Ensure that automount(8) flushes autofs by checking fsid

[kib]

Summary

This is a reimplementation of the bug fix from D40961 without changing nmount(2) semantic.

Not tested.

[rmacklem]

Looks fine to me. I don't know enough about
autofs to say if it will fix the problem.

Hopefully the original reporter can test this
patch and confirm that it fixes the problem
for them?

[mjg]

sys/kern/vfs_mount.c

1390

this is open-coded fsidcmp

[kib]

Use fsidcmp()

[andrew_tao173.riddles.org.uk]

The additional check of the fstype in automount.c got lost in this patch; without it, you get a lot of unnecessary errors (since it will log_err for each covered mountpoint)

[kib]

In D42023#959415, @andrew_tao173.riddles.org.uk wrote:

The additional check of the fstype in automount.c got lost in this patch; without it, you get a lot of unnecessary errors (since it will log_err for each covered mountpoint)

Do you mean your patch from D40961 for automount.c? If yes, please send me it as git-formatted patch by mail, with all metadata and perhaps 'Differential revision: https://reviews.freebsd.org/D40961' tag. It is your change.

[andrew_tao173.riddles.org.uk]

Sent.

[kib]

Include relevant chunk from D40961

[kib]

fsid for MNT_UPDATE call should be taken from the last statfs, I believe

[andrew_tao173.riddles.org.uk]

In D42023#959423, @kib wrote:

fsid for MNT_UPDATE call should be taken from the last statfs, I believe

I don't know of any legitimate case where one autofs would be mounted directly over another one, so it should not matter much; using the last statfs is probably cleanest.

I will test this version shortly.

[andrew_tao173.riddles.org.uk]

tested, seems to work.

[trasz]

I'm not sure what exactly happened here, but I suspect there was something wrong with testing it: while one of previous Kib's commits fixed the instapanic on "automount -c", it still doesn't work, see last few entries at https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=274992.

[andrew_tao173.riddles.org.uk]

Yeah, something's still wrong, looking into it.

[andrew_tao173.riddles.org.uk]

sys/kern/vfs_mount.c

1390

Surely this is misusing fsidcmp? the first arg should be fsid_up, not the address of fsid_up which is a pointer.

[kib]

sys/kern/vfs_mount.c

1390

Did it helped? Should I just go ahead with the change alone anyway?

[trasz]

It appears to be working correctly now. Thank you all :)