Build issue of nativesdk toolchain

Build issue of nativesdk toolchain

[Luo Zhenhua-B19537]

When I build nativesdk toolchain with recent poky + meta-fsl-ppc + FSL SDK specific layers, build issue similar to error-1 happens. This issue might appear for different gcc related package during different trial, e.g. libgcc, gcc, gcc-cross-initial, gcc-crosssdk, etc. 

When I revert http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=9bfd92243eced921f166d84b5a72d6f774e13413, error-1 build issue goes away, error-2 appears, error-2 also can't be reproduced every time. Does anyone ever meet the similar build issue and any comment on the build error?

[Error-1 logs]: 
OSError: [Errno 39] Directory not empty: '/local/cache-build-32bit/poky/build_p1022ds_release_master/tmp/work-shared/gcc-4.7.2+fsl-r19/gcc-4.7.2+fsl/gcc/cp'

ERROR: The stack trace of python calls that resulted in this exception/failure was:
ERROR:   File "base_do_unpack", line 19, in <module>
ERROR: 
ERROR:   File "base_do_unpack", line 14, in base_do_unpack
ERROR: 
ERROR:   File "/local/cache-build-32bit/poky/bitbake/lib/bb/fetch2/__init__.py", line 1466, in unpack
ERROR:     ud.method.unpack(ud, root, self.d)
ERROR: 
ERROR:   File "/local/cache-build-32bit/poky/bitbake/lib/bb/fetch2/git.py", line 239, in unpack
ERROR:     bb.utils.prunedir(destdir)
ERROR: 
ERROR:   File "/local/cache-build-32bit/poky/bitbake/lib/bb/utils.py", line 561, in prunedir
ERROR:     os.rmdir(os.path.join(root, name))
ERROR: 
ERROR: The code that was being executed was:
ERROR:      0015:    except bb.fetch2.BBFetchException, e:
ERROR:      0016:        raise bb.build.FuncFailed(e)
ERROR:      0017:
ERROR:      0018:
ERROR:  *** 0019:base_do_unpack(d)
ERROR:      0020:
ERROR: [From file: 'base_do_unpack', lineno: 19, function: <module>]
ERROR:      0010:    rootdir = localdata.getVar('WORKDIR', True)
ERROR:      0011:
ERROR:      0012:    try:
ERROR:      0013:        fetcher = bb.fetch2.Fetch(src_uri, localdata)
ERROR:  *** 0014:        fetcher.unpack(rootdir)
ERROR:      0015:    except bb.fetch2.BBFetchException, e:
ERROR:      0016:        raise bb.build.FuncFailed(e)
ERROR:      0017:
ERROR:      0018:
ERROR: [From file: 'base_do_unpack', lineno: 14, function: base_do_unpack]
ERROR: Function failed: base_do_unpack
ERROR: Logfile of failure stored in: /local/cache-build-32bit/poky/build_p1022ds_release_master/tmp/work-shared/gcc-4.7.2+fsl-r19/temp/log.do_unpack.13886
[...snip...]
ERROR: Command Error: exit status: 1  Output:
Applying patch optional_libstdc.patch
patching file gcc/cp/g++spec.c
can't find file to patch at input line 30
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|Index: gcc-4_7-branch/gcc/doc/invoke.texi
|===================================================================
|--- gcc-4_7-branch.orig/gcc/doc/invoke.texi	2012-04-10 10:37:09.343387385 -0700
|+++ gcc-4_7-branch/gcc/doc/invoke.texi	2012-04-10 10:37:49.483389340 -0700
--------------------------
No file to patch.  Skipping patch.
3 out of 3 hunks ignored
can't find file to patch at input line 64
Perhaps you used the wrong -p or --strip option?
The text leading up to this was:
--------------------------
|Index: gcc-4_7-branch/gcc/c-family/c.opt
|===================================================================
|--- gcc-4_7-branch.orig/gcc/c-family/c.opt	2012-04-10 10:17:24.667330076 -0700
|+++ gcc-4_7-branch/gcc/c-family/c.opt	2012-04-10 10:37:49.483389340 -0700
--------------------------
No file to patch.  Skipping patch.
1 out of 1 hunk ignored
patching file gcc/gcc.c
Hunk #1 succeeded at 679 (offset -2 lines).
Patch optional_libstdc.patch does not apply (enforce with -f)
ERROR: Function failed: patch_do_patch
ERROR: Logfile of failure stored in: /local/cache-build-32bit/poky/build_p1022ds_release_master/tmp/work-shared/gcc-4.7.2+fsl-r19/temp/log.do_patch.13815

[Error-2 logs]:
[...snip...]
| log_check: Using /local/cache-build-32bit/poky/build_p5020ds-64b_release_master/tmp/work/ppc64e5500-fsl_networking-linux/fsl-toolchain/1.0-r7/temp/log.do_populate_sdk.26285 as logfile
| Logfile is clean
| DEBUG: Shell function populate_sdk_image finished
| DEBUG: SITE files ['endian-big', 'bit-64', 'powerpc-common', 'common-linux', 'common-glibc', 'powerpc-linux', 'powerpc64-linux', 'common']
| DEBUG: Executing shell function create_sdk_files
| DEBUG: Shell function create_sdk_files finished
| DEBUG: SITE files ['endian-big', 'bit-32', 'powerpc-common', 'common-linux', 'common-glibc', 'powerpc32-linux', 'powerpc-linux', 'common']
| DEBUG: Executing shell function create_sdk_files
| DEBUG: Shell function create_sdk_files finished
| DEBUG: Executing shell function tar_sdk
| tar: ./sysroots/ppc64e5500-fsl_networking-linux/var/lib/rpm/__db.002: file changed as we read it
| DEBUG: Python function do_populate_sdk finished
| ERROR: Function failed: tar_sdk (see /local/cache-build-32bit/poky/build_p5020ds-64b_release_master/tmp/work/ppc64e5500-fsl_networking-linux/fsl-toolchain/1.0-r7/temp/log.do_populate_sdk.26285 for further information)
NOTE: recipe fsl-toolchain-1.0-r7: task do_populate_sdk: Failed


Best Regards,

Zhenhua

Re: Build issue of nativesdk toolchain

[Richard Purdie]

On Tue, 2013-03-26 at 14:29 +0000, Luo Zhenhua-B19537 wrote:
> When I build nativesdk toolchain with recent poky + meta-fsl-ppc + FSL SDK specific layers, build issue similar to error-1 happens. This issue might appear for different gcc related package during different trial, e.g. libgcc, gcc, gcc-cross-initial, gcc-crosssdk, etc. 
> 
> When I revert http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=9bfd92243eced921f166d84b5a72d6f774e13413, error-1 build issue goes away, error-2 appears, error-2 also can't be reproduced every time. Does anyone ever meet the similar build issue and any comment on the build error?

It sounds like you've changes in gcc between some of the different
variants, e.g. different patch sets or SRC_URI.

gcc uses a shared source directory so the same patchset must be used for
all variants. It has nothing to do with the above commit.

Cheers,

Richard

Re: Build issue of nativesdk toolchain

[Luo Zhenhua-B19537]

FSL gcc source is used for following gcc variants, and other gcc variants use Yocto default source. "+fsl" suffix is appended into PV in below gcc recipes. My understanding is that below gcc variants should use a different build folder with gcc-crosssdk, am I misunderstanding something? 

I might need to unify the source for all gcc variants if Yocto doesn't support above scenario.  

Gcc variants with FSL source: 
gcc_4.7.bbappend        gcc-cross-canadian_4.7.bbappend  gcc-fsl.inc               libgcc_4.7.bbappend
gcc-cross_4.7.bbappend  gcc-cross-initial_4.7.bbappend   gcc-runtime_4.7.bbappend

$ cat gcc-fsl.inc 
PV = "4.7.2+fsl"

BINV = "4.7.2"

SRCREV = "c88a2f851197a103296175e0a7a1cb94ab91bcc3"

SRC_URI = "git://git.am.freescale.net/gitolite/sdk/gcc.git;destsuffix=gcc-${PV} \
        file://optional_libstdc.patch \
        file://use-defaults.h-and-t-oe-in-B.patch \
        file://GLIBC_DYNAMIC_LINKER.patch \
"

FILESEXTRAPATHS_prepend := "${THISDIR}/files:" 

Below is the content of the bbappend: 
require gcc-fsl.inc


Best Regards,

Zhenhua


> -----Original Message-----
> From: Richard Purdie [mailto:richard.purdie@linuxfoundation.org]
> Sent: Tuesday, March 26, 2013 11:45 PM
> To: Luo Zhenhua-B19537
> Cc: openembedded-core@lists.openembedded.org
> Subject: Re: [OE-core] Build issue of nativesdk toolchain
> 
> On Tue, 2013-03-26 at 14:29 +0000, Luo Zhenhua-B19537 wrote:
> > When I build nativesdk toolchain with recent poky + meta-fsl-ppc + FSL
> SDK specific layers, build issue similar to error-1 happens. This issue
> might appear for different gcc related package during different trial,
> e.g. libgcc, gcc, gcc-cross-initial, gcc-crosssdk, etc.
> >
> > When I revert
> http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=9bfd92243eced92
> 1f166d84b5a72d6f774e13413, error-1 build issue goes away, error-2 appears,
> error-2 also can't be reproduced every time. Does anyone ever meet the
> similar build issue and any comment on the build error?
> 
> It sounds like you've changes in gcc between some of the different
> variants, e.g. different patch sets or SRC_URI.
> 
> gcc uses a shared source directory so the same patchset must be used for
> all variants. It has nothing to do with the above commit.
> 
> Cheers,
> 
> Richard
>