Re: [Xen-devel] [Xen-unstable] regression in pci passthrough to HVM guests due to commit 568da4f8c43d2e5b614964c6aefd768de3e3af14 "pt-irq fixes and improvements".

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 04/08/14 13:29, Sander Eikelenboom wrote:
> Hi Jan / Andrew,
>
> I'm experiencing a regression in pci passthrough to HVM guests due to commit 
> 568da4f8c43d2e5b614964c6aefd768de3e3af14 "pt-irq fixes and improvements".
>
> Before it used to be fine to shutdown and restart guests with the same pci 
> devices passed through. After this commit it doesn't, the device is passed 
> through and visible but doesn't function properly (for instance when passing 
> through a USB card, a "lsusb" fails.
>
> From the logs i see there is (at least) a problem with unmapping the irqs at 
> the 
> shutdown of the guest, after this commit it gives:
>
> (XEN) [2014-08-04 11:15:48.783] irq.c:2119: dom1: forcing unbind of pirq 87
> (XEN) [2014-08-04 11:15:48.783] irq.c:2119: dom1: forcing unbind of pirq 86
> (XEN) [2014-08-04 11:15:48.783] irq.c:2119: dom1: forcing unbind of pirq 85
> (XEN) [2014-08-04 11:15:48.783] irq.c:2119: dom1: forcing unbind of pirq 84
>
> While before this commit it gives:
> (XEN) [2014-08-04 09:00:02.361] io.c:305: d2: unbind: m_gsi=87 g_gsi=16 
> device=0 intx=0
> (XEN) [2014-08-04 09:00:02.361] io.c:363: d2 unmap: m_irq=87 device=0 intx=0
> (XEN) [2014-08-04 09:00:02.361] io.c:305: d2: unbind: m_gsi=86 g_gsi=27 
> device=64 intx=195
> (XEN) [2014-08-04 09:00:02.361] io.c:363: d2 unmap: m_irq=86 device=64 
> intx=195
> (XEN) [2014-08-04 09:00:02.361] io.c:305: d2: unbind: m_gsi=85 g_gsi=27 
> device=64 intx=195
> (XEN) [2014-08-04 09:00:02.361] io.c:363: d2 unmap: m_irq=85 device=64 
> intx=195
> (XEN) [2014-08-04 09:00:02.361] io.c:305: d2: unbind: m_gsi=84 g_gsi=27 
> device=64 intx=195
> (XEN) [2014-08-04 09:00:02.361] io.c:363: d2 unmap: m_irq=84 device=64 
> intx=195
> (XEN) [2014-08-04 09:00:04.497] AMD-Vi: Disable: device id = 0x400, domain = 
> 2, paging mode = 4
> (XEN) [2014-08-04 09:00:04.497] AMD-Vi: Setup I/O page table: device id = 
> 0x400, type = 0x1, root table = 0x54ef79000, domain = 0, paging mode = 3
> (XEN) [2014-08-04 09:00:04.497] AMD-Vi: Re-assign 0000:04:00.0 from dom2 to 
> dom0
>
> This is for a device with MSI-X enabled, lspci from the guest:
>
> 00:05.0 USB controller [0c03]: NEC Corporation uPD720200 USB 3.0 Host 
> Controller [1033:0194] (rev 03) (prog-if 30 [XHCI])
>         Subsystem: Micro-Star International Co., Ltd. Device [1462:4257]
>         Physical Slot: 5
>         Control: I/O- Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- 
> Stepping- SERR- FastB2B- DisINTx+
>         Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- 
> <TAbort- <MAbort- >SERR- <PERR- INTx-
>         Latency: 0, Cache Line Size: 64 bytes
>         Interrupt: pin A routed to IRQ 36
>         Region 0: Memory at f3070000 (64-bit, non-prefetchable) [size=8K]
>         Capabilities: [50] Power Management version 3
>                 Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA 
> PME(D0-,D1-,D2-,D3hot-,D3cold-)
>                 Status: D0 NoSoftRst+ PME-Enable- DSel=0 DScale=0 PME-
>         Capabilities: [70] MSI: Enable- Count=1/1 Maskable- 64bit+
>                 Address: 0000000000000000  Data: 0000
>         Capabilities: [90] MSI-X: Enable+ Count=8 Masked-
>                 Vector table: BAR=0 offset=00001000
>                 PBA: BAR=0 offset=00001080
>         Capabilities: [a0] Express (v2) Endpoint, MSI 00
>                 DevCap: MaxPayload 128 bytes, PhantFunc 0, Latency L0s 
> unlimited, L1 unlimited
>                         ExtTag- AttnBtn- AttnInd- PwrInd- RBE+ FLReset-
>                 DevCtl: Report errors: Correctable- Non-Fatal- Fatal- 
> Unsupported-
>                         RlxdOrd+ ExtTag- PhantFunc- AuxPwr- NoSnoop+
>                         MaxPayload 128 bytes, MaxReadReq 512 bytes
>                 DevSta: CorrErr- UncorrErr- FatalErr- UnsuppReq- AuxPwr+ 
> TransPend-
>                 LnkCap: Port #0, Speed 5GT/s, Width x1, ASPM L0s L1, Latency 
> L0 <4us, L1 unlimited
>                         ClockPM+ Surprise- LLActRep- BwNot-
>                 LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- 
> CommClk-
>                         ExtSynch- ClockPM- AutWidDis- BWInt- AutBWInt-
>                 LnkSta: Speed 5GT/s, Width x1, TrErr- Train- SlotClk+ 
> DLActive- BWMgmt- ABWMgmt-
>                 DevCap2: Completion Timeout: Not Supported, TimeoutDis+
>                 DevCtl2: Completion Timeout: 50us to 50ms, TimeoutDis-
>                 LnkCtl2: Target Link Speed: 5GT/s, EnterCompliance- 
> SpeedDis-, Selectable De-emphasis: -6dB
>                          Transmit Margin: Normal Operating Range, 
> EnterModifiedCompliance- ComplianceSOS-
>                          Compliance De-emphasis: -6dB
>                 LnkSta2: Current De-emphasis Level: -3.5dB, 
> EqualizationComplete-, EqualizationPhase1-
>                          EqualizationPhase2-, EqualizationPhase3-, 
> LinkEqualizationRequest-
>         Capabilities: [100 v4] #1033
>         Kernel driver in use: xhci_hcd
>
>
>
> I also tried using "pci=nomsi" for the guest, to rule out anything msi(-x 
> specific) and i end up with:
>
> Before this commit:
> (XEN) [2014-08-04 11:57:20.346] irq.c:270: Dom1 PCI link 0 changed 5 -> 0
> (XEN) [2014-08-04 11:57:20.352] irq.c:270: Dom1 PCI link 1 changed 10 -> 0
> (XEN) [2014-08-04 11:57:20.357] irq.c:270: Dom1 PCI link 2 changed 11 -> 0
> (XEN) [2014-08-04 11:57:20.363] irq.c:270: Dom1 PCI link 3 changed 5 -> 0
> (XEN) [2014-08-04 11:58:17.382] AMD-Vi: Disable: device id = 0x400, domain = 
> 1, paging mode = 4
> (XEN) [2014-08-04 11:58:17.382] AMD-Vi: Setup I/O page table: device id = 
> 0x400, type = 0x1, root table = 0x55d00e000, domain = 0, paging mode = 3
> (XEN) [2014-08-04 11:58:17.382] AMD-Vi: Re-assign 0000:04:00.0 from dom1 to 
> dom0
>
> After this commit:
> (XEN) [2014-08-04 11:30:08.923] irq.c:270: Dom1 PCI link 0 changed 5 -> 0
> (XEN) [2014-08-04 11:30:08.928] irq.c:270: Dom1 PCI link 1 changed 10 -> 0
> (XEN) [2014-08-04 11:30:08.934] irq.c:270: Dom1 PCI link 2 changed 11 -> 0
> (XEN) [2014-08-04 11:30:08.939] irq.c:270: Dom1 PCI link 3 changed 5 -> 0
> (XEN) [2014-08-04 11:31:16.112] AMD-Vi: Disable: device id = 0x400, domain = 
> 1, paging mode = 4
> (XEN) [2014-08-04 11:31:16.112] AMD-Vi: Setup I/O page table: device id = 
> 0x400, type = 0x1, root table = 0x55d00e000, domain = 0, paging mode = 3
> (XEN) [2014-08-04 11:31:16.112] AMD-Vi: Re-assign 0000:04:00.0 from dom1 to 
> dom0
>
> So that doesn't seem to be changed .. so it's probably a msi(-x) specific 
> issue.
>
> I also checked if the setting of "pci_msitranslate=1" in the guest config had 
> any effect, but "pci_msitranslate=0" gave the same results.
>
> Complete dmesg and xl dmesg from after this commit (with MSI-X enabled), 
> starting and shutting down the guest is attached.
>
> --
> Sander

Are there any qemu logs?  The idenitified changeset changed the tools as
well as Xen when it came to this side of things.  It would be
interesting to see if Qemu noticed a difference in the results of the
hypercalls it makes.

~Andrew

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