similar to: RE: [Ext2-devel] Re: ext3 htree brelse problems look to be fixed!

I just booted 2.5-bk current as of last night with the below patch¹ (which was recently posted to ext3-users) that un-static-ifies a struct dx_frame in namei.c. I then did my best torture test for the brelse bug: starting gnus (3600+ nnmh folders² with a total of XXX messages; it does a readdir on each of those folders) while doing bk consistancy checks in 2.5 and/or 2.4 kernel trees. All

hm, this backtrace actually has info... btw, I think we need this: --- 25/fs/ext3/namei.c~ext3-brelse-fix Sun Dec 29 06:53:39 2002 +++ 25-akpm/fs/ext3/namei.c Sun Dec 29 06:53:54 2002 @@ -576,8 +576,10 @@ int ext3_htree_fill_tree(struct file *di (hinfo.minor_hash < start_minor_hash))) continue; if ((err = ext3_htree_store_dirent(dir_file, - hinfo.hash,

I've backported all of the bugfixes to the 2.5 dxdir/htree patches to 2.4, and have created a new set of patches for Linux 2.4.21rc5. At this point it *looks* like we've fixed all of the htree bugs that people have reported, including the brelse bug, the memory leak bugs, and the NFS compatibility problems. I've done *very* light testing, and things seem to work, but I'm now

I was flipping through the code recently and noticed that we still have the double whammy of allocating dir entry positions with parent_dir->counter++ and that weird setting of f_pos to 2^31-1. So after enough creates (and deletes :)) in a directory we end up with an entry item whose key is past that value. f_pos gets rewound instead of being set to that magical EOF. readdir() gets stuck

Two patches follow this message. One fixes the default implementation of SEEK_HOLE/DATA. This patch applies atop Josef's last posted patch. The second patch implements the same on ocfs2. The test tool for the same is available here. http://oss.oracle.com/~smushran/seek_data/seek_test.c It is improved since the last post. It runs cleanly on zfs, ocfs2 and ext3 (default behavior). Users

Two patches follow this message. One fixes the default implementation of SEEK_HOLE/DATA. This patch applies atop Josef's last posted patch. The second patch implements the same on ocfs2. The test tool for the same is available here. http://oss.oracle.com/~smushran/seek_data/seek_test.c It is improved since the last post. It runs cleanly on zfs, ocfs2 and ext3 (default behavior). Users

Over the last two days we've been seeing a fair bit of this: ---- # ls -laR > /dev/null ... ls: ./server2/b/user/bxyz/392.: Input/output error ---- This is with the latest htree patches applied to 2.4.19, and latest e2fsprogs-test, on a dual AMD system, with 5x73GB SCSI drives on a MegaRAID controller. We're using the gcc 2.96 that comes with RH7.3. esfsck shows "Inodes that

Hello When using readdir() on a directory with many files or long file names it can happen that it returns the same file name twice. Attached is a program that demonstrates this. I have traced this problem back to linux-2.6.10-rc1-bk18 and all kernels after this one are effected. linux-2.6.10-rc1-bk17 is still okay. If I reverse the following patch in linux-2.6.10-rc1-bk18, readdir() works again

Since the d_off in the first dirent for "." (that originates from the 4th argument "offset" of filldir() for the 2nd dirent for "..") is wrongly assigned in btrfs_real_readdir(), telldir returns same offset for different locations. | # mkfs.btrfs /dev/sdb1 | # mount /dev/sdb1 fs0 | # cd fs0 | # touch file0 file1 | # ../test | telldir: 0 | readdir: d_off = 2,

Hello Dan Johnson, hello list, It's really weird that there's no working command line player for FLAC files. Dan Johnson created flac123 (licensed under the GPL) to fill that gap, and started the flac-tools project. Unfortunately, flac123 doesn't work anymore with the latest FLAC, and the flac-tools project has been idle for more than a year. I created a patch that makes flac123

In ocfs2_find_entry_dx(), if ocfs2_read_inode_block() failed, both di_bh and dx_root_bh are both released. In this case, for dx_root_bh, it's unnecessary (even buggy if refcount of dx_root_bh is non-zero). This patch fixes this issue by change brelse order of di_bh and dx_root_bh. Signed-off-by: Coly Li <coly.li at suse.de> --- fs/ocfs2/dir.c | 5 +++-- 1 files changed, 3

I got this error in my log reports this morning, from the machine I use as my firewall. This is the first time it has occurred, but the machine is running a very new kernel: Linux fendrian.rimspace.net 2.6.0-test11-fendrian #1 Wed Dec 10 22:25:59 EST 2003 i486 GNU/Linux The kernel was up to date as per the CVS repository at that point. This was just before the CDROM_SEND_PACKET IOCTL fix went

Hello, When drop_progress isn't zero, the root->node's usage count is increased in btrfs_search_slot. Therefore, the get_bh in the body of while loop is redundant in most cases. (this change is in accordance with btrfs_defrag_leaves). The second change is decrease root->node's usage count when drop a snapshot. Regards YZ diff -r 29b8cc7794ac extent-tree.c --- a/extent-tree.c

I decided that I would try out 2.5.44, and I noticed that htree was merged. If I don't do the tune2fs -O dir_index, and e2fsck -D, the (exisintg) fs won't use htree, right? Once I do the tune2fs and e2fsck, will I still be able to go back to a non-htree kernel if needed? (Will a htree-ized fs work on a non-htree kernel?) I'm guessing that it won't. I've seen a 2.4 htree

Apologies for the newbie question: I have a (stock) 2.4.20 build (*not* -ac), and I'm trying to work with large ext3 directories. By large, I mean 160,000 files per directory. (Yes, I know it would be better in nested directories but such is life). I feel htree would benefit me. Having upgraded from an earlier version of 2.4, I don't see any change, and close reading of the 2.4 changelog

Well here is what I got from testing htree with NFS. http://labs.zianet.com/benchmarks_html/postmark_benchmarks_NFS_100_htree.html It looks like htree improves performance when the storage device is local(as per my previous post) but it looks like htree degrades performance when used in conjunction with NFS. I am going to try it with gigabit ethernet when I get the time to see if maybe I am

Apologies for the newbie question: I have a (stock) 2.4.20 build (*not* -ac), and I'm trying to work with large ext3 directories. By large, I mean 160,000 files per directory. (Yes, I know it would be better in nested directories but such is life). I feel htree would benefit me. Close reading of the 2.4 changelog suggests that htree isn't in there - only a patch to prevent non-htree

Hello, Could someone create some newer htree patches for the 2.4.x kernel? There are some bug fixes in the 2.4.21-rc releases that would be nice to have but the htree patches preceed those and they won't apply cleanly to later pre/rc kernels. It sounds like Marcelo may be about ready to put out 2.4.21 so it may just be worth holding off until that happens, but I don't suspect much will

Guys, I have recently applied the latest 2.4 htree patch on a heavily loaded nfs server. The nfs server serves around four very busy clients that deliver email in maildir format and pop3/imap clients. Being maildir I presumed that the htree patch would improve performance - but I was wrong. Load on the server went up by around 25-40%. After 3-4 hours of heavy use the clients load went up to

Hi Guys, I am planning on upgrading an existing NFS exported filesystem to the ext3 htree patch on kernel v2.4.23. Will the patch index directories automatically after tune2fs -O dir_index or should I do an e2fsck -Dfy on the filesystem before remounting?