two generic block layer fixes for 5.9
cover two generic block layer fixes for 5.9
Hi Jens,
These two patches are posted for a while, and have reviewed by several other developers. Could you please to take them for Linux v5.9 ?
Thanks in advance.
Coly Li
---
Coly Li (2):
block: change REQ_OP_ZONE_RESET and REQ_OP_ZONE_RESET_ALL to be odd
numbers
block: improve discard bio alignment in __blkdev_issue_discard()
block/blk-lib.c | 25 +++++++++++++++++++++++-- block/blk.h | 14 ++++++++++++++ include/linux/blk_types.h | 8 ++++---- 3 files changed, 41 insertions(+), 6 deletions(-)
Signature · 1 line
2.26.2
PATCH 1/2 block: change REQ_OP_ZONE_RESET and REQ_OP_ZONE_RESET_ALL to be odd numbers
Currently REQ_OP_ZONE_RESET and REQ_OP_ZONE_RESET_ALL are defined as even numbers 6 and 8, such zone reset bios are treated as READ bios by bio_data_dir(), which is obviously misleading.
The macro bio_data_dir() is defined in include/linux/bio.h as,
55 #define bio_data_dir(bio) \
56 (op_is_write(bio_op(bio)) ? WRITE : READ)
And op_is_write() is defined in include/linux/blk_types.h as, 397 static inline bool op_is_write(unsigned int op) 398 {
399 return (op & 1);
400 }
The convention of op_is_write() is when there is data transfer then the op code should be odd number, and treat as a write op. bio_data_dir() treats all bio direction as READ if op_is_write() reports false, and WRITE if op_is_write() reports true.
Because REQ_OP_ZONE_RESET and REQ_OP_ZONE_RESET_ALL are even numbers, although they don't transfer data but reporting them as READ bio by bio_data_dir() is misleading and might be wrong. Because these two commands will reset the writer pointers of the resetting zones, and all content after the reset write pointer will be invalid and unaccessible, obviously they are not READ bios in any means.
This patch changes REQ_OP_ZONE_RESET from 6 to 15, and changes REQ_OP_ZONE_RESET_ALL from 8 to 17. Now bios with these two op code can be treated as WRITE by bio_data_dir(). Although they don't transfer data, now we keep them consistent with REQ_OP_DISCARD and REQ_OP_WRITE_ZEROES with the ituition that they change on-media content and should be WRITE request.
--- include/linux/blk_types.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
include/linux/blk_types.h +4 -4
@@ -300,12 +300,8 @@ enum req_opf {REQ_OP_DISCARD =3 ,/* securely erase sectors */ REQ_OP_SECURE_ERASE =5 , -/* reset a zone write pointer */ -REQ_OP_ZONE_RESET =6 ,/* write the same sector many times */ REQ_OP_WRITE_SAME =7 , -/* reset all the zone present on the device */ -REQ_OP_ZONE_RESET_ALL =8 ,/* write the zero filled sector many times */ REQ_OP_WRITE_ZEROES = 9,/* Open a zone */
@@ -316,6 +312,10 @@ enum req_opf {REQ_OP_ZONE_FINISH =12 ,/* write data at the current zone write pointer */ REQ_OP_ZONE_APPEND =13 , +/* reset a zone write pointer */ +REQ_OP_ZONE_RESET =15 , +/* reset all the zone present on the device */ +REQ_OP_ZONE_RESET_ALL =17 ,/* SCSI passthrough using struct scsi_request */ REQ_OP_SCSI_IN = 32,
Signature · 1 line
2.26.2
Re: [PATCH 1/2] block: change REQ_OP_ZONE_RESET and REQ_OP_ZONE_RESET_ALL to be odd numbers
Quoted 26 lines · Coly Li
Currently
REQ_OP_ZONE_RESETandREQ_OP_ZONE_RESET_ALLare defined as even numbers 6 and 8, such zone reset bios are treated as READ bios bybio_data_dir(), which is obviously misleading.The macro
bio_data_dir()is defined ininclude/linux/bio.has,
55 #definebio_data_dir(bio)\
56 (op_is_write(bio_op(bio)) ? WRITE : READ)And
op_is_write()is defined ininclude/linux/blk_types.has, 397 static inline boolop_is_write(unsigned int op)398 {
399 return (op & 1);
400 }The convention of
op_is_write()is when there is data transfer then the op code should be odd number, and treat as a write op.bio_data_dir()treats all bio direction as READ ifop_is_write()reports false, and WRITE ifop_is_write()reports true.Because
REQ_OP_ZONE_RESETandREQ_OP_ZONE_RESET_ALLare even numbers, although they don't transfer data but reporting them as READ bio bybio_data_dir()is misleading and might be wrong. Because these two commands will reset the writer pointers of the resetting zones, and all content after the reset write pointer will be invalid and unaccessible, obviously they are not READ bios in any means.
Quoted 5 lines
This patch changes
REQ_OP_ZONE_RESETfrom 6 to 15, and changesREQ_OP_ZONE_RESET_ALLfrom 8 to 17. Now bios with these two op code can be treated as WRITE bybio_data_dir(). Although they don't transfer data, now we keep them consistent withREQ_OP_DISCARDandREQ_OP_WRITE_ZEROESwith the ituition that they change on-media content
s/ituition/assumption
Quoted 34 lines
and should be WRITE request.
Signed-off-by Coly Li <[email protected]>Reviewed-by Damien Le Moal <[email protected]>Reviewed-by Chaitanya Kulkarni <[email protected]>Cc Christoph Hellwig <[email protected]>Cc Hannes Reinecke <[email protected]>Cc Jens Axboe <[email protected]>Cc Johannes Thumshirn <[email protected]>Cc Keith Busch <[email protected]>Cc Shaun Tancheff <[email protected]>--- include/linux/blk_types.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)include/linux/blk_types.h +0
-4@@ -300,12 +300,8 @@ enum req_opf {REQ_OP_DISCARD =3 ,/* securely erase sectors */ REQ_OP_SECURE_ERASE =5 , -/* reset a zone write pointer */ -REQ_OP_ZONE_RESET =6 ,/* write the same sector many times */ REQ_OP_WRITE_SAME =7 , -/* reset all the zone present on the device */ -REQ_OP_ZONE_RESET_ALL =8 ,/* write the zero filled sector many times */ REQ_OP_WRITE_ZEROES = 9,/* Open a zone */
include/linux/blk_types.h enum req_opf { · line 321
Quoted 10 lines
include/linux/blk_types.h +4
-0@@ -316,6 +312,10 @@ enum req_opf {REQ_OP_ZONE_FINISH =12 ,/* write data at the current zone write pointer */ REQ_OP_ZONE_APPEND =13 , +/* reset a zone write pointer */ +REQ_OP_ZONE_RESET =15 , +/* reset all the zone present on the device */ +REQ_OP_ZONE_RESET_ALL =17 ,/* SCSI passthrough using struct scsi_request */ REQ_OP_SCSI_IN = 32,
Looks good to me. Zone reset is very similar to a discard, albeit stronger (zone reset is not a hint). So defining these ops as having the same data dir makes sense.
Signature · 2 lines
Damien Le Moal Western Digital Research
PATCH 2/2 block: improve discard bio alignment in __blkdev_issue_discard()
In reply to Coly LiThis patch improves discard bio split for address and size alignment in __blkdev_issue_discard(). The aligned discard bio may help underlying device controller to perform better discard and internal garbage collection, and avoid unnecessary internal fragment.
Current discard bio split algorithm in __blkdev_issue_discard() may have non-discarded fregment on device even the discard bio LBA and size are both aligned to device's discard granularity size.
Here is the example steps on how to reproduce the above problem.
- On a VMWare ESXi 6.5 update3 installation, create a 51GB virtual disk with thin mode and give it to a Linux virtual machine.
- Inside the Linux virtual machine, if the 50GB virtual disk shows up as /dev/sdb, fill data into the first 50GB by,
# dd if=/dev/zero of=/dev/sdb bs=4096 count=13107200
- Discard the 50GB range from offset 0 on /dev/sdb,
# blkdiscard /dev/sdb -o 0 -l 53687091200
- Observe the underlying mapping status of the device
# sg_get_lba_status /dev/sdb -m 1048 --lba=0
descriptor LBA: 0x0000000000000000 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000000000800 blocks: 16773120 deallocated
descriptor LBA: 0x0000000000fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000001000000 blocks: 8386560 deallocated
descriptor LBA: 0x00000000017ff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000001800000 blocks: 8386560 deallocated
descriptor LBA: 0x0000000001fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000002000000 blocks: 8386560 deallocated
descriptor LBA: 0x00000000027ff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000002800000 blocks: 8386560 deallocated
descriptor LBA: 0x0000000002fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000003000000 blocks: 8386560 deallocated
descriptor LBA: 0x00000000037ff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000003800000 blocks: 8386560 deallocated
descriptor LBA: 0x0000000003fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000004000000 blocks: 8386560 deallocated
descriptor LBA: 0x00000000047ff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000004800000 blocks: 8386560 deallocated
descriptor LBA: 0x0000000004fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000005000000 blocks: 8386560 deallocated
descriptor LBA: 0x00000000057ff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000005800000 blocks: 8386560 deallocated
descriptor LBA: 0x0000000005fff800 blocks: 2048 mapped (or unknown)
descriptor LBA: 0x0000000006000000 blocks: 6291456 deallocated
descriptor LBA: 0x0000000006600000 blocks: 0 deallocatedAlthough the discard bio starts at LBA 0 and has 50<<30 bytes size which are perfect aligned to the discard granularity, from the above list these are many 1MB (2048 sectors) internal fragments exist unexpectedly.
The problem is in __blkdev_issue_discard(), an improper algorithm causes an improper bio size which is not aligned.
25 int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,
26 sector_t nr_sects, gfp_t gfp_mask, int flags,
27 struct bio **biop)
28 {
29 struct request_queue *q = bdev_get_queue(bdev);
[snipped]
56
57 while (nr_sects) {
58 sector_t req_sects = min_t(sector_t, nr_sects,
59 bio_allowed_max_sectors(q));
60
61 WARN_ON_ONCE((req_sects << 9) > UINT_MAX);
62
63 bio = blk_next_bio(bio, 0, gfp_mask); 64 bio->bi_iter.bi_sector = sector; 65 bio_set_dev(bio, bdev); 66 bio_set_op_attrs(bio, op, 0); 67 68 bio->bi_iter.bi_size = req_sects << 9; 69 sector += req_sects; 70 nr_sects -= req_sects;
[snipped]
79 }
80
81 *biop = bio;
82 return 0;
83 }
84 EXPORT_SYMBOL(__blkdev_issue_discard);
At line 58-59, to discard a 50GB range, req_sects is set as return value of bio_allowed_max_sectors(q), which is 8388607 sectors. In the above case, the discard granularity is 2048 sectors, although the start LBA and discard length are aligned to discard granularity, req_sects never has chance to be aligned to discard granularity. This is why there are some still-mapped 2048 sectors fragment in every 4 or 8 GB range.
If req_sects at line 58 is set to a value aligned to discard_granularity and close to UNIT_MAX, then all consequent split bios inside device driver are (almostly) aligned to discard_granularity of the device queue. The 2048 sectors still-mapped fragment will disappear.
This patch introduces bio_aligned_discard_max_sectors() to return the the value which is aligned to q->limits.discard_granularity and closest to UINT_MAX. Then this patch replaces bio_allowed_max_sectors() with this new routine to decide a more proper split bio length.
But we still need to handle the situation when discard start LBA is not aligned to q->limits.discard_granularity, otherwise even the length is aligned, current code may still leave 2048 fragment around every 4GB range. Therefore, to calculate req_sects, firstly the start LBA of discard range is checked, if it is not aligned to discard granularity, the first split location should make sure following bio has bi_sector aligned to discard granularity. Then there won't be still-mapped fragment in the middle of the discard range.
The above is how this patch improves discard bio alignment in __blkdev_issue_discard(). Now with this patch, after discard with same command line mentiond previously, sg_get_lba_status returns, descriptor LBA: 0x0000000000000000 blocks: 106954752 deallocated descriptor LBA: 0x0000000006600000 blocks: 0 deallocated
We an see there is no 2048 sectors segment anymore, everything is clean.
--- block/blk-lib.c | 25 +++++++++++++++++++++++-- block/blk.h | 14 ++++++++++++++ 2 files changed, 37 insertions(+), 2 deletions(-)
block/blk-lib.c +23 -2
@@ -55,8 +55,29 @@ int __blkdev_issue_discard(struct block_device *bdev, sector_t sector,return -EINVAL ;while (nr_sects) { -sector_t req_sects =min_t (sector_t, nr_sects, -bio_allowed_max_sectors (q)); +sector_t granularity_aligned_lba; +sector_t req_sects; + + granularity_aligned_lba =round_up (sector, + q->limits.discard_granularity >>SECTOR_SHIFT ); + +/* +* Check whether the discard bio starts at a discard_granularity +* aligned LBA, +* - If no: set (granularity_aligned_lba - sector) to bi_size of +* the first split bio, then the second bio will start at a +* discard_granularity aligned LBA. +* - If yes: use bio_aligned_discard_max_sectors() as the max +* possible bi_size of the first split bio. Then when this bio +* is split in device drive, the split ones are very probably +* to be aligned to discard_granularity of the device's queue. +*/ +if (granularity_aligned_lba == sector) + req_sects =min_t (sector_t, nr_sects, +bio_aligned_discard_max_sectors (q)); +else + req_sects =min_t (sector_t, nr_sects, + granularity_aligned_lba - sector);WARN_ON_ONCE ((req_sects <<9 ) >UINT_MAX );
block/blk.h +14 -0
@@ -281,6 +281,20 @@ static inline unsigned int bio_allowed_max_sectors(struct request_queue *q)return round_down (UINT_MAX ,queue_logical_block_size (q)) >>9 ; } +/* +* The max bio size which is aligned to q->limits.discard_granularity. This +* is a hint to split large discard bio in generic block layer, then if device +* driver needs to split the discard bio into smaller ones, their bi_size can +* be very probably and easily aligned to discard_granularity of the device's +* queue. +*/ +static inline unsigned int bio_aligned_discard_max_sectors ( +struct request_queue *q) +{ +return round_down (UINT_MAX , q->limits.discard_granularity) >> +SECTOR_SHIFT ; +} +/* * Internal io_context interface */
Signature · 1 line
2.26.2
Re: [PATCH 2/2] block: improve discard bio alignment in __blkdev_issue_discard()
Hi Coly!
Quoted 4 lines
This patch improves discard bio split for address and size alignment in
__blkdev_issue_discard(). The aligned discard bio may help underlying device controller to perform better discard and internal garbage collection, and avoid unnecessary internal fragment.
If the aim is to guarantee that all discard requests, except for head and tail, are aligned multiples of the discard_granularity, you also need to take the discard_alignment queue limit and the partition offset into consideration.
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Re: [PATCH 2/2] block: improve discard bio alignment in __blkdev_issue_discard()
Quoted 8 lines · Martin K. Petersen
Hi Coly!
Quoted 4 lines
This patch improves discard bio split for address and size alignment in
__blkdev_issue_discard(). The aligned discard bio may help underlying device controller to perform better discard and internal garbage collection, and avoid unnecessary internal fragment.
Hi Martin,
Quoted 5 lines
If the aim is to guarantee that all discard requests, except for head and tail, are aligned multiples of the
discard_granularity, you also need to take thediscard_alignmentqueue limit and the partition offset into consideration.
The discard_granularity was considered and my though is, discard_alignment normally is multiples of discard granularity, if the discard bio bi_sector and bi_size are aligned to discard granularity, when underlying driver splits its discard bio by discard_alignment, the split bio bi_sector and bi_size can still be aligned to discard granularity. Another reason I don't handle discard_alignment alignment in __blkdev_issue_discard() is performance. Handling discard_alignment bio split in driver's split loop may call blk_next_bio() much less in __blkdev_issue_discard(), and be more friendly for memory and cache footprint.
For the partition offset, my original idea was to suggest the partition or dm target starts on offset 2048 sectors. But your opinion sound better, if considering partition or target offset maybe users misconfigured the partition offset may also gain the benefit of discard alignment.
Let me try to improve a v3 patch to handle the partition offset too.
Thanks for the cool idea :-)
Coly Li