[OE-core] [PATCH] automake: Fix issue with tar configure failing with large UID/GIDs

Kumar Gala galak at kernel.crashing.org
Tue Aug 2 16:08:27 UTC 2011


On Aug 2, 2011, at 11:04 AM, Tom Rini wrote:

> On 08/02/2011 09:01 AM, Kumar Gala wrote:
>> 
>> On Aug 2, 2011, at 10:34 AM, Tom Rini wrote:
>> 
>>> On 08/02/2011 08:19 AM, Phil Blundell wrote:
>>>> On Tue, 2011-08-02 at 03:14 -0500, Kumar Gala wrote:
>>>>> +-_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) pax cpio none'
>>>>> ++_am_tools='gnutar m4_if([$1], [ustar], [plaintar]) cpio pax none'
>>>> 
>>>> Have you discussed that with upstream?  If not, are you confident that
>>>> preferring cpio will not simply mean that we are swapping one deficiency
>>>> for another?
>>> 
>>> A bug was filed with upstream automake.  At issue is that pax does not
>>> fail gracefully when it cannot create ustar archives while cpio does.
>>> 
>>>>> +Index: automake-1.11.1/Makefile.in
>>>>> +===================================================================
>>>>> +--- automake-1.11.1.orig/Makefile.in
>>>>> ++++ automake-1.11.1/Makefile.in
>>>>> +@@ -44,7 +44,7 @@ am__aclocal_m4_deps = $(top_srcdir)/m4/a
>>>>> + 	$(top_srcdir)/m4/missing.m4 $(top_srcdir)/m4/mkdirp.m4 \
>>>>> + 	$(top_srcdir)/m4/options.m4 $(top_srcdir)/m4/runlog.m4 \
>>>>> + 	$(top_srcdir)/m4/sanity.m4 $(top_srcdir)/m4/strip.m4 \
>>>>> +-	$(top_srcdir)/m4/substnot.m4 $(top_srcdir)/m4/tar.m4 \
>>>>> ++	$(top_srcdir)/m4/substnot.m4 \
>>>>> + 	$(top_srcdir)/configure.ac
>>>>> + am__configure_deps = $(am__aclocal_m4_deps) $(CONFIGURE_DEPENDENCIES) \
>>>>> + 	$(ACLOCAL_M4)
>>>> 
>>>> Can you explain why this hunk is necessary?
>>> 
>>> That I don't recall.  Possibly some sort of re-gen hell I was trying /
>>> needing to avoid.
>>> 
>>>> 
>>>>> +do_configure () {
>>>>> +	touch ${S}/Makefile.in
>>>>> +	autotools_do_configure
>>>>> +}
>>>> 
>>>> ... and this one?
>>> 
>>> Didn't do that one.  Kumar? :)
>> 
>> I think I pulled that from the SB3 change you made.
> 
> Gah, you're right.  I think it was more "avoid re-gen" bits.

Tom,

Do you mind take a stab at updating this patch based on feedback as your far more familiar with that I am ;)

- k



More information about the Openembedded-core mailing list