* [PATCH RFC 1/8] btrfs: prepare compression folio alloc/free for bs > ps cases
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 2/8] btrfs: prepare zstd to support " Qu Wenruo
` (6 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
This includes the following preparation for bs > ps cases:
- Always alloc/free the folio directly if bs > ps
This adds a new @fs_info parameter for btrfs_alloc_compr_folio(), thus
affecting all compression algorithms.
For btrfs_free_compr_folio() it needs no parameter for now, as we can
use the folio size to skip the caching part.
For now the change is just to passing a @fs_info into the function,
all the folio size assumption is still based on page size.
- Properly zero the last folio in compress_file_range()
Since the compressed folios can be larger than a page, we need to
properly zero the whole folio.
- Use correct folio size for btrfs_add_compressed_bio_folios()
Instead of page size, use the correct folio size.
- Use correct folio size/shift for btrfs_compress_filemap_get_folio()
As we are not only using simple page sized folios anymore.
- Use correct folio size for btrfs_decompress()
There is an ASSERT() making sure the decompressed range is no larger
than a page, which will be triggered for bs > ps cases.
- Skip readahead for compressed pages
Similar to subpage cases.
- Make btrfs_alloc_folio_array() to accept a new @order parameter
- Add a helper to calculate the minimal folio size
All those changes should not affect the existing bs <= ps handling.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/compression.c | 42 ++++++++++++++++++++++++++++++------------
fs/btrfs/compression.h | 2 +-
fs/btrfs/extent_io.c | 7 +++++--
fs/btrfs/extent_io.h | 3 ++-
fs/btrfs/fs.h | 6 ++++++
fs/btrfs/inode.c | 16 +++++++++-------
fs/btrfs/lzo.c | 18 ++++++++++--------
fs/btrfs/zlib.c | 13 +++++++------
fs/btrfs/zstd.c | 15 ++++++++-------
9 files changed, 78 insertions(+), 44 deletions(-)
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 068339e86123..f653c237bdfc 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -223,10 +223,14 @@ static unsigned long btrfs_compr_pool_scan(struct shrinker *sh, struct shrink_co
/*
* Common wrappers for page allocation from compression wrappers
*/
-struct folio *btrfs_alloc_compr_folio(void)
+struct folio *btrfs_alloc_compr_folio(struct btrfs_fs_info *fs_info)
{
struct folio *folio = NULL;
+ /* For bs > ps cases, no cached folio pool for now. */
+ if (fs_info->block_min_order)
+ goto alloc;
+
spin_lock(&compr_pool.lock);
if (compr_pool.count > 0) {
folio = list_first_entry(&compr_pool.list, struct folio, lru);
@@ -238,13 +242,18 @@ struct folio *btrfs_alloc_compr_folio(void)
if (folio)
return folio;
- return folio_alloc(GFP_NOFS, 0);
+alloc:
+ return folio_alloc(GFP_NOFS, fs_info->block_min_order);
}
void btrfs_free_compr_folio(struct folio *folio)
{
bool do_free = false;
+ /* The folio is from bs > ps fs, no cached pool for now. */
+ if (folio_order(folio))
+ goto free;
+
spin_lock(&compr_pool.lock);
if (compr_pool.count > compr_pool.thresh) {
do_free = true;
@@ -257,6 +266,7 @@ void btrfs_free_compr_folio(struct folio *folio)
if (!do_free)
return;
+free:
ASSERT(folio_ref_count(folio) == 1);
folio_put(folio);
}
@@ -344,16 +354,19 @@ static void end_bbio_compressed_write(struct btrfs_bio *bbio)
static void btrfs_add_compressed_bio_folios(struct compressed_bio *cb)
{
+ struct btrfs_fs_info *fs_info = cb->bbio.fs_info;
struct bio *bio = &cb->bbio.bio;
u32 offset = 0;
while (offset < cb->compressed_len) {
+ struct folio *folio;
int ret;
- u32 len = min_t(u32, cb->compressed_len - offset, PAGE_SIZE);
+ u32 len = min_t(u32, cb->compressed_len - offset,
+ btrfs_min_folio_size(fs_info));
+ folio = cb->compressed_folios[offset >> (PAGE_SHIFT + fs_info->block_min_order)];
/* Maximum compressed extent is smaller than bio size limit. */
- ret = bio_add_folio(bio, cb->compressed_folios[offset >> PAGE_SHIFT],
- len, 0);
+ ret = bio_add_folio(bio, folio, len, 0);
ASSERT(ret);
offset += len;
}
@@ -443,6 +456,10 @@ static noinline int add_ra_bio_pages(struct inode *inode,
if (fs_info->sectorsize < PAGE_SIZE)
return 0;
+ /* For bs > ps cases, we don't support readahead for compressed folios for now. */
+ if (fs_info->block_min_order)
+ return 0;
+
end_index = (i_size_read(inode) - 1) >> PAGE_SHIFT;
while (cur < compressed_end) {
@@ -606,14 +623,15 @@ void btrfs_submit_compressed_read(struct btrfs_bio *bbio)
btrfs_free_extent_map(em);
- cb->nr_folios = DIV_ROUND_UP(compressed_len, PAGE_SIZE);
+ cb->nr_folios = DIV_ROUND_UP(compressed_len, btrfs_min_folio_size(fs_info));
cb->compressed_folios = kcalloc(cb->nr_folios, sizeof(struct folio *), GFP_NOFS);
if (!cb->compressed_folios) {
status = BLK_STS_RESOURCE;
goto out_free_bio;
}
- ret = btrfs_alloc_folio_array(cb->nr_folios, cb->compressed_folios);
+ ret = btrfs_alloc_folio_array(cb->nr_folios, fs_info->block_min_order,
+ cb->compressed_folios);
if (ret) {
status = BLK_STS_RESOURCE;
goto out_free_compressed_pages;
@@ -1033,12 +1051,12 @@ int btrfs_compress_filemap_get_folio(struct address_space *mapping, u64 start,
* - compression algo are 0-3
* - the level are bits 4-7
*
- * @out_pages is an in/out parameter, holds maximum number of pages to allocate
- * and returns number of actually allocated pages
+ * @out_folios is an in/out parameter, holds maximum number of folios to allocate
+ * and returns number of actually allocated folios
*
* @total_in is used to return the number of bytes actually read. It
* may be smaller than the input length if we had to exit early because we
- * ran out of room in the pages array or because we cross the
+ * ran out of room in the folios array or because we cross the
* max_out threshold.
*
* @total_out is an in/out parameter, must be set to the input length and will
@@ -1093,11 +1111,11 @@ int btrfs_decompress(int type, const u8 *data_in, struct folio *dest_folio,
int ret;
/*
- * The full destination page range should not exceed the page size.
+ * The full destination folio range should not exceed the folio size.
* And the @destlen should not exceed sectorsize, as this is only called for
* inline file extents, which should not exceed sectorsize.
*/
- ASSERT(dest_pgoff + destlen <= PAGE_SIZE && destlen <= sectorsize);
+ ASSERT(dest_pgoff + destlen <= folio_size(dest_folio) && destlen <= sectorsize);
workspace = get_workspace(fs_info, type, 0);
ret = compression_decompress(type, workspace, data_in, dest_folio,
diff --git a/fs/btrfs/compression.h b/fs/btrfs/compression.h
index 760d4aac74e6..41d93a977d69 100644
--- a/fs/btrfs/compression.h
+++ b/fs/btrfs/compression.h
@@ -112,7 +112,7 @@ void btrfs_submit_compressed_read(struct btrfs_bio *bbio);
int btrfs_compress_str2level(unsigned int type, const char *str);
-struct folio *btrfs_alloc_compr_folio(void);
+struct folio *btrfs_alloc_compr_folio(struct btrfs_fs_info *fs_info);
void btrfs_free_compr_folio(struct folio *folio);
struct workspace_manager {
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index ca7174fa0240..258658856195 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -618,6 +618,7 @@ static void end_bbio_data_read(struct btrfs_bio *bbio)
* Populate every free slot in a provided array with folios using GFP_NOFS.
*
* @nr_folios: number of folios to allocate
+ * @order: the order of the folios to be allocated
* @folio_array: the array to fill with folios; any existing non-NULL entries in
* the array will be skipped
*
@@ -625,12 +626,13 @@ static void end_bbio_data_read(struct btrfs_bio *bbio)
* -ENOMEM otherwise, the partially allocated folios would be freed and
* the array slots zeroed
*/
-int btrfs_alloc_folio_array(unsigned int nr_folios, struct folio **folio_array)
+int btrfs_alloc_folio_array(unsigned int nr_folios, unsigned int order,
+ struct folio **folio_array)
{
for (int i = 0; i < nr_folios; i++) {
if (folio_array[i])
continue;
- folio_array[i] = folio_alloc(GFP_NOFS, 0);
+ folio_array[i] = folio_alloc(GFP_NOFS, order);
if (!folio_array[i])
goto error;
}
@@ -639,6 +641,7 @@ int btrfs_alloc_folio_array(unsigned int nr_folios, struct folio **folio_array)
for (int i = 0; i < nr_folios; i++) {
if (folio_array[i])
folio_put(folio_array[i]);
+ folio_array[i] = NULL;
}
return -ENOMEM;
}
diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 61130786b9a3..5fcbfe44218c 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -366,7 +366,8 @@ void btrfs_clear_buffer_dirty(struct btrfs_trans_handle *trans,
int btrfs_alloc_page_array(unsigned int nr_pages, struct page **page_array,
bool nofail);
-int btrfs_alloc_folio_array(unsigned int nr_folios, struct folio **folio_array);
+int btrfs_alloc_folio_array(unsigned int nr_folios, unsigned int order,
+ struct folio **folio_array);
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
bool find_lock_delalloc_range(struct inode *inode,
diff --git a/fs/btrfs/fs.h b/fs/btrfs/fs.h
index bf4a1b75b0cf..4a48b19ea863 100644
--- a/fs/btrfs/fs.h
+++ b/fs/btrfs/fs.h
@@ -921,6 +921,12 @@ static inline gfp_t btrfs_alloc_write_mask(struct address_space *mapping)
return mapping_gfp_constraint(mapping, ~__GFP_FS);
}
+/* Return the minimal folio size of the fs. */
+static inline unsigned int btrfs_min_folio_size(struct btrfs_fs_info *fs_info)
+{
+ return 1 << (PAGE_SHIFT + fs_info->block_min_order);
+}
+
static inline u64 btrfs_get_fs_generation(const struct btrfs_fs_info *fs_info)
{
return READ_ONCE(fs_info->generation);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index ad876779289e..6b52ab164f45 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -854,6 +854,8 @@ static void compress_file_range(struct btrfs_work *work)
struct btrfs_inode *inode = async_chunk->inode;
struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct address_space *mapping = inode->vfs_inode.i_mapping;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
+ const u32 min_folio_size = btrfs_min_folio_size(fs_info);
u64 blocksize = fs_info->sectorsize;
u64 start = async_chunk->start;
u64 end = async_chunk->end;
@@ -864,7 +866,7 @@ static void compress_file_range(struct btrfs_work *work)
unsigned long nr_folios;
unsigned long total_compressed = 0;
unsigned long total_in = 0;
- unsigned int poff;
+ unsigned int loff;
int i;
int compress_type = fs_info->compress_type;
int compress_level = fs_info->compress_level;
@@ -902,8 +904,8 @@ static void compress_file_range(struct btrfs_work *work)
actual_end = min_t(u64, i_size, end + 1);
again:
folios = NULL;
- nr_folios = (end >> PAGE_SHIFT) - (start >> PAGE_SHIFT) + 1;
- nr_folios = min_t(unsigned long, nr_folios, BTRFS_MAX_COMPRESSED_PAGES);
+ nr_folios = (end >> min_folio_shift) - (start >> min_folio_shift) + 1;
+ nr_folios = min_t(unsigned long, nr_folios, BTRFS_MAX_COMPRESSED >> min_folio_shift);
/*
* we don't want to send crud past the end of i_size through
@@ -965,12 +967,12 @@ static void compress_file_range(struct btrfs_work *work)
goto mark_incompressible;
/*
- * Zero the tail end of the last page, as we might be sending it down
+ * Zero the tail end of the last folio, as we might be sending it down
* to disk.
*/
- poff = offset_in_page(total_compressed);
- if (poff)
- folio_zero_range(folios[nr_folios - 1], poff, PAGE_SIZE - poff);
+ loff = total_compressed & (min_folio_size - 1);
+ if (loff)
+ folio_zero_range(folios[nr_folios - 1], loff, min_folio_size - loff);
/*
* Try to create an inline extent.
diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
index 047d90e216f6..c5a25fd872bd 100644
--- a/fs/btrfs/lzo.c
+++ b/fs/btrfs/lzo.c
@@ -132,13 +132,14 @@ static inline size_t read_compress_length(const char *buf)
*
* Will allocate new pages when needed.
*/
-static int copy_compressed_data_to_page(char *compressed_data,
+static int copy_compressed_data_to_page(struct btrfs_fs_info *fs_info,
+ char *compressed_data,
size_t compressed_size,
struct folio **out_folios,
unsigned long max_nr_folio,
- u32 *cur_out,
- const u32 sectorsize)
+ u32 *cur_out)
{
+ const u32 sectorsize = fs_info->sectorsize;
u32 sector_bytes_left;
u32 orig_out;
struct folio *cur_folio;
@@ -156,7 +157,7 @@ static int copy_compressed_data_to_page(char *compressed_data,
cur_folio = out_folios[*cur_out / PAGE_SIZE];
/* Allocate a new page */
if (!cur_folio) {
- cur_folio = btrfs_alloc_compr_folio();
+ cur_folio = btrfs_alloc_compr_folio(fs_info);
if (!cur_folio)
return -ENOMEM;
out_folios[*cur_out / PAGE_SIZE] = cur_folio;
@@ -182,7 +183,7 @@ static int copy_compressed_data_to_page(char *compressed_data,
cur_folio = out_folios[*cur_out / PAGE_SIZE];
/* Allocate a new page */
if (!cur_folio) {
- cur_folio = btrfs_alloc_compr_folio();
+ cur_folio = btrfs_alloc_compr_folio(fs_info);
if (!cur_folio)
return -ENOMEM;
out_folios[*cur_out / PAGE_SIZE] = cur_folio;
@@ -217,8 +218,9 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
u64 start, struct folio **folios, unsigned long *out_folios,
unsigned long *total_in, unsigned long *total_out)
{
+ struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct workspace *workspace = list_entry(ws, struct workspace, list);
- const u32 sectorsize = inode->root->fs_info->sectorsize;
+ const u32 sectorsize = fs_info->sectorsize;
struct address_space *mapping = inode->vfs_inode.i_mapping;
struct folio *folio_in = NULL;
char *sizes_ptr;
@@ -268,9 +270,9 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
goto out;
}
- ret = copy_compressed_data_to_page(workspace->cbuf, out_len,
+ ret = copy_compressed_data_to_page(fs_info, workspace->cbuf, out_len,
folios, max_nr_folio,
- &cur_out, sectorsize);
+ &cur_out);
if (ret < 0)
goto out;
diff --git a/fs/btrfs/zlib.c b/fs/btrfs/zlib.c
index d72566a87afa..ccf77a0fa96c 100644
--- a/fs/btrfs/zlib.c
+++ b/fs/btrfs/zlib.c
@@ -136,6 +136,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
u64 start, struct folio **folios, unsigned long *out_folios,
unsigned long *total_in, unsigned long *total_out)
{
+ struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct workspace *workspace = list_entry(ws, struct workspace, list);
struct address_space *mapping = inode->vfs_inode.i_mapping;
int ret;
@@ -147,7 +148,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
unsigned long len = *total_out;
unsigned long nr_dest_folios = *out_folios;
const unsigned long max_out = nr_dest_folios * PAGE_SIZE;
- const u32 blocksize = inode->root->fs_info->sectorsize;
+ const u32 blocksize = fs_info->sectorsize;
const u64 orig_end = start + len;
*out_folios = 0;
@@ -156,7 +157,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = zlib_deflateInit(&workspace->strm, workspace->level);
if (unlikely(ret != Z_OK)) {
- btrfs_err(inode->root->fs_info,
+ btrfs_err(fs_info,
"zlib compression init failed, error %d root %llu inode %llu offset %llu",
ret, btrfs_root_id(inode->root), btrfs_ino(inode), start);
ret = -EIO;
@@ -166,7 +167,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
workspace->strm.total_in = 0;
workspace->strm.total_out = 0;
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
@@ -224,7 +225,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = zlib_deflate(&workspace->strm, Z_SYNC_FLUSH);
if (unlikely(ret != Z_OK)) {
- btrfs_warn(inode->root->fs_info,
+ btrfs_warn(fs_info,
"zlib compression failed, error %d root %llu inode %llu offset %llu",
ret, btrfs_root_id(inode->root), btrfs_ino(inode),
start);
@@ -249,7 +250,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = -E2BIG;
goto out;
}
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
@@ -285,7 +286,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = -E2BIG;
goto out;
}
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
diff --git a/fs/btrfs/zstd.c b/fs/btrfs/zstd.c
index b11a87842cda..28e2e99a2463 100644
--- a/fs/btrfs/zstd.c
+++ b/fs/btrfs/zstd.c
@@ -400,6 +400,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
u64 start, struct folio **folios, unsigned long *out_folios,
unsigned long *total_in, unsigned long *total_out)
{
+ struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct workspace *workspace = list_entry(ws, struct workspace, list);
struct address_space *mapping = inode->vfs_inode.i_mapping;
zstd_cstream *stream;
@@ -412,7 +413,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
unsigned long len = *total_out;
const unsigned long nr_dest_folios = *out_folios;
const u64 orig_end = start + len;
- const u32 blocksize = inode->root->fs_info->sectorsize;
+ const u32 blocksize = fs_info->sectorsize;
unsigned long max_out = nr_dest_folios * PAGE_SIZE;
unsigned int cur_len;
@@ -425,7 +426,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
stream = zstd_init_cstream(&workspace->params, len, workspace->mem,
workspace->size);
if (unlikely(!stream)) {
- btrfs_err(inode->root->fs_info,
+ btrfs_err(fs_info,
"zstd compression init level %d failed, root %llu inode %llu offset %llu",
workspace->req_level, btrfs_root_id(inode->root),
btrfs_ino(inode), start);
@@ -443,7 +444,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
workspace->in_buf.size = cur_len;
/* Allocate and map in the output buffer */
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
@@ -459,7 +460,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret2 = zstd_compress_stream(stream, &workspace->out_buf,
&workspace->in_buf);
if (unlikely(zstd_is_error(ret2))) {
- btrfs_warn(inode->root->fs_info,
+ btrfs_warn(fs_info,
"zstd compression level %d failed, error %d root %llu inode %llu offset %llu",
workspace->req_level, zstd_get_error_code(ret2),
btrfs_root_id(inode->root), btrfs_ino(inode),
@@ -491,7 +492,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = -E2BIG;
goto out;
}
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
@@ -532,7 +533,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret2 = zstd_end_stream(stream, &workspace->out_buf);
if (unlikely(zstd_is_error(ret2))) {
- btrfs_err(inode->root->fs_info,
+ btrfs_err(fs_info,
"zstd compression end level %d failed, error %d root %llu inode %llu offset %llu",
workspace->req_level, zstd_get_error_code(ret2),
btrfs_root_id(inode->root), btrfs_ino(inode),
@@ -556,7 +557,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
ret = -E2BIG;
goto out;
}
- out_folio = btrfs_alloc_compr_folio();
+ out_folio = btrfs_alloc_compr_folio(fs_info);
if (out_folio == NULL) {
ret = -ENOMEM;
goto out;
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 2/8] btrfs: prepare zstd to support bs > ps cases
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 1/8] btrfs: prepare compression folio alloc/free for bs > ps cases Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 3/8] btrfs: prepare lzo " Qu Wenruo
` (5 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
This involves converting the following functions to use proper folio
sizes/shifts:
- zstd_compress_folios()
- zstd_decompress_bio()
The function zstd_decompress() is already using block size correctly
without using page size, thus it needs no modification.
And since zstd compression is calling kmap_local_folio(), the existing
code can not handle large folios with HIGHMEM, as kmap_local_folio()
requires us to handle one page range each time.
I do not really think it's worthy to spend time on some feature that
will be deprecated eventually.
So here just add an extra explicit rejection for bs > ps with HIGHMEM
feature enabled kernels.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/fs.c | 17 +++++++++++++++++
fs/btrfs/zstd.c | 29 ++++++++++++++++-------------
2 files changed, 33 insertions(+), 13 deletions(-)
diff --git a/fs/btrfs/fs.c b/fs/btrfs/fs.c
index 014fb8b12f96..35084b4e498b 100644
--- a/fs/btrfs/fs.c
+++ b/fs/btrfs/fs.c
@@ -79,6 +79,23 @@ bool __attribute_const__ btrfs_supported_blocksize(u32 blocksize)
if (blocksize == PAGE_SIZE || blocksize == SZ_4K || blocksize == BTRFS_MIN_BLOCKSIZE)
return true;
#ifdef CONFIG_BTRFS_EXPERIMENTAL
+ /*
+ * For bs > ps support it's done by specifying a minimal folio order
+ * for filemap, thus implying large data folios.
+ * For HIGHMEM systems, we can not always access the content of a (large)
+ * folio in one go, but go through them page by page.
+ *
+ * A lot of features doesn't implement a proper PAGE sized loop for large
+ * folios, this includes:
+ * - compression
+ * - verity
+ * - encoded write
+ *
+ * Considering HIGHMEM is such a pain in the backend and it's going
+ * to be deprecated eventually, just reject HIGHMEM && bs > ps cases.
+ */
+ if (IS_ENABLED(CONFIG_HIGHMEM) && blocksize > PAGE_SIZE)
+ return false;
if (blocksize <= PAGE_SIZE)
return true;
#endif
diff --git a/fs/btrfs/zstd.c b/fs/btrfs/zstd.c
index 28e2e99a2463..2f1593ddef4a 100644
--- a/fs/btrfs/zstd.c
+++ b/fs/btrfs/zstd.c
@@ -414,7 +414,8 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
const unsigned long nr_dest_folios = *out_folios;
const u64 orig_end = start + len;
const u32 blocksize = fs_info->sectorsize;
- unsigned long max_out = nr_dest_folios * PAGE_SIZE;
+ const u32 min_folio_size = btrfs_min_folio_size(fs_info);
+ unsigned long max_out = nr_dest_folios * min_folio_size;
unsigned int cur_len;
workspace->params = zstd_get_btrfs_parameters(workspace->req_level, len);
@@ -452,7 +453,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
folios[nr_folios++] = out_folio;
workspace->out_buf.dst = folio_address(out_folio);
workspace->out_buf.pos = 0;
- workspace->out_buf.size = min_t(size_t, max_out, PAGE_SIZE);
+ workspace->out_buf.size = min_t(size_t, max_out, min_folio_size);
while (1) {
size_t ret2;
@@ -486,8 +487,8 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
/* Check if we need more output space */
if (workspace->out_buf.pos == workspace->out_buf.size) {
- tot_out += PAGE_SIZE;
- max_out -= PAGE_SIZE;
+ tot_out += min_folio_size;
+ max_out -= min_folio_size;
if (nr_folios == nr_dest_folios) {
ret = -E2BIG;
goto out;
@@ -501,7 +502,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
workspace->out_buf.dst = folio_address(out_folio);
workspace->out_buf.pos = 0;
workspace->out_buf.size = min_t(size_t, max_out,
- PAGE_SIZE);
+ min_folio_size);
}
/* We've reached the end of the input */
@@ -551,8 +552,8 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
goto out;
}
- tot_out += PAGE_SIZE;
- max_out -= PAGE_SIZE;
+ tot_out += min_folio_size;
+ max_out -= min_folio_size;
if (nr_folios == nr_dest_folios) {
ret = -E2BIG;
goto out;
@@ -565,7 +566,7 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
folios[nr_folios++] = out_folio;
workspace->out_buf.dst = folio_address(out_folio);
workspace->out_buf.pos = 0;
- workspace->out_buf.size = min_t(size_t, max_out, PAGE_SIZE);
+ workspace->out_buf.size = min_t(size_t, max_out, min_folio_size);
}
if (tot_out >= tot_in) {
@@ -587,14 +588,16 @@ int zstd_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
{
+ struct btrfs_fs_info *fs_info = cb_to_fs_info(cb);
struct workspace *workspace = list_entry(ws, struct workspace, list);
struct folio **folios_in = cb->compressed_folios;
size_t srclen = cb->compressed_len;
zstd_dstream *stream;
int ret = 0;
- const u32 blocksize = cb_to_fs_info(cb)->sectorsize;
+ const u32 blocksize = fs_info->sectorsize;
+ const unsigned int min_folio_size = btrfs_min_folio_size(fs_info);
unsigned long folio_in_index = 0;
- unsigned long total_folios_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
+ unsigned long total_folios_in = DIV_ROUND_UP(srclen, min_folio_size);
unsigned long buf_start;
unsigned long total_out = 0;
@@ -612,7 +615,7 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
workspace->in_buf.src = kmap_local_folio(folios_in[folio_in_index], 0);
workspace->in_buf.pos = 0;
- workspace->in_buf.size = min_t(size_t, srclen, PAGE_SIZE);
+ workspace->in_buf.size = min_t(size_t, srclen, min_folio_size);
workspace->out_buf.dst = workspace->buf;
workspace->out_buf.pos = 0;
@@ -657,11 +660,11 @@ int zstd_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
ret = -EIO;
goto done;
}
- srclen -= PAGE_SIZE;
+ srclen -= min_folio_size;
workspace->in_buf.src =
kmap_local_folio(folios_in[folio_in_index], 0);
workspace->in_buf.pos = 0;
- workspace->in_buf.size = min_t(size_t, srclen, PAGE_SIZE);
+ workspace->in_buf.size = min_t(size_t, srclen, min_folio_size);
}
}
ret = 0;
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 3/8] btrfs: prepare lzo to support bs > ps cases
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 1/8] btrfs: prepare compression folio alloc/free for bs > ps cases Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 2/8] btrfs: prepare zstd to support " Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 4/8] btrfs: prepare zlib " Qu Wenruo
` (4 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
This involves converting the following functions to use correct folio
sizes/shifts:
- copy_compress_data_to_page()
- lzo_compress_folios()
- lzo_decompress_bio()
Just like zstd, lzo has some extra incorrect usage of kmap_local_folio()
that the offset is always 0.
This will not handle HIGHMEM large folios correct, but those cases are
already rejected explicitly so it should not cause problems when bs > ps
support is enabled.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/lzo.c | 41 ++++++++++++++++++++++-------------------
1 file changed, 22 insertions(+), 19 deletions(-)
diff --git a/fs/btrfs/lzo.c b/fs/btrfs/lzo.c
index c5a25fd872bd..bc0890f3c2bb 100644
--- a/fs/btrfs/lzo.c
+++ b/fs/btrfs/lzo.c
@@ -140,12 +140,13 @@ static int copy_compressed_data_to_page(struct btrfs_fs_info *fs_info,
u32 *cur_out)
{
const u32 sectorsize = fs_info->sectorsize;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
u32 sector_bytes_left;
u32 orig_out;
struct folio *cur_folio;
char *kaddr;
- if ((*cur_out / PAGE_SIZE) >= max_nr_folio)
+ if ((*cur_out >> min_folio_shift) >= max_nr_folio)
return -E2BIG;
/*
@@ -154,18 +155,17 @@ static int copy_compressed_data_to_page(struct btrfs_fs_info *fs_info,
*/
ASSERT((*cur_out / sectorsize) == (*cur_out + LZO_LEN - 1) / sectorsize);
- cur_folio = out_folios[*cur_out / PAGE_SIZE];
+ cur_folio = out_folios[*cur_out >> min_folio_shift];
/* Allocate a new page */
if (!cur_folio) {
cur_folio = btrfs_alloc_compr_folio(fs_info);
if (!cur_folio)
return -ENOMEM;
- out_folios[*cur_out / PAGE_SIZE] = cur_folio;
+ out_folios[*cur_out >> min_folio_shift] = cur_folio;
}
- kaddr = kmap_local_folio(cur_folio, 0);
- write_compress_length(kaddr + offset_in_page(*cur_out),
- compressed_size);
+ kaddr = kmap_local_folio(cur_folio, offset_in_folio(cur_folio, *cur_out));
+ write_compress_length(kaddr, compressed_size);
*cur_out += LZO_LEN;
orig_out = *cur_out;
@@ -177,20 +177,20 @@ static int copy_compressed_data_to_page(struct btrfs_fs_info *fs_info,
kunmap_local(kaddr);
- if ((*cur_out / PAGE_SIZE) >= max_nr_folio)
+ if ((*cur_out >> min_folio_shift) >= max_nr_folio)
return -E2BIG;
- cur_folio = out_folios[*cur_out / PAGE_SIZE];
+ cur_folio = out_folios[*cur_out >> min_folio_shift];
/* Allocate a new page */
if (!cur_folio) {
cur_folio = btrfs_alloc_compr_folio(fs_info);
if (!cur_folio)
return -ENOMEM;
- out_folios[*cur_out / PAGE_SIZE] = cur_folio;
+ out_folios[*cur_out >> min_folio_shift] = cur_folio;
}
kaddr = kmap_local_folio(cur_folio, 0);
- memcpy(kaddr + offset_in_page(*cur_out),
+ memcpy(kaddr + offset_in_folio(cur_folio, *cur_out),
compressed_data + *cur_out - orig_out, copy_len);
*cur_out += copy_len;
@@ -221,6 +221,7 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct workspace *workspace = list_entry(ws, struct workspace, list);
const u32 sectorsize = fs_info->sectorsize;
+ const u32 min_folio_size = btrfs_min_folio_size(fs_info);
struct address_space *mapping = inode->vfs_inode.i_mapping;
struct folio *folio_in = NULL;
char *sizes_ptr;
@@ -287,8 +288,8 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
goto out;
}
- /* Check if we have reached page boundary */
- if (PAGE_ALIGNED(cur_in)) {
+ /* Check if we have reached folio boundary */
+ if (IS_ALIGNED(cur_in, min_folio_size)) {
folio_put(folio_in);
folio_in = NULL;
}
@@ -305,7 +306,7 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
out:
if (folio_in)
folio_put(folio_in);
- *out_folios = DIV_ROUND_UP(cur_out, PAGE_SIZE);
+ *out_folios = DIV_ROUND_UP(cur_out, min_folio_size);
return ret;
}
@@ -317,15 +318,16 @@ int lzo_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
static void copy_compressed_segment(struct compressed_bio *cb,
char *dest, u32 len, u32 *cur_in)
{
+ struct btrfs_fs_info *fs_info = cb_to_fs_info(cb);
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
u32 orig_in = *cur_in;
while (*cur_in < orig_in + len) {
- struct folio *cur_folio;
- u32 copy_len = min_t(u32, PAGE_SIZE - offset_in_page(*cur_in),
- orig_in + len - *cur_in);
+ struct folio *cur_folio = cb->compressed_folios[*cur_in >> min_folio_shift];
+ u32 copy_len = min_t(u32, orig_in + len - *cur_in,
+ folio_size(cur_folio) - offset_in_folio(cur_folio, *cur_in));
ASSERT(copy_len);
- cur_folio = cb->compressed_folios[*cur_in / PAGE_SIZE];
memcpy_from_folio(dest + *cur_in - orig_in, cur_folio,
offset_in_folio(cur_folio, *cur_in), copy_len);
@@ -339,6 +341,7 @@ int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
struct workspace *workspace = list_entry(ws, struct workspace, list);
const struct btrfs_fs_info *fs_info = cb->bbio.inode->root->fs_info;
const u32 sectorsize = fs_info->sectorsize;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
char *kaddr;
int ret;
/* Compressed data length, can be unaligned */
@@ -385,10 +388,10 @@ int lzo_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
*/
ASSERT(cur_in / sectorsize ==
(cur_in + LZO_LEN - 1) / sectorsize);
- cur_folio = cb->compressed_folios[cur_in / PAGE_SIZE];
+ cur_folio = cb->compressed_folios[cur_in >> min_folio_shift];
ASSERT(cur_folio);
kaddr = kmap_local_folio(cur_folio, 0);
- seg_len = read_compress_length(kaddr + offset_in_page(cur_in));
+ seg_len = read_compress_length(kaddr + offset_in_folio(cur_folio, cur_in));
kunmap_local(kaddr);
cur_in += LZO_LEN;
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 4/8] btrfs: prepare zlib to support bs > ps cases
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
` (2 preceding siblings ...)
2025-09-17 22:46 ` [PATCH RFC 3/8] btrfs: prepare lzo " Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 5/8] btrfs: prepare scrub " Qu Wenruo
` (3 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
This involves converting the following functions to use correct folio
sizes/shifts:
- zlib_compress_folios()
- zlib_decompress_bio()
There is a special handling for s390 hardware acceleration.
With bs > ps cases, we can go with 16K block size on s390 (which uses
fixed 4K page size).
In that case we do not need to do the buffer copy as our folio is large
enough for hardware acceleration.
So extract the s390 specific and folio size check into a helper,
need_special_buffer().
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/zlib.c | 47 ++++++++++++++++++++++++++++++++---------------
1 file changed, 32 insertions(+), 15 deletions(-)
diff --git a/fs/btrfs/zlib.c b/fs/btrfs/zlib.c
index ccf77a0fa96c..889af188a924 100644
--- a/fs/btrfs/zlib.c
+++ b/fs/btrfs/zlib.c
@@ -53,6 +53,22 @@ void zlib_free_workspace(struct list_head *ws)
kfree(workspace);
}
+/*
+ * For s390 hardware acceleration, the buffer size should be at least
+ * ZLIB_DFLTCC_BUF_SIZE to achieve the best performance.
+ *
+ * But if bs > ps we can have large enough folios that meets the s390 hardware
+ * handling.
+ */
+static bool need_special_buffer(struct btrfs_fs_info *fs_info)
+{
+ if (!zlib_deflate_dfltcc_enabled())
+ return false;
+ if (btrfs_min_folio_size(fs_info) >= ZLIB_DFLTCC_BUF_SIZE)
+ return false;
+ return true;
+}
+
struct list_head *zlib_alloc_workspace(struct btrfs_fs_info *fs_info, unsigned int level)
{
const u32 blocksize = fs_info->sectorsize;
@@ -68,11 +84,7 @@ struct list_head *zlib_alloc_workspace(struct btrfs_fs_info *fs_info, unsigned i
workspace->strm.workspace = kvzalloc(workspacesize, GFP_KERNEL | __GFP_NOWARN);
workspace->level = level;
workspace->buf = NULL;
- /*
- * In case of s390 zlib hardware support, allocate lager workspace
- * buffer. If allocator fails, fall back to a single page buffer.
- */
- if (zlib_deflate_dfltcc_enabled()) {
+ if (need_special_buffer(fs_info)) {
workspace->buf = kmalloc(ZLIB_DFLTCC_BUF_SIZE,
__GFP_NOMEMALLOC | __GFP_NORETRY |
__GFP_NOWARN | GFP_NOIO);
@@ -139,6 +151,8 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
struct btrfs_fs_info *fs_info = inode->root->fs_info;
struct workspace *workspace = list_entry(ws, struct workspace, list);
struct address_space *mapping = inode->vfs_inode.i_mapping;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
+ const u32 min_folio_size = btrfs_min_folio_size(fs_info);
int ret;
char *data_in = NULL;
char *cfolio_out;
@@ -147,7 +161,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
struct folio *out_folio = NULL;
unsigned long len = *total_out;
unsigned long nr_dest_folios = *out_folios;
- const unsigned long max_out = nr_dest_folios * PAGE_SIZE;
+ const unsigned long max_out = nr_dest_folios << min_folio_shift;
const u32 blocksize = fs_info->sectorsize;
const u64 orig_end = start + len;
@@ -179,7 +193,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
workspace->strm.next_in = workspace->buf;
workspace->strm.avail_in = 0;
workspace->strm.next_out = cfolio_out;
- workspace->strm.avail_out = PAGE_SIZE;
+ workspace->strm.avail_out = min_folio_size;
while (workspace->strm.total_in < len) {
/*
@@ -191,10 +205,11 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
unsigned int copy_length = min(bytes_left, workspace->buf_size);
/*
- * This can only happen when hardware zlib compression is
- * enabled.
+ * For s390 hardware accelerated zlib, and our folio is smaller
+ * than the copy_length, we need to fill the buffer so that
+ * we can take full advantage of hardware acceleration.
*/
- if (copy_length > PAGE_SIZE) {
+ if (need_special_buffer(fs_info)) {
ret = copy_data_into_buffer(mapping, workspace,
start, copy_length);
if (ret < 0)
@@ -258,7 +273,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
cfolio_out = folio_address(out_folio);
folios[nr_folios] = out_folio;
nr_folios++;
- workspace->strm.avail_out = PAGE_SIZE;
+ workspace->strm.avail_out = min_folio_size;
workspace->strm.next_out = cfolio_out;
}
/* we're all done */
@@ -294,7 +309,7 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
cfolio_out = folio_address(out_folio);
folios[nr_folios] = out_folio;
nr_folios++;
- workspace->strm.avail_out = PAGE_SIZE;
+ workspace->strm.avail_out = min_folio_size;
workspace->strm.next_out = cfolio_out;
}
}
@@ -320,20 +335,22 @@ int zlib_compress_folios(struct list_head *ws, struct btrfs_inode *inode,
int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
{
+ struct btrfs_fs_info *fs_info = cb_to_fs_info(cb);
struct workspace *workspace = list_entry(ws, struct workspace, list);
+ const u32 min_folio_size = btrfs_min_folio_size(fs_info);
int ret = 0, ret2;
int wbits = MAX_WBITS;
char *data_in;
size_t total_out = 0;
unsigned long folio_in_index = 0;
size_t srclen = cb->compressed_len;
- unsigned long total_folios_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
+ unsigned long total_folios_in = DIV_ROUND_UP(srclen, min_folio_size);
unsigned long buf_start;
struct folio **folios_in = cb->compressed_folios;
data_in = kmap_local_folio(folios_in[folio_in_index], 0);
workspace->strm.next_in = data_in;
- workspace->strm.avail_in = min_t(size_t, srclen, PAGE_SIZE);
+ workspace->strm.avail_in = min_t(size_t, srclen, min_folio_size);
workspace->strm.total_in = 0;
workspace->strm.total_out = 0;
@@ -394,7 +411,7 @@ int zlib_decompress_bio(struct list_head *ws, struct compressed_bio *cb)
data_in = kmap_local_folio(folios_in[folio_in_index], 0);
workspace->strm.next_in = data_in;
tmp = srclen - workspace->strm.total_in;
- workspace->strm.avail_in = min(tmp, PAGE_SIZE);
+ workspace->strm.avail_in = min(tmp, min_folio_size);
}
}
if (unlikely(ret != Z_STREAM_END)) {
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 5/8] btrfs: prepare scrub to support bs > ps cases
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
` (3 preceding siblings ...)
2025-09-17 22:46 ` [PATCH RFC 4/8] btrfs: prepare zlib " Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 6/8] btrfs: fix symbolic link reading when bs > ps Qu Wenruo
` (2 subsequent siblings)
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
This involves:
- Migrate scrub_stripe::pages[] to folios[]
- Use btrfs_alloc_folio_array() and folio_put() to alloc above array.
- Migrate scrub_stripe_get_kaddr() and scrub_stripe_get_paddr() to use
folio interfaces
- Migrate raid56_parity_cache_data_pages() to
raid56_parity_cache_data_folios()
Since scrub is the only caller still using pages.
This helper will copy the folio array contents into rbio::stripe_pages,
with sector uptodate flags updated.
And a new ASSERT() to make sure bs > ps cases will not hit this path.
Since most scrub code is based on kaddr/paddr, the migration itself is
pretty straightforward.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/raid56.c | 43 ++++++++++++++++++++++++++++-----------
fs/btrfs/raid56.h | 4 ++--
fs/btrfs/scrub.c | 51 +++++++++++++++++++++++++++--------------------
3 files changed, 62 insertions(+), 36 deletions(-)
diff --git a/fs/btrfs/raid56.c b/fs/btrfs/raid56.c
index 2b4f577dcf39..c1aa4b3edcec 100644
--- a/fs/btrfs/raid56.c
+++ b/fs/btrfs/raid56.c
@@ -2844,19 +2844,24 @@ void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio)
* This is for scrub call sites where we already have correct data contents.
* This allows us to avoid reading data stripes again.
*
- * Unfortunately here we have to do page copy, other than reusing the pages.
+ * Unfortunately here we have to do folio copy, other than reusing the pages.
* This is due to the fact rbio has its own page management for its cache.
*/
-void raid56_parity_cache_data_pages(struct btrfs_raid_bio *rbio,
- struct page **data_pages, u64 data_logical)
+void raid56_parity_cache_data_folios(struct btrfs_raid_bio *rbio,
+ struct folio **data_folios, u64 data_logical)
{
+ struct btrfs_fs_info *fs_info = rbio->bioc->fs_info;
const u64 offset_in_full_stripe = data_logical -
rbio->bioc->full_stripe_logical;
- const int page_index = offset_in_full_stripe >> PAGE_SHIFT;
- const u32 sectorsize = rbio->bioc->fs_info->sectorsize;
- const u32 sectors_per_page = PAGE_SIZE / sectorsize;
+ const u32 sectors_per_page = PAGE_SIZE >> fs_info->sectorsize_bits;
+ u32 cur_off = offset_in_full_stripe;
+ unsigned int findex = 0;
+ unsigned int foffset = 0;
int ret;
+ /* We shouldn't hit RAID56 for bs > ps cases for now. */
+ ASSERT(sectors_per_page > 0);
+
/*
* If we hit ENOMEM temporarily, but later at
* raid56_parity_submit_scrub_rbio() time it succeeded, we just do
@@ -2873,14 +2878,28 @@ void raid56_parity_cache_data_pages(struct btrfs_raid_bio *rbio,
ASSERT(IS_ALIGNED(offset_in_full_stripe, BTRFS_STRIPE_LEN));
ASSERT(offset_in_full_stripe < (rbio->nr_data << BTRFS_STRIPE_LEN_SHIFT));
- for (int page_nr = 0; page_nr < (BTRFS_STRIPE_LEN >> PAGE_SHIFT); page_nr++) {
- struct page *dst = rbio->stripe_pages[page_nr + page_index];
- struct page *src = data_pages[page_nr];
+ while (cur_off <= offset_in_full_stripe + BTRFS_STRIPE_LEN) {
+ const unsigned int pindex = cur_off >> PAGE_SHIFT;
+ unsigned int copy_len = min(PAGE_SIZE, folio_size(data_folios[findex]));
+ void *kaddr;
- memcpy_page(dst, 0, src, 0, PAGE_SIZE);
- for (int sector_nr = sectors_per_page * page_index;
- sector_nr < sectors_per_page * (page_index + 1);
+ ASSERT(IS_ALIGNED(cur_off, PAGE_SIZE));
+
+ kaddr = kmap_local_page(rbio->stripe_pages[pindex]);
+ memcpy_from_folio(kaddr, data_folios[findex], foffset, copy_len);
+ kunmap_local(kaddr);
+
+ for (int sector_nr = sectors_per_page * pindex;
+ sector_nr < sectors_per_page * (pindex + 1);
sector_nr++)
rbio->stripe_sectors[sector_nr].uptodate = true;
+
+ cur_off += copy_len;
+ foffset += copy_len;
+ ASSERT(foffset <= folio_size(data_folios[findex]));
+ if (foffset == folio_size(data_folios[findex])) {
+ findex++;
+ foffset = 0;
+ }
}
}
diff --git a/fs/btrfs/raid56.h b/fs/btrfs/raid56.h
index 0d7b4c2fb6ae..84c4d1d29c7a 100644
--- a/fs/btrfs/raid56.h
+++ b/fs/btrfs/raid56.h
@@ -201,8 +201,8 @@ struct btrfs_raid_bio *raid56_parity_alloc_scrub_rbio(struct bio *bio,
unsigned long *dbitmap, int stripe_nsectors);
void raid56_parity_submit_scrub_rbio(struct btrfs_raid_bio *rbio);
-void raid56_parity_cache_data_pages(struct btrfs_raid_bio *rbio,
- struct page **data_pages, u64 data_logical);
+void raid56_parity_cache_data_folios(struct btrfs_raid_bio *rbio,
+ struct folio **data_folios, u64 data_logical);
int btrfs_alloc_stripe_hash_table(struct btrfs_fs_info *info);
void btrfs_free_stripe_hash_table(struct btrfs_fs_info *info);
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index 979d33d8c193..bf50556f186e 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -130,7 +130,7 @@ enum {
scrub_bitmap_nr_last,
};
-#define SCRUB_STRIPE_PAGES (BTRFS_STRIPE_LEN / PAGE_SIZE)
+#define SCRUB_STRIPE_MAX_FOLIOS (BTRFS_STRIPE_LEN / PAGE_SIZE)
/*
* Represent one contiguous range with a length of BTRFS_STRIPE_LEN.
@@ -139,7 +139,7 @@ struct scrub_stripe {
struct scrub_ctx *sctx;
struct btrfs_block_group *bg;
- struct page *pages[SCRUB_STRIPE_PAGES];
+ struct folio *folios[SCRUB_STRIPE_MAX_FOLIOS];
struct scrub_sector_verification *sectors;
struct btrfs_device *dev;
@@ -339,10 +339,10 @@ static void release_scrub_stripe(struct scrub_stripe *stripe)
if (!stripe)
return;
- for (int i = 0; i < SCRUB_STRIPE_PAGES; i++) {
- if (stripe->pages[i])
- __free_page(stripe->pages[i]);
- stripe->pages[i] = NULL;
+ for (int i = 0; i < SCRUB_STRIPE_MAX_FOLIOS; i++) {
+ if (stripe->folios[i])
+ folio_put(stripe->folios[i]);
+ stripe->folios[i] = NULL;
}
kfree(stripe->sectors);
kfree(stripe->csums);
@@ -355,6 +355,7 @@ static void release_scrub_stripe(struct scrub_stripe *stripe)
static int init_scrub_stripe(struct btrfs_fs_info *fs_info,
struct scrub_stripe *stripe)
{
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
int ret;
memset(stripe, 0, sizeof(*stripe));
@@ -367,7 +368,9 @@ static int init_scrub_stripe(struct btrfs_fs_info *fs_info,
atomic_set(&stripe->pending_io, 0);
spin_lock_init(&stripe->write_error_lock);
- ret = btrfs_alloc_page_array(SCRUB_STRIPE_PAGES, stripe->pages, false);
+ ASSERT(BTRFS_STRIPE_LEN >> min_folio_shift <= SCRUB_STRIPE_MAX_FOLIOS);
+ ret = btrfs_alloc_folio_array(BTRFS_STRIPE_LEN >> min_folio_shift,
+ fs_info->block_min_order, stripe->folios);
if (ret < 0)
goto error;
@@ -687,27 +690,30 @@ static int fill_writer_pointer_gap(struct scrub_ctx *sctx, u64 physical)
static void *scrub_stripe_get_kaddr(struct scrub_stripe *stripe, int sector_nr)
{
- u32 offset = (sector_nr << stripe->bg->fs_info->sectorsize_bits);
- const struct page *page = stripe->pages[offset >> PAGE_SHIFT];
+ struct btrfs_fs_info *fs_info = stripe->bg->fs_info;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
+ u32 offset = (sector_nr << fs_info->sectorsize_bits);
+ const struct folio *folio = stripe->folios[offset >> min_folio_shift];
- /* stripe->pages[] is allocated by us and no highmem is allowed. */
- ASSERT(page);
- ASSERT(!PageHighMem(page));
- return page_address(page) + offset_in_page(offset);
+ /* stripe->folios[] is allocated by us and no highmem is allowed. */
+ ASSERT(folio);
+ ASSERT(!folio_test_partial_kmap(folio));
+ return folio_address(folio) + offset_in_folio(folio, offset);
}
static phys_addr_t scrub_stripe_get_paddr(struct scrub_stripe *stripe, int sector_nr)
{
struct btrfs_fs_info *fs_info = stripe->bg->fs_info;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
u32 offset = (sector_nr << fs_info->sectorsize_bits);
- const struct page *page = stripe->pages[offset >> PAGE_SHIFT];
+ const struct folio *folio = stripe->folios[offset >> min_folio_shift];
- /* stripe->pages[] is allocated by us and no highmem is allowed. */
- ASSERT(page);
- ASSERT(!PageHighMem(page));
- /* And the range must be contained inside the page. */
- ASSERT(offset_in_page(offset) + fs_info->sectorsize <= PAGE_SIZE);
- return page_to_phys(page) + offset_in_page(offset);
+ /* stripe->folios[] is allocated by us and no highmem is allowed. */
+ ASSERT(folio);
+ ASSERT(!folio_test_partial_kmap(folio));
+ /* And the range must be contained inside the folio. */
+ ASSERT(offset_in_folio(folio, offset) + fs_info->sectorsize <= folio_size(folio));
+ return page_to_phys(folio_page(folio, 0)) + offset_in_folio(folio, offset);
}
static void scrub_verify_one_metadata(struct scrub_stripe *stripe, int sector_nr)
@@ -1872,6 +1878,7 @@ static void scrub_submit_initial_read(struct scrub_ctx *sctx,
{
struct btrfs_fs_info *fs_info = sctx->fs_info;
struct btrfs_bio *bbio;
+ const u32 min_folio_shift = PAGE_SHIFT + fs_info->block_min_order;
unsigned int nr_sectors = stripe_length(stripe) >> fs_info->sectorsize_bits;
int mirror = stripe->mirror_num;
@@ -1884,7 +1891,7 @@ static void scrub_submit_initial_read(struct scrub_ctx *sctx,
return;
}
- bbio = btrfs_bio_alloc(SCRUB_STRIPE_PAGES, REQ_OP_READ, fs_info,
+ bbio = btrfs_bio_alloc(BTRFS_STRIPE_LEN >> min_folio_shift, REQ_OP_READ, fs_info,
scrub_read_endio, stripe);
bbio->bio.bi_iter.bi_sector = stripe->logical >> SECTOR_SHIFT;
@@ -2215,7 +2222,7 @@ static int scrub_raid56_parity_stripe(struct scrub_ctx *sctx,
for (int i = 0; i < data_stripes; i++) {
stripe = &sctx->raid56_data_stripes[i];
- raid56_parity_cache_data_pages(rbio, stripe->pages,
+ raid56_parity_cache_data_folios(rbio, stripe->folios,
full_stripe_start + (i << BTRFS_STRIPE_LEN_SHIFT));
}
raid56_parity_submit_scrub_rbio(rbio);
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 6/8] btrfs: fix symbolic link reading when bs > ps
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
` (4 preceding siblings ...)
2025-09-17 22:46 ` [PATCH RFC 5/8] btrfs: prepare scrub " Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 7/8] btrfs: add extra ASSERT()s to catch unaligned bios Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 8/8] btrfs: enable experimental bs > ps support Qu Wenruo
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
[BUG DURING BS > PS TEST]
When running the following script on a btrfs whose block size is larger
than page size, e.g. 8K block size and 4K page size, it will trigger a
kernel BUG:
# mkfs.btrfs -s 8k $dev
# mount $dev $mnt
# mkdir $mnt/dir
# ln -s dir $mnt/link
# ls $mnt/link
The call trace looks like this:
BTRFS warning (device dm-2): support for block size 8192 with page size 4096 is experimental, some features may be missing
BTRFS info (device dm-2): checking UUID tree
BTRFS info (device dm-2): enabling ssd optimizations
BTRFS info (device dm-2): enabling free space tree
NOTICE: Automounting of tracing to debugfs is deprecated and will be removed in 2030
------------[ cut here ]------------
kernel BUG at /home/adam/linux/include/linux/highmem.h:275!
Oops: invalid opcode: 0000 [#1] SMP
CPU: 8 UID: 0 PID: 667 Comm: ls Tainted: G OE 6.17.0-rc4-custom+ #283 PREEMPT(full)
Tainted: [O]=OOT_MODULE, [E]=UNSIGNED_MODULE
Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS unknown 02/02/2022
RIP: 0010:zero_user_segments.constprop.0+0xdc/0xe0 [btrfs]
Call Trace:
<TASK>
btrfs_get_extent.cold+0x85/0x101 [btrfs 7453c70c03e631c8d8bfdd4264fa62d3e238da6f]
btrfs_do_readpage+0x244/0x750 [btrfs 7453c70c03e631c8d8bfdd4264fa62d3e238da6f]
btrfs_read_folio+0x9c/0x100 [btrfs 7453c70c03e631c8d8bfdd4264fa62d3e238da6f]
filemap_read_folio+0x37/0xe0
do_read_cache_folio+0x94/0x3e0
__page_get_link.isra.0+0x20/0x90
page_get_link+0x16/0x40
step_into+0x69b/0x830
path_lookupat+0xa7/0x170
filename_lookup+0xf7/0x200
? set_ptes.isra.0+0x36/0x70
vfs_statx+0x7a/0x160
do_statx+0x63/0xa0
__x64_sys_statx+0x90/0xe0
do_syscall_64+0x82/0xae0
entry_SYSCALL_64_after_hwframe+0x4b/0x53
</TASK>
Please note bs > ps support is still under development and the
enablement patch is not even in btrfs development branch.
[CAUSE]
Btrfs reuses its data folio read path to handle symbolic links, as the
symbolic link target is stored as an inline data extent.
But for newly created inodes, btrfs only set the minimal order if the
target inode is a regular file.
Thus for above newly created symbolic link, it doesn't properly respect
the minimal folio order, and triggered the above crash.
[FIX]
Call btrfs_set_inode_mapping_order() unconditionally inside
btrfs_create_new_inode().
For symbolic links this will fix the crash as now the folio will meet
the minimal order.
For regular files this brings no change.
For directory/bdev/char and all the other types of inodes, they won't
go through the data read path, thus no effect either.
Fixes: cc38d178ff33 ("btrfs: enable large data folio support under CONFIG_BTRFS_EXPERIMENTAL")
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/inode.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 6b52ab164f45..601a20396171 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6506,6 +6506,7 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
if (!args->subvol)
btrfs_inherit_iflags(BTRFS_I(inode), BTRFS_I(dir));
+ btrfs_set_inode_mapping_order(BTRFS_I(inode));
if (S_ISREG(inode->i_mode)) {
if (btrfs_test_opt(fs_info, NODATASUM))
BTRFS_I(inode)->flags |= BTRFS_INODE_NODATASUM;
@@ -6513,7 +6514,6 @@ int btrfs_create_new_inode(struct btrfs_trans_handle *trans,
BTRFS_I(inode)->flags |= BTRFS_INODE_NODATACOW |
BTRFS_INODE_NODATASUM;
btrfs_update_inode_mapping_flags(BTRFS_I(inode));
- btrfs_set_inode_mapping_order(BTRFS_I(inode));
}
ret = btrfs_insert_inode_locked(inode);
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 7/8] btrfs: add extra ASSERT()s to catch unaligned bios
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
` (5 preceding siblings ...)
2025-09-17 22:46 ` [PATCH RFC 6/8] btrfs: fix symbolic link reading when bs > ps Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
2025-09-17 22:46 ` [PATCH RFC 8/8] btrfs: enable experimental bs > ps support Qu Wenruo
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
Btrfs uses btrfs_bio to handle read/write of logical address, for the
incoming bs > ps support, btrfs has extra requirements:
- One folio must contain at least one fs block
- No fs block can cross folio boundaries
This requirement is not hard to maintain, thanks to the address space's
minimal folio order.
But not all btrfs bios are generated through address space, e.g.
compression and scrub.
To catch possible unaligned bios, introduce a helper,
assert_bbio_alginment(), for each btrfs_bio in btrfs_submit_bbio().
This will check the following things:
- bv_offset is aligned to block size
- bv_len is aligned to block size
With a btrfs bio passing above checks, unless it's empty it will ensure
the requirements for bs > ps support.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/bio.c | 27 +++++++++++++++++++++++++++
1 file changed, 27 insertions(+)
diff --git a/fs/btrfs/bio.c b/fs/btrfs/bio.c
index 909b208f9ef3..db2deaa4aad4 100644
--- a/fs/btrfs/bio.c
+++ b/fs/btrfs/bio.c
@@ -779,11 +779,38 @@ static bool btrfs_submit_chunk(struct btrfs_bio *bbio, int mirror_num)
return true;
}
+static void assert_bbio_alignment(struct btrfs_bio *bbio)
+{
+#ifdef CONFIG_BTRFS_ASSERT
+ struct btrfs_fs_info *fs_info = bbio->fs_info;
+ struct bio_vec bvec;
+ struct bvec_iter iter;
+ const u32 blocksize = fs_info->sectorsize;
+
+ /* Metadata has no extra bs > ps alignment requirement. */
+ if (!is_data_bbio(bbio))
+ return;
+
+ bio_for_each_bvec(bvec, &bbio->bio, iter)
+ ASSERT(IS_ALIGNED(bvec.bv_offset, blocksize) &&
+ IS_ALIGNED(bvec.bv_len, blocksize),
+ "root=%llu inode=%llu logical=%llu length=%u index=%u bv_offset=%u bv_len=%u",
+ btrfs_root_id(bbio->inode->root),
+ btrfs_ino(bbio->inode),
+ bbio->bio.bi_iter.bi_sector << SECTOR_SHIFT,
+ bbio->bio.bi_iter.bi_size, iter.bi_idx,
+ bvec.bv_offset,
+ bvec.bv_len);
+#endif
+}
+
void btrfs_submit_bbio(struct btrfs_bio *bbio, int mirror_num)
{
/* If bbio->inode is not populated, its file_offset must be 0. */
ASSERT(bbio->inode || bbio->file_offset == 0);
+ assert_bbio_alignment(bbio);
+
while (!btrfs_submit_chunk(bbio, mirror_num))
;
}
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread* [PATCH RFC 8/8] btrfs: enable experimental bs > ps support
2025-09-17 22:46 [PATCH RFC 0/9] btrfs: initial bs > ps support Qu Wenruo
` (6 preceding siblings ...)
2025-09-17 22:46 ` [PATCH RFC 7/8] btrfs: add extra ASSERT()s to catch unaligned bios Qu Wenruo
@ 2025-09-17 22:46 ` Qu Wenruo
7 siblings, 0 replies; 9+ messages in thread
From: Qu Wenruo @ 2025-09-17 22:46 UTC (permalink / raw)
To: linux-btrfs
With all the preparation patches, we're able to finally enable btrfs
block size (sector size) larger than page size support and give it a
full fstests run.
And obvious this new feature is hidden behind experimental flags, and
should not be considered as a core feature yet as btrfs' default block
size is still 4K.
But this is still a feature that will shine in the future where 16K
block sized device are widely adapted.
For now there are some features explicitly disabled:
- Direct IO
This is the most complex part to support, the root reason is we can
not control the pages of iov iter passed in.
User space programs can only ensure the virtual addresses are
contiguous, but no control on their physical addresses.
Our bs > ps support heavily relies on large folios, and direct IO
memory can easily break it.
So direct IO is disabled and will always fall back to buffered IO.
- RAID56
In theory we can convert RAID56 to use large folios, but it will need
to be converted back to page based if we want to support direct IO in
the future.
So just reject it for now.
- Encoded send
- Encoded read
Both are utilizing btrfs_encoded_read_regular_fill_pages(), and send
is utilizing vmallocated memory.
Unfortunately for vmallocated memory we can not guarantee the minimal
folio order.
For send, it will just always fallback to regular writes, which reads
from page cache and will follow the existing folio order requirement.
- Encoded write
Encoded write itself is allocating pages by themselves, and we can
easily change it to follow the minimal order.
But since encoded read is already disabled, there is no need to only
enable encoded write.
Finally just like what we did for bs < ps support in the past, add a
warning message for bs > ps mounts.
Signed-off-by: Qu Wenruo <wqu@suse.com>
---
fs/btrfs/direct-io.c | 12 ++++++++++++
fs/btrfs/disk-io.c | 14 ++++++++++++--
fs/btrfs/fs.c | 3 +--
fs/btrfs/ioctl.c | 35 +++++++++++++++++++++++++----------
fs/btrfs/send.c | 9 ++++++++-
5 files changed, 58 insertions(+), 15 deletions(-)
diff --git a/fs/btrfs/direct-io.c b/fs/btrfs/direct-io.c
index fe9a4bd7e6e6..abe0d5c42baa 100644
--- a/fs/btrfs/direct-io.c
+++ b/fs/btrfs/direct-io.c
@@ -786,6 +786,18 @@ static ssize_t check_direct_IO(struct btrfs_fs_info *fs_info,
if (iov_iter_alignment(iter) & blocksize_mask)
return -EINVAL;
+ /*
+ * For bs > ps support, btrfs heavily relies on large folios to make
+ * sure no block will cross large folio boundaries.
+ *
+ * But memory provided by direct IO is only virtually contiguous, not
+ * physically contiguous, and will break the btrfs' large folio requirement.
+ *
+ * So for bs > ps support, all direct IOs should fallback to buffered ones.
+ */
+ if (fs_info->sectorsize > PAGE_SIZE)
+ return -EINVAL;
+
return 0;
}
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index a2eba8bc4336..d3dce247f3d4 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3243,18 +3243,24 @@ int btrfs_check_features(struct btrfs_fs_info *fs_info, bool is_rw_mount)
}
/*
- * Subpage runtime limitation on v1 cache.
+ * Subpage/bs > ps runtime limitation on v1 cache.
*
* V1 space cache still has some hard coded PAGE_SIZE usage, while
* we're already defaulting to v2 cache, no need to bother v1 as it's
* going to be deprecated anyway.
*/
- if (fs_info->sectorsize < PAGE_SIZE && btrfs_test_opt(fs_info, SPACE_CACHE)) {
+ if (fs_info->sectorsize != PAGE_SIZE && btrfs_test_opt(fs_info, SPACE_CACHE)) {
btrfs_warn(fs_info,
"v1 space cache is not supported for page size %lu with sectorsize %u",
PAGE_SIZE, fs_info->sectorsize);
return -EINVAL;
}
+ if (fs_info->sectorsize > PAGE_SIZE && btrfs_fs_incompat(fs_info, RAID56)) {
+ btrfs_err(fs_info,
+ "RAID56 is not supported for page size %lu with sectorsize %u",
+ PAGE_SIZE, fs_info->sectorsize);
+ return -EINVAL;
+ }
/* This can be called by remount, we need to protect the super block. */
spin_lock(&fs_info->super_lock);
@@ -3389,6 +3395,10 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
fs_info->stripesize = stripesize;
fs_info->fs_devices->fs_info = fs_info;
+ if (fs_info->sectorsize > PAGE_SIZE)
+ btrfs_warn(fs_info,
+ "support for block size %u with page size %zu is experimental, some features may be missing",
+ fs_info->sectorsize, PAGE_SIZE);
/*
* Handle the space caching options appropriately now that we have the
* super block loaded and validated.
diff --git a/fs/btrfs/fs.c b/fs/btrfs/fs.c
index 35084b4e498b..3b4040e9c45d 100644
--- a/fs/btrfs/fs.c
+++ b/fs/btrfs/fs.c
@@ -96,8 +96,7 @@ bool __attribute_const__ btrfs_supported_blocksize(u32 blocksize)
*/
if (IS_ENABLED(CONFIG_HIGHMEM) && blocksize > PAGE_SIZE)
return false;
- if (blocksize <= PAGE_SIZE)
- return true;
+ return true;
#endif
return false;
}
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 063291519b36..0e9e2b999392 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -4418,6 +4418,10 @@ static int btrfs_ioctl_encoded_read(struct file *file, void __user *argp,
goto out_acct;
}
+ if (fs_info->sectorsize > PAGE_SIZE) {
+ ret = -ENOTTY;
+ goto out_acct;
+ }
if (compat) {
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
struct btrfs_ioctl_encoded_io_args_32 args32;
@@ -4509,6 +4513,7 @@ static int btrfs_ioctl_encoded_read(struct file *file, void __user *argp,
static int btrfs_ioctl_encoded_write(struct file *file, void __user *argp, bool compat)
{
+ struct btrfs_fs_info *fs_info = inode_to_fs_info(file->f_inode);
struct btrfs_ioctl_encoded_io_args args;
struct iovec iovstack[UIO_FASTIOV];
struct iovec *iov = iovstack;
@@ -4522,6 +4527,11 @@ static int btrfs_ioctl_encoded_write(struct file *file, void __user *argp, bool
goto out_acct;
}
+ if (fs_info->sectorsize > PAGE_SIZE) {
+ ret = -ENOTTY;
+ goto out_acct;
+ }
+
if (!(file->f_mode & FMODE_WRITE)) {
ret = -EBADF;
goto out_acct;
@@ -4780,14 +4790,14 @@ static int btrfs_uring_read_extent(struct kiocb *iocb, struct iov_iter *iter,
static int btrfs_uring_encoded_read(struct io_uring_cmd *cmd, unsigned int issue_flags)
{
+ struct file *file = cmd->file;
+ struct btrfs_inode *inode = BTRFS_I(file->f_inode);
+ struct extent_io_tree *io_tree = &inode->io_tree;
+ struct btrfs_fs_info *fs_info = inode->root->fs_info;
size_t copy_end_kernel = offsetofend(struct btrfs_ioctl_encoded_io_args, flags);
size_t copy_end;
int ret;
u64 disk_bytenr, disk_io_size;
- struct file *file;
- struct btrfs_inode *inode;
- struct btrfs_fs_info *fs_info;
- struct extent_io_tree *io_tree;
loff_t pos;
struct kiocb kiocb;
struct extent_state *cached_state = NULL;
@@ -4803,10 +4813,11 @@ static int btrfs_uring_encoded_read(struct io_uring_cmd *cmd, unsigned int issue
ret = -EPERM;
goto out_acct;
}
- file = cmd->file;
- inode = BTRFS_I(file->f_inode);
- fs_info = inode->root->fs_info;
- io_tree = &inode->io_tree;
+ if (fs_info->sectorsize > PAGE_SIZE) {
+ ret = -ENOTTY;
+ goto out_acct;
+ }
+
sqe_addr = u64_to_user_ptr(READ_ONCE(cmd->sqe->addr));
if (issue_flags & IO_URING_F_COMPAT) {
@@ -4933,9 +4944,10 @@ static int btrfs_uring_encoded_read(struct io_uring_cmd *cmd, unsigned int issue
static int btrfs_uring_encoded_write(struct io_uring_cmd *cmd, unsigned int issue_flags)
{
+ struct file *file = cmd->file;
+ struct btrfs_fs_info *fs_info = inode_to_fs_info(file->f_inode);
loff_t pos;
struct kiocb kiocb;
- struct file *file;
ssize_t ret;
void __user *sqe_addr;
struct io_btrfs_cmd *bc = io_uring_cmd_to_pdu(cmd, struct io_btrfs_cmd);
@@ -4948,8 +4960,11 @@ static int btrfs_uring_encoded_write(struct io_uring_cmd *cmd, unsigned int issu
ret = -EPERM;
goto out_acct;
}
+ if (fs_info->sectorsize > PAGE_SIZE) {
+ ret = -ENOTTY;
+ goto out_acct;
+ }
- file = cmd->file;
sqe_addr = u64_to_user_ptr(READ_ONCE(cmd->sqe->addr));
if (!(file->f_mode & FMODE_WRITE)) {
diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index c5771df3a2c7..cf561a1ae2e8 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -5654,7 +5654,14 @@ static int send_extent_data(struct send_ctx *sctx, struct btrfs_path *path,
ei = btrfs_item_ptr(leaf, path->slots[0],
struct btrfs_file_extent_item);
- if ((sctx->flags & BTRFS_SEND_FLAG_COMPRESSED) &&
+ /*
+ * Do not go through encoded read for bs > ps cases.
+ *
+ * Encoded send is using vmallocated pages as buffer, which we can
+ * not ensure every folio is large enough to contain a block.
+ */
+ if (sctx->send_root->fs_info->sectorsize <= PAGE_SIZE &&
+ (sctx->flags & BTRFS_SEND_FLAG_COMPRESSED) &&
btrfs_file_extent_compression(leaf, ei) != BTRFS_COMPRESS_NONE) {
bool is_inline = (btrfs_file_extent_type(leaf, ei) ==
BTRFS_FILE_EXTENT_INLINE);
--
2.50.1
^ permalink raw reply related [flat|nested] 9+ messages in thread