Date: Fri, 10 Jan 2020 06:25:23 +0000 From: bugzilla-noreply@freebsd.org To: virtualization@FreeBSD.org Subject: [Bug 241774] FreeBSD 11.3 & 12.0 has broken SCSI & Networking on KVM/QEMU Q35 with OVMF Message-ID: <bug-241774-27103-tCAWyF7kej@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-241774-27103@https.bugs.freebsd.org/bugzilla/> References: <bug-241774-27103@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=241774 --- Comment #22 from John Hartley <drum@graphica.com.au> --- (In reply to Vincenzo Maffione from comment #21) Hi Vincenzo, thank you very much of for responding. My testing has been to try to isolate where the problems lies. The issues appears when you go from 11.2 -> 11.3 . As the problem relates to device detection / initialisation, my initial thought was that problem might be in PCI/PCIe code. To test this I tried to create a custom 11.3 kernel with 11.2 PCI subsystem. The impact of regression to 11.2 PCI was that I also had to regress netmap, network and driver code back to 11.2 as well. The resulting 11.3 (with 11.2 pci / networking) kernel no longer had the issue (ie e1000, vmxnet3 & rtl8139 all working ok) So I then rolled forward the pci code back to 11.3 . Again testing worked with each of e1000, vmxnet3 & rtl8139 working. So this seemed to have issue as being in netmap or drivers ... In looking at record of code updates, the netmap related changes (refactoring etc) and adding support for netmap in e1000 & vmxnet3 drivers all related to MFC r339639 (which I believe was big upstream netmap alignment and refactoring). So right now I have not confirmed that the bug is in netmap, but it might be... On a related front this VirtIO bug: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=236922 required changes to both VirtIO driver and netmap. Right now I am a bit stuck in finding the defect, as netmap and driver code is more complex and requires deeper understand than I have. My objective was just try to pin point where the problem was. Having looked at this I now have a much greater appreciation of what the dev/netmap code does and why it impacts such a large portion of network code. On my working 11.3 / 11.2 custom kernel here is dmesg result from em0: em0: <Intel(R) PRO/1000 Legacy Network Connection 1.1.0> port 0x9000-0x903f mem 0xc8600000-0xc861ffff irq 23 at device 1.0 on pci2 em0: Ethernet address: 52:54:00:4e:50:91 em0: netmap queues/slots: TX 1/256, RX 1/256 So it appears that netmap is enabled by default. The only configuration I have done is set DHCP on: # cat /etc/rc.conf clear_tmp_enable="YES" ... ifconfig_em0="DHCP" local_unbound_enable="YES" sshd_enable="YES" # Set dumpdev to "AUTO" to enable crash dumps, "NO" to disable dumpdev="AUTO" Right now I have not managed to get problem isolated so specific code item... Your advise on helping to get to the bottom of problem would be appreciated. Cheers, John Hartley -- You are receiving this mail because: You are the assignee for the bug.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-241774-27103-tCAWyF7kej>