|
From: Florian K. <br...@ac...> - 2011-05-10 20:27:00
Attachments:
r11734-static-analysis
|
Greetings. Attached are the results from a new run of IBM's BEAM checker. Good progress since last time. Here's the breakdown of the complaints: coregrind 1 gdbserver 3 callgrind 4 cachegrind 1 memcheck 2 helgrind 1 All the other components were clean. Like the previous run this was on valgrind configured for x86 assuming an ILP32 target. Florian |
|
From: Philippe W. <phi...@sk...> - 2011-05-10 23:45:18
|
> gdbserver 3 Thanks for this (as the signal to noise ratio of BEAM is ok, it is nice to receive plenty of complaints of this tool :). 2 of 3 are real bugs. I have uploaded a patch (comment 71 in bug 214909) fixing the fd leak and the missing break. For what concerns the "useless" 3rd argument in VG_(fcntl) : I do not understand how the checker was able to detect that this is matching fcntl usage : the reported error is ok for fcntl; so I tried to fix it but then it does not compile anymore as VG_(fcntl) is defined in Valgrind as a 3 arg function. Is there a way to "silence" the BEAM checker for the next run ? (e.g. with a special comment ) ? |
|
From: Florian K. <br...@ac...> - 2011-05-11 00:27:44
|
On 05/10/2011 07:45 PM, Philippe Waroquiers wrote:
>
> For what concerns the "useless" 3rd argument in VG_(fcntl) : I do not
> understand how the checker was able to detect that this is matching
> fcntl usage
The checker did not figure this out. I did :) and I told BEAM about it.
> the reported error is ok for fcntl; so I tried to fix it
> but then it does not compile anymore as VG_(fcntl) is defined
> in Valgrind as a 3 arg function.
>
I understand. I will suppress the complaint in future runs.
> Is there a way to "silence" the BEAM checker for the next run ?
> (e.g. with a special comment ) ?
Yes, it would be possible to suppress it with a structured comment. But
I don't like polluting code with comments to placate checkers.
If you look at the complaint it has this hash code beginning with >>>
and that can be used as well to suppress the complaint. No code
modifications are needed in that case.
Florian
|
|
From: Florian K. <br...@ac...> - 2011-05-13 13:46:36
|
On 05/12/2011 10:46 PM, Scott Pakin wrote: > > > On 05/10/2011 02:26 PM, Florian Krohm wrote: >> Attached are the results from a new run of IBM's BEAM checker. Good >> progress since last time. Here's the breakdown of the complaints: > > This is the first I've heard of BEAM, and I'm curious: Is BEAM a > commercial product? Freely available? I cannot speak for IBM but I think that the answer is "no". Twice. > Not available? BEAM is made available to the valgrind project by special agreement. Florian |