Re: [PATCH] kernel: Export mm_access

From: Andrew Morton
Date: Tue May 23 2017 - 17:14:25 EST

On Tue, 23 May 2017 16:17:29 -0400 Felix Kuehling <felix.kuehling@xxxxxxx> wrote:

> From: Harish Kasiviswanathan <Harish.Kasiviswanathan@xxxxxxx>
> Exporting mm_access, which is required for implementing accelerated
> equivalents of process_vm_readv/process_vm_writev for GPU memory in KFD
> (AMD's GPU compute driver). This allows us to apply all the same remote
> process memory access policies.
> Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@xxxxxxx>
> Reviewed-by: Felix Kuehling <Felix.Kuehling@xxxxxxx>
> ---
> Current KFD with AMD discrete GPU support is not yet upstream, but we
> are working on getting it upstream for 4.13 or 4.14 depending on how we
> line up with Dave Airlie's merge windows.
> For reference, recent releases of ROCm (AMD's Radeon Open Compute
> Platform software stack) with fully open-source user mode and kernel
> code can be found on GitHub:

Well if we're going to export it, I suppose we should actually document
it. Could you please include some commentary above mm_access() which
describes what it does, why it does it, return value, etc?

Then please include this patch in whatever tree contains the KFD code,
with my Acked-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>.

> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -1060,6 +1060,7 @@ struct mm_struct *mm_access(struct task_struct
> *task, unsigned int mode)
> return mm;
> }
> +EXPORT_SYMBOL_GPL(mm_access);
> static void complete_vfork_done(struct task_struct *tsk)
> {

fyi, your email client is word-wrapping the patches.