On 2019/7/21 12:05, Al Viro wrote: > On Sun, Jul 21, 2019 at 11:08:42AM +0800, Gao Xiang wrote: > >> It is for debugging use as you said below, mainly for our internal >> testers whose jobs are >> to read kmsg logs and catch kernel problems. sb->s_id (device number) >> maybe not >> straight-forward for them compared with dev_name... > > Huh? ->s_id is something like "sdb7" - it's bdev_name(), not a device > number... You are right. Forgive me, actually we use /dev/block/by-name/system to mount fs... we have to do some lookup if using sdbX instead. > >> The initial purpose of erofs_mount_private was to passing multi private >> data from erofs_mount >> to erofs_read_super, which was written before fs_contest was introduced. > > That has nothing to do with fs_context (well, other than fs_context conversions > affecting the code very close to that). OK. That is fine. > >> I agree with you, it seems better to just use s_id in community and >> delete erofs_mount_private stuffs... >> Yet I don't look into how to use new fs_context, could I keep using >> legacy mount interface and fix them all? > > Sure. > >> I guess if I don't misunderstand, that is another suggestion -- in >> short, leave all destructors to .kill_sb() and >> cleanup fill_super(). > > Just be careful with that iput() there - AFAICS, if fs went live (i.e. > if ->s_root is non-NULL), you really need it done only from put_super(); > OTOH, for the case of NULL ->s_root ->put_super() won't be called at all, > so in that case you need it directly in ->kill_sb(). I got it. I will do a quick try now :) But in case of introducing issues, I guess I need to do some fault injection by hand..... Thanks, Gao Xiang >
