Re: [PATCH] [2.4.20-pre10] dpt_i2o fix

[Alan Cox]

On Tue, 2002-10-15 at 14:40, Ducrot Bruno wrote:
> The first chunk of attached patch alloc wait_data with
> kmalloc(..., GFP_ATOMIC) instead of GPF_KERNEL
> in the function adpt_i2o_post_wait() because this function
> is called in the function adpt_i2o_passthru() (line 1717 or
> so) but with a spin_lock held.

Given the nature of the usage I think its probably better to fix it
properly than to just hack it up to be GFP_ATOMIC. If the caller was to
pass in the wait_data buffer then the problem could be fixed cleanly