[Pkg-shadow-devel] redhat patches
nicolas.francois at centraliens.net
Tue Nov 13 12:51:58 UTC 2007
On Tue, Nov 13, 2007 at 10:58:59AM +0100, Peter Vrabec wrote:
> do you mind to look at some redhat patches, which might be interesting to
> I'm sending one example. It's fixing this problem:
> If you have a group file that looks something like the following:
> And you try to "userdel deleteme", userdel will go into an endless loop.
As it fixes this issue, it shall be included upstream.
I will have a look.
> Also I'd like to ask:
> You fixed getgr_nam_gid() in usermod.c but didn't do the same in useradd.c. Is
> there any reason?
The part in usermod was already fixed in the PLD CVS, but useradd was not
I haded very few changes between the PLD CVS and 220.127.116.11.
I also noticed this one later when preparing the Debian package.
This will be fixed in 4.0.19.
Here are some comments on the RedHat patches (based on
Is it still needed?
What are /etc/ptmp and /tmp/gtmp?
Are they standard lock files used by other common/recent software?
I've committed the part which implement the following comment:
/* note: the original util-linux newgrp didn't ask for
* pasword if there is no password. It's better directly give up.
* -- kzak at redhat.com
I don't think the rest of the patch introduce any functionnal change.
Am I wrong?
RedHat specific (we have something similar for Debian).
I'm not sure about the chunk:
- if (strlen (name) > sizeof (ut.ut_user))
+ if (strlen(name) + 1 > sizeof(ut.ut_user))
Not sure neither for the group name length.
I will keep it for latter.
This adds a new useradd option.
This will make the useradd implementation incompatible with other
implementation, but people are free not to use that option.
I think I will commit it.
Why is the current nscd_flush_cache implementation not sufficient?
Fixed by Tomasz => in 18.104.22.168
Also fixed in 22.214.171.124
Config file => IMO, they have to be tuned by distributors
I will check later if upstream shall config files shall be updated.
Only useful when shadow-4.0.16-nscd.c is used?
Upstream implementation does not execve, but uses a socket.
The oflg stuff is not needed (find_new_uid is always protected by
!oflg). Maybe an assert could be nice.
For the main part of the patch (new options), it seems to be controversial.
My opinion is that they should be included, maybe with a note that
software willing to be compatible with other implementation shall not
use these options.
I would prefer to have some kind of script or a set of script in a
directory for this.
To be checked later.
Fixed in 126.96.36.199
Fixed in 188.8.131.52 + 4.0.19
Committed. (maybe with some indentation changes).
I don't remember;)
Probably has to be checked again after inclusion of
Distribution specific (but I'm not sure there is a distribution which
choose to distribute the shadow library).
More information about the Pkg-shadow-devel