Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
Hi Pankaj,
On 9/24/2022 1:36 AM, Pankaj Raghav wrote:
Hi Jens,
Please consider this patch series for the 6.1 release.
- Background and Motivation:
The zone storage implementation in Linux, introduced since v4.10, first
targetted SMR drives which have a power of 2 (po2) zone size alignment
requirement. The po2 zone size was further imposed implicitly by the
block layer's blk_queue_chunk_sectors(), used to prevent IO merging
across chunks beyond the specified size, since v3.16 through commit
762380ad9322 ("block: add notion of a chunk size for request merging").
But this same general block layer po2 requirement for blk_queue_chunk_sectors()
was removed on v5.10 through commit 07d098e6bbad ("block: allow 'chunk_sectors'
to be non-power-of-2").
NAND, which is the media used in newer zoned storage devices, does not
naturally align to po2. In these devices, zone capacity(cap) is not the
same as the po2 zone size. When the zone cap != zone size, then unmapped
LBAs are introduced to cover the space between the zone cap and zone size.
po2 requirement does not make sense for these type of zone storage devices.
This patch series aims to remove these unmapped LBAs for zoned devices when
zone cap is npo2. This is done by relaxing the po2 zone size constraint
in the kernel and allowing zoned device with npo2 zone sizes if zone cap
== zone size.
I came across function sd_zbc_check_capacity() in sd_zbc.c, it still
errors out in case of npo2.
I don't see this series touching sd_zbc.c. Is there plan or existing
change to relax this check?
if(!*is_power_of_2*
(*zone_blocks*
)){
*sd_printk*
(*KERN_ERR*
,sdkp,
"Zone size %llu is not a power of two.\n",
*zone_blocks*
);
return-*EINVAL* ;
}
Thanks.
Regards,
Can Guo.
--
dm-devel mailing list
dm-devel@redhat.com
https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 9/30/22 14:24, Jens Axboe wrote: Noted. I'll find some time to review this as well separately, once we're on the other side of the merge window. Hi Jens, Now that we are on the other side of the merge window: do you perhaps want Pankaj to repost this patch series? From what I have heard in several fora (JEDEC, SNIA) all flash storage vendors except one (WDC) are in favor of a contiguous LBA space and hence are in favor of supporting zone sizes that are not a power of two. As you may know in JEDEC we are working on standardizing zoned storage for UFS devices. We (JEDEC JC-64.1 committee members) would like to know whether or not we should require that the UFS zone size should be a power of two. Thank you, Bart. -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 9/30/22 17:45, Damien Le Moal wrote: On 10/1/22 04:38, Bart Van Assche wrote: Since this has not been mentioned in the cover letter, I want to add that in the near future we will need these patches for Android devices. JEDEC is working on supporting zoned storage for UFS devices, the storage devices used in all modern Android phones. Although it would be possible to make the offset between zone starts a power of two by inserting gap zones between data zones, UFS vendors asked not to do this and hence need support for zone sizes that are not a power of two. An advantage of not having to deal with gap zones is better filesystem performance since filesystem extents cannot span gap zones. Having to split filesystem extents because of gap zones reduces filesystem performance. As mentioned many times, my opinion is that a good implementation should *not* have any extent span zone boundaries. So personally, I do not consider such argument as a valid justification for the non-power-of-2 zone size support. Hi Damien, Although the filesystem extent issue probably can be solved in software, the argument that UFS vendors strongly prefer not to have gap zones and hence need support for zone sizes that are not a power of two remains. Thanks, Bart. -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 10/1/22 04:38, Bart Van Assche wrote: > On 9/30/22 08:13, Jens Axboe wrote: >> On 9/29/22 12:31 AM, Pankaj Raghav wrote: Hi Jens, Please consider this patch series for the 6.1 release. >>> >>> Hi Jens, Christoph, and Keith, >>> All the patches have a Reviewed-by tag at this point. Can we queue this up >>> for 6.1? >> >> It's getting pretty late for 6.1 and I'd really like to have both Christoph >> and Martin sign off on these changes. > > Hi Jens, > > Agreed that it's getting late for 6.1. > > Since this has not been mentioned in the cover letter, I want to add > that in the near future we will need these patches for Android devices. > JEDEC is working on supporting zoned storage for UFS devices, the > storage devices used in all modern Android phones. Although it would be > possible to make the offset between zone starts a power of two by > inserting gap zones between data zones, UFS vendors asked not to do this > and hence need support for zone sizes that are not a power of two. An > advantage of not having to deal with gap zones is better filesystem > performance since filesystem extents cannot span gap zones. Having to > split filesystem extents because of gap zones reduces filesystem > performance. As mentioned many times, my opinion is that a good implementation should *not* have any extent span zone boundaries. So personally, I do not consider such argument as a valid justification for the non-power-of-2 zone size support. > > Thanks, > > Bart. > > -- Damien Le Moal Western Digital Research -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 9/30/22 1:38 PM, Bart Van Assche wrote: > On 9/30/22 08:13, Jens Axboe wrote: >> On 9/29/22 12:31 AM, Pankaj Raghav wrote: Hi Jens, ?? Please consider this patch series for the 6.1 release. >>> >>> Hi Jens, Christoph, and Keith, >>> ? All the patches have a Reviewed-by tag at this point. Can we queue this up >>> for 6.1? >> >> It's getting pretty late for 6.1 and I'd really like to have both Christoph >> and Martin sign off on these changes. > > Hi Jens, > > Agreed that it's getting late for 6.1. > > Since this has not been mentioned in the cover letter, I want to add > that in the near future we will need these patches for Android > devices. JEDEC is working on supporting zoned storage for UFS devices, > the storage devices used in all modern Android phones. Although it > would be possible to make the offset between zone starts a power of > two by inserting gap zones between data zones, UFS vendors asked not > to do this and hence need support for zone sizes that are not a power > of two. An advantage of not having to deal with gap zones is better > filesystem performance since filesystem extents cannot span gap zones. > Having to split filesystem extents because of gap zones reduces > filesystem performance. Noted. I'll find some time to review this as well separately, once we're on the other side of the merge window. -- Jens Axboe -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 9/30/22 08:13, Jens Axboe wrote: On 9/29/22 12:31 AM, Pankaj Raghav wrote: Hi Jens, Please consider this patch series for the 6.1 release. Hi Jens, Christoph, and Keith, All the patches have a Reviewed-by tag at this point. Can we queue this up for 6.1? It's getting pretty late for 6.1 and I'd really like to have both Christoph and Martin sign off on these changes. Hi Jens, Agreed that it's getting late for 6.1. Since this has not been mentioned in the cover letter, I want to add that in the near future we will need these patches for Android devices. JEDEC is working on supporting zoned storage for UFS devices, the storage devices used in all modern Android phones. Although it would be possible to make the offset between zone starts a power of two by inserting gap zones between data zones, UFS vendors asked not to do this and hence need support for zone sizes that are not a power of two. An advantage of not having to deal with gap zones is better filesystem performance since filesystem extents cannot span gap zones. Having to split filesystem extents because of gap zones reduces filesystem performance. Thanks, Bart. -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
On 9/29/22 12:31 AM, Pankaj Raghav wrote: >> Hi Jens, >> Please consider this patch series for the 6.1 release. >> > > Hi Jens, Christoph, and Keith, > All the patches have a Reviewed-by tag at this point. Can we queue this up > for 6.1? It's getting pretty late for 6.1 and I'd really like to have both Christoph and Martin sign off on these changes. -- Jens Axboe -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
Re: [dm-devel] [PATCH v15 00/13] support zoned block devices with non-power-of-2 zone sizes
> Hi Jens, > Please consider this patch series for the 6.1 release. > Hi Jens, Christoph, and Keith, All the patches have a Reviewed-by tag at this point. Can we queue this up for 6.1? -- Pankaj -- dm-devel mailing list dm-devel@redhat.com https://listman.redhat.com/mailman/listinfo/dm-devel
