All of lore.kernel.org
 help / color / mirror / Atom feed
From: Scott Garman <scott.a.garman@intel.com>
To: "yocto@yoctoproject.org" <yocto@yoctoproject.org>
Subject: Re: Trying to re-use sstate.
Date: Thu, 09 Feb 2012 15:14:32 -0800	[thread overview]
Message-ID: <4F345358.8060103@intel.com> (raw)
In-Reply-To: <CABcZAN=kZCy5=puLzDofDAf5SHuGJBZt_+H+pAUp-HbnoG+VOQ@mail.gmail.com>

On 02/06/2012 07:08 AM, Chris Larson wrote:
> On Mon, Feb 6, 2012 at 12:53 AM, David Nyström<david.nystrom@enea.com>  wrote:
>> On 02/04/2012 05:51 AM, Scott Garman wrote:
>>>
>>> Hello,
>>>
>>> I'm trying to re-use the sstate from one build directory and move it to
>>> another. It's not working - bitbake is rebuilding *everything*.
>>>
>>> My hunch is the reason for this is because bblayers.conf must differ
>>> between the two build directories, because it uses full paths to the
>>> layer directories.
>>>
>>> By default, when you source oe-init-build-env for the first time, it
>>> generates a bblayers.conf file with full paths to the default layers,
>>> meta and meta-yocto.
>>>
>>> I thought the point of sstate was that I could copy it between build
>>> directories - or even different hosts - and save myself the effort of
>>> rebuilding everything if the remaining metadata was the same?
>>>
>>> Scott
>>>
>> We have the same issue,
>>
>> I tried to setup a buildbot to warm up sstate caches over night,
>> and let developers use this common sstate to speed up build times.
>> Epic fail, since paths usually are not consistent.
>> And it seems you can't even fake  an "identical path" with symlinks
>> either, bitbake detects the symlink and rebuilds everything.
>
>
> Afaik your BBLAYERS should not be in your signatures at all, unless
> something in the metadata is accessing it. For what it's worth, I
> reuse sstate every day, but don't use the poky/oe-core scripts
> (oe-init-build-env).

As a follow-up to this thread, I spent a bit of time today trying to 
reproduce/isolate the issue, and I could not do so. Changing the paths 
to layers correctly rebuilt from sstate, and adding an empty layer also 
didn't trigger a rebuild. So I'm not sure what I saw previously, but 
unfortunately my original example of this is no longer available for me 
to work with.

I apologize for causing any confusion, but it's still quite possible we 
have some issues - I'll definitely use bitbake-diffsig to dig deeper the 
next time I see it.

Scott

-- 
Scott Garman
Embedded Linux Engineer - Yocto Project
Intel Open Source Technology Center


  reply	other threads:[~2012-02-09 23:14 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-02-04  4:51 Trying to re-use sstate Scott Garman
2012-02-06  7:53 ` David Nyström
2012-02-06 15:08   ` Chris Larson
2012-02-09 23:14     ` Scott Garman [this message]
2012-02-07 14:59 ` McClintock Matthew-B29882
2012-02-14 17:58   ` [yocto] " Martin Jansa
2012-02-14 17:58     ` Martin Jansa
2012-02-15 15:34     ` [yocto] " Martin Jansa
2012-02-15 15:34       ` Martin Jansa

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4F345358.8060103@intel.com \
    --to=scott.a.garman@intel.com \
    --cc=yocto@yoctoproject.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.