On 2024/9/16 21:56, Yiyang Wu via Linux-erofs wrote:
>
> This patch introduces the same on-disk erofs data structure
> in rust and also introduces multiple helpers for inode i_format
> and chunk_indexing and later can be used to implement map_blocks.
>
> Signed-off-by: Yiyang Wu
> ---
> fs/ero
On 2024/9/6 10:31, Gao Xiang wrote:
>
> Hi Jianan,
>
> On 2024/9/6 10:22, Hongzhen Luo wrote:
>> The current `fsck --extract` does not support exporting the extended
>> attributes of files. This patch adds `--xattrs` option to dump the
>> extended attributes, as well as the `--no-xattrs` option t
On 2024/9/5 19:37, Hongzhen Luo wrote:
>
> The current `fsck --extract` does not support exporting the extended
> attributes of files. This patch adds `--xattrs` option to dump the
> extended attributes, as well as the `--no-xattrs` option to not dump
> the extended attributes.
>
> Signed-off-by:
Fix some minor issues, including:
- Spelling mistakes;
- Remove redundant spaces and parenthesis;
- clean up file headers;
- Match parameters with format parameters.
Signed-off-by: Huang Jianan
---
Changes since v1:
- Removed modification related to alignment format;
fuse/main.c
Execute fclose before return error.
Signed-off-by: Huang Jianan
---
lib/compress_hints.c | 10 +++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/lib/compress_hints.c b/lib/compress_hints.c
index 25adf35..92964eb 100644
--- a/lib/compress_hints.c
+++ b/lib/compress_hints.c
在 2022/2/18 14:20, Gao Xiang 写道:
Hi Jianan,
On Fri, Feb 18, 2022 at 11:11:37AM +0800, Huang Jianan via Linux-erofs wrote:
Fix some minor issues, including:
- Align with the left parenthesis;
- Spelling mistakes;
- Remove redundant spaces and parenthesis;
- clean up file headers
在 2022/2/18 15:02, Gao Xiang 写道:
On Fri, Feb 18, 2022 at 11:11:36AM +0800, Huang Jianan via Linux-erofs wrote:
Execute fclose before return error.
Signed-off-by: Huang Jianan
It's actually a file descriptor leakage?
Thanks,
Gao Xiang
Yes, I will fix this in the next version.
T
Execute fclose before return error.
Signed-off-by: Huang Jianan
---
lib/compress_hints.c | 10 +++---
1 file changed, 7 insertions(+), 3 deletions(-)
diff --git a/lib/compress_hints.c b/lib/compress_hints.c
index c3f3d48..c52e2d3 100644
--- a/lib/compress_hints.c
+++ b/lib/compress_hints.c
Fix some minor issues, including:
- Align with the left parenthesis;
- Spelling mistakes;
- Remove redundant spaces and parenthesis;
- clean up file headers;
- Match parameters with format parameters.
Signed-off-by: Huang Jianan
---
dump/main.c| 53 +++--
Need to check if we got a normal length.
Signed-off-by: Huang Jianan
---
lib/blobchunk.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/blobchunk.c b/lib/blobchunk.c
index 4a440d6..483362b 100644
--- a/lib/blobchunk.c
+++ b/lib/blobchunk.c
@@ -221,6 +221,8 @@ int erofs_blob_remap(void
Sort shared xattr before writing to disk to ensure the consistency
of reproducible builds.
Signed-off-by: Huang Jianan
---
since v2:
- Only traverse shared_xattrs_list once to clean up code.
since v1:
- Use strncmp instead of strcmp.
lib/xattr.c | 42 --
在 2021/12/15 11:22, Gao Xiang 写道:
On Wed, Dec 15, 2021 at 10:43:40AM +0800, Huang Jianan via Linux-erofs wrote:
Hi Xiang,
在 2021/12/14 23:31, Gao Xiang 写道:
Hi Jianan,
On Tue, Dec 14, 2021 at 05:52:01PM +0800, Huang Jianan via Linux-erofs wrote:
Sort shared xattr before writing to disk to
Hi Xiang,
在 2021/12/14 23:31, Gao Xiang 写道:
Hi Jianan,
On Tue, Dec 14, 2021 at 05:52:01PM +0800, Huang Jianan via Linux-erofs wrote:
Sort shared xattr before writing to disk to ensure the consistency
of reproducible builds.
Signed-off-by: Huang Jianan
I still fail to understand why the
Add basic function check for xattr.
Signed-off-by: Huang Jianan
---
tests/Makefile.am | 3 +++
tests/erofs/019 | 63 +
tests/erofs/019.out | 2 ++
3 files changed, 68 insertions(+)
create mode 100755 tests/erofs/019
create mode 100644 tests/e
Sort shared xattr before writing to disk to ensure the consistency
of reproducible builds.
Signed-off-by: Huang Jianan
---
since v1:
- use strncmp instead of strcmp.
lib/xattr.c | 36 ++--
1 file changed, 34 insertions(+), 2 deletions(-)
diff --git a/lib/xattr.c
Let's release liberofs under GPL-2.0+ OR Apache-2.0 license for
better integration.
Signed-off-by: Huang Jianan
---
include/erofs/blobchunk.h | 2 +-
include/erofs/block_list.h | 2 +-
include/erofs/cache.h | 2 +-
include/erofs/compress.h | 2 +-
include/erofs/compress_h
Add myself to the AUTHORS file.
Signed-off-by: Huang Jianan
---
AUTHORS | 1 +
1 file changed, 1 insertion(+)
diff --git a/AUTHORS b/AUTHORS
index 812042a..6b41df8 100644
--- a/AUTHORS
+++ b/AUTHORS
@@ -1,6 +1,7 @@
EROFS USERSPACE UTILITIES
M: Li Guifu
M: Gao Xiang
+M: Huang Jianan
R: Ch
在 2021/11/23 10:59, Jianhua1 Hao 郝建华 via Linux-erofs 写道:
*We also found that it is easy to cause deadlock in the kswap scene，
We observed the following deadlock in the stress test under low memory
scenario，****Same as "erofs: fix deadlock when shrink erofs slab".*
**
Thread A: Thread B:
erof
We observed the following deadlock in the stress test under low
memory scenario:
Thread A Thread B
- erofs_shrink_scan
- erofs_try_to_release_workgroup
- erofs_workgroup_try_to_freeze -- A
- z_erofs_do_read_page
Need to check whether the allocation of erofs_dentry is successful.
Signed-off-by: Huang Jianan
---
lib/inode.c | 4
1 file changed, 4 insertions(+)
diff --git a/lib/inode.c b/lib/inode.c
index 5cbfc78..2fa74e2 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -162,11 +162,15 @@ int erofs_prep
Release memory allocated for chunk before return error.
Signed-off-by: Huang Jianan
---
lib/blobchunk.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/blobchunk.c b/lib/blobchunk.c
index 5cbb831..b605b0b 100644
--- a/lib/blobchunk.c
+++ b/lib/blobchunk.c
@@ -78,6 +78,7 @@ static struct
在 2021/11/12 13:36, Gao Xiang 写道:
Hi Jianan,
On Fri, Nov 12, 2021 at 10:30:02AM +0800, Huang Jianan wrote:
Add sysfs interface to configure erofs related parameters later.
Signed-off-by: Huang Jianan
Reviewed-by: Chao Yu
---
since v4:
- Resend in a clean chain.
since v3:
- Add description o
在 2021/11/11 14:17, Gao Xiang 写道:
Hi David,
On Thu, Nov 11, 2021 at 05:30:31AM +, David Anderson via Linux-erofs wrote:
When using the --block-list-file option, add block mapping lines for
chunked files. The extent printing code has been slightly refactored to
accommodate multiple extent ra
Add sysfs interface to configure erofs related parameters later.
Signed-off-by: Huang Jianan
Reviewed-by: Chao Yu
---
since v4:
- Resend in a clean chain.
since v3:
- Add description of sysfs in erofs documentation.
since v2:
- Check whether t in erofs_attr_store is illegal.
- Print raw value
Although readpage is a synchronous path, there will be no additional
kworker scheduling overhead in non-atomic contexts. So add a sysfs
node to allow disable sync decompression.
Signed-off-by: Huang Jianan
Reviewed-by: Chao Yu
---
since v4:
- Resend in a clean chain.
since v3:
- Clean up the sy
在 2021/11/12 4:17, kernel test robot 写道:
Hi Huang,
Thank you for the patch! Yet something to improve:
[auto build test ERROR on xiang-erofs/dev-test]
[also build test ERROR on next-2021]
[cannot apply to v5.15]
[If your patch is applied to the wrong git tree, kindly drop us a note.
And when
Although readpage is a synchronous path, there will be no additional
kworker scheduling overhead in non-atomic contexts. So add a sysfs
node to allow disable sync decompression.
Signed-off-by: Huang Jianan
---
changes since v1:
- leave auto default
- add a disable strategy for sync_decompress
D
Add sysfs interface to configure erofs related parameters later.
Signed-off-by: Huang Jianan
---
changes since v1:
- Add sysfs API documentation.
- Use sysfs_emit over snprintf.
Documentation/ABI/testing/sysfs-fs-erofs | 7 +
fs/erofs/Makefile| 2 +-
fs/erofs/interna
Hi Xiang,
在 2021/11/9 11:16, Gao Xiang 写道:
On Tue, Nov 09, 2021 at 10:54:45AM +0800, Huang Jianan via Linux-erofs wrote:
Although readpage is a synchronous path, there will be no additional
kworker scheduling overhead in non-atomic contexts. So we can add a
sysfs node to allow disable sync
在 2021/11/9 11:14, Joe Perches 写道:
On Tue, 2021-11-09 at 10:54 +0800, Huang Jianan wrote:
Add sysfs interface to configure erofs related parameters in the
future.
[]
diff --git a/fs/erofs/sysfs.c b/fs/erofs/sysfs.c
[]
+static ssize_t erofs_attr_show(struct kobject *kobj,
+
Hi Xiang,
在 2021/11/9 11:10, Gao Xiang 写道:
Hi Jianan,
On Tue, Nov 09, 2021 at 10:54:44AM +0800, Huang Jianan via Linux-erofs wrote:
You might need to add a "From:" tag here, otherwise, the author will
show "Huang Jianan via Linux-erofs" due to your mailing server...
I hav
Although readpage is a synchronous path, there will be no additional
kworker scheduling overhead in non-atomic contexts. So we can add a
sysfs node to allow disable sync decompression.
Signed-off-by: Huang Jianan
---
fs/erofs/internal.h | 2 +-
fs/erofs/super.c| 2 +-
fs/erofs/sysfs.c| 6
Add sysfs interface to configure erofs related parameters in the
future.
Signed-off-by: Huang Jianan
---
fs/erofs/Makefile | 2 +-
fs/erofs/internal.h | 10 ++
fs/erofs/super.c| 12 +++
fs/erofs/sysfs.c| 239
4 files changed, 262 inserti
在 2021/10/21 15:25, Gao Xiang 写道:
Hi Jianan,
On Thu, Oct 21, 2021 at 10:58:47AM +0800, Huang Jianan via Linux-erofs wrote:
Sort shared xattr before writing to disk to ensure the consistency
of reproducible builds.
How about adding it as an option?
Can we consider turning on this by
The compiled product should be added to gitignore.
Signed-off-by: Huang Jianan
---
.gitignore | 1 +
1 file changed, 1 insertion(+)
diff --git a/.gitignore b/.gitignore
index 7bc3f58..27403d4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -29,3 +29,4 @@ stamp-h
stamp-h1
/mkfs/mkfs.erofs
/fuse/e
---
.gitignore | 1 +
1 file changed, 1 insertion(+)
diff --git a/.gitignore b/.gitignore
index 7bc3f58..27403d4 100644
--- a/.gitignore
+++ b/.gitignore
@@ -29,3 +29,4 @@ stamp-h
stamp-h1
/mkfs/mkfs.erofs
/fuse/erofsfuse
+/dump/dump.erofs
--
2.25.1
Sort shared xattr before writing to disk to ensure the consistency
of reproducible builds.
---
lib/xattr.c | 34 ++
1 file changed, 30 insertions(+), 4 deletions(-)
diff --git a/lib/xattr.c b/lib/xattr.c
index 196133a..f17e57e 100644
--- a/lib/xattr.c
+++ b/lib/xat
在 2021/9/27 17:38, Miklos Szeredi 写道:
On Wed, Sep 22, 2021 at 04:00:47PM +0200, Miklos Szeredi wrote:
First let's fix the oops: ovl_read_iter()/ovl_write_iter() must check
real file's ->direct_IO if IOCB_DIRECT is set in iocb->ki_flags and
return -EINVAL if not.
And here's that fix. Please te
在 2021/9/22 16:06, Chengguang Xu 写道:
在 2021/9/22 15:23, Huang Jianan 写道:
From: Huang Jianan
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
From: Huang Jianan
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
squashfs, don't support direct_IO.
Users who use f_mapping->a_ops->direct_IO t
在 2021/9/22 13:09, Chengguang Xu 写道:
在 2021/9/22 11:47, Huang Jianan 写道:
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
squashfs, don't support
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
squashfs, don't support direct_IO.
Users who use f_mapping->a_ops->direct_IO to check O_DIRECT supp
在 2021/9/22 9:56, Chengguang Xu 写道:
在 2021/9/18 20:13, Huang Jianan 写道:
From: Huang Jianan
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
squa
From: Huang Jianan
At present, overlayfs provides overlayfs inode to users. Overlayfs
inode provides ovl_aops with noop_direct_IO to avoid open failure
with O_DIRECT. But some compressed filesystems, such as erofs and
squashfs, don't support direct_IO.
Users who use f_mapping->a_ops->direct_IO t
Signed-off-by: Huang Jianan
---
tests/Makefile.am | 5 ++-
tests/common/rc | 2 +-
tests/erofs/017 | 78 +
tests/erofs/017.out | 2 ++
4 files changed, 85 insertions(+), 2 deletions(-)
create mode 100755 tests/erofs/017
create mode 100644
在 2021/9/12 0:13, Gao Xiang 写道:
(+Cc Jianan.)
On Sat, Sep 11, 2021 at 09:46:33PM +0800, Guo Xuenan wrote:
From: mpiglet
Signed-off-by: Guo Xuenan
Signed-off-by: mpiglet
---
dump/main.c | 474
1 file changed, 474 insertions(+)
diff --g
Previously, there is no need to get the full decompressed length since
EROFS supports partial decompression. However for some other cases
such as fiemap, the full decompressed length is necessary for iomap to
make it work properly.
This patch adds a way to get the full decompressed length. Note th
在 2021/9/7 21:20, Gao Xiang 写道:
On Tue, Sep 07, 2021 at 11:53:45AM +0800, Huang Jianan via Linux-erofs wrote:
If the file size is block-aligned for uncompressed files, i_u is
meaningless for erofs_inode on disk, but it's not cleared when
datalayout is seted in erofs_prepare_inode_buffer.
If the file size is block-aligned for uncompressed files, i_u is
meaningless for erofs_inode on disk, but it's not cleared when
datalayout is seted in erofs_prepare_inode_buffer.
This problem will cause inconsistent reproducible builds. Clear the
entire erofs_inode to zero to fix this.
Signed-off
在 2021/9/7 8:46, Gao Xiang 写道:
On Tue, Sep 07, 2021 at 08:10:45AM +0800, Gao Xiang wrote:
On Mon, Sep 06, 2021 at 04:13:59PM +0800, Huang Jianan via Linux-erofs wrote:
If the file size is block-aligned for uncompressed files, i_u is
meaningless for erofs_inode on disk, but it's not cl
在 2021/9/6 1:59, Gao Xiang 写道:
On Wed, Aug 25, 2021 at 11:35:23AM +0800, Huang Jianan via Linux-erofs wrote:
Add an option to configure per-inode compression strategy. Each line
of the file should be in the following form:
When pcluster is 0, it means that the file shouldn't be compr
If the file size is block-aligned for uncompressed files, i_u is
meaningless for erofs_inode on disk, but it's not cleared when
datalayout is seted in erofs_prepare_inode_buffer. Clear the entire
erofs_inode to zero to fix this.
Signed-off-by: Huang Jianan
---
lib/inode.c | 2 +-
1 file changed,
Signed-off-by: Huang Jianan
---
lib/inode.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/inode.c b/lib/inode.c
index 5bad75e..0ad703d 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@@ -744,7 +744,7 @@ int erofs_droid_inode_fsconfig(struct erofs_inode *inode,
Use larger numbers for AOSP-specific options in case of bump up
generic options to >= 256 in the future.
Signed-off-by: Huang Jianan
---
changes since v1:
- use 512 for AOSP-specific options in case of bump up generic options
to >= 256 in the future. (Gao Xiang)
mkfs/main.c | 16 ---
在 2021/8/25 15:25, Gao Xiang 写道:
Hi Jianan,
On Wed, Aug 25, 2021 at 11:34:16AM +0800, Huang Jianan via Linux-erofs wrote:
Subject: AOSP: erofs-utils: increase val for AOSP-specific long options
Signed-off-by: Huang Jianan
---
mkfs/main.c | 16
1 file changed, 8
Add an option to configure per-inode compression strategy. Each line
of the file should be in the following form:
When pcluster is 0, it means that the file shouldn't be compressed.
Signed-off-by: Huang Jianan
---
changes since v2:
- change compress_rule to compress_hints for better understa
Signed-off-by: Huang Jianan
---
mkfs/main.c | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/mkfs/main.c b/mkfs/main.c
index 10fe14d..9369b72 100644
--- a/mkfs/main.c
+++ b/mkfs/main.c
@@ -45,10 +45,10 @@ static struct option long_options[] = {
#endif
在 2021/8/25 9:27, Gao Xiang 写道:
On Wed, Aug 25, 2021 at 09:17:38AM +0800, Gao Xiang wrote:
On Wed, Aug 18, 2021 at 12:27:15PM +0800, Huang Jianan via Linux-erofs wrote:
Add an option to configure per-inode compression strategy. Each line
of the file should be in the following form:
When
---
.gitignore | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/.gitignore b/.gitignore
index 8bdd505..7bc3f58 100644
--- a/.gitignore
+++ b/.gitignore
@@ -27,4 +27,5 @@ configure.scan
libtool
stamp-h
stamp-h1
-
+/mkfs/mkfs.erofs
+/fuse/erofsfuse
--
2.25.1
On 2021/8/19 0:49, Gao Xiang wrote:
On Wed, Aug 18, 2021 at 12:35:39PM +0800, Huang Jianan via Linux-erofs wrote:
---
.gitignore | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/.gitignore b/.gitignore
index 8bdd505..0e54836 100644
--- a/.gitignore
+++ b/.gitignore
---
.gitignore | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/.gitignore b/.gitignore
index 8bdd505..0e54836 100644
--- a/.gitignore
+++ b/.gitignore
@@ -27,4 +27,5 @@ configure.scan
libtool
stamp-h
stamp-h1
-
+mkfs.erofs
+erofsfuse
--
2.25.1
Add an option to configure per-inode compression strategy. Each line
of the file should be in the following form:
When pcluster is 0, it means that the file shouldn't be compressed.
Signed-off-by: Huang Jianan
---
changes since v1:
- rename c_pclusterblks to c_physical_clusterblks and place
Add an option to configure per-inode compression strategy. Each line
of the file should be in the following form:
When pcluster is 0, it means that the file shouldn't be compressed.
Signed-off-by: Huang Jianan
---
include/erofs/compress_rule.h | 25
include/erofs/config.h|
Hi Xiang,
I have tested this patch based on the erofs dax support below, It works
well for me.
https://git.kernel.org/pub/scm/linux/kernel/git/xiang/linux.git/log/?h=erofs/dax
Tested-by: Huang Jianan
Thanks,
Jianan
On 2021/7/23 9:26, Gao Xiang wrote:
Hi Darrick,
On Thu, Jul 22, 2021 at
Hi Xiang,
This verison works well.
Tested-by: Huang Jianan
Thanks,
Jianan
On 2021/6/24 16:04, Gao Xiang wrote:
[hsiang...@linux.alibaba.com 与之前向你发送电子邮件的用户相似，但可能不是其人。请访问
http://aka.ms/LearnAboutSenderIdentification，了解这种情况的风险所在。]
On Thu, Jun 24, 2021 at 03:02:34PM +0800, Yue Hu wrote:
From
Hi all,
This patch works well for us.
Tested-by: Huang Jianan
Thanks,
Jianan
On 2021/6/22 11:02, Yue Hu wrote:
From: Yue Hu
Android update engine will treat EROFS filesystem image as one single
file. Let's add block list support to optimize OTA size.
Change-Id: I21d6177dff0ee65d3c57023b1
__func__ has been printed in the macro definition of erofs_err.
Signed-off-by: Huang Jianan
---
lib/inode.c | 4 ++--
lib/xattr.c | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/lib/inode.c b/lib/inode.c
index 40189fe..b964b98 100644
--- a/lib/inode.c
+++ b/lib/inode.c
@
Sync decompression was introduced to get rid of additional kworker
scheduling overhead. But there is no such overhead in non-atomic
contexts. Therefore, it should be better to turn off sync decompression
to avoid the current thread waiting in z_erofs_runqueue.
Signed-off-by: Huang Jianan
Signed-o
This patch set was separated form erofs: decompress in endio if possible
since it does these things:
- combine dm-verity and erofs workqueue
- change policy of decompression in context of thread
Huang Jianan (2):
erofs: use workqueue decompression for atomic contexts only
erofs: use sync deco
z_erofs_decompressqueue_endio may not be executed in the atomic
context, for example, when dm-verity is turned on. In this scenario,
data can be decompressed directly to get rid of additional kworker
scheduling overhead.
Signed-off-by: Huang Jianan
Signed-off-by: Guo Weichao
Reviewed-by: Gao Xia
On 2021/3/16 16:26, Chao Yu wrote:
Hi Jianan,
On 2021/3/16 11:15, Huang Jianan via Linux-erofs wrote:
z_erofs_decompressqueue_endio may not be executed in the atomic
context, for example, when dm-verity is turned on. In this scenario,
data can be decompressed directly to get rid of
z_erofs_decompressqueue_endio may not be executed in the atomic
context, for example, when dm-verity is turned on. In this scenario,
data can be decompressed directly to get rid of additional kworker
scheduling overhead. Also, it makes no sense to apply synchronous
decompression for such case.
Sig
Currently, err would be treated as io error. Therefore, it'd be
better to ensure memory allocation during rolling decompression
to avoid such io error.
In the long term, we might consider adding another !Uptodate case
for such case.
Signed-off-by: Huang Jianan
Signed-off-by: Guo Weichao
Reviewe
z_erofs_decompressqueue_endio may not be executed in the atomic
context, for example, when dm-verity is turned on. In this scenario,
data can be decompressed directly to get rid of additional kworker
scheduling overhead. Also, it makes no sense to apply synchronous
decompression for such case.
Sig
On 2021/3/5 16:08, Huang Jianan wrote:
z_erofs_decompressqueue_endio may not be executed in the atomic
context, for example, when dm-verity is turned on. In this scenario,
data can be decompressed directly to get rid of additional kworker
scheduling overhead. Also, it makes no sense to apply sy
Currently, err would be treated as io error. Therefore, it'd be
better to ensure memory allocation during rolling decompression
to avoid such io error.
In the long term, we might consider adding another !Uptodate case
for such case.
Signed-off-by: Huang Jianan
Signed-off-by: Guo Weichao
---
fs
It should be better to ensure memory allocation during rolling
decompression to avoid io error.
Signed-off-by: Huang Jianan
Signed-off-by: Guo Weichao
---
fs/erofs/decompressor.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/erofs/decompressor.c b/fs/erofs/decompressor.
z_erofs_decompressqueue_endio may not be executed in the interrupt
context, for example, when dm-verity is turned on. In this scenario,
io should be decompressed directly to avoid additional scheduling
overhead. Also there is no need to wait for endio to execute
synchronous decompression.
Signed-o
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large number of concurrent
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large number of concurrent
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large number of concurrent
From: huangjianan
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large n
Hi Xiang,
On 2021/2/22 12:44, Gao Xiang wrote:
Hi Jianan,
On Thu, Feb 18, 2021 at 08:00:49PM +0800, Huang Jianan via Linux-erofs wrote:
From: huangjianan
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will
From: huangjianan
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large n
lz4 uses LZ4_DISTANCE_MAX to record history preservation. When
using rolling decompression, a block with a higher compression
ratio will cause a larger memory allocation (up to 64k). It may
cause a large resource burden in extreme cases on devices with
small memory and a large number of concurrent
85 matches
Mail list logo
