(resending,... seems this hasn't gotten through to the list, when I've
sent it the first time)
Hi.
On Debian's 4.19.28-2 kernel (which includes the recent read-
corruption
on compression fix) the following happens:
As a consequence of the bug from the "Reproducer for "compressed data +
hole data corruption bug, 2018 edition" still works on 4.20.7" mail
thread I started (trying) to verify whether any of my data was
affected.
Part of this was looking for files which actually are compressed by the
two methods Zygo mentioned (compsize and filefrag -v).
For this I used two scripts like the attached ones (yes I know, bad
performance) being fed by find path -type f -exec script {} \; .
I've did this already on one of my disks, which I btrfs checked before
(both normal and lowmem mode with no error), blockdev --setro'ed the
device and mounted it ro.
The filefrag seems to cause all kinds of errors and call traces, giving
the dmesg output attached.
Any ideas what causes that?
These days I unfortunately strongly loose trust in the stability and
integrity of btrfs :-(
Cheers,
Chris.
Attachment:
compsize-compressed
Description: application/shellscript
Attachment:
filefrag-compressed
Description: application/shellscript
Attachment:
dmesg.xz
Description: application/xz
