Re: [PATCH 18/24] scsi_ioctl: move the "block layer" SCSI ioctl handling to drivers/scsi

On Mon, 23 Aug 2021 08:49:36 +0200
Christoph Hellwig [off-list ref] wrote:

On Mon, Aug 23, 2021 at 08:43:16AM +0200, Halil Pasic wrote:

quoted

I believe there is a small problem with this patch. I think it is
easiest to explain with the diff that fixes it. Please see the patch
at the end of this email.

Otherwise your patch looks great!

This may or may not be related to the problem reported here:
https://lkml.org/lkml/2021/7/29/157
Adding Oliver, maybe he can test if this fixes his testcases as well.
(It did fix ours.:)

If you like I can respin my fix with an extended patch description.  

No this looks good, but to make sure Martin picks it up please send it
as a separate thread.  It would be great it this fixes Olives issue,
but at least on my Debian systems blkid don't even call into SG_IO.
But maybe he has a different one or it is a cascading effect on that
particular setup.

Reviewed-by: Christoph Hellwig <hch@lst.de>

This patch is directly based on f2542a3be327 ("scsi: scsi_ioctl: Move
the "block layer" SCSI ioctl handling to drivers/scsi") from linux-next, 
and a simple rebase onto the tip of linux-next does not work because the
block of code I'm about to modify got factored out into a function
called scsi_ioctl_sg_io().

I'm not sure about the process of fixes in linux-next, so can please
somebody (Christoph, Martin) tell me against what base should I post the
respin (in a separate thread)?

Thanks in advance!

Halil