* [PATCH 1/3] btrfs: introduce rbio_nr_data_sectors() helper
2026-10-22 9:39 [PATCH 0/3] btrfs: raid56: fully support stripe_len < PAGE_SIZE Qu Wenruo
@ 2026-10-22 9:39 ` Qu Wenruo
2026-10-26 15:22 ` Johannes Thumshirn
2026-10-22 9:39 ` [PATCH 2/3] btrfs: add stripe_len < PAGE_SIZE support for steal_rbio() Qu Wenruo
2026-10-22 9:39 ` [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages() Qu Wenruo
2 siblings, 1 reply; 8+ messages in thread
From: Qu Wenruo @ 2026-10-22 9:39 UTC (permalink / raw)
To: linux-btrfs
The pattern "rbio->nr_data * rbio->stripe_nsectors" is utilized more
than 5 times inside raid56.c, it's worthy to extract it into a new
helper.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/raid56.c | 27 ++++++++++++++++-----------
1 file changed, 16 insertions(+), 11 deletions(-)
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index 59aad807f98a..0c792d85f520 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -246,6 +246,12 @@ static void memcpy_from_bio_to_stripe(struct btrfs_raid_bio *rbio, unsigned int
}
}
+/* Return how many data blocks there is inside the rbio. */
+static unsigned int rbio_nr_data_sectors(const struct btrfs_raid_bio *rbio)
+{
+ return rbio->nr_data * rbio->stripe_nsectors;
+}
+
/*
* caching an rbio means to copy anything from the
* bio_sectors array into the stripe_pages array. We
@@ -272,7 +278,7 @@ static void cache_rbio_pages(struct btrfs_raid_bio *rbio)
* a data sector it should still be uptodate as it is
* read from disk.
*/
- if (i < rbio->nr_data * rbio->stripe_nsectors)
+ if (i < rbio_nr_data_sectors(rbio))
ASSERT(test_bit(i, rbio->stripe_uptodate_bitmap));
continue;
}
@@ -402,7 +408,7 @@ static bool is_data_stripe_page(struct btrfs_raid_bio *rbio, int page_nr)
* Thus if the first sector of the page belongs to data stripes, then
* the full page belongs to data stripes.
*/
- return (sector_nr < rbio->nr_data * rbio->stripe_nsectors);
+ return (sector_nr < rbio_nr_data_sectors(rbio));
}
/*
@@ -1559,7 +1565,7 @@ static void set_rbio_range_error(struct btrfs_raid_bio *rbio, struct bio *bio)
rbio->bioc->full_stripe_logical;
int total_nr_sector = offset >> fs_info->sectorsize_bits;
- ASSERT(total_nr_sector < rbio->nr_data * rbio->stripe_nsectors);
+ ASSERT(total_nr_sector < rbio_nr_data_sectors(rbio));
bitmap_set(rbio->error_bitmap, total_nr_sector,
bio->bi_iter.bi_size >> fs_info->sectorsize_bits);
@@ -1669,7 +1675,7 @@ static void verify_bio_data_sectors(struct btrfs_raid_bio *rbio,
return;
/* P/Q stripes, they have no data csum to verify against. */
- if (total_sector_nr >= rbio->nr_data * rbio->stripe_nsectors)
+ if (total_sector_nr >= rbio_nr_data_sectors(rbio))
return;
for (struct bvec_iter iter = init_bvec_iter_for_bio(bio);
@@ -2303,9 +2309,9 @@ static void fill_data_csums(struct btrfs_raid_bio *rbio)
{
struct btrfs_fs_info *fs_info = rbio->bioc->fs_info;
struct btrfs_root *csum_root;
+ const unsigned int nr_data_sectors = rbio_nr_data_sectors(rbio);
const u64 start = rbio->bioc->full_stripe_logical;
- const u32 len = (rbio->nr_data * rbio->stripe_nsectors) <<
- fs_info->sectorsize_bits;
+ const u32 len = nr_data_sectors << fs_info->sectorsize_bits;
int ret;
/* The rbio should not have its csum buffer initialized. */
@@ -2326,10 +2332,8 @@ static void fill_data_csums(struct btrfs_raid_bio *rbio)
rbio->bioc->map_type & BTRFS_BLOCK_GROUP_METADATA)
return;
- rbio->csum_buf = kzalloc(rbio->nr_data * rbio->stripe_nsectors *
- fs_info->csum_size, GFP_NOFS);
- rbio->csum_bitmap = bitmap_zalloc(rbio->nr_data * rbio->stripe_nsectors,
- GFP_NOFS);
+ rbio->csum_buf = kzalloc(nr_data_sectors * fs_info->csum_size, GFP_NOFS);
+ rbio->csum_bitmap = bitmap_zalloc(nr_data_sectors, GFP_NOFS);
if (!rbio->csum_buf || !rbio->csum_bitmap) {
ret = -ENOMEM;
goto error;
@@ -2445,9 +2449,10 @@ static void submit_write_bios(struct btrfs_raid_bio *rbio,
*/
static bool need_read_stripe_sectors(struct btrfs_raid_bio *rbio)
{
+ const unsigned int nr_data_sectors = rbio_nr_data_sectors(rbio);
int i;
- for (i = 0; i < rbio->nr_data * rbio->stripe_nsectors; i++) {
+ for (i = 0; i < nr_data_sectors; i++) {
phys_addr_t paddr = rbio->stripe_paddrs[i * rbio->sector_nsteps];
/*
--
2.56.0
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 2/3] btrfs: add stripe_len < PAGE_SIZE support for steal_rbio()
2026-10-22 9:39 [PATCH 0/3] btrfs: raid56: fully support stripe_len < PAGE_SIZE Qu Wenruo
2026-10-22 9:39 ` [PATCH 1/3] btrfs: introduce rbio_nr_data_sectors() helper Qu Wenruo
@ 2026-10-22 9:39 ` Qu Wenruo
2026-10-26 15:33 ` Johannes Thumshirn
2026-10-22 9:39 ` [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages() Qu Wenruo
2 siblings, 1 reply; 8+ messages in thread
From: Qu Wenruo @ 2026-10-22 9:39 UTC (permalink / raw)
To: linux-btrfs
The function steal_rbio() is utilized to use the pages that belongs to
uptodate data blocks, without doing a page copy.
That is done by just updating the dest->stripe_pages[] pointer to use
the point inside @src directly.
That function previously will reject stripe_len < PAGE_SIZE cases, as a
page can contain both data and P/Q blocks.
But that shouldn't be a big deal, as long as the corresponding P/Q
blocks are not marked uptodate, we can still steal the page that also
contains P/Q blocks.
So rework steal_rbio() to support stripe_len < PAGE_SIZE by:
- Calculate the last page index correctly
Instead of iterating all pages but skip P/Q pages, calculate the last
page index properly.
The last page may contain both data and P/Q blocks, but we can still
steal the page.
- Update the bitmap in one go
Instead of updating the stripe_uptodate_bitmap for every page, we can
do the update in just one go, out of the main loop.
And that update should not touch the P/Q blocks.
- Remove unused helpers
* page_nr_to_sector_nr()
* page_nr_to_num_sectors()
* steal_rbio_page()
* is_data_stripe_page()
* full_page_sectors_uptodate()
This one is very annoying, because of its __maybe_unused
attribution.
They are based on the old BTRFS_STRIPE_LEN >= PAGE_SIZE assumption,
and no longer utilized.
- Make index_stripe_sectors() update the stripe_paddrs[] without a page
This doesn't cause any functional change for now, as there is no one
accessing the source stripe_paddrs[] after a steal.
But this should still make the function a little more robust.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/raid56.c | 127 +++++++++-------------------------------------
1 file changed, 23 insertions(+), 104 deletions(-)
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index 0c792d85f520..c81f7a0297e4 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -307,55 +307,8 @@ static int rbio_bucket(struct btrfs_raid_bio *rbio)
return hash_64(num >> 16, BTRFS_STRIPE_HASH_TABLE_BITS);
}
-/* Get the sector number of the first sector covered by @page_nr. */
-static u32 page_nr_to_sector_nr(struct btrfs_raid_bio *rbio, unsigned int page_nr)
-{
- u32 sector_nr;
-
- ASSERT(page_nr < rbio->nr_pages);
-
- sector_nr = (page_nr << PAGE_SHIFT) >> rbio->bioc->fs_info->sectorsize_bits;
- ASSERT(sector_nr < rbio->nr_sectors);
- return sector_nr;
-}
-
/*
- * Get the number of sectors covered by @page_nr.
- *
- * For bs > ps cases, the result will always be 1.
- * For bs <= ps cases, the result will be ps / bs.
- */
-static u32 page_nr_to_num_sectors(struct btrfs_raid_bio *rbio, unsigned int page_nr)
-{
- struct btrfs_fs_info *fs_info = rbio->bioc->fs_info;
- u32 nr_sectors;
-
- ASSERT(page_nr < rbio->nr_pages);
-
- nr_sectors = round_up(PAGE_SIZE, fs_info->sectorsize) >> fs_info->sectorsize_bits;
- ASSERT(nr_sectors > 0);
- return nr_sectors;
-}
-
-static __maybe_unused bool full_page_sectors_uptodate(struct btrfs_raid_bio *rbio,
- unsigned int page_nr)
-{
- const u32 sector_nr = page_nr_to_sector_nr(rbio, page_nr);
- const u32 nr_bits = page_nr_to_num_sectors(rbio, page_nr);
- int i;
-
- ASSERT(page_nr < rbio->nr_pages);
- ASSERT(sector_nr + nr_bits < rbio->nr_sectors);
-
- for (i = sector_nr; i < sector_nr + nr_bits; i++) {
- if (!test_bit(i, rbio->stripe_uptodate_bitmap))
- return false;
- }
- return true;
-}
-
-/*
- * Update the stripe_sectors[] array to use correct page and pgoff
+ * Update the stripe_paddrs[] array to use correct paddr.
*
* Should be called every time any page pointer in stripes_pages[] got modified.
*/
@@ -370,82 +323,48 @@ static void index_stripe_sectors(struct btrfs_raid_bio *rbio)
int page_index = offset >> PAGE_SHIFT;
ASSERT(page_index < rbio->nr_pages);
- if (!rbio->stripe_pages[page_index])
+ if (!rbio->stripe_pages[page_index]) {
+ rbio->stripe_paddrs[i] = INVALID_PADDR;
continue;
+ }
rbio->stripe_paddrs[i] = page_to_phys(rbio->stripe_pages[page_index]) +
offset_in_page(offset);
}
}
-static void steal_rbio_page(struct btrfs_raid_bio *src,
- struct btrfs_raid_bio *dest, int page_nr)
-{
- const u32 sector_nr = page_nr_to_sector_nr(src, page_nr);
- const u32 nr_bits = page_nr_to_num_sectors(src, page_nr);
-
- ASSERT(page_nr < src->nr_pages);
- ASSERT(sector_nr + nr_bits < src->nr_sectors);
-
- if (dest->stripe_pages[page_nr])
- __free_page(dest->stripe_pages[page_nr]);
- dest->stripe_pages[page_nr] = src->stripe_pages[page_nr];
- src->stripe_pages[page_nr] = NULL;
-
- /* Also update the stripe_uptodate_bitmap bits. */
- bitmap_set(dest->stripe_uptodate_bitmap, sector_nr, nr_bits);
-}
-
-static bool is_data_stripe_page(struct btrfs_raid_bio *rbio, int page_nr)
-{
- const int sector_nr = page_nr_to_sector_nr(rbio, page_nr);
-
- /*
- * The caller has ensured stripe_len >= PAGE_SIZE, and PAGE_SIZE is
- * aligned with sectorsize, thus we won't have a page which is half
- * data half parity.
- *
- * Thus if the first sector of the page belongs to data stripes, then
- * the full page belongs to data stripes.
- */
- return (sector_nr < rbio_nr_data_sectors(rbio));
-}
-
/*
- * Stealing an rbio means taking all the uptodate pages from the stripe array
- * in the source rbio and putting them into the destination rbio.
+ * Stealing an rbio means taking all pages containing data blocks from the
+ * stripe array in the source rbio and putting them into the destination rbio.
*
- * This will also update the involved stripe_sectors[] which are referring to
+ * This will also update the involved stripe_paddrs[] which are referring to
* the old pages.
*/
static void steal_rbio(struct btrfs_raid_bio *src, struct btrfs_raid_bio *dest)
{
- int i;
+ const u32 nr_data_blocks = rbio_nr_data_sectors(src);
+ const u32 last_data_pos = nr_data_blocks << src->bioc->fs_info->sectorsize_bits;
+ const u32 last_pindex = round_up(last_data_pos, PAGE_SIZE) >> PAGE_SHIFT;
if (!test_bit(RBIO_CACHE_READY_BIT, &src->flags))
return;
- /* Data and P/Q can be mixed into a page, skip stealing. */
- if (src->bioc->stripe_len < PAGE_SIZE)
- return;
- for (i = 0; i < dest->nr_pages; i++) {
- struct page *p = src->stripe_pages[i];
+ /* All data blocks in the src should be uptodate. */
+ ASSERT(bitmap_full(src->stripe_uptodate_bitmap, nr_data_blocks));
- /*
- * We don't need to steal P/Q pages as they will always be
- * regenerated for RMW or full write anyway.
- */
- if (!is_data_stripe_page(src, i))
- continue;
+ /* Steal all pages that contains any data block. */
+ for (int i = 0; i < last_pindex; i++) {
+ ASSERT(src->stripe_pages[i]);
- /*
- * If @src already has RBIO_CACHE_READY_BIT, it should have
- * all data stripe pages present and uptodate.
- */
- ASSERT(p);
- ASSERT(full_page_sectors_uptodate(src, i));
- steal_rbio_page(src, dest, i);
+ if (dest->stripe_pages[i])
+ __free_page(dest->stripe_pages[i]);
+ dest->stripe_pages[i] = src->stripe_pages[i];
+ src->stripe_pages[i] = NULL;
}
+
+ /* Only data blocks should be set uptodate. */
+ bitmap_set(dest->stripe_uptodate_bitmap, 0, nr_data_blocks);
+
index_stripe_sectors(dest);
index_stripe_sectors(src);
}
--
2.56.0
^ permalink raw reply related [flat|nested] 8+ messages in thread* [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages()
2026-10-22 9:39 [PATCH 0/3] btrfs: raid56: fully support stripe_len < PAGE_SIZE Qu Wenruo
2026-10-22 9:39 ` [PATCH 1/3] btrfs: introduce rbio_nr_data_sectors() helper Qu Wenruo
2026-10-22 9:39 ` [PATCH 2/3] btrfs: add stripe_len < PAGE_SIZE support for steal_rbio() Qu Wenruo
@ 2026-10-22 9:39 ` Qu Wenruo
2026-10-26 15:36 ` Johannes Thumshirn
2 siblings, 1 reply; 8+ messages in thread
From: Qu Wenruo @ 2026-10-22 9:39 UTC (permalink / raw)
To: linux-btrfs
The functions alloc_rbio_parity_pages() and alloc_rbio_data_pages()
didn't support stripe_len < PAGE_SIZE cases, and fallback to
alloc_rbio_pages() instead.
That's mostly to avoid the case where both data and P/Q blocks can exist
in the same page.
However even in that case, it's not a big deal, the existing
infrastructures like index_stripe_sectors() can handle it correctly.
Update those functions to support stripe_len < PAGE_SIZE cases, with two
points to note:
- Extra atttention needed for the round direction
For alloc_rbio_data_pages() we need to round up, so the last data page
gets allocated.
For alloc_rbio_parity_pages() we need to round down, so the first
P/Q page get allocated.
- Mixed page allocation
We can call alloc_rbio_parity_pages() first, then
alloc_rbio_data_pages().
In that case, it's alloc_pages_bulk() saving us, which always skips
allocated slot. Thus we're still safe.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/raid56.c | 30 +++++++++++++-----------------
1 file changed, 13 insertions(+), 17 deletions(-)
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index c81f7a0297e4..7dd937fa0c26 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -1061,22 +1061,21 @@ static int alloc_rbio_pages(struct btrfs_raid_bio *rbio)
return 0;
}
-/* only allocate pages for p/q stripes */
+/*
+ * Allocate pages that cover P/Q blocks.
+ *
+ * For stripe_len < PAGE_SIZE case, we can allocate a page that cover both
+ * data and P/Q blocks.
+ */
static int alloc_rbio_parity_pages(struct btrfs_raid_bio *rbio)
{
- int data_pages;
+ const unsigned int first_pq_page = round_down(rbio->nr_data *
+ rbio->bioc->stripe_len, PAGE_SIZE) >>
+ PAGE_SHIFT;
int ret;
- /*
- * Page size larger than stripe size, we can not only allocate
- * pages for P/Q stripes.
- */
- if (rbio->bioc->stripe_len < PAGE_SIZE)
- return alloc_rbio_pages(rbio);
-
- data_pages = rbio->nr_data * (rbio->bioc->stripe_len >> PAGE_SHIFT);
- ret = btrfs_alloc_page_array(rbio->nr_pages - data_pages,
- rbio->stripe_pages + data_pages, GFP_NOFS);
+ ret = btrfs_alloc_page_array(rbio->nr_pages - first_pq_page,
+ rbio->stripe_pages + first_pq_page, GFP_NOFS);
if (ret < 0)
return ret;
@@ -1653,13 +1652,10 @@ static void submit_read_wait_bio_list(struct btrfs_raid_bio *rbio,
static int alloc_rbio_data_pages(struct btrfs_raid_bio *rbio)
{
- int data_pages;
+ const unsigned int data_pages = round_up(rbio->nr_data *
+ rbio->bioc->stripe_len, PAGE_SIZE) >> PAGE_SHIFT;
int ret;
- if (rbio->bioc->stripe_len < PAGE_SIZE)
- return alloc_rbio_pages(rbio);
-
- data_pages = rbio->nr_data * (rbio->bioc->stripe_len >> PAGE_SHIFT);
ret = btrfs_alloc_page_array(data_pages, rbio->stripe_pages, GFP_NOFS);
if (ret < 0)
return ret;
--
2.56.0
^ permalink raw reply related [flat|nested] 8+ messages in thread* Re: [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages()
2026-10-22 9:39 ` [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages() Qu Wenruo
@ 2026-10-26 15:36 ` Johannes Thumshirn
2026-10-26 20:56 ` Qu Wenruo
0 siblings, 1 reply; 8+ messages in thread
From: Johannes Thumshirn @ 2026-10-26 15:36 UTC (permalink / raw)
To: Qu Wenruo, linux-btrfs
On 10/22/26 11:39 AM, Qu Wenruo wrote:
> + const unsigned int first_pq_page = round_down(rbio->nr_data *
> + rbio->bioc->stripe_len, PAGE_SIZE) >>
> + PAGE_SHIFT;
> static int alloc_rbio_data_pages(struct btrfs_raid_bio *rbio)
> {
> - int data_pages;
> + const unsigned int data_pages = round_up(rbio->nr_data *
> + rbio->bioc->stripe_len, PAGE_SIZE) >> PAGE_SHIFT;
This is the same calculation twice, maybe extract into a 'unsigned int
rbio_nr_data_pages(rbio)'?
^ permalink raw reply [flat|nested] 8+ messages in thread* Re: [PATCH 3/3] btrfs: add stripe_len < PAGE_SIZE support for alloc_rbio_*_pages()
2026-10-26 15:36 ` Johannes Thumshirn
@ 2026-10-26 20:56 ` Qu Wenruo
0 siblings, 0 replies; 8+ messages in thread
From: Qu Wenruo @ 2026-10-26 20:56 UTC (permalink / raw)
To: Johannes Thumshirn, Qu Wenruo, linux-btrfs
在 2026/10/27 02:06, Johannes Thumshirn 写道:
> On 10/22/26 11:39 AM, Qu Wenruo wrote:
>> + const unsigned int first_pq_page = round_down(rbio->nr_data *
>> + rbio->bioc->stripe_len, PAGE_SIZE) >>
>> + PAGE_SHIFT;
>
>> static int alloc_rbio_data_pages(struct btrfs_raid_bio *rbio)
>> {
>> - int data_pages;
>> + const unsigned int data_pages = round_up(rbio->nr_data *
>> + rbio->bioc->stripe_len, PAGE_SIZE) >> PAGE_SHIFT;
>
> This is the same calculation twice, maybe extract into a 'unsigned int
> rbio_nr_data_pages(rbio)'?
Please note that, one is round_up(), one is round_down(). So it's not
the same unfortunately.
Thanks,
Qu
^ permalink raw reply [flat|nested] 8+ messages in thread