This patch introduces a kcalloc() and replaces ALSA magic allocator
snd_kcalloc() and snd_magic_kcalloc() with it. I also fixed a memory
leak in the arm sa11xx driver and added a BUG() in seq_oss_synth to
catch failing allocations.
I kept the snd_magic_cast macro in place just in case someone wants to
add a generic type checking facility in the kernel and convert the users
to use it.
The patch is against 2.6.6 and was compile-tested with allyesconfig.
Signed-off-by: Pekka Enberg <[email protected]>
On Tue, 2004-06-08 at 23:24, Pekka J Enberg wrote:
> This patch introduces a kcalloc()
too bad you didn't make kcalloc an array allocator, which then could
check for multiplication overfow nicely.....
On Wed, 9 Jun 2004, Pekka J Enberg wrote:
> This patch introduces a kcalloc() and replaces ALSA magic allocator
> snd_kcalloc() and snd_magic_kcalloc() with it. I also fixed a memory
> leak in the arm sa11xx driver and added a BUG() in seq_oss_synth to
> catch failing allocations.
>
> I kept the snd_magic_cast macro in place just in case someone wants to
> add a generic type checking facility in the kernel and convert the users
> to use it.
I suggest to Linus and Andrew to postpone applying of these patches into
mainstream. We'll integrate them (maybe in a slightly modified form -
after discussion on the alsa-devel mailing list) to our ALSA tree at
first. Thanks.
Jaroslav
-----
Jaroslav Kysela <[email protected]>
Linux Kernel Sound Maintainer
ALSA Project, SuSE Labs
At Wed, 9 Jun 2004 00:24:53 +0300,
Pekka J Enberg wrote:
>
> This patch introduces a kcalloc() and replaces ALSA magic allocator
> snd_kcalloc() and snd_magic_kcalloc() with it. I also fixed a memory
> leak in the arm sa11xx driver and added a BUG() in seq_oss_synth to
> catch failing allocations.
>
> I kept the snd_magic_cast macro in place just in case someone wants to
> add a generic type checking facility in the kernel and convert the users
> to use it.
>
> The patch is against 2.6.6 and was compile-tested with allyesconfig.
no, this patch breaks when CONFIG_SND_DEBUG=y. the compile is ok but
it won't work, as long as you leave snd_magic_cast.
Takashi
At Wed, 09 Jun 2004 10:57:45 +0200,
I wrote:
>
> At Wed, 9 Jun 2004 00:24:53 +0300,
> Pekka J Enberg wrote:
> >
> > This patch introduces a kcalloc() and replaces ALSA magic allocator
> > snd_kcalloc() and snd_magic_kcalloc() with it. I also fixed a memory
> > leak in the arm sa11xx driver and added a BUG() in seq_oss_synth to
> > catch failing allocations.
> >
> > I kept the snd_magic_cast macro in place just in case someone wants to
> > add a generic type checking facility in the kernel and convert the users
> > to use it.
> >
> > The patch is against 2.6.6 and was compile-tested with allyesconfig.
>
> no, this patch breaks when CONFIG_SND_DEBUG=y. the compile is ok but
> it won't work, as long as you leave snd_magic_cast.
sorry, i overlooked the patch 8/8.
but i don't see no reason to leave only snd_magic_cast.
if there is no cast checking facility, we should remove them all.
Takashi