Re: [qemu-upstream-4.2-testing test] 77180: regressions - FAIL

[Ian Campbell <ian.campbell@citrix.com>]

On Wed, 2016-01-06 at 18:28 +0000, osstest service owner wrote:
> flight 77180 qemu-upstream-4.2-testing real [real]
> http://logs.test-lab.xenproject.org/osstest/logs/77180/
> 
> Regressions :-(
> 
> Tests which did not succeed and are blocking,
> including tests which could not be run:
>  build-i386                    5 xen-build                 fail REGR. vs. 62044
>  build-amd64                   5 xen-build                 fail REGR. vs. 62044

This is "man/xl.pod.1 around line 854: Expected text after =item, not a
bullet" exposed by the Jessie upgrade.

However per Ian in <22154.35021.750846.695785@mariner.uk.xensource.com> [0]
:

    ...] 4.2 has had no commits since October - in particular, none
    of the recent security fixes - and I would be reluctant to give it a
    veneer of activity.

So our choices WRT these fixes in qemu-xen.git#staging-4.2, given they have
already been pushed, are:

   1. Fix xen.git#staging-4.2 to build on Jessie and wait for it to propagate.
   2. Revert the fixes from qemu-xen.git#staging-4.2 and force push the
      resulting tree (which would be identical to something which previously
      passed).
   3. Rollback qemu-xen.git#staging-4.2.
   4. Force push.
   5. Drop a stop file.
   6. Shave yakks in osstest to allow per-branch selection of the Debian suite
      and pin xen-4.2 and earlier to Wheezy.

#1 is contrary to the quote above, which makes a reasonable argument IMHO.

#3, #4 and #5 all seem like bad ideas to me.

#2 is a bit odd (to have the fixes in the branch but reverted), but seems
least bad compared with #3..#5.

#6 is potentially a lot of work, but might be the right long term answer.

Ian.

[0] http://lists.xenproject.org/archives/html/xen-devel/2016-01/msg00112.html
> 
> Tests which did not succeed, but are not blocking:
>  build-i386-libvirt            1 build-
> check(1)               blocked  n/a
>  test-amd64-i386-xl-qemuu-win7-amd64  1 build-
> check(1)              blocked n/a
>  test-amd64-i386-qemuu-rhel6hvm-intel  1 build-
> check(1)             blocked n/a
>  test-i386-i386-xl-qemuu-winxpsp3  1 build-
> check(1)               blocked  n/a
>  test-amd64-i386-xl-qemuu-ovmf-amd64  1 build-
> check(1)              blocked n/a
>  test-amd64-i386-qemuu-rhel6hvm-amd  1 build-
> check(1)               blocked n/a
>  test-amd64-i386-xl-qemuu-winxpsp3-vcpus1  1 build-
> check(1)         blocked n/a
>  test-amd64-i386-xend-qemuu-winxpsp3  1 build-
> check(1)              blocked n/a
>  test-amd64-i386-xl-qemuu-debianhvm-amd64  1 build-
> check(1)         blocked n/a
>  build-amd64-libvirt           1 build-
> check(1)               blocked  n/a
>  test-amd64-amd64-xl-qemuu-debianhvm-amd64  1 build-
> check(1)        blocked n/a
>  test-amd64-amd64-xl-qemuu-win7-amd64  1 build-
> check(1)             blocked n/a
>  test-amd64-amd64-xl-qemuu-winxpsp3  1 build-
> check(1)               blocked n/a
>  test-amd64-amd64-xl-qemuu-ovmf-amd64  1 build-
> check(1)             blocked n/a
> 
> version targeted for testing:
>  qemuu                5081fc1c773d2a83ec7a867f030323b8b6956cd1
> baseline version:
>  qemuu                c17e602ae64fb24405ebd256679ba9a6f5819086
> 
> Last test of basis    62044  2015-09-15 15:06:42 Z  113 days
> Testing same since    66542  2015-12-18 16:37:10 Z   19 days   11
> attempts
> 
> ------------------------------------------------------------
> People who touched revisions under test:
>   Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> 
> jobs:
>  build-amd64                                                  fail    
>  build-i386                                                   fail    
>  build-amd64-libvirt                                          blocked 
>  build-i386-libvirt                                           blocked 
>  build-amd64-pvops                                            pass    
>  build-i386-pvops                                             pass    
>  test-amd64-i386-qemuu-rhel6hvm-amd                           blocked 
>  test-amd64-amd64-xl-qemuu-debianhvm-amd64                    blocked 
>  test-amd64-i386-xl-qemuu-debianhvm-amd64                     blocked 
>  test-amd64-amd64-xl-qemuu-ovmf-amd64                         blocked 
>  test-amd64-i386-xl-qemuu-ovmf-amd64                          blocked 
>  test-amd64-amd64-xl-qemuu-win7-amd64                         blocked 
>  test-amd64-i386-xl-qemuu-win7-amd64                          blocked 
>  test-amd64-i386-qemuu-rhel6hvm-intel                         blocked 
>  test-amd64-i386-xl-qemuu-winxpsp3-vcpus1                     blocked 
>  test-amd64-i386-xend-qemuu-winxpsp3                          blocked 
>  test-amd64-amd64-xl-qemuu-winxpsp3                           blocked 
>  test-i386-i386-xl-qemuu-winxpsp3                             blocked 
> 
> 
> ------------------------------------------------------------
> sg-report-flight on osstest.test-lab.xenproject.org
> logs: /home/logs/logs
> images: /home/logs/images
> 
> Logs, config files, etc. are available at
>     http://logs.test-lab.xenproject.org/osstest/logs
> 
> Explanation of these reports, and of osstest in general, is at
>     http://xenbits.xen.org/gitweb/?p=osstest.git;a=blob;f=README.email;hb
> =master
>     http://xenbits.xen.org/gitweb/?p=osstest.git;a=blob;f=README;hb=maste
> r
> 
> Test harness code can be found at
>     http://xenbits.xen.org/gitweb?p=osstest.git;a=summary
> 
> 
> Not pushing.
> 
> ------------------------------------------------------------
> commit 5081fc1c773d2a83ec7a867f030323b8b6956cd1
> Author: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> Date:   Fri Dec 18 15:45:14 2015 +0000
> 
>     xenfb: avoid reading twice the same fields from the shared page
>     
>     Reading twice the same field could give the guest an attack of
>     opportunity. In the case of event->type, gcc could compile the switch
>     statement into a jump table, effectively ending up reading the type
>     field multiple times.
>     
>     This is part of XSA-155.
>     
>     Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> 
> commit 74fab2ef4c0ba42af477e9e445c9883cc45cf9e6
> Author: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
> Date:   Fri Dec 18 15:44:58 2015 +0000
> 
>     xen/blkif: Avoid double access to src->nr_segments
>     
>     src is stored in shared memory and src->nr_segments is dereferenced
>     twice at the end of the function.  If a compiler decides to compile
> this
>     into two separate memory accesses then the size limitation could be
>     bypassed.
>     
>     Fix it by removing the double access to src->nr_segments.
>     
>     This is part of XSA-155.
>     
>     Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel