Cong Wang <cwang@xxxxxxxxxxxxxxxx> writes:
> On Mon, Feb 3, 2014 at 9:26 PM, Eric W. Biederman <ebiederm@xxxxxxxxxxxx> wrote:
>> diff --git a/fs/file.c b/fs/file.c
>> index 771578b33fb6..db25c2bdfe46 100644
>> --- a/fs/file.c
>> +++ b/fs/file.c
>> @@ -34,7 +34,7 @@ static void *alloc_fdmem(size_t size)
>> * vmalloc() if the allocation size will be considered "large" by the VM.
>> */
>> if (size <= (PAGE_SIZE << PAGE_ALLOC_COSTLY_ORDER)) {
>> - void *data = kmalloc(size, GFP_KERNEL|__GFP_NOWARN);
>> + void *data = kmalloc(size, GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY);
>> if (data != NULL)
>> return data;
>> }
>
> Or try again without __GFP_NORETRY like we do in nelink mmap?
I think I would much rather keep the current semantics of return -ENOMEM
and keep the problem localized then trigger a box wide OOM thank you
very much.
Retrying the kmalloc without __GFP_NORETRY is pointless. If you are in
the unlikely 0.01% of the time when the kmalloc fails it is almost
certainly going to fail again. Writing out_of_memory() as kmalloc()
is pointless and very confusing.
The vmalloc won't fail unless you are on a 32bit box. So it isn't a
case that anyone has to deal with in practice.
Eric
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>