commit b6391ac73400eff38377a4a7364bd3df5efb5178 upstream.
Complete read error handling paths for all three kinds of
compressed pages:
1) For cache-managed pages, PG_uptodate will be checked since
read_endio will unlock and SetPageUptodate for these pages;
2) For inplaced pages, read_endio cannot SetPageUptodate directly
since it should be used to mark the final decompressed data,
PG_error will be set with page locked for IO error instead;
3) For staging pages, PG_error is used, which is similar to
what we do for inplaced pages.
Fixes: 3883a79abd02 ("staging: erofs: introduce VLE decompression support")
Cc: <stable@vger.kernel.org> # 4.19+
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 8bce6dcede65139a087ff240127e3f3c01363eed upstream.
erofs_vmap() wrapped vmap() and vm_map_ram() to return virtual
continuous memory, but both of them can failed due to a lot of
reason, previously, erofs_vmap()'s callers didn't handle them,
which can potentially cause NULL pointer access, fix it.
Fixes: 3883a79abd02 ("staging: erofs: introduce VLE decompression support")
Fixes: 0d40d6e399c1 ("staging: erofs: add a generic z_erofs VLE decompressor")
Cc: <stable@vger.kernel.org> # 4.19+
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 51232df5e4b268936beccde5248f312a316800be upstream.
When the managed cache is enabled, the last reference count
of a workgroup must be used for its workstation.
Otherwise, it could lead to incorrect (un)freezes in
the reclaim path, and it would be harmful.
A typical race as follows:
Thread 1 (In the reclaim path) Thread 2
workgroup_freeze(grp, 1) refcnt = 1
...
workgroup_unfreeze(grp, 1) refcnt = 1
workgroup_get(grp) refcnt = 2 (x)
workgroup_put(grp) refcnt = 1 (x)
...unexpected behaviors
* grp is detached but still used, which violates cache-managed
freeze constraint.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 419d6efc50e94bcf5d6b35cd8c71f79edadec564 upstream.
As Al pointed out, "
... and while we are at it, what happens to
unsigned int nameoff = le16_to_cpu(de[mid].nameoff);
unsigned int matched = min(startprfx, endprfx);
struct qstr dname = QSTR_INIT(data + nameoff,
unlikely(mid >= ndirents - 1) ?
maxsize - nameoff :
le16_to_cpu(de[mid + 1].nameoff) - nameoff);
/* string comparison without already matched prefix */
int ret = dirnamecmp(name, &dname, &matched);
if le16_to_cpu(de[...].nameoff) is not monotonically increasing? I.e.
what's to prevent e.g. (unsigned)-1 ending up in dname.len?
Corrupted fs image shouldn't oops the kernel.. "
Revisit the related lookup flow to address the issue.
Fixes: d72d1ce60174 ("staging: erofs: add namei functions")
Cc: <stable@vger.kernel.org> # 4.19+
Suggested-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 62dc45979f3f8cb0ea67302a93bff686f0c46c5a upstream.
In real scenario, there could be several threads accessing xattrs
of the same xattr-uninitialized inode, and init_inode_xattrs()
almost at the same time.
That's actually an unexpected behavior, this patch closes the race.
Fixes: b17500a0fdba ("staging: erofs: introduce xattr & acl support")
Cc: <stable@vger.kernel.org> # 4.19+
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit af692e117cb8cd9d3d844d413095775abc1217f9 upstream.
This patch resolves the following page use-after-free issue,
z_erofs_vle_unzip:
...
for (i = 0; i < nr_pages; ++i) {
...
z_erofs_onlinepage_endio(page); (1)
}
for (i = 0; i < clusterpages; ++i) {
page = compressed_pages[i];
if (page->mapping == mngda) (2)
continue;
/* recycle all individual staging pages */
(void)z_erofs_gather_if_stagingpage(page_pool, page); (3)
WRITE_ONCE(compressed_pages[i], NULL);
}
...
After (1) is executed, page is freed and could be then reused, if
compressed_pages is scanned after that, it could fall info (2) or
(3) by mistake and that could finally be in a mess.
This patch aims to solve the above issue only with little changes
as much as possible in order to make the fix backport easier.
Fixes: 3883a79abd02 ("staging: erofs: introduce VLE decompression support")
Cc: <stable@vger.kernel.org> # 4.19+
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 1e5ceeab6929585512c63d05911d6657064abf7b upstream.
Considering a read request with two decompressed file pages,
If a decompression work cannot be started on the previous page
due to memory pressure but in-memory LTP map lookup is done,
builder->work should be still NULL.
Moreover, if the current page also belongs to the same map,
it won't try to start the decompression work again and then
run into trouble.
This patch aims to solve the above issue only with little changes
as much as possible in order to make the fix backport easier.
kernel message is:
<4>[1051408.015930s]SLUB: Unable to allocate memory on node -1, gfp=0x2408040(GFP_NOFS|__GFP_ZERO)
<4>[1051408.015930s] cache: erofs_compress, object size: 144, buffer size: 144, default order: 0, min order: 0
<4>[1051408.015930s] node 0: slabs: 98, objs: 2744, free: 0
* Cannot allocate the decompression work
<3>[1051408.015960s]erofs: z_erofs_vle_normalaccess_readpages, readahead error at page 1008 of nid 5391488
* Note that the previous page was failed to read
<0>[1051408.015960s]Internal error: Accessing user space memory outside uaccess.h routines: 96000005 [#1] PREEMPT SMP
...
<4>[1051408.015991s]Hardware name: kirin710 (DT)
...
<4>[1051408.016021s]PC is at z_erofs_vle_work_add_page+0xa0/0x17c
<4>[1051408.016021s]LR is at z_erofs_do_read_page+0x12c/0xcf0
...
<4>[1051408.018096s][<ffffff80c6fb0fd4>] z_erofs_vle_work_add_page+0xa0/0x17c
<4>[1051408.018096s][<ffffff80c6fb3814>] z_erofs_vle_normalaccess_readpages+0x1a0/0x37c
<4>[1051408.018096s][<ffffff80c6d670b8>] read_pages+0x70/0x190
<4>[1051408.018127s][<ffffff80c6d6736c>] __do_page_cache_readahead+0x194/0x1a8
<4>[1051408.018127s][<ffffff80c6d59318>] filemap_fault+0x398/0x684
<4>[1051408.018127s][<ffffff80c6d8a9e0>] __do_fault+0x8c/0x138
<4>[1051408.018127s][<ffffff80c6d8f90c>] handle_pte_fault+0x730/0xb7c
<4>[1051408.018127s][<ffffff80c6d8fe04>] __handle_mm_fault+0xac/0xf4
<4>[1051408.018157s][<ffffff80c6d8fec8>] handle_mm_fault+0x7c/0x118
<4>[1051408.018157s][<ffffff80c8c52998>] do_page_fault+0x354/0x474
<4>[1051408.018157s][<ffffff80c8c52af8>] do_translation_fault+0x40/0x48
<4>[1051408.018157s][<ffffff80c6c002f4>] do_mem_abort+0x80/0x100
<4>[1051408.018310s]---[ end trace 9f4009a3283bd78b ]---
Fixes: 3883a79abd02 ("staging: erofs: introduce VLE decompression support")
Cc: <stable@vger.kernel.org> # 4.19+
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit a112152f6f3a2a88caa6f414d540bd49e406af60 upstream.
EROFS has an optimized path called TAIL merging, which is designed
to merge multiple reads and the corresponding decompressions into
one if these requests read continuous pages almost at the same time.
In general, it behaves as follows:
________________________________________________________________
... | TAIL . HEAD | PAGE | PAGE | TAIL . HEAD | ...
_____|_combined page A_|________|________|_combined page B_|____
1 ] -> [ 2 ] -> [ 3
If the above three reads are requested in the order 1-2-3, it will
generate a large work chain rather than 3 individual work chains
to reduce scheduling overhead and boost up sequential read.
However, if Read 2 is processed slightly earlier than Read 1,
currently it still generates 2 individual work chains (chain 1, 2)
but it does in-place decompression for combined page A, moreover,
if chain 2 decompresses ahead of chain 1, it will be a race and
lead to corrupted decompressed page. This patch fixes it.
Fixes: 3883a79abd02 ("staging: erofs: introduce VLE decompression support")
Cc: <stable@vger.kernel.org> # 4.19+
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 948bbdb1818b7ad6e539dad4fbd2dd4650793ea9 upstream.
Just like other generic locks, insert a full barrier
in case of memory reorder.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 73f5c66df3e26ab750cefcb9a3e08c71c9f79cad upstream.
There are two minor issues in the current freeze interface:
1) Freeze interfaces have not related with CONFIG_DEBUG_SPINLOCK,
therefore fix the incorrect conditions;
2) For SMP platforms, it should also disable preemption before
doing atomic_cmpxchg in case that some high priority tasks
preempt between atomic_cmpxchg and disable_preempt, then spin
on the locked refcount later.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit df134b8d17b90c1e7720e318d36416b57424ff7a upstream.
It's better to use atomic_cond_read_relaxed, which is implemented
in hardware instructions to monitor a variable changes currently
for ARM64, instead of open-coded busy waiting.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit e9c892465583c8f42d61fafe30970d36580925df upstream.
There is actually no need at all to d_rehash() for the root dentry
as Al pointed out, fix it.
Reported-by: Al Viro <viro@ZenIV.linux.org.uk>
Cc: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit e5e3abbadf0dbd1068f64f8abe70401c5a178180 upstream.
Multiref support means that a compressed page could have
more than one reference, which is designed for on-disk data
deduplication. However, mkfs doesn't support this mode
at this moment, and the kernel implementation is also broken.
Let's drop multiref support. If it is fully implemented
in the future, it can be reverted later.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 1e05ff36e6921ca61bdbf779f81a602863569ee3 upstream.
This patch completes error handing code of z_erofs_do_read_page.
PG_error will be set when some read error happens, therefore
z_erofs_onlinepage_endio will unlock this page without setting
PG_uptodate.
Reviewed-by: Chao Yu <yucxhao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
commit 0734ffbf574ee813b20899caef2fe0ed502bb783 upstream.
As described in Kconfig, the last compressed pack should be cached
for further reading for either `EROFS_FS_ZIP_CACHE_UNIPOLAR' or
`EROFS_FS_ZIP_CACHE_BIPOLAR' by design.
However, there is a bug in z_erofs_do_read_page, it will
switch `initial' to `false' at the very beginning before it decides
to cache the last compressed pack.
caching strategy should work properly after appling this patch.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
[ Upstream commit eef168789866514e5d4316f030131c9fe65b643f ]
It's better not to positively BUG_ON the kernel, however developers
need a way to locate issues as soon as possible.
DBG_BUGON is introduced and it could only crash when EROFS_FS_DEBUG
(EROFS developping feature) is on. It is helpful for developers
to find and solve bugs quickly by eng builds.
Previously, DBG_BUGON is defined as ((void)0) if EROFS_FS_DEBUG is off,
but some unused variable warnings as follows could occur:
drivers/staging/erofs/unzip_vle.c: In function `init_alway:':
drivers/staging/erofs/unzip_vle.c:61:33: warning: unused variable `work' [-Wunused-variable]
struct z_erofs_vle_work *const work =
^~~~
Fix it to #define DBG_BUGON(x) ((void)(x)).
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch follows commit 1751e8a6cb93 ("Rename superblock
flags (MS_xyz -> SB_xyz)") and after commit ("vfs: Suppress
MS_* flag defs within the kernel unless explicitly enabled"),
there is no MS_RDONLY and MS_NOATIME at all.
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: David Howells <dhowells@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This reverts commit 156c3df8d4db4e693c062978186f44079413d74d.
Since XArray and the new mount apis aren't merged in 4.19-rc1
merge window, the BROKEN mark can be reverted directly without
any problems.
Fixes: 156c3df8d4db ("staging: erofs: disable compiling temporarile")
Cc: Matthew Wilcox <willy@infradead.org>
Cc: David Howells <dhowells@redhat.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
As Stephen Rothwell reported:
"After merging the staging tree, today's linux-next build (x86_64
allmodconfig) failed like this:
drivers/staging/erofs/super.c: In function 'erofs_read_super':
drivers/staging/erofs/super.c:343:17: error: 'MS_RDONLY' undeclared (first use in this function); did you mean 'IS_RDONLY'?
sb->s_flags |= MS_RDONLY | MS_NOATIME;
^~~~~~~~~
IS_RDONLY
drivers/staging/erofs/super.c:343:17: note: each undeclared identifier is reported only once for each function it appears in
drivers/staging/erofs/super.c:343:29: error: 'MS_NOATIME' undeclared (first use in this function); did you mean 'S_NOATIME'?
sb->s_flags |= MS_RDONLY | MS_NOATIME;
^~~~~~~~~~
S_NOATIME
drivers/staging/erofs/super.c: In function 'erofs_mount':
drivers/staging/erofs/super.c:501:10: warning: passing argument 5 of 'mount_bdev' makes integer from pointer without a cast [-Wint-conversion]
&priv, erofs_fill_super);
^~~~~~~~~~~~~~~~
In file included from include/linux/buffer_head.h:12:0,
from drivers/staging/erofs/super.c:14:
include/linux/fs.h:2151:23: note: expected 'size_t {aka long unsigned int}' but argument is of type 'int (*)(struct super_block *, void *, int)'
extern struct dentry *mount_bdev(struct file_system_type *fs_type,
^~~~~~~~~~
drivers/staging/erofs/super.c:500:9: error: too few arguments to function 'mount_bdev'
return mount_bdev(fs_type, flags, dev_name,
^~~~~~~~~~
In file included from include/linux/buffer_head.h:12:0,
from drivers/staging/erofs/super.c:14:
include/linux/fs.h:2151:23: note: declared here
extern struct dentry *mount_bdev(struct file_system_type *fs_type,
^~~~~~~~~~
drivers/staging/erofs/super.c: At top level:
drivers/staging/erofs/super.c:518:20: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.mount = erofs_mount,
^~~~~~~~~~~
drivers/staging/erofs/super.c:518:20: note: (near initialization for 'erofs_fs_type.mount')
drivers/staging/erofs/super.c: In function 'erofs_remount':
drivers/staging/erofs/super.c:630:12: error: 'MS_RDONLY' undeclared (first use in this function); did you mean 'IS_RDONLY'?
*flags |= MS_RDONLY;
^~~~~~~~~
IS_RDONLY
drivers/staging/erofs/super.c: At top level:
drivers/staging/erofs/super.c:640:16: error: initialization from incompatible pointer type [-Werror=incompatible-pointer-types]
.remount_fs = erofs_remount,
^~~~~~~~~~~~~
Caused by various commits creating erofs in the staging tree interacting
with various commits redoing the mount infrastructure in the vfs tree.
I have disabed CONFIG_EROFS_FS for now:"
The reason of compiling error is:
Since -next collects and merges developing patches including common vfs
stuff from multi-trees, but those patches didn't cover erofs, such as:
('vfs: Suppress MS_* flag defs within the kernel unless explicitly enabled")
https://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git/commit/?h=for-next&id=109b45090d7d3ce2797bb1ef7f70eead5bfe0ff3
("vfs: Require specification of size of mount data for internal mounts")
https://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs.git/commit/?h=for-next&id=0a191e4505a4f255e6513b49426213da69bf0e80
Above vfs related patches has not been merged in staging tree, if we
submit those erofs patches to staging mailing list and after including
them in staging-{test,nexts} tree, it can easily cause compiling error.
We worked out some patches to adjust those vfs change, but now we just
submit them to -next tree temporarily to avoid compiling error.
For potentail conflict in between erofs and vfs changes in incoming
merge window, Stephen suggested that we can disable CONFIG_EROFS_FS
temporarily to pass merge window, and after that we can do restore by
reenabling CONFIG_EROFS_FS and applying those fixing patches. Also
Greg confirmed this solution.
So, let's disable compiling erofs for a while.
Suggested-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Reviewed-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
There is no need to '#if CONFIG_EROFS_FS_XATTR' in xattr.c,
let's remove it.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
There is a type mismatch in the definition of
Z_EROFS_VLE_VMAP_ONSTACK_PAGES, let's fix it.
Link: https://lists.01.org/pipermail/kbuild-all/2018-July/050707.html
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch adds a TODO to list the things to be done, and
the relevant info to MAINTAINERS so we can take all the blame :)
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch adds an optional choice which can be
enabled by users in order to cache both incomplete
ends of compressed clusters as a complement to
the in-place decompression in order to boost random
read, but it costs more memory than the in-place
decompression only.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch introduces the basic in-place VLE decompression
implementation for the erofs file system.
Compared with fixed-sized input compression, it implements
what we call 'the variable-length extent compression' which
specifies the same output size for each compression block
to make the full use of IO bandwidth (which means almost
all data from block device can be directly used for decomp-
ression), improve the real (rather than just via data caching,
which costs more memory) random read and keep the relatively
lower compression ratios (it saves more storage space than
fixed-sized input compression which is also configured with
the same input block size), as illustrated below:
|--- variable-length extent ---|------ VLE ------|--- VLE ---|
/> clusterofs /> clusterofs /> clusterofs /> clusterofs
++---|-------++-----------++---------|-++-----------++-|---------++-|
...|| | || || | || || | || | ... original data
++---|-------++-----------++---------|-++-----------++-|---------++-|
++->cluster<-++->cluster<-++->cluster<-++->cluster<-++->cluster<-++
size size size size size
\ / / /
\ / / /
\ / / /
++-----------++-----------++-----------++
... || || || || ... compressed clusters
++-----------++-----------++-----------++
++->cluster<-++->cluster<-++->cluster<-++
size size size
The main point of 'in-place' refers to the decompression mode:
Instead of allocating independent compressed pages and data
structures, it reuses the allocated file cache pages at most
to store its compressed data and the corresponding pagevec in
a time-sharing approach by default, which will be useful for
low memory scenario.
In the end, unlike the other filesystems with (de)compression
support using a relatively large compression block size, which
reads and decompresses >= 128KB at once, and gains a more
good-looking random read (In fact it collects small random reads
into large sequential reads and caches all decompressed data
in memory, but it is unacceptable especially for embedded devices
with limited memory, and it is not the real random read), we
select a universal small-sized 4KB compressed cluster, which is
the smallest page size for most architectures, and all compressed
clusters can be read and decompressed independently, which ensures
random read number for all use cases.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch introduces another concept used by the unzip
subsystem called 'workstation'. It can be seen as a sparse
array that stores pointers pointed to data structures
related to the corresponding physical blocks.
All lookup cases are protected by RCU read lock. Besides,
reference count and spin_lock are also introduced to
manage its lifetime and serialize all update operations.
'workstation' is currently implemented on the in-kernel
radix tree approach for backward compatibility.
With the evolution of linux kernel, it could be migrated
into XArray implementation in the future.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch adds a dedicated shrinker targeting to free unneeded
memory consumed by a number of erofs in-memory data structures.
Like F2FS and UBIFS, it also adds:
- sbi->umount_mutex to avoid races on shrinker and put_super
- sbi->shrinker_run_no to not revisit recently scaned objects
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
Currently, this patch only simply implements LZ4
decompressor due to its development priority.
In the future, erofs will support more compression
algorithm and format other than LZ4, thus a generic
decompressor interface will be needed.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
We have to reduce the memory cost as much as possible,
so we don't want to decompress more data beyond
the output buffer size, however "LZ4_decompress_safe_partial"
doesn't guarantee to stop at the arbitary end position,
but stop just after its current LZ4 "sequence" is completed.
Link: https://groups.google.com/forum/#!topic/lz4c/_3kkz5N6n00
Therefore, I hacked the LZ4 decompression logic by hand,
probably NOT the fastest approach, and hope for better
implementation.
Signed-off-by: Miao Xie <miaoxie@huawei.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch introduces an temporary _on-stack_ page
pool to reuse the freed page directly as much as
it can for better performance and release all pages
at a time, it also slightly reduces the possibility of
the potential memory allocation failure.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
This patch introduces an iterable L2P mapping
operation 'erofs_map_blocks_iter'.
Compared with 'erofs_map_blocks', it avoids
a number of redundant 'release and regrab'
processes if they request the same meta page.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
For each compressed cluster, there is a straight-forward
way of allocating a fixed or variable-sized (for VLE) array
to record the corresponding file pages for its decompression
if we decide to decompress these pages asynchronously (eg.
read-ahead case), however it could take much extra on-heap
memory compared with traditional uncompressed filesystems.
This patch introduces a pagevec solution to reuse some
allocated file page in the time-sharing approach storing
parts of the array itself in order to minimize the extra
memory overhead, thus only a constant and small-sized array
used for booting the whole array itself up will be needed.
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
Currently kernel has scattered tagged pointer usages hacked
by hand in plain code, without a unique and portable functionset
to highlight the tagged pointer itself and wrap these hacked code
in order to clean up all over meaningless magic masks.
Therefore, this patch introduces simple generic methods to fold
tags into a pointer integer. It currently supports the last n bits
of the pointer for tags, which can be selected by users.
In addition, it will also be used for the upcoming EROFS filesystem,
which heavily uses tagged pointer approach for high performance
and reducing extra memory allocation.
Link: https://en.wikipedia.org/wiki/Tagged_pointer
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Cyber Knight <cyberknight755@gmail.com>
Signed-off-by: azrim <mirzaspc@gmail.com>
