IRC Log for the Virtualization team's meeting on 2013-11-12

2013-11-12 16:02:31     -->     cfontana (~claudio@g226200155.adsl.alicedsl.de) has joined #linaro-meeting-2
2013-11-12 16:54:15     -->     kimphill_ (~kim@2.29.127.85) has joined #linaro-meeting-2
2013-11-12 17:01:28     -->     julieng (~julieng@185.25.64.249) has joined #linaro-meeting-2
2013-11-12 17:02:04     -->     apritzel (~~aprzywar@g224196199.adsl.alicedsl.de) has joined #linaro-meeting-2
2013-11-12 17:07:02     apritzel        isn't there supposed to be a meeting right now?
2013-11-12 17:07:13     kimphill_       I thought so too
2013-11-12 17:07:19     cfontana        I think so
2013-11-12 17:07:35     apritzel        ototo: ^^
2013-11-12 17:08:35     cfontana        maybe hangouts?
2013-11-12 17:08:39     julieng from leaving calendar, ototo is still on holiday.
2013-11-12 17:09:02     apritzel        julieng: good point
2013-11-12 17:09:14     apritzel        actually I don't insist on a meeting ;-)
2013-11-12 17:09:21     -->     pm215 (~petmay01@fw-tnat.cambridge.arm.com) has joined #linaro-meeting-2
2013-11-12 17:09:41     <--     sry1 (~B08248@gate-tx3.freescale.com) has quit (Quit: Leaving.)
2013-11-12 17:09:46     -->     chazy- (chazy@ec2-54-225-101-41.compute-1.amazonaws.com) has joined #linaro-meeting-2
2013-11-12 17:10:16     chazy-  hi eveyone, all here?
2013-11-12 17:10:25     chazy-  sorry I'm late, I was waiting for Serge, but he's on vacation.
2013-11-12 17:10:33     chazy-  stupid me.
2013-11-12 17:10:39     cfontana        hi
2013-11-12 17:10:42      *      ctlaugh_ is here
2013-11-12 17:10:48      *      kimphill_ is here
2013-11-12 17:10:59      *      pm215 is here
2013-11-12 17:11:08      *      julieng is here
2013-11-12 17:11:27      *      apritzel is here (sigh ;-)
2013-11-12 17:11:54     chazy-  ok, looking for actions from last meeting
2013-11-12 17:12:27     chazy-  I was supposed to help ctlaugh_ with mach-virt/QEMU.
2013-11-12 17:12:28     chazy-  I did.
2013-11-12 17:12:32     chazy-  ctlaugh_ are you all set?
2013-11-12 17:12:47     ctlaugh_        Yes - other than Midway being 1/2 dead, it's working great.
2013-11-12 17:13:02     chazy-  half dead? the firmware update did not help?
2013-11-12 17:13:28     ctlaugh_        It still seems to be in progress.  The last I heard was that Calxeda was sending some new boards.
2013-11-12 17:13:32     chazy-  ototo was supposed to fix Xen recognition for patches.linaro.org
2013-11-12 17:13:41     chazy-  julieng: this has been taken care of, yes?
2013-11-12 17:13:48     apritzel        ctlaugh_: are you around later today? Maybe I can help on this
2013-11-12 17:13:57     chazy-  ctlaugh_: ok, if you need something from me, then let me know.
2013-11-12 17:14:02     ctlaugh_        apritzel: Help with what?
2013-11-12 17:14:03     julieng chazy-: Yes.
2013-11-12 17:14:19     apritzel        getting Midway to run again
2013-11-12 17:14:34     apritzel        (1/2 dead -> 0 dead)
2013-11-12 17:14:52     chazy-  apritzel: if you can help ctlaugh_ it would be great.  Thanks.
2013-11-12 17:15:00     chazy-  ok, let's do a quick round-table:
2013-11-12 17:15:03     ctlaugh_        Ah -- well, my node is running (for now) but the whole system seems to be barely functioning.  I've heard new hardware is coming.  Nothing really for me to do.
2013-11-12 17:15:15     kimphill_       pardon me, but is it ok if I go first since there's only a couple of hours left in the hospital visitation hours?
2013-11-12 17:15:27     chazy-  kimphill_: sure, go ahead
2013-11-12 17:15:31     kimphill_       thanks
2013-11-12 17:15:38     kimphill_       CARD-815: "RM: VFIO for platform devices"
2013-11-12 17:15:38     kimphill_       - as mentioned in Connect update slides, PCI rebind sysfs interface is
2013-11-12 17:15:38     kimphill_         inconsistent, yet compatibility needs to be retained.  In theory,
2013-11-12 17:15:39     kimphill_         this could mean that we can just omit the 3rd patch in the series
2013-11-12 17:15:41     kimphill_         (to the vfio-pci driver).  Meanwhile, sry has pinged Greg-KH, who
2013-11-12 17:15:43     kimphill_         responded with 'I might like them more once we're out of the current
2013-11-12 17:15:44     chazy-  #topic Round Table
2013-11-12 17:15:45     kimphill_         merge window'.  So it's not clear whether just the first two patches
2013-11-12 17:15:47     kimphill_         will be acceptable - but there's one way to find out - re-submit them
2013-11-12 17:15:49     kimphill_         come the v3.13-rc2 release.  Note that omission of PCI may bring up
2013-11-12 17:15:51     kimphill_         some additional work items.
2013-11-12 17:15:53     kimphill_       LNG-128: testing KVM networking
2013-11-12 17:15:55     kimphill_         this is now in the lava-ci loop and therefore has been closed.
2013-11-12 17:15:58     kimphill_       "linux-lng doesn't boot up on qemu smp" bug [no associated card]
2013-11-12 17:15:59     kimphill_       [https://bugs.launchpad.net/linaro-networking/+bug/1240283]
2013-11-12 17:16:01     kimphill_         bisecting v3.10..v3.11 points to the fix being commit 70e5975d
2013-11-12 17:16:03     kimphill_         "clockevents: Prefer CPU local devices over global devices"
2013-11-12 17:16:05     kimphill_         yet re-applying that commit (amongst others) doesn't solve the problem
2013-11-12 17:16:08     julieng chazy-: You forgot to start the meeting with the right tag :)
2013-11-12 17:16:09     kimphill_         on v3.10.  Still investigating.
2013-11-12 17:16:11     kimphill_       EOF
2013-11-12 17:17:29     chazy-  kimphill_: are you running with architected timers enabled?
2013-11-12 17:17:41     chazy-  if not, you need to apply an arcane fix by Marc Rutland which I think landed in 3.12
2013-11-12 17:17:43     kimphill_       not sure
2013-11-12 17:18:05     kimphill_       ok, but 3.11 works ok
2013-11-12 17:18:27     cfontana        #startmeeting
2013-11-12 17:18:27     linarobot`      Meeting started Tue Nov 12 17:18:27 2013 UTC.  The chair is cfontana. Information about MeetBot at http://wiki.debian.org/MeetBot.
2013-11-12 17:18:27     linarobot`      Useful Commands: #action #agreed #help #info #idea #link #topic.
2013-11-12 17:18:31     cfontana        ups
2013-11-12 17:18:46     chazy-  cfontana: fine, no worries, I forgot to do it.
2013-11-12 17:18:51     chazy-  (this is Serge's thing really)
2013-11-12 17:18:53     cfontana        #endmeeting
2013-11-12 17:18:54     linarobot`      Meeting ended Tue Nov 12 17:18:53 2013 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
2013-11-12 17:18:54     linarobot`      Minutes:        http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.18.html
2013-11-12 17:18:54     linarobot`      Minutes (text): http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.18.txt
2013-11-12 17:18:54     linarobot`      Log:            http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.18.log.html
2013-11-12 17:19:00     chazy-  argh, stop now :)
2013-11-12 17:19:01     kimphill_       you mean "clocksource: Add generic dummy timer driver"?
2013-11-12 17:19:03     chazy-  #startmeeting
2013-11-12 17:19:03     linarobot`      Meeting started Tue Nov 12 17:19:02 2013 UTC.  The chair is chazy-. Information about MeetBot at http://wiki.debian.org/MeetBot.
2013-11-12 17:19:03     linarobot`      Useful Commands: #action #agreed #help #info #idea #link #topic.
2013-11-12 17:19:07     chazy-  #topic Round Table
2013-11-12 17:19:22     chazy-  kimphill_: let's take it offline.
2013-11-12 17:19:25     chazy-  over e-mail.
2013-11-12 17:19:36     kimphill_       ok
2013-11-12 17:19:52     chazy-  apritzel: you're up (cfontana, chazy)
2013-11-12 17:20:03     apritzel        * (spent last week in Calxeda office in Austin/TX)
2013-11-12 17:20:04     apritzel        * KVM debugging: (Card-414)
2013-11-12 17:20:04     apritzel        ** wading through existing x86 code to learn about this
2013-11-12 17:20:04     apritzel        ** sketching preliminary implementation based on ARMv7 ARM
2013-11-12 17:20:04     apritzel        ** wondering about fight between guest debugging and QEMU gdb stub
2013-11-12 17:20:04     apritzel        *** this does not seem to be handled on x86?
2013-11-12 17:20:11     apritzel        * PCIe QEMU host controller emulation (Card needed!)
2013-11-12 17:20:11     apritzel        ** learning about current Calxeda driver state
2013-11-12 17:20:12     apritzel        ** reading some PCI documentation to understand ARM/x86 differences
2013-11-12 17:20:12     apritzel        ** getting hold of a Midway development board with PCI slots
2013-11-12 17:20:12     apritzel        ** Calxeda has high interest in QEMU model, how about Linaro?
2013-11-12 17:20:12     apritzel        *** does that need to go through TSC or similar?
2013-11-12 17:20:36     apritzel        * KVM bug(?)
2013-11-12 17:20:36     apritzel        ** dmidecode inside a KVM guest kills the guest kernel
2013-11-12 17:20:36     apritzel        ** actually KVM returns an unhandled data abort due to mmapping /dev/mem to QEMU
2013-11-12 17:20:36     apritzel         as an error
2013-11-12 17:20:36     apritzel        ** QEMU then does abort()
2013-11-12 17:20:37     apritzel        ** need to discuss whether inject_dabt() makes more sense here
2013-11-12 17:20:37     apritzel        *** have a simple patch for this, which works
2013-11-12 17:21:05     apritzel        EOF
2013-11-12 17:21:31      *      kimphill_ is off, if that's ok
2013-11-12 17:21:41     chazy-  kimphill_: fine, thanks for joining.
2013-11-12 17:21:46     kimphill_       thanks
2013-11-12 17:22:17     chazy-  apritzel: can you ask the KVM x86 guys (on the kvm list, cc the kvmarm list) about the guest debugging question?
2013-11-12 17:22:23     chazy-  I need some more context to help out there.
2013-11-12 17:22:29     apritzel        yes, started this already
2013-11-12 17:22:47     apritzel        (on IRC)
2013-11-12 17:22:53     chazy-  PCIe, yes, we need a card, yes needs to be approved by TSC/LEG before we start spending too much time.  Yes, Linaro is interested.
2013-11-12 17:23:05     chazy-  (apritzel, yes, I was it on IRC, will take a look).
2013-11-12 17:23:37     apritzel        so should Calxeda push this as a member or is this not needed?
2013-11-12 17:23:53     chazy-  apritzel: it's fine to do initial poking around on the PCIe front so we know what options we have, but don't do any actual work on that before you are assigned to it.
2013-11-12 17:24:04     apritzel        OK
2013-11-12 17:24:07     chazy-  No need to push, I'll create the card and argue for it in front of TSC/Opscom/LEG
2013-11-12 17:24:19     apritzel        Thanks!
2013-11-12 17:24:44     chazy-  apritzel: please also prioritize closing CARD-361 if we have upstream Arndale support.
2013-11-12 17:25:04     chazy-  cfontana: you're up (chazy, ctlaugh_)
2013-11-12 17:25:13     apritzel        chazy-: I checked and didn't see real improvements on that front, but: yes, will do
2013-11-12 17:25:49     cfontana        * I have been working on [CARD-215] Instruction set support
2013-11-12 17:25:49     cfontana        for QEMU.
2013-11-12 17:25:49     cfontana        Developed the initial patch started at Connect with Alex Bennee
2013-11-12 17:25:49     cfontana        a little bit further, and posted RFC patch to the mailing list,
2013-11-12 17:25:49     cfontana        and then a v2 fixing a bug.
2013-11-12 17:25:51     cfontana        Hopefully that gets included so that we can start submitting the
2013-11-12 17:25:54     cfontana        actual stuff based on the new decoder structure.
2013-11-12 17:25:56     chazy-  apritzel: in that case, send me an e-mail explaining why it doesn't work (or which patches are needed) so I have some ammo to the Samsung LT.
2013-11-12 17:26:24     cfontana        * For [VIRT-166] (testing tool) the only thing missing
2013-11-12 17:26:24     cfontana        is to implement further advanced simd and fp instructions.
2013-11-12 17:26:25     cfontana        The existing stuff is enough at the moment to use as testing
2013-11-12 17:26:25     cfontana        for Card-215. Did not mark it as complete yet, since it isn't.
2013-11-12 17:26:33     cfontana        EOF
2013-11-12 17:27:14     <--     kimphill_ (~kim@2.29.127.85) has quit (Ping timeout: 264 seconds)
2013-11-12 17:27:19     chazy-  cfontana: thanks!
2013-11-12 17:27:27     cfontana        :)
2013-11-12 17:27:34     chazy-  I'm up (ctlaugh_, julieng)
2013-11-12 17:28:01     chazy-  Since last:
2013-11-12 17:28:02     chazy-   - Recovering from LCU
2013-11-12 17:28:03     chazy-   - Writing summaries from LCU and follow-up work
2013-11-12 17:28:05     chazy-   - Measurements on KVM/ARM
2013-11-12 17:28:06     chazy-   - Created cards in JIRA based on LCU13 discussions.
2013-11-12 17:28:08     chazy-  Working on:
2013-11-12 17:28:09     chazy-   - CARD-363: Reworking kernel patches, lots of interesting races.
2013-11-12 17:28:09     chazy-  EOF
2013-11-12 17:28:20     chazy-  ctlaugh_: you're up (julieng, pm215)
2013-11-12 17:28:32     ctlaugh_        * Submitted a patch to libvirt to enable it to create virt machines with virtio devices… waiting for feedback / acceptance
2013-11-12 17:28:32     ctlaugh_        * Working on CI for libvirt.  Trying to obtain access to hardware for native build in Jenkins.
2013-11-12 17:28:32     ctlaugh_        * Beginning work to get LAVA tests setup for libvirt and openstack… looking at openstack-ubuntu-testing on launchpad (LEG-329)
2013-11-12 17:28:39     ctlaugh_        EOF
2013-11-12 17:29:16     chazy-  ctlaugh_: is it a requirement that we do native builds for libvirt?
2013-11-12 17:30:05     ctlaugh_        It would be a lot easier.  I spent about 6 hours with Riku at Connect trying to get a cross build working with no luck.  Native build works just fine.  We should have the capability to do that.
2013-11-12 17:30:52     chazy-  ctlaugh_: hmm yeah, but not if it requires a massive amount of infrastructure changes.  But ok, I'll look into it, thanks for the info.
2013-11-12 17:31:03     chazy-  julieng: you're up (pm215)
2013-11-12 17:31:48     julieng Since last meeting:
2013-11-12 17:31:49     julieng     - Was on holiday until yesterday
2013-11-12 17:31:51     julieng     - Ack/Reply to patch on Xen mailing
2013-11-12 17:31:53     julieng Plan:
2013-11-12 17:31:55     julieng     - Will be off friday
2013-11-12 17:31:57     julieng     - Debug xen on the arndale with Lava. Thanks for the mail chazy
2013-11-12 17:31:59     julieng     - Send different patch to allow Xen running on midway
2013-11-12 17:32:01     julieng EOF
2013-11-12 17:32:59     chazy-  julieng: can you talk to Ian and determine if there's anything left for us to do on CARD-733 or if we can close it?
2013-11-12 17:33:27     julieng chazy-: Yes, I will.
2013-11-12 17:33:39     chazy-  thanks!
2013-11-12 17:33:43     chazy-  pm215: you're up
2013-11-12 17:33:48     pm215   Progress:
2013-11-12 17:33:51     pm215    * resent/pinged patchsets that got lost
2013-11-12 17:34:07     pm215    * upgraded Arndale board to kvm-next kernel (took a while)
2013-11-12 17:34:15     pm215    * started cleaning up the kvm v8 control patches
2013-11-12 17:34:19     pm215    * updated cpu-host patches and mach-virt
2013-11-12 17:34:22     pm215   Plans:
2013-11-12 17:34:26     pm215    * v7 mach-virt and cpu-host QEMU patches [prerequisite for v8 kvm control]
2013-11-12 17:34:29     pm215    * v8 KVM control in QEMU
2013-11-12 17:34:43     pm215    * annual performance review
2013-11-12 17:34:44     pm215   EOF
2013-11-12 17:35:05     chazy-  pm215: any idea of remaining time on the v7 and v8 work, respectively?
2013-11-12 17:35:18     chazy-  (the mach-virt and cpu-host and kvm control work that is)
2013-11-12 17:35:43     pm215   the v7 stuff is done except for possible tweaks resulting from v8 cleanup
2013-11-12 17:36:25     pm215   v8 I need to actually test, read through the VOS code a little more carefully, look at Mian's code to do PSCI boot for v8 cores
2013-11-12 17:36:43     chazy-  ok, sounds good. thanks.
2013-11-12 17:36:44     pm215   plan is to get it done (in the sense of "submitted upstream") by end of month
2013-11-12 17:36:58     pm215   I haven't bothered resending cpu-host or mach-virt yet because upstream's in hardfreeze
2013-11-12 17:37:08     chazy-  sounds like that should hold.
2013-11-12 17:37:10     chazy-  ok
2013-11-12 17:37:13     chazy-  #topic Open Floor
2013-11-12 17:37:21     chazy-  Anything else?
2013-11-12 17:37:55     chazy-  going once
2013-11-12 17:38:07     chazy-  going twice
2013-11-12 17:38:21     chazy-  ok, we're done.
2013-11-12 17:38:25     chazy-  thanks everyone!
2013-11-12 17:38:27     chazy-  #endmeeting
2013-11-12 17:38:27     linarobot`      Meeting ended Tue Nov 12 17:38:27 2013 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)
2013-11-12 17:38:27     linarobot`      Minutes:        http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.19.html
2013-11-12 17:38:28     linarobot`      Minutes (text): http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.19.txt
2013-11-12 17:38:28     linarobot`      Log:            http://irclogs.linaro.org/meeting-logs/linaro-meeting-2/2013/linaro-meeting-2.2013-11-12-17.19.log.html

LEG/Engineering/Virtualization/TeamMeetingReports/2013-11-12-log (last modified 2013-11-28 04:16:22)