| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
With outstanding BCBs, cache manager will try to grab BCB
resource before doing the actual flushing, thus cuase hang
|
|
|
|
New buffer-head created during flushing, thus deadlock
|
|
|
|
merged and optimized upon ngkaho123's implmentation
|
|
|
|
to avoid further CcZeroData in Ext2Cleanup
|
|
|
|
Layered fs like Dekcore or lfs could send only paging writing.
|
|
|
|
|
|
|
|
|
|
casued by non-cached reading, read length should be cut by
file size
FIXME 2): WDK sources compiling random failures
caused by ValidDataLength update and pended writing irps
<possibly casued by garbage data in uwritten area. More
research is required to see the root reason>
|
|
1) zero data of pre-allocated blocks (by IRP_MJ_SET_INFO)
2) deallocate of pre-allocated block (by IRP_MJ_CREATE)
3) call CcZeroData in wait state when user can wait
4) FIXME: VDL set to wrong value in request SetEndOfFileInfo
5) fast i/o won't touch VDL, make i/o mgr create IRP request
6) reconstruct writing path (Ext2WriteFile)
|
|
|
|
|
|
MSB of ee_len field for unwritten extent is just a flag, but 0.60 will
treat it as block number, then it might write data to blocks that might
belong any other files or file system's metadata, thus lead to silent
data corruption
|
|
|
|
|
|
|
|
1, Ext4 extent fully supported.
a) Merge ngkaho1234's patch of ext4 extent support.
Thanks to ngkaho1234@gmail.com, see his repository at
http://github.com/ngkaho1234/Ext3Fsd.git
b) i/o path revised for indirect block and extent
2, File deletion optimization
3, Bug fixings ...
|
|
|
|
1, Feature: Windows 8 supported
2, Feature: Force-writing supported
3, FIXME: data loss with TeraCopy when moving directories
4, FIXME: ATTO Disk Benchnmark fails with direct i/o tests
5, FIXME: files become invisible with empty hidden pattens
6, FIXME: possible hang with concurrent access attempts
upon the same file
7, FIXME: BSOD due to pagable code sections
|
|
writing on XP system
2, FIXME: Set ValidDataLength only for FileEndOfFileInformation
w/ AdvanceOnly
3, Support writing to ext4 volumes w/ flex_bg
4, Code cleanup and enhancement for global/volume property settings
|
|
1, FIXME: Possible memory leak for Mcbs of symlinks
2, IMPROVEMENT: don't hold global lock when deleting a file
|
|
1, 0.50 pre-release: release notes and readme.txt & versions update
2, FIXME: DDK compiling failure: build.exe reports obj files are
invalid. caused by fastio, it doesn't update i_size.
|
|
1, uninit_bg (GDT_CSUM)
2, huge_file
3, win7: zero-size reported in cmd
4, symlink open/read/deletiona
|
|
1, htree directory index support
2, memory allocation and check
3, buffer head implemented over cache pages
4, inode optimization, not whole inode needed
|
|
Ext4 extents support, by Bo Branten
|
|
|
