Home > Pci Assign Cannot > Pci-assign Cannot Read From Host Rom

Pci-assign Cannot Read From Host Rom

So we need to fix qemu to get those passed in from the privileged libvirt. Ah ok, that makes sense. LC_ALL=C PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin QEMU_AUDIO_DRV=spice /usr/bin/qemu-kvm -name RHEL6 -S -machine pc-1.2,accel=kvm,usb=off -m 3907 -realtime mlock=off -smp 1,sockets=1,cores=1,threads=1 -uuid cf9c00de-8615-9213-2e0d-5edc89e3b369 -no-user-config -nodefaults -chardev socket,id=charmonitor,path=/var/lib/libvirt/qemu/RHEL6.monitor,server,nowait -mon chardev=charmonitor,id=monitor,mode=control -rtc base=utc -no-shutdown -device piix3-usb-uhci,id=usb,bus=pci.0,addr=0x1.0x2 -device virtio-serial-pci,id=virtio-serial0,bus=pci.0,addr=0x7 -drive So I'm almost certainly wasting my time trying to get the actual responsible party to fix their broken firmware.I'll keep you posted on any issues that I run into as I

The resource files we covered via write in the > previous patch, and I'm assuming Matthew did it that way on purpose. Adding the check in open isn't trivial. Is there something I can do manually to get the answer to the question "Did config space access work"? I plan on doing that today. http://marc.info/?l=qemu-devel&m=134590292008036&w=2

If there is a security issue with the way caps checking is done on these sysfs files, then that changes the discussion somewhat. You might experience some performance hit due to that.Failed to assign irq for "hostdev1": Input/output errorPerhaps you are assigning a device that shares an IRQ with another device?qemu-kvm: -device pci-assign,host=17:04.0,id=hostdev1,configfd=25,bus=pci.0,addr=0x8: Device Mai 00:53 0000:01:00.0 -> ../../../../devices/pci0000:00/0000:00:01.0/0000:01:00.0lrwxrwxrwx 1 root root 0 17.

Comment 30 Josh Boyer 2013-04-09 15:02:24 EDT (In reply to comment #29) > I grabbed the build that Josh points to in Comment 28 and tried it out. Buy now! Device 4093 (rev 0a) Flags: medium devsel, IRQ 45 Memory at de220000 (32-bit, non-prefetchable) [size=128] I/O ports at 6000 [size=128] ScoobySnack View Public Profile Find all posts by ScoobySnack Tags error, I can help you setup libvirt for a quick and simple test so you don't have to keep waiting for me to get results) Comment 36 Kyle McMartin 2013-06-26 17:00:54 EDT

do_early_param+0x88/0x88 [ 0.402827] [] ? Failed to assign irq for "hostdev1": Input/output error Perhaps you are assigning a device that shares an IRQ with another device? The only thing is "DMAR: No ATSR found". I also re-uploaded linux-mainline based on 3.9.2 and qemu-git packages with reset patches, they're working fine on my pc.EDIT: Could you try to load your gpu bios by file and see

It > does definitely solve the problem; I can now configure qemu to run as > non-root and drop all capabilities (i.e. Comment 1 Laine Stump 2013-02-07 14:30:07 EST I have a set of patches for libvirt that manage to set CAP_COMPROMISE_KERNEL for non-root qemu-kvm, and that does indeed solve the problem. I was wondering myself too Offline #17 2013-05-17 02:45:21 nbhs Member From: Montevideo, Uruguay Registered: 2013-05-02 Posts: 402 Re: KVM VGA-Passthrough using the new vfio-vga support in kernel =>3.9 Have you Skip option ROM probe with rombar=0, or load from file with romfile= Segmentation fault (core dumped) Expected result: ---------------- the guest can boot up, and the device works fine in guest.

Could you or Daniel explain what might possibly be going wrong here, since the config space is readable by qemu? Note You need to log in before you can comment on or make changes to this bug. Device 4093 (rev 0a) Flags: medium devsel, IRQ 45 Memory at de220000 (32-bit, non-prefetchable) [size=128] I/O ports at 6000 [size=128] Top Post Reply Print view 2 posts • Page 1 of I really am coming at this from the outside.

Laine said as much in comment #1, and said that vfio will eventually be used. If I run the VM twice, the host will hang, gradually getting slower for a few seconds until it freezes, almost like it's experiencing an interrupt storm or something; eventually even If I run qemu as a non-privileged user, with no extra capabilities, then it still fails. rest_init+0x90/0x90 [ 0.402828] [] kernel_init+0xe/0x190 [ 0.402830] [] ret_from_fork+0x7c/0xb0 [ 0.402832] [] ?

That will evaluate to false on non-SB machines. You will get warning message when you assign it. If someone is still affected, please reopen. Thanks.

rpm_resume+0x12f/0x6a0[55406.929141] [] ? However, this is really more of a general capabilities issue than anything. Kyle's patch added the cap check for the /resourceX files, and the vpd file (if it exists).

Comment 21 Matthew Garrett 2013-03-28 14:45:53 EDT The resourceX files *need* to be constrained by this capability, so if there's an expectation that an unprivileged qemu can open and read/write these

Now I need to buy a USB switch for input devices; or maybe I'll just solder/glue one out of spare parts. Register All Albums FAQ Today's Posts Search Using Fedora General support for current versions. My understanding was that these kernel patches which were supposedly put in to enable Secure Boot must not (on a non-Secure Boot machine) prohibit *anything* that was previously allowed; doing otherwise It's not a > fatal error and probably just means you cannot use the PCI NIC's > network boot ROM (PXE) inside the guest.

ASM1042 SuperSpeed USB Host Controller 06:00.0 USB controller: ASMedia Technology Inc. I don't have to add > CAP_COMPROMISE_KERNEL, or anything else), and PCI passthrough works as it > did in the past. I'm moving this bug to rawhide and leaving it open until some kind of final solution presents itself though. Skip option ROM probe with rombar=0 > or load from file with romfile= > > This is a warning that the option ROM could not be loaded.

With the checks he added there, those open calls were likely failing. inode_has_perm.isra.31.constprop.61+0x2a/0x30[55406.929201] [] ? I'm not sure why that is, but it seems the file in question is accessible to qemu after it is started and /proc//status clearly does not have any capabilities set. However, when I started the \ guest
> by:
>              qemu-img -boot c -hda readhat.img -device
> pci-assign,host=XX:00.0,
>  the network of the guest failed.

vfio is our long term new architecture for PCI passthrough, which we're expecting to be compatible with SB enabled machines. Kyle moved the check to open instead of write as it was in the original patch (note: read did not have a check) after you suggested moving the check to open