Code review comment for ~utkarsh/ubuntu/+source/autofs:merge-1917423-hirsute

Revision history for this message
Bryce Harrington (bryce) wrote :

Hi Utkarsh,

I think that's a good decision in this case. The purpose of my previous post was for explaining the thinking process one would go through to reach that decision. Since you've joined right as FF started, the thought process seemed worth some explanation, particularly since you're going to be facing more of such decisions in the coming weeks with other packages.

What I would probably do in this case is rather than FFe the package, to focus more narrowly on the bugfixes themselves, and cherry-pick any that look likely to affect users. Debian flagged a couple of the upstream changes, so I'd start by examining these myself:

  * New upstream release.
    - autofs: Fix crash in sun_mount() (Closes: #892953).
    - Use PKG_CHECK_MODULES to detect the libxml2 library. (Closes: #949055).

The sun_mount crash is arch-specific so if we're not hitting it in our CI, that one isn't important. The libxml2 fix is going to be needed when libxml2 updates; maybe it's worth pulling proactively?

Since the upstream release includes some bug fixes, are there any worth cherrypicking? These kind of struck my eye:

- fix a regression with map instance lookup.
- fix quoted string length calc in expandsunent().
- mount_nfs.c fix local rdma share not mounting.
- fix ldap sasl reconnect problem.
- fix sss_master_map_wait timing.
- improve sss setautomntent() error handling.
- improve sss getautomntent() error handling.
- improve sss getautomntbyname() error handling.
- fix direct mount unlink_mount_tree() path.
- fix incorrect logical compare in unlink_mount_tree().
- fix remount expire.
- fix empty mounts list return from unlink_mount_tree().

Cherrypicking these fixes right now versus post-release is a bit easier since a release with only these changes could be released without FFe or SRU paperwork. The question would be if any of these are likely to actually affect end users. And in the case of autofs, as this is not a high profile package for the server team I'd probably be on the conservative side of what patches I'd bother with.

« Back to merge proposal