btrfs: Replace kmap_atomic() with kmap_local_page()
PATCH 1/1 btrfs: Replace kmap_atomic() with kmap_local_page()
kmap_atomic() is being deprecated in favor of kmap_local_page() where it is feasible. With kmap_local_page() mappings are per thread, CPU local, and not globally visible.
As far as I can see, the kmap_atomic() calls in compression.c and in inode.c can be safely converted.
Above all else, David Sterba has confirmed that "The context in check_compressed_csum is atomic [...]" and that "kmap_atomic() in inode.c [...] also can be replaced by kmap_local_page().".[1]
Therefore, convert all kmap_atomic() calls currently still left in fs/btrfs to kmap_local_page().
Tested with xfstests on a QEMU + KVM 32-bits VM with 4GB RAM and booting a kernel with HIGHMEM64GB enabled.
[1] https://lore.kernel.org/linux-btrfs/20220601132545.GM20
[email protected]/
---
Tests of groups "quick" and "compress" output several errors largely due to memory leaks and shift-out-of-bounds. However, these errors are exactly the same which are output without this and other conversions of mine to use kmap_local_page(). Therefore, it looks like these changes don't introduce regressions.
The previous RFC PATCH can be ignored:
https://lore.kernel.org/lkml/[email protected]/
With this patch, in fs/btrfs there are no longer call sites of kmap() and kmap_atomic().
fs/btrfs/compression.c | 4 ++-- fs/btrfs/inode.c | 12 ++++++------ 2 files changed, 8 insertions(+), 8 deletions(-)
fs/btrfs/compression.c +2 -2
@@ -175,10 +175,10 @@ static int check_compressed_csum(struct btrfs_inode *inode, struct bio *bio,/* Hash through the page sector by sector */ for (pg_offset =0 ; pg_offset < bytes_left; pg_offset += sectorsize) { - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page);crypto_shash_digest (shash, kaddr + pg_offset, sectorsize, csum); -kunmap_atomic (kaddr); +kunmap_local (kaddr);if (memcmp (&csum, cb_sum, csum_size) !=0 ) { btrfs_print_data_csum_error(inode, disk_start,
fs/btrfs/inode.c +6 -6
@@ -332,9 +332,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, cur_size = min_t(unsigned long, compressed_size,PAGE_SIZE ); - kaddr =kmap_atomic (cpage); + kaddr =kmap_local_page (cpage);write_extent_buffer (leaf, kaddr, ptr, cur_size); -kunmap_atomic (kaddr); +kunmap_local (kaddr); i++; ptr += cur_size;
@@ -345,9 +345,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, }else { page =find_get_page (inode->vfs_inode.i_mapping,0 );btrfs_set_file_extent_compression (leaf, ei,0 ); - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page);write_extent_buffer (leaf, kaddr, ptr, size); -kunmap_atomic (kaddr); +kunmap_local (kaddr);put_page (page); }btrfs_mark_buffer_dirty (leaf);
@@ -3357,11 +3357,11 @@ static int check_data_csum(struct inode *inode, struct btrfs_bio *bbio, offset_sectors = bio_offset >> fs_info->sectorsize_bits; csum_expected = ((u8 *)bbio->csum) + offset_sectors * csum_size; - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page); shash->tfm = fs_info->csum_shash;crypto_shash_digest (shash, kaddr + pgoff, len, csum); -kunmap_atomic (kaddr); +kunmap_local (kaddr);if (memcmp (csum, csum_expected, csum_size)) goto zeroit;
Signature · 1 line
2.36.1
Quoted 14 lines · Fabio M. De Francesco
kmap_atomic()is being deprecated in favor ofkmap_local_page()where it is feasible. Withkmap_local_page()mappings are per thread, CPU local, and not globally visible.As far as I can see, the
kmap_atomic()calls incompression.cand ininode.ccan be safely converted.Above all else, David Sterba has confirmed that "The context in
check_compressed_csumis atomic [...]" and that "kmap_atomic()ininode.c[...] also can be replaced by kmap_local_page().".[1]Therefore, convert all
kmap_atomic()calls currently still left in fs/btrfs tokmap_local_page().
Quoted 7 lines
Tested with xfstests on a QEMU + KVM 32-bits VM with 4GB RAM and booting a kernel with HIGHMEM64GB enabled.
[1] https://lore.kernel.org/linux-btrfs/20220601132545.GM20
[email protected]/Suggested-by Ira Weiny <[email protected]>
Test failures aside this looks ok to me.
Quoted 82 lines
Suggested-by David Sterba <[email protected]>Signed-off-by Fabio M. De Francesco <[email protected]>---
Tests of groups "quick" and "compress" output several errors largely due to memory leaks and shift-out-of-bounds. However, these errors are exactly the same which are output without this and other conversions of mine to use
kmap_local_page(). Therefore, it looks like these changes don't introduce regressions.The previous RFC PATCH can be ignored:
https://lore.kernel.org/lkml/[email protected]/With this patch, in fs/btrfs there are no longer call sites of
kmap()andkmap_atomic().fs/btrfs/compression.c | 4 ++-- fs/btrfs/inode.c | 12 ++++++------ 2 files changed, 8 insertions(+), 8 deletions(-)fs/btrfs/compression.c +2
-2@@ -175,10 +175,10 @@ static int check_compressed_csum(struct btrfs_inode *inode, struct bio *bio,/* Hash through the page sector by sector */ for (pg_offset =0 ; pg_offset < bytes_left; pg_offset += sectorsize) { - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page);crypto_shash_digest (shash, kaddr + pg_offset, sectorsize, csum); -kunmap_atomic (kaddr); +kunmap_local (kaddr);if (memcmp (&csum, cb_sum, csum_size) !=0 ) { btrfs_print_data_csum_error(inode, disk_start,fs/btrfs/inode.c +6
-6@@ -332,9 +332,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, cur_size = min_t(unsigned long, compressed_size,PAGE_SIZE ); - kaddr =kmap_atomic (cpage); + kaddr =kmap_local_page (cpage);write_extent_buffer (leaf, kaddr, ptr, cur_size); -kunmap_atomic (kaddr); +kunmap_local (kaddr); i++; ptr += cur_size;@@ -345,9 +345,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, }else { page =find_get_page (inode->vfs_inode.i_mapping,0 );btrfs_set_file_extent_compression (leaf, ei,0 ); - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page);write_extent_buffer (leaf, kaddr, ptr, size); -kunmap_atomic (kaddr); +kunmap_local (kaddr);put_page (page); }btrfs_mark_buffer_dirty (leaf);@@ -3357,11 +3357,11 @@ static int check_data_csum(struct inode *inode, struct btrfs_bio *bbio, offset_sectors = bio_offset >> fs_info->sectorsize_bits; csum_expected = ((u8 *)bbio->csum) + offset_sectors * csum_size; - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page); shash->tfm = fs_info->csum_shash;crypto_shash_digest (shash, kaddr + pgoff, len, csum); -kunmap_atomic (kaddr); +kunmap_local (kaddr);if (memcmp (csum, csum_expected, csum_size)) goto zeroit;Signature · 1 line
2.36.1
Quoted 17 lines · Fabio M. De Francesco
kmap_atomic()is being deprecated in favor ofkmap_local_page()where it is feasible. Withkmap_local_page()mappings are per thread, CPU local, and not globally visible.As far as I can see, the
kmap_atomic()calls incompression.cand ininode.ccan be safely converted.Above all else, David Sterba has confirmed that "The context in
check_compressed_csumis atomic [...]" and that "kmap_atomic()ininode.c[...] also can be replaced by kmap_local_page().".[1]Therefore, convert all
kmap_atomic()calls currently still left in fs/btrfs tokmap_local_page().Tested with xfstests on a QEMU + KVM 32-bits VM with 4GB RAM and booting a kernel with HIGHMEM64GB enabled.
Quoted 6 lines
[1] https://lore.kernel.org/linux-btrfs/20220601132545.GM20
[email protected]/Suggested-by Ira Weiny <[email protected]>Suggested-by David Sterba <[email protected]>Signed-off-by Fabio M. De Francesco <[email protected]>
Added to the kmap patches, thanks.
Quoted 17 lines
---
Tests of groups "quick" and "compress" output several errors largely due to memory leaks and shift-out-of-bounds. However, these errors are exactly the same which are output without this and other conversions of mine to use
kmap_local_page(). Therefore, it looks like these changes don't introduce regressions.The previous RFC PATCH can be ignored:
https://lore.kernel.org/lkml/[email protected]/With this patch, in fs/btrfs there are no longer call sites of
kmap()andkmap_atomic().
fs/btrfs/inode.c static int insert_inline_extent(struct btrfs_trans_handle *trans, · line 335
Quoted 5 lines
fs/btrfs/inode.c +1
-1@@ -332,9 +332,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, cur_size = min_t(unsigned long, compressed_size,PAGE_SIZE ); - kaddr =kmap_atomic (cpage); + kaddr =kmap_local_page (cpage);
After some cleanups and simplifications in checksumming functions (that mapped the buffers) only kmap_atomic in insert_inline_extent remains, so the final patch will be a bit shorter.
Quoted 18 lines
fs/btrfs/inode.c +3
-3write_extent_buffer (leaf, kaddr, ptr, cur_size); -kunmap_atomic (kaddr); +kunmap_local (kaddr); i++; ptr += cur_size;@@ -345,9 +345,9 @@ static int insert_inline_extent(struct btrfs_trans_handle *trans, }else { page =find_get_page (inode->vfs_inode.i_mapping,0 );btrfs_set_file_extent_compression (leaf, ei,0 ); - kaddr =kmap_atomic (page); + kaddr =kmap_local_page (page);write_extent_buffer (leaf, kaddr, ptr, size); -kunmap_atomic (kaddr); +kunmap_local (kaddr);put_page (page); }btrfs_mark_buffer_dirty (leaf);