killall -SOMESIGNAL broken by psmisc-23.2

[ken@…]

I'm sure I mentioned this some months ago, either in October or November, but I'm not sure on which list and I can't find it in the archives. I reverted to psmisc-23.1 for my own builds. I use killall -HUP (or whatever) fairly frequently, e.g. to save state in a browser before rebooting. With 23.2 that breaks (any named signal appears to generate an error).

Apart from general usage, I used to stop syslogd when rotating my logs with

 /bin/killall -HUP syslogd

but that became ineffective in 23.2 (daemon was still running, log did not get properly rotated). I think Bruce's view was "wait for the next release of psmisc", but since we are now heading for our own release ...

Hopefully, someone using 23.2 can try killing a named process to see if it gets killed (in which case this ticket can be closed as invalid), or else report the error message.

Looking at the BLFS book just now (pulseaudio) I see that I mention killall there.

[Xi Ruoyao]

I confirm this issue. And this change may also make a bash-5.0 test hangs.

[Xi Ruoyao]

$ sleep 100 &
[1] 3347
$ /bin/killall -HUP sleep
Usage: killall [OPTION]... [--] NAME...
       killall -l, --list
       killall -V, --version

[Douglas R. Reno]

I am able to confirm as well.

[Xi Ruoyao]

We are using the wrong tarball. The 23.2 release is actually

​https://sourceforge.net/projects/psmisc/files/psmisc/psmisc-23.2.tar.xz

This tarball doesn't have the issue.

[Bruce Dubbs]

I agree that the URLs are different:

https://sourceforge.net/projects/psmisc/files/psmisc/psmisc-23.2.tar.xz
https://sourceforge.net/projects/psmisc/files/psmisc%20devel/psmisc-23.2.tar.xz

0524258861f00be1a02d27d39d8e5e62  psmisc-23.2.tar.xz.1
17b72c193b090f379fedf573123e89b8  psmisc-23.2.tar.xz

The one we have int the book is dated 2018-08-13. The good one is dated 2018-09-24.

[Bruce Dubbs]

I updated the URL and stats in the book. Should be fixed at revision 11509.