[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Matt Weber]

On some systems the host system or even the cross sysroot can
contain a version.h.  This leads to the wrong file being picked
up and a PLAIN_VERSION undefined error.

This workaround symlinks the version.h into the tool folder to
allow reordering of search folders.

Fixes
http://autobuild.buildroot.net/results/770/7702d5df36a6532aafdbe6e9e62709bbfa058b54/build-end.log
http://autobuild.buildroot.net/results/e34/e3401027d2fb3ce565ca9e2456a427afd3610a87/build-end.log
... additional can be found with these queries ...
http://autobuild.buildroot.net/?reason=uboot-tools-2018.03
http://autobuild.buildroot.net/?reason=uboot-tools-2018.05

Upstream: https://lists.denx.de/pipermail/u-boot/2018-July/336283.html

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
---
 ...rsion.h-workaround-sysroot-inc-order.patch | 53 +++++++++++++++++++
 1 file changed, 53 insertions(+)
 create mode 100644 package/uboot-tools/0004-include-version.h-workaround-sysroot-inc-order.patch

diff --git a/package/uboot-tools/0004-include-version.h-workaround-sysroot-inc-order.patch b/package/uboot-tools/0004-include-version.h-workaround-sysroot-inc-order.patch
new file mode 100644
index 0000000000..89cec5bb34
--- /dev/null
+++ b/package/uboot-tools/0004-include-version.h-workaround-sysroot-inc-order.patch
@@ -0,0 +1,53 @@
+From 546ab8f1e4b91f12b4cc1b4b73bb89a1645a99a4 Mon Sep 17 00:00:00 2001
+From: Matt Weber <matthew.weber@rockwellcollins.com>
+Date: Thu, 26 Jul 2018 22:16:17 -0500
+Subject: [PATCH] include/version.h: workaround sysroot inc order
+
+On some systems the host system or even the cross sysroot can
+contain a version.h.  This leads to the wrong file being picked
+up and a PLAIN_VERSION undefined error.
+
+This workaround symlinks the version.h into the tool folder to
+allow reordering of search folders.
+
+Fixes
+http://autobuild.buildroot.net/results/770/7702d5df36a6532aafdbe6e9e62709bbfa058b54/build-end.log
+http://autobuild.buildroot.net/results/e34/e3401027d2fb3ce565ca9e2456a427afd3610a87/build-end.log
+... additional can be found with these queries ...
+http://autobuild.buildroot.net/?reason=uboot-tools-2018.03
+http://autobuild.buildroot.net/?reason=uboot-tools-2018.05
+
+Upstream: https://lists.denx.de/pipermail/u-boot/2018-July/336283.html
+
+Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
+---
+ tools/env/Makefile | 3 ++-
+ tools/version.h    | 1 +
+ 2 files changed, 3 insertions(+), 1 deletion(-)
+ create mode 120000 tools/version.h
+
+diff --git a/tools/env/Makefile b/tools/env/Makefile
+index 95a03c9..865fbc3 100644
+--- a/tools/env/Makefile
++++ b/tools/env/Makefile
+@@ -11,7 +11,8 @@
+ override HOSTCC = $(CC)
+ 
+ # Compile for a hosted environment on the target
+-HOST_EXTRACFLAGS  = $(patsubst -I%,-idirafter%, $(filter -I%, $(UBOOTINCLUDE))) \
++HOST_EXTRACFLAGS  = -I$(srctree)/tools \
++		$(patsubst -I%,-idirafter%, $(filter -I%, $(UBOOTINCLUDE))) \
+ 		-idirafter $(srctree)/tools/env \
+ 		-DUSE_HOSTCC \
+ 		-DTEXT_BASE=$(TEXT_BASE)
+diff --git a/tools/version.h b/tools/version.h
+new file mode 120000
+index 0000000..bb57607
+--- /dev/null
++++ b/tools/version.h
+@@ -0,0 +1 @@
++../include/version.h
+\ No newline at end of file
+-- 
+2.17.0
+
-- 
2.17.0

[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Thomas Petazzoni]

Hello,

On Thu, 26 Jul 2018 22:50:44 -0500, Matt Weber wrote:
> On some systems the host system or even the cross sysroot can
> contain a version.h.  This leads to the wrong file being picked
> up and a PLAIN_VERSION undefined error.
> 
> This workaround symlinks the version.h into the tool folder to
> allow reordering of search folders.
> 
> Fixes
> http://autobuild.buildroot.net/results/770/7702d5df36a6532aafdbe6e9e62709bbfa058b54/build-end.log
> http://autobuild.buildroot.net/results/e34/e3401027d2fb3ce565ca9e2456a427afd3610a87/build-end.log
> ... additional can be found with these queries ...
> http://autobuild.buildroot.net/?reason=uboot-tools-2018.03
> http://autobuild.buildroot.net/?reason=uboot-tools-2018.05
> 
> Upstream: https://lists.denx.de/pipermail/u-boot/2018-July/336283.html
> 
> Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>

This issue has already been fixed. It was btrfs-progs that was
installing a version.h header directly in usr/include. It has been
fixed by:

  https://git.buildroot.org/buildroot/commit/package/btrfs-progs?id=83b40e1ea2e3ce046d7db442b01c3bfe52886ce8

I agree it is nice to fix it properly upstream so that the presence of
a header with a conflicting name doesn't confuse U-Boot, but I don't
think it's worth having this patch in Buildroot since the problem no
longer occurs.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Matthew Weber]

Thomas,

On Fri, Jul 27, 2018 at 2:03 AM, Thomas Petazzoni
<thomas.petazzoni@bootlin.com> wrote:
> Hello,
>
> On Thu, 26 Jul 2018 22:50:44 -0500, Matt Weber wrote:
>> On some systems the host system or even the cross sysroot can
>> contain a version.h.  This leads to the wrong file being picked
>> up and a PLAIN_VERSION undefined error.
>>
>> This workaround symlinks the version.h into the tool folder to
>> allow reordering of search folders.
>>
>> Fixes
>> http://autobuild.buildroot.net/results/770/7702d5df36a6532aafdbe6e9e62709bbfa058b54/build-end.log
>> http://autobuild.buildroot.net/results/e34/e3401027d2fb3ce565ca9e2456a427afd3610a87/build-end.log
>> ... additional can be found with these queries ...
>> http://autobuild.buildroot.net/?reason=uboot-tools-2018.03
>> http://autobuild.buildroot.net/?reason=uboot-tools-2018.05
>>
>> Upstream: https://lists.denx.de/pipermail/u-boot/2018-July/336283.html
>>
>> Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
>
> This issue has already been fixed. It was btrfs-progs that was
> installing a version.h header directly in usr/include. It has been
> fixed by:
>
>   https://git.buildroot.org/buildroot/commit/package/btrfs-progs?id=83b40e1ea2e3ce046d7db442b01c3bfe52886ce8
>
> I agree it is nice to fix it properly upstream so that the presence of
> a header with a conflicting name doesn't confuse U-Boot, but I don't
> think it's worth having this patch in Buildroot since the problem no
> longer occurs.
>

Figured out why I though it was still an error, 2018.05.x doesn't have
that btrfs patch and occasionally fails.

Matt

[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Thomas Petazzoni]

Hello,

On Fri, 27 Jul 2018 08:46:37 -0500, Matthew Weber wrote:

> > This issue has already been fixed. It was btrfs-progs that was
> > installing a version.h header directly in usr/include. It has been
> > fixed by:
> >
> >   https://git.buildroot.org/buildroot/commit/package/btrfs-progs?id=83b40e1ea2e3ce046d7db442b01c3bfe52886ce8
> >
> > I agree it is nice to fix it properly upstream so that the presence of
> > a header with a conflicting name doesn't confuse U-Boot, but I don't
> > think it's worth having this patch in Buildroot since the problem no
> > longer occurs.
> 
> Figured out why I though it was still an error, 2018.05.x doesn't have
> that btrfs patch and occasionally fails.

I notified Peter about this, and he said he backported it to 2018.05:

  http://lists.busybox.net/pipermail/buildroot/2018-July/226228.html

And indeed:

  https://git.buildroot.org/buildroot/commit/?h=2018.05.x&id=2bc7184ba3b86149e473a266c185352418120abd

So it should already be fixed in 2018.05.

If you look at
http://autobuild.buildroot.net/?reason=uboot-tools-2018.03, the
PLAIN_VERSION issue stopped on July 19. The more recent issue on July
27 is different, it is:

/home/peko/autobuild/instance-0/output/host/lib/gcc/arm-buildroot-linux-musleabihf/6.4.0/libgcc.a(_dvmd_lnx.o): In function `__aeabi_idiv0':
/home/peko/autobuild/instance-0/output/build/host-gcc-final-6.4.0/build/arm-buildroot-linux-musleabihf/libgcc/../../../libgcc/config/arm/lib1funcs.S:1354: undefined reference to `raise'
collect2: error: ld returned 1 exit status

This is due to -static missing while linking, in a purely static musl
configuration. So that's obviously not related to version.h/btrfs-progs.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin (formerly Free Electrons)
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [PATCH] package/uboot-tools: version.h collision

[Matthew Weber]

Thomas,

On Fri, Jul 27, 2018 at 9:00 AM, Thomas Petazzoni
<thomas.petazzoni@bootlin.com> wrote:
> Hello,
>
> On Fri, 27 Jul 2018 08:46:37 -0500, Matthew Weber wrote:
>
>> > This issue has already been fixed. It was btrfs-progs that was
>> > installing a version.h header directly in usr/include. It has been
>> > fixed by:
>> >
>> >   https://git.buildroot.org/buildroot/commit/package/btrfs-progs?id=83b40e1ea2e3ce046d7db442b01c3bfe52886ce8
>> >
>> > I agree it is nice to fix it properly upstream so that the presence of
>> > a header with a conflicting name doesn't confuse U-Boot, but I don't
>> > think it's worth having this patch in Buildroot since the problem no
>> > longer occurs.
>>
>> Figured out why I though it was still an error, 2018.05.x doesn't have
>> that btrfs patch and occasionally fails.
>
> I notified Peter about this, and he said he backported it to 2018.05:
>
>   http://lists.busybox.net/pipermail/buildroot/2018-July/226228.html
>
> And indeed:
>
>   https://git.buildroot.org/buildroot/commit/?h=2018.05.x&id=2bc7184ba3b86149e473a266c185352418120abd
>
> So it should already be fixed in 2018.05.
>
> If you look at
> http://autobuild.buildroot.net/?reason=uboot-tools-2018.03, the
> PLAIN_VERSION issue stopped on July 19. The more recent issue on July
> 27 is different, it is:

Ah true.  Thanks for the clarification.

>
> /home/peko/autobuild/instance-0/output/host/lib/gcc/arm-buildroot-linux-musleabihf/6.4.0/libgcc.a(_dvmd_lnx.o): In function `__aeabi_idiv0':
> /home/peko/autobuild/instance-0/output/build/host-gcc-final-6.4.0/build/arm-buildroot-linux-musleabihf/libgcc/../../../libgcc/config/arm/lib1funcs.S:1354: undefined reference to `raise'
> collect2: error: ld returned 1 exit status
>
> This is due to -static missing while linking, in a purely static musl
> configuration. So that's obviously not related to version.h/btrfs-progs.
>

Yeah, noticed that one in my inbox this morning.

Matt