[PATCH 0/2] bc: upgrade 1.06 to 1.07.1

[PATCH 0/2] bc: upgrade 1.06 to 1.07.1

[Jose Lamego]

bc recipe must be upgraded to latest version in upstream.

Changes in this series:
    include ed to the hosttools required to run builds with oe-core,
    since the new bc version requires the ed command during build

    upgrade bc version from 1.06 to 1.07.1

    update bc license from "GPLv2+ & LGPLv2.1" to "GPLv3+ & LGPLv3"

    unset PARALLEL_MAKE variable in bc recipe to prevent a
    Parallel Make Race during build

Jose Lamego (2):
  conf/bitbake.conf: include ed in hosttools
  bc: upgrade 1.06 to 1.07.1

 meta/conf/bitbake.conf                             |  2 +-
 .../bc/{bc_1.06.bb => bc_1.07.1.bb}                | 22 +++++++++++-----------
 2 files changed, 12 insertions(+), 12 deletions(-)
 rename meta/recipes-extended/bc/{bc_1.06.bb => bc_1.07.1.bb} (28%)

-- 
2.7.4

[PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Jose Lamego]

Some recipes build fail due to missing ed command.

This change includes ed to the hosttools required to
run builds with oe-core.

Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
---
 meta/conf/bitbake.conf | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/conf/bitbake.conf b/meta/conf/bitbake.conf
index 8e4f4bb..64f1a54 100644
--- a/meta/conf/bitbake.conf
+++ b/meta/conf/bitbake.conf
@@ -459,7 +459,7 @@ HOSTTOOLS_DIR = "${TMPDIR}/hosttools"
 # Tools needed to run builds with OE-Core
 HOSTTOOLS += " \
     [ ar as awk basename bash bzip2 cat chgrp chmod chown chrpath cmp cp cpio \
-    cpp cut date dd diff diffstat dirname du echo egrep env expand expr false \
+    cpp cut date dd diff diffstat dirname du echo ed egrep env expand expr false \
     fgrep file find flock g++ gawk gcc getconf getopt git grep gunzip gzip \
     head hostname install ld ldd ln ls make makeinfo md5sum mkdir mknod \
     mktemp mv nm objcopy objdump od patch perl pod2man pr printf pwd python python2 \
-- 
2.7.4

[PATCH 2/2] bc: upgrade 1.06 to 1.07.1

[Jose Lamego]

bc recipe must be upgraded to latest version in upstream.

bc license changed from "GPLv2+ & LGPLv2.1" to "GPLv3+ & LGPLv3"
The new version exhibits a Parallel Make Race that required
unsetting PARALLEL_MAKE variable.
This change was tested with core-image-minimal in qemux86.

Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com>
---
 .../bc/{bc_1.06.bb => bc_1.07.1.bb}                | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
 rename meta/recipes-extended/bc/{bc_1.06.bb => bc_1.07.1.bb} (28%)

diff --git a/meta/recipes-extended/bc/bc_1.06.bb b/meta/recipes-extended/bc/bc_1.07.1.bb
similarity index 28%
rename from meta/recipes-extended/bc/bc_1.06.bb
rename to meta/recipes-extended/bc/bc_1.07.1.bb
index d8c8a86..55b4422 100644
--- a/meta/recipes-extended/bc/bc_1.06.bb
+++ b/meta/recipes-extended/bc/bc_1.07.1.bb
@@ -1,25 +1,25 @@
 SUMMARY = "Arbitrary precision calculator language"
 HOMEPAGE = "http://www.gnu.org/software/bc/bc.html"
 
-LICENSE = "GPLv2+ & LGPLv2.1"
-LIC_FILES_CHKSUM = "file://COPYING;md5=94d55d512a9ba36caa9b7df079bae19f \
-                    file://COPYING.LIB;md5=d8045f3b8f929c1cb29a1e3fd737b499 \
-                    file://bc/bcdefs.h;endline=31;md5=46dffdaf10a99728dd8ce358e45d46d8 \
-                    file://dc/dc.h;endline=25;md5=2f9c558cdd80e31b4d904e48c2374328 \
-                    file://lib/number.c;endline=31;md5=99434a0898abca7784acfd36b8191199"
+LICENSE = "GPLv3+ & LGPLv3"
+LIC_FILES_CHKSUM = "file://COPYING;md5=d32239bcb673463ab874e80d47fae504 \
+                    file://COPYING.LIB;md5=6a6a8e020838b23406c81b19c1d46df6 \
+                    file://bc/bcdefs.h;endline=27;md5=1bf4215734d4b9002ffda7b90b14e8e9 \
+                    file://dc/dc.h;endline=20;md5=be0fc95c3503cb8116eea6acb63a5922 \
+                    file://lib/number.c;endline=31;md5=5c437fc63f83a46d4d769d4dd1e74525"
 
 SECTION = "base"
 DEPENDS = "flex"
-PR = "r3"
 
-SRC_URI = "${GNU_MIRROR}/bc/bc-${PV}.tar.gz \
-           file://fix-segment-fault.patch "
+SRC_URI = "${GNU_MIRROR}/bc/bc-${PV}.tar.gz"
 
-SRC_URI[md5sum] = "d44b5dddebd8a7a7309aea6c36fda117"
-SRC_URI[sha256sum] = "4ef6d9f17c3c0d92d8798e35666175ecd3d8efac4009d6457b5c99cea72c0e33"
+SRC_URI[md5sum] = "cda93857418655ea43590736fc3ca9fc"
+SRC_URI[sha256sum] = "62adfca89b0a1c0164c2cdca59ca210c1d44c3ffc46daf9931cf4942664cb02a"
 
 inherit autotools texinfo update-alternatives
 
+PARALLEL_MAKE = ""
+
 ALTERNATIVE_${PN} = "dc"
 ALTERNATIVE_PRIORITY = "100"
 
-- 
2.7.4

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Richard Purdie]

On Thu, 2017-05-18 at 17:28 -0500, Jose Lamego wrote:
> Some recipes build fail due to missing ed command.
> 
> This change includes ed to the hosttools required to
> run builds with oe-core.
> 
> Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
> Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.in
> tel.com>

We really don't want to add ed here. Which recipes are failing and why
do we need ed to build them? 

Cheers,

Richard

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Khem Raj]

On Thu, May 18, 2017 at 4:12 PM, Richard Purdie
<richard.purdie@linuxfoundation.org> wrote:
> On Thu, 2017-05-18 at 17:28 -0500, Jose Lamego wrote:
>> Some recipes build fail due to missing ed command.
>>
>> This change includes ed to the hosttools required to
>> run builds with oe-core.
>>
>> Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
>> Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.in
>> tel.com>
>
> We really don't want to add ed here. Which recipes are failing and why
> do we need ed to build them?

we also need whoami which kernel uses to identify the build user id

>
> Cheers,
>
> Richard
>
>
> --
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Anders Darander]

* Khem Raj <raj.khem@gmail.com> [170519 01:17]:

> On Thu, May 18, 2017 at 4:12 PM, Richard Purdie
> <richard.purdie@linuxfoundation.org> wrote:
> > On Thu, 2017-05-18 at 17:28 -0500, Jose Lamego wrote:
> >> Some recipes build fail due to missing ed command.

> >> This change includes ed to the hosttools required to
> >> run builds with oe-core.

> >> Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
> >> Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.in
> >> tel.com>

> > We really don't want to add ed here. Which recipes are failing and why
> > do we need ed to build them?

> we also need whoami which kernel uses to identify the build user id


No, we don't want to add whoami...

As fas as I can (in a recent kernel), whoami is only used in
scripts/mkcompile_h. And there, it's only used if $KBUILD_BUILD_USER is
empty. I'd rather see that we set $KBUILD_BUILD_USER to something
deterministic in our kernel.bbclass.

Cheers,
Anders

-- 
Anders Darander, Senior System Architect
ChargeStorm AB / eStorm AB

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 354 bytes --]

On 19 May 2017 at 10:02, Anders Darander <anders@chargestorm.se> wrote:

> As fas as I can (in a recent kernel), whoami is only used in
> scripts/mkcompile_h. And there, it's only used if $KBUILD_BUILD_USER is
> empty. I'd rather see that we set $KBUILD_BUILD_USER to something
> deterministic in our kernel.bbclass.
>

Definitely this.

Ross

[-- Attachment #2: Type: text/html, Size: 776 bytes --]

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Jose Lamego]

[-- Attachment #1.1: Type: text/plain, Size: 1290 bytes --]



On 05/19/2017 04:02 AM, Anders Darander wrote:
> * Khem Raj <raj.khem@gmail.com> [170519 01:17]:
> 
>> On Thu, May 18, 2017 at 4:12 PM, Richard Purdie
>> <richard.purdie@linuxfoundation.org> wrote:
>>> On Thu, 2017-05-18 at 17:28 -0500, Jose Lamego wrote:
>>>> Some recipes build fail due to missing ed command.
> 
>>>> This change includes ed to the hosttools required to
>>>> run builds with oe-core.
> 
>>>> Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
>>>> Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.in
>>>> tel.com>
> 
>>> We really don't want to add ed here. Which recipes are failing and why
>>> do we need ed to build them?

This patch is a requisite for the bc recipe update from 1.06 to 1.07.1,
which fails due to ed command not found during building.

> 
>> we also need whoami which kernel uses to identify the build user id
> 
> 
> No, we don't want to add whoami...
> 
> As fas as I can (in a recent kernel), whoami is only used in
> scripts/mkcompile_h. And there, it's only used if $KBUILD_BUILD_USER is
> empty. I'd rather see that we set $KBUILD_BUILD_USER to something
> deterministic in our kernel.bbclass.
> 
> Cheers,
> Anders
> 

-- 
Jose Lamego | OTC Embedded Platforms & Tools | GDC


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 484 bytes --]

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Richard Purdie]

On Fri, 2017-05-19 at 12:56 +0100, Burton, Ross wrote:
> 
> On 19 May 2017 at 10:02, Anders Darander <anders@chargestorm.se>
> wrote:
> > As fas as I can (in a recent kernel), whoami is only used in
> > scripts/mkcompile_h. And there, it's only used if
> > $KBUILD_BUILD_USER is
> > empty. I'd rather see that we set $KBUILD_BUILD_USER to something
> > deterministic in our kernel.bbclass.
> > 
> Definitely this.

Agreed, I've mentioned this would be preferred already. I'm just living
in hope someone sends a patch!

-- 
RP

Re: [PATCH 2/2] bc: upgrade 1.06 to 1.07.1

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 1389 bytes --]

On 18 May 2017 at 23:28, Jose Lamego <jose.a.lamego@linux.intel.com> wrote:

> bc recipe must be upgraded to latest version in upstream.
>
> bc license changed from "GPLv2+ & LGPLv2.1" to "GPLv3+ & LGPLv3"
> The new version exhibits a Parallel Make Race that required
> unsetting PARALLEL_MAKE variable.
> This change was tested with core-image-minimal in qemux86.


It would be appreciated if you could spend a few minutes looking at what
the build race is and pushing a fix upstream.  Or at least include a
comment showing the failure.

I chased the sed thing as previous releases didn't need it: my build
machine doesn't even have ed installed so HOSTTOOLS won't help.  This
appears to be a regression as 1.06 shipped the file that needs ed to
build.  The makefile now always builds a binary to generate a file that it
then runs through ed, and it doesn't build this tool correctly in a cross
environment either:

| /bin/bash: ./fbc: cannot execute binary file: Exec format error

(it's trying to run an arm binary on x86)

I'd suggest that we approach the upstream maintainer and discuss:
1) should the tarballs ship libmath.h instead of requiring it to be
generated
2) using CC_FOR_BUILD etc from autoconf-archive to build fbc with the host
compiler in cross-compile environments
3) can sed be used instead of ed to reduce the host requirements

Ross

[-- Attachment #2: Type: text/html, Size: 2230 bytes --]

Re: [PATCH 1/2] conf/bitbake.conf: include ed in hosttools

[Burton, Ross]

[-- Attachment #1: Type: text/plain, Size: 316 bytes --]

On 19 May 2017 at 13:54, Jose Lamego <jose.a.lamego@linux.intel.com> wrote:

> This patch is a requisite for the bc recipe update from 1.06 to 1.07.1,
> which fails due to ed command not found during building.
>

I just replied to the bc patch, the new version is broken in several
interesting ways.

Ross

[-- Attachment #2: Type: text/html, Size: 761 bytes --]

Re: [PATCH 2/2] bc: upgrade 1.06 to 1.07.1

[Jose Lamego]

[-- Attachment #1.1: Type: text/plain, Size: 1760 bytes --]



On 05/19/2017 10:25 AM, Burton, Ross wrote:
> 
> On 18 May 2017 at 23:28, Jose Lamego <jose.a.lamego@linux.intel.com
> <mailto:jose.a.lamego@linux.intel.com>> wrote:
> 
>     bc recipe must be upgraded to latest version in upstream.
> 
>     bc license changed from "GPLv2+ & LGPLv2.1" to "GPLv3+ & LGPLv3"
>     The new version exhibits a Parallel Make Race that required
>     unsetting PARALLEL_MAKE variable.
>     This change was tested with core-image-minimal in qemux86.
> 
> 
> It would be appreciated if you could spend a few minutes looking at what
> the build race is and pushing a fix upstream.  Or at least include a
> comment showing the failure.
> 
> I chased the sed thing as previous releases didn't need it: my build
> machine doesn't even have ed installed so HOSTTOOLS won't help.  This
> appears to be a regression as 1.06 shipped the file that needs ed to
> build.  The makefile now always builds a binary to generate a file that
> it then runs through ed, and it doesn't build this tool correctly in a
> cross environment either:
> 
> | /bin/bash: ./fbc: cannot execute binary file: Exec format error
> 
> (it's trying to run an arm binary on x86)
> 
> I'd suggest that we approach the upstream maintainer and discuss:
> 1) should the tarballs ship libmath.h instead of requiring it to be
> generated
> 2) using CC_FOR_BUILD etc from autoconf-archive to build fbc with the
> host compiler in cross-compile environments
> 3) can sed be used instead of ed to reduce the host requirements
> 
> Ross

Agree. I will contact the bc maintainer and work on the build race. I'll
get back to you with the findings. Thanks for the suggestions!

-- 
Jose Lamego | OTC Embedded Platforms & Tools | GDC


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 484 bytes --]

Re: [PATCH 2/2] bc: upgrade 1.06 to 1.07.1

[Jose Lamego]

On 05/19/2017 02:17 PM, Jose Lamego wrote:
>
> On 05/19/2017 10:25 AM, Burton, Ross wrote:
>> On 18 May 2017 at 23:28, Jose Lamego <jose.a.lamego@linux.intel.com
>> <mailto:jose.a.lamego@linux.intel.com>> wrote:
>>
>>     bc recipe must be upgraded to latest version in upstream.
>>
>>     bc license changed from "GPLv2+ & LGPLv2.1" to "GPLv3+ & LGPLv3"
>>     The new version exhibits a Parallel Make Race that required
>>     unsetting PARALLEL_MAKE variable.
>>     This change was tested with core-image-minimal in qemux86.
>>
>>
>> It would be appreciated if you could spend a few minutes looking at what
>> the build race is and pushing a fix upstream.  Or at least include a
>> comment showing the failure.
>>
>> I chased the sed thing as previous releases didn't need it: my build
>> machine doesn't even have ed installed so HOSTTOOLS won't help.  This
>> appears to be a regression as 1.06 shipped the file that needs ed to
>> build.  The makefile now always builds a binary to generate a file that
>> it then runs through ed, and it doesn't build this tool correctly in a
>> cross environment either:
>>
>> | /bin/bash: ./fbc: cannot execute binary file: Exec format error
>>
>> (it's trying to run an arm binary on x86)
>>
>> I'd suggest that we approach the upstream maintainer and discuss:
>> 1) should the tarballs ship libmath.h instead of requiring it to be
>> generated
>> 2) using CC_FOR_BUILD etc from autoconf-archive to build fbc with the
>> host compiler in cross-compile environments
>> 3) can sed be used instead of ed to reduce the host requirements
>>
>> Ross
> Agree. I will contact the bc maintainer and work on the build race. I'll
> get back to you with the findings. Thanks for the suggestions!
>
Update: bc maintainer is working on version 1.07.2, which includes fixes
for the ed and fbc errors and the Parallel Make race. This upgrade will
be released soon, so we may skip 1.07.1 upgrade and wait for the new
stable release.

-- 
Jose Lamego | OTC Embedded Platform & Tools | GDC