Re: Bug#666110: pvbatch / openmpi refuses under chroot
- To: debian-science@lists.debian.org
- Cc: 666110@bugs.debian.org
- Subject: Re: Bug#666110: pvbatch / openmpi refuses under chroot
- From: Michael Banck <mbanck@debian.org>
- Date: Wed, 2 May 2012 13:15:46 +0200
- Message-id: <[🔎] 20120502111546.GN18895@nighthawk.chemicalconnection.dyndns.org>
- In-reply-to: <20120418222742.GB10883@nighthawk.chemicalconnection.dyndns.org>
- References: <1330704488.3213.13.camel@hamburg.upc.es> <CALjQY_FNQBHaSh=FF7TcQ-q9MH1F9DquwBE5XbbW58OQdk4pXw@mail.gmail.com> <20120328220142.GA6746@nighthawk.chemicalconnection.dyndns.org> <CALjQY_E2e4X+FiP-vjHF2bq4wy3ULVpv3kE4=qHsPAXz+OWxgA@mail.gmail.com> <CALF6qJnM3ER3Kbt2+MvB_MSR7BfcdxwOcxjwtXPhtGbwsQaL2A@mail.gmail.com> <4F74A7FB.40203@debian.org> <20120403162150.GB17836@crater1.logilab.fr> <4F7B2654.5090102@debian.org> <20120418222742.GB10883@nighthawk.chemicalconnection.dyndns.org>
tags 666110 +pending
thanks
Hi,
On Thu, Apr 19, 2012 at 12:27:43AM +0200, Michael Banck wrote:
> On Tue, Apr 03, 2012 at 06:33:24PM +0200, Sylvestre Ledru wrote:
> > On 03/04/2012 18:21, Julien Cristau wrote:
> > >On Thu, Mar 29, 2012 at 20:20:43 +0200, Sylvestre Ledru wrote:
> > >
> > >>On 29/03/2012 20:06, Anton Gladky wrote:
> > >>>I have similar problem on liggghts/lammps packages, which are running
> > >>>under openmpi [1] and [2].
> > >>>
> > >>>It would be good to try a patch from LP-bug [3].
> > >>Indeed. I am currently building openmpi 1.4.3-3 with the fix.
> > >>
> > >Are you also going to upload it? :)
> > Not yet, see the comment #38:
> > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=666110#38
> > I have a build issue with openmpi. I reported to upstream but no answer yet:
> > http://lists.alioth.debian.org/pipermail/pkg-openmpi-maintainers/2012-February/003106.html
> > (I pinged them recently)
>
> I had no trouble building openmpi_1.4.3-2.1 in a current unstable
> chroot. I think this bug is grave enough it warrants a fix to the
> unstable package before tending to new upstream versions, I believe you
> have trouble with 1.4.5?
>
> I could NMU the current unstable version with the patch if you agree.
I have just uploaded an NMU to DELAYED/3-day. Let me know if I should
move it to 0-day. Debdiff attached.
Michael
diff -u openmpi-1.4.3/debian/changelog openmpi-1.4.3/debian/changelog
--- openmpi-1.4.3/debian/changelog
+++ openmpi-1.4.3/debian/changelog
@@ -1,3 +1,11 @@
+openmpi (1.4.3-2.2) unstable; urgency=low
+
+ * Non-maintainer upload
+ * debian/patches/arm_support: fix ARM timer code, by Leif Lindholm.
+ Closes: #666110.
+
+ -- Michael Banck <mbanck@debian.org> Wed, 02 May 2012 00:34:48 +0200
+
openmpi (1.4.3-2.1) unstable; urgency=low
* Non-maintainer upload
diff -u openmpi-1.4.3/debian/patches/arm_support openmpi-1.4.3/debian/patches/arm_support
--- openmpi-1.4.3/debian/patches/arm_support
+++ openmpi-1.4.3/debian/patches/arm_support
@@ -255,7 +255,7 @@
===================================================================
--- /dev/null 1970-01-01 00:00:00.000000000 +0000
+++ openmpi-1.4.3/opal/include/opal/sys/arm/timer.h 2011-01-06 14:49:50.000000000 +0200
-@@ -0,0 +1,32 @@
+@@ -0,0 +1,40 @@
+/*
+ * Copyright (c) 2004-2005 The Trustees of Indiana University and Indiana
+ * University Research and Technology
@@ -279,12 +279,20 @@
+
+#define OPAL_HAVE_SYS_TIMER_GET_CYCLES 1
+
++#include <sys/times.h>
++
+typedef uint64_t opal_timer_t;
+
+static inline opal_timer_t
+opal_sys_timer_get_cycles(void)
+{
-+ return 0;
++ opal_timer_t ret;
++ struct tms accurate_clock;
++
++ times(&accurate_clock);
++ ret = accurate_clock.tms_utime + accurate_clock.tms_stime;
++
++ return ret;
+}
+
+#endif /* ! OMPI_SYS_ARCH_TIMER_H */
Reply to: