On Tue, 2005-09-20 at 12:56 -0700, Mark Haverkamp wrote:
> + && ((hw_fib_pool = kmalloc(sizeof
> (struct hw_fib *) * num, GFP_ATOMIC|GFP_KERNEL)))
> + && ((fib_pool = kmalloc(sizeof(struct
> fib *) * num, GFP_ATOMIC|GFP_KERNEL)))) {
> + hw_fib_p = hw_fib_pool;
> + fib_p = fib_pool;
> + while (hw_fib_p < &hw_fib_pool
> [num]) {
> + if (!(*(hw_fib_p++) =
> kmalloc(sizeof(struct hw_fib), GFP_ATOMIC|GFP_KERNEL))) {
> + --hw_fib_p;
> + break;
> + }
> + if (!(*(fib_p++) =
> kmalloc(sizeof(struct fib), GFP_ATOMIC|GFP_KERNEL))) {
all of these allocations should be either GFP_ATOMIC or GFP_KERNEL, but
not both (it looks like they should be GFP_KERNEL). What GFP_KERNEL|
GFP_ATOMIC actually gives you is a potentially sleeping allocation that
will exhaust the emergency pools, which sounds really undesirable.
James
-
: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
