[poky] SSTATE builds more broken than I thought

[Tian, Kevin]

>From: Richard Purdie
>Sent: Monday, November 15, 2010 1:58 AM
>
>On Fri, 2010-11-12 at 16:55 -0700, Gary Thomas wrote:
>> On 11/12/2010 04:31 PM, Richard Purdie wrote:
>> > On Fri, 2010-11-12 at 01:35 -0700, Gary Thomas wrote:
>> >> On 11/12/2010 01:30 AM, Richard Purdie wrote:
>> >>> On Fri, 2010-11-12 at 01:18 -0700, Gary Thomas wrote:
>> >>>>> Are these all files or are they symlinks? Are they just present for
>> >>>>> deploy-ipk or are there others?
>> >>>>
>> >>>> They are files.
>> >>>>
>> >>>> Sorry, but I don't understand your other question.  These are the only
>> >>>> extra files polluting the tree, except for Python *.py? object files
>> >>>> which are also always created.
>> >>>
>> >>> Are there any files related to other sstate tasks such as
>> >>> populate-sysroot*.tgz or package*.tgz or is there just *deploy-ipk*.tgz?
>> >>>
>> >>> In other words I'm wondering if this is something specific to the
>> >>> deploy-ipk sstate task...
>> >>
>> >> The only extra files created are *deploy-ipk*.tgz
>> >> This most certainly is restricted to the deploy-ipk sstate task and only
>> >> happens if SSTATE_MIRRORS is defined.
>> >
>> > This should be fixed in ae98f7eacb9e61fe086d88dc694b4c651af9fee3, it
>> > turns out to be an issue with the local fetcher not searching DL_DIR.
>>
>> Trying the same process as in bug #526, it failed with this error:
>[...]
>
>Right, I messed up those commits. I should know better than make commits
>when travelling and jetlagged.
>
>I ended up reverting the first fixes and pushing a set of different
>fixes.
>
>Previously you (and I think Kevin Tian) reporting bitbake spending a
>huge amount of time thinking about shared state. When I was debugging
>this I noticed a potential cause of that, my local test setups were
>evidently just too fast to notice it. I added another commit:
>
>http://git.pokylinux.org/cgit.cgi/poky/commit/?id=89929e1f283c8508c505c9731ad93388
>0abf22a1

thanks for the quick fix. I'll verify it later. If basehash/taskhash don't change,
the result should be soon. Or else another full rebuild is required as the prebuilt base.

btw, I have a quick scan about cases failing to reuse prebuilt (on a built in the weekend
at a commit before above commits), there're 31:

NOTE: package calibrateproto-0.0+git0+1da6fd1e2c7a49648245c98481fabea8b9690a8c-r2: task do_install: Succeeded
NOTE: package sat-solver-0.0-git0+aa799f7bae0ec055e0e527203635001bb7346dbc-r1: task do_install: Succeeded
NOTE: package bzip2-1.0.5-r3: task do_install: Succeeded
NOTE: package libxcalibrate-0.0+git0+209d83af61ed38a002c8096377deac292b3e396c-r0: task do_install: Succeeded
NOTE: package hal-0.5.14-r2: task do_install: Succeeded
NOTE: package task-poky-boot-1.0-r7: task do_install: Succeeded
NOTE: package trace-cmd-1.0.4+git0+0d252224626bd6926324f023a65f20c165232891-r1: task do_install: Succeeded
NOTE: package procps-3.2.7-r9: task do_install: Succeeded
NOTE: package xtscal-0.6.3-r12: task do_install: Succeeded
NOTE: package kexec-tools-2.0.1-r0: task do_install: Succeeded
NOTE: package poky-image-sato-1.0-r0: task do_install: Succeeded
NOTE: package connman-0.56-r2: task do_install: Succeeded
NOTE: package gnome-vfs-2.24.3-r0: task do_install: Succeeded
NOTE: package libzypp-0.0-git0+4494797d5b0369365b1af63921de45b197ead64f-r3: task do_install: Succeeded
NOTE: package oprofileui-0.0+svnr197-r0: task do_install: Succeeded
NOTE: package zypper-1.4.7-git0+9eb0e248e06c8d20ad054be2439149d9ede37531-r1: task do_install: Succeeded
NOTE: package gst-plugins-base-0.10.29-r1: task do_install: Succeeded
NOTE: package libowl-av-0.0+svnr398-r4: task do_install: Succeeded
NOTE: package gaku-0.0+svnr399-r3: task do_install: Succeeded
NOTE: package owl-video-0.0+svnr394-r2: task do_install: Succeeded
NOTE: package bluez4-4.66-r0: task do_install: Succeeded
NOTE: package gst-plugins-bad-0.10.19-r2: task do_install: Succeeded
NOTE: package eds-dbus-2.30+git0+7337d11aed576e7caaa12b4e881ad8d33668799f-r1: task do_install: Succeeded
NOTE: package tasks-0.16-r0: task do_install: Succeeded
NOTE: package dates-0.4.11+git0+514185dc1f6588085fda41eb59898b93d0487dd4-r0: task do_install: Succeeded
NOTE: package contacts-0.12+git0+19853893fdb595de6aa59db0d9dc2f9451ed2933-r0: task do_install: Succeeded
NOTE: package gst-plugins-good-0.10.23-r0: task do_install: Succeeded
NOTE: package gst-meta-base-0.10-r8: task do_install: Succeeded
NOTE: package linux-wrs-2.6.34+git0+5adcf6fbcd7491d8d3e1805929f575041d439235_0+aae69fdf104b0a9d7b3710f808aac6ab303490f7-r12: task do_install: Succeeded
NOTE: package webkit-gtk-1.3.2+svnr62027-r0: task do_install: Succeeded
NOTE: package web-webkit-0.0+svnr110-r2: task do_install: Succeeded

I haven't got time looking into all of them. For random 3 (bzip2, gnome-vfs, webkit-gtk),
they have the same symptom on populate_sysroot, e.g:

basehash changed from c7c1d0c7ea052eef6f6665b8959b44fc to 3eb4a010cbfb3b15df49d433f6593b0c
Variable BB_TASKHASH value changed from 9e077b25e1e2c50d4f8a09a9356a6a6c to 0d6608b425d8a59b21f5874b4914b412
Hash for dependent task /home/poky/tasks/poky/meta/recipes-extended/bzip2/bzip2_1.0.5.bb.do_install changed from c4cba89fa491ecb0916a56f1e75c5f17 to c4e7f58c385e384749cc20ca1bd8fccf

* basehash varies while no variable is explicitly listed by bitbake-diffsigs. I expected
to see the variable causing difference to be marked explicitly here.
* taskhash varies due to do_install, however bitbake-diffsigs doesn't list how do_install
checksum is calculated. I made a quick comparison between run.do_install scripts, with
major difference still on expanded variables. There may still have some whitelist work
to be done.

I used same source tree for two builds, which share same sstate-cache directory.

>
>which should make things faster (n^2 faster where n is the number of
>tasks).
>
>Thanks again for testing and reporting the issues, I believe this code
>will make a significant difference for people once it works reliably and
>we're close to making that happen.
>

Agree. :-)

Thanks,
Kevin