On 2021/03/25 15:19, Tetsuo Handa wrote: > On 2021/03/25 0:20, Jeff Mahoney wrote: >> Tetsuo's patch is fine but it needs a similar fix in reiserfs_xattr_set, >> as you noted. Whether it's required is another question. ReiserFS is >> absolutely loaded with fuzzer bugs. > > Can we apply this patch as-is? Since this is currently 5th top crasher, > applying this patch as soon as possible helps utilizing syzbot's resource > for finding further bugs. > Will you explain why we need a similar fix in reiserfs_xattr_set() ? Debug print patch ( https://syzkaller.appspot.com/x/patch.diff?x=1112d621d00000 ) and console output ( https://syzkaller.appspot.com/x/log.txt?x=13e76921d00000 ) says "reiserfs_xattr_init returns -95" which indicates that reiserfs_fill_super() from mount attempts for such crafted filesystem images fails with -EOPNOTSUPP error. Given that such crafted filesystem images cannot be mounted, how can reiserfs_xattr_set() be called and cause problems?
