gmane.comp.lib.uclibc.buildroot

Re: [PATCH 1/2] vala: do not append a -dirty suffix tothe valac version number

Peter Korsgaard — 2012-05-18T08:44:58

 spdawson> From: Simon Dawson <spdawson-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>
 spdawson> Signed-off-by: Simon Dawson <spdawson-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>
 spdawson> ---
 spdawson>  package/vala/vala-fix-dirty-version-suffix.patch |   15 +++++++++++++++
 spdawson>  1 file changed, 15 insertions(+)
 spdawson>  create mode 100644 package/vala/vala-fix-dirty-version-suffix.patch

 spdawson> diff --git a/package/vala/vala-fix-dirty-version-suffix.patch b/package/vala/vala-fix-dirty-version-suffix.patch
 spdawson> new file mode 100644
 spdawson> index 0000000..4b0044d
 spdawson> --- /dev/null
 spdawson> +++ b/package/vala/vala-fix-dirty-version-suffix.patch
 spdawson> < at >< at > -0,0 +1,15 < at >< at >
 spdawson> +Do not append a -dirty suffix to the valac version number. The presence of
 spdawson> +the suffix can caused problems for packages (e.g. midori) which check the
 spdawson> +valac version number during configuration.

Why is this needed? We don't apply any patches to vala, and I don't see
any "-dirty" suf

[git commit] ccache: allow dynamic selection of cachedirectory

Peter Korsgaard — 2012-05-18T08:42:29

commit: http://git.buildroot.net/buildroot/commit/?id=433290761fceb476b095548eec10adf72405e050
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

The existing ccache infrastructure sets the cache directory hardcoded in the
ccache binary. As this directory was set to ~/.buildroot-ccache, the cache
is not necessarily local (e.g. in corporate environments the home directories
may be mounted over NFS.)
Previous versions of buildroot did allow to set the cache directory, but this
was also hardcoded (so you had to rebuild ccache to change it), plus that
support was removed.
See http://lists.busybox.net/pipermail/buildroot/2011-July/044511.html for
a discussion on this.

This patch modifies ccache to respect a new shell variable (exported from
the Makefile, based on a configuration option) instead of CCACHE_DIR.
The name CCACHE_DIR itself is already used by autotargets for the ccache
package.

Signed-off-by: Thomas De Schampheleire <thomas.de.schampheleire-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.o

Re: [PATCH v3] ccache: allow dynamic selection of cachedirectory

Peter Korsgaard — 2012-05-18T08:43:42

 Thomas> The existing ccache infrastructure sets the cache directory
 Thomas> hardcoded in the ccache binary. As this directory was set to
 Thomas> ~/.buildroot-ccache, the cache is not necessarily local
 Thomas> (e.g. in corporate environments the home directories may be
 Thomas> mounted over NFS.)  Previous versions of buildroot did allow to
 Thomas> set the cache directory, but this was also hardcoded (so you
 Thomas> had to rebuild ccache to change it), plus that support was
 Thomas> removed.  See
 Thomas> http://lists.busybox.net/pipermail/buildroot/2011-July/044511.html
 Thomas> for a discussion on this.

Committed, thanks.

[autobuild.buildroot.net] Build results for 2012-05-17

Thomas Petazzoni — 2012-05-18T06:30:56

Hello,

On 2012-05-17, 109 random build tests have been done and
submitted on autobuild.buildroot.net.
 91 builds have been successful
 18 builds have failed

Build de781a2c5be1f1fb537263d9125b64f0f0c22acc
==============================================

Status         : OK
Architecture   : i686
Submitted by   : Peter Korsgaard (gcc110)
Submitted at   : 2012-05-17 00:17:00
Git commit ID  : http://git.buildroot.net/buildroot/commit/?id=cb30e130519fcb50e1a2e0d7a02aa628e52ed24c
End of log     : http://autobuild.buildroot.net/results/de781a2c5be1f1fb537263d9125b64f0f0c22acc/build-end.log
Complete log   : http://autobuild.buildroot.net/results/de781a2c5be1f1fb537263d9125b64f0f0c22acc/build.log.bz2
Configuration  : http://autobuild.buildroot.net/results/de781a2c5be1f1fb537263d9125b64f0f0c22acc/config
Defconfig      : http://autobuild.buildroot.net/results/de781a2c5be1f1fb537263d9125b64f0f0c22acc/defconfig

Build 7ef0e6cbabbaf403440b32fedcc74ac011863a2a
==============================================

Status

Re: [PATCH] gdb: adjust behavior when thread debug isnot available

Thomas Petazzoni — 2012-05-18T06:21:49

Le Thu, 17 May 2012 23:54:06 +0200,
Peter Korsgaard <jacmet< at >uclibc.org> a écrit :


Will look at it, yes.

Thanks,

Thomas

Re: connman requirement for resolver in toolchain

Michael J. Hammel — 2012-05-18T04:38:21

Howdy Thomas.


Oops. That's what I missed.  I didn't check my uClibc config.  I only
looked at Crosstool-NG's config (well, and Buildroot's).  I'll check
uClibc.


Yup.  I just have to remember which ones to check.  :-)

Thanks again Thomas.

Re: dbus missing /var/lib/dbus directory

Michael J. Hammel — 2012-05-18T04:33:24


Ah.  There's the reason.  I wipe /tmp on boot so I can create a custom
directory structure in a tmpfs.  I'll add support for /var/lib/dbus
there if I decide to keep dbus.

Thanks for the tip Thomas.

[Bug 5222] pango1.14.x build failed for ARM

2012-05-17T23:37:33

https://bugs.busybox.net/show_bug.cgi?id=5222

--- Comment #3 from rdghx <rdghx-9Onoh4P/yGk< at >public.gmane.org> 2012-05-17 23:37:32 UTC ---
1. first buliding error is in host-libpng-1.14.11
   >>> host-libpng 1.4.11 Building
PATH=/usr/work/buildroot/buildroot-2012.02/output/host/bin:/usr/work/buildroot/buildroot-2012.02/output/host/usr/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/opt/FriendlyARM/toolschain/4.5.1/bin
LD_LIBRARY_PATH="/usr/work/buildroot/buildroot-2012.02/output/host/usr/lib:"
PKG_CONFIG="/usr/work/buildroot/buildroot-2012.02/output/host/usr/bin/pkg-config"
PKG_CONFIG_LIBDIR="/usr/work/buildroot/buildroot-2012.02/output/host/usr/lib/pkgconfig"
PERLLIB="/usr/work/buildroot/buildroot-2012.02/output/host/usr/lib/perl" 
/usr/bin/make -j2  -C
/usr/work/buildroot/buildroot-2012.02/output/build/host-libpng-1.4.11/
make[1]: Entering directory
`/usr/work/buildroot/buildroot-2012.02/output/build/host-libpng-1.4.11'
/usr/bin/make  all-am
make[2]: Entering directory
`/usr/wor

Re: Building 2 root filesystems with Buildroot

Peter Korsgaard — 2012-05-17T22:22:51

Hi,

Please ask buildroot questions on the buildroot list instead of mailing
me directly.

 Chung> Hello,
 Chung> I am trying to get Buildroot to build 2 filesystems; a smaller one
 Chung> with the system files and essentials, and a larger one to hold the big
 Chung> packages (i.e. Qt).

 Chung> The idea is that the kernel will boot up using the smaller one, then
 Chung> mount the larger one during the final stages of initialization. And
 Chung> the two filesystems, or partitions, may be of different filesystem
 Chung> type (i.e. one is yaffs, while the other is ext3).

 Chung> What is the most elegant way to do this in Buildroot?

 Chung> To my knowledge, Buildroot is hardwired to create a system with
 Chung> only one filesystem partition, so some customization is
 Chung> evitable. But I just wonder if you have any suggestions.

Buildroot indeed only handles a single filesystem, so I think the
easiest is to just build a single big filesystem with buildroot (E.G. a
tarball), and then write a small script to

[git commit] docs/news.html: add 2012.05-rc1announcement link

Peter Korsgaard — 2012-05-17T21:59:21

commit: http://git.buildroot.net/buildroot/commit/?id=4d1c69dd1021b0e94930c97f6f8af7bdff2fd667
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

Signed-off-by: Peter Korsgaard <jacmet-OfajU3CKLf1/SzgSGea1oA< at >public.gmane.org>
---
 docs/news.html |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/docs/news.html b/docs/news.html
index 8c4871f..0032dfa 100644
--- a/docs/news.html
+++ b/docs/news.html
< at >< at > -10,7 +10,8 < at >< at >
     <p>We have a new release candidate! Lots of changes all over the
     tree, see the <a
     href="http://git.buildroot.net/buildroot/plain/CHANGES?id=2012.05_rc1">CHANGES</a>
-    file for details</p>
+    file for details, and read the
+    <a href="http://lists.busybox.net/pipermail/buildroot/2012-May/053736.html">announcement</a>.</p>
 
     <p>Head to the <a href="/downloads/">downloads page</a> to pick up the
     <a href="/downloads/buildroot-2012.05-rc1.tar.bz2">2012.05-rc1

Re: [PATCH 1/2] socat: fix build failure due toext2_fs.h header, bump version

Peter Korsgaard — 2012-05-17T21:46:50

 Thomas> In kernel >= 3.3, the ext2_fs.h can no longer be imported from
 Thomas> userspace. This has been fixed for internal toolchains by adding a
 Thomas> patch to kernel headers, but this doesn't work with toolchains
 Thomas> generated by Crosstool-NG, or potentially upcoming external
 Thomas> toolchains.

 Thomas> socat in fact has a test in its configure.in, but the configure was
 Thomas> generated too long ago, and the generated test relies on the
 Thomas> preprocessor result and not the compiler result (but warns that in the
 Thomas> future, the compiler result will be used instead of the preprocessor
 Thomas> result).

 Thomas> So, by running autoconf on this package, we fix the problem: it
 Thomas> properly checks whether ext2_fs is usable or not, and acts
 Thomas> accordingly. Of course, it means that with recent versions of the
 Thomas> kernel, ext2-specific features of socat are unavailable, and we'll
 Thomas> have to wait for the socat developers to adapt their code so that they
 Thomas> use th

Re: [PATCH] gdb: adjust behavior when thread debug isnot available

Peter Korsgaard — 2012-05-17T21:54:06

 Thomas> When an external toolchain without thread debug is used, the gdb
 Thomas> package can be selected, but no version can be choosen, since none
 Thomas> match any of the requirements. This leads Buildroot to try to build
 Thomas> gdb for the target without a version being defined, as in the
 Thomas> following build log:

 Thomas>  http://autobuild.buildroot.org/results/84e8fd2df0cc22448052a572c2e9a6e03dd137eb/build-end.log

 Thomas> To fix this, we adjust the dependencies of the BR2_PACKAGE_GDB option
 Thomas> so that the package as a whole is not selectable when the required
 Thomas> conditions are not met. Basically, we have the choice of:

 Thomas>  * Having a toolchain that supports thread debugging, which is needed
 Thomas>    for gdb >= 7.x

 Thomas>  * Having BR2_DEPRECATED enabled, which allows gdb 6.8 to be selected,
 Thomas>    which doesn't require thread debugging

 Thomas>  * Using either avr32 of bfin, since those architectures have special
 Thomas>    old gdb versions

Thanks, but this

[git commit] apr: requires shared library support

Peter Korsgaard — 2012-05-17T21:42:26

commit: http://git.buildroot.net/buildroot/commit/?id=02b00dc9dfb7216145277a303c7c2f52004f85b2
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

The apr library requires shared library support unconditionally, so
make it depend on !BR2_PREFER_STATIC_LIB, so that architectures that
support static libraries only do not make this package available.

Solves
http://autobuild.buildroot.org/results/f4cd9751e85e9ee7fde2e9479e0f01ab2be93e84/build-end.log.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8< at >public.gmane.org>
Signed-off-by: Peter Korsgaard <jacmet-OfajU3CKLf1/SzgSGea1oA< at >public.gmane.org>
---
 package/apr-util/Config.in |    2 ++
 package/apr/Config.in      |    2 ++
 2 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/package/apr-util/Config.in b/package/apr-util/Config.in
index 99b32f2..895dbb8 100644
--- a/package/apr-util/Config.in
+++ b/package/apr-util/Config.in
< at >< at > -5,6 +5,8 < at >< at > config BR2_PACKAGE_APR_UTIL
 select BR2_PACKAGE_NEON

[git commit] opencv: add zlib dependency

Peter Korsgaard — 2012-05-17T19:00:16

commit: http://git.buildroot.net/buildroot/commit/?id=60cc0ba4b3a97804c9230cddbdeb1999f5023368
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

The opencv package selects zlib, but it does not depend on it. This
leads to cases where opencv gets built before zlib, and in this case,
opencv uses its internal version of zlib, which doesn't build properly
(it has some conditional code for ARM that is probably broken), see:

  http://autobuild.buildroot.org/results/45f4e195fe0cc5acace10287f5ff33aa953d5430/build-end.log

We fix this by properly adding the opencv dependency on zlib.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8< at >public.gmane.org>
Signed-off-by: Peter Korsgaard <jacmet-OfajU3CKLf1/SzgSGea1oA< at >public.gmane.org>
---
 package/opencv/opencv.mk |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/package/opencv/opencv.mk b/package/opencv/opencv.mk
index d415e51..ec94715 100644
--- a/package/opencv/opencv.mk
+++ b/package/open

[git commit] socat: fix build failure due to ext2_fs.hheader, bump version

Peter Korsgaard — 2012-05-17T21:43:00

commit: http://git.buildroot.net/buildroot/commit/?id=878401a9d06e260a0d3d86a92c21aab4a622d2c8
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

In kernel >= 3.3, the ext2_fs.h can no longer be imported from
userspace. This has been fixed for internal toolchains by adding a
patch to kernel headers, but this doesn't work with toolchains
generated by Crosstool-NG, or potentially upcoming external
toolchains.

socat in fact has a test in its configure.in, but the configure was
generated too long ago, and the generated test relies on the
preprocessor result and not the compiler result (but warns that in the
future, the compiler result will be used instead of the preprocessor
result).

So, by running autoconf on this package, we fix the problem: it
properly checks whether ext2_fs is usable or not, and acts
accordingly. Of course, it means that with recent versions of the
kernel, ext2-specific features of socat are unavailable, and we'll
have to wait for the socat developers to adapt their c

[git commit] socat: disable documentationbuild/installation

Peter Korsgaard — 2012-05-17T21:43:43

commit: http://git.buildroot.net/buildroot/commit/?id=556fd4f81588dff2e9ff288a87517d4ae18bf0ed
branch: http://git.buildroot.net/buildroot/commit/?id=refs/heads/master

The documentation generation process requires a special yold2man
program, for which we don't have a package in Buildroot. Since we
generally don't care much about documentation of packages, just adjust
the package Makefile.in to not build/install its documentation.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8< at >public.gmane.org>
Signed-off-by: Peter Korsgaard <jacmet-OfajU3CKLf1/SzgSGea1oA< at >public.gmane.org>
---
 package/socat/socat-no-documentation.patch |   37 ++++++++++++++++++++++++++++
 1 files changed, 37 insertions(+), 0 deletions(-)

diff --git a/package/socat/socat-no-documentation.patch b/package/socat/socat-no-documentation.patch
new file mode 100644
index 0000000..ce088df
--- /dev/null
+++ b/package/socat/socat-no-documentation.patch
< at >< at > -0,0 +1,37 < at >< at >
+socat: disable documentation build/installati

Re: [PATCH] apr: requires shared library support

Peter Korsgaard — 2012-05-17T21:42:34

 Thomas> The apr library requires shared library support unconditionally, so
 Thomas> make it depend on !BR2_PREFER_STATIC_LIB, so that architectures that
 Thomas> support static libraries only do not make this package available.

Committed, thanks.

Re: cp210x needed

Ron Eggler — 2012-05-17T20:56:54

with make linux-menuconfig I get this:
reg< at >SemiosDesktop-reg:/usr/local/src/E10-buildroot/br1202-main$ make
linux-menuconfig
mkdir -p /usr/local/src/E10-buildroot/br1202-main/output/target
if ! [ -d "/usr/local/src/E10-buildroot/br1202-main/output/target/bin" ];
then \
                if [ -d
"/usr/local/src/E10-buildroot/br1202-main/fs/skeleton" ]; then \
                        cp -fa
/usr/local/src/E10-buildroot/br1202-main/fs/skeleton/*
/usr/local/src/E10-buildroot/br1202-main/output/target/; \
                fi; \
                touch
/usr/local/src/E10-buildroot/br1202-main/output/host/usr/arm-unknown-linux-uclibcgnueabi/sysroot/.fakeroot.00000;
\
        fi
find /usr/local/src/E10-buildroot/br1202-main/output/target -type d -name
CVS -print0 -o -name .svn -print0 | xargs -0 rm -rf
find /usr/local/src/E10-buildroot/br1202-main/output/target -type f \(
-name .empty -o -name '*~' \) -print0 | xargs -0 rm -rf
touch /usr/local/src/E10-buildroot/br1202-main/output/build/.root
cp "board/synapse/e10/linux-3

Re: [PATCH v4 09/22] manual: add advice about GPL compliance for Buildroot

Thomas De Schampheleire — 2012-05-17T20:56:29

Op 17 mei 2012 19:35 schreef "Luca Ceresoli" <luca-ep08DiM7hqZH3+C6az13tg< at >public.gmane.org> het
volgende:
packages.
product:
machine, not
should
packages
installation
should

Acked-by: Thomas De Schampheleire <thomas.de.schampheleire-Re5JQEeQqe8AvxtiuMwx3w< at >public.gmane.org>
_______________________________________________
buildroot mailing list
buildroot-9GAsQqxh4YTR7s880joybQ< at >public.gmane.org
http://lists.busybox.net/mailman/listinfo/buildroot

Re: cp210x needed

Thomas Petazzoni — 2012-05-17T20:45:05

Le Thu, 17 May 2012 13:35:46 -0700,
Ron Eggler <reggler< at >semiosbio.com> a écrit :


In Buildroot, if you run "make menuconfig", you get the Buildroot
options.

If you have the kernel options, you have to run "make linux-menuconfig".

Thomas

Re: cp210x needed

Ron Eggler — 2012-05-17T20:35:46

Thomas,
Right but I haven't been able to find it so I wasn't sure if "make
menuconfig"
is the right way to do it...?

Thanks.
Ron

On Thu, May 17, 2012 at 1:21 PM, Thomas Petazzoni <
thomas.petazzoni-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8< at >public.gmane.org> wrote:

_______________________________________________
buildroot mailing list
buildroot-9GAsQqxh4YTR7s880joybQ< at >public.gmane.org
http://lists.busybox.net/mailman/listinfo/buildroot