Re: [PATCH 2/2] ALSA: ctl: refactoring for read operation

[Takashi Sakamoto <o-takashi@sakamocchi.jp>]

On Apr 10 2015 19:43, Takashi Iwai wrote:
>> I'm unaware of it... These code should be:
>>
>> if (copy_to_user(buffer, &ev, sizeof(struct snd_ctl_event))) {
>>     if (result == 0)
>>         result = -EFAULT;
>>     break;
> 
> Well, it's not much better than the original code, IMO.

OK. I dropped this patch.

>> And I realize sound/firewire/fireworks/fireworks_hwdep.c has the same
>> bug. I'll post another patch for this bug.
> 
> This is an implementation detail and I believe it rather depends on
> each driver.  (But I should reread POSIX definition again before
> concluding it.)
> 
> That said, it isn't wrong to return -EFAULT immediately, per se.
> The problem here is that you change the existing behavior.  Especially
> a core code like this should be treated carefully even for such a
> small behavior change.

Thanks

Takashi Sakamoto