Re: [PATCH 1/1] libarchive: set CLEANBROKEN

[Robert Yang <liezhi.yang@windriver.com>]

On 04/21/2015 03:39 PM, Richard Purdie wrote:
> On Tue, 2015-03-24 at 22:41 -0700, Robert Yang wrote:
>> Errors when rebuild:
>> checking for xml2-config... /path/to/tmp/sysroots/qemux86-64/usr/bin/crossscripts/xml2-config
>> checking for xmlInitParser in -lxml2... no
>> configure: error: in `/path/to/tmp/work/core2-64-poky-linux/libarchive/3.1.2-r0/libarchive-3.1.2':
>> configure: error: Missing xml2 library
>>
>> Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
>> ---
>>   .../libarchive/libarchive_3.1.2.bb                 |    2 ++
>>   1 file changed, 2 insertions(+)
>>
>> diff --git a/meta/recipes-extended/libarchive/libarchive_3.1.2.bb b/meta/recipes-extended/libarchive/libarchive_3.1.2.bb
>> index 96e2d50..2ac8c5e 100644
>> --- a/meta/recipes-extended/libarchive/libarchive_3.1.2.bb
>> +++ b/meta/recipes-extended/libarchive/libarchive_3.1.2.bb
>> @@ -7,6 +7,8 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=b4e3ffd607d6686c6cb2f63394370841"
>>
>>   DEPENDS = "e2fsprogs-native"
>>
>> +CLEANBROKEN = "1"
>> +
>>   PACKAGECONFIG ?= "libxml2 zlib bz2"
>>
>>   PACKAGECONFIG_append_class-target = "\
>
> I just looked into this I will explain why this worries me. We patch
> libarchive to remove the reference to xml2-config in SRC_URI. For this
> error to appear, the patch must have been removed.
>
> The question is therefore why does it get removed and why isn't that an
> error? Does the CVE patch get removed too?
>
> FWIW, "bitbake libarchive -c configure -f" does not reproduce the error
> and the clean broken is therefore a workaround/hack.
>
> Actually, I just found what does break it is "bitbake libarchive -C
> unpack". Why? Running "make clean" triggers configure to run again since
> after unpack, the configure timestamp is more recent that config.status.
>
> So to fix this, we just need to touch config.status (if it exists)
> before running make clean. We should probably do this in the autotools
> class.

Thanks, I will do more investigations on it.

// Robert

>
> Cheers,
>
> Richard
>
>
>
>
>