Re: [dpdk-dev] IXGBE, IOMMU DMAR DRHD handling fault issue

["Burakov, Anatoly" <anatoly.burakov@intel.com>]

On 10-Feb-18 5:53 PM, Ravi Kerur wrote:
> 
> 
> On Sat, Feb 10, 2018 at 2:58 AM, Burakov, Anatoly 
> <anatoly.burakov@intel.com <mailto:anatoly.burakov@intel.com>> wrote:
> 
>     On 29-Jan-18 10:35 PM, Ravi Kerur wrote:
> 
>         Hi Burakov,
> 
>         When using vfio-pci on host both VF and PF interfaces works fine
>         with dpdk i.e. I don't see DMAR fault messages anymore. However,
>         when I attach a VF interface to a VM and start DPDK with
>         vfio-pci inside VM I still see DMAR fault messages on host. Both
>         host and VM are booted with 'intel-iommu=on' on GRUB. Ping from
>         VM with DPDK/vfio-pci doesn't work (I think it's expected
>         because of DMAR faults), however, when VF interface uses ixgbevf
>         driver ping works.
> 
>         Following are some details
> 
>         /*****************On VM***************/
>         dpdk-devbind -s
> 
>         Network devices using DPDK-compatible driver
>         ============================================
>         0000:00:07.0 '82599 Ethernet Controller Virtual Function'
>         drv=vfio-pci unused=ixgbevf
> 
>         Network devices using kernel driver
>         ===================================
>         0000:03:00.0 'Device 1041' if=eth0 drv=virtio-pci
>         unused=vfio-pci *Active*
>         0000:04:00.0 'Device 1041' if=eth1 drv=virtio-pci unused=vfio-pci
>         0000:05:00.0 'Device 1041' if=eth2 drv=virtio-pci unused=vfio-pci
> 
>         Other network devices
>         =====================
>         <none>
> 
>         Crypto devices using DPDK-compatible driver
>         ===========================================
>         <none>
> 
>         Crypto devices using kernel driver
>         ==================================
>         <none>
> 
>         Other crypto devices
>         ====================
>         <none>
> 
> 
>         00:07.0 Ethernet controller: Intel Corporation 82599 Ethernet
>         Controller Virtual Function (rev 01)
>                   Subsystem: Intel Corporation 82599 Ethernet Controller
>         Virtual Function
>                   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-
>                   Region 0: Memory at fda00000 (64-bit, prefetchable)
>         [size=16K]
>                   Region 3: Memory at fda04000 (64-bit, prefetchable)
>         [size=16K]
>                   Capabilities: [70] MSI-X: Enable+ Count=3 Masked-
>                           Vector table: BAR=3 offset=00000000
>                           PBA: BAR=3 offset=00002000
>                   Capabilities: [a0] Express (v1) Root Complex
>         Integrated Endpoint, MSI 00
>                           DevCap: MaxPayload 128 bytes, PhantFunc 0
>                                   ExtTag- RBE-
>                           DevCtl: Report errors: Correctable- Non-Fatal-
>         Fatal- Unsupported-
>                                   RlxdOrd- ExtTag- PhantFunc- AuxPwr-
>         NoSnoop-
>                                   MaxPayload 128 bytes, MaxReadReq 128 bytes
>                           DevSta: CorrErr- UncorrErr- FatalErr-
>         UnsuppReq- AuxPwr- TransPend-
>                   Capabilities: [100 v1] Advanced Error Reporting
>                           UESta:  DLP- SDES- TLP- FCP- CmpltTO-
>         CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
>                           UEMsk:  DLP- SDES- TLP- FCP- CmpltTO-
>         CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
>                           UESvrt: DLP- SDES- TLP- FCP- CmpltTO-
>         CmpltAbrt- UnxCmplt- RxOF- MalfTLP- ECRC- UnsupReq- ACSViol-
>                           CESta:  RxErr- BadTLP- BadDLLP- Rollover-
>         Timeout- NonFatalErr-
>                           CEMsk:  RxErr- BadTLP- BadDLLP- Rollover-
>         Timeout- NonFatalErr-
>                           AERCap: First Error Pointer: 00, GenCap-
>         CGenEn- ChkCap- ChkEn-
>                   Kernel driver in use: vfio-pci
>                   Kernel modules: ixgbevf
> 
>         /***************on Host*************/
>         dmesg | grep DMAR
>         ...
>         [  978.268143] DMAR: DRHD: handling fault status reg 2
>         [  978.268147] DMAR: [DMA Read] *Request device [04:10.0]* fault
>         addr 33a128000 [fault reason 06] PTE Read access is not set
>         [ 1286.677726] DMAR: DRHD: handling fault status reg 102
>         [ 1286.677730] DMAR: [DMA Read] Request device [04:10.0] fault
>         addr fb663000 [fault reason 06] PTE Read access is not set
>         [ 1676.436145] DMAR: DRHD: handling fault status reg 202
>         [ 1676.436149] DMAR: [DMA Read] Request device [04:10.0] fault
>         addr 33a128000 [fault reason 06] PTE Read access is not set
>         [ 1734.433649] DMAR: DRHD: handling fault status reg 302
>         [ 1734.433652] DMAR: [DMA Read] Request device [04:10.0] fault
>         addr 33a128000 [fault reason 06] PTE Read access is not set
>         [ 2324.428938] DMAR: DRHD: handling fault status reg 402
>         [ 2324.428942] DMAR: [DMA Read] Request device [04:10.0] fault
>         addr 7770c000 [fault reason 06] PTE Read access is not set
>         [ 2388.553640] DMAR: DRHD: handling fault status reg 502
>         [ 2388.553643] DMAR: [DMA Read] *Request device [04:10.0]* fault
>         addr 33a128000 [fault reason 06] PTE Read access is not set
> 
> 
> 
>     Going back to this, i would like to suggest run a few tests to
>     ensure that we have all information that we can gather.
> 
>     First of all, i'm assuming that you're using native ixgbe Linux
>     driver on the host, and that you're only passing through the VF
>     device to the VM using VFIO. Is my understanding correct here?
> 
>     Now, let's forget about the iommu=pt and igb_uio for a moment. Boot
>     both your host and your VM with iommu=on and intel_iommu=on (or
>     whatever command-line enables full IOMMU support on both host and
>     guest) and do the same tests you've done before. Do you still see
>     your issues?
> 
>     It would also be very useful to also try native Linux kernel driver
>     on the guest *with traffic forwarding* and see how it works in your
>     VM. Therefore i would suggest you to compile DPDK with PCAP support,
>     bind your (VM) interface to native Linux driver, and use the
>     interface via our pcap driver (creating a vdev should do the trick -
>     please refer to PCAP PMD documentation [1]). Simple forwarding test
>     should be enough - just make sure to pass traffic to and from DPDK
>     in both cases, and that it doesn't give you any DMAR errors.
> 
>     We can go from there.
> 
> 
> Let me just give you what has been tested and working/nonworking 
> scenarios. Some of your questions might get answered as well. Test bed 
> is very simple with 2 VF's created under IXGBE PF on host with one VF 
> interface added to ovs-bridge on host and another VF interface given to 
> guest. Test connectivity between VF's via ping.
> 
> Host and guest -- Kernel 4.9
> Host -- Qemu 2.11.50 (tried both released 2.11 and tip of the git (2.11.50))
> DPDK -- 17.05.1 on host and guest
> Host and guest -- booted with GRUB intel_iommu=on (which enables IOMMU). 
> Have tried with "iommu=on and intel_iommu=on" as well, but iommu=on is 
> not needed when intel_iommu=on is set.
> 
> Test-scenario-1: Host -- ixgbe_vf driver, Guest ixgbe_vf driver ping works
> Test-scenario-2: Host -- DPDK vfio-pci driver, Guest ixgbe_vf driver 
> ping works
> Test-scenario-3: Host -- DPDK vfio-pci driver, Guest DPDK vfio-pci 
> driver, DMAR errors seen on host, ping doesn't work

OK, that makes it clearer, thanks. Does the third scenario work in other 
DPDK versions?

> 
> DPDK works fine on host with vfio-pci, however, has issues when used 
> inside the guest. Please let me know if more information is needed.
> 
> Thanks,
> Ravi
> 
>     [1] http://dpdk.org/doc/guides/nics/pcap_ring.html
>     <http://dpdk.org/doc/guides/nics/pcap_ring.html>
> 
>     -- 
>     Thanks,
>     Anatoly
> 
> 


-- 
Thanks,
Anatoly