[poky] [PULL] two sstate fixes

[Gary Thomas]

On 11/07/2010 05:49 PM, Tian, Kevin wrote:
>> From: Richard Purdie [mailto:rpurdie at linux.intel.com]
>> Sent: Monday, November 08, 2010 4:42 AM
>>
>> On Fri, 2010-11-05 at 15:53 +0800, Tian, Kevin wrote:
>>> meta/classes/package.bbclass |    1 +
>>>   meta/conf/distro/poky.conf   |    2 +-
>>>   2 files changed, 2 insertions(+), 1 deletion(-)
>>>
>>> commit bb9d4efa4076cfcf0998ecafb6d6b0969b97d224
>>> Author: Kevin Tian<kevin.tian at intel.com>
>>> Date:   Fri Nov 5 15:16:50 2010 +0800
>>>
>>>      poky.conf: add PWD to BB_HASHBASE_WHITELIST
>>>
>>>      Without doing this I can't reuse prebuilts generated from one directory in another
>>>      build directory which is really not comfortable for developers. Due to difference
>>>      on PWD variable, base hash becomes different for every sstate package.
>>>
>>>      So here PWD is added to the hase base whitelist, which makes most prebuilts reusable
>>>      now in a different build directory. However there still remains one issue about
>>>      do_populate_sysroot sstate package. Absolute path is encoded in this task which
>>>      then blocks reuse on them. Perhaps we need some redesign for do_populate_sysroot
>> into
>>>      two stages, with all absolute path related operations moved to the 2nd stage. This
>>>      way then the 1st stage can be reused.
>>
>>>      Partially fix [BUGID #522]
>>>
>>>      Signed-off-by Kevin Tian<kevin.tian at intel.com>
>>>
>>> commit 599ccdfa9d53e43fb2e44fdfd57de5d91868428b
>>> Author: Kevin Tian<kevin.tian at intel.com>
>>> Date:   Tue Oct 26 15:54:43 2010 +0800
>>>
>>>      package.bbclass: make sure 'sysroots' created before lockfile
>>>
>>>      package sstate requires a lock under sysroots/, which however may not be
>>>      created when sstate_setscene functions are executed and then causes failures.
>>>      here we make sure 'sysroots' created before do_package_setscene is executed.
>>>
>>>      Signed-off-by: Kevin Tian<kevin.tian at intel.com>
>>> Pull URL: http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=tk/master
>>
>> Good catches, I've merged these, thanks!
>>
>> You're right btw, we shouldn't be using hardcoded paths and we should
>> fix that. Can you confirm you were using staging packages from a recent
>> poky build though (0.9 release at least) as there were some bugs fixed
>> in that area coming up to release.
>>
>
> I forgot the exact commit after the rebase before sending pull request, but I know
> it's around the 0.9 release. So I'll make another build to make sure it's still the problem. :-)

These changes seem to have vastly improved this mechanism.  I'm still
testing, but I was [finally] able to actually use the sstate info to
make a new build.

Hooray :-)

-- 
------------------------------------------------------------
Gary Thomas                 |  Consulting for the
MLB Associates              |    Embedded world
------------------------------------------------------------