[bfs] always panics, removing "syslog.old" (can't boot up any more)

[ttcoder]

This is my PM partition, which I've been keeping up to date with "pkgman update" for a couple weeks now.

It panics (after the last icon, as the desktop appears) every time I try to boot up, so quite a serious condition.

If that matters, it first occured after rebooting from a session on the other (r 45824) partition where I was running tests with the cdda driver re. its suspected memory-corruption/use-after-delete bug... Speculation: maybe the cdda driver corrupted an area of memory that is "shared" between reboots? Still, the BFS code seemingly has a vulnerability that should be fixed..

I'll keep that PM partition as-is until > next week in case I am advised to run diagnosing KDL commands ..etc, then I'll try to take corrective actions from the other partition, 'remotely' removing the offending syslog.old file and so on so I can get that PM partition back in play.

[ttcoder]

syslog daemon failed Assert(), removing file syslog.old

[ttcoder]

I'm about to remotely delete the problematic file but I've first run this:

~/Desktop> checkfs --check-only /SamCharlie/
Tracker (inode = 61114027), could not be opened
syslog.old (inode = 2555902), some blocks weren't allocated
activated-packages (inode = 7077851), has blocks already set
        5581 nodes checked,
        25 blocks not allocated,
        1 blocks already set,
        34473 blocks could be freed

        files           4331
        directories     976
        attributes      173
        attr. dirs      73
        indices         28

        direct block runs               7144 (85.29 GiB)
        indirect block runs             724 (in 61 array blocks, 36.28 GiB)
        double indirect block runs      0 (in 0 array blocks, 0 bytes)

I'm curious as to what 'Tracker .... could not be opened ' means..

[ttcoder]

Probably nothing.. Everything works again now, I'm posting this from the PM partition :-)

Ticket can be low-prioritized I guess, seeing as the fix is simply to run checkbfs (no need to delete any file).

[pulkomandy]

mh... anything more we can do here? Not being able to boot when the FS is corrupt sounds ok to me, especially if checkfs can fix it.

[ttcoder]

Right not a #7665 blocker (as to the issue itself, a couple BFS bugs were fixed during the Code Sprint so maybe this can be closed somewhat...? If someone runs in the future into the same stack trace with code words BlockAllocator ..etc we'll know the fixes did not affect this ticket and it should be re-opened).