Dear Maintainer,
* What led up to the situation?
Upgrade to bookworm fragmented the EFI paritition (due to changing boot logo) and defrag was in order. As this BIOS
currently has broken legacy boot, I used dosbox-x to defrag the EFI partition rather than boot the FreeDOS CD-ROM.
Thus this uncovered the bug. On minimizing the bug, it got very simple. I can reliably reproduce as follows:
# hexedit /dev/sda
G 100000 (your value may vary)
PgDn
PgDn
^C
# hexedit /dev/sda1
PgDn
write a marker to the padding between the BPB and the first FAT sector.
(If your system doesn't have one, edit an error message in the boot sector)
# hexedit /dev/sda1
G 100000
PgDn
PgDn
Look for marker written above, find it's not there !!!
^C
For demonstration of the fact the cache is wrong but the data is sved, you can do the following:
# reboot
# hexedit /dev/sda
G 10000
PgDn
PgDn
Look for marker written above, find it's there !!!
* What exactly did you do (or not do) that was effective (or
ineffective)?
Only effective workaround I could find was reboot at certain locations.
* What outcome did you expect instead?
I seriously expect no possability of block level cache incoherency between
different partition devices on the same block device.
Quite frankly, I am amazed this doesn't cause tons and tons of issues.
Bug verified *upstream* on stock kernel version 6.4.12
Hi This bug was filed for a very old kernel or the bug is old itself without resolution. If you can reproduce it with - the current version in unstable/testing - the latest kernel from backports please reopen the bug, see https://www.debian.org/Bugs/server-control for details. Regards, Salvatore
Hi This bug was filed for a very old kernel or the bug is old itself without resolution. If you can reproduce it with - the current version in unstable/testing - the latest kernel from backports please reopen the bug, see https://www.debian.org/Bugs/server-control for details. Regards, Salvatore
Bug is still active on upstream kernel today. Verification is trivial.
Bug is still active on upstream kernel today. Verification is trivial.
Bug is still active on upstream kernel today. Verification is trivial.
Hi Joshua, Thanks for reporting back. In this case can you please report it to upstream and reference the upstream report here please? I'm reopening the bug. Thanks already! Regards, Salvatore
Hi Joshua, Thanks for reporting back. In this case can you please report it to upstream and reference the upstream report here please? I'm reopening the bug. Thanks already! Regards, Salvatore
I did long ago; no one cares.
I did long ago; no one cares.
Hi Joshua, Went trough the upstream posting at https://lore.kernel.org/lkml/CA+jjjYTk=5wn2o46uNB+bJYX8xLgMP==dsJuvC94DvtN2f_6Yw@mail.gmail.com/ . So as long there is nothing ongoing upstream related to it, we can neither do something downstream. Regards, Salvatore
Hi Joshua, Went trough the upstream posting at https://lore.kernel.org/lkml/CA+jjjYTk=5wn2o46uNB+bJYX8xLgMP==dsJuvC94DvtN2f_6Yw@mail.gmail.com/ . So as long there is nothing ongoing upstream related to it, we can neither do something downstream. Regards, Salvatore
As this usage of block devices is not supported by upstream and that isn't likely to change, I'm marking this wontfix and reducing severity. Ben.