6a69631ec9
[BUG] If we have a filesystem with 4k sectorsize, and an inlined compressed extent created like this: item 4 key (257 INODE_ITEM 0) itemoff 15863 itemsize 160 generation 8 transid 8 size 4096 nbytes 4096 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 sequence 1 flags 0x0(none) item 5 key (257 INODE_REF 256) itemoff 15839 itemsize 24 index 2 namelen 14 name: source_inlined item 6 key (257 EXTENT_DATA 0) itemoff 15770 itemsize 69 generation 8 type 0 (inline) inline extent data size 48 ram_bytes 4096 compression 2 (lzo) Then trying to reflink that extent in an aarch64 system with 64K page size, the reflink would just fail: # xfs_io -f -c "reflink $mnt/source_inlined 0 60k 4k" $mnt/dest XFS_IOC_CLONE_RANGE: Input/output error [CAUSE] In zlib_decompress(), we didn't treat @start_byte as just a page offset, but also use it as an indicator on whether we should error out, without any proper explanation (this is from the very beginning of btrfs). In reality, for subpage cases, although @start_byte can be non-zero, we should never switch input/output buffer nor error out, since the whole input/output buffer should never exceed one sector. Note: The above assumption is only not true if we're going to support multi-page sectorsize. Thus the current code using @start_byte as a condition to switch input/output buffer or finish the decompression is completely incorrect. [FIX] The fix involves several modifications: - Rename @start_byte to @dest_pgoff to properly express its meaning - Use @sectorsize other than PAGE_SIZE to properly initialize the output buffer size - Use correct destination offset inside the destination page - Use memcpy_to_page() to copy the contents to the destination page - Use memzero_page() to zero out the tailing part - Consider early end as an error After the fix, even on 64K page sized aarch64, above reflink now works as expected: # xfs_io -f -c "reflink $mnt/source_inlined 0 60k 4k" $mnt/dest linked 4096/4096 bytes at offset 61440 And results the correct file layout: item 9 key (258 INODE_ITEM 0) itemoff 15542 itemsize 160 generation 10 transid 10 size 65536 nbytes 4096 block group 0 mode 100600 links 1 uid 0 gid 0 rdev 0 sequence 1 flags 0x0(none) item 10 key (258 INODE_REF 256) itemoff 15528 itemsize 14 index 3 namelen 4 name: dest item 11 key (258 XATTR_ITEM 3817753667) itemoff 15445 itemsize 83 location key (0 UNKNOWN.0 0) type XATTR transid 10 data_len 37 name_len 16 name: security.selinux data unconfined_u:object_r:unlabeled_t:s0 item 12 key (258 EXTENT_DATA 61440) itemoff 15392 itemsize 53 generation 10 type 1 (regular) extent data disk byte 13631488 nr 4096 extent data offset 0 nr 4096 ram 4096 extent compression 0 (none) Signed-off-by: Qu Wenruo <wqu@suse.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
478 lines
12 KiB
C
478 lines
12 KiB
C
// SPDX-License-Identifier: GPL-2.0
|
|
/*
|
|
* Copyright (C) 2008 Oracle. All rights reserved.
|
|
*/
|
|
|
|
#include <linux/kernel.h>
|
|
#include <linux/slab.h>
|
|
#include <linux/mm.h>
|
|
#include <linux/init.h>
|
|
#include <linux/err.h>
|
|
#include <linux/sched.h>
|
|
#include <linux/pagemap.h>
|
|
#include <linux/bio.h>
|
|
#include <linux/lzo.h>
|
|
#include <linux/refcount.h>
|
|
#include "messages.h"
|
|
#include "compression.h"
|
|
#include "ctree.h"
|
|
#include "super.h"
|
|
#include "btrfs_inode.h"
|
|
|
|
#define LZO_LEN 4
|
|
|
|
/*
|
|
* Btrfs LZO compression format
|
|
*
|
|
* Regular and inlined LZO compressed data extents consist of:
|
|
*
|
|
* 1. Header
|
|
* Fixed size. LZO_LEN (4) bytes long, LE32.
|
|
* Records the total size (including the header) of compressed data.
|
|
*
|
|
* 2. Segment(s)
|
|
* Variable size. Each segment includes one segment header, followed by data
|
|
* payload.
|
|
* One regular LZO compressed extent can have one or more segments.
|
|
* For inlined LZO compressed extent, only one segment is allowed.
|
|
* One segment represents at most one sector of uncompressed data.
|
|
*
|
|
* 2.1 Segment header
|
|
* Fixed size. LZO_LEN (4) bytes long, LE32.
|
|
* Records the total size of the segment (not including the header).
|
|
* Segment header never crosses sector boundary, thus it's possible to
|
|
* have at most 3 padding zeros at the end of the sector.
|
|
*
|
|
* 2.2 Data Payload
|
|
* Variable size. Size up limit should be lzo1x_worst_compress(sectorsize)
|
|
* which is 4419 for a 4KiB sectorsize.
|
|
*
|
|
* Example with 4K sectorsize:
|
|
* Page 1:
|
|
* 0 0x2 0x4 0x6 0x8 0xa 0xc 0xe 0x10
|
|
* 0x0000 | Header | SegHdr 01 | Data payload 01 ... |
|
|
* ...
|
|
* 0x0ff0 | SegHdr N | Data payload N ... |00|
|
|
* ^^ padding zeros
|
|
* Page 2:
|
|
* 0x1000 | SegHdr N+1| Data payload N+1 ... |
|
|
*/
|
|
|
|
#define WORKSPACE_BUF_LENGTH (lzo1x_worst_compress(PAGE_SIZE))
|
|
#define WORKSPACE_CBUF_LENGTH (lzo1x_worst_compress(PAGE_SIZE))
|
|
|
|
struct workspace {
|
|
void *mem;
|
|
void *buf; /* where decompressed data goes */
|
|
void *cbuf; /* where compressed data goes */
|
|
struct list_head list;
|
|
};
|
|
|
|
static struct workspace_manager wsm;
|
|
|
|
void lzo_free_workspace(struct list_head *ws)
|
|
{
|
|
struct workspace *workspace = list_entry(ws, struct workspace, list);
|
|
|
|
kvfree(workspace->buf);
|
|
kvfree(workspace->cbuf);
|
|
kvfree(workspace->mem);
|
|
kfree(workspace);
|
|
}
|
|
|
|
struct list_head *lzo_alloc_workspace(unsigned int level)
|
|
{
|
|
struct workspace *workspace;
|
|
|
|
workspace = kzalloc(sizeof(*workspace), GFP_KERNEL);
|
|
if (!workspace)
|
|
return ERR_PTR(-ENOMEM);
|
|
|
|
workspace->mem = kvmalloc(LZO1X_MEM_COMPRESS, GFP_KERNEL | __GFP_NOWARN);
|
|
workspace->buf = kvmalloc(WORKSPACE_BUF_LENGTH, GFP_KERNEL | __GFP_NOWARN);
|
|
workspace->cbuf = kvmalloc(WORKSPACE_CBUF_LENGTH, GFP_KERNEL | __GFP_NOWARN);
|
|
if (!workspace->mem || !workspace->buf || !workspace->cbuf)
|
|
goto fail;
|
|
|
|
INIT_LIST_HEAD(&workspace->list);
|
|
|
|
return &workspace->list;
|
|
fail:
|
|
lzo_free_workspace(&workspace->list);
|
|
return ERR_PTR(-ENOMEM);
|
|
}
|
|
|
|
static inline void write_compress_length(char *buf, size_t len)
|
|
{
|
|
__le32 dlen;
|
|
|
|
dlen = cpu_to_le32(len);
|
|
memcpy(buf, &dlen, LZO_LEN);
|
|
}
|
|
|
|
static inline size_t read_compress_length(const char *buf)
|
|
{
|
|
__le32 dlen;
|
|
|
|
memcpy(&dlen, buf, LZO_LEN);
|
|
return le32_to_cpu(dlen);
|
|
}
|
|
|
|
/*
|
|
* Will do:
|
|
*
|
|
* - Write a segment header into the destination
|
|
* - Copy the compressed buffer into the destination
|
|
* - Make sure we have enough space in the last sector to fit a segment header
|
|
* If not, we will pad at most (LZO_LEN (4)) - 1 bytes of zeros.
|
|
*
|
|
* Will allocate new pages when needed.
|
|
*/
|
|
static int copy_compressed_data_to_page(char *compressed_data,
|
|
size_t compressed_size,
|
|
struct page **out_pages,
|
|
unsigned long max_nr_page,
|
|
u32 *cur_out,
|
|
const u32 sectorsize)
|
|
{
|
|
u32 sector_bytes_left;
|
|
u32 orig_out;
|
|
struct page *cur_page;
|
|
char *kaddr;
|
|
|
|
if ((*cur_out / PAGE_SIZE) >= max_nr_page)
|
|
return -E2BIG;
|
|
|
|
/*
|
|
* We never allow a segment header crossing sector boundary, previous
|
|
* run should ensure we have enough space left inside the sector.
|
|
*/
|
|
ASSERT((*cur_out / sectorsize) == (*cur_out + LZO_LEN - 1) / sectorsize);
|
|
|
|
cur_page = out_pages[*cur_out / PAGE_SIZE];
|
|
/* Allocate a new page */
|
|
if (!cur_page) {
|
|
cur_page = btrfs_alloc_compr_page();
|
|
if (!cur_page)
|
|
return -ENOMEM;
|
|
out_pages[*cur_out / PAGE_SIZE] = cur_page;
|
|
}
|
|
|
|
kaddr = kmap_local_page(cur_page);
|
|
write_compress_length(kaddr + offset_in_page(*cur_out),
|
|
compressed_size);
|
|
*cur_out += LZO_LEN;
|
|
|
|
orig_out = *cur_out;
|
|
|
|
/* Copy compressed data */
|
|
while (*cur_out - orig_out < compressed_size) {
|
|
u32 copy_len = min_t(u32, sectorsize - *cur_out % sectorsize,
|
|
orig_out + compressed_size - *cur_out);
|
|
|
|
kunmap_local(kaddr);
|
|
|
|
if ((*cur_out / PAGE_SIZE) >= max_nr_page)
|
|
return -E2BIG;
|
|
|
|
cur_page = out_pages[*cur_out / PAGE_SIZE];
|
|
/* Allocate a new page */
|
|
if (!cur_page) {
|
|
cur_page = btrfs_alloc_compr_page();
|
|
if (!cur_page)
|
|
return -ENOMEM;
|
|
out_pages[*cur_out / PAGE_SIZE] = cur_page;
|
|
}
|
|
kaddr = kmap_local_page(cur_page);
|
|
|
|
memcpy(kaddr + offset_in_page(*cur_out),
|
|
compressed_data + *cur_out - orig_out, copy_len);
|
|
|
|
*cur_out += copy_len;
|
|
}
|
|
|
|
/*
|
|
* Check if we can fit the next segment header into the remaining space
|
|
* of the sector.
|
|
*/
|
|
sector_bytes_left = round_up(*cur_out, sectorsize) - *cur_out;
|
|
if (sector_bytes_left >= LZO_LEN || sector_bytes_left == 0)
|
|
goto out;
|
|
|
|
/* The remaining size is not enough, pad it with zeros */
|
|
memset(kaddr + offset_in_page(*cur_out), 0,
|
|
sector_bytes_left);
|
|
*cur_out += sector_bytes_left;
|
|
|
|
out:
|
|
kunmap_local(kaddr);
|
|
return 0;
|
|
}
|
|
|
|
int lzo_compress_pages(struct list_head *ws, struct address_space *mapping,
|
|
u64 start, struct page **pages, unsigned long *out_pages,
|
|
unsigned long *total_in, unsigned long *total_out)
|
|
{
|
|
struct workspace *workspace = list_entry(ws, struct workspace, list);
|
|
const u32 sectorsize = btrfs_sb(mapping->host->i_sb)->sectorsize;
|
|
struct page *page_in = NULL;
|
|
char *sizes_ptr;
|
|
const unsigned long max_nr_page = *out_pages;
|
|
int ret = 0;
|
|
/* Points to the file offset of input data */
|
|
u64 cur_in = start;
|
|
/* Points to the current output byte */
|
|
u32 cur_out = 0;
|
|
u32 len = *total_out;
|
|
|
|
ASSERT(max_nr_page > 0);
|
|
*out_pages = 0;
|
|
*total_out = 0;
|
|
*total_in = 0;
|
|
|
|
/*
|
|
* Skip the header for now, we will later come back and write the total
|
|
* compressed size
|
|
*/
|
|
cur_out += LZO_LEN;
|
|
while (cur_in < start + len) {
|
|
char *data_in;
|
|
const u32 sectorsize_mask = sectorsize - 1;
|
|
u32 sector_off = (cur_in - start) & sectorsize_mask;
|
|
u32 in_len;
|
|
size_t out_len;
|
|
|
|
/* Get the input page first */
|
|
if (!page_in) {
|
|
page_in = find_get_page(mapping, cur_in >> PAGE_SHIFT);
|
|
ASSERT(page_in);
|
|
}
|
|
|
|
/* Compress at most one sector of data each time */
|
|
in_len = min_t(u32, start + len - cur_in, sectorsize - sector_off);
|
|
ASSERT(in_len);
|
|
data_in = kmap_local_page(page_in);
|
|
ret = lzo1x_1_compress(data_in +
|
|
offset_in_page(cur_in), in_len,
|
|
workspace->cbuf, &out_len,
|
|
workspace->mem);
|
|
kunmap_local(data_in);
|
|
if (ret < 0) {
|
|
pr_debug("BTRFS: lzo in loop returned %d\n", ret);
|
|
ret = -EIO;
|
|
goto out;
|
|
}
|
|
|
|
ret = copy_compressed_data_to_page(workspace->cbuf, out_len,
|
|
pages, max_nr_page,
|
|
&cur_out, sectorsize);
|
|
if (ret < 0)
|
|
goto out;
|
|
|
|
cur_in += in_len;
|
|
|
|
/*
|
|
* Check if we're making it bigger after two sectors. And if
|
|
* it is so, give up.
|
|
*/
|
|
if (cur_in - start > sectorsize * 2 && cur_in - start < cur_out) {
|
|
ret = -E2BIG;
|
|
goto out;
|
|
}
|
|
|
|
/* Check if we have reached page boundary */
|
|
if (PAGE_ALIGNED(cur_in)) {
|
|
put_page(page_in);
|
|
page_in = NULL;
|
|
}
|
|
}
|
|
|
|
/* Store the size of all chunks of compressed data */
|
|
sizes_ptr = kmap_local_page(pages[0]);
|
|
write_compress_length(sizes_ptr, cur_out);
|
|
kunmap_local(sizes_ptr);
|
|
|
|
ret = 0;
|
|
*total_out = cur_out;
|
|
*total_in = cur_in - start;
|
|
out:
|
|
if (page_in)
|
|
put_page(page_in);
|
|
*out_pages = DIV_ROUND_UP(cur_out, PAGE_SIZE);
|
|
return ret;
|
|
}
|
|
|
|
/*
|
|
* Copy the compressed segment payload into @dest.
|
|
*
|
|
* For the payload there will be no padding, just need to do page switching.
|
|
*/
|
|
static void copy_compressed_segment(struct compressed_bio *cb,
|
|
char *dest, u32 len, u32 *cur_in)
|
|
{
|
|
u32 orig_in = *cur_in;
|
|
|
|
while (*cur_in < orig_in + len) {
|
|
struct page *cur_page;
|
|
u32 copy_len = min_t(u32, PAGE_SIZE - offset_in_page(*cur_in),
|
|
orig_in + len - *cur_in);
|
|
|
|
ASSERT(copy_len);
|
|
cur_page = cb->compressed_pages[*cur_in / PAGE_SIZE];
|
|
|
|
memcpy_from_page(dest + *cur_in - orig_in, cur_page,
|
|
offset_in_page(*cur_in), copy_len);
|
|
|
|
*cur_in += copy_len;
|
|
}
|
|
}
|
|
|
|
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;
|
|
char *kaddr;
|
|
int ret;
|
|
/* Compressed data length, can be unaligned */
|
|
u32 len_in;
|
|
/* Offset inside the compressed data */
|
|
u32 cur_in = 0;
|
|
/* Bytes decompressed so far */
|
|
u32 cur_out = 0;
|
|
|
|
kaddr = kmap_local_page(cb->compressed_pages[0]);
|
|
len_in = read_compress_length(kaddr);
|
|
kunmap_local(kaddr);
|
|
cur_in += LZO_LEN;
|
|
|
|
/*
|
|
* LZO header length check
|
|
*
|
|
* The total length should not exceed the maximum extent length,
|
|
* and all sectors should be used.
|
|
* If this happens, it means the compressed extent is corrupted.
|
|
*/
|
|
if (len_in > min_t(size_t, BTRFS_MAX_COMPRESSED, cb->compressed_len) ||
|
|
round_up(len_in, sectorsize) < cb->compressed_len) {
|
|
btrfs_err(fs_info,
|
|
"invalid lzo header, lzo len %u compressed len %u",
|
|
len_in, cb->compressed_len);
|
|
return -EUCLEAN;
|
|
}
|
|
|
|
/* Go through each lzo segment */
|
|
while (cur_in < len_in) {
|
|
struct page *cur_page;
|
|
/* Length of the compressed segment */
|
|
u32 seg_len;
|
|
u32 sector_bytes_left;
|
|
size_t out_len = lzo1x_worst_compress(sectorsize);
|
|
|
|
/*
|
|
* We should always have enough space for one segment header
|
|
* inside current sector.
|
|
*/
|
|
ASSERT(cur_in / sectorsize ==
|
|
(cur_in + LZO_LEN - 1) / sectorsize);
|
|
cur_page = cb->compressed_pages[cur_in / PAGE_SIZE];
|
|
ASSERT(cur_page);
|
|
kaddr = kmap_local_page(cur_page);
|
|
seg_len = read_compress_length(kaddr + offset_in_page(cur_in));
|
|
kunmap_local(kaddr);
|
|
cur_in += LZO_LEN;
|
|
|
|
if (seg_len > WORKSPACE_CBUF_LENGTH) {
|
|
/*
|
|
* seg_len shouldn't be larger than we have allocated
|
|
* for workspace->cbuf
|
|
*/
|
|
btrfs_err(fs_info, "unexpectedly large lzo segment len %u",
|
|
seg_len);
|
|
return -EIO;
|
|
}
|
|
|
|
/* Copy the compressed segment payload into workspace */
|
|
copy_compressed_segment(cb, workspace->cbuf, seg_len, &cur_in);
|
|
|
|
/* Decompress the data */
|
|
ret = lzo1x_decompress_safe(workspace->cbuf, seg_len,
|
|
workspace->buf, &out_len);
|
|
if (ret != LZO_E_OK) {
|
|
btrfs_err(fs_info, "failed to decompress");
|
|
return -EIO;
|
|
}
|
|
|
|
/* Copy the data into inode pages */
|
|
ret = btrfs_decompress_buf2page(workspace->buf, out_len, cb, cur_out);
|
|
cur_out += out_len;
|
|
|
|
/* All data read, exit */
|
|
if (ret == 0)
|
|
return 0;
|
|
ret = 0;
|
|
|
|
/* Check if the sector has enough space for a segment header */
|
|
sector_bytes_left = sectorsize - (cur_in % sectorsize);
|
|
if (sector_bytes_left >= LZO_LEN)
|
|
continue;
|
|
|
|
/* Skip the padding zeros */
|
|
cur_in += sector_bytes_left;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
int lzo_decompress(struct list_head *ws, const u8 *data_in,
|
|
struct page *dest_page, unsigned long dest_pgoff, size_t srclen,
|
|
size_t destlen)
|
|
{
|
|
struct workspace *workspace = list_entry(ws, struct workspace, list);
|
|
struct btrfs_fs_info *fs_info = btrfs_sb(dest_page->mapping->host->i_sb);
|
|
const u32 sectorsize = fs_info->sectorsize;
|
|
size_t in_len;
|
|
size_t out_len;
|
|
size_t max_segment_len = WORKSPACE_BUF_LENGTH;
|
|
int ret = 0;
|
|
|
|
if (srclen < LZO_LEN || srclen > max_segment_len + LZO_LEN * 2)
|
|
return -EUCLEAN;
|
|
|
|
in_len = read_compress_length(data_in);
|
|
if (in_len != srclen)
|
|
return -EUCLEAN;
|
|
data_in += LZO_LEN;
|
|
|
|
in_len = read_compress_length(data_in);
|
|
if (in_len != srclen - LZO_LEN * 2) {
|
|
ret = -EUCLEAN;
|
|
goto out;
|
|
}
|
|
data_in += LZO_LEN;
|
|
|
|
out_len = sectorsize;
|
|
ret = lzo1x_decompress_safe(data_in, in_len, workspace->buf, &out_len);
|
|
if (ret != LZO_E_OK) {
|
|
pr_warn("BTRFS: decompress failed!\n");
|
|
ret = -EIO;
|
|
goto out;
|
|
}
|
|
|
|
ASSERT(out_len <= sectorsize);
|
|
memcpy_to_page(dest_page, dest_pgoff, workspace->buf, out_len);
|
|
/* Early end, considered as an error. */
|
|
if (unlikely(out_len < destlen)) {
|
|
ret = -EIO;
|
|
memzero_page(dest_page, dest_pgoff + out_len, destlen - out_len);
|
|
}
|
|
out:
|
|
return ret;
|
|
}
|
|
|
|
const struct btrfs_compress_op btrfs_lzo_compress = {
|
|
.workspace_manager = &wsm,
|
|
.max_level = 1,
|
|
.default_level = 1,
|
|
};
|