On 2019/7/9 上午8:49, Jungyeon Yoon wrote: > Hi btrfs developers, > > I'm Jungyeon Yoon. I have reported btrfs bug before. > Some of them have marked to fixed thanks to your efforts. > Additionally following bugs seems also fixed as I've checked. > If okay, I would like to close following bugs, too. > > https://bugzilla.kernel.org/show_bug.cgi?id=202839 > https://bugzilla.kernel.org/show_bug.cgi?id=202751 > https://bugzilla.kernel.org/show_bug.cgi?id=202753 > > > In addition to those bugs, > I would like for following 16 reports to be checked. > Most of following 9 cases end up reporting kernel panic or kernel BUG messages. > > https://bugzilla.kernel.org/show_bug.cgi?id=202817 > https://bugzilla.kernel.org/show_bug.cgi?id=202819 > https://bugzilla.kernel.org/show_bug.cgi?id=202821 > https://bugzilla.kernel.org/show_bug.cgi?id=202823 > https://bugzilla.kernel.org/show_bug.cgi?id=202825 > https://bugzilla.kernel.org/show_bug.cgi?id=202827 > https://bugzilla.kernel.org/show_bug.cgi?id=202829 > https://bugzilla.kernel.org/show_bug.cgi?id=202831 > https://bugzilla.kernel.org/show_bug.cgi?id=202833 > https://bugzilla.kernel.org/show_bug.cgi?id=202837 Have addressed all above cases with the new patchset titled "[PATCH 0/5] btrfs: Enhanced runtime defence against fuzzed images". Although more than half of the reports are already fixed in v5.2. Just one recommendation for later reports. Please, please don't use DUP metadata/data profile. Use SINGLE instead. It will make our life MUCH MUCH MUCH easier. The problem here is, btrfs-progs' check on tree block is not as strict as kernel, since btrfs-progs needs to repair them, rejecting just like kernel is not acceptable. This leads to one problem, btrfs dump-tree can sometimes uses the incorrect mirror while kernel uses the other mirror (correct one), causing unexpected result for manual analyze. This makes manual analyze more time consuming, normally I need to either mount the fs and trigger all read on all tree blocks and let kernel to fix the incorrect mirror and hopes no other modification is done to the fs. Or for the worst case, I need to manually corruption certain mirrors to force btrfs-progs to use the other mirror. If you could change the metadata profile to SINGLE, it would save us a lot of time. Thanks, Qu > > > Following 7 cases run into assertion with btrfs check integrity options on. > https://bugzilla.kernel.org/show_bug.cgi?id=203279 > https://bugzilla.kernel.org/show_bug.cgi?id=203253 > https://bugzilla.kernel.org/show_bug.cgi?id=203255 > https://bugzilla.kernel.org/show_bug.cgi?id=203257 > https://bugzilla.kernel.org/show_bug.cgi?id=203259 > https://bugzilla.kernel.org/show_bug.cgi?id=203261 > https://bugzilla.kernel.org/show_bug.cgi?id=203251 > > > Thank you, > Jungyeon >
Attachment:
signature.asc
Description: OpenPGP digital signature
