Your message dated Sat, 15 Sep 2012 15:28:12 +0100
with message-id <[email protected]>
and subject line Re: Bug#687736: unblock: cp2k/2.2.426-6
has caused the Debian Bug report #687736,
regarding unblock: cp2k/2.2.426-6
to be marked as done.
This means that you claim that the problem has been dealt with.
If this is not the case it is now your responsibility to reopen the
Bug report if necessary, and/or fix the problem forthwith.
(NB: If you are a system administrator and have no idea what this
message is talking about, this may indicate a serious mail system
misconfiguration somewhere. Please contact [email protected]
immediately.)
--
687736: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=687736
Debian Bug Tracking System
Contact [email protected] with problems
--- Begin Message ---
Package: release.debian.org
Severity: normal
User: [email protected]
Usertags: unblock
Please unblock package cp2k, it fixes #686874, namely segfaults in most
test cases on i386 due to not-16byte-aligned arrays being passed to
(since recently) SSE2-using fftw3, see
https://buildd.debian.org/status/fetch.php?pkg=cp2k&arch=i386&ver=2.2.426-5&stamp=1346699763
|-------------------------regtesting cp2k----------------------------------
|Starting tests in
|/build/buildd-cp2k_2.2.426-5-i386-rUGkvz/cp2k-2.2.426/tools/..//TEST-Linux-i686-gfortran-popt-2012-09-03T19:08:13+0000/tests/QS/regtest-hfx
|Running H2O-hfx-1.inp
|Running H2O-hfx-2.inp
|Running H2O-hfx-3.inp
|./do_regtest: line 146: 29062 Segmentation fault ${cp2k_prefix}
|${input_file} ${cp2k_postfix} &>${output_file}
|Running CH-hfx-md.inp
|./do_regtest: line 146: 29079 Segmentation fault ${cp2k_prefix}
|${input_file} ${cp2k_postfix} &>${output_file}
|[...]
|number of FAILED tests 72
|number of WRONG tests 28
|number of CORRECT tests 141
vs.
https://buildd.debian.org/status/fetch.php?pkg=cp2k&arch=i386&ver=2.2.426-6&stamp=1346979234
|-------------------------regtesting cp2k----------------------------------
|Starting tests in
|/build/buildd-cp2k_2.2.426-6-i386-bTTI0T/cp2k-2.2.426/tools/..//TEST-Linux-i686-gfortran-popt-2012-09-07T00:42:13+0000/tests/QS/regtest-hfx
|Running H2O-hfx-1.inp
|Running H2O-hfx-2.inp
|Running H2O-hfx-3.inp
|Running CH-hfx-md.inp
|[...]
|number of FAILED tests 1
|number of WRONG tests 37
|number of CORRECT tests 203
The one remaining FAILED test is likely due to a timeout on the buildd,
the cp2k regression testsuite is setting a timer for each test, it
succeeds with no FAILED tests on my local i386 sbuild.
The issue had been diagnosed by the fftw3 maintainer (Julian Taylor) and
discussed with upstream; they have since committed a more disruptive
patch, but the current minimal patch should be fine.
unblock cp2k/2.2.426-6
Thanks in advance, regards
Michael
diff -Nru cp2k-2.2.426/debian/changelog cp2k-2.2.426/debian/changelog
--- cp2k-2.2.426/debian/changelog 2012-09-03 19:18:02.000000000 +0200
+++ cp2k-2.2.426/debian/changelog 2012-09-07 00:33:20.000000000 +0200
@@ -1,3 +1,16 @@
+cp2k (2.2.426-6) unstable; urgency=low
+
+ * debian/rules (FFTW3_ALIGN): New variable, set to '-D__FFTW3_UNALIGNED' on
+ i386.
+ * debian/patches/makeflags.patch: Include $(FFTW3_ALIGN) environment
+ variable in $(DFLAGS).
+ * debian/patches/fftw3_unaligned.patch: New patch, unconditionally adds the
+ FFTW_UNALIGNED flag to the FFTW3 plan if __FFTW3_UNALIGNED is defined,
+ thanks to Julian Taylor (Closes: #686874).
+ * debian/control (Description): Added Metadynamics to capabilities.
+
+ -- Michael Banck <[email protected]> Fri, 07 Sep 2012 00:27:39 +0200
+
cp2k (2.2.426-5) unstable; urgency=low
[ Michael Banck ]
diff -Nru cp2k-2.2.426/debian/control cp2k-2.2.426/debian/control
--- cp2k-2.2.426/debian/control 2012-09-03 19:16:24.000000000 +0200
+++ cp2k-2.2.426/debian/control 2012-09-06 21:43:40.000000000 +0200
@@ -71,6 +71,8 @@
* Classical Molecular Dynamics (MD) simulations in microcanonical ensemble
(NVE) or canonical ensmble (NVT) with Nose-Hover and canonical sampling
through velocity rescaling (CSVR) thermostats
+ * Metadynamics including well-tempered Metadynamics for Free Energy
+ calculations
* Classical Force-Field (MM) simulations
* Monte-Carlo (MC) KS-DFT simulations
* HFX module for linear-scaling MD simulations using hybrid functionals
diff -Nru cp2k-2.2.426/debian/patches/fftw3_unaligned.patch
cp2k-2.2.426/debian/patches/fftw3_unaligned.patch
--- cp2k-2.2.426/debian/patches/fftw3_unaligned.patch 1970-01-01
01:00:00.000000000 +0100
+++ cp2k-2.2.426/debian/patches/fftw3_unaligned.patch 2012-09-06
23:18:44.000000000 +0200
@@ -0,0 +1,30 @@
+--- ./src/fft_lib/fftw3_lib.F.orig 2012-09-06 23:09:54.067428479 +0200
++++ ./src/fft_lib/fftw3_lib.F 2012-09-06 23:18:17.691429179 +0200
+@@ -127,9 +127,13 @@
+ STOP "fftw3_create_plan_3d"
+ END SELECT
+
++#if defined ( __FFTW3_UNALIGNED )
++ fftw_plan_type = fftw_plan_type + FFTW_UNALIGNED
++#else
+ IF ( .NOT. aligned ) THEN
+ fftw_plan_type = fftw_plan_type + FFTW_UNALIGNED
+ END IF
++#endif
+
+ n1 = plan%n_3d(1)
+ n2 = plan%n_3d(2)
+@@ -257,9 +261,13 @@
+ STOP "fftw3_create_plan_1dm"
+ END SELECT
+
++#if defined ( __FFTW3_UNALIGNED )
++ fftw_plan_type = fftw_plan_type + FFTW_UNALIGNED
++#else
+ IF ( .NOT. aligned ) THEN
+ fftw_plan_type = fftw_plan_type + FFTW_UNALIGNED
+ END IF
++#endif
+
+ num_threads = 1
+ !$omp parallel default(none), &
diff -Nru cp2k-2.2.426/debian/patches/makeflags.patch
cp2k-2.2.426/debian/patches/makeflags.patch
--- cp2k-2.2.426/debian/patches/makeflags.patch 2012-02-19 13:55:30.000000000
+0100
+++ cp2k-2.2.426/debian/patches/makeflags.patch 2012-09-06 23:36:31.000000000
+0200
@@ -13,7 +13,7 @@
+
+AR = ar -r
+
-+DFLAGS = -D__GFORTRAN -D__FFTSG -D__FFTW3 -D__LIBINT
++DFLAGS = -D__GFORTRAN -D__FFTSG -D__FFTW3 $(FFTW3_ALIGN) -D__LIBINT
+CPPFLAGS = -C -traditional $(DFLAGS)
+FCFLAGS = -g -Wall -O2 -ffast-math -funroll-loops -ftree-vectorize
-ffree-form -ffree-line-length-none $(DFLAGS)
+LDFLAGS = $(FCFLAGS)
@@ -35,7 +35,7 @@
+
+AR = ar -r
+
-+DFLAGS = -D__GFORTRAN -D__FFTSG -D__FFTW3 -D__LIBINT -D__parallel -D__BLACS
-D__SCALAPACK
++DFLAGS = -D__GFORTRAN -D__FFTSG -D__FFTW3 $(FFTW3_ALIGN) -D__LIBINT
-D__parallel -D__BLACS -D__SCALAPACK
+CPPFLAGS = -C -traditional $(DFLAGS)
+FCFLAGS = -g -Wall -O2 -ffast-math -funroll-loops -ftree-vectorize
-ffree-form -ffree-line-length-none $(DFLAGS)
+LDFLAGS = $(FCFLAGS)
diff -Nru cp2k-2.2.426/debian/patches/series cp2k-2.2.426/debian/patches/series
--- cp2k-2.2.426/debian/patches/series 2012-03-24 17:59:41.000000000 +0100
+++ cp2k-2.2.426/debian/patches/series 2012-09-06 23:19:34.000000000 +0200
@@ -3,3 +3,4 @@
reset_cppflags_makdepf90.patch
data_files_standard_location.patch
testsuite_output.patch
+fftw3_unaligned.patch
diff -Nru cp2k-2.2.426/debian/rules cp2k-2.2.426/debian/rules
--- cp2k-2.2.426/debian/rules 2012-09-03 18:41:44.000000000 +0200
+++ cp2k-2.2.426/debian/rules 2012-09-06 23:40:15.000000000 +0200
@@ -3,6 +3,14 @@
include /usr/share/mpi-default-dev/debian_defaults
+DEB_BUILD_ARCH=$(shell dpkg-architecture -qDEB_BUILD_ARCH)
+
+ifeq ($(DEB_BUILD_ARCH), i386)
+ export FFTW3_ALIGN=-D__FFTW3_UNALIGNED
+else
+ export FFTW3_ALIGN=
+endif
+
export LIB_SCALAPACK=-lscalapack-$(ARCH_DEFAULT_MPI_IMPL)
export LIB_BLACS=-lblacsCinit-$(ARCH_DEFAULT_MPI_IMPL)
-lblacs-$(ARCH_DEFAULT_MPI_IMPL)
export OMPI_MCA_plm_rsh_agent=/bin/false
--- End Message ---
--- Begin Message ---
On Sat, 2012-09-15 at 16:03 +0200, Michael Banck wrote:
> Please unblock package cp2k, it fixes #686874, namely segfaults in most
> test cases on i386 due to not-16byte-aligned arrays being passed to
> (since recently) SSE2-using fftw3, see
>
> https://buildd.debian.org/status/fetch.php?pkg=cp2k&arch=i386&ver=2.2.426-5&stamp=1346699763
Unblocked; thanks.
Regards,
Adam
--- End Message ---
