Re: [PATCH bpf-next v4 1/3] bpftool: Add bpf_token show

[Tao Chen <chen.dylane@linux.dev>]

在 2025/9/16 00:21, Andrii Nakryiko 写道:

Hi Andrii, Chris,

> On Sat, Sep 13, 2025 at 10:18 AM Chris Mason <clm@meta.com> wrote:
>>
>> On Wed, 23 Jul 2025 22:44:40 +0800 Tao Chen <chen.dylane@linux.dev> wrote:
>>
>> [ ... ]
>>
>>> diff --git a/tools/bpf/bpftool/token.c b/tools/bpf/bpftool/token.c
>>> new file mode 100644
>>> index 00000000000..6312e662a12
>>> --- /dev/null
>>> +++ b/tools/bpf/bpftool/token.c
>>> @@ -0,0 +1,225 @@
>>
>> [ ... ]
>>
>>> +
>>> +static char *get_delegate_value(const char *opts, const char *key)
>>> +{
>>> +     char *token, *rest, *ret = NULL;
>>> +     char *opts_copy = strdup(opts);
>>> +
>>> +     if (!opts_copy)
>>> +             return NULL;
>>> +
>>> +     for (token = strtok_r(opts_copy, ",", &rest); token;
>>> +                     token = strtok_r(NULL, ",", &rest)) {
>>> +             if (strncmp(token, key, strlen(key)) == 0 &&
>>> +                 token[strlen(key)] == '=') {
>>> +                     ret = token + strlen(key) + 1;
>>> +                     break;
>>> +             }
>>> +     }
>>> +     free(opts_copy);
>>> +
>>> +     return ret;
>>
>> The ret pointer is pointing inside opts_copy, but opts_copy gets freed
>> before returning?
> 
> Thanks for the bug report, Chris!
> 
> Tao, the fix probably should be something along the lines of:
> 
> ret = token + strlen(key) + 1 - opts_copy + opts;
> 
> to translate that pointer back into the original string? Can you
> please send a patch?
> 

Of course, i will fix it, thanks for the bug report.

>>
>> -chris
-- 
Best Regards
Tao Chen