Re: string.h: Mark 34 functions with __must_check

2019-10-10 Thread Markus Elfring
> +our $PositionalAttribute = qr{ > + __must_check| > + __printf| I suggest to put all key words which share the leading underscores into another alternation for an improved regular expression. Regards, Markus

Re: string.h: Mark 34 functions with __must_check

2019-10-10 Thread Markus Elfring
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/lib/test_kasan.c?id=b92a953cb7f727c42a15ac2ea59bf3cf9c39370d#n595 > > The *test* word must have given you a clue that the code you a looking > at is not an ordinary one. The proposed extension of function annotations can be

Re: string.h: Mark 34 functions with __must_check

2019-10-09 Thread Andy Shevchenko
On Wed, Oct 9, 2019 at 11:11 PM Markus Elfring wrote: > > > I'm curious. How many warnings showed up when you applied this patch? > > I suggest to take another look at six places in a specific source file > (for example). > https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/li

Re: string.h: Mark 34 functions with __must_check

2019-10-09 Thread Markus Elfring
> I'm curious. How many warnings showed up when you applied this patch? I suggest to take another look at six places in a specific source file (for example). https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/lib/test_kasan.c?id=b92a953cb7f727c42a15ac2ea59bf3cf9c39370d#n595 ht

Re: string.h: Mark 34 functions with __must_check

2019-10-09 Thread Nick Desaulniers
On Wed, Oct 9, 2019 at 10:04 AM Markus Elfring wrote: > > > Ah, granted, I was surprised, too. > > Thanks for this view. I mean, it's a good thing that we don't have any issues that this patch would catch today. Seems Steven and I were surprised (pessimistic?). > > > > Maybe would be helpful to

Re: string.h: Mark 34 functions with __must_check

2019-10-09 Thread Markus Elfring
> Ah, granted, I was surprised, too. Thanks for this view. > Maybe would be helpful to mention that in the commit message. My Linux software build resources might be too limited to take more system configuration variations safely into account for this issue. Would you like to achieve further ch

Re: string.h: Mark 34 functions with __must_check

2019-10-09 Thread Markus Elfring
> You're also not consistent - strlen() is not annotated. Would you like to integrate any additional function annotations? > And, for the standard C functions, -Wall already seems to warn about > an unused call: This detail is nice, isn't it? > a.c:5:2: warning: statement with no effect [-Wun