On 29/03/18 08:31, Eugene V. Lyubimkin wrote:
> Package: coreutils
> Version: 8.28-1
> Severity: normal
>
> Control: affects -1 cupt
>
>
> Hello,
>
> Thank you for maintaining coreutils. I use timeout(1) to conveniently
> detect hanging problems in the testsuite of cupt. Unfortunately, on hurd
> and hppa it makes the test suite fail.
>
> This issue is best illustrated through a test case:
>
> On hurd-i386 the commands which take less time to execute still make
> 'timeout' wait full time and exit with the error code:
> --8<-
> $ mkdir abc
> $ cd abc
> $ timeout 5s ls
> $ echo $?
> 124
> $ timeout 5s cat
> $ echo $?
> 124
> -->8-
>
> On, say, amd64 the same scenario works as expected:
> --8<-
> $ mkdir abc
> $ cd abc
> $ timeout 5s ls
> $ echo $?
> 0
> $ timeout 5s cat
> $ echo $?
> 124
> -->8-
>
> I couldn't check this scenario on hppa (as there are no porterboxes for
> it), but the bug report #882238 suggests a similar issue has been
> encountered on a hppa buildd box.
>
> On hurd-i386 portexbox, I did verify that the program run under
> timeout(1) does exit, but timeout(1) itself continues to run even
> though it has no child processes anymore.
>
>
There was a bug fixed wrt SIGCHLD in coreutils-8.29
