Arnaldo Carvalho de Melo <acme@...nel.org>
Cc: Ian Rogers <irogers@...gle.com>,
"Aneesh Kumar K.V" <aneesh.kumar@...ux.ibm.com>,
Athira Rajeev <atrajeev@...ux.vnet.ibm.com>,
Disha Goel <disgoel@...ux.vnet.ibm.com>,
Jiri Olsa <jolsa@...nel.org>, Kajol Jain <kjain@...ux.ibm.com>,
Madhavan Srinivasan <maddy@...ux.ibm.com>,
Namhyung Kim <namhyung@...nel.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH 1/1] perf dlfilter: Initialize addr_location before
passing it to thread__find_symbol_fb()
On 28/07/23 22:18, Adrian Hunter wrote:
> On 27/07/23 17:13, Arnaldo Carvalho de Melo wrote:
>> Em Tue, Jul 25, 2023 at 02:51:58PM +0300, Adrian Hunter escreveu:
>>> On 25/07/23 00:31, Arnaldo Carvalho de Melo wrote:
>>>> Em Mon, Jul 24, 2023 at 02:16:05PM -0700, Ian Rogers escreveu:
>>>>> On Mon, Jul 24, 2023 at 2:09 PM Arnaldo Carvalho de Melo
>>>>> <acme@...nel.org> wrote:
>>>>>>
>>>>>> As thread__find_symbol_fb() will end up calling thread__find_map() and
>>>>>> it in turn will call these on uninitialized memory:
>>>>>>
>>>>>> maps__zput(al->maps);
>>>>>> map__zput(al->map);
>>>>>> thread__zput(al->thread);
>>>>>
>>>>> Normally there would be a addr_location__exit. It looks here like most
>>>>> values are copied in al_to_d_al, which copies reference counted values
>>>>> without doing appropriate gets. Perhaps add the gets in al_to_d_al and
>>>>> add the exit to make it clear there can't be addr_location related
>>>>> leaks here.
>>
>> Or just make al_to_d_al use strdups for the strings it gets from
>> reference counter structs.
>>
>> That or make the caller of ->resolve_address() provide a initialized
>> addr_location that it then exits.
>
> OK, I will send a patch.
2 patches plus yours:
https://lore.kernel.org/linux-perf-users/20230731091857.10681-1-adrian.hunter@intel.com/T/#t
>
>>
>> - Arnaldo
>>
>>>> I'll do that as well and add the addr_location__exit() then send a V2.
>>>
>>> The data is not valid outside the context of ->filter_event() or
>>> ->filter_event_early(), so the reference counts should not be changed.
>>> That probably needs to be made clearer in the documentation.
>>
>>
>>
>>>>
>>>> - Arnaldo
>>>>
>>>>> Thanks,
>>>>> Ian
>>>>>
>>>>>> Fixes: 0dd5041c9a0eaf8c ("perf addr_location: Add init/exit/copy functions")
>>>>>> Cc: Adrian Hunter <adrian.hunter@...el.com>
>>>>>> Cc: Aneesh Kumar K.V <aneesh.kumar@...ux.ibm.com>
>>>>>> Cc: Athira Rajeev <atrajeev@...ux.vnet.ibm.com>
>>>>>> Cc: Disha Goel <disgoel@...ux.vnet.ibm.com>
>>>>>> Cc: Ian Rogers <irogers@...gle.com>
>>>>>> Cc: Jiri Olsa <jolsa@...nel.org>
>>>>>> Cc: Kajol Jain <kjain@...ux.ibm.com>
>>>>>> Cc: Madhavan Srinivasan <maddy@...ux.ibm.com>
>>>>>> Cc: Namhyung Kim <namhyung@...nel.org>
>>>>>> Link: https://lore.kernel.org/lkml/
>>>>>> Signed-off-by: Arnaldo Carvalho de Melo <acme@...hat.com>
>>>>>> ---
>>>>>> tools/perf/util/dlfilter.c | 1 +
>>>>>> 1 file changed, 1 insertion(+)
>>>>>>
>>>>>> diff --git a/tools/perf/util/dlfilter.c b/tools/perf/util/dlfilter.c
>>>>>> index 46f74b2344dbb34c..798a53d7e6c9dfc5 100644
>>>>>> --- a/tools/perf/util/dlfilter.c
>>>>>> +++ b/tools/perf/util/dlfilter.c
>>>>>> @@ -166,6 +166,7 @@ static __s32 dlfilter__resolve_address(void *ctx, __u64 address, struct perf_dlf
>>>>>> if (!thread)
>>>>>> return -1;
>>>>>>
>>>>>> + addr_location__init(&al);
>>>>>> thread__find_symbol_fb(thread, d->sample->cpumode, address, &al);
>>>>>>
>>>>>> al_to_d_al(&al, &d_al);
>>>>>> --
>>>>>> 2.37.1
>>>>>>
>>>>
>>>
>>
>
Powered by blists - more mailing lists