[PATCH] cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[PATCH] cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[Chris Larson]

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 classes/cross.bbclass |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/classes/cross.bbclass b/classes/cross.bbclass
index 7debde6..72a0fb7 100644
--- a/classes/cross.bbclass
+++ b/classes/cross.bbclass
@@ -2,6 +2,11 @@
 # no need for them to be a direct target of 'world'
 EXCLUDE_FROM_WORLD = "1"
 
+# In order to keep TARGET_PREFIX decoupled from TARGET_SYS, let's force the
+# binary names to match the former, rather than relying on autoconf's implicit
+# prefixing based on the latter.
+EXTRA_OECONF_append = " --program-prefix=${TARGET_PREFIX}"
+
 # Save PACKAGE_ARCH before changing HOST_ARCH
 OLD_PACKAGE_ARCH := "${PACKAGE_ARCH}"
 PACKAGE_ARCH = "${OLD_PACKAGE_ARCH}"
-- 
1.6.0

[PATCH] patch.bbclass: catch exceptions raised in the Resolve().

[Chris Larson]

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 classes/patch.bbclass |    7 +++----
 1 files changed, 3 insertions(+), 4 deletions(-)

diff --git a/classes/patch.bbclass b/classes/patch.bbclass
index 2f99e4c..2aa63c0 100644
--- a/classes/patch.bbclass
+++ b/classes/patch.bbclass
@@ -530,10 +530,9 @@ python patch_do_patch() {
 		bb.note("Applying patch '%s' (%s)" % (pname, base_path_out(unpacked, d)))
 		try:
 			patchset.Import({"file":unpacked, "remote":url, "strippath": pnum}, True)
-		except:
-			import sys
-			raise bb.build.FuncFailed(str(sys.exc_value))
-		resolver.Resolve()
+			resolver.Resolve()
+		except Exception, e:
+			raise bb.build.FuncFailed(str(e))
 }
 
 EXPORT_FUNCTIONS do_patch
-- 
1.6.0

[PATCH] patch.bbclass: Add "git" patchtool mechanism, which uses git-apply

[Chris Larson]

From: Dale Farnsworth <dfarnsworth@mvista.com>

It can be selected by setting PATCHTOOL = "git".

This is useful because git-apply honors the permissions information
produced by git-format-patch.

Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
---
 classes/patch.bbclass |   20 ++++++++++++++++++++
 1 files changed, 20 insertions(+), 0 deletions(-)

diff --git a/classes/patch.bbclass b/classes/patch.bbclass
index 2aa63c0..650d9c1 100644
--- a/classes/patch.bbclass
+++ b/classes/patch.bbclass
@@ -189,6 +189,24 @@ def patch_init(d):
 		def Clean(self):
 			""""""
 
+	class GitApplyTree(PatchTree):
+		def __init__(self, dir, d):
+			PatchTree.__init__(self, dir, d)
+
+		def _applypatch(self, patch, force = False, reverse = False, run = True):
+			shellcmd = ["git", "--git-dir=.", "apply", "-p%s" % patch['strippath']]
+
+			if reverse:
+				shellcmd.append('-R')
+
+			shellcmd.append(patch['file'])
+
+			if not run:
+				return "sh" + "-c" + " ".join(shellcmd)
+
+			return runcmd(["sh", "-c", " ".join(shellcmd)], self.dir)
+
+
 	class QuiltTree(PatchSet):
 		def _runcmd(self, args, run = True):
 			quiltrc = bb.data.getVar('QUILTRCFILE', self.d, 1)
@@ -424,6 +442,7 @@ def patch_init(d):
 	g["PatchSet"] = PatchSet
 	g["PatchTree"] = PatchTree
 	g["QuiltTree"] = QuiltTree
+	g["GitApplyTree"] = GitApplyTree
 	g["Resolver"] = Resolver
 	g["UserResolver"] = UserResolver
 	g["NOOPResolver"] = NOOPResolver
@@ -449,6 +468,7 @@ python patch_do_patch() {
 	patchsetmap = {
 		"patch": PatchTree,
 		"quilt": QuiltTree,
+		"git": GitApplyTree,
 	}
 
 	cls = patchsetmap[bb.data.getVar('PATCHTOOL', d, 1) or 'quilt']
-- 
1.6.0

[PATCH] kernel.bbclass: Set EXTRA_OEMAKE to null for kernel

[Chris Larson]

From: Dale Farnsworth <dfarnsworth@mvista.com>

Currently the only thing in EXTRA_OEMAKE is '-e MAKEFLAGS='.  We don't
want to overide the kernel's Makefile variables from the environment.
It caused the passed -j<N> parameter from PARALLEL_MAKE to be ignored.

Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
---
 classes/kernel.bbclass |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass
index 3ee7f53..2947d4d 100644
--- a/classes/kernel.bbclass
+++ b/classes/kernel.bbclass
@@ -78,6 +78,10 @@ PACKAGE_ARCH = "${MACHINE_ARCH}"
 UBOOT_ENTRYPOINT ?= "20008000"
 UBOOT_LOADADDRESS ?= "${UBOOT_ENTRYPOINT}"
 
+# For the kernel, we don't want the '-e MAKEFLAGS=' in EXTRA_OEMAKE.
+# We don't want to override kernel Makefile variables from the environment
+EXTRA_OEMAKE = ""
+
 kernel_do_compile() {
 	unset CFLAGS CPPFLAGS CXXFLAGS LDFLAGS MACHINE
 	oe_runmake include/linux/version.h CC="${KERNEL_CC}" LD="${KERNEL_LD}"
-- 
1.6.0

[PATCH] kernel.bbclass: Install & package symvers as /boot/Module.symvers-<ver>.

[Chris Larson]

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 classes/kernel.bbclass |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass
index 2947d4d..a8b2b0d 100644
--- a/classes/kernel.bbclass
+++ b/classes/kernel.bbclass
@@ -179,6 +179,7 @@ kernel_do_install() {
 	install -m 0644 System.map ${D}/boot/System.map-${KERNEL_VERSION}
 	install -m 0644 .config ${D}/boot/config-${KERNEL_VERSION}
 	install -m 0644 vmlinux ${D}/boot/vmlinux-${KERNEL_VERSION}
+	[ -e Module.symvers ] && install -m 0644 Module.symvers ${D}/boot/Module.symvers-${KERNEL_VERSION}
 	install -d ${D}/etc/modutils
 	if [ "${KERNEL_MAJOR_VERSION}" = "2.6" ]; then
 		install -d ${D}/etc/modprobe.d
@@ -229,7 +230,7 @@ EXPORT_FUNCTIONS do_compile do_install do_stage do_configure
 PACKAGES = "kernel kernel-base kernel-image kernel-dev kernel-vmlinux"
 FILES = ""
 FILES_kernel-image = "/boot/${KERNEL_IMAGETYPE}*"
-FILES_kernel-dev = "/boot/System.map* /boot/config*"
+FILES_kernel-dev = "/boot/System.map* /boot/Module.symvers* /boot/config*"
 FILES_kernel-vmlinux = "/boot/vmlinux*"
 RDEPENDS_kernel = "kernel-base"
 RRECOMMENDS_kernel-module-hostap-cs += '${@base_version_less_or_equal("KERNEL_VERSION", "2.6.17", "", "apm-wifi-suspendfix", d)}'
-- 
1.6.0

[PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Chris Larson]

To re-enable, set CCACHE = "ccache ".

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 conf/bitbake.conf |    1 -
 1 files changed, 0 insertions(+), 1 deletions(-)

diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index c4af34e..0b35ea8 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -390,7 +390,6 @@ export PATH
 # Build utility info.
 ##################################################################
 
-CCACHE = "${@bb.which(bb.data.getVar('PATH', d, 1), 'ccache') and 'ccache '}"
 TOOLCHAIN_OPTIONS = ""
 
 export CC = "${CCACHE}${HOST_PREFIX}gcc ${HOST_CC_ARCH}${TOOLCHAIN_OPTIONS}"
-- 
1.6.0

[PATCH] bitbake.conf: Add export for cross NM.

[Chris Larson]

From: Jeremy Puhlman <jpuhlman@mvista.com>

When not providing a cross nm, the configure for libtool-cross falls back to
the system nm. This can lead to empty "global_symbol_pipe" and
"global_symbol_to_cdeclvarble" variables in the generated libtool script.

Signed-off-by: Jeremy Puhlman <jpuhlman@mvista.com>
Signed-off-by: Chris Larson <clarson@mvista.com>
---
 conf/bitbake.conf |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index 0b35ea8..371bf8e 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -404,6 +404,7 @@ export RANLIB = "${HOST_PREFIX}ranlib"
 export STRIP = "${HOST_PREFIX}strip"
 export OBJCOPY = "${HOST_PREFIX}objcopy"
 export OBJDUMP = "${HOST_PREFIX}objdump"
+export NM = "${HOST_PREFIX}nm"
 PYTHON = "${@sys.executable}"
 
 export BUILD_CC = "${CCACHE}${BUILD_PREFIX}gcc ${BUILD_CC_ARCH}"
-- 
1.6.0

[PATCH] bitbake.conf: BUILD_LDFLAGS: use $ORIGIN in the -rpath rather than a full path.

[Chris Larson]

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 conf/bitbake.conf |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index 371bf8e..9cb6039 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -447,7 +447,7 @@ export SDK_CXXFLAGS = "${SDK_CFLAGS} -fpermissive"
 
 export BUILD_LDFLAGS = "-L${STAGING_LIBDIR_NATIVE} \
                         -Wl,-rpath-link,${STAGING_LIBDIR_NATIVE} \
-                        -Wl,-rpath,${STAGING_LIBDIR_NATIVE} -Wl,-O1"
+                        -Wl,-rpath,\\\$\$ORIGIN/${@base_path_relative(d.getVar('STAGING_BINDIR_NATIVE', 1), d.getVar('STAGING_LIBDIR_NATIVE', 1))} -Wl,-O1"
 
 export LDFLAGS = "${TARGET_LDFLAGS}"
 export TARGET_LDFLAGS = "-L${STAGING_DIR_TARGET}${layout_libdir} \
-- 
1.6.0

[PATCH] bitbake.conf: only pass --no-check-certificate if wget supports it.

[Chris Larson]

Signed-off-by: Chris Larson <clarson@mvista.com>
---
 conf/bitbake.conf |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/conf/bitbake.conf b/conf/bitbake.conf
index 9cb6039..b849a85 100644
--- a/conf/bitbake.conf
+++ b/conf/bitbake.conf
@@ -530,12 +530,13 @@ FETCHCMD_wget = "/usr/bin/env wget -t 5 --no-check-certificate"
 FETCHCMD_bzr = "/usr/bin/env bzr"
 FETCHCMD_hg = "/usr/bin/env hg"
 
+__wget_nocheckcert := "${@['--no-check-certificate', '']['unrecognized option' in os.popen3("wget --no-check-certificate")[2].read()]}"
 FETCHCOMMAND = "ERROR, this must be a BitBake bug"
-FETCHCOMMAND_wget = "/usr/bin/env 'PATH=${PATH}' wget -t 5 --passive-ftp --no-check-certificate -P ${DL_DIR} ${URI}"
+FETCHCOMMAND_wget = "/usr/bin/env 'PATH=${PATH}' wget -t 5 --passive-ftp ${__wget_nocheckcert} -P ${DL_DIR} ${URI}"
 FETCHCOMMAND_cvs = "/usr/bin/env 'PATH=${PATH}' cvs '-d${CVSROOT}' co ${CVSCOOPTS} ${CVSMODULE}"
 FETCHCOMMAND_svn = "/usr/bin/env svn co ${SVNCOOPTS} ${SVNROOT} ${SVNMODULE}"
 RESUMECOMMAND = "ERROR, this must be a BitBake bug"
-RESUMECOMMAND_wget = "/usr/bin/env 'PATH=${PATH}' wget -c -t 5 --passive-ftp --no-check-certificate -P ${DL_DIR} ${URI}"
+RESUMECOMMAND_wget = "/usr/bin/env 'PATH=${PATH}' wget -c -t 5 --passive-ftp ${__wget_nocheckcert} -P ${DL_DIR} ${URI}"
 UPDATECOMMAND = "ERROR, this must be a BitBake bug"
 UPDATECOMMAND_cvs = "/usr/bin/env 'PATH=${PATH}' cvs -d${CVSROOT} update -d -P ${CVSCOOPTS}"
 UPDATECOMMAND_svn = "/usr/bin/env svn update ${SVNCOOPTS}"
-- 
1.6.0

Re: [PATCH] kernel.bbclass: Set EXTRA_OEMAKE to null for kernel

[Denys Dmytriyenko]

On Wed, Aug 19, 2009 at 08:24:35PM -0700, Chris Larson wrote:
> From: Dale Farnsworth <dfarnsworth@mvista.com>
> 
> Currently the only thing in EXTRA_OEMAKE is '-e MAKEFLAGS='.  We don't
> want to overide the kernel's Makefile variables from the environment.
> It caused the passed -j<N> parameter from PARALLEL_MAKE to be ignored.
> 
> Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
> Signed-off-by: Chris Larson <clarson@mvista.com>

Acked-by: Denys Dmytriyenko <denis@denix.org>

Had a similar fix in my own tree locally, but not specifically for -jN...

> ---
>  classes/kernel.bbclass |    4 ++++
>  1 files changed, 4 insertions(+), 0 deletions(-)
> 
> diff --git a/classes/kernel.bbclass b/classes/kernel.bbclass
> index 3ee7f53..2947d4d 100644
> --- a/classes/kernel.bbclass
> +++ b/classes/kernel.bbclass
> @@ -78,6 +78,10 @@ PACKAGE_ARCH = "${MACHINE_ARCH}"
>  UBOOT_ENTRYPOINT ?= "20008000"
>  UBOOT_LOADADDRESS ?= "${UBOOT_ENTRYPOINT}"
>  
> +# For the kernel, we don't want the '-e MAKEFLAGS=' in EXTRA_OEMAKE.
> +# We don't want to override kernel Makefile variables from the environment
> +EXTRA_OEMAKE = ""
> +
>  kernel_do_compile() {
>  	unset CFLAGS CPPFLAGS CXXFLAGS LDFLAGS MACHINE
>  	oe_runmake include/linux/version.h CC="${KERNEL_CC}" LD="${KERNEL_LD}"
> -- 
> 1.6.0
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [PATCH] patch.bbclass: Add "git" patchtool mechanism, which uses git-apply

[Denys Dmytriyenko]

On Wed, Aug 19, 2009 at 08:24:34PM -0700, Chris Larson wrote:
> From: Dale Farnsworth <dfarnsworth@mvista.com>
> 
> It can be selected by setting PATCHTOOL = "git".
> 
> This is useful because git-apply honors the permissions information
> produced by git-format-patch.
> 
> Signed-off-by: Dale Farnsworth <dfarnsworth@mvista.com>
> Signed-off-by: Chris Larson <clarson@mvista.com>

Acked-by: Denys Dmytriyenko <denis@denix.org>

Nice one, thanks!

> ---
>  classes/patch.bbclass |   20 ++++++++++++++++++++
>  1 files changed, 20 insertions(+), 0 deletions(-)
> 
> diff --git a/classes/patch.bbclass b/classes/patch.bbclass
> index 2aa63c0..650d9c1 100644
> --- a/classes/patch.bbclass
> +++ b/classes/patch.bbclass
> @@ -189,6 +189,24 @@ def patch_init(d):
>  		def Clean(self):
>  			""""""
>  
> +	class GitApplyTree(PatchTree):
> +		def __init__(self, dir, d):
> +			PatchTree.__init__(self, dir, d)
> +
> +		def _applypatch(self, patch, force = False, reverse = False, run = True):
> +			shellcmd = ["git", "--git-dir=.", "apply", "-p%s" % patch['strippath']]
> +
> +			if reverse:
> +				shellcmd.append('-R')
> +
> +			shellcmd.append(patch['file'])
> +
> +			if not run:
> +				return "sh" + "-c" + " ".join(shellcmd)
> +
> +			return runcmd(["sh", "-c", " ".join(shellcmd)], self.dir)
> +
> +
>  	class QuiltTree(PatchSet):
>  		def _runcmd(self, args, run = True):
>  			quiltrc = bb.data.getVar('QUILTRCFILE', self.d, 1)
> @@ -424,6 +442,7 @@ def patch_init(d):
>  	g["PatchSet"] = PatchSet
>  	g["PatchTree"] = PatchTree
>  	g["QuiltTree"] = QuiltTree
> +	g["GitApplyTree"] = GitApplyTree
>  	g["Resolver"] = Resolver
>  	g["UserResolver"] = UserResolver
>  	g["NOOPResolver"] = NOOPResolver
> @@ -449,6 +468,7 @@ python patch_do_patch() {
>  	patchsetmap = {
>  		"patch": PatchTree,
>  		"quilt": QuiltTree,
> +		"git": GitApplyTree,
>  	}
>  
>  	cls = patchsetmap[bb.data.getVar('PATCHTOOL', d, 1) or 'quilt']
> -- 
> 1.6.0
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Denys Dmytriyenko]

On Wed, Aug 19, 2009 at 08:24:37PM -0700, Chris Larson wrote:
> To re-enable, set CCACHE = "ccache ".
> 
> Signed-off-by: Chris Larson <clarson@mvista.com>

Acked-by: Denys Dmytriyenko <denis@denix.org>

CCACHE-related issues are one of the most commonly reported...

> ---
>  conf/bitbake.conf |    1 -
>  1 files changed, 0 insertions(+), 1 deletions(-)
> 
> diff --git a/conf/bitbake.conf b/conf/bitbake.conf
> index c4af34e..0b35ea8 100644
> --- a/conf/bitbake.conf
> +++ b/conf/bitbake.conf
> @@ -390,7 +390,6 @@ export PATH
>  # Build utility info.
>  ##################################################################
>  
> -CCACHE = "${@bb.which(bb.data.getVar('PATH', d, 1), 'ccache') and 'ccache '}"
>  TOOLCHAIN_OPTIONS = ""
>  
>  export CC = "${CCACHE}${HOST_PREFIX}gcc ${HOST_CC_ARCH}${TOOLCHAIN_OPTIONS}"
> -- 
> 1.6.0
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [PATCH] bitbake.conf: BUILD_LDFLAGS: use $ORIGIN in the -rpath rather than a full path.

[Denys Dmytriyenko]

On Wed, Aug 19, 2009 at 08:24:39PM -0700, Chris Larson wrote:
> Signed-off-by: Chris Larson <clarson@mvista.com>

Hmm, interesting. I didn't know about $ORIGIN...
But what is the benefit here? To be able to move $TMPDIR?

> ---
>  conf/bitbake.conf |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/conf/bitbake.conf b/conf/bitbake.conf
> index 371bf8e..9cb6039 100644
> --- a/conf/bitbake.conf
> +++ b/conf/bitbake.conf
> @@ -447,7 +447,7 @@ export SDK_CXXFLAGS = "${SDK_CFLAGS} -fpermissive"
>  
>  export BUILD_LDFLAGS = "-L${STAGING_LIBDIR_NATIVE} \
>                          -Wl,-rpath-link,${STAGING_LIBDIR_NATIVE} \
> -                        -Wl,-rpath,${STAGING_LIBDIR_NATIVE} -Wl,-O1"
> +                        -Wl,-rpath,\\\$\$ORIGIN/${@base_path_relative(d.getVar('STAGING_BINDIR_NATIVE', 1), d.getVar('STAGING_LIBDIR_NATIVE', 1))} -Wl,-O1"
>  
>  export LDFLAGS = "${TARGET_LDFLAGS}"
>  export TARGET_LDFLAGS = "-L${STAGING_DIR_TARGET}${layout_libdir} \
> -- 
> 1.6.0
> 
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [PATCH] bitbake.conf: BUILD_LDFLAGS: use $ORIGIN in the -rpath rather than a full path.

[Chris Larson]

Yeah, its one of many changes needed to get things relocatable enough
to be able to use prebuilt binaries / packaged staging packages
amongst a group of developers, for example.  This is just one piece of
the puzzle, of course.  Ideally, more open source projects would be
properly relocatable, and we wouldn't need so many hacks.  We ended up
doing a lot of things to work around reloc issues in native/cross
recipes, for MVL6.  Everything from exporting a pile of new env vars
to writing wrapper scripts and the like.. yech.  Still, it's a worthy
goal, and the benefits would be worthwhile, so I think we should start
taking the steps in that direction.

On Wed, Aug 19, 2009 at 9:18 PM, Denys Dmytriyenko<denis@denix.org> wrote:
> On Wed, Aug 19, 2009 at 08:24:39PM -0700, Chris Larson wrote:
>> Signed-off-by: Chris Larson <clarson@mvista.com>
>
> Hmm, interesting. I didn't know about $ORIGIN...
> But what is the benefit here? To be able to move $TMPDIR?
>
>> ---
>>  conf/bitbake.conf |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/conf/bitbake.conf b/conf/bitbake.conf
>> index 371bf8e..9cb6039 100644
>> --- a/conf/bitbake.conf
>> +++ b/conf/bitbake.conf
>> @@ -447,7 +447,7 @@ export SDK_CXXFLAGS = "${SDK_CFLAGS} -fpermissive"
>>
>>  export BUILD_LDFLAGS = "-L${STAGING_LIBDIR_NATIVE} \
>>                          -Wl,-rpath-link,${STAGING_LIBDIR_NATIVE} \
>> -                        -Wl,-rpath,${STAGING_LIBDIR_NATIVE} -Wl,-O1"
>> +                        -Wl,-rpath,\\\$\$ORIGIN/${@base_path_relative(d.getVar('STAGING_BINDIR_NATIVE', 1), d.getVar('STAGING_LIBDIR_NATIVE', 1))} -Wl,-O1"
>>
>>  export LDFLAGS = "${TARGET_LDFLAGS}"
>>  export TARGET_LDFLAGS = "-L${STAGING_DIR_TARGET}${layout_libdir} \
>> --
>> 1.6.0
>>
>>
>> _______________________________________________
>> Openembedded-devel mailing list
>> Openembedded-devel@lists.openembedded.org
>> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>



-- 
Chris Larson
clarson at kergoth dot com
clarson at mvista dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Software Engineer
MontaVista Software, Inc.

Re: [PATCH] bitbake.conf: BUILD_LDFLAGS: use $ORIGIN in the -rpath rather than a full path.

[Denys Dmytriyenko]

On Wed, Aug 19, 2009 at 09:53:49PM -0700, Chris Larson wrote:
> Yeah, its one of many changes needed to get things relocatable enough
> to be able to use prebuilt binaries / packaged staging packages
> amongst a group of developers, for example.  This is just one piece of
> the puzzle, of course.  Ideally, more open source projects would be
> properly relocatable, and we wouldn't need so many hacks.  We ended up
> doing a lot of things to work around reloc issues in native/cross
> recipes, for MVL6.  Everything from exporting a pile of new env vars
> to writing wrapper scripts and the like.. yech.  Still, it's a worthy
> goal, and the benefits would be worthwhile, so I think we should start
> taking the steps in that direction.

Ah, relocatable pstage - a worthy goal! :)

Acked-by: Denys Dmytriyenko <denis@denix.org>

> On Wed, Aug 19, 2009 at 9:18 PM, Denys Dmytriyenko<denis@denix.org> wrote:
> > On Wed, Aug 19, 2009 at 08:24:39PM -0700, Chris Larson wrote:
> >> Signed-off-by: Chris Larson <clarson@mvista.com>
> >
> > Hmm, interesting. I didn't know about $ORIGIN...
> > But what is the benefit here? To be able to move $TMPDIR?
> >
> >> ---
> >>  conf/bitbake.conf |    2 +-
> >>  1 files changed, 1 insertions(+), 1 deletions(-)
> >>
> >> diff --git a/conf/bitbake.conf b/conf/bitbake.conf
> >> index 371bf8e..9cb6039 100644
> >> --- a/conf/bitbake.conf
> >> +++ b/conf/bitbake.conf
> >> @@ -447,7 +447,7 @@ export SDK_CXXFLAGS = "${SDK_CFLAGS} -fpermissive"
> >>
> >>  export BUILD_LDFLAGS = "-L${STAGING_LIBDIR_NATIVE} \
> >>                          -Wl,-rpath-link,${STAGING_LIBDIR_NATIVE} \
> >> -                        -Wl,-rpath,${STAGING_LIBDIR_NATIVE} -Wl,-O1"
> >> +                        -Wl,-rpath,\\\$\$ORIGIN/${@base_path_relative(d.getVar('STAGING_BINDIR_NATIVE', 1), d.getVar('STAGING_LIBDIR_NATIVE', 1))} -Wl,-O1"
> >>
> >>  export LDFLAGS = "${TARGET_LDFLAGS}"
> >>  export TARGET_LDFLAGS = "-L${STAGING_DIR_TARGET}${layout_libdir} \
> >> --
> >> 1.6.0
> >>
> >>
> >> _______________________________________________
> >> Openembedded-devel mailing list
> >> Openembedded-devel@lists.openembedded.org
> >> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> >
> > _______________________________________________
> > Openembedded-devel mailing list
> > Openembedded-devel@lists.openembedded.org
> > http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
> >
> 
> 
> 
> -- 
> Chris Larson
> clarson at kergoth dot com
> clarson at mvista dot com
> Founder - BitBake, OpenEmbedded, OpenZaurus
> Maintainer - Tslib
> Software Engineer
> MontaVista Software, Inc.
> 
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel

Re: [PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Holger Hans Peter Freyther]

On Thursday 20 August 2009 05:24:37 Chris Larson wrote:
> To re-enable, set CCACHE = "ccache ".
>
> Signed-off-by: Chris Larson <clarson@mvista.com>

Could you elaborate on the need of that? Which platforms do you deal with that 
break due the usage of ccache?

And technically I think you should make it CCACHE = "", out of legacy I think 
we can not really remove keys from bitbake.conf

Re: [PATCH] bitbake.conf: Add export for cross NM.

[Holger Hans Peter Freyther]

On Thursday 20 August 2009 05:24:38 Chris Larson wrote:
> From: Jeremy Puhlman <jpuhlman@mvista.com>
>
> When not providing a cross nm, the configure for libtool-cross falls back
> to the system nm. This can lead to empty "global_symbol_pipe" and
> "global_symbol_to_cdeclvarble" variables in the generated libtool script.
>
> Signed-off-by: Jeremy Puhlman <jpuhlman@mvista.com>
> Signed-off-by: Chris Larson <clarson@mvista.com>

Acked-by: Holger Hans Peter Freyther <zecke@selfish.org>


> ---
>  conf/bitbake.conf |    1 +
>  1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/conf/bitbake.conf b/conf/bitbake.conf
> index 0b35ea8..371bf8e 100644
> --- a/conf/bitbake.conf
> +++ b/conf/bitbake.conf
> @@ -404,6 +404,7 @@ export RANLIB = "${HOST_PREFIX}ranlib"
>  export STRIP = "${HOST_PREFIX}strip"
>  export OBJCOPY = "${HOST_PREFIX}objcopy"
>  export OBJDUMP = "${HOST_PREFIX}objdump"
> +export NM = "${HOST_PREFIX}nm"
>  PYTHON = "${@sys.executable}"
>
>  export BUILD_CC = "${CCACHE}${BUILD_PREFIX}gcc ${BUILD_CC_ARCH}"

Re: [PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Graham Gower]

2009/8/20 Holger Hans Peter Freyther <holger+oe@freyther.de>:
> On Thursday 20 August 2009 05:24:37 Chris Larson wrote:
>> To re-enable, set CCACHE = "ccache ".
>>
>> Signed-off-by: Chris Larson <clarson@mvista.com>
>
> Could you elaborate on the need of that? Which platforms do you deal with that
> break due the usage of ccache?

I run Slackware 12.2, whose build of ccache does not seem to be
accepted. I don't recall the exact mechanism of failure, but
shasum-native fails before getting to any other pacakges. Currently I
have to move the ccache binary out of my $PATH or OE tries to use it.

-Graham

Re: [PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Denys Dmytriyenko]

On Thu, Aug 20, 2009 at 05:02:31PM +0930, Graham Gower wrote:
> 2009/8/20 Holger Hans Peter Freyther <holger+oe@freyther.de>:
> > On Thursday 20 August 2009 05:24:37 Chris Larson wrote:
> >> To re-enable, set CCACHE = "ccache ".
> >>
> >> Signed-off-by: Chris Larson <clarson@mvista.com>
> >
> > Could you elaborate on the need of that? Which platforms do you deal with that
> > break due the usage of ccache?
> 
> I run Slackware 12.2, whose build of ccache does not seem to be
> accepted. I don't recall the exact mechanism of failure, but
> shasum-native fails before getting to any other pacakges. Currently I
> have to move the ccache binary out of my $PATH or OE tries to use it.

Same exact problem has been reported many times with RedHat-based distros as 
well - it just fails to build shasum-native (the very first package) w/o any 
meaningful messages...

-- 
Denys

Re: [PATCH] cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[Michael Smith]

On Wed, 19 Aug 2009, Chris Larson wrote:

> Signed-off-by: Chris Larson <clarson@mvista.com>
> ---
>  classes/cross.bbclass |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)

Hi Chris,

Can you explain for a crosscompile noob -- when would TARGET_PREFIX != 
${TARGET_SYS}-? I understand they're separate variables so they COULD be 
different; just wondering why someone would set them that way.

Thanks,
Mike

Re: [PATCH] kernel.bbclass: Install & package symvers as /boot/Module.symvers-<ver>.

[Michael Smith]

On Wed, 19 Aug 2009, Chris Larson wrote:

> Signed-off-by: Chris Larson <clarson@mvista.com>
> ---
>  classes/kernel.bbclass |    3 ++-
>  1 files changed, 2 insertions(+), 1 deletions(-)

Acked-by: Michael Smith <msmith@cbnco.com>

Re: [PATCH] bitbake.conf: Kill CCACHE. Implicit, automatic use of things which can fail is bad.

[Martyn Welch]

Denys Dmytriyenko wrote:
> On Thu, Aug 20, 2009 at 05:02:31PM +0930, Graham Gower wrote:
>   
>> 2009/8/20 Holger Hans Peter Freyther <holger+oe@freyther.de>:
>>     
>>> On Thursday 20 August 2009 05:24:37 Chris Larson wrote:
>>>       
>>>> To re-enable, set CCACHE = "ccache ".
>>>>
>>>> Signed-off-by: Chris Larson <clarson@mvista.com>
>>>>         
>>> Could you elaborate on the need of that? Which platforms do you deal with that
>>> break due the usage of ccache?
>>>       
>> I run Slackware 12.2, whose build of ccache does not seem to be
>> accepted. I don't recall the exact mechanism of failure, but
>> shasum-native fails before getting to any other pacakges. Currently I
>> have to move the ccache binary out of my $PATH or OE tries to use it.
>>     
>
> Same exact problem has been reported many times with RedHat-based distros as 
> well - it just fails to build shasum-native (the very first package) w/o any 
> meaningful messages...
>   
I have also suffered random failures that disappear once ccache has been 
uninstalled on at least Ubuntu 8.10.

Martyn

-- 
Martyn Welch MEng MPhil MIET (Principal Software Engineer)   T:+44(0)1327322748
GE Fanuc Intelligent Platforms Ltd,        |Registered in England and Wales
Tove Valley Business Park, Towcester,      |(3828642) at 100 Barbirolli Square,
Northants, NN12 6PF, UK T:+44(0)1327359444 |Manchester,M2 3AB  VAT:GB 927559189

Re: [PATCH] cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[Chris Larson]

> Can you explain for a crosscompile noob -- when would TARGET_PREFIX !=
> ${TARGET_SYS}-? I understand they're separate variables so they COULD be
> different; just wondering why someone would set them that way.

The only case that comes to mind is when using external toolchains.
They aren't all so kind as to be arch-vendor-os-<foo>.
-- 
Chris Larson
clarson at kergoth dot com
clarson at mvista dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Software Engineer
MontaVista Software, Inc.

Re: [PATCH] cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[Michael Smith]

Chris Larson wrote:
>> Can you explain for a crosscompile noob -- when would TARGET_PREFIX !=
>> ${TARGET_SYS}-? I understand they're separate variables so they COULD be
>> different; just wondering why someone would set them that way.
> 
> The only case that comes to mind is when using external toolchains.
> They aren't all so kind as to be arch-vendor-os-<foo>.

Cool, thanks for the explanation. For what it's worth,

Acked-by: Michael Smith <msmith@cbnco.com>

Re: cross.bbclass: keep TARGET_SYS & TARGET_PREFIX independent.

[Tom Rini]

On Wed, 2009-08-19 at 17:24 +0000, Christopher Larson wrote:
> Signed-off-by: Chris Larson <clarson@mvista.com>
> Acked-by: Michael Smith <msmith@cbnco.com>

Acked-by: Tom Rini <tom_rini@mentor.com>

> 
> ---
> classes/cross.bbclass |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
> 
> diff --git a/classes/cross.bbclass b/classes/cross.bbclass
> index 7debde6..72a0fb7 100644
> --- a/classes/cross.bbclass
> +++ b/classes/cross.bbclass
> @@ -2,6 +2,11 @@
>  # no need for them to be a direct target of 'world'
>  EXCLUDE_FROM_WORLD = "1"
>  
> +# In order to keep TARGET_PREFIX decoupled from TARGET_SYS, let's force the
> +# binary names to match the former, rather than relying on autoconf's implicit
> +# prefixing based on the latter.
> +EXTRA_OECONF_append = " --program-prefix=${TARGET_PREFIX}"
> +
>  # Save PACKAGE_ARCH before changing HOST_ARCH
>  OLD_PACKAGE_ARCH := "${PACKAGE_ARCH}"
>  PACKAGE_ARCH = "${OLD_PACKAGE_ARCH}"


-- 
Tom Rini <tom_rini@mentor.com>
Mentor Graphics Corporation