There were the patch for this issue in earlier bowtie version:
https://sources.debian.net/src/bowtie/1.1.2-6/debian/patches/no_hash_style_both_for_mips.patch/
This patch prevents using hash style both for mips, in 1.2.0 version it is
empty for some reason.
Control: reassign -1 libmariadb-dev 2.3.1-1
Hi,
This issue is caused by garbage output form mariadb_config.
The issue was reported and the fix has been proposed:
https://bugs.debian.org/cgi-bin/bugreport.cgi?archive=no=850142
I will merge these two bugs.
Dejan
Package: apt-cacher-ng
Version: 1-1
Severity: serious
Tags: sid, patch
Justification: FTBFS
User: debian-m...@lists.debian.org
Usertags: mips-patch
Build for mips and mipsel fails during apt-cacher-ng linking with error:
/usr/include/c++/6/bits/atomic_base.h:514: error: undefined reference to
Package: ngraph-gtk
Version: 6.06.13-5.1
Severity: serious
Tags: sid, patch
Justification: FTBFS
User: debian-m...@lists.debian.org
Usertags: mips-patch
Package ngraph-gtk FTBFS for mips* on testing with an error:
> make[6]: Entering directory '/«PKGBUILDDIR»/doc/html/ja'
> cat object.html.in
Control: tags -1 + patch
The same behavior is noticed on other architectures as well: alpha, arm64,
kfreebsd-amd64, kfreebsd-i386, mips64el, s390x, sparc64.
https://buildd.debian.org/status/logs.php?pkg=cpp-netlib=0.11.2%2Bdfsg1-2%2Bb2=sid
I took a look on github repository and this test has
Any updates on this issue?
I have tested newer releases (0.7.73 and 0.7.83) on local mips machine, both
build failed with Segmentation fault:
> PASS: ags_thread_test
> PASS: ags_audio_test
> PASS: ags_channel_test
> PASS: ags_recycling_test
> PASS: ags_audio_signal_test
> PASS:
Hi,
Another solution could be to add ggc-min-expand=20 to CXXFLAGS.
The solution is tested on mips and mipsel.
A patch that contains needed changes is attached.
Regards,
Dejan
--- icinga2-2.5.4.orig/debian/rules 2016-08-26 08:11:01.0 +
+++ icinga2-2.5.4/debian/rules 2016-09-09
Package: ffmpeg
Version: 7:3.0.2-1
Severity: serious
Tags: sid, patch
Justification: FTBFS
User: debian-m...@lists.debian.org
Usertags: mips-patch
Package ffmpeg FTBFS for mips64el on configure with the following error:
> ERROR: LoadLibrary/dlopen not found for avisynth
Error message from
Hi,
Are there any updates on this issue?
Any concerns for including proposed patch?
Regards,
Dejan
tags 828034 + patch
Workaround for this issue is available here:
https://github.com/gnss-sdr/gnss-sdr/issues/31
Patch that contains backported changes from mentioned workaround is attached.
With this patch applied I was able to build gnss-sdr successfully for i386,
amd64, mips*.
Note:
For
Hi Yaroslav,
did you test the solution mentioned here:
https://github.com/pydata/pandas/issues/10125 ?
Regards,
Dejan
Hi,
I noticed that last 4 times build failed on loongson boards
with same TERM signal error message.
https://buildd.debian.org/status/logs.php?pkg=libv8-3.14=mipsel=sid
Could you please try to build libv8-3.14 on mayer
or other non loongson board if there are any?
Maybe we should blacklist
Hi,
this is obviously a consequence of mips buildd's limitations.
In past all three successfully builds had happened on ball.
I also was able to built freecontact successfully for mips on machine with FPU,
localy.
On machine without FPU, if icme_timeout is significantly increased
tests
Hi,
I had tried to build freecontact locally on two different cavium boards.
On both board build failed with TERM signal error on testing.
It seems that this is the command that causes TERM signal:
/build/freecontact-hkgMtc/freecontact-1.0.21/src/.libs/lt-freecontact
--parprof=psicov
It seems that this test is problematic:
LD_LIBRARY_PATH=./src/.libs/:./gmock/gtest/lib/.libs/:./gmock/lib/.libs:./gtest/lib/.libs/
./src/.libs/protobuf-test --gtest_filter=ExtensionSetTest.RepeatedFields*
Running main() from gtest_main.cc
Note: Google Test filter =
Hi,
I had tested it on a few local MIPS boards,
it builds successfully for both mips and mipsel.
I will ask for give back.
Regards,
Dejan
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Hi,
the problem appears during clear phase, not on configure.
Using BEFORE_CLEAN instead BEFORE_CONFIGURE solves
this issue for me.
The patch that contains needed changes is attached.
I had tested it on mips, mipsel and amd64.
Could you please consider including this changes?diff -uNr
Hello,
I had noticed that the gui part builds with target all
even if the main target is out.
Beside that, in gui/Makefile OCAMLC variable is redefined and ocamlc.opt is
used instead of
ocamlc (defined in Makefile.inc).
The patch that contains needed changes is attached,
With this patch I was
Hi,
On which MIPS machine did you build it ?
As I mentioned before
https://bugs.debian.org/cgi-bin/bugreport.cgi?archive=nobug=770009#45
I had tested it on Cavium board without FPU.
Unfortunately, I do not have more info.
I deleted old imagemagic chroots,
but I just had started builds
on
Hi,
did anyone tried to wait some longer time to see if command will execute?
./magick.sh convert
/home/fourmond/tmp/imagemagick-6.8.9.9/debian/display-im6.svg \
-background none -define filter:blur=0.75 -filter
Gaussian -resize $SIZE \
-gravity
Hi,
I had tried to build samba locally
for mips and mipsel on four different machines,
and I was not able to reproduce this error.
Maybe we should ask for give back and see
the result.
Regards,
Dejan
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of
Hi,
My assumption is that this action on cavium boards need a lot of time to
execute,
and that is the reason for TERM signal.
while read SIZE; do \
mkdir -p
/«PKGBUILDDIR»/debian/tmp-Q16/usr/share/icons/hicolor/$SIZE/apps/ ;\
cd /«PKGBUILDDIR»/debian/build-quantum-Q16; \
Control: tags -1 + patch
Control: user -1 debian-m...@lists.debian.org
Control: usertags -1 + mips-patch
Unfortunately there is no feedback from upstream.
The problem is the condition roundedBits 0x.
On 32-bit system 0x treated as Int is -1,
due to maxBound :: Int is
Hello,
I tried to build cython 0.20.2-1 locally.
On Broadcom machine,
cython builds successfully.
On Cavium machine,
build failed with TERM signal on testing.
Also, I ran tests manually,
and all tests pass on Cavium,
but it needs a lot of time for execution.
Ran 9153 tests in 33413.962s
Hello,
Maybe it is possible to blacklist blitz++ on cavium machines,
and try to build it on broadcom board (ball, swarm).
I had tested it locally.
On cavium board,
build fail with TERM signal.
If tests are ran manually, all tests pass,
but requires a lot of time.
Actually, the array test is
Hello,
Last successfully built metview version
it was built against magics++ 2.18.15-7.
In magics++ 2.18.15-7 there is only static libMagWrapper
library as well.
The difference is that, in last metview version, during linking of
bin/uPlotBatch, are used flags: -fPIE -pie.
Using this flags on
Control: tags -1 + patch
Hi,
I had tested it on mips/mipsel,
the same error appears.
This is happening because of missing
build dependency (libjpeg8-dev).
Patch that fix this issue is attached.
Best Regards,
Dejandiff -uNr freeimage-3.15.4.orig/debian/control
Control: tags -1 + patch
Hello,
Patch that fixes this issue is attached.
This patch contains needed changes for mips mipsel and ppc64el.
(see: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=755816)
Regards,
Dejan
liblog4ada_add_symbols.debdiff
Description:
Control: tags -1 + patch
Hi,
the reason for test failures is a need to
access .sdformat in HOME directory.
Patch that fix this issue is attached.
With this patch I was able to build sdformat
for i386, amd64, mips and mipsel.
Could you please consider including these changes?
Best Regards,
Hello,
The error appears while compiling HQ2xScaler.cc.
Compilation finish successfully but it needs more time
than is allowed before raising TERM signal.
I had tested it locally, and on was able to reproduce
the error with signal TERM on cavium board.
On that board, time of HQ2xScaler.cc
Hi,
The reason for this failure is a difference
in alignment of long long for MIPS ISA and IA32.
This results that size of gu_rse structure on mips is 24,
instead of expected 20 (on ia32).
As long long is first attribute in structure,
so it is safe to use pack(4).
struct gu_rse
{
long
Adding -mips32r2 is not a good idea,
since all Debian on mips* is built for mips2.
Instead, i think that better solution would be to
define WEBP_USE_MIPS32 only in mips32 case.
(src/dsp/dsp.h)
The patch that contains needed changes is attached.
With this patch I was able to successfully
Hello,
The reason for this failure is the fact
that p does not get info about
Additional Authentication Data length (al)
in src/jose/apr_jwe.c for big endian.
Patch that fixes this issue is attached.
This patch should enable build of libapache2-mod-auth-openidc
on big endian architectures.
I
Hi,
I am preparing a patch that fixes this issue.
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Hello,
I took a look at test failures on mips/mipsel.
Test that fails is based on a result of dnal tool.
dnal is part of a debian package wise.
In file: Bio/Wise/__init__.py,
an error was raised because the return value from dnal
is greater than 1.
If you look at dnal.c:
alb =
Control: tags -1 + patch
Hello,
I was able to avoid this error and successfully build seqan on both mips and
mipsel,
adding flags -O0 -mxgot.
-O0 was added to solve issue:
virtual memory exhausted: Cannot allocate memory
during compilation of pair_align.cpp.
-mxgot was added to lift the
Control: block -1 by 741288
Hello,
this problem is caused using an older version of seqan, 1.3.1-1.
With a new seqan (1.4.1-3), built with proposed patch on #741288,
I was able to build tophat successfully on both mips and mipsel.
Best Regards,
Dejan
--
To UNSUBSCRIBE, email to
Hello,
I have investigated this a little bit more
This seems not to be a compile issue.
The reason for build failure on mips/mipsel is
improper usage of inline functions.
For example in a file:
src/age/graph/mesh.cpp
a functions mmalloc is used.
Declaration of mmalloc function is included
Sorry,
i sent a wrong patch.
Here is a good one.
In previous patch I accidentally included mmalloc.h in mmalloc.h :)diff -uNr berusky2-0.10.orig/src/age/utils/mem_alloc.h berusky2-0.10/src/age/utils/mem_alloc.h
--- berusky2-0.10.orig/src/age/utils/mem_alloc.h 2013-03-03 23:38:40.0 +
Hi,
it seams that build of chealpix fails only on Cavium machines.
Even on Cavium, tests execute successfully but
it takes a lot of time.
On Broadcom and Loongson boards,
chealpix builds successfully for me.
The same conclusion could be made from
build logs:
Hi,
I test it with gcc/g++ 4.8 on local board,
and berusky2 builds successfully.
Also,
I tried to build an older version of berusky2 (0.10-1),
it fails with g++-4.9 and
builds fine with g++-4.8.
Still,
I did not yet figured out the reason of this behavior.
--
To UNSUBSCRIBE, email to
Hello,
the same error is noticed for other big endian architectures like
mips and powerpc.
This patch from upstream fixes this issue for me on mips.
https://github.com/stedolan/jq/commit/20e629284f94e0c8a14e8425309c913b6ee5b280
The patch is attached.
Could you please consider including this
Hello Scott,
It seams that this failure
is related to board performance,
since this test failed three times but always on corelli (Cavium).
I was not able to reproduce this test failure on my local boards,
aside from decreasing time limit below 1.0.
To resolve this issue we could increase
Debian patch that removes test_vterm.py
is attached.
Vincent, could you please include this patch
in new Debian package version?
Regards,
Dejan
diff -uNr urwid-1.2.1.orig/urwid/tests/test_vterm.py urwid-1.2.1/urwid/tests/test_vterm.py
--- urwid-1.2.1.orig/urwid/tests/test_vterm.py
Hi,
It seams to me that the problem is in debian/rules file.
If package is being built with both arch specific and indep parts (fakeroot
debian/rules binary),
build finishes successfully .
If we build only arch specific parts of the package (binary-arch),
build fails on:
find
Hello,
Instead of statically linking,
we could replace the private libpcap function pcap_offline_read
with the proper pcap_dispatch function.
A similar issue is reported here:
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=545595
The patch that include this change is attached.
, March 31, 2014 10:26 AM
To: Charles Plessy
Cc: 736...@bugs.debian.org; Dejan Latinovic; debian-m...@lists.debian.org
Subject: Re: Bug#736565: [Debian-med-packaging] Bug#736565: FTBFS on non-PC
architectures: FAIL: scram_mt.test
On Mon, 2014-03-31 18:03:12 +1100, Aníbal Monsalve Salazar wrote
Hi,
Falcon has it own basic memory manager functions.
In these functions, during memory allocation,
the size of allocated memory is expanded by sizeof(size_t)
[4 bytes, on 32-bit MIPS],
which is used for saving an amount of allocated memory.
The first 4 bytes are used to store the amount.
Hello,
It seems that libncurses5-dev is added as build dependency
at latest version (1.5-1) of nettoe.
Due to that, libncurses5-dev and libtinfo-dev
were installed.
On configuration, term.h and libtermcap.so
were found and it is guessed that TERM environment variable is defined.
That was not the
Hello,
The reason for that is a migration to the newer version of bison,
from 2.5 to 3.0.2.
Attached patch contains changes to bison input files (src/core/conffile.y and
src/core/confline.y),
in order to adapt to newer bison version.
Could you please consider applying this patch?
Thanks,
Hello,
It is not a problem to locate libkqueue.pc,
but content of .pc file.
libkqueue.pc in 2.0.2-1:
Name: libkqueue
Description: Emulates FreeBSD kqueue(2) on other platforms
Version: 2.0.2
Requires:
Libs: -lkqueue
Libs.private: -lpthread -lrt
Cflags: -I${includedir}/kqueue
libkqueue.pc in
Hello,
As I see from debian/changelog,
tests for all architectures should be disabled.
Snip from changelog:
opendkim (2.9.0-2) unstable; urgency=low
* Disable test suite on all archs due to undiagnosed race condition in the
test suite that causes multiple FTBFS, but doesn't appear
Hello,
I have attached a patch that solves
this issue for me on mips/mipsel.
Regards,
Dejan
--- libpwiz-3.0.4624.orig/debian/rules 2014-02-11 14:46:15.0 +
+++ libpwiz-3.0.4624/debian/rules 2014-02-27 09:58:59.0 +
@@ -12,6 +12,7 @@
DEB_HOST_GNU_TYPE ?= $(shell
Hello,
I was getting the same error with libpwiz packge.
checking for boost/system/error_code.hpp... yes
checking for the Boost system library... yes
checking boost/thread.hpp usability... yes
checking boost/thread.hpp presence... yes
checking for boost/thread.hpp... yes
configure: error:
Hello,
Apart form mips/mipsel architectures, the declaration of memcpy is reached
through openmpi.
Since the mpi-defaults for mips/mipsel was build against mpich (not openmpi),
this issue appeared.
Therefore, package ray does build succsessfully on other architectures without
changes.
New
Hi,
The bug is fixed upstream.
Can you consider getting the latest source from
upstream?
https://github.com/ClusterLabs/libqb
Regards,
Dejan LatinovićAuthor: Dejan Latinovic dejan.latino...@rt-rk.com
Description: Check if call of function will go through lazy binding stub for mips.
Index
Control: tags -1 + patch
Hi,
this bug also occurs on other architectures:
https://buildd.debian.org/status/logs.php?pkg=argussuite=experimental
Snip of build log:
ArgusSource.o: In function `ArgusGetPackets':
is attached.
Author: Dejan Latinovic dejan.latino...@rt-rk.com
Description: Set RecCounter=0 if rbound_==0 to avoid cast of -inf.
Index: linbox-1.3.2/linbox/algorithms/rational-reconstruction-base.h
===
--- linbox-1.3.2.orig/linbox/algorithms
memory during reading.
This test does not fail on some other architectures
because there is a possibility that my_buf is located after hdr,
char my_buf[512];
and then the overwriting passes unnoticed.
A patch fixing this issue is attached.
Regards,
Dejan Latinović
Author: Dejan Latinovic
LatinovićAuthor: Dejan Latinovic dejan.latino...@rt-rk.com
Description: Remove srcdir due to changes in automake.
diff -uNr lhapdf-5.8.9.orig/octave/Makefile.am lhapdf-5.8.9/octave/Makefile.am
--- lhapdf-5.8.9.orig/octave/Makefile.am 2013-12-03 15:17:38.0 +
+++ lhapdf-5.8.9/octave
Hi,
it seems that scilab does not work correctly,
here is snip of build log:
--
Making all in scilab
make[6]: Entering directory
`/build/getfem++-uJr3zI/getfem++-4.2.1~beta1~svn4453~dfsg/interface/src/scilab'
SCI_DISABLE_TK=1 SCI_JAVA_ENABLE_HEADLESS=1 scilab -nw
Control: tags -1 + patch jessie
Control: found -1 0.5.9-4
Hi,
Mips doesn't support neither -m64 nor -m32 options.
I have attached a patch solving the issue.
Regards,
Dejan Latinovićdiff -uNr jaffl-0.5.9.orig/libtest/GNUmakefile jaffl-0.5.9/libtest/GNUmakefile
---
Package: gammaray
Version: 1.2.2-1
Severity: serious
Tags: jessie sid
Justification: FTBFS
In an attempt to rebuild the package on mips/mipsel, build failed on testing:
Running tests...
/usr/bin/ctest --force-new-ctest-process -j6
Test project
Hi,
the same error occurs on mips/mipsel.
Full build log:
https://buildd.debian.org/status/fetch.php?pkg=python-scryptarch=mipsver=0.6.1-5stamp=1377211402
Header openssl/aes.h is part of libssl-dev package.
After I installed it, package was built successfully.
Is it a correct solution to add
Hi,
gcc for mips/mipsel used in wheezy and jessie (sid) doesn't support
define_register_constraint h.
I have attached a patch solving the issue.
Cheers,
Dejan Latinovićdiff -uNr flint-1.011.orig/longlong.h flint-1.011/longlong.h
--- flint-1.011.orig/longlong.h 2008-06-16 03:31:56.0
Hi,
uc-echo has an unconditional build dependency on g++-multilib, which
is not
available on all architectures.
do you know a better choice?
maybe the better way would be to list all architectures that support
g++-multilib.
For sid those architectures are: amd64, i386, kfreebsd-amd64,
66 matches
Mail list logo
