On 2019/6/6 7:54 下午, Massimo Burcheri wrote: > On Thu, 2019-06-06 at 19:38 +0800, Coly Li wrote: >> What is your gcc version ? > > That Gentoo had some gcc-9.1 enabled, however the running kernel > was created earlier, probably using gcc-8.3.0. > > The OpenSuse TW live where I tried to register the caching device > manually, I don't know what the kernel and tools are built with. > zypper info gcc says 9-1.2, so I guess this was the toolchain for > creation as well. OK, then it seems another gcc9 related issue. > >> This one is important, if I can have the kernel message or call >> trace of this segfault it will be very helpful. > > That one I don't have anymore. I can try with some 5.1.5 kernel > later.. > >>> Is my bcache definitely lost? > >> I am not sure for the dirty data on cache, but for the backing >> device you may have most of data back. Considering there is btrfs >> on top of it, a fsck is required. >> >> You may try to run the backing device wihtout attaching cache >> device by: echo 1 > /sys/block/bcache0/bcache/running > > Read my initial post, I did so. I was able to do a luksOpen. But > the btrfs inside was corrupted when I tried 'filesystem check' and > 'mount' on that. This is bad, it seems btrfs metadata lost. I am not able to help the data back, but now I am working full time on the gcc9 compiled bcache issue. This is my first priority now. -- Coly Li
