brd: Allow to change block sizes
Subsystems (4)
cover brd: Allow to change block sizes
Hi all,
meat to the bone: with this patchset one can change the physical and logical block size of the 'brd' ramdisk driver.
Default is 512 (for both); one can easily increase the physical block size to 16k and the logical block size to 4k.
Increasing the logcial block size beyond 4k gives some 'interesting' crashes.
It should also be possible to use the resulting ram disk as a backend for nvme target, thereby exercising the NVMe stack for larger block sizes, too.
Happy hacking!
Hannes Reinecke (5):
brd: convert to folios
brd: abstract page_size conventions
brd: make sector size configurable
brd: limit maximal block size to 32M
brd: make logical sector size configurable
drivers/block/brd.c | 244 ++++++++++++++++++++++++-------------------- 1 file changed, 136 insertions(+), 108 deletions(-)
Signature · 1 line
2.35.3
PATCH 1/5 brd: convert to folios
Convert the driver to work on folios instead of pages.
--- drivers/block/brd.c | 171 ++++++++++++++++++++++---------------------- 1 file changed, 85 insertions(+), 86 deletions(-)
drivers/block/brd.c +85 -86
@@ -28,8 +28,8 @@#include <linux/uaccess.h> /* -* Each block ramdisk device has a radix_tree brd_pages of pages that stores -* the pages containing the block device's contents. A brd page's ->index is +* Each block ramdisk device has a radix_tree brd_folios of folios that stores +* the folios containing the block device's contents. A brd folio's ->index is * its offset in PAGE_SIZE units. This is similar to, but in no way connected * with, the kernel's pagecache or buffer cache (which sit above our block * device).
@@ -40,25 +40,25 @@ struct brd_device {struct list_head brd_list;/* -* Backing store of pages and lock to protect it. This is the contents -* of the block device. +* Backing store of folios and lock to protect it. +* This is the contents of the block device. */ spinlock_t brd_lock; -struct radix_tree_root brd_pages; -u64 brd_nr_pages; +struct radix_tree_root brd_folios; +u64 brd_nr_folios; };/* -* Look up and return a brd's page for a given sector. +* Look up and return a brd's folio for a given sector. */ -static struct page *brd_lookup_page (struct brd_device *brd,sector_t sector) +static struct folio *brd_lookup_folio (struct brd_device *brd,sector_t sector) {pgoff_t idx; -struct page *page; +struct folio *folio;/* -* The page lifetime is protected by the fact that we have opened the -* device node -- brd pages will never be deleted under us, so we +* The folio lifetime is protected by the fact that we have opened the +* device node -- brd folios will never be deleted under us, so we * don't need any further locking or refcounting. * * This is strictly true for the radix-tree nodes as well (ie. we
@@ -68,49 +68,49 @@ static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector)* here, only deletes). */ rcu_read_lock (); - idx = sector >>PAGE_SECTORS_SHIFT ;/* sector to page index */ - page =radix_tree_lookup (&brd->brd_pages, idx); + idx = sector >>PAGE_SECTORS_SHIFT ;/* sector to folio index */ + folio =radix_tree_lookup (&brd->brd_folios, idx);rcu_read_unlock (); -BUG_ON (page && page->index != idx); +BUG_ON (folio && folio->index != idx); -return page; +return folio; }/* -* Insert a new page for a given sector, if one does not already exist. +* Insert a new folio for a given sector, if one does not already exist. */ -static int brd_insert_page (struct brd_device *brd,sector_t sector,gfp_t gfp) +static int brd_insert_folio (struct brd_device *brd,sector_t sector,gfp_t gfp) {pgoff_t idx; -struct page *page; +struct folio *folio;int ret =0 ; - page =brd_lookup_page (brd, sector); -if (page) + folio =brd_lookup_folio (brd, sector); +if (folio)return 0 ; - page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)return -ENOMEM ;if (radix_tree_maybe_preload (gfp)) { -__free_page (page); +folio_put (folio);return -ENOMEM ; }spin_lock (&brd->brd_lock); idx = sector >>PAGE_SECTORS_SHIFT ; - page->index = idx; -if (radix_tree_insert (&brd->brd_pages, idx, page)) { -__free_page (page); - page =radix_tree_lookup (&brd->brd_pages, idx); -if (!page) + folio->index = idx; +if (radix_tree_insert (&brd->brd_folios, idx, folio)) { +folio_put (folio); + folio =radix_tree_lookup (&brd->brd_folios, idx); +if (!folio) ret = -ENOMEM ; -else if (page->index != idx) +else if (folio->index != idx) ret = -EIO ; }else { - brd->brd_nr_pages++; + brd->brd_nr_folios++; }spin_unlock (&brd->brd_lock);
@@ -119,30 +119,30 @@ static int brd_insert_page(struct brd_device *brd, sector_t sector, gfp_t gfp) }/* -* Free all backing store pages and radix tree. This must only be called when +* Free all backing store folios and radix tree. This must only be called when * there are no other users of the device. */ #define FREE_BATCH 16 -static void brd_free_pages (struct brd_device *brd) +static void brd_free_folios (struct brd_device *brd) {unsigned long pos =0 ; -struct page *pages[FREE_BATCH ]; -int nr_pages; +struct folio *folios[FREE_BATCH ]; +int nr_folios; do { int i; - nr_pages = radix_tree_gang_lookup(&brd->brd_pages, - (void **)pages, pos, FREE_BATCH); + nr_folios = radix_tree_gang_lookup(&brd->brd_folios, + (void **)folios, pos, FREE_BATCH); - for (i = 0; i < nr_pages; i++) { + for (i = 0; i < nr_folios; i++) { void *ret; - BUG_ON(pages[i]->index < pos); - pos = pages[i]->index; - ret = radix_tree_delete(&brd->brd_pages, pos); - BUG_ON(!ret || ret != pages[i]); - __free_page(pages[i]); + BUG_ON(folios[i]->index < pos); + pos = folios[i]->index; + ret = radix_tree_delete(&brd->brd_folios, pos); + BUG_ON(!ret || ret != folios[i]); + folio_put(folios[i]); } pos++;
@@ -155,10 +155,10 @@ static void brd_free_pages(struct brd_device *brd)/* * This assumes radix_tree_gang_lookup always returns as -* many pages as possible. If the radix-tree code changes, +* many folios as possible. If the radix-tree code changes, * so will this have to. */ - }while (nr_pages ==FREE_BATCH ); + }while (nr_folios ==FREE_BATCH ); }/*
@@ -172,12 +172,12 @@ static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n,int ret; copy =min_t (size_t, n,PAGE_SIZE - offset); - ret =brd_insert_page (brd, sector, gfp); + ret =brd_insert_folio (brd, sector, gfp);if (ret)return ret;if (copy < n) { sector += copy >>SECTOR_SHIFT ; - ret =brd_insert_page (brd, sector, gfp); + ret =brd_insert_folio (brd, sector, gfp); }return ret; }
@@ -188,29 +188,29 @@ static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n,static void copy_to_brd (struct brd_device *brd,const void *src,sector_t sector,size_t n) { -struct page *page; +struct folio *folio;void *dst;unsigned int offset = (sector & (PAGE_SECTORS -1 )) <<SECTOR_SHIFT ;size_t copy; copy =min_t (size_t, n,PAGE_SIZE - offset); - page =brd_lookup_page (brd, sector); -BUG_ON (!page); + folio =brd_lookup_folio (brd, sector); +BUG_ON (!folio); - dst =kmap_atomic (page); -memcpy (dst + offset, src, copy); -kunmap_atomic (dst); + dst =kmap_local_folio (folio, offset); +memcpy (dst, src, copy); +kunmap_local (dst);if (copy < n) { src += copy; sector += copy >>SECTOR_SHIFT ; copy = n - copy; - page =brd_lookup_page (brd, sector); -BUG_ON (!page); + folio =brd_lookup_folio (brd, sector); +BUG_ON (!folio); - dst =kmap_atomic (page); + dst =kmap_local_folio (folio,0 );memcpy (dst, src, copy); -kunmap_atomic (dst); +kunmap_local (dst); } }
@@ -220,17 +220,17 @@ static void copy_to_brd(struct brd_device *brd, const void *src,static void copy_from_brd (void *dst,struct brd_device *brd,sector_t sector,size_t n) { -struct page *page; +struct folio *folio;void *src;unsigned int offset = (sector & (PAGE_SECTORS -1 )) <<SECTOR_SHIFT ;size_t copy; copy =min_t (size_t, n,PAGE_SIZE - offset); - page =brd_lookup_page (brd, sector); -if (page) { - src =kmap_atomic (page); -memcpy (dst, src + offset, copy); -kunmap_atomic (src); + folio =brd_lookup_folio (brd, sector); +if (folio) { + src =kmap_local_folio (folio, offset); +memcpy (dst, src, copy); +kunmap_local (src); }else memset (dst,0 , copy);
@@ -238,20 +238,20 @@ static void copy_from_brd(void *dst, struct brd_device *brd, dst += copy; sector += copy >>SECTOR_SHIFT ; copy = n - copy; - page =brd_lookup_page (brd, sector); -if (page) { - src =kmap_atomic (page); + folio =brd_lookup_folio (brd, sector); +if (folio) { + src =kmap_local_folio (folio,0 );memcpy (dst, src, copy); -kunmap_atomic (src); +kunmap_local (src); }else memset (dst,0 , copy); } }/* -* Process a single bvec of a bio. +* Process a single folio of a bio. */ -static int brd_do_bvec (struct brd_device *brd,struct page *page, +static int brd_do_folio (struct brd_device *brd,struct folio *folio,unsigned int len,unsigned int off,blk_opf_t opf,sector_t sector) {
@@ -261,7 +261,7 @@ static int brd_do_bvec(struct brd_device *brd, struct page *page,if (op_is_write (opf)) {/* * Must use NOIO because we don't want to recurse back into the -* block or filesystem layers from page reclaim. +* block or filesystem layers from folio reclaim. */ gfp_t gfp = opf &REQ_NOWAIT ?GFP_NOWAIT :GFP_NOIO ;
@@ -270,15 +270,15 @@ static int brd_do_bvec(struct brd_device *brd, struct page *page, goto out; } - mem =kmap_atomic (page); + mem =kmap_local_folio (folio, off);if (!op_is_write (opf)) { -copy_from_brd (mem + off, brd, sector, len); -flush_dcache_page (page); +copy_from_brd (mem, brd, sector, len); +flush_dcache_folio (folio); }else { -flush_dcache_page (page); -copy_to_brd (brd, mem + off, sector, len); +flush_dcache_folio (folio); +copy_to_brd (brd, mem, sector, len); } -kunmap_atomic (mem); +kunmap_local (mem); out:return err;
@@ -288,19 +288,18 @@ static void brd_submit_bio(struct bio *bio) {struct brd_device *brd = bio->bi_bdev->bd_disk->private_data;sector_t sector = bio->bi_iter.bi_sector; -struct bio_vec bvec; -struct bvec_iter iter; +struct folio_iter iter; -bio_for_each_segment (bvec, bio, iter) { -unsigned int len = bvec.bv_len; +bio_for_each_folio_all (iter, bio) { +unsigned int len = iter.length;int err;/* Don't support un-aligned buffer */ -WARN_ON_ONCE ((bvec.bv_offset & (SECTOR_SIZE -1 )) || +WARN_ON_ONCE ((iter.offset & (SECTOR_SIZE -1 )) || (len & (SECTOR_SIZE -1 ))); - err =brd_do_bvec (brd, bvec.bv_page, len, bvec.bv_offset, - bio->bi_opf, sector); + err =brd_do_folio (brd, iter.folio, len, iter.offset, + bio->bi_opf, sector);if (err) {if (err == -ENOMEM && bio->bi_opf &REQ_NOWAIT ) {bio_wouldblock_error (bio);
@@ -373,12 +372,12 @@ static int brd_alloc(int i)list_add_tail (&brd->brd_list, &brd_devices);spin_lock_init (&brd->brd_lock); -INIT_RADIX_TREE (&brd->brd_pages,GFP_ATOMIC ); +INIT_RADIX_TREE (&brd->brd_folios,GFP_ATOMIC );snprintf (buf,DISK_NAME_LEN ,"ram%d" , i);if (!IS_ERR_OR_NULL (brd_debugfs_dir))debugfs_create_u64 (buf,0444 , brd_debugfs_dir, - &brd->brd_nr_pages); + &brd->brd_nr_folios); disk = brd->brd_disk =blk_alloc_disk (NUMA_NO_NODE );if (!disk)
@@ -434,7 +433,7 @@ static void brd_cleanup(void)list_for_each_entry_safe (brd, next, &brd_devices, brd_list) {del_gendisk (brd->brd_disk);put_disk (brd->brd_disk); -brd_free_pages (brd); +brd_free_folios (brd);list_del (&brd->brd_list);kfree (brd); }
@@ -465,7 +464,7 @@ static int __init brd_init(void)brd_check_and_reset_par (); - brd_debugfs_dir =debugfs_create_dir ("ramdisk_pages" ,NULL ); + brd_debugfs_dir =debugfs_create_dir ("ramdisk_folios" ,NULL );for (i =0 ; i < rd_nr; i++) { err =brd_alloc (i);
Signature · 1 line
2.35.3
kernel test robot
build botRe: [PATCH 1/5] brd: convert to folios
I love your patch! Yet something to improve:
Hi Hannes,
I love your patch! Yet something to improve:
[auto build test ERROR on axboe-block/for-next]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch#_base_tree_information]
url: https://github.com/intel-lab-lkp/linux/commits/Hannes-Reinecke/brd-convert-to-folios/20230306-200223 base: https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git for-next patch link: https://lore.kernel.org/r/20230306120127.21375-2-hare%40suse.de patch subject: [PATCH 1/5] brd: convert to folios config: hexagon-randconfig-r045-20230306 (https://download.01.org/0day-ci/archive/20230306/[email protected]/config) compiler: clang version 17.0.0 (https://github.com/llvm/llvm-project 67409911353323ca5edf2049ef0df54132fa1ca7)
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# https://github.com/intel-lab-lkp/linux/commit/b29fb9873ddbb2efb157e5d6548abf3c88b3458c
git remote add linux-review https://github.com/intel-lab-lkp/linux
git fetch --no-tags linux-review Hannes-Reinecke/brd-convert-to-folios/20230306-200223
git checkout b29fb9873ddbb2efb157e5d6548abf3c88b3458c
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=hexagon olddefconfig COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1 O=build_dir ARCH=hexagon SHELL=/bin/bash drivers/block/
If you fix the issue, kindly add following tag where applicable
| Reported-by: kernel test robot <[email protected]>
| Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/
All errors (new ones prefixed by >>):
In file included from drivers/block/brd.c:17:
In file included from include/linux/blkdev.h:9:
In file included from include/linux/blk_types.h:10:
In file included from include/linux/bvec.h:10:
In file included from include/linux/highmem.h:12:
In file included from include/linux/hardirq.h:11:
In file included from ./arch/hexagon/include/generated/asm/hardirq.h:1:
In file included from include/asm-generic/hardirq.h:17:
In file included from include/linux/irq.h:20:
In file included from include/linux/io.h:13:
In file included from arch/hexagon/include/asm/io.h:334:
include/asm-generic/io.h:547:31: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
val = __raw_readb(PCI_IOBASE + addr);
~~~~~~~~~~ ^
include/asm-generic/io.h:560:61: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
val = __le16_to_cpu((__le16 __force)__raw_readw(PCI_IOBASE + addr));
~~~~~~~~~~ ^
include/uapi/linux/byteorder/little_endian.h:37:51: note: expanded from macro '__le16_to_cpu'
#define __le16_to_cpu(x) ((__force __u16)(__le16)(x))
^
In file included from drivers/block/brd.c:17:
In file included from include/linux/blkdev.h:9:
In file included from include/linux/blk_types.h:10:
In file included from include/linux/bvec.h:10:
In file included from include/linux/highmem.h:12:
In file included from include/linux/hardirq.h:11:
In file included from ./arch/hexagon/include/generated/asm/hardirq.h:1:
In file included from include/asm-generic/hardirq.h:17:
In file included from include/linux/irq.h:20:
In file included from include/linux/io.h:13:
In file included from arch/hexagon/include/asm/io.h:334:
include/asm-generic/io.h:573:61: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
val = __le32_to_cpu((__le32 __force)__raw_readl(PCI_IOBASE + addr));
~~~~~~~~~~ ^
include/uapi/linux/byteorder/little_endian.h:35:51: note: expanded from macro '__le32_to_cpu'
#define __le32_to_cpu(x) ((__force __u32)(__le32)(x))
^
In file included from drivers/block/brd.c:17:
In file included from include/linux/blkdev.h:9:
In file included from include/linux/blk_types.h:10:
In file included from include/linux/bvec.h:10:
In file included from include/linux/highmem.h:12:
In file included from include/linux/hardirq.h:11:
In file included from ./arch/hexagon/include/generated/asm/hardirq.h:1:
In file included from include/asm-generic/hardirq.h:17:
In file included from include/linux/irq.h:20:
In file included from include/linux/io.h:13:
In file included from arch/hexagon/include/asm/io.h:334:
include/asm-generic/io.h:584:33: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
__raw_writeb(value, PCI_IOBASE + addr);
~~~~~~~~~~ ^
include/asm-generic/io.h:594:59: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
__raw_writew((u16 __force)cpu_to_le16(value), PCI_IOBASE + addr);
~~~~~~~~~~ ^
include/asm-generic/io.h:604:59: warning: performing pointer arithmetic on a null pointer has undefined behavior [-Wnull-pointer-arithmetic]
__raw_writel((u32 __force)cpu_to_le32(value), PCI_IOBASE + addr);
~~~~~~~~~~ ^
>> drivers/block/brd.c:325:8: error: call to undeclared function 'brd_do_bvec'; ISO C99 and later do not support implicit function declarations [-Wimplicit-function-declaration]
err = brd_do_bvec(brd, page, PAGE_SIZE, 0, op, sector);
^6 warnings and 1 error generated.
vim +/brd_do_bvec +325 drivers/block/brd.c
9db5579be4bb53 Nicholas Piggin 2008-02-08 316
a72132c31d5809 Matthew Wilcox 2014-06-04 317 static int brd_rw_page(struct block_device *bdev, sector_t sector,
86947df3a92364 Bart Van Assche 2022-07-14 318 struct page *page, enum req_op op)
a72132c31d5809 Matthew Wilcox 2014-06-04 319 {
a72132c31d5809 Matthew Wilcox 2014-06-04 320 struct brd_device *brd = bdev->bd_disk->private_data;
98cc093cba1e92 Huang Ying 2017-09-06 321 int err;
98cc093cba1e92 Huang Ying 2017-09-06 322
98cc093cba1e92 Huang Ying 2017-09-06 323 if (PageTransHuge(page))
98cc093cba1e92 Huang Ying 2017-09-06 324 return -ENOTSUPP;
3f289dcb4b2654 Tejun Heo 2018-07-18 @325 err = brd_do_bvec(brd, page, PAGE_SIZE, 0, op, sector);
3f289dcb4b2654 Tejun Heo 2018-07-18 326 page_endio(page, op_is_write(op), err);
a72132c31d5809 Matthew Wilcox 2014-06-04 327 return err;
a72132c31d5809 Matthew Wilcox 2014-06-04 328 }
a72132c31d5809 Matthew Wilcox 2014-06-04 329 Signature · 2 lines
0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests
Re: [PATCH 1/5] brd: convert to folios
In reply to Hannes Reineckedrivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)
Did you drop HIGHMEM support on purpose?
Re: [PATCH 1/5] brd: convert to folios
Quoted 7 lines · Matthew Wilcox
drivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)Did you drop HIGHMEM support on purpose?
No; I thought that folios would be doing that implicitely. Will be re-adding.
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman
Re: [PATCH 1/5] brd: convert to folios
Quoted 11 lines · Hannes Reinecke
Quoted 7 lines · Matthew Wilcox
drivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)Did you drop HIGHMEM support on purpose?
No; I thought that folios would be doing that implicitely. Will be re-adding.
We can't ... not all filesystems want to allocate every folio from HIGHMEM. eg for superblocks, it often makes more sense to allocate the folio from lowmem than allocate it from highmem and keep it kmapped. The only GFP flag that folios force-set is __GFP_COMP because folios by definition are compound pages.
Re: [PATCH 1/5] brd: convert to folios
Quoted 10 lines · Matthew Wilcox
Quoted 9 lines · Hannes Reinecke
Quoted 7 lines · Matthew Wilcox
drivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)Did you drop HIGHMEM support on purpose?
Quoted 8 lines
Quoted 2 lines
No; I thought that folios would be doing that implicitely. Will be re-adding.
We can't ... not all filesystems want to allocate every folio from HIGHMEM. eg for superblocks, it often makes more sense to allocate the folio from lowmem than allocate it from highmem and keep it kmapped. The only GFP flag that folios force-set is
__GFP_COMPbecause folios by definition are compound pages.
Just some historical information here. Some commit logs would seem to make it seem that __GFP_HIGHMEM was added to support DAX, but it's not the case. When DAX suport was removed from brd the __GFP_HIGHMEM removed was removed by mistake, and later fixed through commit f6b50160a06d4 ("brd: re-enable __GFP_HIGHMEM in brd_insert_page()"). But that doesn't tell us what the default were before.
To avoid future removals maybe we should document why we care?
See these commits:
9db5579be4bb5 ("rewrite rd") initial commit with highmem
cb9cd2ef0bbb4 ("rd: support XIP") forces highmem if XIP, but already set
a7a97fc9ff6c2 ("brd: rename XIP to DAX") renames the config to DAX26defe34e48e1 ("fix brd allocation flags") tries to fix the confusion but makes it worse
And so commit f6b50160a06d4 ("brd: re-enable __GFP_HIGHMEM in brd_insert_page()") seems to be correct in re-adding it, adn commit cb9cd2ef0bbb4 ("rd: support XIP") made things confusing assuming we wanted only highmem when DAX was enabled.
So we only want highmem just because it's been there since it was first added. That's all.
Luis
A message that is not in the archive (sent off-list or not yet received): [email protected]
Re: [PATCH 1/5] brd: convert to folios
Quoted 22 lines · Hannes Reinecke
Quoted 10 lines · Matthew Wilcox
Quoted 9 lines · Hannes Reinecke
Quoted 7 lines · Matthew Wilcox
drivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)Did you drop HIGHMEM support on purpose?
Quoted 8 lines
Quoted 2 lines
No; I thought that folios would be doing that implicitely. Will be re-adding.
We can't ... not all filesystems want to allocate every folio from HIGHMEM. eg for superblocks, it often makes more sense to allocate the folio from lowmem than allocate it from highmem and keep it kmapped. The only GFP flag that folios force-set is
__GFP_COMPbecause folios by definition are compound pages.Oh well.
Quoted 7 lines
However, when playing with the modified brd and setting the logical&physical blocksize to 16k the whole thing crashes
(not unexpectedly).
It does crash, however, inblock_read_full_folio(), which rather surprisingly (at least for me) is usingcreate_page_buffers();
I would have expected something likecreate_folio_buffers().
Is this work in progress or what is the plan here?
Supporting folios > PAGE_SIZE in blockdev is definitely still WIP. I know of at least one bug, which is:
#define bh_offset(bh) ((unsigned long)(bh)->b_data & ~PAGE_MASK)
That needs to be something like
static size_t bh_offset(const struct buffer_head *bh)
{
return (unsigned long)bh->b_data & (folio_size(bh->b_folio) - 1);
}I haven't done a thorough scan for folio-size problems in the block layer; I've just been fixing up things as I notice them.
Yes, create_page_buffers() should now be create_folio_buffers(). Just didn't get round to it yet.
Re: [PATCH 1/5] brd: convert to folios
Quoted 38 lines · Matthew Wilcox
Quoted 22 lines · Hannes Reinecke
Quoted 10 lines · Matthew Wilcox
Quoted 9 lines · Hannes Reinecke
Quoted 7 lines · Matthew Wilcox
drivers/block/brd.c static struct page *brd_lookup_page(struct brd_device *brd, sector_t sector) · line 94
Quoted 4 lines · Hannes Reinecke
drivers/block/brd.c +2
-2- page =alloc_page (gfp | __GFP_ZERO | __GFP_HIGHMEM); -if (!page) + folio =folio_alloc (gfp | __GFP_ZERO,0 ); +if (!folio)Did you drop HIGHMEM support on purpose?
Quoted 8 lines
Quoted 2 lines
No; I thought that folios would be doing that implicitely. Will be re-adding.
We can't ... not all filesystems want to allocate every folio from HIGHMEM. eg for superblocks, it often makes more sense to allocate the folio from lowmem than allocate it from highmem and keep it kmapped. The only GFP flag that folios force-set is
__GFP_COMPbecause folios by definition are compound pages.Oh well.
Quoted 7 lines
However, when playing with the modified brd and setting the logical&physical blocksize to 16k the whole thing crashes
(not unexpectedly).
It does crash, however, inblock_read_full_folio(), which rather surprisingly (at least for me) is usingcreate_page_buffers();
I would have expected something likecreate_folio_buffers().
Is this work in progress or what is the plan here?Supporting folios >
PAGE_SIZEin blockdev is definitely still WIP. I know of at least one bug, which is:#define
bh_offset(bh)((unsigned long)(bh)->b_data & ~PAGE_MASK)That needs to be something like
Quoted 8 lines
static
size_tbh_offset(const struct buffer_head *bh){ return (unsigned long)bh->b_data & (folio_size(bh->b_folio) - 1); }I haven't done a thorough scan for folio-size problems in the block layer; I've just been fixing up things as I notice them.
And not to mention the various instances of (PAGE_SHIFT - blkbits) which will get happily into negative numbers for large block sizes, resulting in interesting effects for a shift left operation ...
Quoted 2 lines
Yes,
create_page_buffers()should now becreate_folio_buffers(). Just didn't get round to it yet.
Ah, so I was on the right track after all.
But I was wondering ... couldn't we use the physical block size as a hint on how many pages to allocate?
With that we can work on updating the stack even with existing hardware, and we wouldn't crash immediately if we miss the odd conversion ...
Hmm?
Cheers,
Hannes
Re: [PATCH 1/5] brd: convert to folios
In reply to Hannes ReineckeQuoted 12 lines · Hannes Reinecke
Convert the driver to work on folios instead of pages.
Signed-off-by Hannes Reinecke <[email protected]>--- drivers/block/brd.c | 171 ++++++++++++++++++++++---------------------- 1 file changed, 85 insertions(+), 86 deletions(-)drivers/block/brd.c +0
-0
drivers/block/brd.c line 31
Quoted 6 lines
drivers/block/brd.c +1
-2@@ -28,8 +28,8 @@#include <linux/uaccess.h> /* -* Each block ramdisk device has a radix_tree brd_pages of pages that stores -* the pages containing the block device's contents. A brd page's ->index is +* Each block ramdisk device has a radix_tree brd_folios of folios that stores
^^^^^^^^^
drivers/block/brd.c line 32
Quoted 1 line
drivers/block/brd.c +1
-0+* the folios containing the block device's contents. A brd folio's ->index is
^^^^^^^^^^
So we end up with:
"a radix_tree brd_folios of folios that stores the folios containing ..."
What about:
"a radix_tree brd_folios that stores folios containing"
Other than that, looks good:
So another thing, I think I counted about 5-8 grammatical rules which could be bundled up into *one* SmPL grammar patch which could then be used to automatically do similar tasks elsewhere.
Luis
PATCH 2/5 brd: abstract page_size conventions
In reply to Hannes ReineckeIn preparation for changing the block sizes abstract away references to PAGE_SIZE and friends.
--- drivers/block/brd.c | 30 +++++++++++++++++++++--------- 1 file changed, 21 insertions(+), 9 deletions(-)
drivers/block/brd.c +21 -9
@@ -55,6 +55,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) {pgoff_t idx;struct folio *folio; +unsigned int rd_sector_shift =PAGE_SHIFT -SECTOR_SHIFT ;/* * The folio lifetime is protected by the fact that we have opened the
@@ -68,7 +69,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector)* here, only deletes). */ rcu_read_lock (); - idx = sector >>PAGE_SECTORS_SHIFT ;/* sector to folio index */ + idx = sector >> rd_sector_shift;/* sector to folio index */ folio =radix_tree_lookup (&brd->brd_folios, idx);rcu_read_unlock ();
@@ -84,13 +85,15 @@ static int brd_insert_folio(struct brd_device *brd, sector_t sector, gfp_t gfp) {pgoff_t idx;struct folio *folio; +unsigned int rd_sector_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sector_order =get_order (PAGE_SIZE );int ret =0 ; folio =brd_lookup_folio (brd, sector);if (folio)return 0 ; - folio =folio_alloc (gfp | __GFP_ZERO,0 ); + folio =folio_alloc (gfp | __GFP_ZERO, rd_sector_order);if (!folio)return -ENOMEM ;
@@ -100,7 +103,7 @@ static int brd_insert_folio(struct brd_device *brd, sector_t sector, gfp_t gfp) }spin_lock (&brd->brd_lock); - idx = sector >>PAGE_SECTORS_SHIFT ; + idx = sector >> rd_sector_shift; folio->index = idx;if (radix_tree_insert (&brd->brd_folios, idx, folio)) {folio_put (folio);
@@ -167,11 +170,14 @@ static void brd_free_folios(struct brd_device *brd)static int copy_to_brd_setup (struct brd_device *brd,sector_t sector,size_t n,gfp_t gfp) { -unsigned int offset = (sector & (PAGE_SECTORS -1 )) <<SECTOR_SHIFT ; +unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors =1 << rd_sectors_shift; +unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy;int ret; - copy =min_t (size_t, n,PAGE_SIZE - offset); + copy =min_t (size_t, n, rd_sector_size - offset); ret =brd_insert_folio (brd, sector, gfp);if (ret)return ret;
@@ -190,10 +196,13 @@ static void copy_to_brd(struct brd_device *brd, const void *src, {struct folio *folio;void *dst; -unsigned int offset = (sector & (PAGE_SECTORS -1 )) <<SECTOR_SHIFT ; +unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors =1 << rd_sectors_shift; +unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy; - copy =min_t (size_t, n,PAGE_SIZE - offset); + copy =min_t (size_t, n, rd_sector_size - offset); folio =brd_lookup_folio (brd, sector);BUG_ON (!folio);
@@ -222,10 +231,13 @@ static void copy_from_brd(void *dst, struct brd_device *brd, {struct folio *folio;void *src; -unsigned int offset = (sector & (PAGE_SECTORS -1 )) <<SECTOR_SHIFT ; +unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors =1 << rd_sectors_shift; +unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy; - copy =min_t (size_t, n,PAGE_SIZE - offset); + copy =min_t (size_t, n, rd_sector_size - offset); folio =brd_lookup_folio (brd, sector);if (folio) { src =kmap_local_folio (folio, offset);
Signature · 1 line
2.35.3
PATCH 3/5 brd: make sector size configurable
In reply to Hannes ReineckeAdd a module option 'rd_blksize' to allow the user to change the sector size of the RAM disks.
--- drivers/block/brd.c | 30 +++++++++++++++++++----------- 1 file changed, 19 insertions(+), 11 deletions(-)
drivers/block/brd.c +19 -11
@@ -30,7 +30,7 @@/* * Each block ramdisk device has a radix_tree brd_folios of folios that stores * the folios containing the block device's contents. A brd folio's ->index is -* its offset in PAGE_SIZE units. This is similar to, but in no way connected +* its offset in brd_blksize units. This is similar to, but in no way connected * with, the kernel's pagecache or buffer cache (which sit above our block * device). */
@@ -46,6 +46,8 @@ struct brd_device {spinlock_t brd_lock;struct radix_tree_root brd_folios;u64 brd_nr_folios; +unsigned int brd_sector_shift; +unsigned int brd_sector_size; };/*
@@ -55,7 +57,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ;/* * The folio lifetime is protected by the fact that we have opened the
@@ -85,8 +87,8 @@ static int brd_insert_folio(struct brd_device *brd, sector_t sector, gfp_t gfp) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift =PAGE_SHIFT -SECTOR_SHIFT ; -unsigned int rd_sector_order =get_order (PAGE_SIZE ); +unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sector_order =get_order (brd->brd_sector_size);int ret =0 ; folio =brd_lookup_folio (brd, sector);
@@ -170,9 +172,9 @@ static void brd_free_folios(struct brd_device *brd)static int copy_to_brd_setup (struct brd_device *brd,sector_t sector,size_t n,gfp_t gfp) { -unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ;unsigned int rd_sectors =1 << rd_sectors_shift; -unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int rd_sector_size = brd->brd_sector_size;unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy;int ret;
@@ -196,9 +198,9 @@ static void copy_to_brd(struct brd_device *brd, const void *src, {struct folio *folio;void *dst; -unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ;unsigned int rd_sectors =1 << rd_sectors_shift; -unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int rd_sector_size = brd->brd_sector_size;unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy;
@@ -231,9 +233,9 @@ static void copy_from_brd(void *dst, struct brd_device *brd, {struct folio *folio;void *src; -unsigned int rd_sectors_shift =PAGE_SHIFT -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ;unsigned int rd_sectors =1 << rd_sectors_shift; -unsigned int rd_sector_size =PAGE_SIZE ; +unsigned int rd_sector_size = brd->brd_sector_size;unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ;size_t copy;
@@ -346,6 +348,10 @@ static int max_part = 1;module_param (max_part, int,0444 );MODULE_PARM_DESC (max_part,"Num Minors to reserve between devices" ); +static unsigned int rd_blksize =PAGE_SIZE ; +module_param (rd_blksize, uint,0444 ); +MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." ); +MODULE_LICENSE ("GPL" );MODULE_ALIAS_BLOCKDEV_MAJOR (RAMDISK_MAJOR );MODULE_ALIAS ("rd" );
@@ -382,6 +388,8 @@ static int brd_alloc(int i)return -ENOMEM ; brd->brd_number = i;list_add_tail (&brd->brd_list, &brd_devices); + brd->brd_sector_shift =ilog2 (rd_blksize); + brd->brd_sector_size = rd_blksize;spin_lock_init (&brd->brd_lock);INIT_RADIX_TREE (&brd->brd_folios,GFP_ATOMIC );
@@ -410,7 +418,7 @@ static int brd_alloc(int i)* otherwise fdisk will align on 1M. Regardless this call * is harmless) */ -blk_queue_physical_block_size (disk->queue,PAGE_SIZE ); +blk_queue_physical_block_size (disk->queue, rd_blksize);/* Tell the block layer that this is not a rotational device */ blk_queue_flag_set (QUEUE_FLAG_NONROT , disk->queue);
Signature · 1 line
2.35.3
Re: [PATCH 3/5] brd: make sector size configurable
Quoted 13 lines · Hannes Reinecke
Add a module option '
rd_blksize' to allow the user to change the sector size of the RAM disks.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/block/brd.c | 30 +++++++++++++++++++----------- 1 file changed, 19 insertions(+), 11 deletions(-)drivers/block/brd.c +0
-0
drivers/block/brd.c struct brd_device { · line 51
Quoted 6 lines
drivers/block/brd.c +2
-0@@ -46,6 +46,8 @@ struct brd_device {spinlock_t brd_lock;struct radix_tree_root brd_folios;u64 brd_nr_folios; +unsigned int brd_sector_shift; +unsigned int brd_sector_size; };
Why not just do this first and initialize this to the defaults set without the module parameter. Then you don't need to declare over and over unsigned int rd_sectors, etc in tons of routines and just pass the brd which most routines get.
Then most of this and the prior patch can be squeezed into one.
The functional changes would be the addition of the module parameter.
drivers/block/brd.c static int brd_alloc(int i) · line 421
Quoted 6 lines
drivers/block/brd.c +1
-1@@ -410,7 +418,7 @@ static int brd_alloc(int i)* otherwise fdisk will align on 1M. Regardless this call * is harmless) */ -blk_queue_physical_block_size (disk->queue,PAGE_SIZE ); +blk_queue_physical_block_size (disk->queue, rd_blksize);
And this was added for DAX support, but DAX support was long removed from brd see commit 7a862fbbdec6 ("brd: remove dax support"). This nugget was added by Boaz via commit c8fa31730fc72a8 ("brd: Request from fdisk 4k alignment"), but if we don't support DAX, can't we just kill that line?
Current doc for blk_queue_physical_block_size() says:
* This should be set to the lowest possible sector size that the * hardware can operate on without reverting to read-modify-write * operations.
But since we're working directly with RAM, do we care?
The comment above that line referring to direct_access should be killed at the very least.
While you're at it, can you then also update Documentation/filesystems/dax.rst to remove the brd as an example driver with DAX support.
That leaves us with only a few block drivers with DAX:
- dcssblk: s390 dcss block device driver
- pmem: NVDIMM persistent memory driver
- some dm targets
- fuse virtio_fs
Wonder which is the right example these days for DAX, now that persistant memory has End of Life'd with Optane dead, curious also of the value of the above and DAX in general other than support for whatever made it out.
Should we EOL DAX too?
Luis
Re: [PATCH 3/5] brd: make sector size configurable
Quoted 45 lines · Luis Chamberlain
Quoted 13 lines · Hannes Reinecke
Add a module option '
rd_blksize' to allow the user to change the sector size of the RAM disks.Signed-off-by Hannes Reinecke <[email protected]>--- drivers/block/brd.c | 30 +++++++++++++++++++----------- 1 file changed, 19 insertions(+), 11 deletions(-)drivers/block/brd.c +0
-0drivers/block/brd.c struct brd_device { · line 51
Quoted 6 lines
drivers/block/brd.c +2
-0@@ -46,6 +46,8 @@ struct brd_device {spinlock_t brd_lock;struct radix_tree_root brd_folios;u64 brd_nr_folios; +unsigned int brd_sector_shift; +unsigned int brd_sector_size; };Why not just do this first and initialize this to the defaults set without the module parameter. Then you don't need to declare over and over unsigned int
rd_sectors, etc in tons of routines and just pass the brd which most routines get.Then most of this and the prior patch can be squeezed into one.
The functional changes would be the addition of the module parameter.
drivers/block/brd.c static int brd_alloc(int i) · line 421
Quoted 6 lines
drivers/block/brd.c +1
-1@@ -410,7 +418,7 @@ static int brd_alloc(int i)* otherwise fdisk will align on 1M. Regardless this call * is harmless) */ -blk_queue_physical_block_size (disk->queue,PAGE_SIZE ); +blk_queue_physical_block_size (disk->queue, rd_blksize);And this was added for DAX support, but DAX support was long removed from brd see commit
7a862fbbdec6("brd: remove dax support"). This nugget was added by Boaz via commitc8fa31730fc72a8("brd: Request from fdisk 4k alignment"), but if we don't support DAX, can't we just kill that line?Current doc for
blk_queue_physical_block_size()says:
Quoted 8 lines
* This should be set to the lowest possible sector size that the * hardware can operate on without reverting to read-modify-write * operations.But since we're working directly with RAM, do we care?
The comment above that line referring to
direct_accessshould be killed at the very least.
I was curious whether or not Martin's comments about "mkfs already considers the reported queue limits (for the filesystems most people use, anyway)" applies to say XFS [0] and whether or not the above helps guide mkfs that way, so I took a look now.
[0] https://lore.kernel.org/all/[email protected]/
There's a default struct mkfs_default_params which sets the blocksize to 1 << XFS_DFL_BLOCKSIZE_LOG and so at least validate_blocksize() will use that if no parameter was passed on the cli. We also set the default sector size to XFS_MIN_SECTORSIZE.
There are two aspects to what we use for the final mkfs, first we use validate_blocksize(), then validate_sectorsize(). The first will set the default block size to 1 << XFS_DFL_BLOCKSIZE_LOG (4096 bytes) if the user did not specify anything. Then validate_sectorsize() uses get_topology() to get the physical / logical block sizes and both:
* blkid_topology_get_minimum_io_size()
* blkid_topology_get_optimal_io_size()
These come from utilil-linux:
https://github.com/util-linux/util-linux.git
https://www.kernel.org/doc/Documentation/ABI/testing/sysfs-block
So that's just
/sys/block/<disk>/queue/physical_block_size
/sys/block/<disk>/queue/logical_block_size
Then:
/sys/block/<disk>/queue/minimum_io_size
The documentation suggests " For disk drives this is often the physical block size."
/sys/block/<disk>/queue/optimal_io_size
The documentation suggests this is "This is rarely reported for disk drives."
From my review of xfs's mkfs is we essentially use the physical block size as a default sector size if set, otherwise we use the device's logical block size if set otherwise xfsprog's default and so 4096.
And so the above comment referring to DAX could be killed and replaced with allowing userspace mkfs utils to use a more appropriate block size. Or just kill the comment all together and we update the docs for blk_queue_physical_block_size() instead.
Quoted 10 lines
While you're at it, can you then also update
Documentation/filesystems/dax.rstto remove the brd as an example driver with DAX support.That leaves us with only a few block drivers with DAX:
- dcssblk: s390 dcss block device driver
- pmem: NVDIMM persistent memory driver
- some dm targets
- fusevirtio_fs
Quoted 4 lines
Wonder which is the right example these days for DAX, now that persistant memory has End of Life'd with Optane dead, curious also of the value of the above and DAX in general other than support for whatever made it out.
Looking at this:
https://lore.kernel.org/all/[email protected]/
Quoted 1 line
Should we EOL DAX too?
I think the answer is no as a generic kernel thing, however, it probably means we should really update the DAX documentation to reflect what our real expecations are for its current and future uses. And that is to highlight the non-PMEM use cases as *those* probably are EOL'd now.
Luis
Re: [PATCH 3/5] brd: make sector size configurable
Luis,
Quoted 9 lines
/sys/block/<disk>/queue/minimum_io_size
The documentation suggests " For disk drives this is often the physical block size."
/sys/block/<disk>/queue/optimal_io_size
The documentation suggests this is "This is rarely reported for disk drives."
min_io and opt_io are used to key mkfs.xfs' sunit/swidth. So if you're using a hardware RAID, MD, or DM, we'll attempt to align allocations on stripe boundaries.
Back when that "rarely reported" blurb was written (2009), we did not have any individual disk drives which reported min_io/opt_io. Reporting those parameters was mostly a storage array thing. These days it's fairly common for both disk drives and SSDs to fill out these fields.
Quoted 3 lines
From my review of xfs's mkfs is we essentially use the physical block size as a default sector size if set, otherwise we use the device's logical block size if set otherwise xfsprog's default and so 4096.
Yep.
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering
Re: [PATCH 3/5] brd: make sector size configurable
Quoted 13 lines · Martin K. Petersen
Luis,
Quoted 9 lines
/sys/block/<disk>/queue/minimum_io_size
The documentation suggests " For disk drives this is often the physical block size."
/sys/block/<disk>/queue/optimal_io_size
The documentation suggests this is "This is rarely reported for disk drives."
Quoted 8 lines
min_ioandopt_ioare used to key mkfs.xfs' sunit/swidth. So if you're using a hardware RAID, MD, or DM, we'll attempt to align allocations on stripe boundaries.Back when that "rarely reported" blurb was written (2009), we did not have any individual disk drives which reported min_io/opt_io. Reporting those parameters was mostly a storage array thing. These days it's fairly common for both disk drives and SSDs to fill out these fields.
Glad you mentioned this, I followed up in my review of these and I see even the names, swidth, sunit, are all "stripe" indicative. Based on what you are saying, it seems we may need to update docs to reflect actual / new uses.
Quoted 5 lines
Quoted 3 lines
From my review of xfs's mkfs is we essentially use the physical block size as a default sector size if set, otherwise we use the device's logical block size if set otherwise xfsprog's default and so 4096.
Yep.
The use case for swidth / sunit on mkfs.xfs seemed pretty tied to striping, and it was not obvious or clear / if using it for hints could be used today as perhaps intended. At least all the naming and validation stuff seems to make it very "stripy" still.
Luis
PATCH 4/5 brd: limit maximal block size to 32M
In reply to Hannes ReineckeUse an arbitrary limit of 32M for the maximal blocksize so as not to overflow the page cache.
--- drivers/block/brd.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
drivers/block/brd.c +8 -8
@@ -348,6 +348,7 @@ static int max_part = 1;module_param (max_part, int,0444 );MODULE_PARM_DESC (max_part,"Num Minors to reserve between devices" ); +#define RD_MAX_SECTOR_SIZE 65536static unsigned int rd_blksize =PAGE_SIZE ;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." );
@@ -410,15 +411,14 @@ static int brd_alloc(int i) disk->private_data = brd;strscpy (disk->disk_name, buf,DISK_NAME_LEN );set_capacity (disk, rd_size *2 ); - -/* -* This is so fdisk will align partitions on 4k, because of -* direct_access API needing 4k alignment, returning a PFN -* (This is only a problem on very small devices <= 4M, -* otherwise fdisk will align on 1M. Regardless this call -* is harmless) -*/ + +if (rd_blksize >RD_MAX_SECTOR_SIZE ) { +/* Arbitrary limit maximum block size to 32M */ + err = -EINVAL ; + goto out_cleanup_disk; + }blk_queue_physical_block_size (disk->queue, rd_blksize); +blk_queue_max_hw_sectors (disk->queue,RD_MAX_SECTOR_SIZE );/* Tell the block layer that this is not a rotational device */ blk_queue_flag_set (QUEUE_FLAG_NONROT , disk->queue);
Signature · 1 line
2.35.3
Re: [PATCH 4/5] brd: limit maximal block size to 32M
Quoted 2 lines · Hannes Reinecke
Use an arbitrary limit of 32M for the maximal blocksize so as not to overflow the page cache.
The page allocator tops out at MAX_ORDER (generally 4MB), so that might be a better limit. Also, this has nothing to do with the page cache, right?
Re: [PATCH 4/5] brd: limit maximal block size to 32M
Quoted 7 lines · Matthew Wilcox
Quoted 2 lines · Hannes Reinecke
Use an arbitrary limit of 32M for the maximal blocksize so as not to overflow the page cache.
The page allocator tops out at
MAX_ORDER(generally 4MB), so that might be a better limit. Also, this has nothing to do with the page cache, right?
Indeed. Sloppy description.
Will be fixing it up.
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman
Re: [PATCH 4/5] brd: limit maximal block size to 32M
In reply to Hannes ReineckeQuoted 24 lines · Hannes Reinecke
Use an arbitrary limit of 32M for the maximal blocksize so as not to overflow the page cache.
Signed-off-by Hannes Reinecke <[email protected]>--- drivers/block/brd.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)drivers/block/brd.c +1
-0@@ -348,6 +348,7 @@ static int max_part = 1;module_param (max_part, int,0444 );MODULE_PARM_DESC (max_part,"Num Minors to reserve between devices" ); +#define RD_MAX_SECTOR_SIZE 65536static unsigned int rd_blksize =PAGE_SIZE ;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." );@@ -410,15 +411,14 @@ static int brd_alloc(int i) disk->private_data = brd;strscpy (disk->disk_name, buf,DISK_NAME_LEN );set_capacity (disk, rd_size *2 );
drivers/block/brd.c static int brd_alloc(int i) · line 415
Quoted 10 lines
drivers/block/brd.c +2
-8- -/* -* This is so fdisk will align partitions on 4k, because of -* direct_access API needing 4k alignment, returning a PFN -* (This is only a problem on very small devices <= 4M, -* otherwise fdisk will align on 1M. Regardless this call -* is harmless) -*/ + +if (rd_blksize >RD_MAX_SECTOR_SIZE ) {
rd_blkside is in bytes, but the above is ccomapring it to something in units of SECTOR_SIZE.
drivers/block/brd.c static int brd_alloc(int i) · line 419
Quoted 4 lines
drivers/block/brd.c +4
-0+/* Arbitrary limit maximum block size to 32M */ + err = -EINVAL ; + goto out_cleanup_disk; + }
rd_blksize also needs to be >= 512, and a power of 2.
Quoted 5 lines
drivers/block/brd.c +1
-0blk_queue_physical_block_size (disk->queue, rd_blksize); +blk_queue_max_hw_sectors (disk->queue,RD_MAX_SECTOR_SIZE );/* Tell the block layer that this is not a rotational device */ blk_queue_flag_set (QUEUE_FLAG_NONROT , disk->queue);
Re: [PATCH 4/5] brd: limit maximal block size to 32M
Quoted 30 lines · Keith Busch
Quoted 29 lines · Hannes Reinecke
Use an arbitrary limit of 32M for the maximal blocksize so as not to overflow the page cache.
Signed-off-by Hannes Reinecke <[email protected]>--- drivers/block/brd.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)drivers/block/brd.c +1
-5@@ -348,6 +348,7 @@ static int max_part = 1;module_param (max_part, int,0444 );MODULE_PARM_DESC (max_part,"Num Minors to reserve between devices" ); +#define RD_MAX_SECTOR_SIZE 65536static unsigned int rd_blksize =PAGE_SIZE ;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." );@@ -410,15 +411,14 @@ static int brd_alloc(int i) disk->private_data = brd;strscpy (disk->disk_name, buf,DISK_NAME_LEN );set_capacity (disk, rd_size *2 ); - -/* -* This is so fdisk will align partitions on 4k, because of -* direct_access API needing 4k alignment, returning a PFN -* (This is only a problem on very small devices <= 4M,
Quoted 9 lines
drivers/block/brd.c static int brd_alloc(int i) · line 415
Quoted 5 lines
drivers/block/brd.c +2
-3-* otherwise fdisk will align on 1M. Regardless this call -* is harmless) -*/ + +if (rd_blksize >RD_MAX_SECTOR_SIZE ) {
rd_blksideis in bytes, but the above is ccomapring it to something in units ofSECTOR_SIZE.
Ok.
Quoted 7 lines
drivers/block/brd.c static int brd_alloc(int i) · line 419
Quoted 4 lines
drivers/block/brd.c +4
-0+/* Arbitrary limit maximum block size to 32M */ + err = -EINVAL ; + goto out_cleanup_disk; + }
rd_blksizealso needs to be >= 512, and a power of 2.
Yes; I'll be adding the checks accordingly.
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Maxfeldstr. 5, 90409 Nürnberg HRB 36809 (AG Nürnberg), Geschäftsführer: Ivo Totev, Andrew Myers, Andrew McDonald, Martje Boudien Moerman
PATCH 5/5 brd: make logical sector size configurable
In reply to Hannes ReineckeAdd a module option 'rd_logical_blksize' to allow the user to change the logical sector size of the RAM disks.
--- drivers/block/brd.c | 37 +++++++++++++++++++++++-------------- 1 file changed, 23 insertions(+), 14 deletions(-)
drivers/block/brd.c +23 -14
@@ -48,6 +48,8 @@ struct brd_device {u64 brd_nr_folios;unsigned int brd_sector_shift;unsigned int brd_sector_size; +unsigned int brd_logical_sector_shift; +unsigned int brd_logical_sector_size; };/*
@@ -57,7 +59,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sector_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;/* * The folio lifetime is protected by the fact that we have opened the
@@ -87,7 +89,7 @@ static int brd_insert_folio(struct brd_device *brd, sector_t sector, gfp_t gfp) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sector_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;unsigned int rd_sector_order =get_order (brd->brd_sector_size);int ret =0 ;
@@ -172,10 +174,10 @@ static void brd_free_folios(struct brd_device *brd)static int copy_to_brd_setup (struct brd_device *brd,sector_t sector,size_t n,gfp_t gfp) { -unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;unsigned int rd_sectors =1 << rd_sectors_shift;unsigned int rd_sector_size = brd->brd_sector_size; -unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ; +unsigned int offset = (sector & (rd_sectors -1 )) << brd->brd_logical_sector_shift;size_t copy;int ret;
@@ -184,7 +186,7 @@ static int copy_to_brd_setup(struct brd_device *brd, sector_t sector, size_t n,if (ret)return ret;if (copy < n) { - sector += copy >>SECTOR_SHIFT ; + sector += copy >> brd->brd_logical_sector_shift; ret =brd_insert_folio (brd, sector, gfp); }return ret;
@@ -198,10 +200,10 @@ static void copy_to_brd(struct brd_device *brd, const void *src, {struct folio *folio;void *dst; -unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;unsigned int rd_sectors =1 << rd_sectors_shift;unsigned int rd_sector_size = brd->brd_sector_size; -unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ; +unsigned int offset = (sector & (rd_sectors -1 )) << brd->brd_logical_sector_shift;size_t copy; copy =min_t (size_t, n, rd_sector_size - offset);
@@ -214,7 +216,7 @@ static void copy_to_brd(struct brd_device *brd, const void *src,if (copy < n) { src += copy; - sector += copy >>SECTOR_SHIFT ; + sector += copy >> brd->brd_logical_sector_shift; copy = n - copy; folio =brd_lookup_folio (brd, sector);BUG_ON (!folio);
@@ -233,10 +235,10 @@ static void copy_from_brd(void *dst, struct brd_device *brd, {struct folio *folio;void *src; -unsigned int rd_sectors_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sectors_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;unsigned int rd_sectors =1 << rd_sectors_shift;unsigned int rd_sector_size = brd->brd_sector_size; -unsigned int offset = (sector & (rd_sectors -1 )) <<SECTOR_SHIFT ; +unsigned int offset = (sector & (rd_sectors -1 )) << brd->brd_logical_sector_shift;size_t copy; copy =min_t (size_t, n, rd_sector_size - offset);
@@ -250,7 +252,7 @@ static void copy_from_brd(void *dst, struct brd_device *brd,if (copy < n) { dst += copy; - sector += copy >>SECTOR_SHIFT ; + sector += copy >> brd->brd_logical_sector_shift; copy = n - copy; folio =brd_lookup_folio (brd, sector);if (folio) {
@@ -309,8 +311,8 @@ static void brd_submit_bio(struct bio *bio)int err;/* Don't support un-aligned buffer */ -WARN_ON_ONCE ((iter.offset & (SECTOR_SIZE -1 )) || - (len & (SECTOR_SIZE -1 ))); +WARN_ON_ONCE ((iter.offset & (brd->brd_logical_sector_size -1 )) || + (len & (brd->brd_logical_sector_size -1 ))); err =brd_do_folio (brd, iter.folio, len, iter.offset, bio->bi_opf, sector);
@@ -322,7 +324,7 @@ static void brd_submit_bio(struct bio *bio)bio_io_error (bio);return ; } - sector += len >>SECTOR_SHIFT ; + sector += len >> brd->brd_logical_sector_shift; }bio_endio (bio);
@@ -353,6 +355,10 @@ static unsigned int rd_blksize = PAGE_SIZE;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." ); +static unsigned int rd_logical_blksize =SECTOR_SIZE ; +module_param (rd_logical_blksize, uint,0444 ); +MODULE_PARM_DESC (rd_logical_blksize,"Logical blocksize of each RAM disk in bytes." ); +MODULE_LICENSE ("GPL" );MODULE_ALIAS_BLOCKDEV_MAJOR (RAMDISK_MAJOR );MODULE_ALIAS ("rd" );
@@ -391,6 +397,8 @@ static int brd_alloc(int i)list_add_tail (&brd->brd_list, &brd_devices); brd->brd_sector_shift =ilog2 (rd_blksize); brd->brd_sector_size = rd_blksize; + brd->brd_logical_sector_shift =ilog2 (rd_logical_blksize); + brd->brd_logical_sector_size = rd_logical_blksize;spin_lock_init (&brd->brd_lock);INIT_RADIX_TREE (&brd->brd_folios,GFP_ATOMIC );
@@ -418,6 +426,7 @@ static int brd_alloc(int i) goto out_cleanup_disk; }blk_queue_physical_block_size (disk->queue, rd_blksize); +blk_queue_logical_block_size (disk->queue, rd_logical_blksize);blk_queue_max_hw_sectors (disk->queue,RD_MAX_SECTOR_SIZE );/* Tell the block layer that this is not a rotational device */
Signature · 1 line
2.35.3
Re: [PATCH 5/5] brd: make logical sector size configurable
drivers/block/brd.c static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) · line 62
Quoted 6 lines
drivers/block/brd.c +1
-1@@ -57,7 +59,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sector_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;
Could we create a simple macro instead of repeating this everywhere? #define RD_SECTOR_SHIFT(brd) (brd->brd_sector_shift - brd->brd_logical_sector_shift)
Quoted 23 lines
drivers/block/brd.c +0
-0/* * The folio lifetime is protected by the fact that we have opened the bio_io_error(bio); drivers/block/brd.c +5
-1return ; } - sector += len >>SECTOR_SHIFT ; + sector += len >> brd->brd_logical_sector_shift; }bio_endio (bio);@@ -353,6 +355,10 @@ static unsigned int rd_blksize = PAGE_SIZE;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." ); +static unsigned int rd_logical_blksize =SECTOR_SIZE ; +module_param (rd_logical_blksize, uint,0444 ); +MODULE_PARM_DESC (rd_logical_blksize,"Logical blocksize of each RAM disk in bytes." ); +MODULE_LICENSE ("GPL" );MODULE_ALIAS_BLOCKDEV_MAJOR (RAMDISK_MAJOR );MODULE_ALIAS ("rd" );
drivers/block/brd.c static int brd_alloc(int i) · line 401
Quoted 5 lines
drivers/block/brd.c +2
-0@@ -391,6 +397,8 @@ static int brd_alloc(int i)list_add_tail (&brd->brd_list, &brd_devices); brd->brd_sector_shift =ilog2 (rd_blksize); brd->brd_sector_size = rd_blksize; + brd->brd_logical_sector_shift =ilog2 (rd_logical_blksize); + brd->brd_logical_sector_size = rd_logical_blksize;
We should a check here to see if logical block > rd_blksize similar to what is done in blk_queue_logical_block_size()?
// physical block size should not be less than the logical block size
if (rd_blksize < rd_logical_blksize) {
brd->brd_logical_sector_shift = ilog2(rd_blksize);
brd->brd_logical_sector_size = rd_blksize;}
Quoted 14 lines
drivers/block/brd.c +1
-0spin_lock_init (&brd->brd_lock);INIT_RADIX_TREE (&brd->brd_folios,GFP_ATOMIC );@@ -418,6 +426,7 @@ static int brd_alloc(int i) goto out_cleanup_disk; }blk_queue_physical_block_size (disk->queue, rd_blksize); +blk_queue_logical_block_size (disk->queue, rd_logical_blksize);blk_queue_max_hw_sectors (disk->queue,RD_MAX_SECTOR_SIZE );/* Tell the block layer that this is not a rotational device */ Signature · 1 line
2.35.3
Re: [PATCH 5/5] brd: make logical sector size configurable
Quoted 10 lines · Pankaj Raghav
drivers/block/brd.c static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) · line 62
Quoted 6 lines
drivers/block/brd.c +1
-1@@ -57,7 +59,7 @@ static struct folio *brd_lookup_folio(struct brd_device *brd, sector_t sector) {pgoff_t idx;struct folio *folio; -unsigned int rd_sector_shift = brd->brd_sector_shift -SECTOR_SHIFT ; +unsigned int rd_sector_shift = brd->brd_sector_shift - brd->brd_logical_sector_shift;Could we create a simple macro instead of repeating this everywhere? #define
RD_SECTOR_SHIFT(brd)(brd->brd_sector_shift-brd->brd_logical_sector_shift)
Yeah, I'm not utterly happy with that one, too; this patchset is primarily a mechanical conversion to avoid errors. Will be changing it.
Quoted 31 lines
Quoted 23 lines
drivers/block/brd.c +0
-0/* * The folio lifetime is protected by the fact that we have opened the bio_io_error(bio); return; }drivers/block/brd.c +5
-1- sector += len >>SECTOR_SHIFT ; + sector += len >> brd->brd_logical_sector_shift; }bio_endio (bio);@@ -353,6 +355,10 @@ static unsigned int rd_blksize = PAGE_SIZE;module_param (rd_blksize, uint,0444 );MODULE_PARM_DESC (rd_blksize,"Blocksize of each RAM disk in bytes." ); +static unsigned int rd_logical_blksize =SECTOR_SIZE ; +module_param (rd_logical_blksize, uint,0444 ); +MODULE_PARM_DESC (rd_logical_blksize,"Logical blocksize of each RAM disk in bytes." ); +MODULE_LICENSE ("GPL" );MODULE_ALIAS_BLOCKDEV_MAJOR (RAMDISK_MAJOR );MODULE_ALIAS ("rd" );drivers/block/brd.c static int brd_alloc(int i) · line 401
Quoted 5 lines
drivers/block/brd.c +2
-0@@ -391,6 +397,8 @@ static int brd_alloc(int i)list_add_tail (&brd->brd_list, &brd_devices); brd->brd_sector_shift =ilog2 (rd_blksize); brd->brd_sector_size = rd_blksize; + brd->brd_logical_sector_shift =ilog2 (rd_logical_blksize); + brd->brd_logical_sector_size = rd_logical_blksize;We should a check here to see if logical block >
rd_blksizesimilar to what is done inblk_queue_logical_block_size()?
Quoted 6 lines
Review static int brd_alloc(int i) · line 402
Quoted 1 line
(quoted diff) +0
-0// physical block size should not be less than the logical block sizeif (rd_blksize < rd_logical_blksize) { brd->brd_logical_sector_shift = ilog2(rd_blksize); brd->brd_logical_sector_size = rd_blksize;}
Sure. Keith already complained about it.
Cheers,
Hannes
Signature · 5 lines
Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions Germany GmbH, Frankenstr. 146, 90461 Nürnberg Managing Directors: I. Totev, A. Myers, A. McDonald, M. B. Moerman (HRB 36809, AG Nürnberg)
Re: [PATCH 5/5] brd: make logical sector size configurable
In reply to Hannes ReineckeHi Hannes,
drivers/block/brd.c static void brd_submit_bio(struct bio *bio) · line 311
Quoted 2 lines
drivers/block/brd.c +0
-0@@ -309,8 +311,8 @@ static void brd_submit_bio(struct bio *bio)int err;
I think you missed the conversion from "kernel" logical sectors to the device logical sector here:sector_t sector = bio->bi_iter.bi_sector >>
(brd->brd_logical_sector_shift - SECTOR_SHIFT);
This fixes the issue when you try to do a fio with -verify on a 4k logical and physical block size.
Quoted 10 lines
drivers/block/brd.c +2
-2/* Don't support un-aligned buffer */ -WARN_ON_ONCE ((iter.offset & (SECTOR_SIZE -1 )) || - (len & (SECTOR_SIZE -1 ))); +WARN_ON_ONCE ((iter.offset & (brd->brd_logical_sector_size -1 )) || + (len & (brd->brd_logical_sector_size -1 ))); err =brd_do_folio (brd, iter.folio, len, iter.offset, bio->bi_opf, sector);
drivers/block/brd.c static void brd_submit_bio(struct bio *bio) · line 327
Quoted 5 lines
drivers/block/brd.c +1
-1@@ -322,7 +324,7 @@ static void brd_submit_bio(struct bio *bio)bio_io_error (bio);return ; } - sector += len >>SECTOR_SHIFT ; + sector += len >> brd->brd_logical_sector_shift;
As you can see here, you divide the len with device logical block size to go to the next sector.
Quoted 3 lines
drivers/block/brd.c +0
-0}bio_endio (bio);
Hi Hannes,
Quoted 8 lines · Hannes Reinecke
Hi all,
meat to the bone: with this patchset one can change the physical and logical block size of the 'brd' ramdisk driver.
Default is 512 (for both); one can easily increase the physical block size to 16k and the logical block size to 4k.
Increasing the logcial block size beyond 4k gives some 'interesting' crashes.
I did something similar for an internal prototype to test large block size. I ran the perf test suite I created on your changes and I can clearly see a perf increase for large block IOs.
I enabled huge pages and used the iomem=mmaphuge option in fio to test large block IOs. Here are my results:
base: next-20230307
new: base with your changes on top
bw: bandwith in MiB/s
For each set of rd_blksize and rd_logical_blksize, I ran the test with different fio blocksize as indicated by io_uring_iod_128_bs_4k
rd_blksize=8192 and rd_logical_blksize=4096
+------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_4k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +------------------------+-------------+------------+-------+----------+---------+-------+ | read | 567 | 605 | 6.7 | 2214 | 2362 | 6.68 | | randread | 517 | 529 | 2.32 | 2019 | 2066 | 2.33 | | write | 551 | 558 | 1.27 | 2154 | 2179 | 1.16 | | randwrite | 481 | 502 | 4.37 | 1880 | 1962 | 4.36 | +------------------------+-------------+------------+-------+----------+---------+-------+ +------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_8k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +------------------------+-------------+------------+-------+----------+---------+-------+ | read | 462 | 512 | 10.82 | 3611 | 3997 | 10.69 | | randread | 426 | 445 | 4.46 | 3326 | 3480 | 4.63 | | write | 442 | 472 | 6.79 | 3454 | 3687 | 6.75 | | randwrite | 401 | 426 | 6.23 | 3134 | 3328 | 6.19 | +------------------------+-------------+------------+-------+----------+---------+-------+ +-------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_16k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +-------------------------+-------------+------------+-------+----------+---------+-------+ | read | 343 | 390 | 13.7 | 5360 | 6086 | 13.54 | | randread | 317 | 364 | 14.83 | 4946 | 5694 | 15.12 | | write | 335 | 346 | 3.28 | 5235 | 5414 | 3.42 | | randwrite | 305 | 327 | 7.21 | 4759 | 5106 | 7.29 | +-------------------------+-------------+------------+-------+----------+---------+-------+
rd_blksize=16384 and rd_logical_blksize=4096
+------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_4k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +------------------------+-------------+------------+-------+----------+---------+-------+ | read | 576 | 586 | 1.74 | 2250 | 2291 | 1.82 | | randread | 524 | 548 | 4.58 | 2046 | 2139 | 4.55 | | write | 533 | 545 | 2.25 | 2081 | 2129 | 2.31 | | randwrite | 484 | 496 | 2.48 | 1892 | 1938 | 2.43 | +------------------------+-------------+------------+-------+----------+---------+-------+ +------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_8k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +------------------------+-------------+------------+-------+----------+---------+-------+ | read | 461 | 491 | 6.51 | 3601 | 3836 | 6.53 | | randread | 425 | 472 | 11.06 | 3323 | 3684 | 10.86 | | write | 454 | 465 | 2.42 | 3543 | 3632 | 2.51 | | randwrite | 395 | 430 | 8.86 | 3086 | 3357 | 8.78 | +------------------------+-------------+------------+-------+----------+---------+-------+ +-------------------------+-------------+------------+-------+----------+---------+-------+ | io_uring_iod_128_bs_16k | base[kiops] | new[kiops] | delta | base[bw] | new[bw] | delta | +-------------------------+-------------+------------+-------+----------+---------+-------+ | read | 338 | 400 | 18.34 | 5282 | 6255 | 18.42 | | randread | 317 | 384 | 21.14 | 4959 | 5997 | 20.93 | | write | 335 | 354 | 5.67 | 5239 | 5525 | 5.46 | | randwrite | 303 | 326 | 7.59 | 4728 | 5097 | 7.8 | +-------------------------+-------------+------------+-------+----------+---------+-------+