All of lore.kernel.org
 help / color / mirror / Atom feed
From: Martin Jansa <martin.jansa@gmail.com>
To: Mike Looijmans <mike.looijmans@topic.nl>
Cc: openembedded-core@lists.openembedded.org
Subject: Re: Multiple MACHINE building is broken in OE-core?
Date: Tue, 29 Jan 2013 12:27:06 +0100	[thread overview]
Message-ID: <20130129112706.GL16904@jama.palm1.palmone.com> (raw)
In-Reply-To: <5107A295.2060307@topic.nl>

[-- Attachment #1: Type: text/plain, Size: 2977 bytes --]

On Tue, Jan 29, 2013 at 11:21:09AM +0100, Mike Looijmans wrote:
> On 01/29/2013 10:40 AM, Martin Jansa wrote:
> > On Tue, Jan 29, 2013 at 10:29:07AM +0100, Mike Looijmans wrote:
> >> I often have several hardware boards sharing 99% of periferals and
> >> configuration. Usually they all have the exact same CPU, and the
> >> difference is in the minor details like the screen resolution.
> >>
> >> In "classic" OE I was used to sharing the work, by just changing the
> >> MACHINE variable I could create an image for a slightly changed board in
> >> a matter of minutes.
> >>
> >> In OE-core this seems to be horribly broken, unless there's some other
> >> procedure to follow.
> >>
> >> If I change the MACHINE, it insists on rebuilding all kinds of packages
> >> that don't need rebuilding at all. Instead of minutes, it takes an hour
> >> to switch machines. And if I change back, the same thing happens all
> >> over again.
> >>
> >> In addition to needlessly rebuilding dozens of packages for the same
> >> architecture, it often "starts" halfway through, so it tries to run the
> >> "deploy" phase for a package while the workspace has already been
> >> removed in the rm_work phase of the previous build. Which fails, so I
> >> have to manually reset stamps for that package and let it start over
> >> again (while usually the package shouldn't be rebuilt in the first place).
> >>
> >> Is building for several machines no longer supported in OE-core? Or is
> >> there some secret way of setting up directories and variables that will
> >> make it work as it used to?
> >
> > It's more strict and every machine now has own sysroot (so it's doing a
> > bit more after MACHINE change then it did with OE-Classic), but it still
> > works. I'm using it like this.
> 
> How are you using it? Because for me it doesn't work at all, it results 
> in build errors and needless rebuilds on every switch. I've tried in 
> three different environments, and it never works.
> 
> (Needless rebuild: Built package X for machine A. Switch to B, and it 
> will build X again, which isn't needed because they're for the same 
> architecture and the package isn't dependent on the machine. Switch back 
> to A, and it will AGAIN build package X, which is beyond just needless, 
> it's a bug in my opinion.)

There were some bugs (lot of them reported by me in bugzilla) but in
current state it works correctly as long as checksums are correct.

Also depends on what you mean by "build",
do_configure+do_compile+do_install+do_package? or just
do_populate_sysroot?

> > Use openembedded-core/scripts/sstate-diff-machines.sh to compare your
> > machines and make sure that different checksums are only for
> > MACHINE_ARCH recipes.
> 
> How? The tool only explains its commandline, but doesn't explain what 
> it's supposed to do...

There is example in script header.

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 205 bytes --]

  parent reply	other threads:[~2013-01-29 11:42 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-01-29  9:29 Multiple MACHINE building is broken in OE-core? Mike Looijmans
2013-01-29  9:40 ` Martin Jansa
2013-01-29 10:21   ` Mike Looijmans
2013-01-29 10:48     ` Burton, Ross
     [not found]       ` <510B73EB.8060907@topic.nl>
     [not found]         ` <CAJTo0LbkOMXqZkHd-JAcNKxiOmXTy+TXPEEJ-CkPc1Jx3Ujekg@mail.gmail.com>
2013-02-01 12:59           ` Mike Looijmans
2013-02-01 14:00             ` Burton, Ross
2013-02-09 19:39             ` Mike Looijmans
2013-02-10 14:55               ` Martin Jansa
2013-02-10 19:02                 ` Mike Looijmans
2013-02-10 19:59                   ` Martin Jansa
2013-02-18 11:04                     ` OE-core] How do I find out why are packages being rebuilt? Mike Looijmans
2013-02-18 11:11                       ` Burton, Ross
2013-02-18 13:14                         ` Mike Looijmans
2013-02-18 13:19                           ` Burton, Ross
2013-02-18 17:28                           ` Martin Jansa
2013-01-29 11:27     ` Martin Jansa [this message]
2013-02-06  8:02 Multiple MACHINE building is broken in OE-core? Cooper Jr., Franklin
2013-02-06  8:56 ` Martin Jansa
2013-02-07 18:00   ` Cooper Jr., Franklin

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=20130129112706.GL16904@jama.palm1.palmone.com \
    --to=martin.jansa@gmail.com \
    --cc=mike.looijmans@topic.nl \
    --cc=openembedded-core@lists.openembedded.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.