linux-kernel - Re: [PATCH v1 4/5] perf mem: Support record for hybrid platform

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]

Message-ID: <6d6f1040-6c96-7d1d-c766-5fb0057e1cc4@linux.intel.com>
Date:   Wed, 26 May 2021 09:51:34 +0800
From:   "Jin, Yao" <yao.jin@...ux.intel.com>
To:     Jiri Olsa <jolsa@...hat.com>
Cc:     acme@...nel.org, jolsa@...nel.org, peterz@...radead.org,
        mingo@...hat.com, alexander.shishkin@...ux.intel.com,
        Linux-kernel@...r.kernel.org, ak@...ux.intel.com,
        kan.liang@...el.com, yao.jin@...el.com
Subject: Re: [PATCH v1 4/5] perf mem: Support record for hybrid platform

Hi Jiri,

On 5/25/2021 3:39 PM, Jin, Yao wrote:
> Hi Jiri,
> 
>>>>       rec_argv = calloc(rec_argc + 1, sizeof(char *));
>>>>       if (!rec_argv)
>>>>           return -1;
>>>> +    /*
>>>> +     * Save the allocated event name strings.
>>>> +     */
>>>> +    rec_tmp = calloc(rec_argc + 1, sizeof(char *));
>>>> +    if (!rec_tmp) {
>>>> +        free(rec_argv);
>>>> +        return -1;
>>>> +    }
>>>
>>> why not do strdup on all of them and always call free instead?
>>> that would get rid of the rec_tmp and tmp_nr
>>>
>>
>> That is also one method. Let me try it.
>>
> 
> If we do strdup on all of them, such as,
> 
>      if (e->record)
>          rec_argv[i++] = strdup("-W");
> 
>      rec_argv[i++] = strdup("-d");
> 
>      if (mem->phys_addr)
>          rec_argv[i++] = strdup("--phys-data");
>      ....
> 
> That looks too much strdup used here. So I choose to use a rec_tmp[] to record the allocated string 
> and free them before exit the function.
> 
> Or we record the start index and end index in rec_argv[] for the allocated event string, use strdup 
> on them and call free before exit the function.
> 

This method looks also not OK.

The rec_argv[] is changed in cmd_record() for some complex command lines.

For example,

./perf mem record -- ./memtest -R0d -b2000 -d64 -n100

Before cmd_record(), rec_argv[3] = "-e".
After cmd_record(), rec_argv[3] = "-d64"

Even we do strdup on all of rec_argv[], but the entries are probably changed in cmd_record(), so we 
can't free the entries at the end of __cmd_record().

Maybe we have to use the original way which just records the allocated event string to a temporary 
array and free the whole array at the end of __cmd_record().

What do you think?

Thanks
Jin Yao

> What do you think?
> 
> Thanks
> Jin Yao