I really don't like Bugzilla. Bugzilla has a tendency of accumulating useless information, as it is impossible to delete anything from it. Reading Bugzilla entries is like reading a book that still contains every deleted passage, editor comment, etc.
So this is the current state of known bugs in LogFS. If you find a new bug that is not reported in this list, please add it. You can create a dedicated page for each bug, if you have to add a lot of information to it.
Current list of known bugs is short. Jörn tries hard to keep it short by fixing bugs. You should try hard to make it longer by detecting and reporting new problems. 
Contents
Known Bugs
LogFS as a root filessystem causes problems
When used as a root filesystem, the system does not boot successfully. This is weird, I have no idea what is failing. Any tips how one can debug this problem are appreciated.
Rare garbage collection failure
Can be triggered by filling the filesystem 100%, then randomly replacing data for several minutes. GC can checks how many free segments are available and backs off to an easier target when the number gets too low. However, between two checks more than one free segment can be consumed, violating one of the GC invariants. Will require a bit of surgery to fix this properly.
Warning when copying large amount of data to a freshly formatted logfs partition
Was triggered with a freshly formatted logfs partition (around 2 GB). When copying data from an existing ext3 partition with rsync, the following warning was logged to /var/log/messages:
May 10 04:00:06 debian kernel:
[ cut here ]
May 10 04:00:06 debian kernel: WARNING: at fs/logfs/readwrite.c:569 logfs_seek_hole+0xc3/0x1d6() May 10 04:00:06 debian kernel: Modules linked in: loop pcnet32 intel_agp agpgart May 10 04:00:06 debian kernel: Pid: 2790, comm: rsync Not tainted 2.6.25.4 #2 May 10 04:00:06 debian kernel: [<c011edd3>] warn_on_slowpath+0x40/0x4f May 10 04:00:06 debian kernel: [<c013527e>] clockevents_program_event+0xc4/0xd2 May 10 04:00:06 debian kernel: [<c01318b8>] hrtimer_run_pending+0x20/0xd0 May 10 04:00:06 debian kernel: [<c0407bcd>] _spin_lock_irq+0xe/0x24 May 10 04:00:06 debian kernel: [<c0407e61>] _spin_unlock_irq+0xd/0x20 May 10 04:00:06 debian kernel: [<c01263ae>] run_timer_softirq+0x16a/0x171 May 10 04:00:06 debian kernel: [<c0123018>] do_softirq+0x7b/0xbf May 10 04:00:07 debian kernel: [<c0407ea4>] _read_unlock_irq+0xe/0x21 May 10 04:00:07 debian kernel: [<c0157467>] find_lock_page+0x6d/0x73 May 10 04:00:07 debian kernel: [<c01feacd>] seek_holedata_loop+0x161/0x16d May 10 04:00:07 debian kernel: [<c0200ad3>] logfs_seek_hole+0xc3/0x1d6 May 10 04:00:07 debian kernel: [<c01fa63c>] logfs_write_dir+0x8d/0xaf May 10 04:00:07 debian kernel: [<c01fabce>] logfs_create+0x10b/0x17a May 10 04:00:07 debian kernel: [<c01fb0ea>] logfs_mknod+0x43/0x51 May 10 04:00:07 debian kernel: [<c01fb0a7>] logfs_mknod+0x0/0x51 May 10 04:00:07 debian kernel: [<c017af4a>] vfs_mknod+0x80/0xe1 May 10 04:00:07 debian kernel: [<c017ced9>] sys_mknodat+0x12a/0x161 May 10 04:00:07 debian kernel: [<c0177121>] sys_lstat64+0xf/0x23 May 10 04:00:07 debian kernel: [<c018e4d1>] sys_futimesat+0x71/0x81 May 10 04:00:07 debian kernel: [<c017cf23>] sys_mknod+0x13/0x17 May 10 04:00:08 debian kernel: [<c0103b6a>] sysenter_past_esp+0x5f/0x85 May 10 04:00:08 debian kernel: ======================= May 10 04:00:08 debian kernel: ---[ end trace ca143223eefdc828 ]---
Command used to copy the data was: rsync -avz /media/sdc1/ /media/sdb2 /media/sdb2 was the fresh logfs partition, sdc1 contained the data to copy.
Kernel 2.6.25.4 logfs.patch.773
The copy process hangs after 288 MB, with 99% IO wait (for almost 2 hours now).