Re: [PATCH v5 01/11] block: make generic_make_request handle arbitrarily sized bios

[Ming Lin <mlin@kernel.org>]

On Mon, 2015-08-10 at 12:22 -0400, Martin K. Petersen wrote:
> >>>>> "Mike" == Mike Snitzer <snitzer@redhat.com> writes:
> 
> Mike> Shouldn't we also be using MAX_BIO_SECTORS in
> Mike> blkdev_issue_write_same (instead of UINT_MAX >> 9)?
> 
> The granularity of WRITE SAME is always 1 logical block and there is no
> reason to round it down to the next power of two.
> 
> +/*
> + * Ensure that max discard sectors doesn't overflow bi_size and hopefully
> + * it is of the proper granularity as long as the granularity is a power
> + * of two.
> + */
> +#define MAX_DISCARD_SECTORS ((1U << 31) >> 9)
> +
> 
> That's fine for SATA since we're already capping at 2TB minus change.
> But it means we'll be capping unnecessarily on SCSI. And larger range
> counts are impending in SATA as well.
> 
> So this goes back to my original comment: The only place there can be a
> discard granularity is for SCSI UNMAP. And consequently, we should only
> enforce alignment and granularity when that code path is taken in sd.c.
> 
> I'm OK with Ming's patch series in general. Let's leave the discard cap
> at UINT_MAX and I'll twiddle the rest in SCSI.

Just to make sure I didn't misunderstand it.

Did you mean still use (UINT_MAX >> 9) in blkdev_issue_discard()?

	req_sects = min_t(sector_t, nr_sects, UINT_MAX >> 9);
instead of:
	req_sects = min_t(sector_t, nr_sects, MAX_DISCARD_SECTORS);

But that doesn't work for dm-thinp. See Kent's suggestion to use 1<<31.
https://www.redhat.com/archives/dm-devel/2015-August/msg00053.html

> 
> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
> 

Thanks!