Ext2/3/4 manages generation numbers differently than UFS so adopt
some rules that should work well. When allocating a new inode,
make sure we generate a "good" random value specifically avoiding
zero.
Don't interfere with the numbers that are already generated in
the filesystem: ext2fs doesn't have the backwards compatibility
issues where there were no generation numbers.
Reviewed by: kevlo
MFC after: 1 week
h_levels_num, as most data structs in ext2fs, is unsigned so
the index that addresses it has to be unsigned as well.
To get to overflow here we would probably be considering a
degenerate case though.
MFC after: 5 days
rounddown2 tends to produce longer lines than the original code
and when the code has a high indentation level it was not really
advantageous to do the replacement.
This tries to strike a balance between readability using the macros
and flexibility of having the expressions, so not everything is
converted.
for limiting disk (actually filesystem) IO.
Note that in some cases these limits are not quite precise. It's ok,
as long as it's within some reasonable bounds.
Testing - and review of the code, in particular the VFS and VM parts - is
very welcome.
MFC after: 1 month
Relnotes: yes
Sponsored by: The FreeBSD Foundation
Differential Revision: https://reviews.freebsd.org/D5080
This adopts the same change as r291936 for UFS.
Directly clear IN_ACCESS or IN_UPDATE when user supplied the time, and
copy the value into the inode.
This keeps the behaviour cleaner and is consistent with UFS.
Reviewed by: bde
MFC after: 1 month (only 10)
Sync with r84642 from UFS:
The panics are inappropriate because the IN_RENAME flag only fixes a
few of the huge number of race conditions that can result in the
source path becoming invalid even prior to the VOP_RENAME() call.
Found accidentally while checking an issue from PVS Static Analysis.
MFC after: 3 days
This is ongoing work from Damjan Jovanovic to improve ext4 read support
with sparse files:
Keep track of the first and last block in each extent as it descends down
the extent tree, thus being able to work out that some blocks are sparse
earlier. This solves an issue on r293680.
In ext4_bmapext() start supporting the runb parameter, which appears to be
the number of adjacent blocks prior to the block being converted in the
same way that runp is the number of blocks after, speding up random access
to mmaped files.
PR: 206652
ext2fs: passthrough any extra timestamps to the dinode struct.
While it passed the classic testing, the change appears to have
caused some regression and still requires some more precautions.
PR: 206820
MFC after: 3 days
In general we don't trust any of the extended timestamps unless the
EXT2F_ROCOMPAT_EXTRA_ISIZE feature is set. However, in the case where
we freshly allocated a new inode the information is valid and it is
better to pass it along instead of leaving the value undefined.
This should have no practical effect but should reduce the amount of
garbage if EXT2F_ROCOMPAT_EXTRA_ISIZE is set, like in cases where the
filesystem is converted from ext3 to ext4.
MFC after: 4 days
Directory index was introduced in ext3. We don't always use the
prefix to denote the ext2 variant they belong to but when we
do we should try to be accurate.
We use i_flag to carry some flags like IN_E4INDEX which newer
ext2fs variants uses internally.
fsck.ext3 rightfully complains after our implementation tags
non-directory inodes with INDEX_FL.
Initializing i_flag during allocation removes the noise factor
and quiets down fsck.
Patch from: Damjan Jovanovic
PR: 206530
The htree dir_index is perhaps one of the most characteristic
features of the linux ext3 implementation. It was removed
in r281670, due to repeated bug reports.
Damjan Jovanic detected and fixed three bugs and did some
stress testing by building Apache OpenOffice on top of it
so it is now in good shape to bring back.
Differential Revision: https://reviews.freebsd.org/D5007
Submitted by: Damjan Jovanovic
Reviewed by: pfg
Tested by: pho
Relnotes: Yes
MFC after: 2 months (only 10.x)
Add support for sparse files in ext4. Also implement read-ahead, which
greatly increases the performance when transferring files from ext4.
Both features implemented by Damjan Jovanovic.
PR: 205816
MFC after: 1 week
Merge the filesystem specific part from r274914 to ext2fs.
I only did regular testing with the change but UFS and our ext2fs
are similar enough that the code should just work with the new
sendfile.
Discussed with: glebius
The htree directory index is a highly desirable feature for research
purposes and was meant to improve performance in our ext2/3 driver.
Unfortunately our implementation has two problems:
- It never really delivered any performance improvement.
- It appears to corrupt the filesystem in undetermined circumstances.
Strictly speaking dir_index is not required for read/write support in
ext2/3 and our limited ext4 support still works fine without it.
Regain stability in the ext2 driver by removing it. We may need it back
(fixed) if we want to support encrypted ext4 support but thanks to the
wonders of version control we can always revert this change and bring it
back.
PR: 191895
PR: 198731
PR: 199309
MFC after: 5 days
use VOP_FSYNC() to perform the NFS server's Commit operation.
This patch adds a mnt_kern_flag called MNTK_USES_BCACHE which
is set by file systems that use the buffer cache. If this flag
is not set, the NFS server always does a VOP_FSYNC().
This should be ok for old file system modules that do not set
MNTK_USES_BCACHE, since calling VOP_FSYNC() is correct, although
it might not be optimal for file systems that use the buffer cache.
Reviewed by: kib
MFC after: 2 weeks
The e2fs_gd struct was not being initialized and garbage was
being used for hinting the ext2 allocator variant.
Use malloc to clear the values and also initialize e2fs_contigdirs
during allocation to keep consistency.
While here clean up small style issues.
Reported by: Clang static analyser
MFC after: 1 week
After the ext2 variant of the "orlov allocator" was implemented,
the case for a negative or zero dirsize disappeared.
Drop the dead code and unsign dirsize given that it can't be
negative anyways.
CID: 1008669
MFC after: 1 week
into namecache, to avoid cache trashing when doing large operations.
E.g., tar archive extraction is not usually followed by access to many
of the files created.
Right now, each VOP_LOOKUP() implementation explicitely knowns about
this quirk and tests for both MAKEENTRY flag presence and op != CREATE
to make the call to cache_enter(). Centralize the handling of the
quirk into VFS, by deciding to cache only by MAKEENTRY flag in VOP.
VFS now sets NOCACHE flag for CREATE namei() calls.
Note that the change in semantic is backward-compatible and could be
merged to the stable branch, and is compatible with non-changed
third-party filesystems which correctly handle MAKEENTRY.
Suggested by: Chris Torek <torek@pi-coral.com>
Reviewed by: mckusick
Tested by: pho
Sponsored by: The FreeBSD Foundation
MFC after: 2 weeks
Overrunning buffer pointed to by (caddr_t)&oip->i_db[0] of 48 bytes by
passing it to a function which accesses it at byte offset 59 using
argument 60UL.
The issue was inherited from an older FFS implementation and
fixed there with by merging UFS2 in r98542. We follow the
FFS fix.
Discussed with: bde
CID: 1007665
MFC after: 3 days
ext2_print_inode is not really used but it was nice to
have for initial development work. #ifdef it under a
new EXT2FS_DEBUG knob so that we don't spend time
compiling it.
MFC after: 3 days
survives remount in rw, also it is set for vnodes on rootfs before
noatime can be set or clock is adjusted. All conditions result in
wrong atime for accessed vnodes.
Submitted by: bde
MFC after: 1 week
by ffs and ext2fs. Remove duplicated call to vm_page_zero_invalid(),
done by VOP and by vm_pager_getpages(). Use vm_pager_free_nonreq().
Reviewed by: alc (previous version)
Sponsored by: The FreeBSD Foundation
MFC after: 6 weeks (after r271596)
