This link is always being generated and it is assumed that it must
always point to /lib/libdl.so.2.
But this script is not needed and points to the wrong location when a
multilib toolchain is being used where the dynamic linker must be in
/lib64
Signed-off-by: Remy Bohmer
---
v2: Repost to keep
It is assumed by the makefiles that all dynamic linkers must be put in
/lib. This is however not true. Especially on multilib configurations this
information cannot be guessed. By looking in the sysroot of the compiler
toolchain the right paths can be found. Default the scripts already look
there,
Busybox make is not properly called in case:
* compiler name tuple does not match target tuple (as used in multilib
compilers, for example: i686-pc-linux-gnu-gcc can compile for
x86_64-unknown-linux-gnu targets. To solve this the 'SUBARCH' need to be
set)
* CONFIG_EXTRA_CFLAGS need to e set to s
For Sourcery Codebench toolchains the 'gcc -print-file-name' only returns
the proper sysroot directory when the proper architecture flags are
provided via '-m64 -march=' flags. These flags are provided via the
ptxdist platformconfig PTXCONF_TARGET_EXTRA_CFLAGS setting.
These flags therefore also ne
The platformconfig can specify some valuable global flags for each
package in the configuration to build, for example:
* CPU optimisations (-march= and friends)
* multilib toolchain selection between x86_64 and ia32 such as used for
the Sourcery Codebench compiler toolchains as specified here:
ht
Hi,
2012/3/13 Michael Olbrich :
> On Tue, Mar 13, 2012 at 01:46:26PM +0100, Remy Bohmer wrote:
>> It seems I have to stick with the bare minimal options I can set
>> here... Or any other ideas?
>
> Don't set any flags in BUSYBOX_MAKE_OPT. From what I can tell, the current
> busybox Makefile uses C
Some tools that mtd-utils provide can be also provided by busybox. Add
the dependencies here as done in other configuration files to prevent
the user from misconfiguration.
Signed-off-by: Bernhard Walle
---
rules/mtd-utils.in | 28
1 file changed, 28 insertions(+)
I had to rework some patches. Parallel build has been fixed upstream due
in commits 5b1cfb9d09a8a40577e79262c6d79dfcde36b1b7 and
05f56bf49ee0de8979c29c11f61a01e30e96, so we should be able to enable
it here.
flash-info has been dropped and mtdinfo should be used instead.
Signed-off-by: Bernhar
This is the 2nd/3rd version of the patches. Hopefully the last round.
Sorry for the confusion today. Too many machines, branches and too much work at
the same time. :(
--
ptxdist mailing list
ptxdist@pengutronix.de
On Tue, Mar 13, 2012 at 05:21:15PM +, Jerry Kirk wrote:
> Within my embedded project I'd like to modify Busybox's FTPD command to allow
> more general access to the filesystem. I'm not quite sure the right way to do
> this. I've been considering
>
> 1. A local "patch" to the busybox file tha
Signed-off-by: Alexander Dahl
---
rules/file.make |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/rules/file.make b/rules/file.make
index 8840bc4..5c7d739 100644
--- a/rules/file.make
+++ b/rules/file.make
@@ -16,8 +16,8 @@ PACKAGES-$(PTXCONF_FILE) += file
#
# Paths
Signed-off-by: Alexander Dahl
---
rules/dropbear.make |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/rules/dropbear.make b/rules/dropbear.make
index a3ba905..fe52ad8 100644
--- a/rules/dropbear.make
+++ b/rules/dropbear.make
@@ -18,8 +18,8 @@ PACKAGES-$(PTXCONF_DROPB
Within my embedded project I'd like to modify Busybox's FTPD command to allow
more general access to the filesystem. I'm not quite sure the right way to do
this. I've been considering
1. A local "patch" to the busybox file that gets copied
2. A new package that runs SED and modifies the busybox
On Tue, Mar 13, 2012 at 12:17:38PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> Some tools that mtd-utils provide can be also provided by busybox. Add
> the dependencies here as done in other configuration files to prevent
> the user from misconfiguration.
>
> Signed-off-by: Bernh
On Tue, Mar 13, 2012 at 02:31:14PM +0100, Michael Olbrich wrote:
> On Tue, Mar 13, 2012 at 01:00:11PM +0100, bernh...@bwalle.de wrote:
> > From: Bernhard Walle
> >
> > I had to rework some patches. Parallel build has been fixed upstream due
> > in commits 5b1cfb9d09a8a40577e79262c6d79dfcde36b1b7
On Tue, Mar 13, 2012 at 01:00:11PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> I had to rework some patches. Parallel build has been fixed upstream due
> in commits 5b1cfb9d09a8a40577e79262c6d79dfcde36b1b7 and
> 05f56bf49ee0de8979c29c11f61a01e30e96, so we should be able to ena
On Tue, Mar 13, 2012 at 12:30:23PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> ptxdist heavily uses "readlink -f" which only works with GNU readlink.
> So add the usual check here.
>
> However, because "readlink -f" is also used before the symlinks in the
> ptxdist directory is a
On Tue, Mar 13, 2012 at 01:46:26PM +0100, Remy Bohmer wrote:
> Hi,
>
> 2012/3/13 Michael Olbrich :
> > On Mon, Mar 12, 2012 at 10:25:02PM +0100, Remy Bohmer wrote:
> >> Busybox make is not properly called in case:
> >> * compiler name tuple does not match target tuple (as used in multilib
> >> c
Hi,
2012/3/13 Michael Olbrich :
> On Mon, Mar 12, 2012 at 10:25:02PM +0100, Remy Bohmer wrote:
>> Busybox make is not properly called in case:
>> * compiler name tuple does not match target tuple (as used in multilib
>> compilers, for example: i686-pc-linux-gnu-gcc can compile for
>> x86_64-un
From: Bernhard Walle
I had to rework some patches. Parallel build has been fixed upstream due
in commits 5b1cfb9d09a8a40577e79262c6d79dfcde36b1b7 and
05f56bf49ee0de8979c29c11f61a01e30e96, so we should be able to enable
it here.
flash-info has been dropped and mtdinfo should be used instead.
* Michael Olbrich [2012-03-13 10:09]:
> >
> > flash-info has been dropped in favour to mtdinfo.
>
> An option for mtdinfo already exists.
Indeed, I missed that. But it's in the ubi directory, so I adjusted the
dependencies in Kconfig.
> > +++
> > b/patches/mtd-utils-1.4.9/0003-Makefile-Use-LZ
From: Bernhard Walle
ptxdist heavily uses "readlink -f" which only works with GNU readlink.
So add the usual check here.
However, because "readlink -f" is also used before the symlinks in the
ptxdist directory is added to the PATH, that "readlink -f" call is
replaced by a bash-only implementatio
* Michael Olbrich [2012-03-13 11:14]:
>
> Not the 'local', the whole function. Getting all corner-cases right for
> something like this is not trivial.
Ah, ok. Indeed. That's why I started using something from the
Internet[tm] and that's also why I only replaced that initial
"readlink -f" call t
From: Bernhard Walle
ptxdist heavily uses "readlink -f" which only works with GNU readlink.
So add the usual check here.
However, because "readlink -f" is also used before the symlinks in the
ptxdist directory is added to the PATH, that "readlink -f" call is
replaced by a bash-only implementatio
From: Bernhard Walle
Some tools that mtd-utils provide can be also provided by busybox. Add
the dependencies here as done in other configuration files to prevent
the user from misconfiguration.
Signed-off-by: Bernhard Walle
---
rules/mtd-utils.in | 32
1 file
On Tue, Mar 13, 2012 at 11:34:52AM +0100, Remy Bohmer wrote:
> FYI: With these 5 patches (and a few unfinished architecture.in
> patches I have here locally) I at least get the glibc and the dynamic
> linkers installed in the right place and all binaries we used to be
> compiled with the right flag
Hi,
FYI: With these 5 patches (and a few unfinished architecture.in
patches I have here locally) I at least get the glibc and the dynamic
linkers installed in the right place and all binaries we used to be
compiled with the right flags.
Furthermore, I noticed that the extra flags are not passed pr
Hi,
2012/3/13 Michael Olbrich :
> On Mon, Mar 12, 2012 at 10:25:03PM +0100, Remy Bohmer wrote:
>> It is assumed by the makefiles that all dynamic linkers must be put in
>> /lib. This is however not true. Especially on multilib configurations this
>> information cannot be guessed. By looking in the
Hi,
2012/3/13 Michael Olbrich :
> On Mon, Mar 12, 2012 at 10:25:02PM +0100, Remy Bohmer wrote:
>> BUSYBOX_MAKE_OPT := \
>> ARCH=$(PTXCONF_ARCH_STRING) \
>> + SUBARCH=$(PTXCONF_ARCH_STRING) \
>> CROSS_COMPILE=$(COMPILER_PREFIX) \
>> CONFIG_EXTRA_CFLAGS="$(CROSS_CFLAGS)" \
>>
Hi,
2012/3/13 Michael Olbrich :
>>
>> lib="${1}"
>> + extra_cflags=$(source ${PTXDIST_PLATFORMCONFIG} && echo
>> "${PTXCONF_TARGET_EXTRA_CFLAGS}")
>
> I think ptxd_get_ptxconf should work here. And I think you should use
> PTXCONF_TARGET_EXTRA_CFLAGS and PTXCONF_TARGET_EXTRA_CPPFLAGS.
I
Hi,
2012/3/13 Michael Olbrich :
>> +-CFLAGS = -D_GNU_SOURCE -Wall -Wno-nonnull -Isrc/include
>> ++CFLAGS = -D_GNU_SOURCE -Wall -Wno-nonnull -Isrc/include $(CXXFLAGS)
>
> This should be:
> CFLAGS = $(CPPFLAGS) -D_GNU_SOURCE -Wall -Wno-nonnull -Isrc/include $(CFLAGS)
>
> CFLAGS is for C CXXFLAGS for
On Mon, Mar 12, 2012 at 08:56:33PM +0100, Bernhard Walle wrote:
> ptxdist heavily uses "readlink -f" which only works with GNU readlink.
> So add the usual check here.
>
> However, because "readlink -f" is also used before the symlinks in the
> ptxdist directory is added to the PATH, that "readlin
On Mon, Mar 12, 2012 at 08:56:17PM +0100, Bernhard Walle wrote:
> Hi Michael,
>
> thanks again for the review.
>
> * Michael Olbrich [2012-02-14 23:05]:
> > > +# bash implementation of realpath / readlink -f
> > > +# arg1 - filename
> > > +realpath()
> > > +{
> > > + local fname oldfname
> > >
On Mon, Mar 12, 2012 at 10:25:03PM +0100, Remy Bohmer wrote:
> It is assumed by the makefiles that all dynamic linkers must be put in
> /lib. This is however not true. Especially on multilib configurations this
> information cannot be guessed. By looking in the sysroot of the compiler
> toolchain t
On Mon, Mar 12, 2012 at 10:25:02PM +0100, Remy Bohmer wrote:
> Busybox make is not properly called in case:
> * compiler name tuple does not match target tuple (as used in multilib
> compilers, for example: i686-pc-linux-gnu-gcc can compile for
> x86_64-unknown-linux-gnu targets. To solve this
On Mon, Mar 12, 2012 at 10:25:01PM +0100, Remy Bohmer wrote:
> For Sourcery Codebench toolchains the 'gcc -print-file-name' only returns
> the proper sysroot directory when the proper architecture flags are
> provided via '-m64 -march=' flags. These flags are provided via the
> ptxdist platformconf
On Mon, Mar 12, 2012 at 10:25:00PM +0100, Remy Bohmer wrote:
> The platformconfig can specify some valuable global flags for each
> package in the configuration to build, for example:
> * CPU optimisations (-march= and friends)
> * multilib toolchain selection between x86_64 and ia32 such as used f
On Mon, Mar 12, 2012 at 08:56:32PM +0100, Bernhard Walle wrote:
> We check for a xargs binary that supports the '-r' flag, so put the
> usual symlink into the bin directory once we have found a proper
> implementation.
Applied.
Michael
> Signed-off-by: Bernhard Walle
> ---
> Makefile.in |1
On Tue, Mar 13, 2012 at 10:18:12AM +0100, Bernhard Walle wrote:
> * Michael Olbrich [2012-03-13 10:12]:
> > On Mon, Mar 12, 2012 at 03:09:04PM +0100, bernh...@bwalle.de wrote:
> > > From: Bernhard Walle
> > >
> > > Some tools that mtd-utils provide can be also provided by busybox. Add
> > > the
On Mon, Mar 12, 2012 at 05:00:14PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> The lsof build system uses host ar and ranlib. With that patch it uses
> the tools provided by the cross toolchain.
>
> On a Gentoo system, without that patch, I get following error:
>
> ar cr lib
* Michael Olbrich [2012-03-13 10:12]:
> On Mon, Mar 12, 2012 at 03:09:04PM +0100, bernh...@bwalle.de wrote:
> > From: Bernhard Walle
> >
> > Some tools that mtd-utils provide can be also provided by busybox. Add
> > the dependencies here as done in other configuration files to prevent
> > the us
On Mon, Mar 12, 2012 at 03:09:05PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
Thanks, applied.
Michael
> Signed-off-by: Bernhard Walle
> ---
> rules/procps.in |8
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/rules/procps.in b/rules/procps.in
>
On Mon, Mar 12, 2012 at 03:09:04PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> Some tools that mtd-utils provide can be also provided by busybox. Add
> the dependencies here as done in other configuration files to prevent
> the user from misconfiguration.
>
> Signed-off-by: Bernh
On Mon, Mar 12, 2012 at 03:09:03PM +0100, bernh...@bwalle.de wrote:
> From: Bernhard Walle
>
> I had to rework some patches. Parallel build has been fixed upstream due
> in commits 5b1cfb9d09a8a40577e79262c6d79dfcde36b1b7 and
> 05f56bf49ee0de8979c29c11f61a01e30e96, so we should be able to ena
Hi Dániel,
for Mini2440 related questions better use our community mailing list at [1].
Dániel Szálas wrote:
> I have a mini2440 with l80 8" lcd screen. As I see there is no support in
> pdxdist for this device. :(
Do you have any information about this display (resolution, signal
timing, ...)?
45 matches
Mail list logo