RTEMS / Tools / RTEMS Source Builder¶
Go to Issues or Merge Requests
Issues¶
23 - Microblaze Tools Fails to Build¶
Id |
23 |
State |
closed |
Type |
ISSUE |
Author |
Joel Sherrill |
Closed by |
Chris Johns |
Created |
2024-07-12T15:02:17.691Z |
Closed |
2024-07-16T22:30:37.013Z |
Updated |
2024-07-16T22:30:37.534Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/23 |
Merges |
1 |
Summary¶
Tools for microblaze-rtems6 fail to build with the issue being an XXX version conflict that results in a compile error for gcc. This is the error message from the RSB log:
libtool: compile: gcc -O2 -g -pipe -I/home/joel/rtems-work/rtems-source-builder/rtems/build/tmp/sb-1000/6/rtems-microblaze/home/joel/rtems-work/tools/6/include -I/home/joel/rtems-work/rtems-source-builder/rtems/build/tmp/sb-1000-internal/include -DHAVE_CONFIG_H -I. -I../../../gnu-mirror-gcc-d04fe55/mpc/src -I.. -g -O2 -MT mul.lo -MD -MP -MF .deps/mul.Tpo -c ../../../gnu-mirror-gcc-d04fe55/mpc/src/mul.c -o mul.o
../../../gnu-mirror-gcc-d04fe55/mpc/src/mul.c:175:1: error: conflicting types for \ |md_0_8216|\ mpfr_fmma\ |md_0_8217|\ ; have \ |md_0_8216|\ int(__mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, int, mpfr_rnd_t)\ |md_0_8217|\
175 | mpfr_fmma (mpfr_ptr z, mpfr_srcptr a, mpfr_srcptr b, mpfr_srcptr c,
| ^~~~~~~~~
In file included from ../../../gnu-mirror-gcc-d04fe55/mpc/src/mpc.h:25,
from ../../../gnu-mirror-gcc-d04fe55/mpc/src/mpc-impl.h:30,
from ../../../gnu-mirror-gcc-d04fe55/mpc/src/mul.c:22:
/usr/include/mpfr.h:765:21: note: previous declaration of \ |md_0_8216|\ mpfr_fmma\ |md_0_8217|\ with type \ |md_0_8216|\ int(__mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, const __mpfr_struct *, mpfr_rnd_t)\ |md_0_8217|\
765 | __MPFR_DECLSPEC int mpfr_fmma (mpfr_ptr, mpfr_srcptr, mpfr_srcptr, mpfr_srcptr,
| ^~~~~~~~~
I ran git bisect and it identified this as the breaking commit:
423d98d2058a7dac0921f9a13113b024696236b9 is the first bad commit
commit 423d98d2058a7dac0921f9a13113b024696236b9
Author: Chris Johns <chrisj@rtems.org>
Date: Sat Jun 8 09:04:11 2024 +1000
gdb: Build MPFR before GDB as it now depeneds on it
Closes #7
bare/config/devel/expat-internal.bset | 14 +++++++++
bare/config/devel/gmp-internal.bset | 14 +++++++++
bare/config/devel/mpfr-4.2.1.cfg | 18 +++++++++++
bare/config/devel/mpfr-internal.bset | 14 +++++++++
rtems/config/6/rtems-default.bset | 1 +
rtems/config/tools/rtems-default-tools.bset | 16 ++++++++--
source-builder/config/expat-2-1.cfg | 47 ++++++++++++++++++++++++---
source-builder/config/gcc-13.cfg | 8 ++---
source-builder/config/gcc-common-1.cfg | 16 ++--------
source-builder/config/gdb-common-1.cfg | 2 +-
source-builder/config/gmp.cfg | 48 +++++++++++++++++++++++++---
source-builder/config/mpfr.cfg | 49 ++++++++++++++++++++++++++---
source-builder/defaults.mc | 19 +++++++----
13 files changed, 223 insertions(+), 43 deletions(-)
create mode 100644 bare/config/devel/expat-internal.bset
create mode 100644 bare/config/devel/gmp-internal.bset
create mode 100644 bare/config/devel/mpfr-4.2.1.cfg
create mode 100644 bare/config/devel/mpfr-internal.bset
This is the git bisect log:
$ git bisect log
git bisect start
# status: waiting for both good and bad commits
# bad: [c085b3bafb66e7cf2f9afd8b2622b5c1799e1616] rtems-tools: Fetch rtems-syms improvements
git bisect bad c085b3bafb66e7cf2f9afd8b2622b5c1799e1616
# status: waiting for good commit(s), bad commit known
# good: [93321bacf219039a02b3150c83ed1114d5bd1657] Change README to markdown
git bisect good 93321bacf219039a02b3150c83ed1114d5bd1657
# good: [b7be268a79710ad267e967a112644b105b2c4133] config/rtems-kernel: Update for gitlab
git bisect good b7be268a79710ad267e967a112644b105b2c4133
# bad: [80f8bdb11dd373dfa26376c1beb378f0c16d7943] bare/expat-internal: Fix the default version
git bisect bad 80f8bdb11dd373dfa26376c1beb378f0c16d7943
# bad: [423d98d2058a7dac0921f9a13113b024696236b9] gdb: Build MPFR before GDB as it now depeneds on it
git bisect bad 423d98d2058a7dac0921f9a13113b024696236b9
# good: [d68c65b2e03ac7512922dc167f3573badc7d1cfe] config/rtems-net-services: Update for gitlab
git bisect good d68c65b2e03ac7512922dc167f3573badc7d1cfe
# good: [f39e9584775a0bccdbd6959a9c55d473330d08e2] Migrate devel links to gitlab
git bisect good f39e9584775a0bccdbd6959a9c55d473330d08e2
# first bad commit: [423d98d2058a7dac0921f9a13113b024696236b9] gdb: Build MPFR before GDB as it now depeneds on it
The microblaze target is using a specific version of binutils and gcc with patches extracted from the Xilinx repo. Updating to FSF versions is not possible. There is an effort underway to submit work but the person who reached out to the binutils and gcc communities says it may take a year. So we are stuck needing to get the Microblaze target back to its state before this patch.
The contrib/download_prerequisites script with the Xilinx gcc version cites these:
gmp-6.1.0.tar.bz2
mpfr-3.1.6.tar.bz2
mpc-1.0.3.tar.gz
isl-0.18.tar.bz2
Steps to reproduce¶
cd
rtems
../source-builder/sb-set-builder --prefix=${HOME}/rtems-work/tools/6 --log=l-microblaze.txt 6/rtems-microblaze
/milestone %6.1
Author: Joel Sherrill
2024-07-12T15:35:07.099Z
changed the description
Author: Alex White
2024-07-16T14:32:33.824Z
mentioned in merge request !39
24 - Update newlib hash for RUSAGE_PTHREAD and remove sys/tree.h¶
Id |
24 |
State |
closed |
Type |
ISSUE |
Author |
Joel Sherrill |
Assignee(s) |
Joel Sherrill |
Closed by |
Chris Johns |
Created |
2024-07-15T19:13:17.634Z |
Closed |
2024-07-15T22:18:49.160Z |
Updated |
2024-07-15T22:18:49.341Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/24 |
Merges |
1 |
Summary¶
Update newlib hash per Title.
Author: Joel Sherrill
2024-07-15T19:13:17.790Z
assigned to @joel
Author: Joel Sherrill
2024-07-15T19:19:59.241Z
mentioned in merge request !37
22 - All Builds Fail on Cygwin From chmod Returning an Error¶
Id |
22 |
State |
closed |
Type |
ISSUE |
Author |
Joel Sherrill |
Closed by |
Chris Johns |
Created |
2024-07-08T16:58:19.497Z |
Closed |
2024-07-09T20:49:28.912Z |
Updated |
2024-07-09T20:49:28.992Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/22 |
Merges |
1 |
Summary¶
For some unknown reason, the _chmod_ has begun to fail on Cygwin. This is done after untar-ing the source files and results in all Cygwin RSB builds failing.
Steps to reproduce¶
Build any target on Cygwin. The generated do-build script is run with “sh -ex” which will exit on any command returning non-zero. The following shows the results of manually “sh -ex do-build ; echo $?”
+ tar_exit=0
+ cd sourceware-mirror-newlib-cygwin-7947581
+ chmod -R a+rX,g-w,o-w .
1
Author: Amar Takhar
2024-07-08T17:00:20.169Z
What version of Windows?
Author: Amar Takhar
2024-07-08T17:00:33.055Z
marked this issue as related to #21
Author: Amar Takhar
2024-07-08T17:08:22.824Z
If it’s Windows 10+ chmod isn’t going to be needed and those flags will cause issues for sure especially the group perms. If we want to just say “You need at least Windows 10” then deleting all of them would be fine.
Author: Joel Sherrill
2024-07-08T17:11:58.600Z
mentioned in merge request !33
13 - sb-rtems-pkg does not update rtems-tools checksum¶
Id |
13 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-06-15T03:22:23.484Z |
Closed |
2024-06-15T18:26:25.620Z |
Updated |
2024-06-15T18:26:25.705Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/13 |
Merges |
1 |
Summary¶
Running the sb-rtems-pkg
tool does not update the checksum
Steps to reproduce¶
cd rtems
../source-builder/sb-rtems-pkg
git diff
The error is no checksum update when the hash is updated:
diff --git a/rtems/config/tools/rtems-tools-6.cfg b/rtems/config/tools/rtems-tools-6.cfg
index 7e4bfb0..8e7edc8 100644
--- a/rtems/config/tools/rtems-tools-6.cfg
+++ b/rtems/config/tools/rtems-tools-6.cfg
@@ -10,7 +10,7 @@
%define rtems_tools_source rtems-tools-%{rtems_tools_version}
%define rtems_tools_ext xz
%else
-%define rtems_tools_version ed300ed6061779ca6802b183e09d86e99649407d
+%define rtems_tools_version 9cd5599e468eaf847b690a4608906ca9bece10c7
%define rtems_tools_ext bz2
%endif
Author: Chris Johns
2024-06-15T03:22:23.637Z
assigned to @chris
Author: Chris Johns
2024-06-15T05:26:40.062Z
mentioned in merge request !29
12 - GIT links to gitlab.rtems.org need to use protocol=https¶
Id |
12 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-06-13T06:15:14.004Z |
Closed |
2024-06-13T07:18:12.069Z |
Updated |
2024-06-13T07:18:12.130Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/12 |
Merges |
1 |
Summary¶
Moving to Gitlab has changed the git repo path from git://git.rtems.org
to https://gitlab.rtems.org
for read-only access. The RSB code download.py
assumes anything with http
/ https
is a file download. The URLs should use protocol=https
.
Steps to reproduce¶
../source-builder/sb-rtems-pkg
Solution¶
Change the URLs back to git://gitlab.rtems.org
Author: Chris Johns
2024-06-13T06:15:14.100Z
assigned to @chris
Author: Amar Takhar
2024-06-13T07:00:44.135Z
Can’t you fix this by treating anything that ends in
.git
as a repo? I’ve never seen a clone URL without.git
at the end.
Author: Chris Johns
2024-06-13T07:00:44.135Z
Maybe but it seems fragile. For
download.py
’sgit
support a URL is normally:https://gitlab.rtems.org/rtems/rtos/rtems.git?fetch?checkout=mainSo I will make sure only the supported list of
git
commands are supported. Anything else is a file download
Author: Chris Johns
2024-06-13T07:00:44.135Z
Oh hang on, it is already covered in the code here https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/blob/main/source-builder/sb/download.py?ref_type=heads#L490 :
if _as[0] == 'protocol': if len(_as) != 2: raise error.general('invalid git protocol option: %s' % (_as)) if _as[1] == 'none': # remove the rest of the protocol header leaving nothing. us[0] = url_base[len('://'):] else: if _as[1] not in [ 'ssh', 'git', 'http', 'https', 'ftp', 'ftps', 'rsync' ]: raise error.general('unknown git protocol: %s' % (_as[1])) us[0] = _as[1] + url_baseThis issue is now about updating the
git
URLs
Author: Amar Takhar
2024-06-13T07:00:44.135Z
Well, since we’re the ones controlling it can’t we state that anything ending in
.git
is a repository? Also Python has urlparse which can easily get the base URL.
Author: Amar Takhar
2024-06-13T07:00:44.135Z
hmm okay should be using urlparse though instead of hacking the URL directly
Author: Chris Johns
2024-06-13T07:00:44.135Z
Yeah, old code and not related to this issue.
I will find the URLs that got changed and change them back.
Author: Amar Takhar
2024-06-13T07:00:44.135Z
What do you mean?
git://
is not a supported protocol on GitLab.
Author: Chris Johns
2024-06-13T07:00:44.135Z
Yes I know
git://
is not supported. The URL is internal to the RSB and how it works and not related to gitlab. A URL of:git://gitlab.rtems.org/rtems/tools/rtems-tools.git?protocol=httpsWill be turned into something like:
url: get: git://gitlab.rtems.org/rtems/tools/rtems-tools.git?protocol=https?fetch?checkout=main -> /Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git git: clone: https://gitlab.rtems.org/rtems/tools/rtems-tools.git -> sources/rtems-tools.git cmd: (None) /usr/bin/git clone https://gitlab.rtems.org/rtems/tools/rtems-tools.git /Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git exe: spawn: /usr/bin/git clone https://gitlab.rtems.org/rtems/tools/rtems-tools.git /Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git exe: ['/usr/bin/git', 'clone', 'https://gitlab.rtems.org/rtems/tools/rtems-tools.git', '/Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git'] Cloning into '/Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git'... git: fetch: https://gitlab.rtems.org/rtems/tools/rtems-tools.git -> sources/rtems-tools.git cmd: (/Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git) /usr/bin/git fetch exe: spawn: /usr/bin/git fetch exe: ['/usr/bin/git', 'fetch'] git: checkout: https://gitlab.rtems.org/rtems/tools/rtems-tools.git => main cmd: (/Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git) /usr/bin/git checkout main exe: spawn: /usr/bin/git checkout main exe: ['/usr/bin/git', 'checkout', 'main'] Already on 'main' Your branch is up to date with 'origin/main'. cmd: (/Users/chris/development/rtems/rsb/rtems-source-builder.git/rtems/sources/rtems-tools.git) git log -n 1 exe: spawn: git log -n 1 exe: ['git', 'log', '-n', '1'] commit ed300ed6061779ca6802b183e09d86e99649407d Author: Amar Takhar <amar@rtems.org> Date: Sat Jun 8 00:04:03 2024 -0400 Fix Python 3.12 PEP-701 Escapes are no longer allowed in strings. Fixes #3
Author: Amar Takhar
2024-06-13T07:00:44.135Z
Oh, yeah I see how it works now that is a strange way to do it but I guess if it works. The nice thing of just checking for .git at the ending is you can just cut and paste any urls in there without worrying about switching protocols in the url arguments. Also url.parse will get the argument of a URL even if it’s encoded.
Author: Chris Johns
2024-06-13T07:07:33.181Z
The change was made a while ago to support other sites. It is a way of fitting in a solution into something existing.
Author: Amar Takhar
2024-06-13T07:00:44.135Z
Fair enough.
Author: Chris Johns
2024-06-13T07:09:24.635Z
changed title from GIT {-download using http/https has broken download.py-} to GIT {+links to gitlab.rtems.org need to use protocol=https+}
Author: Chris Johns
2024-06-13T07:09:24.660Z
changed the description
Author: Chris Johns
2024-06-13T07:09:53.056Z
changed the description
Author: Chris Johns
2024-06-13T07:13:10.496Z
mentioned in merge request !25
11 - Source downloads fail¶
Id |
11 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-06-11T23:29:31.177Z |
Closed |
2024-06-12T01:41:26.609Z |
Updated |
2024-06-12T01:41:26.668Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/11 |
Merges |
1 |
Summary¶
Running sb-get-sources
fails because of incorrect logic in:
config/tools/rtems-kernel-common.cfg
source-builder/config/glib-2-meson.cfg
The changes are:
-%if %{_dry_run} && %{defined with_download}
+%if %{_dry_run} || %{defined with_download}
GNU GCC git upgrades has broken building gcc-4.9
compilers as the common support contains patches from git. Remove the GCC 4.9 builds from bare
.
Changes to fix #9 in !21 breaks building RTEMS 7. The MPFR define is not present.
Steps to reproduce¶
cd rtems
../source-builder/sb-get-sources --stop-on-error
Author: Chris Johns
2024-06-11T23:29:31.261Z
assigned to @chris
Author: Chris Johns
2024-06-11T23:34:32.748Z
changed the description
Author: Chris Johns
2024-06-12T00:29:28.446Z
Reviewing the comments in
config/tools/rtems-kernel-common.cfg
it says:# If a dry-run and with download ignore errors and correct setting for tools # and BSPs. Only after the source to download.and this means
--dry-run
and--with-download
should be set. Reviewing thesb-get-sources
command it is only providing--with-download
as a default option:opts = simhost.load_options(args, argopts, extras=['--with-download'])Add
--dry-run
is a better solution.Further review of
source-builder/sb/getsources.py
dry run returnsTrue
and the%{_dry_run}
macro is0
. I will add a set of extra defaults tosimhost.py
to match the option defaults.
Author: Chris Johns
2024-06-12T01:01:30.305Z
changed title from Source downloads fail{- with incorrect logic-} to Source downloads fail
Author: Chris Johns
2024-06-12T01:01:30.329Z
changed the description
Author: Chris Johns
2024-06-12T01:36:16.362Z
mentioned in merge request !22
9 - GDB 14.2 fails to build on MacOS¶
Id |
9 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-06-11T04:28:05.144Z |
Closed |
2024-06-11T04:48:36.646Z |
Updated |
2024-06-12T01:01:30.418Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/9 |
Merges |
1 |
From https://lists.rtems.org/pipermail/users/2024-June/068991.html
when trying to build the arm-tools with the Rtems Source Builder on osX and I get this error:
It starts with Python 3.12.3
RTEMS Tools Project - Source Builder Error Report
Build: error: building arm-rtems6-gdb-14.2-arm64-apple-darwin23.5.0-1
Command Line: ../source-builder/sb-set-builder --jobs=20 --prefix=/Volumes/Epics/LONG_ISLAND/RTEMS_ARM_libBsd_stack/rtems/6 6/rtems-arm
Python: 3.12.3 (main, Apr 9 2024, 08:09:14) [Clang 15.0.0 (clang-1500.3.9.4)]
https://gitlab.rtems.org/rtems/tools/rtems-source-builder.git/origin/ac974f70ecdb9c9ee64ff0611b78960437efc8da
Darwin Zarquon.local 23.5.0 Darwin Kernel Version 23.5.0: Wed May 1 20:14:38 PDT 2024; root:xnu-10063.121.3~5/RELEASE_ARM64_T6020 arm64
Tail of the build log:
but at the end /usr/local/bin/python3
gets called … which exists, but older Version???
checking how to link with libexpat... /Volumes/Epics/LONG_ISLAND/RTEMS_ARM_libBsd_stack/rsb/rtems/build/tmp/sb-501-internal/lib/libexpat.a
checking for XML_StopParser... yes
checking whether to use python... /usr/local/bin/python3
checking for python... no
configure: error: no usable python found at /usr/local/bin/python3
make[1]: *** [configure-gdb] Error 1
make: *** [all] Error 2
shell cmd failed: /bin/sh -ex /Volumes/Epics/LONG_ISLAND/RTEMS_ARM_libBsd_stack/rsb/rtems/build/arm-rtems6-gdb-14.2-arm64-apple-darwin23.5.0-1/do-build
error: building arm-rtems6-gdb-14.2-arm64-apple-darwin23.5.0-1%
junkes at Zarquon RTEMS_ARM_libBsd_stack % /usr/local/bin/python3
Python 3.12.2 (v3.12.2:6abddd9f6a, Feb 6 2024, 17:02:06) [Clang 13.0.0 (clang-1300.0.29.30)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
Even if I had previously actived a python environment ``./py3.12/bin/activate``
On Linux everything works \ |md_0_8230|\ as usual ...
Author: Chris Johns
2024-06-11T04:28:05.300Z
assigned to @chris
Author: Chris Johns
2024-06-11T04:28:49.029Z
I have investigated the problem and the MacOS specific patch to work round this issue did not make it across to the GDB 14.2 configuration.
Author: Chris Johns
2024-06-11T04:34:42.191Z
mentioned in merge request !21
Author: Chris Johns
2024-06-12T01:01:30.397Z
mentioned in issue #11
7 - GDB requires GMP 4.2+, and MPFR 3.1.0+¶
Id |
7 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Amar Takhar, Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-06-06T20:49:21.554Z |
Closed |
2024-06-08T19:42:26.969Z |
Updated |
2024-06-08T19:42:27.056Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/7 |
Merges |
1 |
The change to gdb-14
has changed the dependencies it has when building when a host does not have packages installed. The change is:
GMP 4.2+
MPDR 3.1.0+
We should check the change worked on MacOS with M2 processors
Author: Chris Johns
2024-06-06T20:49:21.618Z
changed due date to June 09, 2024
Author: Chris Johns
2024-06-06T20:49:21.668Z
assigned to @joel
Author: Chris Johns
2024-06-07T05:18:39.764Z
The RTEMS defaults tools for 6 builds GMP 6.3.0 so that is ok. The error must be MPFR.
The RSB
bare
has support for MPFR 4.2.0. This will be added to the defaults tools build.
Author: Chris Johns
2024-06-07T05:21:49.779Z
GCC 13 is building MPFR 4.2.1. The
bare
package should be moved to 4.2.1 and support in GCC 13 removed.
Author: Chris Johns
2024-06-07T22:30:20.731Z
assigned to @chris and unassigned @joel
Author: Chris Johns
2024-06-07T22:32:29.250Z
assigned to @joel
Author: Chris Johns
2024-06-08T04:44:19.340Z
@joel it seems MPFR and GDB use
libtool
and this is the error:libtool: link: warning: library `/opt/work/chris/rtems/deploy/rtems-deployment.git/build/tmp/sb-500/tools/rtems-default-tools/opt/rtems/deploy/lib/libmpfr.la' was moved. libtool: link: cannot find the library `/opt/rtems/deploy/lib/libgmp.la' or unhandled argument `/opt/rtems/deploy/lib/libgmp.la' make[2]: *** [Makefile:2173: gdb] Error 1It seems
libtool
does not like using staged libraries in a build and wants the library in the final location.
Author: Chris Johns
2024-06-08T04:44:19.340Z
If you have an existing install under the build prefix the build may work as
libgmp.a
etc are found. If you use:--prefix=/never-exist --no-installyou will see the problem. This is a simple RSB set of options for the command line tests the RSB build works for a clean install.
The issue is the
.la
fileslibtool
uses being installed into the_tmproot
path used to stage internally used libraries. If they are deleted before the GDB build runs with:rm -f $(find %{_tmproot} -name \\*.la)before
make
runs in thegdb-common-1.cfg
the build and GDB and GCC build.I think the solution is to make GPM and MPFR internal packages so the prefix they have is internal and
libtool
should not see a problem.
Author: Chris Johns
2024-06-08T04:44:19.340Z
I have made the changes to build Expat, GMP and MPFR as internal packages. The build installs static libraries into a temporary internal path in the RSB’s build area and GDB and GCC link against these. The packages are not installed which I think is a good thing.
The MR !16 has the changes.
Author: Chris Johns
2024-06-07T23:06:16.806Z
assigned to @amar and unassigned @chris and @joel
Author: Chris Johns
2024-06-07T23:07:27.322Z
I have a patch on https://gitlab.rtems.org/chris/rtems-source-builder/-/tree/7-gdb-requires-gmp-4-2-and-mpfr-3-1-0?ref_type=heads which shows the error
Author: Chris Johns
2024-06-07T23:30:44.676Z
mentioned in issue #8
Author: Chris Johns
2024-06-08T00:47:06.901Z
mentioned in merge request !16
Author: Chris Johns
2024-06-08T04:44:42.974Z
assigned to @chris
8 - Change GDB back to 13¶
Id |
8 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Chris Johns |
Created |
2024-06-07T23:30:44.306Z |
Closed |
2024-06-08T01:12:24.941Z |
Updated |
2024-06-08T01:12:24.952Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/8 |
Merges |
0 |
Change the 6 version of GDB back to 13 until #7 is resolved.
Author: Chris Johns
2024-06-07T23:30:44.394Z
assigned to @chris
Author: Chris Johns
2024-06-07T23:53:49.093Z
Changing GDB to 13.2 lets it build however GCC fails with:
libtool: warning: library '/opt/work/chris/rtems/deploy/rtems-deployment.git/build/tmp/sb-500/tools/rtems-default-tools/opt/rtems/deploy/lib/libmpfr.la' was moved. /usr/bin/grep: /opt/rtems/deploy/lib/libgmp.la: No such file or directory /usr/bin/sed: can't read /opt/rtems/deploy/lib/libgmp.la: No such file or directory libtool: error: '/opt/rtems/deploy/lib/libgmp.la' is not a valid libtool archiveAnother but similar error to the GDB MPFR one reported in #7
This is on Rocky 9
Author: Chris Johns
2024-06-08T01:12:11.875Z
I am closing this. It looks like the problem in #7 is related to deployment builds in
rpmbuild
to make an RPM. The updates from @joel build cleanly on FreeBSD 14 and Rocky 9 from the RSB command line.
4 - Move devel.rtems.org patches to gitlab.rtems.org¶
Id |
4 |
State |
closed |
Type |
ISSUE |
Author |
Gedare Bloom |
Assignee(s) |
Chris Johns |
Closed by |
Kinsey Moore |
Created |
2024-05-16T18:06:24.945Z |
Closed |
2024-06-06T23:26:44.614Z |
Updated |
2024-06-06T23:26:44.626Z |
Milestone |
6.1 |
Labels |
infra::migration, tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/4 |
Merges |
0 |
Summary¶
There are approx. 73 references to devel.rtems.org in RSB and most of them appear to be linking to patches. The patch attachment links appear to be mechanically fixable by replacing:
https://devel.rtems.org/raw-attachment/ticket/####/FILENAME
with
https://gitlab.rtems.org/rtems/rtos/rtems/-/blob/main/assets/tracmigration/ticket_attachments/FILENAME.patch
Steps to reproduce¶
Author: Gedare Bloom
2024-05-16T18:06:25.054Z
assigned to @chris
Author: Amar Takhar
2024-05-16T18:23:13.241Z
This won’t work. The directory structure was flattened so there are no more subdirectories I did this to future-proof in the hopes that they can actually be imported into GitLab in the future. So if there were previously 5 patch.diff files they will now be patch.diff, patch-1.diff, patch-2.diff etc.
I’ll see if I can generate a log so we can easily replace these not sure though due to how tracboat is written.
Author: Kinsey Moore
2024-06-05T21:04:36.562Z
mentioned in merge request !15
Author: Kinsey Moore
2024-06-06T23:26:44.321Z
This is fixed now that !15 is merged.
3 - Move git.rtems.org configurations to gitlab.rtems.org¶
Id |
3 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Kinsey Moore |
Created |
2024-05-16T05:54:50.654Z |
Closed |
2024-06-06T23:26:30.996Z |
Updated |
2024-06-06T23:26:31.011Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/3 |
Merges |
0 |
Summary¶
A number of configuration files reference git.rtems.org
and they need to be changed to reference gitlab.rtems.org
.
Author: Chris Johns
2024-05-16T05:54:50.758Z
assigned to @chris
Author: Kinsey Moore
2024-06-05T16:58:06.143Z
I’m working on this patch set, just FYI.
Author: Amar Takhar
2024-06-05T17:03:31.699Z
If there are any comments required outside of the changes we should keep them here and point users to this ticket if they have issues the MR should give them an indication of what URLs they need to switch ot.
Author: Kinsey Moore
2024-06-05T18:20:17.005Z
mentioned in merge request !15
Author: Kinsey Moore
2024-06-05T18:20:44.964Z
created branch 3-move-git-rtems-org-configurations-to-gitlab-rtems-org to address this issue
Author: Kinsey Moore
2024-06-05T18:22:05.220Z
Ignore that created branch…the MR is !15 to fix main/6/7.
Author: Kinsey Moore
2024-06-06T23:26:30.695Z
This is fixed now that !15 is merged.
6 - CI to update file cache on ftp.rtems.org¶
Id |
6 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Amar Takhar |
Closed by |
Amar Takhar |
Created |
2024-05-24T06:35:32.626Z |
Closed |
2024-05-26T19:27:41.058Z |
Updated |
2024-05-26T19:29:13.811Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/6 |
Merges |
0 |
Summary¶
This issue is to add a CI that runs after a merge is pushed to the repo. It runs the RSB’s source download. There is a file cache per active branch in the RSB.
The URL is https://ftp.rtems.org/pub/rtems/cache/rsb and branches are subdirectories under this. For example the main
branch is https://ftp.rtems.org/pub/rtems/cache/rsb/main
Reason for the Cache¶
The cache is provided for user sites that are locked down. The RSB will not work because it fetches source from a number of hosts and IT support staff tend to not want to add the list of places the RSB references. The cache provides a single machine which is can be made accessible allowing the RSB to build tools.
Command¶
The command to run is:
cd rtems-source-builder/rtems
../source-builder/sb-get-sources
The command will check all buildsets and configuration files for all support hosts.
The files can be found in sources
and patches
and the contents of both can be copied to the cache directory.
Author: Chris Johns
2024-05-24T06:35:32.739Z
assigned to @amar
Author: Amar Takhar
2024-05-26T19:27:41.184Z
moved to administration/integration#6
Author: Amar Takhar
2024-05-26T19:29:13.796Z
Moved because the correct place for these issues is in the integration group otherwise they will get lost.
5 - The –url command line option does not dowload all paths¶
Id |
5 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Chris Johns |
Created |
2024-05-16T22:40:13.794Z |
Closed |
2024-05-23T22:08:28.464Z |
Updated |
2024-05-23T22:08:28.507Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/5 |
Merges |
1 |
Summary¶
When using --url
some paths are not redirected:
download: https://sourceware.org/bugzilla/attachment.cgi?id=14783&format=raw -> patches/gdb-14067-python-config.diff
Steps to reproduce¶
Use:
../source-builder/sb-set-builder --log=ppc-6-ppc.txt --prefix=/opt/rtems/6 --url=https://ftp.rtems.org/pub/rtems/cache/rsb/main 6/rtems-powerpc
Author: Chris Johns
2024-05-16T22:40:13.883Z
assigned to @chris
Author: Chris Johns
2024-05-17T05:59:35.715Z
changed title from The –url command line option does not {-move-} all paths to The –url command line option does not {+dowload+} all paths
Author: Chris Johns
2024-05-17T06:00:56.961Z
I will clean up releases with the change. Releases where being treated as special when the
--url
option should be the same.The requirement for
--url
is the path is to a cache of all the files.
Author: Chris Johns
2024-05-17T06:03:35.561Z
mentioned in merge request !11
2 - QEMU failed to build on FreeBSD 14¶
Id |
2 |
State |
closed |
Type |
ISSUE |
Author |
Chris Johns |
Assignee(s) |
Chris Johns |
Closed by |
Amar Takhar |
Created |
2024-05-13T07:21:47.963Z |
Closed |
2024-05-16T05:07:07.237Z |
Updated |
2024-05-16T05:07:07.302Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/2 |
Merges |
1 |
Pixman and GLib fail to build on FreeBSD 14¶
Pixman update to 0.42.2 build however GLib 2.69.3 requires meson
and ninja
to build. This is a major changed.
Steps to reproduce¶
cd bare
../source-builder/sb-set-builder --prefix=/opt/rtems/qemu --log=qemu.txt devel/qemu
Author: Chris Johns
2024-05-13T07:21:48.106Z
assigned to @chris
Author: Chris Johns
2024-05-14T04:16:16.059Z
Glib uses
pkg-config
viameson
andmeson
checks for--version
to determine ifpkg-config
is usable. I have added--version
to the RSBpkg-config
andmeson
seems happier:Found pkg-config: YES (/opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config) version: 1.2.3The Glib build now fails due to:
Run-time dependency libpcre found: NO (tried pkgconfig and cmake) Run-time dependency libpcre found: NO (tried pkgconfig and cmake) Looking for a fallback subproject for the dependency libpcre Downloading libpcre source from https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 <urlopen error [Errno 8] Name does not resolve> WARNING: failed to download with error: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?. Trying after a delay... <urlopen error [Errno 8] Name does not resolve> WARNING: failed to download with error: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?. Trying after a delay... <urlopen error [Errno 8] Name does not resolve> WARNING: failed to download with error: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?. Trying after a delay... <urlopen error [Errno 8] Name does not resolve> WARNING: failed to download with error: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?. Trying after a delay... <urlopen error [Errno 8] Name does not resolve> WARNING: failed to download with error: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?. Trying after a delay... <urlopen error [Errno 8] Name does not resolve> A fallback URL could be specified using source_fallback_url key in the wrap file meson.build:2002:9: ERROR: could not get https://ftp.pcre.org/pub/pcre/pcre-8.37.tar.bz2 is the internet available?Oh hmm
Author: Chris Johns
2024-05-14T04:16:16.059Z
FreeBSD has
pcre2-10.43
installed and the following is present:$ pkg which /usr/local/lib/libpcre2-16.a /usr/local/lib/libpcre2-16.a was installed by package pcre2-10.43and:
$ find /usr/local/libdata/ -name \\*.pc | grep pcre /usr/local/libdata/pkgconfig/libpcre2-32.pc /usr/local/libdata/pkgconfig/libpcre2-posix.pc /usr/local/libdata/pkgconfig/libpcre2-16.pc /usr/local/libdata/pkgconfig/libpcre2-8.pcThe RSB’s
pkg-config
hasmeson
reporting:Pkg-config binary missing from cross or native file, or env var undefined. Trying a default Pkg-config fallback at pkg-config Found pkg-config: YES (/opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config) version: 1.2.3 Determining dependency 'libpcre' with pkg-config executable '/opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config' env[PKG_CONFIG_BUILD_TOP_DIR]: /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/tmp/sb-1001/devel/qemu env[PKG_CONFIG_PATH]: /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/tmp/sb-1001/devel/qemu/opt/work/rtems/qemu/lib/pkgconfig env[PKG_CONFIG]: /opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config ----------- Called: `/opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config --modversion libpcre` -> 1Using the FreeBSD
pkg
installedpkg-config
gives:Pkg-config binary missing from cross or native file, or env var undefined. Trying a default Pkg-config fallback at pkg-config Found pkg-config: YES (/usr/local/bin/pkg-config) 2.0.3 Determining dependency 'libpcre' with pkg-config executable '/usr/local/bin/pkg-config' env[PKG_CONFIG_BUILD_TOP_DIR]: /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/tmp/sb-1001/devel/qemu env[PKG_CONFIG_PATH]: /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/tmp/sb-1001/devel/qemu/opt/work/rtems/qemu/lib/pkgconfig env[PKG_CONFIG]: /usr/local/bin/pkg-config ----------- Called: `/usr/local/bin/pkg-config --modversion libpcre` -> 1Both report no
pcre
present. The RSBpkg-config
is not the issue. The question is whyglib
and/ormeson
does not findpcre
?The
glib
support inmeson.build
is:pcre = dependency('libpcre', version: '>= 8.31', required : false) # Should check for Unicode support, too. FIXME if not pcre.found() if cc.get_id() == 'msvc' or cc.get_id() == 'clang-cl' # MSVC: Search for the PCRE library by the configuration, which corresponds # to the output of CMake builds of PCRE. Note that debugoptimized # is really a Release build with .PDB files. if vs_crt == 'debug' pcre = cc.find_library('pcred', required : false) else pcre = cc.find_library('pcre', required : false) endif endif endif # Try again with the fallback if not pcre.found() pcre = dependency('libpcre', required : true, fallback : ['libpcre', 'pcre_dep']) use_pcre_static_flag = true elif host_system == 'windows' pcre_static = cc.links('''#define PCRE_STATIC #include <pcre.h> int main() { void *p = NULL; pcre_free(p); return 0; }''', dependencies: pcre, name : 'Windows system PCRE is a static build') use_pcre_static_flag = pcre_static else use_pcre_static_flag = false endif
Author: Chris Johns
2024-05-14T04:16:16.059Z
Updating to 2.80.2 as @amar kindly pointed out has resolved the PCRE issue
Author: Amar Takhar
2024-05-14T02:18:37.043Z
Note that the former ftp.pcre.org FTP site is no longer available. You will need to update any scripts that download PCRE source code to download via HTTPS, Git, or Subversion from the new home on GitHub instead.
Author: Amar Takhar
2024-05-14T04:17:22.081Z
Author: Chris Johns
2024-05-14T04:17:22.081Z
It is not yet clear what the issue is to trigger
meson
to take the path it has. And I would discourage lettingglib
build a package like this. Too easy to be manipulated.
Author: Amar Takhar
2024-05-14T04:17:22.081Z
You mean why glib can’t find libpcre or why it’s trying to download it?
Author: Amar Takhar
2024-05-14T04:17:22.081Z
What version of glib is this?
Author: Chris Johns
2024-05-14T04:17:22.081Z
Attempting to build glib-2.69.3
Author: Chris Johns
2024-05-14T04:17:22.081Z
I wonder if
glib
wants PCRE, ie 8.45 (or something), and not PCRE2?
Author: Amar Takhar
2024-05-14T04:17:22.081Z
glib stable is at 2.80 any reason to build an old version like that? Might be why it’s failing certainly why it’s looking for an old ftp.
Author: Chris Johns
2024-05-14T04:17:22.081Z
Ah thanks and no reason other than scanning the list of files and getting the wrong one. I will take a look.
Author: Chris Johns
2024-05-14T04:17:22.081Z
@amar The
glib
site says 2.81.0 (https://docs.gtk.org/glib/) but I cannot find it in the Releases on their gitlab site. I have paged through a lot of dot releases and got bored?
Author: Amar Takhar
2024-05-14T04:17:22.081Z
I think that is the version of the documentation which is probably HEAD. 2.81 isn’t released yet:
Author: Chris Johns
2024-05-14T04:17:22.081Z
Using 2.80.2 has sorted this issue. Thanks.
Author: Chris Johns
2024-05-14T04:42:02.723Z
Fixed the RSB
pkg-config
to support--modversion
someson
is happy however there is a new issue withglib
:Found pkg-config: YES (/opt/work/chris/rtems/rsb/rtems-source-builder.git/source-builder/pkg-config) version: 1.2.3 Run-time dependency libpcre2-8 found: YES 10.43 meson.build:2214:0: ERROR: Subproject exists but has no meson.build file.
Author: Amar Takhar
2024-05-14T04:42:02.723Z
Do you have any build artifacts around? That error can happen from previous builds/downloads that have failed.
Author: Amar Takhar
2024-05-14T04:42:02.723Z
Are you using the ‘source code’ or ‘release tarball’?? Make sure you’re using the release tarball it has all the submodules already.
Author: Chris Johns
2024-05-14T04:42:02.723Z
Ah OK. I took the source tarball.
Author: Amar Takhar
2024-05-14T04:42:02.723Z
Okay probably source of the error then they don’t do a good job of making it clear you want the release tarball and not source.
Author: Chris Johns
2024-05-14T04:42:02.723Z
The released sources has resolved that issue
Author: Chris Johns
2024-05-14T04:52:24.733Z
The
glib
build has now moved onto fail with:Dependency gvdb found: YES 0.0 (overridden) Library m found: YES Run-time dependency libffi found: YES 3.0.13 Run-time dependency zlib found: YES 1.3 Run-time dependency intl found: YES Checking for function "ngettext" with dependency intl: YES Checking for function "bind_textdomain_codeset" with dependency intl: YES Checking for function "getxattr" : NO Header "attr/xattr.h" has symbol "getxattr" : NO meson.build:2318:4: ERROR: Problem encountered: No getxattr implementation found in C library or libxattr
Author: Chris Johns
2024-05-14T04:52:24.733Z
I do not have:
pxattr-2.1.0 Portable utility to work with file extended attributesinstalled. I do not like adding further dependencies into what we need.
Author: Amar Takhar
2024-05-14T04:52:24.733Z
You can pass
-Dxattr=false
To meson.
Author: Amar Takhar
2024-05-14T04:52:24.733Z
Actually the FreeBSD port may give you some ideas:
Author: Chris Johns
2024-05-14T04:52:24.733Z
Thanks, happy to borrow those options.
Author: Chris Johns
2024-05-15T02:47:29.135Z
Glib’s
meson
support does not like the staging used by the RSB. The RSB builds and installs packages into an internal temporary work space (usingDESTDIR
) and those packages are referenced usingLDFLAGS
etc when building on top of earlier packages. This is mostly working expect when runninggio/glib-compile-resources
. This tool does a compile and link as it does not include theLDFLAGS
orLD
. TheLDFLAGS
are used else where bymeson
without an issue so it seems this is specific to this type of build target. In this caseiconv
is built however there is:[986/1597] /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/glib-2.80.2-x86_64-freebsd14.0-1/glib-2.80.2/_build/gio/glib-compile-resources --compiler=clang --target=gio/tests/plugin-resources.c --sourcedir=/opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/glib-2.80.2-x86_64-freebsd14.0-1/glib-2.80.2/gio/tests --internal --generate-source --c-name _g_plugin ../gio/tests/test4.gresource.xml FAILED: gio/tests/plugin-resources.c /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/glib-2.80.2-x86_64-freebsd14.0-1/glib-2.80.2/_build/gio/glib-compile-resources --compiler=clang --target=gio/tests/plugin-resources.c --sourcedir=/opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/glib-2.80.2-x86_64-freebsd14.0-1/glib-2.80.2/gio/tests --internal --generate-source --c-name _g_plugin ../gio/tests/test4.gresource.xml ld-elf.so.1: /opt/work/chris/rtems/rsb/rtems-source-builder.git/bare/build/glib-2.80.2-x86_64-freebsd14.0-1/glib-2.80.2/_build/gio/../glib/libglib-2.0.so.0: Undefined symbol "libiconv_open"
Author: Amar Takhar
2024-05-15T02:47:29.135Z
There’s probably a different environment variable that’s used to avoid pulling in the system libraries and use the internal ones. Not sure why it isn’t adding the directories it may not be detecting them. Did you look at the FreeBSD Makefile to see what it does?
Author: Chris Johns
2024-05-15T02:47:29.135Z
It is a custom command so env handling would have to be in compiler as a child process.
While looking into the issue I notice the name
gio/tests/test4.gresource.xml
is a test so I checkedmeson_options.txt
and found I could disable the tests. The build completes andqemu
is now building.
Author: Chris Johns
2024-05-16T04:51:20.683Z
QEMU has built and install however the
iconv
shared library is not being installed:$ /opt/work/rtems/qemu/bin/qemu-system-aarch64 --version ld-elf.so.1: Shared object "libintl.so.9" not found, required by "libgio-2.0.so.0"The installed shared libraries are:
$ find /opt/work/rtems/qemu -name \\*.so /opt/work/rtems/qemu/lib/libgirepository-2.0.so /opt/work/rtems/qemu/lib/libgettextsrc.so /opt/work/rtems/qemu/lib/libpixman-1.so /opt/work/rtems/qemu/lib/libgthread-2.0.so /opt/work/rtems/qemu/lib/libffi.so /opt/work/rtems/qemu/lib/libasprintf.so /opt/work/rtems/qemu/lib/libintl.so /opt/work/rtems/qemu/lib/libgettextsrc-0.18.3.so /opt/work/rtems/qemu/lib/libgio-2.0.so /opt/work/rtems/qemu/lib/libgettextlib-0.18.3.so /opt/work/rtems/qemu/lib/libgettextlib.so /opt/work/rtems/qemu/lib/libglib-2.0.so /opt/work/rtems/qemu/lib/libgmodule-2.0.so /opt/work/rtems/qemu/lib/libgobject-2.0.so /opt/work/rtems/qemu/lib/libgettextpo.so
Author: Amar Takhar
2024-05-16T04:51:20.683Z
Is
libintl.so.9
actually there? What is the link pointing to?
Author: Chris Johns
2024-05-16T04:51:20.683Z
It is built as part of the vertical stack if not present. I am looking into where the problem could be. I think this is an RSB issue.
Author: Chris Johns
2024-05-16T04:51:20.683Z
Hmm is is not being built:
config: devel/libiconv-1.14-1.cfg config: devel/gettext-0.18.3.1-1.cfg package: gettext-0.18.3.1-x86_64-freebsd14.0-1 building: gettext-0.18.3.1-x86_64-freebsd14.0-1The package is in the stack but not built. It is only built on Windows.
Author: Amar Takhar
2024-05-16T04:51:20.683Z
If it’s not built how is there a
libintl.so
?? Is that being created somewhere else?
Author: Chris Johns
2024-05-16T04:51:20.683Z
I cleaned the build and install
${prefix}
path and it is now failing with:ld-elf.so.1: Shared object "libpixman-1.so.0" not found, required by "qemu-system-aarch64"I cannot build pixman as static and adding
--static
to the qemu build gave me an error looking forlibxml2
which I have not looked into. The shared librarylibpixman-1.so.0
is installed into${prefix}/lib
however qemu does not know to search there.
Author: Chris Johns
2024-05-16T04:51:20.683Z
The RSB configures qemu with:
--extra-ldflags="-Wl,-rpath -Wl,/$SB_PREFIX_CLEAN/lib -L$SYSROOT/lib ${VDE_LDFLAGS}"to set the search path in the executable. I am not sure why this has broken?
Author: Amar Takhar
2024-05-16T04:51:20.683Z
How come
-rpath
is empty? That might be your problem.
Author: Chris Johns
2024-05-16T04:51:20.683Z
It should not be empty as
$SB_PREFIX_CLEAN
is a clean version of the$prefix
which is/opt/work/rtems/qemu
? These options are (should be?) passed tocc
which calls theld
with-rpath /opt/work/rtems/qemu/lib
. The build system ismeson
and amake
wraper which is calling ninja. I am not yet sure how to get a verbose option into that path.Building with the path hard coded to the prefix:
--extra-ldflags="-Wl,-rpath -Wl,/opt/work/rtems/qemu/lib"results in:
$ readelf -d /opt/work/rtems/qemu/bin/qemu-system-aarch64 | grep RUNPATH $If I change the path to something else:
--extra-ldflags="-Wl,-rpath -Wl,/chris/lib"I get:
$ readelf -d /opt/work/rtems/qemu/bin/qemu-system-aarch64 | grep RUNPATH 0x000000000000001d RUNPATH Library runpath: [/chris/lib] $Something must be touching these options if the path is based on the
$prefix
is not being set. The linker does not know about prefixing.
Author: Chris Johns
2024-05-16T04:51:20.683Z
And to prove there must be a simple pattern match I have made the path
/$SB_PREFIX_CLEAN/lib/../lib
and it works:$ readelf -d /opt/work/rtems/qemu/bin/qemu-system-aarch64 | grep RUNPATH 0x000000000000001d RUNPATH Library runpath: [/opt/work/rtems/qemu/lib/../lib]Weird
Author: Chris Johns
2024-05-16T04:51:20.683Z
I think reason
libintl.so.9
was not found bylibgio-2.0.so.0
isgio
is not built with a suitable-rpath
option. I guess therpath
needs to be set for all packages in the vertical stack?
Author: Amar Takhar
2024-05-16T04:51:20.683Z
Oh sorry I was confused by the separate use of -Wl I usually use -Wl,-rpath=<path>
Yes it might need to be set for all of them but your find only showed the *.so files was libintl.so.9 actually there or another version?
Author: Chris Johns
2024-05-16T04:51:20.683Z
I am reluctant to change
-rpath /a/b
to-rpath=/a/b
given it has been working for years.Yes that library is installed but it is not found by
gpio
.Also using
/$SB_PREFIX_CLEAN/lib/
works as an-rpath
.
Author: Chris Johns
2024-05-16T04:51:20.683Z
I have added
-rpath
to the packages in the stack and it works:$ /opt/work/rtems/qemu/bin/qemu-system-aarch64 --version QEMU emulator version 5.1.91 Copyright (c) 2003-2020 Fabrice Bellard and the QEMU Project developers@amar thanks for your help and feedback, it has saved my sanity.
Author: Amar Takhar
2024-05-16T04:51:20.683Z
hah no problem I deal with this sort of thing a _lot_.
Also the only reason I use rpath=<path> is to make sure the path always sticks with the argument. I’ve had that not be the case but that was a long time ago.
Author: Chris Johns
2024-05-16T04:51:20.683Z
I agree on using
=
but I think the GNU doco was separate at the time but I cannot remember.
Author: Amar Takhar
2024-05-16T04:51:20.683Z
I think it was a bug in automake this was a really long time ago I’m sure it’s not an issue anymore but who knows..
Author: Chris Johns
2024-05-16T04:51:50.363Z
mentioned in merge request !10
1 - Migrate commits from master to main after gitlab transition¶
Id |
1 |
State |
closed |
Type |
ISSUE |
Author |
Gedare Bloom |
Assignee(s) |
Chris Johns |
Closed by |
Gedare Bloom |
Created |
2024-04-29T15:53:22.953Z |
Closed |
2024-05-02T04:57:55.003Z |
Updated |
2024-05-02T04:57:55.066Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/1 |
Merges |
2 |
Summary¶
sync commits from master
Steps to reproduce¶
Author: Gedare Bloom
2024-04-29T15:53:28.235Z
created branch 1-migrate-commits-from-master-to-main-after-gitlab-transition to address this issue
Author: Gedare Bloom
2024-04-29T15:54:05.830Z
mentioned in merge request !1
Author: Gedare Bloom
2024-05-01T16:12:37.261Z
assigned to @chris
Author: Chris Johns
2024-05-02T00:29:13.078Z
Is this issue done?
Author: Gedare Bloom
2024-05-02T00:46:20.672Z
No
Author: Gedare Bloom
2024-05-02T04:27:57.779Z
mentioned in merge request !7
18 - Try ftp.rtems.org first to download files before anywhere else (opened)¶
Id |
18 |
State |
opened |
Type |
ISSUE |
Author |
Amar Takhar |
Created |
2024-06-27T19:52:15.412Z |
Updated |
2024-07-31T18:52:34.679Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/18 |
Merges |
0 |
Summary¶
We should set the RSB to use ftp.rtems.org
first before anywhere else. Reasons:
Right now
*.gnu.org
is dead so the RSB sits and waits.It will be faster for most people but also consistent speeds.
The chance of something breaking is lower from changing URLs or remotely deleted files
If we start with RTEMS 6 we can have a subdirectory and do that going forward. Any versions before that can use the old system just fine but we should still have an archive of all files just in case the older ones go away which they usually do, eventually.
This includes patches – literally all the files. It will solve the issue of older branches breaking when we move repositories around. We’ve already spoken about a ‘legacy’ archive. GitLab does automatically put redirects in but that is not guaranteed to always exist there are cases where they get removed.
We’ll be able to support old branches with no extra work using this method.
These will also break the next time we change platforms staying on GitLab forever would be great but we don’t know what the future entails ftp.rtems.org
will always be ours to use and in our control.
Author: Amar Takhar
2024-07-31T18:51:26.683Z
changed the description
Author: Amar Takhar
2024-07-31T18:51:54.750Z
Edited the description to included downloaded diffs and patches that will break when we move repositories or change platforms.
Author: Amar Takhar
2024-07-31T18:52:34.756Z
changed the description
16 - Re-evaluate mirrors on our GitHub (opened)¶
Id |
16 |
State |
opened |
Type |
ISSUE |
Author |
Amar Takhar |
Assignee(s) |
Amar Takhar, Chris Johns |
Created |
2024-06-25T23:06:11.856Z |
Updated |
2024-06-27T05:53:08.430Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/issues/16 |
Merges |
0 |
Summary¶
We have several mirrors on our GitHub located at https://www.github.com/RTEMS/
Motivations for moving these:
It’s confusing on our GitHub page
We now have GitLab if we need to run our own patches it would be easier if we mirror it there so we can fork it.
At least GCC has their own official GitHub mirror now and the other two are small.
Steps to reproduce¶
Author: Amar Takhar
2024-06-25T23:06:20.654Z
assigned to @chris and @amar
Author: Sebastian Huber
2024-06-27T05:53:08.442Z
The GCC mirror contains branches from the unofficial SVN GCC Git mirror. In general, these mirrors are no longer needed except if you want to rebuild older tool chains.
Author: Amar Takhar
2024-06-27T05:53:08.405Z
Thanks the current plan is to archive them so they don’t show up on our page this will keep them intact for older tool chains.
I know GCC has their own official mirror on git:
But it seems sourceware doe snot. We could mirror the sourceware repositories on our GitLab instance if we need to. I’d first like to see how stable the current mirrors are seems they use HTTP cloning now which is better.
Merge Requests¶
41 - Convert LICENSE to MarkDown¶
Id |
41 |
State |
merged |
Merged by |
Kinsey Moore |
Merged at |
2024-07-25T22:32:43.392Z |
Assignee(s) |
Amar Takhar |
Created |
2024-07-20T04:59:00.116Z |
Updated |
2024-07-25T22:32:45.164Z |
Milestone |
6.1 |
Labels |
doc |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/41 |
Commit |
Summary¶
Author: Amar Takhar
2024-07-20T04:59:00.634Z
assigned to @amar
Author: Kinsey Moore
2024-07-25T22:32:29.547Z
added 2 commits
8d70e4b5 - 1 commit from branch
rtems/tools:main
892e966e - Convert LICENSE to MarkDown
Author: Kinsey Moore
2024-07-25T22:32:40.087Z
approved this merge request
29 - Fix the handling of the hash in sb-rtems-pkg¶
Id |
29 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-15T18:26:25.386Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-15T05:26:37.423Z |
Updated |
2024-06-15T18:26:27.273Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/29 |
Commit |
Fix the handling of the hash in sb-rtems-pkg
. The logic was confused of the tar file name was in another macro. Expand the macros and check for the package name.
Close #13
Author: Chris Johns
2024-06-15T05:26:37.569Z
assigned to @chris
Author: Amar Takhar
2024-06-15T18:26:18.385Z
added 2 commits
a961b7c5 - 1 commit from branch
rtems/tools:main
d282a172 - sb/rtemspkg: Fix the handling of the hash
Author: Amar Takhar
2024-06-15T18:26:22.089Z
approved this merge request
30 - rtems/tools,kernel: Update tools after elftoolchain update¶
Id |
30 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-15T06:57:34.690Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-15T05:34:23.681Z |
Updated |
2024-06-15T06:57:35.796Z |
Milestone |
6.1 |
Labels |
rtems::kernel, tool::rtems-tools |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/30 |
Commit |
Update tools after elftoolchain update
Pick up the latest kernel changes
Author: Chris Johns
2024-06-15T05:34:23.857Z
assigned to @chris
Author: Amar Takhar
2024-06-15T06:57:31.704Z
approved this merge request
28 - RSB tar strip components double percent¶
Id |
28 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-14T05:07:38.037Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-14T04:34:06.598Z |
Updated |
2024-06-14T05:07:39.100Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/28 |
Commit |
A bug was introduced when lowering the tar output in the log
Author: Chris Johns
2024-06-14T04:34:06.736Z
assigned to @chris
Author: Amar Takhar
2024-06-14T05:07:34.147Z
approved this merge request
27 - rtems-tools-6.cfg: Checksum updated¶
Id |
27 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-13T17:03:37.914Z |
Assignee(s) |
Joel Sherrill |
Created |
2024-06-13T15:41:54.484Z |
Updated |
2024-06-14T04:08:05.369Z |
Milestone |
6.1 |
Labels |
tool::rtems-tools |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/27 |
Commit |
Summary¶
rtems-tools-6.cfg: Checksum updated
The hash was updated without the checksum being updated.
Author: Joel Sherrill
2024-06-13T15:41:54.643Z
assigned to @joel
Author: Amar Takhar
2024-06-13T17:03:34.057Z
approved this merge request
Author: Chris Johns
2024-06-14T04:08:05.352Z
This checksum is automatically generated in the
sb-rtems-pkg
tool. This means something is broken in that tool.
26 - Update RTEMS packages after moving to gitlab.rtems.org¶
Id |
26 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-13T07:28:05.156Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-13T07:17:52.631Z |
Updated |
2024-06-13T07:28:06.311Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/26 |
Commit |
Summary¶
This merge request updates the RTEMS packages after moving to gitlab.rtems.org
Author: Chris Johns
2024-06-13T07:17:52.762Z
assigned to @chris
Author: Chris Johns
2024-06-13T07:23:56.857Z
added 2 commits
86016e9a - 1 commit from branch
rtems/tools:main
87a855a7 - rtems: Update RTMES packages after moving to gitlab.rtems.org
Author: Chris Johns
2024-06-13T07:24:25.875Z
changed title from Update RT{-M-}ES packages after moving to gitlab.rtems.org to Update RTE{+M+}S packages after moving to gitlab.rtems.org
Author: Amar Takhar
2024-06-13T07:28:00.134Z
approved this merge request
25 - sb/rtemspkg: Use git:// paths for download URLs¶
Id |
25 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-13T07:18:11.809Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-13T07:13:07.911Z |
Updated |
2024-06-13T07:18:13.372Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/25 |
Commit |
Use git://
paths for download URLs
Add protocol=https
to support gitlab.rtems.org
Close #12
Author: Chris Johns
2024-06-13T07:13:08.088Z
assigned to @chris
Author: Amar Takhar
2024-06-13T07:18:08.291Z
approved this merge request
21 - Fix MacOS python patch to the GDB 14.2 build and reduce down tar output¶
Id |
21 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-11T04:48:36.449Z |
Assignee(s) |
Chris Johns |
Created |
2024-06-11T04:34:38.110Z |
Updated |
2024-06-12T01:01:30.448Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/21 |
Commit |
Add the MacOS python patch to the GDB 14.2 build
No verbose tar output when
--trace
is not used
Closes #9
Author: Chris Johns
2024-06-11T04:34:38.258Z
assigned to @chris
Author: Amar Takhar
2024-06-11T04:48:20.561Z
Are these changes and the ones below intentional? Is it a different change that slipped through?
Author: Chris Johns
2024-06-11T04:48:20.561Z
Forgot the change was in the repo. I have updated the MR details to reflect the changed.
Author: Chris Johns
2024-06-11T04:47:34.175Z
changed title from {-rtems/6: Add the-} MacOS python patch to the GDB 14.2 build to {+Fix+} MacOS python patch to the GDB 14.2 build{+ and reduce down tar output+}
Author: Chris Johns
2024-06-11T04:47:34.203Z
changed the description
Author: Chris Johns
2024-06-11T04:48:20.601Z
resolved all threads
Author: Amar Takhar
2024-06-11T04:48:30.371Z
approved this merge request
Author: Chris Johns
2024-06-12T01:01:30.426Z
mentioned in issue #11
16 - gdb: Build MPFR before GDB as it now depeneds on it¶
Id |
16 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-06-08T19:42:26.396Z |
Assignee(s) |
Chris Johns |
Reviewer(s) |
Joel Sherrill, Amar Takhar |
Created |
2024-06-08T00:47:02.398Z |
Updated |
2024-06-08T19:42:28.385Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/16 |
Commit |
Summary¶
Build MPFR before GDB as it now depeneds on it
Closes #7
Author: Chris Johns
2024-06-08T00:47:02.540Z
requested review from @amar and @joel
Author: Chris Johns
2024-06-08T00:47:02.556Z
assigned to @chris
Author: Chris Johns
2024-06-08T01:18:12.315Z
added 11 commits
f1bfcd6e…f39e9584 - 10 commits from branch
rtems/tools:main
148922b4 - gdb: Build MPFR before GDB as it now depeneds on it
Author: Chris Johns
2024-06-08T04:35:59.768Z
added 1 commit
423d98d2 - gdb: Build MPFR before GDB as it now depeneds on it
Author: Chris Johns
2024-06-08T04:37:10.262Z
This MR builds Expat, GMP and MPFR as internal libraries for static linking
Author: Chris Johns
2024-06-08T04:44:17.260Z
mentioned in issue #7
Author: Amar Takhar
2024-06-08T19:42:20.345Z
This builds and works fine now.
Author: Amar Takhar
2024-06-08T19:42:22.698Z
approved this merge request
15 - Update for gitlab changes and fix minor issues¶
Id |
15 |
State |
merged |
Merged by |
Chris Johns |
Merged at |
2024-06-06T23:25:31.643Z |
Assignee(s) |
Kinsey Moore |
Reviewer(s) |
Chris Johns |
Created |
2024-06-05T18:20:12.821Z |
Updated |
2024-06-06T23:26:44.870Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/15 |
Commit |
This MR is to fix the gitlab migration issues for main (6/7) branch for #3
The rtems_waf changes and the rtemspkg.py changes in particular need additional eyes as I’m not familiar with all parts of the RSB workflow.
Author: Kinsey Moore
2024-06-05T18:20:12.973Z
assigned to @opticron
Author: Kinsey Moore
2024-06-05T18:22:05.654Z
mentioned in issue #3
Author: Kinsey Moore
2024-06-05T18:25:08.992Z
added 16 commits
3fc04f12…108b765c - 7 commits from branch
rtems/tools:main
9b231161 - rtems/config: Cleanup orphaned configs
1039f6af - sis: Handle renaming
892e6217 - rtems-tools: Update location to gitlab
bcd838fc - config/rtems-kernel: Update error text
ff51caa9 - config/rtems-kernel: Update for gitlab
31602629 - config/rtems-libbsd: Update for gitlab
55084f6f - config/rtems-net-legacy: Update for gitlab
0ffb6192 - config/rtems-net-services: Update for gitlab
514e6253 - config: Update rtems_waf for gitlab
Author: Kinsey Moore
2024-06-05T19:25:19.809Z
requested review from @chris
Author: Kinsey Moore
2024-06-05T20:50:25.012Z
It looks like the libffi patch path isn’t quite right.
Author: Kinsey Moore
2024-06-05T20:55:25.360Z
added 9 commits
c261531d - rtems/config: Cleanup orphaned configs
263444aa - sis: Handle renaming
819d5e92 - rtems-tools: Update location to gitlab
3fdb8e9f - config/rtems-kernel: Update error text
b7be268a - config/rtems-kernel: Update for gitlab
29396592 - config/rtems-libbsd: Update for gitlab
8a4c26f7 - config/rtems-net-legacy: Update for gitlab
d68c65b2 - config/rtems-net-services: Update for gitlab
10f494df - config: Update rtems_waf for gitlab
Author: Kinsey Moore
2024-06-05T20:56:13.597Z
I had a stale patch in patches that prevented it from picking up the failure to find the patch on the 6 branch that doesn’t exist yet. The updated patch set fixes the libffi patch download issue.
Author: Kinsey Moore
2024-06-05T21:04:15.822Z
added 1 commit
227582a6 - Migrate devel links to gitlab
Author: Kinsey Moore
2024-06-05T21:04:36.118Z
The final patch also addresses #4
Author: Kinsey Moore
2024-06-06T14:22:12.671Z
Also of note, I don’t have a lot of the platforms that these patches affect, so I can’t verify the download of patches for things like darwin.
Author: Chris Johns
2024-06-06T14:22:12.671Z
The
sb-get-source --stop-on-error
is your friend here. It will make sure all build sets and so all configurations are run through all possible host types. The--stop-on-error
option is needed or you miss issues.Note, delete the
sources
andpatches
directories before running or the RSB will see the cached files and assume they are all OK.
Author: Kinsey Moore
2024-06-06T14:22:12.671Z
I have updated the one hash that threw an error.
Author: Chris Johns
2024-06-06T00:50:46.658Z
marked this merge request as draft
Author: Chris Johns
2024-06-06T14:24:41.075Z
I suspect this change in this
.bset
file is not accessible in.cfg
files so I wonder if this file is copied tortems-urls.cfg
andrtems-urls.bset
includes the.cfg
file? We may not need a.bset
version if no extension is added to the include but I would need to check this to be certain.Over time references to our git repos have been added to a number of config files. I am fine with the changes in this MR staying as they are and then a separate issue added to clean up the build set and config files to have a single reference if that is too much for this change?
Author: Kinsey Moore
2024-06-06T14:24:41.075Z
I’d prefer the cleanup to happen in a different MR/issue. I definitely noticed that the defined URLs were not consistently used or available and were often redefined instead of pulling in the global set.
Author: Chris Johns
2024-06-06T01:07:05.678Z
@opticron awesome set of changes, thank you. I have made the MR draft until the
sb-get-sources
command has been run and completes. The https://gitlab.rtems.org/administration/integration/-/issues/6 issue will use this command to update the cache on ftp.rtems.org so it will be checked once we have the CI running.
Author: Kinsey Moore
2024-06-06T14:21:36.053Z
added 2 commits
e4a69442 - config: Update rtems_waf for gitlab
1549a7b9 - Migrate devel links to gitlab
Author: Kinsey Moore
2024-06-06T14:24:41.091Z
resolved all threads
Author: Kinsey Moore
2024-06-06T14:24:57.667Z
marked this merge request as ready
Author: Kinsey Moore
2024-06-06T23:10:13.801Z
added 2 commits
0ed7e8c6 - config: Update rtems_waf for gitlab
f39e9584 - Migrate devel links to gitlab
Author: Kinsey Moore
2024-06-06T23:11:33.510Z
The patch set has been updated for the rtems-waf -> rtems_waf renaming and passes sb-get-sources.
Author: Chris Johns
2024-06-06T23:24:45.448Z
I have also done my testing and it looks good. I have found an unrelated bug I will post in a separate merge request. Thanks for the change.
Author: Chris Johns
2024-06-06T23:24:55.676Z
approved this merge request
Author: Kinsey Moore
2024-06-06T23:26:44.724Z
mentioned in issue #4
13 - Update binutils, gdb, and gcc¶
Id |
13 |
State |
merged |
Merged by |
Chris Johns |
Merged at |
2024-06-03T21:36:14.381Z |
Assignee(s) |
Joel Sherrill |
Created |
2024-06-01T16:21:13.425Z |
Updated |
2024-06-03T21:36:15.618Z |
Milestone |
6.1 |
Labels |
tool::binutils |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/13 |
Commit |
Summary¶
Update binutils, gdb, and gcc
Update to binutils 2.41
Update to gdb 14.2
Update to gcc 13.3
updates #4768
Author: Joel Sherrill
2024-06-01T16:21:13.597Z
assigned to @joel
Author: Chris Johns
2024-06-03T21:36:09.620Z
approved this merge request
12 - Format all code with yapf defaults¶
Id |
12 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-05-25T21:48:38.307Z |
Assignee(s) |
Chris Johns |
Created |
2024-05-25T05:17:51.466Z |
Updated |
2024-05-25T21:48:39.157Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/12 |
Commit |
Summary¶
Format the RSB source code with yapf
defaults to meet the coding standard’s Python formatting rules
Author: Chris Johns
2024-05-25T05:17:51.763Z
assigned to @chris
Author: Amar Takhar
2024-05-25T21:48:34.994Z
approved this merge request
11 - Fix –url paths so they download the –rsb-file¶
Id |
11 |
State |
merged |
Merged by |
Chris Johns |
Merged at |
2024-05-23T22:08:28.262Z |
Assignee(s) |
Chris Johns |
Created |
2024-05-17T06:03:32.582Z |
Updated |
2024-05-23T22:08:29.517Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/11 |
Commit |
Summary¶
Fix --url
paths so they download the --rsb-file
This lets --url
point to a cache of files
Closes #5
Author: Chris Johns
2024-05-17T06:03:32.740Z
assigned to @chris
Author: Chris Johns
2024-05-17T06:04:10.123Z
added 3 commits
229afea7…484c7a40 - 2 commits from branch
rtems/tools:main
d8ac75e2 - sb/download: Fix –url paths so they download the –rsb-file
Author: Chris Johns
2024-05-17T06:04:59.528Z
marked this merge request as draft
Author: Chris Johns
2024-05-17T06:49:46.811Z
added 1 commit
a3819b8e - sb/download: Fix –url paths so they download the –rsb-file
Author: Chris Johns
2024-05-17T06:50:14.914Z
marked this merge request as ready
Author: Gedare Bloom
2024-05-23T20:57:55.882Z
approved this merge request
10 - Update QEMU to build on FreeBSD 14.0¶
Id |
10 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-05-16T05:07:07.000Z |
Assignee(s) |
Chris Johns |
Created |
2024-05-16T04:51:47.589Z |
Updated |
2024-05-16T05:07:08.068Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/10 |
Commit |
Update QEMU¶
This update allows QEMU to build on FreeBSD. The updates are:
sb/pkg-config
:Support
--version
. Meson uses--version
to determine ifpkg-config
is valid.Supoort
--modversion
. Meson uses--modversion
to determine the version of a library.Update
pixman
because it did not buildUpdate
glib
because it did not build. This meant adding support formeson
.Update
qemu
build to resolve shared libraries. For some reasonqemu
does not accept an-rpath
that is$prefix/lib
. The option seems to be removed. Add a trailing/
to the path letsqemu
keep the-rpath
option and the executable can find the installed shared libraries.Add
-rpath
to all the packages in the vertical stack so any shared libraries loaded can find other shared libraries they depend on
Closes #2
Author: Chris Johns
2024-05-16T04:51:47.748Z
assigned to @chris
Author: Chris Johns
2024-05-16T04:59:38.773Z
added 15 commits
20ede602…35ef965f - 13 commits from branch
rtems/tools:main
89a24469 - sb/pkg-config: Support –version and –modversion
484c7a40 - bare/devel/qemu: Update to glib and pixman to build QEMU
Author: Chris Johns
2024-05-16T05:01:46.222Z
changed the description
Author: Chris Johns
2024-05-16T05:02:25.991Z
changed the description
Author: Amar Takhar
2024-05-16T05:06:54.570Z
approved this merge request
9 - qemu-xilinx: Allow build with conflicting libspice¶
Id |
9 |
State |
merged |
Merged by |
Chris Johns |
Merged at |
2024-05-07T21:11:17.532Z |
Created |
2024-05-07T21:08:30.689Z |
Updated |
2024-05-07T21:11:18.461Z |
Milestone |
6.1 |
Labels |
arch:aarch64 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/9 |
Commit |
Summary¶
qemu-xilinx: Allow build with conflicting libspice
Allow Xilinx QEMU to build on systems where there is a version of libspice available that isn’t what the QEMU build system expects. This resolves linking errors against libspice and libgstreamer.
Author: Chris Johns
2024-05-07T21:11:10.847Z
approved this merge request
8 - Update Newlib, Binutils, GDB, GCC for RTEMS 6/7¶
Id |
8 |
State |
merged |
Merged by |
Chris Johns |
Merged at |
2024-05-07T05:09:19.478Z |
Created |
2024-05-02T14:56:44.484Z |
Updated |
2024-05-07T07:35:45.464Z |
Milestone |
6.1 |
Labels |
tool::rtems-source-builder |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/8 |
Commit |
Summary¶
Keep RTEMS up to date with the upstream development.
Author: Chris Johns
2024-05-07T07:35:46.183Z
Is the NIOS2 going to be 7?
Author: Sebastian Huber
2024-05-07T07:35:46.183Z
The nios2 target is still supported in GCC 14. If we ship RTEMS 7 with GCC 14, then I would say yes, with a later GCC no.
Author: Chris Johns
2024-05-07T07:35:46.183Z
We need to announce removal of an arch in 7 when we release 6. If we do not ship GCC 14 with 7 we could let the nios2 stay on GCC 14 for 7 and we announce it’s removal?
Author: Sebastian Huber
2024-05-07T07:35:46.183Z
I am not sure why you rush for the nios2 removal. Currently it builds fine.
Author: Chris Johns
2024-05-07T07:35:46.183Z
We track GCC and this merge request enables obsolete architectures. That raises a warning to me.
We cannot just remove the BSP as we have a policy of tagging it for removal in one release for removal in the next. If we are not proactive in managing this we end up carrying dead architectures. We have a few now.
Author: Sebastian Huber
2024-05-07T07:35:46.183Z
The nios2 target is definitely not recommended for new designs. We can still support it on a best effort basis. Right now the tools are fine.
Author: Chris Johns
2024-05-07T05:08:41.254Z
approved this merge request
Author: Chris Johns
2024-05-07T05:09:10.702Z
resolved all threads
7 - sync in changes from master¶
Id |
7 |
State |
merged |
Merged by |
Gedare Bloom |
Merged at |
2024-05-02T04:57:54.704Z |
Assignee(s) |
Gedare Bloom |
Reviewer(s) |
Chris Johns |
Created |
2024-05-02T04:27:52.540Z |
Updated |
2024-05-02T04:57:56.894Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/7 |
Commit |
Summary¶
Replay commits from master that were missed during the transition to gitlab.
Closes #1
Author: Gedare Bloom
2024-05-02T04:27:52.990Z
requested review from @chris
Author: Gedare Bloom
2024-05-02T04:27:53.027Z
assigned to @gedare
Author: Chris Johns
2024-05-02T04:43:33.841Z
approved this merge request
2 - Change README to markdown¶
Id |
2 |
State |
merged |
Merged by |
Amar Takhar |
Merged at |
2024-05-01T03:42:13.781Z |
Assignee(s) |
Chris Johns |
Reviewer(s) |
Amar Takhar |
Created |
2024-05-01T02:35:15.609Z |
Updated |
2024-05-01T03:42:14.724Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/2 |
Commit |
Change README to markdown¶
Author: Chris Johns
2024-05-01T02:35:15.724Z
requested review from @amar
Author: Chris Johns
2024-05-01T02:35:15.755Z
assigned to @chris
Author: Amar Takhar
2024-05-01T03:42:07.211Z
Sorry for the noise we were testing the CODEOWNER setup.
Author: Amar Takhar
2024-05-01T03:42:08.881Z
approved this merge request
1 - Resolve “Migrate commits from master to main after gitlab transition” (closed)¶
Id |
1 |
State |
closed |
Assignee(s) |
Gedare Bloom |
Reviewer(s) |
Chris Johns |
Closed by |
Gedare Bloom |
Created |
2024-04-29T15:54:04.265Z |
Closed |
2024-05-01T16:10:10.483Z |
Updated |
2024-05-01T16:10:10.469Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/1 |
Commit |
Summary¶
Closes #1
Author: Gedare Bloom
2024-04-29T15:54:04.396Z
requested review from @chris
Author: Gedare Bloom
2024-04-29T15:54:04.428Z
assigned to @gedare
Author: Gedare Bloom
2024-04-29T15:54:35.486Z
added 8 commits
e959a3e7 - rtems-tools-6.cfg: Bump hash to account for 5 months of changes
35493576 - rtems-gcc-*-newlib-head.cfg: Bump hash for newlib
b0b7454c - rtems: Update RTEMS packages to their HEAD
00344e21 - sb: Add sb-rtems-pkg to update the RTEMS package hashes and checksums
801daf78 - glib: Update to 2.56.4
c479ed3a - qemu-xilinx: Update to 2023.2
2a3323cb - rtems: Checksum fixes
92265b77 - 6/rtems-gcc: Revert to the gcc-13 release branch for MacOS fixes
Author: Gedare Bloom
2024-04-29T15:54:47.482Z
marked this merge request as ready
5 - rsb: Minor README change (closed)¶
Id |
5 |
State |
closed |
Assignee(s) |
Chris Johns |
Reviewer(s) |
Amar Takhar |
Closed by |
Chris Johns |
Created |
2024-05-01T04:30:37.579Z |
Closed |
2024-05-01T04:32:59.718Z |
Updated |
2024-05-01T04:32:59.709Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/5 |
Commit |
Summary¶
Author: Chris Johns
2024-05-01T04:30:37.697Z
requested review from @amar
Author: Chris Johns
2024-05-01T04:30:37.717Z
assigned to @chris
4 - rsb: Minor README change (closed)¶
Id |
4 |
State |
closed |
Assignee(s) |
Chris Johns |
Reviewer(s) |
Amar Takhar |
Closed by |
Chris Johns |
Created |
2024-05-01T04:24:31.506Z |
Closed |
2024-05-01T04:27:01.044Z |
Updated |
2024-05-01T04:27:01.034Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/4 |
Commit |
Minor README update¶
Author: Chris Johns
2024-05-01T04:24:31.619Z
requested review from @amar
Author: Chris Johns
2024-05-01T04:24:31.639Z
assigned to @chris
3 - rsb: Minor README change (closed)¶
Id |
3 |
State |
closed |
Assignee(s) |
Chris Johns |
Reviewer(s) |
Amar Takhar |
Closed by |
Chris Johns |
Created |
2024-05-01T04:02:14.215Z |
Closed |
2024-05-01T04:22:02.238Z |
Updated |
2024-05-01T04:22:02.224Z |
Milestone |
6.1 |
Link |
https://gitlab.rtems.org/rtems/tools/rtems-source-builder/-/merge_requests/3 |
Commit |
Minor change to the README¶
This is part of the GitLab testing
Author: Chris Johns
2024-05-01T04:02:14.335Z
requested review from @amar
Author: Chris Johns
2024-05-01T04:02:14.479Z
assigned to @chris
Author: Chris Johns
2024-05-01T04:19:06.422Z
Closing to remake for testing.
Author: Chris Johns
2024-05-01T04:22:01.711Z
Closing. Part of testing.