2006-10-22 20:38:06

by Amit Choudhary

Subject: [PATCH 2.6.19-rc2] mm/slab.c: check kmalloc() return value.

Description: Check the return value of kmalloc() in function setup_cpu_cache(), in file mm/slab.c.

Signed-off-by: Amit Choudhary <[email protected]>

diff --git a/mm/slab.c b/mm/slab.c
index 84c631f..613ae61 100644
--- a/mm/slab.c
+++ b/mm/slab.c
@@ -2021,6 +2021,7 @@ static int setup_cpu_cache(struct kmem_c
} else {
cachep->array[smp_processor_id()] =
kmalloc(sizeof(struct arraycache_init), GFP_KERNEL);
+ BUG_ON(!cachep->array[smp_processor_id()]);

if (g_cpucache_up == PARTIAL_AC) {
set_up_list3s(cachep, SIZE_L3);

2006-10-22 21:16:10

by Josef Sipek

Subject: Re: [PATCH 2.6.19-rc2] mm/slab.c: check kmalloc() return value.

On Sun, Oct 22, 2006 at 01:37:51PM -0700, Amit Choudhary wrote:
> Description: Check the return value of kmalloc() in function setup_cpu_cache(), in file mm/slab.c.
>
> Signed-off-by: Amit Choudhary <[email protected]>
>
> diff --git a/mm/slab.c b/mm/slab.c
> index 84c631f..613ae61 100644
> --- a/mm/slab.c
> +++ b/mm/slab.c
> @@ -2021,6 +2021,7 @@ static int setup_cpu_cache(struct kmem_c
> } else {
> cachep->array[smp_processor_id()] =
> kmalloc(sizeof(struct arraycache_init), GFP_KERNEL);
> + BUG_ON(!cachep->array[smp_processor_id()]);
>
> if (g_cpucache_up == PARTIAL_AC) {
> set_up_list3s(cachep, SIZE_L3);

Any reason to BUG instead of trying to return as gracefully as possible?

Josef "Jeff" Sipek.

--
My public GPG key can be found at
http://www.josefsipek.net/gpg/public-0xC7958FFE.txt

2006-10-22 22:10:43

by Dave Jones

Subject: Re: [PATCH 2.6.19-rc2] mm/slab.c: check kmalloc() return value.

On Sun, Oct 22, 2006 at 05:16:05PM -0400, Josef Sipek wrote:
> On Sun, Oct 22, 2006 at 01:37:51PM -0700, Amit Choudhary wrote:
> > Description: Check the return value of kmalloc() in function setup_cpu_cache(), in file mm/slab.c.
> >
> > Signed-off-by: Amit Choudhary <[email protected]>
> >
> > diff --git a/mm/slab.c b/mm/slab.c
> > index 84c631f..613ae61 100644
> > --- a/mm/slab.c
> > +++ b/mm/slab.c
> > @@ -2021,6 +2021,7 @@ static int setup_cpu_cache(struct kmem_c
> > } else {
> > cachep->array[smp_processor_id()] =
> > kmalloc(sizeof(struct arraycache_init), GFP_KERNEL);
> > + BUG_ON(!cachep->array[smp_processor_id()]);
> >
> > if (g_cpucache_up == PARTIAL_AC) {
> > set_up_list3s(cachep, SIZE_L3);
>
> Any reason to BUG instead of trying to return as gracefully as possible?

If we fail to allocate memory this early in boot, we're hosed anyway.

Dave

--
http://www.codemonkey.org.uk

2006-10-23 06:38:15

by Pekka Enberg

Subject: Re: [PATCH 2.6.19-rc2] mm/slab.c: check kmalloc() return value.

On 10/22/06, Amit Choudhary <[email protected]> wrote:
> diff --git a/mm/slab.c b/mm/slab.c
> index 84c631f..613ae61 100644
> --- a/mm/slab.c
> +++ b/mm/slab.c
> @@ -2021,6 +2021,7 @@ static int setup_cpu_cache(struct kmem_c
> } else {
> cachep->array[smp_processor_id()] =
> kmalloc(sizeof(struct arraycache_init), GFP_KERNEL);
> + BUG_ON(!cachep->array[smp_processor_id()]);
>
> if (g_cpucache_up == PARTIAL_AC) {
> set_up_list3s(cachep, SIZE_L3);

Looks good. You might want to send this to [email protected] directly.