*) Bug it is NOT FS related: we have reports so far for jfs, xfs, reiserfs, ext3 and ext4
*) Apart from a dubitable case, is x86_64 related, but concurrency is not the issue (even going UP it persists).
*) Is FOR SURE hardware related, ata_piix driver being the most probable candidate to host the bug.
*) It is critical for a lot of people as can trigger complete loss of disk stored data
About the method:
-) I can be considered responsible for not pointing out at first the problem. I apologize. My faults are:
*) I nominated ext4 in my first post, in a moment where lots of bugs were being solved and reported on that filesystem, and this for sure triggered down the importance of the bug.
*) I entitled the bug at first "kernel 2.6.28-11 kernel destroy system", which was for sure an overshoot I apologize for, and which caused developers think I was a N00b playing with its new shining toy.
-) I nevertheless consider the forced inclusion of the ext4 filesystem in jaunty the head problem of this: it triggered lots of bugs in a critical timeframe (just weeks before the release) and patches to make it work in a non ready for it kernel possibly triggered bugs in other drivers considered stable.
What we know:
*) Bug it is NOT FS related: we have reports so far for jfs, xfs, reiserfs, ext3 and ext4
*) Apart from a dubitable case, is x86_64 related, but concurrency is not the issue (even going UP it persists).
*) Is FOR SURE hardware related, ata_piix driver being the most probable candidate to host the bug.
*) It is critical for a lot of people as can trigger complete loss of disk stored data
About the method:
-) I can be considered responsible for not pointing out at first the problem. I apologize. My faults are:
*) I nominated ext4 in my first post, in a moment where lots of bugs were being solved and reported on that filesystem, and this for sure triggered down the importance of the bug.
*) I entitled the bug at first "kernel 2.6.28-11 kernel destroy system", which was for sure an overshoot I apologize for, and which caused developers think I was a N00b playing with its new shining toy.
-) I nevertheless consider the forced inclusion of the ext4 filesystem in jaunty the head problem of this: it triggered lots of bugs in a critical timeframe (just weeks before the release) and patches to make it work in a non ready for it kernel possibly triggered bugs in other drivers considered stable.
just my 2€c