Re: [PATCH 15/18] scsi: allocate scsi_cmnd structures as part of struct request

On Wed, 2017-01-25 at 18:25 +0100, Christoph Hellwig wrote:

-unsigned char *scsi_alloc_sense_buffer(struct Scsi_Host *shost, gfp_t gf=

p_mask,

-=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0int numa_node)
+static unsigned char *scsi_alloc_sense_buffer(struct Scsi_Host *shost,
+=A0=A0=A0=A0=A0=A0=A0gfp_t gfp_mask, int numa_node)
=A0{
=A0=A0=A0=A0=A0=A0=A0=A0return kmem_cache_alloc_node(scsi_select_sense_ca=

che(shost), gfp_mask,

=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0n=

uma_node);

nt error,

=A0
=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0if (bidi_bytes)
=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0s=

csi_release_bidi_buffers(cmd);

+=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0scsi_release_buffers(cmd);
+=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0scsi_put_command(cmd);
=A0
=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0spin_lock_irqsave(q->queu=

e_lock, flags);

=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0blk_finish_request(req, e=

rror);

=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0spin_unlock_irqrestore(q-=
queue_lock, flags);
=A0
-=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0scsi_release_buffers(cmd);
-
-=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0scsi_put_command(cmd);
=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0=A0scsi_run_queue(q);
=A0=A0=A0=A0=A0=A0=A0=A0}

Hello Christoph,

Why=A0have the=A0scsi_release_buffers() and scsi_put_command(cmd) calls bee=
n
moved up? I haven't found an explanation for this change in the patch
description.

Please also consider to remove the cmd->request->special =3D NULL assignmen=
ts
via this patch. Since this patch makes the lifetime of struct scsi_cmnd and
struct request identical these assignments are no longer needed.

This patch introduces the function scsi_exit_rq(). Having two functions
for the single-queue path that release resources (scsi_release_buffers()
and scsi_exit_rq()) is confusing. Since every scsi_release_buffers() call
is followed by a blk_unprep_request() call, have you considered to move
the scsi_release_buffers() call into scsi_unprep_fn() via an additional
patch?

Thanks,

Bart.