12 Oct, 2006
40 commits
-
Currently ioprio_best function first checks wethere aioprio or bioprio equals
IOPRIO_CLASS_NONE (ioprio_valid() macros does that) and if it is so it returns
bioprio/aioprio appropriately. Thus the next four lines, that set aclass/bclass
to IOPRIO_CLASS_BE, if aclass/bclass == IOPRIO_CLASS_NONE, are never executed.The second problem: if aioprio from class IOPRIO_CLASS_NONE and bioprio from
class IOPRIO_CLASS_IDLE are passed to ioprio_best function, it will return
IOPRIO_CLASS_IDLE. It means that during __make_request we can merge two
requests and set the priority of merged request to IDLE, while one of
the initial requests originates from a process with NONE (default) priority.
So we can get a situation when a process with default ioprio will experience
IO starvation, while there is no process from real-time class in the system.Just removing ioprio_valid check should correct situation.
Signed-off-by: Vasily Tarasov
Signed-off-by: Jens Axboe -
Don't jump to the unlock+release path, we already did that.
Signed-off-by: Jens Axboe
-
Signed-off-by: Al Viro
Signed-off-by: Linus Torvalds -
- Calculate a variable in bvec_alloc_bs() only once needed, not earlier
(bio.o down from 18408 to 18376 Bytes, 32 Bytes saved, probably due to
data locality improvements).- Init variable idx to silence a gcc warning which already existed in the
unmodified original base file (bvec_alloc_bs() handles idx correctly, so
there's no need for the warning):fs/bio.c: In function `bio_alloc_bioset':
fs/bio.c:169: warning: `idx' may be used uninitialized in this functionSigned-off-by: Andreas Mohr
Acked-by: Jens Axboe
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The attached patch destroys all the dentries attached to a superblock in one go
by:(1) Destroying the tree rooted at s_root.
(2) Destroying every entry in the anon list, one at a time.
(3) Each entry in the anon list has its subtree consumed from the leaves
inwards.This reduces the amount of work generic_shutdown_super() does, and avoids
iterating through the dentry_unused list.Note that locking is almost entirely absent in the shrink_dcache_for_umount*()
functions added by this patch. This is because:(1) at the point the filesystem calls generic_shutdown_super(), it is not
permitted to further touch the superblock's set of dentries, and nor may
it remove aliases from inodes;(2) the dcache memory shrinker now skips dentries that are being unmounted;
and(3) the superblock no longer has any external references through which the VFS
can reach it.Given these points, the only locking we need to do is when we remove dentries
from the unused list and the name hashes, which we do a directory's worth at a
time.We also don't need to guard against reference counts going to zero unexpectedly
and removing bits of the tree we're working on as nothing else can call dput().A cut down version of dentry_iput() has been folded into
shrink_dcache_for_umount_subtree() function. Apart from not needing to unlock
things, it also doesn't need to check for inotify watches.In this version of the patch, the complaint about a dentry still being in use
has been expanded from a single BUG_ON() and now gives much more information.Signed-off-by: David Howells
Acked-by: NeilBrown
Acked-by: Ian Kent
Cc: Trond Myklebust
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Make sure all dentries refs are released before calling kill_anon_super() so
that the assumption that generic_shutdown_super() can completely destroy the
dentry tree for there will be no external references holds true.What was being done in the put_super() superblock op, is now done in the
kill_sb() filesystem op instead, prior to calling kill_anon_super().This makes the struct autofs_sb_info::root member variable redundant (since
sb->s_root is still available), and so that is removed. The calls to
shrink_dcache_sb() are also removed since they're also redundant as
shrink_dcache_for_umount() will now be called after the cleanup routine.Signed-off-by: David Howells
Acked-by: Ian Kent
Cc: Trond Myklebust
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Make sure all dentries refs are released before calling kill_block_super()
so that the assumption that generic_shutdown_super() can completely destroy
the dentry tree for there will be no external references holds true.What was being done in the put_super() superblock op, is now done in the
kill_sb() filesystem op instead, prior to calling kill_block_super().Changes made in [try #2]:
(*) reiserfs_kill_sb() now checks that the superblock FS info pointer is set
before trying to dereference it.Signed-off-by: David Howells
Cc: "Rafael J. Wysocki"
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Alexey Dobriyan
Cc: David Woodhouse
Cc: David Howells
Cc: Mark Fasheh
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
A couple of flush_dcache_page()s are missing on the I/O-error paths.
Cc: "David S. Miller"
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Aince all callers dereference sb, and this function does so earlier too, we
dont need the check.Signed-off-by: Eric Sesterhenn
Acked-by: OGAWA Hirofumi
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
A couple of HDIO IOCTLs are not yet handled and a few others are marked
as using a pointer rather than an unsigned long. The formers include:HDIO_GET_WCACHE, HDIO_GET_ACOUSTIC, HDIO_GET_ADDRESS and
HDIO_GET_BUSSTATE. The latters are: HDIO_SET_MULTCOUNT,
HDIO_SET_UNMASKINTR, HDIO_SET_KEEPSETTINGS, HDIO_SET_32BIT,
HDIO_SET_NOWERR, HDIO_SET_DMA, HDIO_SET_PIO_MODE and HDIO_SET_NICE.Additionally 0x330 used to be HDIO_GETGEO_BIG and may be issued by 32-bit
`hdparm' run on a 64-bit kernel making Linux complain loudly.This is a fix for these issues.
Signed-off-by: Maciej W. Rozycki
Cc: Alan Cox
Acked-by: Arnd Bergmann
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Current error behaviour for ext2 and ext3 filesystems does not fully
correspond to the documentation and should be fixed.According to man 8 mount, ext2 and ext3 file systems allow to set one of 3
different on-errors behaviours:---- start of quote man 8 mount ----
errors=continue / errors=remount-ro / errors=panic
Define the behaviour when an error is encountered. (Either ignore
errors and just mark the file system erroneous and continue, or remount
the file system read-only, or panic and halt the system.) The default is
set in the filesystem superblock, and can be changed using tune2fs(8).---- end of quote ----
However EXT3_ERRORS_CONTINUE is not read from the superblock, and thus
ERRORS_CONT is not saved on the sbi->s_mount_opt. It leads to the incorrect
handle of errors on ext3.Then we've checked corresponding code in ext2 and discovered that it is buggy
as well:- EXT2_ERRORS_CONTINUE is not read from the superblock (the same);
- parse_option() does not clean the alternative values and thus something
like (ERRORS_CONT|ERRORS_RO) can be set;- if options are omitted, parse_option() does not set any of these options.
Therefore it is possible to set any combination of these options on the ext2:
- none of them may be set: EXT2_ERRORS_CONTINUE on superblock / empty mount
options;- any of them may be set using mount options;
- 2 any options may be set: by using EXT2_ERRORS_RO/EXT2_ERRORS_PANIC on the
superblock and other value in mount options;- and finally all three options may be set by adding third option in remount.
Currently ext2 uses these values only in ext2_error() and it is not leading to
any noticeable troubles. However somebody may be discouraged when he will try
to workaround EXT2_ERRORS_PANIC on the superblock by using errors=continue in
mount options.This patch:
EXT2_ERRORS_CONTINUE should be read from the superblock as default value for
error behaviour. parse_option() should clean the alternative options and
should not change default value taken from the superblock.Signed-off-by: Vasily Averin
Acked-by: Kirill Korotaev
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Current error behaviour for ext2 and ext3 filesystems does not fully
correspond to the documentation and should be fixed.According to man 8 mount, ext2 and ext3 file systems allow to set one of 3
different on-errors behaviours:---- start of quote man 8 mount ----
errors=continue / errors=remount-ro / errors=panic
Define the behaviour when an error is encountered. (Either ignore
errors and just mark the file system erroneous and continue, or remount
the file system read-only, or panic and halt the system.) The default is
set in the filesystem superblock, and can be changed using tune2fs(8).---- end of quote ----
However EXT3_ERRORS_CONTINUE is not read from the superblock, and thus
ERRORS_CONT is not saved on the sbi->s_mount_opt. It leads to the incorrect
handle of errors on ext3.Then we've checked corresponding code in ext2 and discovered that it is buggy
as well:- EXT2_ERRORS_CONTINUE is not read from the superblock (the same);
- parse_option() does not clean the alternative values and thus something
like (ERRORS_CONT|ERRORS_RO) can be set;- if options are omitted, parse_option() does not set any of these options.
Therefore it is possible to set any combination of these options on the ext2:
- none of them may be set: EXT2_ERRORS_CONTINUE on superblock / empty mount
options;- any of them may be set using mount options;
- 2 any options may be set: by using EXT2_ERRORS_RO/EXT2_ERRORS_PANIC on the
superblock and other value in mount options;- and finally all three options may be set by adding third option in remount.
Currently ext2 uses these values only in ext2_error() and it is not leading to
any noticeable troubles. However somebody may be discouraged when he will try
to workaround EXT2_ERRORS_PANIC on the superblock by using errors=continue in
mount options.This patch:
EXT3_ERRORS_CONTINUE should be taken from the superblock as default value for
error behaviour.Signed-off-by: Dmitry Mishin
Acked-by: Vasily Averin
Acked-by: Kirill Korotaev
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
If grow_buffers() is for some reason passed a block number which wants to lie
outside the maximum-addressable pagecache range (PAGE_SIZE * 4G bytes) then it
will accidentally truncate `index' and will then instnatiate a page at the
wrong pagecache offset. This causes __getblk_slow() to go into an infinite
loop.This can happen with corrupted disks, or with software errors elsewhere.
Detect that, and handle it.
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Implement the epoll_pwait system call, that extend the event wait mechanism
with the same logic ppoll and pselect do. The definition of epoll_pwait
is:int epoll_pwait(int epfd, struct epoll_event *events, int maxevents,
int timeout, const sigset_t *sigmask, size_t sigsetsize);The difference between the vanilla epoll_wait and epoll_pwait is that the
latter allows the caller to specify a signal mask to be set while waiting
for events. Hence epoll_pwait will wait until either one monitored event,
or an unmasked signal happen. If sigmask is NULL, the epoll_pwait system
call will act exactly like epoll_wait. For the POSIX definition of
pselect, information is available here:http://www.opengroup.org/onlinepubs/009695399/functions/select.html
Signed-off-by: Davide Libenzi
Cc: David Woodhouse
Cc: Andi Kleen
Cc: Michael Kerrisk
Cc: Ulrich Drepper
Cc: Roland McGrath
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Someone's tab key is emitting spaces. Attempt to repair some of the damage.
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Current error behaviour for ext2 and ext3 filesystems does not fully
correspond to the documentation and should be fixed.According to man 8 mount, ext2 and ext3 file systems allow to set one of 3
different on-errors behaviours:---- start of quote man 8 mount ----
errors=continue / errors=remount-ro / errors=panic
Define the behaviour when an error is encountered. (Either ignore
errors and just mark the file system erroneous and continue, or remount
the file system read-only, or panic and halt the system.) The default is
set in the filesystem superblock, and can be changed using tune2fs(8).---- end of quote ----
However EXT3_ERRORS_CONTINUE is not read from the superblock, and thus
ERRORS_CONT is not saved on the sbi->s_mount_opt. It leads to the incorrect
handle of errors on ext3.Then we've checked corresponding code in ext2 and discovered that it is buggy
as well:- EXT2_ERRORS_CONTINUE is not read from the superblock (the same);
- parse_option() does not clean the alternative values and thus something
like (ERRORS_CONT|ERRORS_RO) can be set;- if options are omitted, parse_option() does not set any of these options.
Therefore it is possible to set any combination of these options on the ext2:
- none of them may be set: EXT2_ERRORS_CONTINUE on superblock / empty mount
options;- any of them may be set using mount options;
- 2 any options may be set: by using EXT2_ERRORS_RO/EXT2_ERRORS_PANIC on the
superblock and other value in mount options;- and finally all three options may be set by adding third option in remount.
Currently ext2 uses these values only in ext2_error() and it is not leading to
any noticeable troubles. However somebody may be discouraged when he will try
to workaround EXT2_ERRORS_PANIC on the superblock by using errors=continue in
mount options.This patch:
EXT4_ERRORS_CONTINUE should be taken from the superblock as default value for
error behaviour.Signed-off-by: Dmitry Mishin
Acked-by: Vasily Averin
Acked-by: Kirill Korotaev
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
I assume this means "logical sb block". So call it that.
I still don't understand the name though. A block is a block. What's
different about this one?Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
With CONFIG_LBD=n, sector_div() expands to a plain old divide. But ext4 is
_not_ passing in a sector_t as the first argument, so...fs/built-in.o: In function `ext4_get_group_no_and_offset':
fs/ext4/balloc.c:39: undefined reference to `__umoddi3'
fs/ext4/balloc.c:41: undefined reference to `__udivdi3'
fs/built-in.o: In function `find_group_orlov':
fs/ext4/ialloc.c:278: undefined reference to `__udivdi3'
fs/built-in.o: In function `ext4_fill_super':
fs/ext4/super.c:1488: undefined reference to `__udivdi3'
fs/ext4/super.c:1488: undefined reference to `__umoddi3'
fs/ext4/super.c:1594: undefined reference to `__udivdi3'
fs/ext4/super.c:1601: undefined reference to `__umoddi3'Fix that up by calling do_div() directly.
Also cast the arg to u64. do_div() is only defined on u64, and ext4_fsblk_t
is supposed to be opaque.Note especially the changes to find_group_orlov(). It was attempting to do
do_div(int, unsigned long long);
which is royally screwed up. Switched it to plain old divide.
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Way too big to inline.
Cc:
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
move '_hi' bits of block numbers in the larger part of the
block group descriptor structureSigned-off-by: Alexandre Ratchov
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
make block group descriptor larger.
Signed-off-by: Alexandre Ratchov
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Similar to ext4, change blocks in JBD2 from sector_t to unsigned long long.
Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Previously when in-kernel ext4 block type is sector_t, it's only 4 bits long
under some 32bit arch (when CONFIG_LBD is not on). So we need to check the
size of sector_t before we read 48bit long on-disk blocks to in-kernel blocks.These checks are unnecessary now as we changed the in-kernel blocks to
unsigned longlong.Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Change ext4 in-kernel block type (ext4_fsblk_t) from sector_t to unsigned
long long. Remove ext4 block type string micro E3FSBLK, replaced with "%llu"[akpm@osdl.org: build fix]
Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
In-kernel super block changes to support >32 bit free blocks numbers.
Signed-off-by: Laurent Vivier
Signed-off-by: Dave Kleikamp
Signed-off-by: Alexandre Ratchov
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
As we are planning to support 48-bit block numbers for ext4, we need to
support 48-bit block numbers for extended attributes. In the short term, we
can do this by reuse (on-disk) 16-bit padding (linux2.i_pad1 currently used
only by "hurd") as high order bits for xattr. This patch basically does that.Signed-off-by: Badari Pulavarty
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
JBD layer in-kernel block varibles type fixes to support >32 bit block number
and convert to sector_t type.Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Here is the patch to JBD to handle 64 bit block numbers, originally from Zach
Brown. This patch is useful only after adding support for 64-bit block
numbers in the filesystem.Signed-off-by: Badari Pulavarty
Signed-off-by: Zach Brown
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Randy Dunlap
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Make it possible to add file preallocation support in future as an RO_COMPAT
feature by recognizing uninitialized extents as holes and limiting extent
length to keep the top bit of ee_len free for marking uninitialized extents.Signed-off-by: Suparna Bhattacharya
Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Alex Tomas
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Redefine ext3 in-kernel filesystem block type (ext3_fsblk_t) from unsigned
long to sector_t, to allow kernel to handle >32 bit ext3 blocks.Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
On disk extents format:
/*
* this is extent on-disk structure
* it's used at the bottom of the tree
*/
struct ext3_extent {
__le32 ee_block; /* first logical block extent covers */
__le16 ee_len; /* number of blocks covered by extent */
__le16 ee_start_hi; /* high 16 bits of physical block */
__le32 ee_start; /* low 32 bigs of physical block */
};Signed-off-by: Alex Tomas
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Reworked from a patch by Mingming Cao and Randy Dunlap
Signed-off-By: Randy Dunlap
Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
jbd and jbd2 currently use the same slab names which must be unique. The
patch below just renames jbd2's slabs.Signed-off-by: Johann Lombardi
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Mingming Cao originally did this work, and Shaggy reproduced it using some
scripts from her.Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This is a simple copy of the files in fs/jbd to fs/jbd2 and
/usr/incude/linux/[ext4_]jbd.h to /usr/include/[ext4_]jbd2.hSigned-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Originally part of a patch from Mingming Cao and Randy Dunlap. Reorganized
by Shaggy.Signed-off-by: Randy Dunlap
Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Mingming Cao originally did this work, and Shaggy reproduced it using some
scripts from her.Signed-off-by: Mingming Cao
Signed-off-by: Dave Kleikamp
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
