'kill 0' from cron script kills crond

Bernhard Reutner-Fischer rep.dot.nop at gmail.com
Wed Dec 3 13:19:06 UTC 2008


On Wed, Dec 03, 2008 at 11:47:37AM +0100, Denys Vlasenko wrote:
>On Wednesday 03 December 2008 09:24, Bernhard Reutner-Fischer wrote:
>> >This means that "standard" cron puts each child into new process group.
>> >(BTW, this "standard" crond - what --version and/or --help says?)
>> >
>> >Please try this patch, does it fix the problem?
>> 
>> please use bb_setpgrp (see platform.h).
>
>Done.
>
>
>/* use legacy setpgrp(pid_t,pid_t) for now.  move to platform.c */
>#define bb_setpgrp() do { pid_t __me = getpid(); setpgrp(__me,__me); } while (0)
>
>Does __digital__ *really* have setpgrp(x,y) with two params?!

yes, it really has (if _XOPEN_SOURCE_EXTENDED is not defined).
We could probably also fallback to setpgid(), but setpgid(x,x) worked ok.

I do not know if that was changed in OSF/1 > 5.1 though. These later versions
don't support the HW i have so i can't check.

>Maybe it's a typo, should be setpgid(x,y)?



More information about the busybox mailing list