From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga07.intel.com (mga07.intel.com [134.134.136.100]) by dpdk.org (Postfix) with ESMTP id 179101B613 for ; Fri, 10 Nov 2017 02:40:17 +0100 (CET) Received: from fmsmga003.fm.intel.com ([10.253.24.29]) by orsmga105.jf.intel.com with ESMTP; 09 Nov 2017 17:40:16 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.44,371,1505804400"; d="scan'208";a="369135" Received: from fyigit-mobl1.ger.corp.intel.com (HELO [10.241.224.59]) ([10.241.224.59]) by FMSMGA003.fm.intel.com with ESMTP; 09 Nov 2017 17:40:15 -0800 To: Chas Williams <3chas3@gmail.com> Cc: Thomas Monjalon , dev@dpdk.org, Jianfeng Tan , Jingjing Wu , Shijith Thotton , Gregory Etelson , Harish Patil , George Prekas , Sergio Gonzalez Monroy , Rasesh Mody , Lee Roberts , Stephen Hemminger References: <20171103223822.28852-1-ferruh.yigit@intel.com> <2004961.P5XXAOnQC2@xps> <1537486.rYyyKMFLXg@xps> <3c0fb383-552b-b212-c0a0-9267a12afad9@intel.com> From: Ferruh Yigit Message-ID: Date: Thu, 9 Nov 2017 17:40:13 -0800 User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101 Thunderbird/52.4.0 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 8bit Subject: Re: [dpdk-dev] [dpdk-stable] [PATCH v2] igb_uio: prevent reset for a list of devices X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 10 Nov 2017 01:40:18 -0000 On 11/8/2017 4:00 AM, Chas Williams wrote: > > > On Tue, Nov 7, 2017 at 5:26 PM, Ferruh Yigit > wrote: > > On 11/7/2017 12:47 PM, Chas Williams wrote: > > I will confess I haven't looked into the issue too hard since I have a > > workaround.  My first guess is that there is something going on with the IOMMU > > and quiescing a PCI pass-through device/function from the guest (since I don't > > think the IOMMU is "visible" to the guest) seems iffy. > > > > Most devices have some sort of reset to put the device into a known state for > > setup/configuration (or enable/disable for the DMA engines).  If this is done at > > .dev_close(), shouldn't that be as sufficient as resetting the function? > > This is for the cases DPDK app terminated unexpectedly, proper exit path already > does cleanup. > > > Call a usermode helper from igb_uio that does an open/close on the device about > to be released? Can a generic userspace code know how to cleaup various devices? I guess driver required for this work and dpdk application that has drivers already exit in that stage. >   > > > > > > On Tue, Nov 7, 2017 at 1:49 PM, Ferruh Yigit > > >> wrote: > > > >     On 11/7/2017 10:12 AM, Chas Williams wrote: > >     > Environment: Dell PowerEdge R730, Intel Corporation 82599ES 10-Gigabit > >     SFI/SFP+ > >     > Network Connection shared via PCI pass-through > >     > Host: Debian 8 > >     > Guest: Custom Debian 8 with DPDK application based on 17.11 > >     > > >     > When we shutdown the guest, the kernel panics with: > >     > > >     > [  279.021818] Do you have a strange power saving mode enabled? > >     > [  279.021819] Dazed and confused, but trying to continue > >     > [  279.021847] {1}[Hardware Error]: Hardware error from APEI Generic > Hardware > >     > Error Source: 3 > >     > [  279.021849] {1}[Hardware Error]: event severity: fatal > >     > [  279.021850] {1}[Hardware Error]:  Error 0, type: fatal > >     > [  279.021851] {1}[Hardware Error]:   section_type: PCIe error > >     > [  279.021852] {1}[Hardware Error]:   port_type: 0, PCIe end point > >     > [  279.021853] {1}[Hardware Error]:   version: 1.16 > >     > [  279.021854] {1}[Hardware Error]:   command: 0x0507, status: 0x4010 > >     > [  279.021855] {1}[Hardware Error]:   device_id: 0000:03:00.0 > >     > [  279.021855] {1}[Hardware Error]:   slot: 0 > >     > [  279.021856] {1}[Hardware Error]:   secondary_bus: 0x00 > >     > [  279.021857] {1}[Hardware Error]:   vendor_id: 0x8086, device_id: > 0x10fb > >     > [  279.021858] {1}[Hardware Error]:   class_code: 000002 > >     > [  279.021859] Kernel panic - not syncing: Fatal hardware error! > >     > [  279.021977] sched: Unexpected reschedule of offline CPU#1! > >     > [  279.021984] ------------[ cut here ]------------ > >     > [  279.021992] WARNING: CPU: 43 PID: 2807 at > >     > /build/linux-fHlJSJ/linux-4.12.6/arch/x86/kernel/smp.c:128 > >     > native_smp_send_reschedule+0x34/0x40 > >     > [  279.021993] Modules linked in: vfio_pci vfio_virqfd > vfio_iommu_type1 vfio > >     > openvswitch nf_conntrack_ipv6 nf_nat_ipv6 nf_conntrack_ipv4 > nf_defrag_ipv4 > >     > nf_nat_ipv4 nf_defrag_ipv6 nf_nat nf_conntrack libcrc32c > crc32c_generic nfsd > >     > nfs_aclr > >     > pcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver nfs lockd grace sunrpc > >     fscache tun > >     > intel_rapl sb_edac x86_pkg_temp_thermal intel_powerclamp coretemp > >     kvm_intel kvm > >     > irqbypass mgag200 ttm drm_kms_helper drm joydev crct10dif_pclmul > crc32_pclmu > >     > l ghash_clmulni_intel i2c_algo_bit ipmi_si ipmi_devintf iTCO_wdt > intel_cstate > >     > iTCO_vendor_support evdev intel_uncore mxm_wmi lpc_ich ipmi_msghandler > >     mfd_core > >     > ioatdma intel_rapl_perf dcdbas pcspkr shpchp mei_me button wmi mei > >     acpi_power_m > >     > eter tpm_crb autofs4 ext4 crc16 jbd2 fscrypto mbcache sr_mod cdrom sg > >     > hid_generic usbhid hid sd_mod > >     > [  279.022044]  crc32c_intel aesni_intel aes_x86_64 crypto_simd cryptd > >     > glue_helper ahci ehci_pci libahci ehci_hcd ixgbe libata megaraid_sas > >     usbcore dca > >     > i40e usb_common ptp pps_core scsi_mod mdio > >     > [  279.022060] CPU: 43 PID: 2807 Comm: revalidator85 Not tainted > >     4.12.0-1-amd64 > >     > #1 Debian 4.12.6-1 > >     > [  279.022061] Hardware name: Dell Inc. PowerEdge R730/0WCJNT, BIOS > 2.3.4 > >     11/08/2016 > >     > [  279.022062] task: ffff91d0473f7100 task.stack: ffffafef8f4a4000 > >     > [  279.022066] RIP: 0010:native_smp_send_reschedule+0x34/0x40 > >     > [  279.022067] RSP: 0018:ffffafef8f4a7c98 EFLAGS: 00010082 > >     > [  279.022069] RAX: 000000000000002e RBX: ffff91d059d24080 RCX: > >     0000000000000001 > >     > [  279.022070] RDX: 0000000000000000 RSI: 0000000000000002 RDI: > >     0000000000000046 > >     > [  279.022071] RBP: ffff91d04691d100 R08: 0000000000000000 R09: > >     000000000000002e > >     > [  279.022072] R10: ffffafef8f4a7c90 R11: 00000000001cbb78 R12: > >     ffff91d85d21ae80 > >     > [  279.022073] R13: ffff91d059d24000 R14: 0000000000000002 R15: > >     0000000000000008 > >     > [  279.022075] FS:  00007f726affd700(0000) GS:ffff91d85d740000(0000) > >     > knlGS:0000000000000000 > >     > [  279.022076] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > >     > [  279.022077] CR2: 00007fd422a52c48 CR3: 000000042d90f000 CR4: > >     00000000003426e0 > >     > [  279.022078] DR0: 0000000000000000 DR1: 0000000000000000 DR2: > >     0000000000000000 > >     > [  279.022079] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: > >     0000000000000400 > >     > [  279.022080] Call Trace: > >     > [  279.022086]  ? check_preempt_wakeup+0x181/0x220 > >     > [  279.022091]  ? check_preempt_curr+0x74/0x80 > >     > [  279.022094]  ? ttwu_do_wakeup+0x19/0x140 > >     > [  279.022098]  ? try_to_wake_up+0x1b8/0x470 > >     > [  279.022101]  ? wake_up_q+0x3f/0x70 > >     > [  279.022106]  ? futex_wake+0x15a/0x170 > >     > [  279.022108]  ? do_futex+0x2df/0xa90 > >     > [  279.022111]  ? SyS_futex+0x7a/0x170 > >     > [  279.022113]  ? SyS_read+0x76/0xc0 > >     > [  279.022118]  ? system_call_fast_compare_end+0xc/0x97 > >     > [  279.022119] Code: a3 05 51 fb cc 00 73 15 48 8b 05 28 74 a3 00 be > fd 00 > >     00 00 > >     > 48 8b 80 a0 00 00 00 ff e0 89 fe 48 c7 c7 88 5c de b6 e8 e2 c9 13 00 > <0f> > >     ff c3 > >     > 66 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00 8b 05 5d 00 > >     > [  279.022151] ---[ end trace eddc980dc8648163 ]--- > >     > [  279.454274] Kernel Offset: 0x35400000 from 0xffffffff81000000 > (relocation > >     > range: 0xffffffff80000000-0xffffffffbfffffff) > >     > > >     > The test engineer says this doesn't happen if we use SRIOV (which > makes sense > >     > since the device isn't directly shared between the guest and the > host).  If I > >     > remove the pci_reset_function() from igb_uio's .release, then all is > well. > > > >     This was tougher than expected, so many unexpected behavior. Why resetting > >     pass-through device in guest cause a crash in the host? > > > >     Finally, I will send a patch to remove the reset. Hopefully no more > surprises > >     for release. > > > >     Still there will remain two improvement in igb_uio for better security, > >     disabling device interrupt on exit and clear master on exit. > > > >     > > >     > > >     > On Tue, Nov 7, 2017 at 8:02 AM, Thomas Monjalon > > >     > > >>> wrote: > >     > > >     >     07/11/2017 12:50, Chas Williams: > >     >     > We still have an issue with this and PCI pass-through.  If a > guest is > >     >     > restarted while using PCI pass-through and igb_uio issues a > >     >     > pci_reset_function(), this causes the host to crash. > >     > > >     >     Please, could you better explain the exact scenario and the cause of > >     the crash? > >     >     Thanks > >     > > >     > > > > > > >