tag:blogger.com,1999:blog-8694303781453133223.comments2023-12-12T08:07:13.883-07:00VFIO tips and tricksAlex Williamsonhttp://www.blogger.com/profile/02071923591707250496noreply@blogger.comBlogger370125tag:blogger.com,1999:blog-8694303781453133223.post-38350987205853455262016-10-25T13:54:03.431-06:002016-10-25T13:54:03.431-06:00Should I disable IE8 in Windows Feature? I'm u...Should I disable IE8 in Windows Feature? I'm using a different browser and never use IE.<br /><br /><a href="http://www.socialtech360.com/advanced-windows-tips-and-tricks/" rel="nofollow">Best tips and tricks for windows<br /></a><br />Anonymoushttps://www.blogger.com/profile/17729393614633723007noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-24157480383867849372016-10-15T18:26:15.183-06:002016-10-15T18:26:15.183-06:00I was looking for something that did this, so than...I was looking for something that did this, so thank you.<br /><br />However, on another note I was having issues with Q35, PCI Passthrough and AMD. My host would hang every time the VM started. Entering this configuration has solved my issues and I can only assume it has something to do with timing, the VM and Q35; based this analysis. I am very new to KVM and the deeper sides of this topic, so it would be interesting to know if timing is the reason.<br /><br />Thanks again for posting.Anonymoushttps://www.blogger.com/profile/06947498855662607128noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-46440835409165059742016-10-06T11:44:07.198-06:002016-10-06T11:44:07.198-06:00Thanks for this one. I always see a large number o...Thanks for this one. I always see a large number of qemu args in posts regarding getting vga=on to work, which seemed messy to me compared to adding the devices in virtmanager, but I hadn't worked out how to avoid defining it that way and still get the desired results.Anonymoushttps://www.blogger.com/profile/12024613530599102297noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-39235400753770871972016-10-02T10:21:50.364-06:002016-10-02T10:21:50.364-06:00Thanks for this post! I'm one of those guilty ...Thanks for this post! I'm one of those guilty spreading incorrect info.powerhousehttps://www.blogger.com/profile/10717660094370336211noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-28300671501560685522016-08-21T23:30:38.288-06:002016-08-21T23:30:38.288-06:00It appears to me that Nvidia purposely cripples th...It appears to me that Nvidia purposely cripples their cards, except for the "Multi-OS" Quadro cards they deem VT-d enabled. I have a GTX 960 card with a Xeon X5690 cpu, and I'm observing that the frame rate is consistently under 20 FPS for newer games such as Borderlands 2, while slightly older games such as Supreme Commander 2 will maintain a solid 60 FPS. The GTX 960 is normally capable of playing BL2 at max video specs, and I've verified that cpu is untaxed at levels under 20% usage and less than 15% physical ram used. Enabling MSI was the last optimization I had left to try for this. I'm going to try the Radeon HD 5850 to see how it fares, otherwise I suspect I will go for a higher end Radeon card since Nvidia seems relatively nonviable and unnecessarily difficult to optimize for game play.Ryan Kuzakishttps://www.blogger.com/profile/09139799516044342759noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-90907852875536831872016-08-12T01:16:50.420-06:002016-08-12T01:16:50.420-06:001. Enable VT-d support in BIOS
2. Add "intel_...1. Enable VT-d support in BIOS<br />2. Add "intel_iommu=on" to boot option<br />3. Reboot, check "dmesg | grep IOMMU" to see if get supported<br />4. Insert all vfio_* modules<br />dbphttps://www.blogger.com/profile/05713075470085083144noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-8607329654875568172016-08-12T01:12:29.017-06:002016-08-12T01:12:29.017-06:00If the kernel has changed too differently, the pat...If the kernel has changed too differently, the patch file could not be applied. This is what "FAILED" the messages mean.<br /><br />You need to find the similar version of patch file (if ever exist).dbphttps://www.blogger.com/profile/05713075470085083144noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-20229883675666329582016-08-11T19:44:57.841-06:002016-08-11T19:44:57.841-06:00This reset issue didn't work for me, R9 380. G...This reset issue didn't work for me, R9 380. Got a BSOD with "thread stuck in device driver"Lucretia9https://www.blogger.com/profile/12266346493096832604noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-16340922385176046522016-08-03T10:01:48.208-06:002016-08-03T10:01:48.208-06:00>Another couple other bonuses for v4.1 and newe...>Another couple other bonuses for v4.1 and newer kernels is that by binding devices statically to vfio-pci, they will be placed into a low power state when not in use. Before you get your hopes too high, this generally only saves a few watts and does not stop the fan.<br /><br />Maybe this is common knowledge by now, but using libvirt 2.0.0 on Linux 4.6 with a VFIO configured graphics card the fan may turn off completely if the graphics card has a "Zero RPM" mode. In my case I'm using EVGA GTX 970 SSC, and the RPM control seems to work as if native on Windows.Timhttps://www.blogger.com/profile/01798658893076761254noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-79540141287749267262016-08-02T10:30:34.815-06:002016-08-02T10:30:34.815-06:00If you experience problems with BSODs in Windows, ...If you experience problems with BSODs in Windows, try enabling the kvm module option ignore_msrs. The risk of this is that unknown/unsupported MSRs will return 0, which may not always be the correct return value and may lead to other issues. The Intel graphics drivers are known to need this for now though.Alex Williamsonhttps://www.blogger.com/profile/02071923591707250496noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-46196360180554769612016-07-26T14:53:28.018-06:002016-07-26T14:53:28.018-06:00If anybody interested I've found some workarou...If anybody interested I've found some workaround. It is necessary to suspend-to-ram windows guest further you may kill qemu process. After that VM restart was successful.Anonymoushttps://www.blogger.com/profile/17683389459125996788noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-20742092103728439152016-07-26T01:52:27.020-06:002016-07-26T01:52:27.020-06:0026.07.2016 is there any progress?26.07.2016 is there any progress?Anonymoushttps://www.blogger.com/profile/17683389459125996788noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-27881179972509390322016-07-24T15:01:03.162-06:002016-07-24T15:01:03.162-06:00This comment has been removed by a blog administrator.vfio-testuserhttps://www.blogger.com/profile/02669270231500657770noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-35447323600868089182016-07-24T13:22:33.837-06:002016-07-24T13:22:33.837-06:00This comment has been removed by a blog administrator.vfio-testuserhttps://www.blogger.com/profile/02669270231500657770noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-50905862855703990692016-07-21T13:46:31.438-06:002016-07-21T13:46:31.438-06:00Bingo! Thank you very much sir! I've just chan...Bingo! Thank you very much sir! I've just changed to the new SeaBIOS and it all works now. <br /><br />I've also got it working by blacklisting only i915 in /etc/modprobe.d/blacklist.conf so I don't have to execute 'virsh nodedev-detach pci_0000_00_02_0' after boot.<br /><br /><br />Stuff In Generalhttps://www.blogger.com/profile/04406966851045338936noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-54120784821224966162016-07-21T13:31:30.681-06:002016-07-21T13:31:30.681-06:00You're using /home/user/vfio-bios/bios.bin for...You're using /home/user/vfio-bios/bios.bin for your BIOS and /usr/local/qemu-2.7.0-git-20072106/bin/qemu-system-x86_64 for the emulator, but qemu.git has the correct bios in it, so why aren't you using /usr/local/qemu-2.7.0-git-20072106/share/qemu/bios.bin for the BIOS? Where does vfio-bios come from? Are there any error messages in /var/log/libvirt/qemu/arch-gns3-2016-idg.log? I don't spot any other issues, but using the correct BIOS is critical and there was churn on that during development, so anything other than the final bits that went upstream won't work.Alex Williamsonhttps://www.blogger.com/profile/02071923591707250496noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-29991210060912894452016-07-21T13:21:56.357-06:002016-07-21T13:21:56.357-06:00Here's the XML : http://pastebin.com/mLwrK4Bw
...Here's the XML : http://pastebin.com/mLwrK4Bw<br /><br />The processor is an i7-4770s (Haswell)Stuff In Generalhttps://www.blogger.com/profile/04406966851045338936noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-12128046464427709092016-07-21T12:42:55.456-06:002016-07-21T12:42:55.456-06:00Can we see your guest XML/command line? What'...Can we see your guest XML/command line? What's the processor version?Alex Williamsonhttps://www.blogger.com/profile/02071923591707250496noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-85005944275609131252016-07-21T12:38:00.596-06:002016-07-21T12:38:00.596-06:00I can see a difference now after letting i915 rab ...I can see a difference now after letting i915 rab the IGD and running 'virsh nodedev-detach pci_0000_00_02_0' as the console output to screen disappears. However the guest still fails to boot with no sign of any output on the screen:<br /><br />Here's the pastebin to the outputs after running 'virsh nodedev-detach pci_0000_00_02_0' <br /><br />http://pastebin.com/7WFFpG1TStuff In Generalhttps://www.blogger.com/profile/04406966851045338936noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-74813653816844737382016-07-20T14:26:17.800-06:002016-07-20T14:26:17.800-06:00Can you pastebin /proc/iomem, /proc/ioport, and &#...Can you pastebin /proc/iomem, /proc/ioport, and 'lspci -vvvs 00:02.0', you'll need sudo for all of these on recent kernels. dmesg may also be useful. Clearly some console driver is still attaching to IGD. You might also find that you have more success letting i915 grab IGD, unbinding and preventing future binding. Generally the problems I see with i915 are re-binding after use by vfio-pci. 'virsh nodedev-detach pci_0000_00_02_0' at boot should do this.Alex Williamsonhttps://www.blogger.com/profile/02071923591707250496noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-37599567523992927072016-07-20T14:12:55.975-06:002016-07-20T14:12:55.975-06:00Nice update Alex. I have this a quick go but could...Nice update Alex. I have this a quick go but couldn't get anywhere with it.<br /><br />I'm running Linux 4.7.0-rc7 and the very latest qemu-git. I blacklisted i915, and some other modules after troubleshooting by inserting the following in to /etc/modprobe.d/blacklist.conf:<br /><br />blacklist snd_hda_intel<br />blacklist i915<br />blacklist drm<br />blacklist drm_kms_helper<br />blacklist i2c_algo_bit<br />install snd_hda_intel /bin/false<br />install i915 /bin/false<br /><br />I also added video=vesafb:off,efifb:off to my kernel boot parameters. I can see that 00:02.0 has no kernel module in use when the system boots however I still get a console output on the screen. When I start the guest the console disappears and I see the below in dmesg but the guest never starts and I get no output on the screen:<br /><br />[ 67.980294] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=io+mem,decodes=io+mem:owns=io+mem<br />[ 68.025793] br0: port 3(tap11) entered blocking state<br />[ 68.025796] br0: port 3(tap11) entered disabled state<br />[ 68.025854] device tap11 entered promiscuous mode<br />[ 68.052476] br0: port 3(tap11) entered blocking state<br />[ 68.052479] br0: port 3(tap11) entered forwarding state<br />[ 69.158471] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=io+mem:owns=none<br />[ 69.158473] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=io+mem,decodes=io+mem:owns=io+mem<br />[ 78.675389] br0: port 3(tap11) entered disabled state<br />[ 78.676919] device tap11 left promiscuous mode<br />[ 78.676922] br0: port 3(tap11) entered disabled state<br />[ 78.912853] vgaarb: device changed decodes: PCI:0000:00:02.0,olddecodes=io+mem,decodes=io+mem:owns=io+mem<br /><br /><br /><br />Stuff In Generalhttps://www.blogger.com/profile/04406966851045338936noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-76336545128735517462016-06-24T00:04:21.244-06:002016-06-24T00:04:21.244-06:00Hi Alex. I test VFIO in a x86 computer. But I find...Hi Alex. I test VFIO in a x86 computer. But I find that there's no iommu_group under /sys/kernel/iommu_groups. I have inserted all vfio_* modules. Could you give me a hint? Did I miss some configuration when I compiled kernel image and modules?Anonymoushttps://www.blogger.com/profile/04924132150063339676noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-1757103208930764922016-06-15T14:51:03.651-06:002016-06-15T14:51:03.651-06:00put the 210 in the top slot. Running at slower PCI...put the 210 in the top slot. Running at slower PCI-e speeds is a minimal performace hit anyway. In Linus' 2 player 1 PC video he uses an old PCI-express video card, to use a pair of GPU's (980Ti and Titan) together and not sacrifice much performance (that third x16 slot is likely running at x4 or x8 tops). He never really explains it, he just goes through the setup process and what's in the PC as well as performance testing, but I imagine if you use a PCI video card you can boot the PC using PCI as the "init boot" or "init first" option, which might bind that to UnRAID, leaving the two GPUs to play with. with nVidia, you *need* another video card apparently, I did read a post of someone claiming a Quadro card can be passed easily (and thus flashing your primary card into a Quadro would solve your problems), but literally only the one post, and no evidence behind it.<br /><br />I want to do this, but I need to find a case with 8 PCI slot holes, as my third PCI-express x16 slot sits at the bottom of the case (or I could just notch the x1 slot above the first x16 slot and slap any old GPU in it, or even just use an x1 GPU...)Anonymoushttps://www.blogger.com/profile/04474631390165211898noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-51581372464096558072016-06-15T14:46:19.841-06:002016-06-15T14:46:19.841-06:00This comment has been removed by the author.Anonymoushttps://www.blogger.com/profile/04474631390165211898noreply@blogger.comtag:blogger.com,1999:blog-8694303781453133223.post-82981128954779151762016-05-20T03:21:59.553-06:002016-05-20T03:21:59.553-06:00Try changing you cpu configuration, I had an AMD p...Try changing you cpu configuration, I had an AMD proc, setup with cpu=host shows this BSOD, cpu=phenom solves it...<br />For intel proc, try using different intel arch such as core2duo.<br /><br />Maybe it's too late, but might help someone...Hell_of_BSODhttps://www.blogger.com/profile/05003131095787438062noreply@blogger.com