[Debian-ha-maintainers] Bug#803777: libqb: FTBFS on hurd-i386

[Svante Signell]

On Mon, 2016-02-22 at 12:33 +0100, Christoph Berg wrote:
> Re: Svante Signell 2016-02-20 <1455998664.2888.77.camel at gmail.com>
> > On Sat, 2016-02-20 at 20:21 +0100, Christoph Berg wrote:
> > > Hi,
> > > 
> > > Just a quick note here - the signal being sent here is from sbuild
> > > killing the process because it was stuck for four hours. The problem
> > > is that the sem syscall doesn't return.
> > 
> > Iv'e come a little further: The error code is due to ETIMEDOUT as
> > expected. A new patch will probably follow tomorrow. Thanks!
> 
> Hi,
> 
> I've done some digging as well:
> 
> int main() {
> sem_t sem;
> struct timespec ts = {0, 0};
> if (sem_init(&sem, 0, 0) == -1) {
>         perror ("sem_init");
>         return -1;
> }
> if (sem_timedwait(&sem, &ts)==-1) {
>         perror ("sem_timedwait");
>         return -2;
> }
> return 0;
> }
> 
> sem_timedwait: Connection timed out
> 
> From what I got from the manpages, ts should probably be initialized
> as I've done above. What seems to work is to initialize the semaphore
> to 1, but then I have absolutely no experience with semaphores, nor
> what this test should actually test.

Hi Christoph. I've found out that the test for sem_timedwait is not needed.
There is a program in the manpage that works well in Hurd too. Attached is an
updated patch. This patch and a follow-up patch on bug #803776 will be submitted
upstream ASAP.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: hurd-support.patch
Type: text/x-patch
Size: 5656 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/debian-ha-maintainers/attachments/20160222/83e23b99/attachment.bin>