Re: [Xen-devel] [xen-4.8-testing test] 114505: regressions - FAIL

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 16.10.17 at 11:14, <andrew.cooper3@xxxxxxxxxx> wrote:
> On 15/10/17 20:45, osstest service owner wrote:
>> flight 114505 xen-4.8-testing real [real]
>> http://logs.test-lab.xenproject.org/osstest/logs/114505/ 
>>
>> Regressions :-(
>>
>> Tests which did not succeed and are blocking,
>> including tests which could not be run:
>>  test-xtf-amd64-amd64-2 48 xtf/test-hvm64-lbr-tsx-vmentry fail REGR. vs. 
> 114173
>>
>> Tests which are failing intermittently (not blocking):
>>  test-xtf-amd64-amd64-5 48 xtf/test-hvm64-lbr-tsx-vmentry fail in 114454 
> pass in 114505
> 
> Ian: These tests exercise something very machine specific, and the XTF
> tests really do need tying to specific hardware when making regression
> considerations.
> 
> Jan: This highlights that TSX/VMEntry failure fixes probably want
> backporting to before Xen 4.9.  IIRC, the 6 patches needed are:

So I'm mildly confused by this request:

> e3eb84e33c36 (only as a functional prerequisite)
> 9b93c6b3695b: x86/vmx: introduce vmx_find_msr()
> 7f11aa4b2b1f: x86/vmx: optimize vmx_read/write_guest_msr()
> d6e9f8d4f35d: x86/vmx: fix vmentry failure with TSX bits in LBR
> f97838bbd980: x86: Move microcode loading earlier

Up to here, everything is in 4.9 already afaict. Considering the
context here is a 4.8 test report, did you perhaps mean to ask
for this on 4.8 (and possibly also 4.7)? If so, I'm not really sure -
these changes taken together look a little large for the gain
they provide.

> 20f1976b4419: x86/vmx: Fix vmentry failure because of invalid LER on 
> Broadwell

I'll see to pull this one in for 4.9.1.

Jan


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