[OE-core] [RFC][PATCH] insane.bbclass: Add do_qa_pseudo function to check for common errors listed in pseudo.log

[Martin Jansa]

OK, thanks.

Any insights to the first issue described in commit message? That we know
that is an Actual Problem, but hard to reproduce it.

I'm rebuilding one component over and over again and with pseudo 1.8.1 I
got this issue in 5 rebuilds from 1000, now in the same build with latest
pseudo 1.8.2 backported from oe-core/master I got this issue only 2 times
in 1000 iterations.

Hopefully this insane check will help discover similar issues even without
triggering the QA host-user-contamination check, so that we can detect such
issues even without really trying (it's now included in my regular "bitbake
world" builds which quite often trigger this QA host-user-contamination
issue in glibc-locale. And now we can correlate that with pseudo.log more
easily.

On Thu, Sep 14, 2017 at 12:08 AM, Seebs <seebs at seebs.net> wrote:

> On Wed, 13 Sep 2017 22:50:32 +0200
> Martin Jansa <martin.jansa at gmail.com> wrote:
>
> > inode mismatch:
> > '/OE/build/oe-core/tmp-glibc/work/core2-64-oe-linux/
> systemd/1_234-r0/recipe-sysroot/etc/group'
> > ino 29361846 in db, 29361904 in request.
>
> Just to be clear about this:
>
> This is at least potentially an Actual Problem. It's probably not a
> good idea to habitually ignore these. A thing like this happening means
> that, at some point:
>
> 1. Something created etc/group, and it got inode 29361846 recorded in
> the database.
> 2. Something deleted etc/group, and did NOT record this to the database.
> 3. Something created etc/group, with inode 29361904.
>
> It's possible that the thing creating etc/group didn't record it to the
> database, and something else just referred to it later (like a stat
> call), or that this was generated by the request creating the entry.
>
> Either way, it means changes were made to a pseudo-managed hunk of
> filesystem without pseudo knowing about it, and that is generally bad.
>
> -s
> --
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core at lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-core
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openembedded.org/pipermail/openembedded-core/attachments/20170914/0873aec2/attachment-0002.html>