Re: [Xen-devel] [PATCH for-4.9 v3 3/3] xen/livepatch: Don't crash on encountering STN_UNDEF relocations

[Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>]

On Sat, Jun 24, 2017 at 06:28:16PM +0100, Julien Grall wrote:
> Hi Konrad,
> 
> On 06/23/2017 03:46 PM, Konrad Rzeszutek Wilk wrote:
> > On Fri, Jun 23, 2017 at 03:36:51PM +0100, Julien Grall wrote:
> > > 
> > > 
> > > On 23/06/17 15:35, Konrad Rzeszutek Wilk wrote:
> > > > On Fri, Jun 23, 2017 at 02:45:22PM +0100, Andrew Cooper wrote:
> > > > > On 23/06/17 14:43, Julien Grall wrote:
> > > > > > Hi,
> > > > > > 
> > > > > > On 23/06/17 14:33, Andrew Cooper wrote:
> > > > > > > On 23/06/17 14:32, Julien Grall wrote:
> > > > > > > > Hi Andrew,
> > > > > > > > 
> > > > > > > > I am a bit confused, the title says "PATCH for-4.9 v3 3/3". I 
> > > > > > > > haven't
> > > > > > > > been CCed on the first two patches. Does it mean you are only 
> > > > > > > > looking
> > > > > > > > at this patch to be in 4.9?
> > > > > > > 
> > > > > > > Sorry - I messed up the CC lists.  The correctness of this patch 
> > > > > > > does
> > > > > > > depend on the previous two, so all 3 are looking for inclusion.
> > > > > > 
> > > > > > Given that we don't have livepatch testing in osstest how much test
> > > > > > have we done on those 3 patches?
> > > > > 
> > > > > There is testing in OSSTest.
> > > > 
> > > > Hurray hurray hurray!
> > > > > 
> > > > > I've manually run each of the scenarios, including with my livepatch
> > > > > which has a STN_UNDEF relocation.
> > > > > 
> > > > > I don't know what testing Konrad has done.
> > > > 
> > > > I run a version of the same tests that are in OSSTest (basically an 
> > > > earlier
> > > > version of the Perl code) and I have done it on x86 and on ARM32.
> > > > 
> > > > And I also run the standalone OSSTest (on x86)
> > > > 
> > > > And then I also do a livepatch using the livepatch-build-tools on x86 to
> > > > patch some silly function.
> > > > 
> > > > So from a testing perspective these patches have been tested very 
> > > > exhaustively.
> > > 
> > > Well it has not been tested on ARM64 :). I am about to do that.
> > 
> > /me facepalm.
> > 
> > I really need to get myself a working ARM64 box that is not expensive.
> > 
> > 
> > Also attached is the poor-man livepatch_test.perl script that mirrors
> > what OSSTest does.
> 
> I have got an error when executing the script after applying the "nop"
> livepatch:
> 
> Executing: 'xen-livepatch apply xen_nop:' ..Applying xen_nop... completed
> .. OK!
> Executing: '[ `xl info| grep "xen_m" | grep or | sed s/.*:// | uniq | wc -l`
> == 1 ]:' ..sh: 1: [: 1: unexpected operator
> FAILED (got 512, expected: 0)

Yeah I only get that when I execute it from the serial console.
When I SSH it works fine.

> 
> But this looks like a script error than livepatch. Although, I was not able
> to spot the error in the script.

Neither could I. I do also have an earlier variant of this testing script
in bash (with less test-cases).
> 
> Cheers,
> 
> -- 
> Julien Grall

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