Nasty tar breakage

Nasty tar breakage

[Jack Mitchell]

I upgraded my host distribution today which pulled in tar 1.27 [1] and
caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
would expect this is something we want to fix before 1.5 goes out the door?

DEBUG: Executing python function sstate_task_prefunc
DEBUG: Python function sstate_task_prefunc finished
DEBUG: Executing python function do_populate_sysroot
DEBUG: Executing shell function sysroot_stage_all
tar: --same-order option cannot be used with -c
Try 'tar --help' or 'tar --usage' for more information.
tar: This does not look like a tar archive
tar: Exiting with failure status due to previous errors
WARNING:
/home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/run.sysroot_stage_all.21827:1
exit 2 from
  tar -xf - -C "$dest"
DEBUG: Python function do_populate_sysroot finished
ERROR: Function failed: sysroot_stage_all (log file is located at
/home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/log.do_populate_sysroot.21827)

[1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html

Re: Nasty tar breakage

[Paul Eggleton]

Hi Jack,

On Friday 11 October 2013 16:21:46 Jack Mitchell wrote:
> I upgraded my host distribution today which pulled in tar 1.27 [1] and
> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
> would expect this is something we want to fix before 1.5 goes out the door?

I think this has unfortunately come up too late for 1.5. If we get the fix in 
master it could go into a 1.5.1 point release however.
 
> DEBUG: Executing python function sstate_task_prefunc
> DEBUG: Python function sstate_task_prefunc finished
> DEBUG: Executing python function do_populate_sysroot
> DEBUG: Executing shell function sysroot_stage_all
> tar: --same-order option cannot be used with -c
> Try 'tar --help' or 'tar --usage' for more information.
> tar: This does not look like a tar archive
> tar: Exiting with failure status due to previous errors
> WARNING:
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-na
> tive/0.60-r0/temp/run.sysroot_stage_all.21827:1 exit 2 from
>   tar -xf - -C "$dest"
> DEBUG: Python function do_populate_sysroot finished
> ERROR: Function failed: sysroot_stage_all (log file is located at
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-na
> tive/0.60-r0/temp/log.do_populate_sysroot.21827)
> 
> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html

Could you please file a bug for this so we can track the issue?

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre

Re: Nasty tar breakage

[Mark Hatle]

On 10/11/13 10:55 AM, Paul Eggleton wrote:
> Hi Jack,
>
> On Friday 11 October 2013 16:21:46 Jack Mitchell wrote:
>> I upgraded my host distribution today which pulled in tar 1.27 [1] and
>> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
>> would expect this is something we want to fix before 1.5 goes out the door?
>
> I think this has unfortunately come up too late for 1.5. If we get the fix in
> master it could go into a 1.5.1 point release however.

Workaround would be to use the buildtools-tarball, since it includes tar.

I'd suggest this be documented as a workaround until we can get the patch in for 
1.5.1.

--Mark

>> DEBUG: Executing python function sstate_task_prefunc
>> DEBUG: Python function sstate_task_prefunc finished
>> DEBUG: Executing python function do_populate_sysroot
>> DEBUG: Executing shell function sysroot_stage_all
>> tar: --same-order option cannot be used with -c
>> Try 'tar --help' or 'tar --usage' for more information.
>> tar: This does not look like a tar archive
>> tar: Exiting with failure status due to previous errors
>> WARNING:
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-na
>> tive/0.60-r0/temp/run.sysroot_stage_all.21827:1 exit 2 from
>>    tar -xf - -C "$dest"
>> DEBUG: Python function do_populate_sysroot finished
>> ERROR: Function failed: sysroot_stage_all (log file is located at
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-na
>> tive/0.60-r0/temp/log.do_populate_sysroot.21827)
>>
>> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html
>
> Could you please file a bug for this so we can track the issue?
>
> Cheers,
> Paul
>

Re: Nasty tar breakage

[Richard Purdie]

On Fri, 2013-10-11 at 16:21 +0100, Jack Mitchell wrote:
> I upgraded my host distribution today which pulled in tar 1.27 [1] and
> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
> would expect this is something we want to fix before 1.5 goes out the door?
> 
> DEBUG: Executing python function sstate_task_prefunc
> DEBUG: Python function sstate_task_prefunc finished
> DEBUG: Executing python function do_populate_sysroot
> DEBUG: Executing shell function sysroot_stage_all
> tar: --same-order option cannot be used with -c
> Try 'tar --help' or 'tar --usage' for more information.
> tar: This does not look like a tar archive
> tar: Exiting with failure status due to previous errors
> WARNING:
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/run.sysroot_stage_all.21827:1
> exit 2 from
>   tar -xf - -C "$dest"
> DEBUG: Python function do_populate_sysroot finished
> ERROR: Function failed: sysroot_stage_all (log file is located at
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/log.do_populate_sysroot.21827)
> 
> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html

Just to be clear, the command is:

tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"

from staging.bbclass. The fix is probably just to drop the s from the
-ps.

Cheers,

Richard

Re: Nasty tar breakage

[Jack Mitchell]

On 11/10/13 17:08, Richard Purdie wrote:
> On Fri, 2013-10-11 at 16:21 +0100, Jack Mitchell wrote:
>> I upgraded my host distribution today which pulled in tar 1.27 [1] and
>> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
>> would expect this is something we want to fix before 1.5 goes out the door?
>>
>> DEBUG: Executing python function sstate_task_prefunc
>> DEBUG: Python function sstate_task_prefunc finished
>> DEBUG: Executing python function do_populate_sysroot
>> DEBUG: Executing shell function sysroot_stage_all
>> tar: --same-order option cannot be used with -c
>> Try 'tar --help' or 'tar --usage' for more information.
>> tar: This does not look like a tar archive
>> tar: Exiting with failure status due to previous errors
>> WARNING:
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/run.sysroot_stage_all.21827:1
>> exit 2 from
>>   tar -xf - -C "$dest"
>> DEBUG: Python function do_populate_sysroot finished
>> ERROR: Function failed: sysroot_stage_all (log file is located at
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/log.do_populate_sysroot.21827)
>>
>> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html
> Just to be clear, the command is:
>
> tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"
>
> from staging.bbclass. The fix is probably just to drop the s from the
> -ps.
>
> Cheers,
>
> Richard
>

I'm afraid that doesn't work:

Exception: CalledProcessError: Command 'cd
/home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux;
find . -type d -print | tar -cf - -C
/home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux
-ps --files-from - | tar -xf - -C
/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux'
returned non-zero exit status 2 with output tar: --same-order option
cannot be used with -c
Try 'tar --help' or 'tar --usage' for more information.
tar: This does not look like a tar archive
tar: Exiting with failure status due to previous errors
find: `standard output': Broken pipe
find: write error

Tar command in staging.bbclass

tar -cf - -C "$src" -p . | tar -xf - -C "$dest"

Re: Nasty tar breakage

[Richard Purdie]

On Fri, 2013-10-11 at 17:15 +0100, Jack Mitchell wrote:
> On 11/10/13 17:08, Richard Purdie wrote:
> > On Fri, 2013-10-11 at 16:21 +0100, Jack Mitchell wrote:
> >> I upgraded my host distribution today which pulled in tar 1.27 [1] and
> >> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
> >> would expect this is something we want to fix before 1.5 goes out the door?
> >>
> >> DEBUG: Executing python function sstate_task_prefunc
> >> DEBUG: Python function sstate_task_prefunc finished
> >> DEBUG: Executing python function do_populate_sysroot
> >> DEBUG: Executing shell function sysroot_stage_all
> >> tar: --same-order option cannot be used with -c
> >> Try 'tar --help' or 'tar --usage' for more information.
> >> tar: This does not look like a tar archive
> >> tar: Exiting with failure status due to previous errors
> >> WARNING:
> >> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/run.sysroot_stage_all.21827:1
> >> exit 2 from
> >>   tar -xf - -C "$dest"
> >> DEBUG: Python function do_populate_sysroot finished
> >> ERROR: Function failed: sysroot_stage_all (log file is located at
> >> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/log.do_populate_sysroot.21827)
> >>
> >> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html
> > Just to be clear, the command is:
> >
> > tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"
> >
> > from staging.bbclass. The fix is probably just to drop the s from the
> > -ps.
> >
> > Cheers,
> >
> > Richard
> >
> 
> I'm afraid that doesn't work:
> 
> Exception: CalledProcessError: Command 'cd
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux;
> find . -type d -print | tar -cf - -C
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux
> -ps --files-from - | tar -xf - -C
> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux'
> returned non-zero exit status 2 with output tar: --same-order option
> cannot be used with -c
> Try 'tar --help' or 'tar --usage' for more information.
> tar: This does not look like a tar archive
> tar: Exiting with failure status due to previous errors
> find: `standard output': Broken pipe
> find: write error
> 
> Tar command in staging.bbclass
> 
> tar -cf - -C "$src" -p . | tar -xf - -C "$dest"

I think this command got copy and pasted into a few parts of the system
so we'll need to hunt them all down and fix them. In classes alone:

$ grep \\-ps * -R
libc-package.bbclass:	tar -cf - -C ${LOCALETREESRC}${datadir} -ps i18n | tar -xf - -C $treedir/${datadir}
libc-package.bbclass:	tar -cf - -C ${LOCALETREESRC}${base_libdir} -ps . | tar -xf - -C $treedir/${base_libdir}
libc-package.bbclass:		tar -cf - -C ${STAGING_DIR_NATIVE}/${prefix_native}/${base_libdir} -ps libgcc_s.* | tar -xf - -C $treedir/${base_libdir}
libc-package.bbclass:	tar -cf - -C $treedir/$parent -ps $(basename ${localedir}) | tar -xf - -C ${PKGD}$parent
package.bbclass:    cmd = 'tar -cf - -C %s -ps . | tar -xf - -C %s' % (dest, dvar)
populate_sdk_deb.bbclass:	tar -cf - -C ${STAGING_ETCDIR_NATIVE} -ps apt | tar -xf - -C ${target_rootfs}/etc
staging.bbclass:		tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"

Cheers,

Richard

Re: Nasty tar breakage

[Jack Mitchell]

On 11/10/13 22:47, Richard Purdie wrote:
> On Fri, 2013-10-11 at 17:15 +0100, Jack Mitchell wrote:
>> On 11/10/13 17:08, Richard Purdie wrote:
>>> On Fri, 2013-10-11 at 16:21 +0100, Jack Mitchell wrote:
>>>> I upgraded my host distribution today which pulled in tar 1.27 [1] and
>>>> caused a nasty breakage. Downgrading to tar 1.26 fixed the problem, I
>>>> would expect this is something we want to fix before 1.5 goes out the door?
>>>>
>>>> DEBUG: Executing python function sstate_task_prefunc
>>>> DEBUG: Python function sstate_task_prefunc finished
>>>> DEBUG: Executing python function do_populate_sysroot
>>>> DEBUG: Executing shell function sysroot_stage_all
>>>> tar: --same-order option cannot be used with -c
>>>> Try 'tar --help' or 'tar --usage' for more information.
>>>> tar: This does not look like a tar archive
>>>> tar: Exiting with failure status due to previous errors
>>>> WARNING:
>>>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/run.sysroot_stage_all.21827:1
>>>> exit 2 from
>>>>    tar -xf - -C "$dest"
>>>> DEBUG: Python function do_populate_sysroot finished
>>>> ERROR: Function failed: sysroot_stage_all (log file is located at
>>>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/temp/log.do_populate_sysroot.21827)
>>>>
>>>> [1] https://lists.gnu.org/archive/html/info-gnu/2013-10/msg00004.html
>>> Just to be clear, the command is:
>>>
>>> tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"
>>>
>>> from staging.bbclass. The fix is probably just to drop the s from the
>>> -ps.
>>>
>>> Cheers,
>>>
>>> Richard
>>>
>>
>> I'm afraid that doesn't work:
>>
>> Exception: CalledProcessError: Command 'cd
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux;
>> find . -type d -print | tar -cf - -C
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/work/x86_64-linux/quilt-native/0.60-r0/sysroot-destdir/home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux
>> -ps --files-from - | tar -xf - -C
>> /home/jack/Work/oe-core.git/test-build/tmp-eglibc/sysroots/x86_64-linux'
>> returned non-zero exit status 2 with output tar: --same-order option
>> cannot be used with -c
>> Try 'tar --help' or 'tar --usage' for more information.
>> tar: This does not look like a tar archive
>> tar: Exiting with failure status due to previous errors
>> find: `standard output': Broken pipe
>> find: write error
>>
>> Tar command in staging.bbclass
>>
>> tar -cf - -C "$src" -p . | tar -xf - -C "$dest"
>
> I think this command got copy and pasted into a few parts of the system
> so we'll need to hunt them all down and fix them. In classes alone:
>
> $ grep \\-ps * -R
> libc-package.bbclass:	tar -cf - -C ${LOCALETREESRC}${datadir} -ps i18n | tar -xf - -C $treedir/${datadir}
> libc-package.bbclass:	tar -cf - -C ${LOCALETREESRC}${base_libdir} -ps . | tar -xf - -C $treedir/${base_libdir}
> libc-package.bbclass:		tar -cf - -C ${STAGING_DIR_NATIVE}/${prefix_native}/${base_libdir} -ps libgcc_s.* | tar -xf - -C $treedir/${base_libdir}
> libc-package.bbclass:	tar -cf - -C $treedir/$parent -ps $(basename ${localedir}) | tar -xf - -C ${PKGD}$parent
> package.bbclass:    cmd = 'tar -cf - -C %s -ps . | tar -xf - -C %s' % (dest, dvar)
> populate_sdk_deb.bbclass:	tar -cf - -C ${STAGING_ETCDIR_NATIVE} -ps apt | tar -xf - -C ${target_rootfs}/etc
> staging.bbclass:		tar -cf - -C "$src" -ps . | tar -xf - -C "$dest"
>
> Cheers,
>
> Richard
>

Ah ok, I did wonder but didn't have the time to check. I'll try to get 
round to eliminating all the occurances on Monday and if it works I'll 
put out an RFC with a view for 1.5.1.

Cheers,

-- 
   Jack Mitchell (jack@embed.me.uk)
   Embedded Systems Engineer
   Cambrideshire, UK
   http://www.embed.me.uk
--

Re: Nasty tar breakage

[Jack Mitchell]

>
> Ah ok, I did wonder but didn't have the time to check. I'll try to get
> round to eliminating all the occurances on Monday and if it works I'll
> put out an RFC with a view for 1.5.1.
>
> Cheers,
>

Ignore. I've just seen your patches. Thanks for the help!

-- 
   Jack Mitchell (jack@embed.me.uk)
   Embedded Systems Engineer
   Cambrideshire, UK
   http://www.embed.me.uk
--