Re: [xen-4.4-testing test] 66583: regressions - FAIL

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

On Tue, 2015-12-22 at 14:40 +0000, Stefano Stabellini wrote:
> On Tue, 22 Dec 2015, Jan Beulich wrote:
> > > > > On 22.12.15 at 14:55, <stefano.stabellini@eu.citrix.com> wrote:
> > > On Mon, 21 Dec 2015, Jan Beulich wrote:
> > > > > > > On 20.12.15 at 13:25, <osstest-admin@xenproject.org> wrote:
> > > > > flight 66583 xen-4.4-testing real [real]
> > > > > http://logs.test-lab.xenproject.org/osstest/logs/66583/ 
> > > > > 
> > > > > Regressions :-(
> > > > > 
> > > > > Tests which did not succeed and are blocking,
> > > > > including tests which could not be run:
> > > > >  build-amd64-prev              5 xen-build                 fail
> > > > > REGR. vs. 
> > > 66458
> > > > >  build-i386-prev               5 xen-build                 fail
> > > > > REGR. vs. 
> > > 66458
> > > > 
> > > > Ian, Stefano,
> > > > 
> > > > is one of you two looking into this?
> > > > 
> > > > block-qcow.c: In function 'get_cluster_offset':
> > > > block-qcow.c:444:3: error: 'tmp_ptr' may be used uninitialized in
> > > > this 
> > > function [-Werror=maybe-uninitialized]
> > > >    memcpy(tmp_ptr, l1_ptr, 4096);
> > > >    ^
> > > > block-qcow.c:619:7: error: 'tmp_ptr2' may be used uninitialized in
> > > > this 
> > > function [-Werror=maybe-uninitialized]
> > > >    if (write(s->fd, tmp_ptr2, 4096) != 4096) {
> > > >        ^
> > > > cc1: all warnings being treated as errors
> > > > 
> > > /home/osstest/build.66583.build-amd64-
> > > prev/xen/tools/blktap/drivers/../../../to
> > > ols/Rules.mk:89: recipe for target 'block-qcow.o' failed
> > > > make[5]: *** [block-qcow.o] Error 1
> > > 
> > > What does build-amd64-prev means? I cannot find the explanation of
> > > the
> > > test anywhere.
> > 
> > Iiuc it's the build pf the previous version, needed to test n-1 -> n
> > migration. I.e. (in line with the respective 4.3 failure) it indicates
> > that 4.3's qemu doesn't build.
> 
> Thanks for the explanation, it was very helpful.
> 
> From the logs what is not building is actually tools/blktap, not QEMU
> though, but I cannot reproduce the failure.

I think we want to backport:

commit 345e44a85d71a1a910385f33c7f1ba3683026d18
Author: Dario Faggioli <dario.faggioli@citrix.com>
Date:   Fri Jun 20 16:09:00 2014 +0200

    blktap2: Fix two 'maybe uninitialized' variables
    
    for which gcc 4.9.0 complains about, like this:
    
    block-qcow.c: In function `get_cluster_offset':
    block-qcow.c:431:3: error: `tmp_ptr' may be used uninitialized in this function
    [-Werror=maybe-uninitialized]
       memcpy(tmp_ptr, l1_ptr, 4096);
       ^
    block-qcow.c:606:7: error: `tmp_ptr2' may be used uninitialized in this
    function [-Werror=maybe-uninitialized]
       if (write(s->fd, tmp_ptr2, 4096) != 4096) {
           ^
    cc1: all warnings being treated as errors
    /home/dario/Sources/xen/xen/xen.git/tools/blktap2/drivers/../../../tools/Rules.mk:89:
     recipe for target 'block-qcow.o' failed
    make[5]: *** [block-qcow.o] Error 1
    
    The proper behavior is to return upon allocation failure.
    About what to return, 0 seems the best option, looking
    at both the function and the call sites.
    
    Signed-off-by: Dario Faggioli <dario.faggioli@citrix.com>
    Acked-by: Ian Jackson <ian.jackson@eu.citrix.com>


(this will have been exposed by the update to Jessie's compiler)

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel