Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 29/07/2015 02:47, Andy Lutomirski wrote: If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. Are there many virtio physical devices out there ? We are talking about a virtio flag right ? Or have you been considering something else ? Yes, virtio flag. I dislike having a virtio flag at all, but so far no one has come up with any better ideas. If there was a reliable, cross-platform mechanism for per-device PCI bus properties, I'd be all for using that instead. No, a virtio flag doesn't make sense. Blindly using system memory is a bug in QEMU; it has to be fixed to use the right address space, and then whatever the system provides to describe the right address space can be used (like the DMAR table on x86). On PPC I suppose you could use the host bridge's device tree? If you need a hook, you can add a bool virtio_should_bypass_iommu(void) { /* lookup something in the device tree?!? */ } EXPORT_SYMBOL_GPL(virtio_should_bypass_iommu); in some pseries.c file, and in the driver: static bool virtio_bypass_iommu(void) { bool (*fn)(void); fn = symbol_get(virtio_should_bypass_iommu); return fn fn(); } Awful, but that's what this thing is. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-29 01:21, Benjamin Herrenschmidt wrote: On Tue, 2015-07-28 at 15:43 -0700, Andy Lutomirski wrote: New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. Advertisement of that flag must be configurable, or we won't be able to run older guests anymore which don't know it, thus will reject it. The only precondition: there must be no IOMMU if we turn it off. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-29 10:17, Paolo Bonzini wrote: On 29/07/2015 02:47, Andy Lutomirski wrote: If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. Are there many virtio physical devices out there ? We are talking about a virtio flag right ? Or have you been considering something else ? Yes, virtio flag. I dislike having a virtio flag at all, but so far no one has come up with any better ideas. If there was a reliable, cross-platform mechanism for per-device PCI bus properties, I'd be all for using that instead. No, a virtio flag doesn't make sense. That will create the risk of subtly breaking old guests over new setups. I wouldn't suggest this. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, 2015-07-29 at 10:17 +0200, Paolo Bonzini wrote: On 29/07/2015 02:47, Andy Lutomirski wrote: If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. Are there many virtio physical devices out there ? We are talking about a virtio flag right ? Or have you been considering something else ? Yes, virtio flag. I dislike having a virtio flag at all, but so far no one has come up with any better ideas. If there was a reliable, cross-platform mechanism for per-device PCI bus properties, I'd be all for using that instead. No, a virtio flag doesn't make sense. It wouldn't if we were creating virtio from scratch. However we have to be realistic here, we are contending with existing practices and implementation. The fact is qemu *does* bypass any iommu and has been doing so for a long time, *and* the guest drivers are written today *also* bypassing all DMA mapping mechanisms and just passing everything accross. So if it's a bug, it's a bug on both sides of the fence. We are no longer in bug fixing territory here, it's a fundamental change of ABI. The ABI might not be what was intended (but that's arguable, see below), but it is that way. Arguably it was even known and considered a *feature* by some (including myself) at the time. It somewhat improved performances on archs where otherwise every page would have to be mapped/unmapped in guest iommu. In fact, it also makes vhost a lot easier. So I disagree, it's de-facto a feature (even if unintended) of the existing virtio implementations and changing that would be a major interface change, and thus should be exposed as such. Blindly using system memory is a bug in QEMU; it has to be fixed to use the right address space, and then whatever the system provides to describe the right address space can be used (like the DMAR table on x86). Except that it's not so easy. For example, on PPC PAPR guests, there is no such thing as a no IOMMU space, the concept doesn't exist. So we have at least three things to deal with: - Existing guests, so we must preserve the existing behaviour for backward compatibility. - vhost is made more complex because it now needs to be informed of the guest iommu updates - New guests have the new driver that knows how to map and unmap would have a performance loss unless some mechanism to create a no iommu space exists which for us would need to be added. Either that or we rely on DDW which is a way for a guest to create a permanent mapping of its entire address space in an IOMMU but that incur a significant waste of host kernel memory. On PPC I suppose you could use the host bridge's device tree? If you need a hook, you can add a No because we can mix and match virtio and other devices on the same host bridge. Unless we put a property that only applies to virtio children of the host bridge. bool virtio_should_bypass_iommu(void) { /* lookup something in the device tree?!? */ } EXPORT_SYMBOL_GPL(virtio_should_bypass_iommu); in some pseries.c file, and in the driver: static bool virtio_bypass_iommu(void) { bool (*fn)(void); fn = symbol_get(virtio_should_bypass_iommu); return fn fn(); } Awful, but that's what this thing is. Ben. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Am 28.07.2015 um 03:08 schrieb Andy Lutomirski: On Mon, Sep 1, 2014 at 10:39 AM, Andy Lutomirski l...@amacapital.net wrote: This fixes virtio on Xen guests as well as on any other platform that uses virtio_pci on which physical addresses don't match bus addresses. This can be tested with: virtme-run --xen xen --kimg arch/x86/boot/bzImage --console using virtme from here: https://git.kernel.org/cgit/utils/kernel/virtme/virtme.git Without these patches, the guest hangs forever. With these patches, everything works. Dusting off an ancient thread. Now that the dust has accumulated^Wsettled, is it worth pursuing this? I think the situation is considerably worse than it was when I originally wrote these patches: I think that QEMU now supports a nasty mode in which the guest's PCI bus appears to be behind an IOMMU but the virtio devices on that bus punch straight through that IOMMU. I have a half-hearted port to modern kernels here: https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/log/?h=virtio_ring_xen I didn't implement DMA API access for virtio_pci_modern, and I have no idea what to do about detecting whether a given virtio device honors its IOMMU or not. I think its really tricky. Looking at where virtio came from, the virtio ring was always native access without IOMMU. This was true for the early lguest things and then the early s390 transport, (which is quite close to the lguest interface). virtio-pci used the same scheme - ignoring all iommu considerations. I understand that for PCI we actually might want to follow iommu restrictions from a correctness and security point of view and from the ccw point of view we do not. No idea about virtio-mmio. I think the proper way of handling this is to take this into the TC for virtio - dont know what would be the right thing to do. A feature bit, always iommu for pci, something else? Michael, Conny, do you agree? Christian ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 28/07/2015 03:08, Andy Lutomirski wrote: On Mon, Sep 1, 2014 at 10:39 AM, Andy Lutomirski l...@amacapital.net wrote: This fixes virtio on Xen guests as well as on any other platform that uses virtio_pci on which physical addresses don't match bus addresses. This can be tested with: virtme-run --xen xen --kimg arch/x86/boot/bzImage --console using virtme from here: https://git.kernel.org/cgit/utils/kernel/virtme/virtme.git Without these patches, the guest hangs forever. With these patches, everything works. Dusting off an ancient thread. Now that the dust has accumulated^Wsettled, is it worth pursuing this? I think the situation is considerably worse than it was when I originally wrote these patches: I think that QEMU now supports a nasty mode in which the guest's PCI bus appears to be behind an IOMMU but the virtio devices on that bus punch straight through that IOMMU. That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. Paolo I have a half-hearted port to modern kernels here: https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/log/?h=virtio_ring_xen I didn't implement DMA API access for virtio_pci_modern, and I have no idea what to do about detecting whether a given virtio device honors its IOMMU or not. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 4:21 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Tue, 2015-07-28 at 15:43 -0700, Andy Lutomirski wrote: Let me try to summarize a proposal: Add a feature flag that indicates IOMMU support. New kernels acknowledge that flag on any device that advertises it. New kernels always respect the IOMMU (except on PowerPC). Why ? I disagree, the flag should be honored when set in any architecture. PowerPC is no different than any other platform in that regard. Perhaps I should have said instead someone more familiar with PPC than I am should figure out what PPC should do. For the non-PPC case, there is only one instance that I know of in which ignoring the IOMMU is beneficial, and that case is the experimental Q35 thing. If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. New kernels optionally refuse to talk to devices that don't have that feature flag if the device appears to be behind an IOMMU. (This presumably includes any device whatsoever on an x86 platform with an IOMMU, including Xen's fake IOMMU.) New QEMU always respects the IOMMU, if any, except on PPC. This is just a matter of what is the default of the flag, ie we should have a machine flag that indicates what the default is for new virtio devices, otherwise, it should be specified per device as an attribute of the device instance. On x86, I think that even super-peformance-critical virtio devices should always honor the iommu, but that the iommu in question should be a 1:1 iommu. I *think* that x86 supports that. IOW x86 would always set the feature flag. I would argue that we should default to bypass IOMMU on *all* architecture due to the performance impact, and to essentially default to the same behaviour as today. With things like DDW even powerpc might be able to mostly alleviate the performance impact so we might to change in the long term, but I tend to prefer more incremental approaches. As above, there's a difference between bypass IOMMU and there is no IOMMU. x86 and, I think, most other platforms are capable of the latter. I'm not sure PPC is. I think that, in an ideal world, there would be no feature flag and all virtio devices would always respect the IOMMU. Unfortunately we have existing practice in the form of PPC and Q35 iommu=on that conflict with that. New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. Would any non-PPC user ever configure it differently? I suppose if you want to support old kernels on new QEMU, you'd flip the switch. On PPC, new QEMU will not respect the IOMMU and will not set the flag. New kernels will not talk to devices that set the flag. If someone wants to fix that, then they get to figure out how. I disagree with the kernel bit and I disagree with special casing PPC in any shape or form in the code. The only difference should be a default value for the iommu mode of virtio in qemu set per machine. You can then feel free to change that default (in a separate patch for bisectability) on x86 for the sake of Xen. I think we should flip the default everywhere to respects IOMMU. That's the setting that will work in all cases on new guest + new host, and it's the setting that's safest. vfio will probably always malfunction if given a device that looks like it's behind an IOMMU but doesn't respect it. For people who need the last bit of performance, they should use bus-level controls where available (they should be available everywhere except PPC and maybe arm64) and, ideally, someone would teach PPC how to exclude devices from the IOMMU cleanly if possible. If that can't be done, then there can be an option to bypass the IOMMU the way it's currently done and no one except PPC would do it. PPC really is different from everything except x86 Q35 iommu=on, and the latter is experimental. AFAIK in all other cases, the IOMMU is respected by virtio, but there is no non-1:1 IOMMU. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2015-07-28 at 17:47 -0700, Andy Lutomirski wrote: Yes, virtio flag. I dislike having a virtio flag at all, but so far no one has come up with any better ideas. If there was a reliable, cross-platform mechanism for per-device PCI bus properties, I'd be all for using that instead. There isn't that I know of, so I think it's the best approach we have. .../... - The kernel should just honor what qemu says, ie, whether the qemu device honors or bypasses the iommu. Except for vfio, which maybe just needs a special case: vfio checks if the device claims to be virtio and doesn't set the flag, in which case vfio just refuses to bind the device. Right but passing virtio through isn't the highest priority on the radar, but yes, indeed, it should identify them and reject them. - Qemu default behaviour should be set via a machine attribute which can be overriden both globally (the machine one) or per-device. I think that, in an ideal world, there would be no feature flag and all virtio devices would always respect the IOMMU. Unfortunately we have existing practice in the form of PPC and Q35 iommu=on that conflict with that. And possibly more as in this is how the qemu virtio devices are written today, they do not use the proper DMA accessors, they always bypass, whatever the platform is (so sparc would be in the same boat for example). Except that AFAIK Q35 is the only QEMU platform that supports a nontrivial IOMMU in the first place. Are there pseries hosts that have a working IOMMU? Maybe I've just misunderstood. You may well be correct, I remember that we actually created the iommu infrastructure to a large extent in qemu for ppc/pseries, then it got extended when q35 came in. New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. Would any non-PPC user ever configure it differently? I suppose if you want to support old kernels on new QEMU, you'd flip the switch. Possibly, have we looked at what ia64, sparc, arm, ... do ? At least sparc has iommus as well. I think (I hope!) that ia64 is irrelevant, and last I checked ARM didn't have a QEMU-emulated IOMMU. Maybe things have changed. Not yet... .../... On new machine types, we shouldn't change the behaviour of an existing machine type, and we should keep the default to 0 on ppc/pseries because of backward compatibility issue. But that should be the only place that is ppc specific, ie, a default value in a machine def structure. Fair enough, except I still think we should change the default to be respect IOMMU on machine types that don't have an IOMMU in the first place. Ok, but do it in a separate patch because it *is* a behaviour change to some extent. That way Xen works with old machine types, and I don't think we lose anything. That's the setting that will work in all cases on new guest + new host, and it's the setting that's safest. vfio will probably always malfunction if given a device that looks like it's behind an IOMMU but doesn't respect it. For people who need the last bit of performance, they should use bus-level controls where available (they should be available everywhere except PPC and maybe arm64) and, ideally, someone would teach PPC how to exclude devices from the IOMMU cleanly if possible. If that can't be done, then there can be an option to bypass the IOMMU the way it's currently done and no one except PPC would do it. PPC really is different from everything except x86 Q35 iommu=on, and the latter is experimental. AFAIK in all other cases, the IOMMU is respected by virtio, but there is no non-1:1 IOMMU. What about sparc ? I though it was pretty similar to PPC in that regard... No clue, honestly. I could be wrong about the set of existing QEMU machine types. Ok. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2015-07-28 at 15:43 -0700, Andy Lutomirski wrote: Let me try to summarize a proposal: Add a feature flag that indicates IOMMU support. New kernels acknowledge that flag on any device that advertises it. New kernels always respect the IOMMU (except on PowerPC). Why ? I disagree, the flag should be honored when set in any architecture. PowerPC is no different than any other platform in that regard. New kernels optionally refuse to talk to devices that don't have that feature flag if the device appears to be behind an IOMMU. (This presumably includes any device whatsoever on an x86 platform with an IOMMU, including Xen's fake IOMMU.) New QEMU always respects the IOMMU, if any, except on PPC. This is just a matter of what is the default of the flag, ie we should have a machine flag that indicates what the default is for new virtio devices, otherwise, it should be specified per device as an attribute of the device instance. I would argue that we should default to bypass IOMMU on *all* architecture due to the performance impact, and to essentially default to the same behaviour as today. With things like DDW even powerpc might be able to mostly alleviate the performance impact so we might to change in the long term, but I tend to prefer more incremental approaches. New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. On PPC, new QEMU will not respect the IOMMU and will not set the flag. New kernels will not talk to devices that set the flag. If someone wants to fix that, then they get to figure out how. I disagree with the kernel bit and I disagree with special casing PPC in any shape or form in the code. The only difference should be a default value for the iommu mode of virtio in qemu set per machine. You can then feel free to change that default (in a separate patch for bisectability) on x86 for the sake of Xen. Ben. This results in: New kernels work fine with old QEMU unless iommu=on. New kernels work with new devices (QEMU and physical devices that set the flag) under all circumstances, except on PPC where physical devices are and remain broken. Xen works work new QEMU and cleanly refuses to interoperate with old QEMU. (This is worse than with just my patches, but it's better than the status quo in which the Xen guest corrupts itself and possibly corrupts the Xen hypervisor.) New kernels with old QEMU with iommu=on optionally refuses to interoperate. Old kernels are oblivious. They work exactly the same as they do today except that they fail cleanly with new QEMU with iommu=on. Old kernels continue to fail with physical virtio devices if they're behind an iommu. Old physical virtio devices that don't advertise the flag fail cleanly if the host uses an iommu. The driver could optionally whitelist such devices. PPC works as well as it currently does. I'm unsure about the arm64 situation. Did I get this right? --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2015-07-28 at 16:33 -0700, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 4:21 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Tue, 2015-07-28 at 15:43 -0700, Andy Lutomirski wrote: Let me try to summarize a proposal: Add a feature flag that indicates IOMMU support. New kernels acknowledge that flag on any device that advertises it. New kernels always respect the IOMMU (except on PowerPC). Why ? I disagree, the flag should be honored when set in any architecture. PowerPC is no different than any other platform in that regard. Perhaps I should have said instead someone more familiar with PPC than I am should figure out what PPC should do. For the non-PPC case, there is only one instance that I know of in which ignoring the IOMMU is beneficial, and that case is the experimental Q35 thing. ppc is many fairly different platforms, some with iommu, some without, some benefiting from bypass, some less etc... I think ARM will soon be in a similar basket. If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. Are there many virtio physical devices out there ? We are talking about a virtio flag right ? Or have you been considering something else ? New kernels optionally refuse to talk to devices that don't have that feature flag if the device appears to be behind an IOMMU. (This presumably includes any device whatsoever on an x86 platform with an IOMMU, including Xen's fake IOMMU.) New QEMU always respects the IOMMU, if any, except on PPC. This is just a matter of what is the default of the flag, ie we should have a machine flag that indicates what the default is for new virtio devices, otherwise, it should be specified per device as an attribute of the device instance. On x86, I think that even super-peformance-critical virtio devices should always honor the iommu, but that the iommu in question should be a 1:1 iommu. I *think* that x86 supports that. IOW x86 would always set the feature flag. Ok. I would argue that we should default to bypass IOMMU on *all* architecture due to the performance impact, and to essentially default to the same behaviour as today. With things like DDW even powerpc might be able to mostly alleviate the performance impact so we might to change in the long term, but I tend to prefer more incremental approaches. As above, there's a difference between bypass IOMMU and there is no IOMMU. x86 and, I think, most other platforms are capable of the latter. I'm not sure PPC is. Depends on the platform. pseries isn't since it's already a paravirtualized plaform, but there are other ppc platforms out there which behave differently. That's why I think: - The kernel should just honor what qemu says, ie, whether the qemu device honors or bypasses the iommu. - Qemu default behaviour should be set via a machine attribute which can be overriden both globally (the machine one) or per-device. I think that, in an ideal world, there would be no feature flag and all virtio devices would always respect the IOMMU. Unfortunately we have existing practice in the form of PPC and Q35 iommu=on that conflict with that. And possibly more as in this is how the qemu virtio devices are written today, they do not use the proper DMA accessors, they always bypass, whatever the platform is (so sparc would be in the same boat for example). New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. Would any non-PPC user ever configure it differently? I suppose if you want to support old kernels on new QEMU, you'd flip the switch. Possibly, have we looked at what ia64, sparc, arm, ... do ? At least sparc has iommus as well. Let's try to not make it an architecture issue. As I said above, we have a kernel that just reacts appropriately based on what qemu says it's doing, and what qemu does is a per-machine flag to set the default. On PPC, new QEMU will not respect the IOMMU and will not set the flag. New kernels will not talk to devices that set the flag. If someone wants to fix that, then they get to figure out how. I disagree with the kernel bit and I disagree with special casing PPC in any shape or form in the code. The only difference should be a default value for the iommu mode of virtio in qemu set per machine. You can then feel free to change that default (in a separate patch for bisectability) on x86 for the sake of Xen. I think we should flip the default everywhere to respects IOMMU. On new machine types, we shouldn't change the behaviour of an existing machine type, and we should keep the default to 0 on ppc/pseries because of backward compatibility issue. But
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Let me try to summarize a proposal: Add a feature flag that indicates IOMMU support. New kernels acknowledge that flag on any device that advertises it. New kernels always respect the IOMMU (except on PowerPC). New kernels optionally refuse to talk to devices that don't have that feature flag if the device appears to be behind an IOMMU. (This presumably includes any device whatsoever on an x86 platform with an IOMMU, including Xen's fake IOMMU.) New QEMU always respects the IOMMU, if any, except on PPC. New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. On PPC, new QEMU will not respect the IOMMU and will not set the flag. New kernels will not talk to devices that set the flag. If someone wants to fix that, then they get to figure out how. This results in: New kernels work fine with old QEMU unless iommu=on. New kernels work with new devices (QEMU and physical devices that set the flag) under all circumstances, except on PPC where physical devices are and remain broken. Xen works work new QEMU and cleanly refuses to interoperate with old QEMU. (This is worse than with just my patches, but it's better than the status quo in which the Xen guest corrupts itself and possibly corrupts the Xen hypervisor.) New kernels with old QEMU with iommu=on optionally refuses to interoperate. Old kernels are oblivious. They work exactly the same as they do today except that they fail cleanly with new QEMU with iommu=on. Old kernels continue to fail with physical virtio devices if they're behind an iommu. Old physical virtio devices that don't advertise the flag fail cleanly if the host uses an iommu. The driver could optionally whitelist such devices. PPC works as well as it currently does. I'm unsure about the arm64 situation. Did I get this right? --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 5:36 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Tue, 2015-07-28 at 16:33 -0700, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 4:21 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Tue, 2015-07-28 at 15:43 -0700, Andy Lutomirski wrote: Let me try to summarize a proposal: Add a feature flag that indicates IOMMU support. New kernels acknowledge that flag on any device that advertises it. New kernels always respect the IOMMU (except on PowerPC). Why ? I disagree, the flag should be honored when set in any architecture. PowerPC is no different than any other platform in that regard. Perhaps I should have said instead someone more familiar with PPC than I am should figure out what PPC should do. For the non-PPC case, there is only one instance that I know of in which ignoring the IOMMU is beneficial, and that case is the experimental Q35 thing. ppc is many fairly different platforms, some with iommu, some without, some benefiting from bypass, some less etc... I think ARM will soon be in a similar basket. If new kernels ignore the IOMMU for devices that don't set the flag and there are physical devices that already exist and don't set the flag, then those devices won't work reliably on most modern non-virtual platforms, PPC included. Are there many virtio physical devices out there ? We are talking about a virtio flag right ? Or have you been considering something else ? Yes, virtio flag. I dislike having a virtio flag at all, but so far no one has come up with any better ideas. If there was a reliable, cross-platform mechanism for per-device PCI bus properties, I'd be all for using that instead. New kernels optionally refuse to talk to devices that don't have that feature flag if the device appears to be behind an IOMMU. (This presumably includes any device whatsoever on an x86 platform with an IOMMU, including Xen's fake IOMMU.) New QEMU always respects the IOMMU, if any, except on PPC. This is just a matter of what is the default of the flag, ie we should have a machine flag that indicates what the default is for new virtio devices, otherwise, it should be specified per device as an attribute of the device instance. On x86, I think that even super-peformance-critical virtio devices should always honor the iommu, but that the iommu in question should be a 1:1 iommu. I *think* that x86 supports that. IOW x86 would always set the feature flag. Ok. I would argue that we should default to bypass IOMMU on *all* architecture due to the performance impact, and to essentially default to the same behaviour as today. With things like DDW even powerpc might be able to mostly alleviate the performance impact so we might to change in the long term, but I tend to prefer more incremental approaches. As above, there's a difference between bypass IOMMU and there is no IOMMU. x86 and, I think, most other platforms are capable of the latter. I'm not sure PPC is. Depends on the platform. pseries isn't since it's already a paravirtualized plaform, but there are other ppc platforms out there which behave differently. That's why I think: - The kernel should just honor what qemu says, ie, whether the qemu device honors or bypasses the iommu. Except for vfio, which maybe just needs a special case: vfio checks if the device claims to be virtio and doesn't set the flag, in which case vfio just refuses to bind the device. - Qemu default behaviour should be set via a machine attribute which can be overriden both globally (the machine one) or per-device. I think that, in an ideal world, there would be no feature flag and all virtio devices would always respect the IOMMU. Unfortunately we have existing practice in the form of PPC and Q35 iommu=on that conflict with that. And possibly more as in this is how the qemu virtio devices are written today, they do not use the proper DMA accessors, they always bypass, whatever the platform is (so sparc would be in the same boat for example). Except that AFAIK Q35 is the only QEMU platform that supports a nontrivial IOMMU in the first place. Are there pseries hosts that have a working IOMMU? Maybe I've just misunderstood. New QEMU always advertises this feature flag. If iommu=on, QEMU's virtio devices refuse to work unless the driver acknowledges the flag. This should be configurable. Would any non-PPC user ever configure it differently? I suppose if you want to support old kernels on new QEMU, you'd flip the switch. Possibly, have we looked at what ia64, sparc, arm, ... do ? At least sparc has iommus as well. I think (I hope!) that ia64 is irrelevant, and last I checked ARM didn't have a QEMU-emulated IOMMU. Maybe things have changed. Let's try to not make it an architecture issue. As I said above, we have a kernel that just reacts appropriately based on what qemu says it's doing, and
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 15:06, Michael S. Tsirkin wrote: On Tue, Jul 28, 2015 at 02:46:20PM +0200, Paolo Bonzini wrote: On 28/07/2015 12:12, Benjamin Herrenschmidt wrote: That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. Yes, no known issues with vt-d emulation for almost a year now. Error reporting could be improved, and interrupt remapping is still missing, but those are minor issues in this context. In my testing setups, I also have virtio devices in use, passed through to an L2 guest, but only in 1:1 mapping so that their broken IOMMU support causes no practical problems. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 02:46:20PM +0200, Paolo Bonzini wrote: On 28/07/2015 12:12, Benjamin Herrenschmidt wrote: That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. I would much prefer if the information as to whether it honors or not gets passed to the guest somewhat. My preference goes for passing it via the virtio config space but there were objections that it should be a bus property (which is tricky to do with PCI and doesn't properly reflect the fact that in qemu you can mix match IOMMU-honoring devices and bypassing-virtio on the same bus). Yes, for example on x86 it must be passed through the DMAR table. virtio-pci device must have a separate DRHD for them. In QEMU, you could add an under-iommu property to PCI bridges, and walk the hierarchy of bridges to build the DRHDs. Paolo -- MST ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Jul 28, 2015 6:11 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 15:06, Michael S. Tsirkin wrote: On Tue, Jul 28, 2015 at 02:46:20PM +0200, Paolo Bonzini wrote: On 28/07/2015 12:12, Benjamin Herrenschmidt wrote: That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. Yes, no known issues with vt-d emulation for almost a year now. Error reporting could be improved, and interrupt remapping is still missing, but those are minor issues in this context. In my testing setups, I also have virtio devices in use, passed through to an L2 guest, but only in 1:1 mapping so that their broken IOMMU support causes no practical problems. How are you getting 1:1 to work? Is it something that L0 QEMU can advertise to L1? If so, can we just do that unconditionally, which would make my patch work? I have no objection to 1:1 devices in general. It's only devices that the PCI code on the guest identifies as not 1:1 but that are nonetheless 1:1 that cause problems. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2015-07-28 at 10:16 +0200, Paolo Bonzini wrote: On 28/07/2015 03:08, Andy Lutomirski wrote: On Mon, Sep 1, 2014 at 10:39 AM, Andy Lutomirski l...@amacapital.net wrote: This fixes virtio on Xen guests as well as on any other platform that uses virtio_pci on which physical addresses don't match bus addresses. This can be tested with: virtme-run --xen xen --kimg arch/x86/boot/bzImage --console using virtme from here: https://git.kernel.org/cgit/utils/kernel/virtme/virtme.git Without these patches, the guest hangs forever. With these patches, everything works. Dusting off an ancient thread. Now that the dust has accumulated^Wsettled, is it worth pursuing this? I think the situation is considerably worse than it was when I originally wrote these patches: I think that QEMU now supports a nasty mode in which the guest's PCI bus appears to be behind an IOMMU but the virtio devices on that bus punch straight through that IOMMU. That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. I would much prefer if the information as to whether it honors or not gets passed to the guest somewhat. My preference goes for passing it via the virtio config space but there were objections that it should be a bus property (which is tricky to do with PCI and doesn't properly reflect the fact that in qemu you can mix match IOMMU-honoring devices and bypassing-virtio on the same bus). Ben. Paolo I have a half-hearted port to modern kernels here: https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/log/?h=virtio_ring_xen I didn't implement DMA API access for virtio_pci_modern, and I have no idea what to do about detecting whether a given virtio device honors its IOMMU or not. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 28/07/2015 12:12, Benjamin Herrenschmidt wrote: That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. I would much prefer if the information as to whether it honors or not gets passed to the guest somewhat. My preference goes for passing it via the virtio config space but there were objections that it should be a bus property (which is tricky to do with PCI and doesn't properly reflect the fact that in qemu you can mix match IOMMU-honoring devices and bypassing-virtio on the same bus). Yes, for example on x86 it must be passed through the DMAR table. virtio-pci device must have a separate DRHD for them. In QEMU, you could add an under-iommu property to PCI bridges, and walk the hierarchy of bridges to build the DRHDs. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 28/07/2015 15:11, Jan Kiszka wrote: This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. Yes, no known issues with vt-d emulation for almost a year now. Error reporting could be improved, and interrupt remapping is still missing, but those are minor issues in this context. On the other hand interrupt remapping is absolutely necessary for production use, hence my point that x86 does not promise API stability. (Any kind of stability actually didn't include crashes; those are not expected :)) The Google patches for userspace PIC and IOAPIC are proceeding well, so hopefully we can have interrupt remapping soon. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 9:44 AM, Jan Kiszka jan.kis...@siemens.com wrote: The ability to have virtio on systems with IOMMU in place makes testing much more efficient for us. Ideally, we would have it in non-identity mapping scenarios as well, e.g. to start secondary Linux instances in the test VMs, giving them their own virtio devices. And we will eventually have this need on ARM as well. Virtio needs to be backward compatible, so the change to put these devices under IOMMU control could be advertised during feature negotiations and controlled on QEMU side via a device property. Newer guest drivers would have to acknowledge that they support virtio via IOMMUs. Older ones would refuse to work, and the admin could instead spawn VMs with this feature disabled. The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 28/07/2015 18:42, Jan Kiszka wrote: On the other hand interrupt remapping is absolutely necessary for production use, hence my point that x86 does not promise API stability. Well, we currently implement the features that the Q35 used to expose. Adding interrupt remapping will require a new chipset and/or a hack switch to ignore compatibility. Isn't the VT-d register space separate from other Q35 features and backwards-compatible? You could even add it to PIIX in theory just by adding a DMAR. It's not like for example SMRAM, where the registers are in the northbridge configuration space and move around in every chipset generation. (Any kind of stability actually didn't include crashes; those are not expected :)) The Google patches for userspace PIC and IOAPIC are proceeding well, so hopefully we can have interrupt remapping soon. If the day had 48 hours... I'd love to look into this, first adding QEMU support for the new irqchip architecture. I hope I can squeeze in some time for that... Google also had an intern that was looking at it. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 28/07/2015 19:19, Jan Kiszka wrote: On 2015-07-28 19:15, Paolo Bonzini wrote: On 28/07/2015 18:42, Jan Kiszka wrote: On the other hand interrupt remapping is absolutely necessary for production use, hence my point that x86 does not promise API stability. Well, we currently implement the features that the Q35 used to expose. Adding interrupt remapping will require a new chipset and/or a hack switch to ignore compatibility. Isn't the VT-d register space separate from other Q35 features and backwards-compatible? You could even add it to PIIX in theory just by adding a DMAR. Yes, it's practically working, but it's not accurate /wrt how that hardware looked like in reality. We've done that for a long time. Real PIIX3 didn't have ACPI too, for example (and it had a USB UHCI that is optional in QEMU). Of course I'm not advocating adding the IOMMU to PIIX (assuming that would work even just practically)... but I don't think adding interrupt remapping to Q35 is a big deal. It would be optional, just in case you want to debug something without interrupt remapping, but it can be added. The Google patches for userspace PIC and IOAPIC are proceeding well, so hopefully we can have interrupt remapping soon. If the day had 48 hours... I'd love to look into this, first adding QEMU support for the new irqchip architecture. I hope I can squeeze in some time for that... Google also had an intern that was looking at it. Great! In theory it's easy with the latest series. All you need is support for converting IOAPIC routes to KVM routes (and of course the glue code to enable the capability and create the userspace devices); everything else should work just by reusing the -machine kernel_irqchip=on code. In theory... Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 19:15, Paolo Bonzini wrote: On 28/07/2015 18:42, Jan Kiszka wrote: On the other hand interrupt remapping is absolutely necessary for production use, hence my point that x86 does not promise API stability. Well, we currently implement the features that the Q35 used to expose. Adding interrupt remapping will require a new chipset and/or a hack switch to ignore compatibility. Isn't the VT-d register space separate from other Q35 features and backwards-compatible? You could even add it to PIIX in theory just by adding a DMAR. Yes, it's practically working, but it's not accurate /wrt how that hardware looked like in reality. It's not like for example SMRAM, where the registers are in the northbridge configuration space and move around in every chipset generation. (Any kind of stability actually didn't include crashes; those are not expected :)) The Google patches for userspace PIC and IOAPIC are proceeding well, so hopefully we can have interrupt remapping soon. If the day had 48 hours... I'd love to look into this, first adding QEMU support for the new irqchip architecture. I hope I can squeeze in some time for that... Google also had an intern that was looking at it. Great! Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 18:36, Paolo Bonzini wrote: On 28/07/2015 15:11, Jan Kiszka wrote: This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. Yes, no known issues with vt-d emulation for almost a year now. Error reporting could be improved, and interrupt remapping is still missing, but those are minor issues in this context. On the other hand interrupt remapping is absolutely necessary for production use, hence my point that x86 does not promise API stability. Well, we currently implement the features that the Q35 used to expose. Adding interrupt remapping will require a new chipset and/or a hack switch to ignore compatibility. (Any kind of stability actually didn't include crashes; those are not expected :)) The Google patches for userspace PIC and IOAPIC are proceeding well, so hopefully we can have interrupt remapping soon. If the day had 48 hours... I'd love to look into this, first adding QEMU support for the new irqchip architecture. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 19:10, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 9:44 AM, Jan Kiszka jan.kis...@siemens.com wrote: The ability to have virtio on systems with IOMMU in place makes testing much more efficient for us. Ideally, we would have it in non-identity mapping scenarios as well, e.g. to start secondary Linux instances in the test VMs, giving them their own virtio devices. And we will eventually have this need on ARM as well. Virtio needs to be backward compatible, so the change to put these devices under IOMMU control could be advertised during feature negotiations and controlled on QEMU side via a device property. Newer guest drivers would have to acknowledge that they support virtio via IOMMUs. Older ones would refuse to work, and the admin could instead spawn VMs with this feature disabled. The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 18:11, Andy Lutomirski wrote: On Jul 28, 2015 6:11 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 15:06, Michael S. Tsirkin wrote: On Tue, Jul 28, 2015 at 02:46:20PM +0200, Paolo Bonzini wrote: On 28/07/2015 12:12, Benjamin Herrenschmidt wrote: That is an experimental feature (it's x-iommu), so it can change. The plan was: - for PPC, virtio never honors IOMMU - for non-PPC, either have virtio always honor IOMMU, or enforce that virtio is not under IOMMU. I dislike having PPC special cased. In fact, today x86 guests also assume that virtio bypasses IOMMU I believe. In fact *all* guests do. This doesn't matter much, since the only guests that implement an IOMMU in QEMU are (afaik) PPC and x86, and x86 does not yet promise any kind of stability. Hmm I think Jan (cc) said it was already used out there. Yes, no known issues with vt-d emulation for almost a year now. Error reporting could be improved, and interrupt remapping is still missing, but those are minor issues in this context. In my testing setups, I also have virtio devices in use, passed through to an L2 guest, but only in 1:1 mapping so that their broken IOMMU support causes no practical problems. How are you getting 1:1 to work? Is it something that L0 QEMU can advertise to L1? If so, can we just do that unconditionally, which would make my patch work? The guest hypervisor is Jailhouse and the guest is the root cell that loaded the hypervisor, thus continues with identity mappings. You usually don't have 1:1 mapping with other setups - maybe with some Xen configuration? Dunno. I have no objection to 1:1 devices in general. It's only devices that the PCI code on the guest identifies as not 1:1 but that are nonetheless 1:1 that cause problems. The ability to have virtio on systems with IOMMU in place makes testing much more efficient for us. Ideally, we would have it in non-identity mapping scenarios as well, e.g. to start secondary Linux instances in the test VMs, giving them their own virtio devices. And we will eventually have this need on ARM as well. Virtio needs to be backward compatible, so the change to put these devices under IOMMU control could be advertised during feature negotiations and controlled on QEMU side via a device property. Newer guest drivers would have to acknowledge that they support virtio via IOMMUs. Older ones would refuse to work, and the admin could instead spawn VMs with this feature disabled. Jan -- Siemens AG, Corporate Technology, CT RTC ITP SES-DE Corporate Competence Center Embedded Linux ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 20:22, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 10:17 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 19:10, Andy Lutomirski wrote: The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Except that, with my patches, it would work correctly. ISTM the thing I haven't looked at your patches yet - they make the virtio PCI driver in Linux IOMMU-compatible? Perfect - except for a compatibility check, right? that's broken right now is QEMU and the virtio_pci driver. My patches fix the driver. Last year that would have been the end of the story except for PPC. Now we have to deal with QEMU. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. I still don't see how feature flags solve the problem. Suppose we added a feature flag meaning respects IOMMU. Bad case 1: Build a malicious device that advertises non-IOMMU-respecting virtio. Plug it in behind an IOMMU. Host starts leaking physical addresses to the device (and the device doesn't work, of course). Maybe that's only barely a security problem, but still... I don't see right now how critical such a hypothetical case could be. But the OS / its drivers could still decide to refuse talking to such a device. Bad case 2: Use current QEMU w/ IOMMU enabled. Assign a virtio device provided by L0 QEMU to L2. L1 crashes. I consider *that* to be a security problem, although in practice no one will configure their system that way because it has zero chance of actually working. Nonetheless, the device does work if L1 accesses it directly? The issue is vfio doesn't notice that the device doesn't respect the IOMMU because respects-IOMMU is a property of the PCI bus and the platform IOMMU, and vfio assumes it works correctly. I would have no problem with rejecting configurations in future QEMU that try to expose unconfined virtio devices in the presence of IOMMU emulation. Once we can do better, it's just about letting the guest know about the difference. The current situation is indeed just broken, we don't need to discuss this as we can't change history to prevent this. Bad case 2: Some hypothetical well-behaved new QEMU provides a virtio device that *does* respect the IOMMU and sets the feature flag. They emulate Q35 with an IOMMU. They boot Linux 4.1. Data corruption in the guest. No. In that case, the feature negotiation of virtio-with-iommu-support would have failed for older drivers, and the device would have never been used by the guest. We could make the rule that *all* virtio-pci devices (except on PPC) respect the bus rules. We'd have to fix QEMU so that virtio devices on Q35 iommu=on systems set up a PCI topology where the devices *aren't* behind the IOMMU or are protected by RMRRs or whatever. Then old kernels would work correctly on new hosts, new kernels would work correctly except on old iommu-providing hosts, and Xen would work. I don't see a point in doing anything about old QEMU with IOMMU enabled and virtio devices plugged except declaring such setups broken. No one should have configured this for production purposes, only for test setups (like we, with the knowledge about the limitations). In fact, on Xen, it's impossible without colossal hacks to support non-IOMMU-respecting virtio devices because Xen acts as an intermediate IOMMU between the Linux dom0 guest and the actual host. The QEMU host doesn't even know that Xen is involved. This is why Xen and virtio don't currently work together (without my patches): the device thinks it doesn't respect the IOMMU, the driver thinks the device doesn't respect the IOMMU, and they're both wrong. TL;DR: I think there are only two cases. Either a device respects the IOMMU or a device doesn't know whether it respects the IOMMU. The latter case is problematic. See above, the latter is only problematic on setups that actually use an IOMMU. If that includes Xen, then no one should use it until virtio can declare itself IOMMU
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 10:17 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 19:10, Andy Lutomirski wrote: The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Except that, with my patches, it would work correctly. ISTM the thing that's broken right now is QEMU and the virtio_pci driver. My patches fix the driver. Last year that would have been the end of the story except for PPC. Now we have to deal with QEMU. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. I still don't see how feature flags solve the problem. Suppose we added a feature flag meaning respects IOMMU. Bad case 1: Build a malicious device that advertises non-IOMMU-respecting virtio. Plug it in behind an IOMMU. Host starts leaking physical addresses to the device (and the device doesn't work, of course). Maybe that's only barely a security problem, but still... Bad case 2: Use current QEMU w/ IOMMU enabled. Assign a virtio device provided by L0 QEMU to L2. L1 crashes. I consider *that* to be a security problem, although in practice no one will configure their system that way because it has zero chance of actually working. Nonetheless, the device does work if L1 accesses it directly? The issue is vfio doesn't notice that the device doesn't respect the IOMMU because respects-IOMMU is a property of the PCI bus and the platform IOMMU, and vfio assumes it works correctly. Bad case 2: Some hypothetical well-behaved new QEMU provides a virtio device that *does* respect the IOMMU and sets the feature flag. They emulate Q35 with an IOMMU. They boot Linux 4.1. Data corruption in the guest. We could make the rule that *all* virtio-pci devices (except on PPC) respect the bus rules. We'd have to fix QEMU so that virtio devices on Q35 iommu=on systems set up a PCI topology where the devices *aren't* behind the IOMMU or are protected by RMRRs or whatever. Then old kernels would work correctly on new hosts, new kernels would work correctly except on old iommu-providing hosts, and Xen would work. In fact, on Xen, it's impossible without colossal hacks to support non-IOMMU-respecting virtio devices because Xen acts as an intermediate IOMMU between the Linux dom0 guest and the actual host. The QEMU host doesn't even know that Xen is involved. This is why Xen and virtio don't currently work together (without my patches): the device thinks it doesn't respect the IOMMU, the driver thinks the device doesn't respect the IOMMU, and they're both wrong. TL;DR: I think there are only two cases. Either a device respects the IOMMU or a device doesn't know whether it respects the IOMMU. The latter case is problematic. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 12:06 PM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 20:22, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 10:17 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 19:10, Andy Lutomirski wrote: The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Except that, with my patches, it would work correctly. ISTM the thing I haven't looked at your patches yet - they make the virtio PCI driver in Linux IOMMU-compatible? Perfect - except for a compatibility check, right? Yes. (virtio_pci_legacy, anyway. Presumably virtio_pci_modern is easy to adapt, too.) that's broken right now is QEMU and the virtio_pci driver. My patches fix the driver. Last year that would have been the end of the story except for PPC. Now we have to deal with QEMU. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. I still don't see how feature flags solve the problem. Suppose we added a feature flag meaning respects IOMMU. Bad case 1: Build a malicious device that advertises non-IOMMU-respecting virtio. Plug it in behind an IOMMU. Host starts leaking physical addresses to the device (and the device doesn't work, of course). Maybe that's only barely a security problem, but still... I don't see right now how critical such a hypothetical case could be. But the OS / its drivers could still decide to refuse talking to such a device. How does OS know it's such a device as opposed to a QEMU-supplied thing? Bad case 2: Some hypothetical well-behaved new QEMU provides a virtio device that *does* respect the IOMMU and sets the feature flag. They emulate Q35 with an IOMMU. They boot Linux 4.1. Data corruption in the guest. No. In that case, the feature negotiation of virtio-with-iommu-support would have failed for older drivers, and the device would have never been used by the guest. So are you suggesting that newer virtio devices always provide this feature flag and, if supplied by QEMU with iommu=on, simply refuse to operate of the driver doesn't support that flag? That could work as long as QEMU with the current (broken?) iommu=on never exposes such a device. We could make the rule that *all* virtio-pci devices (except on PPC) respect the bus rules. We'd have to fix QEMU so that virtio devices on Q35 iommu=on systems set up a PCI topology where the devices *aren't* behind the IOMMU or are protected by RMRRs or whatever. Then old kernels would work correctly on new hosts, new kernels would work correctly except on old iommu-providing hosts, and Xen would work. I don't see a point in doing anything about old QEMU with IOMMU enabled and virtio devices plugged except declaring such setups broken. No one should have configured this for production purposes, only for test setups (like we, with the knowledge about the limitations). I'm fine with that. In fact, I proposed these patches before QEMU had this feature in the first place. In fact, on Xen, it's impossible without colossal hacks to support non-IOMMU-respecting virtio devices because Xen acts as an intermediate IOMMU between the Linux dom0 guest and the actual host. The QEMU host doesn't even know that Xen is involved. This is why Xen and virtio don't currently work together (without my patches): the device thinks it doesn't respect the IOMMU, the driver thinks the device doesn't respect the IOMMU, and they're both wrong. TL;DR: I think there are only two cases. Either a device respects the IOMMU or a device doesn't know whether it respects the IOMMU. The latter case is problematic. See above, the latter is only problematic on setups that actually use an IOMMU. If that includes Xen, then no one should use it until virtio can declare itself IOMMU compatible, and drivers exist that process this. Xen works right now with my patches on standard QEMU (as long as iommu=off). Certainly no one except me uses it now with virtio because it doesn't work with mainline kernels. If we apply something
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 2015-07-28 21:24, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 12:06 PM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 20:22, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 10:17 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 19:10, Andy Lutomirski wrote: The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Except that, with my patches, it would work correctly. ISTM the thing I haven't looked at your patches yet - they make the virtio PCI driver in Linux IOMMU-compatible? Perfect - except for a compatibility check, right? Yes. (virtio_pci_legacy, anyway. Presumably virtio_pci_modern is easy to adapt, too.) that's broken right now is QEMU and the virtio_pci driver. My patches fix the driver. Last year that would have been the end of the story except for PPC. Now we have to deal with QEMU. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. I still don't see how feature flags solve the problem. Suppose we added a feature flag meaning respects IOMMU. Bad case 1: Build a malicious device that advertises non-IOMMU-respecting virtio. Plug it in behind an IOMMU. Host starts leaking physical addresses to the device (and the device doesn't work, of course). Maybe that's only barely a security problem, but still... I don't see right now how critical such a hypothetical case could be. But the OS / its drivers could still decide to refuse talking to such a device. How does OS know it's such a device as opposed to a QEMU-supplied thing? It can restrict itself to virtio devices exposing the feature if it feels uncomfortable that it might be talking to some evil piece of silicon (instead of the hypervisor, which has to be trusted anyway). Bad case 2: Some hypothetical well-behaved new QEMU provides a virtio device that *does* respect the IOMMU and sets the feature flag. They emulate Q35 with an IOMMU. They boot Linux 4.1. Data corruption in the guest. No. In that case, the feature negotiation of virtio-with-iommu-support would have failed for older drivers, and the device would have never been used by the guest. So are you suggesting that newer virtio devices always provide this feature flag and, if supplied by QEMU with iommu=on, simply refuse to operate of the driver doesn't support that flag? Exactly. That could work as long as QEMU with the current (broken?) iommu=on never exposes such a device. QEMU would have to be adjusted first so that all its virtio-pci device models take IOMMUs into account - if they exist or not. Only then it could expose the feature and expect the guest to acknowledge it. For compat reasons, QEMU should still be able to expose virtio devices without the flag set - but then without any IOMMU emulation enabled as well. That would prevent the current setup we are using today, but it's trivial to update the guest kernel to a newer virtio driver which would restore our scenario again. We could make the rule that *all* virtio-pci devices (except on PPC) respect the bus rules. We'd have to fix QEMU so that virtio devices on Q35 iommu=on systems set up a PCI topology where the devices *aren't* behind the IOMMU or are protected by RMRRs or whatever. Then old kernels would work correctly on new hosts, new kernels would work correctly except on old iommu-providing hosts, and Xen would work. I don't see a point in doing anything about old QEMU with IOMMU enabled and virtio devices plugged except declaring such setups broken. No one should have configured this for production purposes, only for test setups (like we, with the knowledge about the limitations). I'm fine with that. In fact, I proposed these patches before QEMU had this feature in the first place. In fact, on Xen, it's impossible without colossal hacks to support non-IOMMU-respecting virtio devices because Xen acts as an intermediate IOMMU between the Linux dom0 guest and the actual host. The QEMU host doesn't even know that Xen is
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Jul 28, 2015 at 12:33 PM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 21:24, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 12:06 PM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 20:22, Andy Lutomirski wrote: On Tue, Jul 28, 2015 at 10:17 AM, Jan Kiszka jan.kis...@siemens.com wrote: On 2015-07-28 19:10, Andy Lutomirski wrote: The trouble is that this is really a property of the bus and not of the device. If you build a virtio device that physically plugs into a PCIe slot, the device has no concept of an IOMMU in the first place. If one would build a real virtio device today, it would be broken because every IOMMU would start to translate its requests. Already from that POV, we really need to introduce a feature flag I will be IOMMU-translated so that a potential physical implementation can carry it unconditionally. Except that, with my patches, it would work correctly. ISTM the thing I haven't looked at your patches yet - they make the virtio PCI driver in Linux IOMMU-compatible? Perfect - except for a compatibility check, right? Yes. (virtio_pci_legacy, anyway. Presumably virtio_pci_modern is easy to adapt, too.) that's broken right now is QEMU and the virtio_pci driver. My patches fix the driver. Last year that would have been the end of the story except for PPC. Now we have to deal with QEMU. Similarly, if you take an L0-provided IOMMU-supporting device and pass it through to L2 using current QEMU on L1 (with Q35 emulation and iommu enabled), then, from L2's perspective, the device is 1:1 no matter what the device thinks. IOW, I think the original design was wrong and now we have to deal with it. I think the best solution would be to teach QEMU to fix its ACPI tables so that 1:1 virtio devices are actually exposed as 1:1. Only the current drivers are broken. And we can easily tell them apart from newer ones via feature flags. Sorry, don't get the problem. I still don't see how feature flags solve the problem. Suppose we added a feature flag meaning respects IOMMU. Bad case 1: Build a malicious device that advertises non-IOMMU-respecting virtio. Plug it in behind an IOMMU. Host starts leaking physical addresses to the device (and the device doesn't work, of course). Maybe that's only barely a security problem, but still... I don't see right now how critical such a hypothetical case could be. But the OS / its drivers could still decide to refuse talking to such a device. How does OS know it's such a device as opposed to a QEMU-supplied thing? It can restrict itself to virtio devices exposing the feature if it feels uncomfortable that it might be talking to some evil piece of silicon (instead of the hypervisor, which has to be trusted anyway). Bad case 2: Some hypothetical well-behaved new QEMU provides a virtio device that *does* respect the IOMMU and sets the feature flag. They emulate Q35 with an IOMMU. They boot Linux 4.1. Data corruption in the guest. No. In that case, the feature negotiation of virtio-with-iommu-support would have failed for older drivers, and the device would have never been used by the guest. So are you suggesting that newer virtio devices always provide this feature flag and, if supplied by QEMU with iommu=on, simply refuse to operate of the driver doesn't support that flag? Exactly. That could work as long as QEMU with the current (broken?) iommu=on never exposes such a device. QEMU would have to be adjusted first so that all its virtio-pci device models take IOMMUs into account - if they exist or not. Only then it could expose the feature and expect the guest to acknowledge it. For compat reasons, QEMU should still be able to expose virtio devices without the flag set - but then without any IOMMU emulation enabled as well. That would prevent the current setup we are using today, but it's trivial to update the guest kernel to a newer virtio driver which would restore our scenario again. Seems reasonable. If we apply something similar enough to my patches, then even old hypervisors (e.g. Amazon's hardware virt systems) will support Xen with virtio devices passed in just fine. Then it seems we can make everyone happy - perfect. :) Yay. FWIW, I have no intention to touch the QEMU code for this. I'm willing to do the vring bit and the virtio-pci bit as long as it's well specified. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Mon, Sep 1, 2014 at 10:39 AM, Andy Lutomirski l...@amacapital.net wrote: This fixes virtio on Xen guests as well as on any other platform that uses virtio_pci on which physical addresses don't match bus addresses. This can be tested with: virtme-run --xen xen --kimg arch/x86/boot/bzImage --console using virtme from here: https://git.kernel.org/cgit/utils/kernel/virtme/virtme.git Without these patches, the guest hangs forever. With these patches, everything works. Dusting off an ancient thread. Now that the dust has accumulated^Wsettled, is it worth pursuing this? I think the situation is considerably worse than it was when I originally wrote these patches: I think that QEMU now supports a nasty mode in which the guest's PCI bus appears to be behind an IOMMU but the virtio devices on that bus punch straight through that IOMMU. I have a half-hearted port to modern kernels here: https://git.kernel.org/cgit/linux/kernel/git/luto/linux.git/log/?h=virtio_ring_xen I didn't implement DMA API access for virtio_pci_modern, and I have no idea what to do about detecting whether a given virtio device honors its IOMMU or not. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Fri, Sep 05, 2014 at 12:01:33PM +0930, Rusty Russell wrote: Andy Lutomirski l...@amacapital.net writes: On Sep 2, 2014 11:53 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Xen does expose dma_ops. The trick is knowing when to use it. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. The devices I care about aren't actually Xen devices. They're devices supplied by QEMU/KVM, booting a Xen hypervisor, which in turn passes the virtio device (along with every other PCI device) through to dom0. So this is exactly the same virtio device that regular x86 KVM guests would see. The reason that current code fails is that Xen guest physical addresses aren't the same as the addresses seen by the outer hypervisor. These devices don't know that physical addresses != bus addresses, so they can't advertise that fact. Ah, I see. Then we will need a Xen-specific hack. Grr. This is mostly a result of the fact that virtio_pci devices aren't really PCI devices. I still think that virtio_pci shouldn't have to worry about this; ideally this would all be handled higher up in the device hierarchy. x86 already gets this right. Yes. Adding a feature to say I am a real PCI device is possible, but has other issues (particularly as Michael Tsirkin pointed out, what do you do if the driver doesn't understand the feature). Are there any hypervisors except PPC that use virtio_pci, have IOMMUs on the pci slot that virtio_pci lives in, and that use physical addressing? If not, I think that just quirking PPC will work (at least until someone wants IOMMU support in virtio_pci on PPC, in which case doing something using devicetree seems like a reasonable solution). We can either patch to make PPC weird or make Xen weird. I'm on the fence. Two questions for Paulo: 1) When QEMU support IOMMU on x86, will the virtio devices behind it respect the IOMMU (do they use the right memory access primitives?). 2) Are we really going to be able to exclude virtio devices from using the x86 IOMMU in a portable way which will always work? If it's per-bus granularity, will qemu really put them on their own PCI bus and get this right? Or will it sometimes get it wrong and users will end up using virtio devices via IOMMU by accident? If the answers are both yes, then x86 is going to be able to use virtio+IOMMU, so PPC looks like the odd one out. Otherwise it looks like we're really going to want to stick with the ignore IOMMU rule until (handwave future), and we make an exception for Xen. Cheers, Rusty. In theory, it's yes to both questions. In practice, with patches merged recently it's no to both questions :). It's a work in progress, but some extra effort to support miltiple PCI roots will be needed on the QEMU side. What problems will surface when we try to do multiple roots? Only time will tell. If it's felt that it's much cleaner to make PPC the odd one out, we can defer enabling iommu in qemu on x86 until ways to bypass it are implemented. But I would be inclined, for pre-1.0 drivers, to make Xen weird. For 1.0 drivers, we have a bit of time to consider this, and maybe PPC guys can come up with some way (can be PV) to tell guest these devices bypass the IOMMU. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 09/04/2014 10:57 PM, Andy Lutomirski wrote: On Thu, Sep 4, 2014 at 7:31 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: On Sep 2, 2014 11:53 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Xen does expose dma_ops. The trick is knowing when to use it. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. The devices I care about aren't actually Xen devices. They're devices supplied by QEMU/KVM, booting a Xen hypervisor, which in turn passes the virtio device (along with every other PCI device) through to dom0. So this is exactly the same virtio device that regular x86 KVM guests would see. The reason that current code fails is that Xen guest physical addresses aren't the same as the addresses seen by the outer hypervisor. These devices don't know that physical addresses != bus addresses, so they can't advertise that fact. Ah, I see. Then we will need a Xen-specific hack. Grr. This is mostly a result of the fact that virtio_pci devices aren't really PCI devices. I still think that virtio_pci shouldn't have to worry about this; ideally this would all be handled higher up in the device hierarchy. x86 already gets this right. Yes. Adding a feature to say I am a real PCI device is possible, but has other issues (particularly as Michael Tsirkin pointed out, what do you do if the driver doesn't understand the feature). Are there any hypervisors except PPC that use virtio_pci, have IOMMUs on the pci slot that virtio_pci lives in, and that use physical addressing? If not, I think that just quirking PPC will work (at least until someone wants IOMMU support in virtio_pci on PPC, in which case doing something using devicetree seems like a reasonable solution). We can either patch to make PPC weird or make Xen weird. I'm on the fence. Two questions for Paulo: 1) When QEMU support IOMMU on x86, will the virtio devices behind it respect the IOMMU (do they use the right memory access primitives?). 2) Are we really going to be able to exclude virtio devices from using the x86 IOMMU in a portable way which will always work? If it's per-bus granularity, will qemu really put them on their own PCI bus and get this right? Or will it sometimes get it wrong and users will end up using virtio devices via IOMMU by accident? If the answers are both yes, then x86 is going to be able to use virtio+IOMMU, so PPC looks like the odd one out. Otherwise it looks like we're really going to want to stick with the ignore IOMMU rule until (handwave future), and we make an exception for Xen. There's a third option: try to make virtio-mmio work everywhere (except s390), at least in the long run. This other benefits: it makes minimal hypervisors simpler, I think it'll get rid of the limits on the number of virtio devices in a system. ARM is already going this direction, and I imagine that PPC support would be straightforward (it's already using devicetree). In my opinion, a uniform virt machine for every instruction set would be very beneficial. I would guess that MMIO is more universally available than PCI, and as you point out, simpler to implement. Does virtio-mmio have any reasonable way of doing hotplug? It could also eventually make sense to have a standard for virtio on virtio. I don't think so, but it seems possible. My bystander understanding is that QEMU allocates some fixed number of VirtIO-MMIO devices, maybe a dozen, in the device tree. The ones that don't
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 10, 2014 at 8:36 AM, Christopher Covington c...@codeaurora.org wrote: On 09/04/2014 10:57 PM, Andy Lutomirski wrote: There's a third option: try to make virtio-mmio work everywhere (except s390), at least in the long run. This other benefits: it makes minimal hypervisors simpler, I think it'll get rid of the limits on the number of virtio devices in a system. ARM is already going this direction, and I imagine that PPC support would be straightforward (it's already using devicetree). In my opinion, a uniform virt machine for every instruction set would be very beneficial. I would guess that MMIO is more universally available than PCI, and as you point out, simpler to implement. Except for x86 :( That's presumably fixable, though. Does virtio-mmio have any reasonable way of doing hotplug? It could also eventually make sense to have a standard for virtio on virtio. I don't think so, but it seems possible. My bystander understanding is that QEMU allocates some fixed number of VirtIO-MMIO devices, maybe a dozen, in the device tree. The ones that don't actually get hooked up to something real like a block device or network interface are populated with a dummy device. One naive approach might be to allow the dummy devices to tell the kernel that they are now changing to a real device. My thought (which I completely failed to articulate) was to have a spec for a virtio device that exposes a complete virtio bus along with hotplug and per-cpu interrupts (a la MSI-X). This might be a bit complicated, but it would work everywhere without any firmware or platform issues. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On 05/09/14 04:57, Andy Lutomirski wrote: There's a third option: try to make virtio-mmio work everywhere (except s390), at least in the long run. This other benefits: it makes minimal hypervisors simpler, I think it'll get rid of the limits on the number of virtio devices in a system. ARM is already going this direction, and I imagine that PPC support would be straightforward (it's already using devicetree). Well this chance is gone. When virtio was first introduced we though about abstraction (mmio,hypercalls, pci ops depending on the platform as part of the transport. There was even a virtio over serial line as potential implementation), but we had to do a fully PCI variant to please windows guests IIRC. Christian ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2014-09-02 at 16:11 -0700, Andy Lutomirski wrote: I don't think so. I would argue that it's a straight-up bug for QEMU to expose a physically-addressed virtio-pci device to the guest behind an emulated IOMMU. QEMU may already be doing that on ppc64, but it isn't on x86_64 or arm (yet). Last I looked, it does on everything, it bypasses the DMA layer in qemu which is where IOMMUs are implemented. On x86_64, I'm pretty sure that QEMU can emulate an IOMMU for everything except the virtio-pci devices. The ACPI DMAR stuff is quite expressive. Well, *except* virtio, exactly... On ARM, I hope the QEMU will never implement a PCI IOMMU. As far as I could tell when I looked last week, none of the newer QEMU-emulated ARM machines even support PCI. Even if QEMU were to implement a PCI IOMMU on some future ARM machine, it could continue using virtio-mmio for virtio devices. Possibly... So ppc might actually be the only system that has or will have physically-addressed virtio PCI devices that are behind an IOMMU. Can this be handled in a ppc64-specific way? I wouldn't be so certain, as I said, the way virtio is implemented in qemu bypass the DMA layer which is where IOMMUs sit. The fact that currently x86 doesn't put an IOMMU there is not even garanteed, is it ? What happens if you try to mix and match virtio and other emulated devices that require the iommu on the same bus ? If we could discriminate virtio devices to a specific host bridge and guarantee no mix match, we could probably add a concept of IOMMU-less bus but that would require guest changes which limits the usefulness. Is there any way that the kernel can distinguish a QEMU-provided virtio PCI device from a physical PCIe thing? Not with existing guests which cannot be changed. Existing distros are out with those drivers. If we add a backward compatibility mechanism, then we could add something yes, provided we can segregate virtio onto a dedicated host bridge (which can be a problem with the libvirt trainwreck...) It would be kind of nice to address this without adding complexity to the virtio spec. Maybe virtio 1.0 devices could be assumed to use bus addressing unless a new devicetree property says otherwise. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, 2014-09-03 at 09:47 +0200, Paolo Bonzini wrote: IOMMU support for x86 is going to go in this week. But won't that break virtio on x86 ? Or will virtio continue bypassing it ? IE, the guest side virtio doesn't expect an IOMMU and doesn't call the dma mappings ops. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Well, it's only for virtio and should be on by default on x86 as well if an iommu is installed no ? Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2014-09-02 at 16:42 -0700, Andy Lutomirski wrote: But there aren't any ACPI systems with both virtio-pci and IOMMUs, right? So we could say that, henceforth, ACPI systems must declare whether virtio-pci devices live behind IOMMUs without breaking backward compatibility. I don't know for sure whether that's the case and whether we can rely on that not happening, we'll need x86 folks opinion here. On ARM, I hope the QEMU will never implement a PCI IOMMU. As far as I could tell when I looked last week, none of the newer QEMU-emulated ARM machines even support PCI. Even if QEMU were to implement a PCI IOMMU on some future ARM machine, it could continue using virtio-mmio for virtio devices. Possibly... So ppc might actually be the only system that has or will have physically-addressed virtio PCI devices that are behind an IOMMU. Can this be handled in a ppc64-specific way? I wouldn't be so certain, as I said, the way virtio is implemented in qemu bypass the DMA layer which is where IOMMUs sit. The fact that currently x86 doesn't put an IOMMU there is not even garanteed, is it ? What happens if you try to mix and match virtio and other emulated devices that require the iommu on the same bus ? AFAIK QEMU doesn't support IOMMUs at all on x86, so current versions of QEMU really do guarantee that virtio-pci on x86 has no IOMMU, even if that guarantee is purely accidental. Right. If we could discriminate virtio devices to a specific host bridge and guarantee no mix match, we could probably add a concept of IOMMU-less bus but that would require guest changes which limits the usefulness. Is there any way that the kernel can distinguish a QEMU-provided virtio PCI device from a physical PCIe thing? Not with existing guests which cannot be changed. Existing distros are out with those drivers. If we add a backward compatibility mechanism, then we could add something yes, provided we can segregate virtio onto a dedicated host bridge (which can be a problem with the libvirt trainwreck...) Ugh. So here's an ugly proposal: Step 1: Make virtio-pci use the DMA API only on x86. This will at least fix Xen and people experimenting with virtio hardware on x86, and it won't break anything, since there are no emulated IOMMUs on x86. I think we should make all virtio drivers use the DMA API and just have different set of dma_ops. We can make a simple ifdef powerpc if needed in virtio-pci that force the dma-ops of the device to some direct bypass ops at init time. That way no need to select whether to use the DMA API or not, just always use it, and add a tweak to replace the DMA ops with the direct ones on the archs/platforms that need that. That was my original proposal and I still think it's the best approach. Step 2: Update the virtio spec. Virtio 1.0 PCI devices should set a new bit if they are physically addressed. If that bit is clear, then the device is assumed to be addressed in accordance with the platform's standard addressing model for PCI. Presumably this would be something like VIRTIO_F_BUS_ADDRESSING = 33, and the spec would say something like Physical devices compatible with this specification MUST offer VIRTIO_F_BUS_ADDRESSING. Drivers MUST implement this feature. Alternatively, this could live in a PCI configuration capability. I'll let you sort that out with Rusty but it makes sense. Step 3: Update virtio-pci to use the DMA API for all devices on x86 and for devices that advertise bus addressing on other architectures. I think this proposal will work, but I also think it sucks and I'd really like to see a better counter-proposal. As I said, make it always use the DMA API, but add a quirk to replace the dma_ops with some NULL ops on platforms that need it. The only issue with that is the location of the dma ops is arch specific, so that one function will contain some ifdefs, but the rest of the code can just use the DMA API. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2014-09-02 at 17:32 -0700, Andy Lutomirski wrote: I agree *except* that implementing it will be a real PITA and (I think) can't be done without changing code in arch/. My patches plus an ifdef powerpc will be functionally equivalent, just uglier. So for powerpc, it's a 2 liner inside virtio-pci, but yes, it might be more of a problem for s390, I'm not too sure what they do in that area. Bigger quirk: on a standard s390 virtio guest configuration, dma_map_single etc will fail to link. Yuck I tried this in v1 of these patches. So we can poke at the archdata all day, but we can't build a kernel like that :( I would like the s390 people to chime in here, it still looks like the best way to go if they can fix things on their side :-) So until the dma_ops pointer move into struct device and CONFIG_HAS_DMA becomes mandatory (or mandatory enough that virtio can depend on it), I don't think we can do it this way. I see, it's a bummer because it would be a lot cleaner. I'll send a v5 that is the same as v4 except with physical addressing hardcoded in for powerpc. Thanks. That will do for now, but ideally we want to make it a function of some flag from the implementation, so let's see what Rusty has to say. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Andy Lutomirski l...@amacapital.net writes: On Sep 2, 2014 11:53 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Xen does expose dma_ops. The trick is knowing when to use it. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. The devices I care about aren't actually Xen devices. They're devices supplied by QEMU/KVM, booting a Xen hypervisor, which in turn passes the virtio device (along with every other PCI device) through to dom0. So this is exactly the same virtio device that regular x86 KVM guests would see. The reason that current code fails is that Xen guest physical addresses aren't the same as the addresses seen by the outer hypervisor. These devices don't know that physical addresses != bus addresses, so they can't advertise that fact. Ah, I see. Then we will need a Xen-specific hack. Grr. This is mostly a result of the fact that virtio_pci devices aren't really PCI devices. I still think that virtio_pci shouldn't have to worry about this; ideally this would all be handled higher up in the device hierarchy. x86 already gets this right. Yes. Adding a feature to say I am a real PCI device is possible, but has other issues (particularly as Michael Tsirkin pointed out, what do you do if the driver doesn't understand the feature). Are there any hypervisors except PPC that use virtio_pci, have IOMMUs on the pci slot that virtio_pci lives in, and that use physical addressing? If not, I think that just quirking PPC will work (at least until someone wants IOMMU support in virtio_pci on PPC, in which case doing something using devicetree seems like a reasonable solution). We can either patch to make PPC weird or make Xen weird. I'm on the fence. Two questions for Paulo: 1) When QEMU support IOMMU on x86, will the virtio devices behind it respect the IOMMU (do they use the right memory access primitives?). 2) Are we really going to be able to exclude virtio devices from using the x86 IOMMU in a portable way which will always work? If it's per-bus granularity, will qemu really put them on their own PCI bus and get this right? Or will it sometimes get it wrong and users will end up using virtio devices via IOMMU by accident? If the answers are both yes, then x86 is going to be able to use virtio+IOMMU, so PPC looks like the odd one out. Otherwise it looks like we're really going to want to stick with the ignore IOMMU rule until (handwave future), and we make an exception for Xen. Cheers, Rusty. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Michael S. Tsirkin m...@redhat.com writes: On Wed, Sep 03, 2014 at 04:12:01PM +0930, Rusty Russell wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Why? What's wrong with rings in memory? AFAICT, the card would have to access guest memory to read it, using multiple DMA cycles. That's going to be slow. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Well virtio could probe for xen, it's not a lot of code. We could, but I think this is going to be a more general problem in future. x86 is heading down the IOMMU path, and they're likely to suffer similarly. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. Thoughts? Rusty. OK and it should then be active even if guest does not ack the feature (so in fact, it would have to be a mandatory feature). That can work, but I still find this a bit inelegant: this is a property of the platform, not of the device. True. If a device needs it though, we're no worse of having a device which doesn't work if the driver understand the feature than we were before. Cheers, Rusty. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Thu, Sep 4, 2014 at 7:31 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: On Sep 2, 2014 11:53 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Xen does expose dma_ops. The trick is knowing when to use it. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. The devices I care about aren't actually Xen devices. They're devices supplied by QEMU/KVM, booting a Xen hypervisor, which in turn passes the virtio device (along with every other PCI device) through to dom0. So this is exactly the same virtio device that regular x86 KVM guests would see. The reason that current code fails is that Xen guest physical addresses aren't the same as the addresses seen by the outer hypervisor. These devices don't know that physical addresses != bus addresses, so they can't advertise that fact. Ah, I see. Then we will need a Xen-specific hack. Grr. This is mostly a result of the fact that virtio_pci devices aren't really PCI devices. I still think that virtio_pci shouldn't have to worry about this; ideally this would all be handled higher up in the device hierarchy. x86 already gets this right. Yes. Adding a feature to say I am a real PCI device is possible, but has other issues (particularly as Michael Tsirkin pointed out, what do you do if the driver doesn't understand the feature). Are there any hypervisors except PPC that use virtio_pci, have IOMMUs on the pci slot that virtio_pci lives in, and that use physical addressing? If not, I think that just quirking PPC will work (at least until someone wants IOMMU support in virtio_pci on PPC, in which case doing something using devicetree seems like a reasonable solution). We can either patch to make PPC weird or make Xen weird. I'm on the fence. Two questions for Paulo: 1) When QEMU support IOMMU on x86, will the virtio devices behind it respect the IOMMU (do they use the right memory access primitives?). 2) Are we really going to be able to exclude virtio devices from using the x86 IOMMU in a portable way which will always work? If it's per-bus granularity, will qemu really put them on their own PCI bus and get this right? Or will it sometimes get it wrong and users will end up using virtio devices via IOMMU by accident? If the answers are both yes, then x86 is going to be able to use virtio+IOMMU, so PPC looks like the odd one out. Otherwise it looks like we're really going to want to stick with the ignore IOMMU rule until (handwave future), and we make an exception for Xen. There's a third option: try to make virtio-mmio work everywhere (except s390), at least in the long run. This other benefits: it makes minimal hypervisors simpler, I think it'll get rid of the limits on the number of virtio devices in a system. ARM is already going this direction, and I imagine that PPC support would be straightforward (it's already using devicetree). Does virtio-mmio have any reasonable way of doing hotplug? It could also eventually make sense to have a standard for virtio on virtio. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Thu, Sep 4, 2014 at 7:32 PM, Rusty Russell ru...@rustcorp.com.au wrote: Michael S. Tsirkin m...@redhat.com writes: On Wed, Sep 03, 2014 at 04:12:01PM +0930, Rusty Russell wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Why? What's wrong with rings in memory? AFAICT, the card would have to access guest memory to read it, using multiple DMA cycles. That's going to be slow. I don't personally know all the considerations, but AFAICT NVMe puts its rings in memory, and NMVe is very much focused on performance. There might be an argument for trying to avoid using indirect rings on real hardware to reduce the number of DMA round-trips needed for a comment. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Fri, 2014-09-05 at 12:01 +0930, Rusty Russell wrote: If the answers are both yes, then x86 is going to be able to use virtio+IOMMU, so PPC looks like the odd one out. Well, yes and no ... ppc will be able to do that too, it's just pointless and will suck performances. Additionally, it will be incompatible with existing guests since today, the guest assumes physical (doesn't use the dma mapping routines), so even if x86 grows the ability to have virtio behind an iommu in qemu, that will break existing guests. Otherwise it looks like we're really going to want to stick with the ignore IOMMU rule until (handwave future), and we make an exception for Xen. Either that or we have a capability that can be negociated. There are other reasons for wanting to allow the use of the DMA ops, such as people using virtio as a transport between two physically connected machines (such as a CPU running a PCIe endpoint to a CPU running a PCIe host, or two hosts connected to a non-transparent switch, essentially using PCIe as a fast network fabric). Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Thu, 2014-09-04 at 19:57 -0700, Andy Lutomirski wrote: There's a third option: try to make virtio-mmio work everywhere (except s390), at least in the long run. This other benefits: it makes minimal hypervisors simpler, I think it'll get rid of the limits on the number of virtio devices in a system. ARM is already going this direction, and I imagine that PPC support would be straightforward (it's already using devicetree). PCI has advantages though. Management stacks know about PCI and nothing else really. We already have all the infra to do hotplug with PCI, etc... Does virtio-mmio have any reasonable way of doing hotplug? It could also eventually make sense to have a standard for virtio on virtio. That would be very platform specific. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. Thoughts? Rusty. PS. I cc'd OASIS virtio-dev: it's subscriber only for IP reasons (to subscribe you have to promise we can use your suggestion in the standard). Feel free to remove in any replies, but it's part of the world we live in... ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Il 03/09/2014 01:20, Benjamin Herrenschmidt ha scritto: I wouldn't be so certain, as I said, the way virtio is implemented in qemu bypass the DMA layer which is where IOMMUs sit. The fact that currently x86 doesn't put an IOMMU there is not even garanteed, is it ? What happens if you try to mix and match virtio and other emulated devices that require the iommu on the same bus ? As far as QEMU is concerned, it's trivial to add a property like direct-ram-access that selects whether to bypass the IOMMU or not. And it would have zero performance cost if direct RAM access is enabled, compared to the current code. If possible, I would quirk it in the PPC code. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Il 03/09/2014 02:25, Benjamin Herrenschmidt ha scritto: But there aren't any ACPI systems with both virtio-pci and IOMMUs, right? So we could say that, henceforth, ACPI systems must declare whether virtio-pci devices live behind IOMMUs without breaking backward compatibility. I don't know for sure whether that's the case and whether we can rely on that not happening, we'll need x86 folks opinion here. IOMMU support for x86 is going to go in this week. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Sep 2, 2014 11:53 PM, Rusty Russell ru...@rustcorp.com.au wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Xen does expose dma_ops. The trick is knowing when to use it. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. The devices I care about aren't actually Xen devices. They're devices supplied by QEMU/KVM, booting a Xen hypervisor, which in turn passes the virtio device (along with every other PCI device) through to dom0. So this is exactly the same virtio device that regular x86 KVM guests would see. The reason that current code fails is that Xen guest physical addresses aren't the same as the addresses seen by the outer hypervisor. These devices don't know that physical addresses != bus addresses, so they can't advertise that fact. If we ever end up with a virtio_pci device with physical addressing, behind an IOMMU (but ignoring it), on Xen, we'll have a problem, since neither physical addressing nor dma ops will work. That being said, there are also proposals for virtio devices supplied by Xen dom0 to domU, and these will presumably work the same way, except that the device implementation will know that it's on Xen. Grr. This is mostly a result of the fact that virtio_pci devices aren't really PCI devices. I still think that virtio_pci shouldn't have to worry about this; ideally this would all be handled higher up in the device hierarchy. x86 already gets this right. Are there any hypervisors except PPC that use virtio_pci, have IOMMUs on the pci slot that virtio_pci lives in, and that use physical addressing? If not, I think that just quirking PPC will work (at least until someone wants IOMMU support in virtio_pci on PPC, in which case doing something using devicetree seems like a reasonable solution). --Andy Thoughts? Rusty. PS. I cc'd OASIS virtio-dev: it's subscriber only for IP reasons (to subscribe you have to promise we can use your suggestion in the standard). Feel free to remove in any replies, but it's part of the world we live in... ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 3, 2014 at 12:47 AM, Paolo Bonzini pbonz...@redhat.com wrote: Il 03/09/2014 02:25, Benjamin Herrenschmidt ha scritto: But there aren't any ACPI systems with both virtio-pci and IOMMUs, right? So we could say that, henceforth, ACPI systems must declare whether virtio-pci devices live behind IOMMUs without breaking backward compatibility. I don't know for sure whether that's the case and whether we can rely on that not happening, we'll need x86 folks opinion here. IOMMU support for x86 is going to go in this week. Can you try to make sure that qemu-system-x86_64 -device iommu -device virtio-balloon-pci (or whatever the syntax is) doesn't put the virtio-pci device behind the IOMMU? Because, if it does, then the kernel will have to support that, and it'll be messy. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Il 03/09/2014 09:52, Andy Lutomirski ha scritto: IOMMU support for x86 is going to go in this week. Can you try to make sure that qemu-system-x86_64 -device iommu -device virtio-balloon-pci (or whatever the syntax is) doesn't put the virtio-pci device behind the IOMMU? Because, if it does, then the kernel will have to support that, and it'll be messy. Right now it will not put the device behind the IOMMU, but I'm fairly sure that the DMAR will show the device as being behind the IOMMU. We have time till QEMU 2.2 is out to make it use the IOMMU for virtio-pci devices on x86. I'm not worried about that. The virtio-pci devices do set the bus master bit in the command register, right? I think they do, because otherwise MSIs will not be received by the guest. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Il 03/09/2014 10:05, Benjamin Herrenschmidt ha scritto: On Wed, 2014-09-03 at 09:47 +0200, Paolo Bonzini wrote: IOMMU support for x86 is going to go in this week. But won't that break virtio on x86 ? Or will virtio continue bypassing it ? IE, the guest side virtio doesn't expect an IOMMU and doesn't call the dma mappings ops. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Well, it's only for virtio and should be on by default on x86 as well if an iommu is installed no ? Yes, only for virtio---but for x86 I think it should be off by default, even if that means virtio+IOMMU requires a new kernel. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 03, 2014 at 04:12:01PM +0930, Rusty Russell wrote: Andy Lutomirski l...@amacapital.net writes: There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. Hi Andy, As you're discovering, there's a reason no one has done the DMA API before. So the problem is that ppc64's IOMMU is a platform thing, not a bus thing. They really do carve out an exception for virtio devices, because performance (LOTS of performance). It remains to be seen if other platforms have the same performance issues, but in absence of other evidence, the answer is yes. It's a hack. But having specific virtual-only devices are an even bigger hack. Physical virtio devices have been talked about, but don't actually exist in Real Life. And someone a virtio PCI card is going to have serious performance issues: mainly because they'll want the rings in the card's MMIO region, not allocated by the driver. Why? What's wrong with rings in memory? Being broken on PPC is really the least of their problems. So, what do we do? It'd be nice if Linux virtio Just Worked under Xen, though Xen's IOMMU is outside the virtio spec. Since virtio_pci can be a module, obvious hacks like having xen_arch_setup initialize a dma_ops pointer exposed by virtio_pci.c is out. Well virtio could probe for xen, it's not a lot of code. I think the best approach is to have a new feature bit (25 is free), VIRTIO_F_USE_BUS_MAPPING which indicates that a device really wants to use the mapping for the bus it is on. A real device would set this, or it won't work behind an IOMMU. A Xen device would also set this. Thoughts? Rusty. OK and it should then be active even if guest does not ack the feature (so in fact, it would have to be a mandatory feature). That can work, but I still find this a bit inelegant: this is a property of the platform, not of the device. PS. I cc'd OASIS virtio-dev: it's subscriber only for IP reasons (to subscribe you have to promise we can use your suggestion in the standard). Feel free to remove in any replies, but it's part of the world we live in... ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Sep 3, 2014 5:11 AM, Paolo Bonzini pbonz...@redhat.com wrote: Il 03/09/2014 10:05, Benjamin Herrenschmidt ha scritto: On Wed, 2014-09-03 at 09:47 +0200, Paolo Bonzini wrote: IOMMU support for x86 is going to go in this week. But won't that break virtio on x86 ? Or will virtio continue bypassing it ? IE, the guest side virtio doesn't expect an IOMMU and doesn't call the dma mappings ops. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Well, it's only for virtio and should be on by default on x86 as well if an iommu is installed no ? Yes, only for virtio---but for x86 I think it should be off by default, even if that means virtio+IOMMU requires a new kernel. Just to clarify: is it the direct-ram-access property? If so, I think I might agree. Alternatively, could QEMU easily teach the IOMMU code to generate the ACPI tables such that virtio-pci devices aren't advertised as living behind the IOMMU? This would work both with and without my patches. On the other hand, maybe this gets complicated when hotplug is involved. --Andy Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
Il 03/09/2014 17:07, Andy Lutomirski ha scritto: Yes, only for virtio---but for x86 I think it should be off by default, even if that means virtio+IOMMU requires a new kernel. Just to clarify: is it the direct-ram-access property? If so, I think I might agree. Yes. Alternatively, could QEMU easily teach the IOMMU code to generate the ACPI tables such that virtio-pci devices aren't advertised as living behind the IOMMU? This would work both with and without my patches. On the other hand, maybe this gets complicated when hotplug is involved. That could be possible. For hot-plug, you can simply forbid hotplugging with direct-ram-access=on. If you want hotplug, then we should add the direct-ram-access property to PCI bridges too (with the same limitation on hotplug). All devices under such a bridge would be outside the IOMMU, including virtio devices with direct-ram-access=off. Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 03, 2014 at 08:07:15AM -0700, Andy Lutomirski wrote: On Sep 3, 2014 5:11 AM, Paolo Bonzini pbonz...@redhat.com wrote: Il 03/09/2014 10:05, Benjamin Herrenschmidt ha scritto: On Wed, 2014-09-03 at 09:47 +0200, Paolo Bonzini wrote: IOMMU support for x86 is going to go in this week. But won't that break virtio on x86 ? Or will virtio continue bypassing it ? IE, the guest side virtio doesn't expect an IOMMU and doesn't call the dma mappings ops. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Well, it's only for virtio and should be on by default on x86 as well if an iommu is installed no ? Yes, only for virtio---but for x86 I think it should be off by default, even if that means virtio+IOMMU requires a new kernel. Just to clarify: is it the direct-ram-access property? If so, I think I might agree. Alternatively, could QEMU easily teach the IOMMU code to generate the ACPI tables such that virtio-pci devices aren't advertised as living behind the IOMMU? This would work both with and without my patches. How exactly does this look in ACPI? On the other hand, maybe this gets complicated when hotplug is involved. --Andy Paolo ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 3, 2014 at 9:39 AM, Michael S. Tsirkin m...@redhat.com wrote: On Wed, Sep 03, 2014 at 08:07:15AM -0700, Andy Lutomirski wrote: On Sep 3, 2014 5:11 AM, Paolo Bonzini pbonz...@redhat.com wrote: Il 03/09/2014 10:05, Benjamin Herrenschmidt ha scritto: On Wed, 2014-09-03 at 09:47 +0200, Paolo Bonzini wrote: IOMMU support for x86 is going to go in this week. But won't that break virtio on x86 ? Or will virtio continue bypassing it ? IE, the guest side virtio doesn't expect an IOMMU and doesn't call the dma mappings ops. However, it is and likely will remain niche enough that I don't really care about performance loss from IOMMU support. If you enable it, you want it. So from the QEMU point of view we can simply add the direct-ram-access property, and have the pseries machine turn it on by default (while other machines can leave it off by default---they have no IOMMU and thus no performance cost). Well, it's only for virtio and should be on by default on x86 as well if an iommu is installed no ? Yes, only for virtio---but for x86 I think it should be off by default, even if that means virtio+IOMMU requires a new kernel. Just to clarify: is it the direct-ram-access property? If so, I think I might agree. Alternatively, could QEMU easily teach the IOMMU code to generate the ACPI tables such that virtio-pci devices aren't advertised as living behind the IOMMU? This would work both with and without my patches. How exactly does this look in ACPI? I think that all you need is a PCI device or segment that isn't included in the scope of any DRHD. I could be wrong. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 5:43 PM, Benjamin Herrenschmidt b...@au1.ibm.com wrote: On Tue, 2014-09-02 at 17:32 -0700, Andy Lutomirski wrote: I agree *except* that implementing it will be a real PITA and (I think) can't be done without changing code in arch/. My patches plus an ifdef powerpc will be functionally equivalent, just uglier. So for powerpc, it's a 2 liner inside virtio-pci, but yes, it might be more of a problem for s390, I'm not too sure what they do in that area. Bigger quirk: on a standard s390 virtio guest configuration, dma_map_single etc will fail to link. Yuck I tried this in v1 of these patches. So we can poke at the archdata all day, but we can't build a kernel like that :( I would like the s390 people to chime in here, it still looks like the best way to go if they can fix things on their side :-) So until the dma_ops pointer move into struct device and CONFIG_HAS_DMA becomes mandatory (or mandatory enough that virtio can depend on it), I don't think we can do it this way. I see, it's a bummer because it would be a lot cleaner. I'll send a v5 that is the same as v4 except with physical addressing hardcoded in for powerpc. Thanks. That will do for now, but ideally we want to make it a function of some flag from the implementation, so let's see what Rusty has to say. I've confirmed that ppc64 (on QEMU) breaks without the ppc special case and that ppc64 keeps working with the special case. Once Rusty's patches settle down, I'll rebase onto them and send v5. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Mon, 2014-09-01 at 22:55 -0700, Andy Lutomirski wrote: On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. It will definitely not work on ppc64. We always have IOMMUs on pseries, all PCI busses do, and because it's a paravirtualized environment, napping/unmapping pages means hypercalls - expensive. But our virtio implementation bypasses it in qemu, so if virtio-pci starts using the DMA mapping API without changing the DMA ops under the hood, it will break for us. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Wed, Sep 03, 2014 at 06:53:33AM +1000, Benjamin Herrenschmidt wrote: On Mon, 2014-09-01 at 22:55 -0700, Andy Lutomirski wrote: On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. It will definitely not work on ppc64. We always have IOMMUs on pseries, all PCI busses do, and because it's a paravirtualized environment, napping/unmapping pages means hypercalls - expensive. But our virtio implementation bypasses it in qemu, so if virtio-pci starts using the DMA mapping API without changing the DMA ops under the hood, it will break for us. What is the default dma_ops that the Linux guests start with as guests under ppc64? Thanks! Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2014-09-02 at 16:56 -0400, Konrad Rzeszutek Wilk wrote: On Wed, Sep 03, 2014 at 06:53:33AM +1000, Benjamin Herrenschmidt wrote: On Mon, 2014-09-01 at 22:55 -0700, Andy Lutomirski wrote: On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. It will definitely not work on ppc64. We always have IOMMUs on pseries, all PCI busses do, and because it's a paravirtualized environment, napping/unmapping pages means hypercalls - expensive. But our virtio implementation bypasses it in qemu, so if virtio-pci starts using the DMA mapping API without changing the DMA ops under the hood, it will break for us. What is the default dma_ops that the Linux guests start with as guests under ppc64? On pseries (which is what we care the most about nowadays) it's dma_iommu_ops() which in turn call into the TCE code for populating the IOMMU entries which calls the hypervisor. Cheers, Ben. Thanks! Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Mon, Sep 01, 2014 at 10:55:29PM -0700, Andy Lutomirski wrote: On Mon, Sep 1, 2014 at 3:16 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Mon, 2014-09-01 at 10:39 -0700, Andy Lutomirski wrote: Changes from v1: - Using the DMA API is optional now. It would be nice to improve the DMA API to the point that it could be used unconditionally, but s390 proves that we're not there yet. - Includes patch 4, which fixes DMA debugging warnings from virtio_net. I'm not sure if you saw my reply on the other thread but I have a few comments based on the above it would be nice if ... Yeah, sorry, I sort of thought I responded, but I didn't do a very good job. So here we have both a yes and a no :-) It would be nice to avoid those if () games all over and indeed just use the DMA API, *however* we most certainly don't want to actually create IOMMU mappings for the KVM virio case. This would be a massive loss in performances on several platforms and generally doesn't make much sense. However, we can still use the API without that on any architecture where the dma mapping API ends up calling the generic dma_map_ops, it becomes just a matter of virtio setting up some special nop ops when needed. I'm not quite convinced that this is a good idea. I think that there are three relevant categories of virtio devices: a) Any virtio device where the normal DMA ops are nops. This includes x86 without an IOMMU (e.g. in a QEMU/KVM guest), 32-bit ARM, and probably many other architectures. In this case, what we do only matters for performance, not for correctness. Ideally the arch DMA ops are fast. b) Virtio devices that use physical addressing on systems where DMA ops either don't exist at all (most s390) or do something nontrivial. In this case, we must either override the DMA ops or just not use them. c) Virtio devices that use bus addressing. This includes everything on Xen (because the physical addresses are nonsense) and any actual physical PCI device that speaks virtio on a system with an IOMMU. In this case, we must use the DMA ops. The issue is that, on systems with DMA ops that do something, we need to make sure that we know whether we're in case (b) or (c). In these patches, I've made the assumption that, if the virtio devices lives on the PCI bus, then it uses the same type of addressing that any other device on that PCI bus would use. On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. How exactly does one not advertise an IOMMU for a specific device? Could you please clarify? But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. One simple fix is to defer this all until virtio 1.0. virtio 1.0 has an alternative set of IDs for virtio pci, that can be used if you are making an incompatible change. We can use that if there's an iommu. The difficulty here resides in the fact that we have never completely made the dma_map_ops generic. The ops themselves are defined generically as are the dma_map_* interfaces based on them, but the location of the ops pointer is still more/less arch specific and some architectures still chose not to use that indirection at all I believe. I'd be happy to update the patches if someone does this, but I don't really want to attack the DMA API on all architectures right now. In the mean time, at least s390 requires that we be able to compile out the DMA API calls. I'd rather see s390 provide working no-op dma ops for all of the struct devices that provide virtio interfaces. On a related note, shouldn't virtio be doing something to provide dma ops to the virtio device and any of its children? I don't know how it would even try to do this, given how architecture-dependent this code currently is. Calling dma_map_single on the virtio device (as opposed to its parent) is currently likely to crash on x86. Fortunately, nothing does this. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 1:53 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Mon, 2014-09-01 at 22:55 -0700, Andy Lutomirski wrote: On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. It will definitely not work on ppc64. We always have IOMMUs on pseries, all PCI busses do, and because it's a paravirtualized environment, napping/unmapping pages means hypercalls - expensive. But our virtio implementation bypasses it in qemu, so if virtio-pci starts using the DMA mapping API without changing the DMA ops under the hood, it will break for us. Let's take a step back from from the implementation. What is a driver for a virtio PCI device (i.e. a PCI device with vendor 0x1af4) supposed to do on ppc64? It can send the device physical addresses and ignore the normal PCI DMA semantics, which is what the current virtio_pci driver does. This seems like a layering violation, and this won't work if the device is a real PCI device. Alternatively, it can treat the device like any other PCI device and use the IOMMU. This is a bit slower, and it is also incompatible with current hypervisors. There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. We could teach virtio_pci to use physical addressing on ppc64, but that seems like a pretty awful hack, and it'll start needing quirks as soon as someone tries to plug a virtio-speaking PCI card into a ppc64 machine. Ideas? x86 and arm seem to be safe here, since AFAIK there is no such thing as a physically addressed virtio PCI device on a bus with an IOMMU on x86, arm, or arm64. [1] https://lwn.net/Articles/580186/ Cheers, Ben. -- Andy Lutomirski AMA Capital Management, LLC ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 2:10 PM, Michael S. Tsirkin m...@redhat.com wrote: On Mon, Sep 01, 2014 at 10:55:29PM -0700, Andy Lutomirski wrote: On Mon, Sep 1, 2014 at 3:16 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Mon, 2014-09-01 at 10:39 -0700, Andy Lutomirski wrote: Changes from v1: - Using the DMA API is optional now. It would be nice to improve the DMA API to the point that it could be used unconditionally, but s390 proves that we're not there yet. - Includes patch 4, which fixes DMA debugging warnings from virtio_net. I'm not sure if you saw my reply on the other thread but I have a few comments based on the above it would be nice if ... Yeah, sorry, I sort of thought I responded, but I didn't do a very good job. So here we have both a yes and a no :-) It would be nice to avoid those if () games all over and indeed just use the DMA API, *however* we most certainly don't want to actually create IOMMU mappings for the KVM virio case. This would be a massive loss in performances on several platforms and generally doesn't make much sense. However, we can still use the API without that on any architecture where the dma mapping API ends up calling the generic dma_map_ops, it becomes just a matter of virtio setting up some special nop ops when needed. I'm not quite convinced that this is a good idea. I think that there are three relevant categories of virtio devices: a) Any virtio device where the normal DMA ops are nops. This includes x86 without an IOMMU (e.g. in a QEMU/KVM guest), 32-bit ARM, and probably many other architectures. In this case, what we do only matters for performance, not for correctness. Ideally the arch DMA ops are fast. b) Virtio devices that use physical addressing on systems where DMA ops either don't exist at all (most s390) or do something nontrivial. In this case, we must either override the DMA ops or just not use them. c) Virtio devices that use bus addressing. This includes everything on Xen (because the physical addresses are nonsense) and any actual physical PCI device that speaks virtio on a system with an IOMMU. In this case, we must use the DMA ops. The issue is that, on systems with DMA ops that do something, we need to make sure that we know whether we're in case (b) or (c). In these patches, I've made the assumption that, if the virtio devices lives on the PCI bus, then it uses the same type of addressing that any other device on that PCI bus would use. On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. How exactly does one not advertise an IOMMU for a specific device? Could you please clarify? See https://software.intel.com/en-us/blogs/2009/09/11/decoding-the-dmar-tables-in-acpiiommu-part-2 I think that all that needs to happen is for ACPI to not list the device in the scope of any drhd unit. I don't know whether this works correctly, but it looks like the iommu_dummy and the init_no_remapping_devices code in intel-iommu.c exists for almost exactly this purpose. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. One simple fix is to defer this all until virtio 1.0. virtio 1.0 has an alternative set of IDs for virtio pci, that can be used if you are making an incompatible change. We can use that if there's an iommu. How? If someone builds a physical device compliant with the virtio 1.0 specification, how do can that device know whether it's behind an IOMMU? The IOMMU is part of the host (or Xen, sort of), not the PCI device. I suppose that virtio 1.0 could add a bit indicating that the virtio device is a physical piece of hardware (presumably this should be PCI-specific). --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, 2014-09-02 at 14:37 -0700, Andy Lutomirski wrote: Let's take a step back from from the implementation. What is a driver for a virtio PCI device (i.e. a PCI device with vendor 0x1af4) supposed to do on ppc64? Today, it's supposed to send guest physical addresses. We can make that optional via some nego or capabilities to support more esoteric setups but for backward compatibility, this must remain the default behaviour. It can send the device physical addresses and ignore the normal PCI DMA semantics, which is what the current virtio_pci driver does. This seems like a layering violation, and this won't work if the device is a real PCI device. Correct, it's an original virtio implementation choice for maximum performances. Alternatively, it can treat the device like any other PCI device and use the IOMMU. This is a bit slower, and it is also incompatible with current hypervisors. This is a potentially a LOT slower and is backward incompatible with current qemu/KVM and kvmtool yes. The slowness can be alleviated using various techniques, for example on ppc64 we can create a DMA window that contains a permanent mapping of the entire guest space, so we could use such a thing for virtio. Another think we could do potentially is advertize via the device-tree that such a bus uses a direct mapping and have the guest use appropriate direct map dma_ops. But we need to keep backward compatibility with existing guest/hypervisors so the default must remain as it is. There really are virtio devices that are pieces of silicon and not figments of a hypervisor's imagination [1]. I am aware of that. There are also attempts at using virtio to make two machines communicate via a PCIe link (either with one as endpoint of the other or via a non-transparent switch). Which is why I'm not objecting to what you are trying to do ;-) My suggestion was that it might be a cleaner approach to do that by having the individual virtio drivers always use the dma_map_* API, and limiting the kludgery to a combination of virtio_pci core and arch code by selecting an appropriate set of dma_map_ops, defaulting with a transparent (or direct) one as our current default case (and thus overriding the iommu ones provided by the arch). We could teach virtio_pci to use physical addressing on ppc64, but that seems like a pretty awful hack, and it'll start needing quirks as soon as someone tries to plug a virtio-speaking PCI card into a ppc64 machine. But x86_64 is the same no ? The day it starts growing an iommu emulation in qemu (and I've heard it's happening) it will still want to do direct bypass for virtio for performance. Ideas? x86 and arm seem to be safe here, since AFAIK there is no such thing as a physically addressed virtio PCI device on a bus with an IOMMU on x86, arm, or arm64. Today I wouldn't bet on it to remain that way. The qemu implementation of virtio is physically addressed and you don't necessarily have a choice of which device gets an iommu and which not. Cheers, Ben. [1] https://lwn.net/Articles/580186/ Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 3:10 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Tue, 2014-09-02 at 14:37 -0700, Andy Lutomirski wrote: Let's take a step back from from the implementation. What is a driver for a virtio PCI device (i.e. a PCI device with vendor 0x1af4) supposed to do on ppc64? Today, it's supposed to send guest physical addresses. We can make that optional via some nego or capabilities to support more esoteric setups but for backward compatibility, this must remain the default behaviour. I think it only needs to remain the default in cases where the alternative (bus addressing) won't work. I think that, so far, this is just ppc64. But see below... My suggestion was that it might be a cleaner approach to do that by having the individual virtio drivers always use the dma_map_* API, and limiting the kludgery to a combination of virtio_pci core and arch code by selecting an appropriate set of dma_map_ops, defaulting with a transparent (or direct) one as our current default case (and thus overriding the iommu ones provided by the arch). I think the cleanest way of all would be to get the bus drivers to do the right thing so that all of the virtio code can just use the dma api. I don't know whether this is achievable. We could teach virtio_pci to use physical addressing on ppc64, but that seems like a pretty awful hack, and it'll start needing quirks as soon as someone tries to plug a virtio-speaking PCI card into a ppc64 machine. But x86_64 is the same no ? The day it starts growing an iommu emulation in qemu (and I've heard it's happening) it will still want to do direct bypass for virtio for performance. I don't think so. I would argue that it's a straight-up bug for QEMU to expose a physically-addressed virtio-pci device to the guest behind an emulated IOMMU. QEMU may already be doing that on ppc64, but it isn't on x86_64 or arm (yet). On x86_64, I'm pretty sure that QEMU can emulate an IOMMU for everything except the virtio-pci devices. The ACPI DMAR stuff is quite expressive. On ARM, I hope the QEMU will never implement a PCI IOMMU. As far as I could tell when I looked last week, none of the newer QEMU-emulated ARM machines even support PCI. Even if QEMU were to implement a PCI IOMMU on some future ARM machine, it could continue using virtio-mmio for virtio devices. So ppc might actually be the only system that has or will have physically-addressed virtio PCI devices that are behind an IOMMU. Can this be handled in a ppc64-specific way? Is there any way that the kernel can distinguish a QEMU-provided virtio PCI device from a physical PCIe thing? It would be kind of nice to address this without adding complexity to the virtio spec. Maybe virtio 1.0 devices could be assumed to use bus addressing unless a new devicetree property says otherwise. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 4:20 PM, Benjamin Herrenschmidt b...@au1.ibm.com wrote: On Tue, 2014-09-02 at 16:11 -0700, Andy Lutomirski wrote: I don't think so. I would argue that it's a straight-up bug for QEMU to expose a physically-addressed virtio-pci device to the guest behind an emulated IOMMU. QEMU may already be doing that on ppc64, but it isn't on x86_64 or arm (yet). Last I looked, it does on everything, it bypasses the DMA layer in qemu which is where IOMMUs are implemented. I believe you, but I'm not convinced that this means much from the guest's POV, except on ppc64. On x86_64, I'm pretty sure that QEMU can emulate an IOMMU for everything except the virtio-pci devices. The ACPI DMAR stuff is quite expressive. Well, *except* virtio, exactly... But there aren't any ACPI systems with both virtio-pci and IOMMUs, right? So we could say that, henceforth, ACPI systems must declare whether virtio-pci devices live behind IOMMUs without breaking backward compatibility. On ARM, I hope the QEMU will never implement a PCI IOMMU. As far as I could tell when I looked last week, none of the newer QEMU-emulated ARM machines even support PCI. Even if QEMU were to implement a PCI IOMMU on some future ARM machine, it could continue using virtio-mmio for virtio devices. Possibly... So ppc might actually be the only system that has or will have physically-addressed virtio PCI devices that are behind an IOMMU. Can this be handled in a ppc64-specific way? I wouldn't be so certain, as I said, the way virtio is implemented in qemu bypass the DMA layer which is where IOMMUs sit. The fact that currently x86 doesn't put an IOMMU there is not even garanteed, is it ? What happens if you try to mix and match virtio and other emulated devices that require the iommu on the same bus ? AFAIK QEMU doesn't support IOMMUs at all on x86, so current versions of QEMU really do guarantee that virtio-pci on x86 has no IOMMU, even if that guarantee is purely accidental. If we could discriminate virtio devices to a specific host bridge and guarantee no mix match, we could probably add a concept of IOMMU-less bus but that would require guest changes which limits the usefulness. Is there any way that the kernel can distinguish a QEMU-provided virtio PCI device from a physical PCIe thing? Not with existing guests which cannot be changed. Existing distros are out with those drivers. If we add a backward compatibility mechanism, then we could add something yes, provided we can segregate virtio onto a dedicated host bridge (which can be a problem with the libvirt trainwreck...) Ugh. So here's an ugly proposal: Step 1: Make virtio-pci use the DMA API only on x86. This will at least fix Xen and people experimenting with virtio hardware on x86, and it won't break anything, since there are no emulated IOMMUs on x86. Step 2: Update the virtio spec. Virtio 1.0 PCI devices should set a new bit if they are physically addressed. If that bit is clear, then the device is assumed to be addressed in accordance with the platform's standard addressing model for PCI. Presumably this would be something like VIRTIO_F_BUS_ADDRESSING = 33, and the spec would say something like Physical devices compatible with this specification MUST offer VIRTIO_F_BUS_ADDRESSING. Drivers MUST implement this feature. Alternatively, this could live in a PCI configuration capability. Step 3: Update virtio-pci to use the DMA API for all devices on x86 and for devices that advertise bus addressing on other architectures. I think this proposal will work, but I also think it sucks and I'd really like to see a better counter-proposal. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Tue, Sep 2, 2014 at 5:25 PM, Benjamin Herrenschmidt b...@au1.ibm.com wrote: On Tue, 2014-09-02 at 16:42 -0700, Andy Lutomirski wrote: So here's an ugly proposal: Step 1: Make virtio-pci use the DMA API only on x86. This will at least fix Xen and people experimenting with virtio hardware on x86, and it won't break anything, since there are no emulated IOMMUs on x86. I think we should make all virtio drivers use the DMA API and just have different set of dma_ops. We can make a simple ifdef powerpc if needed in virtio-pci that force the dma-ops of the device to some direct bypass ops at init time. That way no need to select whether to use the DMA API or not, just always use it, and add a tweak to replace the DMA ops with the direct ones on the archs/platforms that need that. That was my original proposal and I still think it's the best approach. I agree *except* that implementing it will be a real PITA and (I think) can't be done without changing code in arch/. My patches plus an ifdef powerpc will be functionally equivalent, just uglier. As I said, make it always use the DMA API, but add a quirk to replace the dma_ops with some NULL ops on platforms that need it. The only issue with that is the location of the dma ops is arch specific, so that one function will contain some ifdefs, but the rest of the code can just use the DMA API. Bigger quirk: on a standard s390 virtio guest configuration, dma_map_single etc will fail to link. I tried this in v1 of these patches. So we can poke at the archdata all day, but we can't build a kernel like that :( So until the dma_ops pointer move into struct device and CONFIG_HAS_DMA becomes mandatory (or mandatory enough that virtio can depend on it), I don't think we can do it this way. I'll send a v5 that is the same as v4 except with physical addressing hardcoded in for powerpc. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
[PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
This fixes virtio on Xen guests as well as on any other platform that uses virtio_pci on which physical addresses don't match bus addresses. This can be tested with: virtme-run --xen xen --kimg arch/x86/boot/bzImage --console using virtme from here: https://git.kernel.org/cgit/utils/kernel/virtme/virtme.git Without these patches, the guest hangs forever. With these patches, everything works. This should be safe on all platforms that I'm aware of. That doesn't mean that there isn't anything that I missed. Thanks to everyone for putting up with the development of this series. Hopefully it'll be the end of DMA issues in virtio. :) Changes from v3: - virtio_pci only asks virtio_ring to use the DMA_API if !PCI_DMA_BUS_IS_PHYS. - Reduce tools/virtio breakage. It's now merely as broken as before instead of being even more broken. - Drop the sg_next changes -- Rusty's version is better. Changes from v2: - Reordered patches. - Fixed a virtio_net OOPS. Changes from v1: - Using the DMA API is optional now. It would be nice to improve the DMA API to the point that it could be used unconditionally, but s390 proves that we're not there yet. - Includes patch 4, which fixes DMA debugging warnings from virtio_net. Andy Lutomirski (4): virtio_ring: Support DMA APIs if requested virtio_pci: Use the DMA API for virtqueues virtio_net: Don't set the end flag on reusable sg entries virtio_net: Stop doing DMA from the stack drivers/lguest/lguest_device.c | 3 +- drivers/misc/mic/card/mic_virtio.c | 2 +- drivers/net/virtio_net.c | 59 +++ drivers/remoteproc/remoteproc_virtio.c | 4 +- drivers/s390/kvm/kvm_virtio.c | 2 +- drivers/s390/kvm/virtio_ccw.c | 4 +- drivers/virtio/virtio_mmio.c | 5 +- drivers/virtio/virtio_pci.c| 41 ++-- drivers/virtio/virtio_ring.c | 187 + include/linux/virtio_ring.h| 1 + tools/virtio/linux/dma-mapping.h | 17 +++ tools/virtio/linux/virtio.h| 1 + tools/virtio/virtio_test.c | 2 +- tools/virtio/vringh_test.c | 3 +- 14 files changed, 268 insertions(+), 63 deletions(-) create mode 100644 tools/virtio/linux/dma-mapping.h -- 1.9.3 ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Mon, 2014-09-01 at 10:39 -0700, Andy Lutomirski wrote: Changes from v1: - Using the DMA API is optional now. It would be nice to improve the DMA API to the point that it could be used unconditionally, but s390 proves that we're not there yet. - Includes patch 4, which fixes DMA debugging warnings from virtio_net. I'm not sure if you saw my reply on the other thread but I have a few comments based on the above it would be nice if ... So here we have both a yes and a no :-) It would be nice to avoid those if () games all over and indeed just use the DMA API, *however* we most certainly don't want to actually create IOMMU mappings for the KVM virio case. This would be a massive loss in performances on several platforms and generally doesn't make much sense. However, we can still use the API without that on any architecture where the dma mapping API ends up calling the generic dma_map_ops, it becomes just a matter of virtio setting up some special nop ops when needed. The difficulty here resides in the fact that we have never completely made the dma_map_ops generic. The ops themselves are defined generically as are the dma_map_* interfaces based on them, but the location of the ops pointer is still more/less arch specific and some architectures still chose not to use that indirection at all I believe. Cheers, Ben. ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization
Re: [PATCH v4 0/4] virtio: Clean up scatterlists and use the DMA API
On Mon, Sep 1, 2014 at 3:16 PM, Benjamin Herrenschmidt b...@kernel.crashing.org wrote: On Mon, 2014-09-01 at 10:39 -0700, Andy Lutomirski wrote: Changes from v1: - Using the DMA API is optional now. It would be nice to improve the DMA API to the point that it could be used unconditionally, but s390 proves that we're not there yet. - Includes patch 4, which fixes DMA debugging warnings from virtio_net. I'm not sure if you saw my reply on the other thread but I have a few comments based on the above it would be nice if ... Yeah, sorry, I sort of thought I responded, but I didn't do a very good job. So here we have both a yes and a no :-) It would be nice to avoid those if () games all over and indeed just use the DMA API, *however* we most certainly don't want to actually create IOMMU mappings for the KVM virio case. This would be a massive loss in performances on several platforms and generally doesn't make much sense. However, we can still use the API without that on any architecture where the dma mapping API ends up calling the generic dma_map_ops, it becomes just a matter of virtio setting up some special nop ops when needed. I'm not quite convinced that this is a good idea. I think that there are three relevant categories of virtio devices: a) Any virtio device where the normal DMA ops are nops. This includes x86 without an IOMMU (e.g. in a QEMU/KVM guest), 32-bit ARM, and probably many other architectures. In this case, what we do only matters for performance, not for correctness. Ideally the arch DMA ops are fast. b) Virtio devices that use physical addressing on systems where DMA ops either don't exist at all (most s390) or do something nontrivial. In this case, we must either override the DMA ops or just not use them. c) Virtio devices that use bus addressing. This includes everything on Xen (because the physical addresses are nonsense) and any actual physical PCI device that speaks virtio on a system with an IOMMU. In this case, we must use the DMA ops. The issue is that, on systems with DMA ops that do something, we need to make sure that we know whether we're in case (b) or (c). In these patches, I've made the assumption that, if the virtio devices lives on the PCI bus, then it uses the same type of addressing that any other device on that PCI bus would use. On x86, at least, I doubt that we'll ever see a physically addressed PCI virtio device for which ACPI advertises an IOMMU, since any sane hypervisor will just not advertise an IOMMU for the virtio device. But are there arm64 or PPC guests that use virtio_pci, that have IOMMUs, and that will malfunction if the virtio_pci driver ends up using the IOMMU? I certainly hope not, since these systems might be very hard-pressed to work right if someone plugged in a physical virtio-speaking PCI device. The difficulty here resides in the fact that we have never completely made the dma_map_ops generic. The ops themselves are defined generically as are the dma_map_* interfaces based on them, but the location of the ops pointer is still more/less arch specific and some architectures still chose not to use that indirection at all I believe. I'd be happy to update the patches if someone does this, but I don't really want to attack the DMA API on all architectures right now. In the mean time, at least s390 requires that we be able to compile out the DMA API calls. I'd rather see s390 provide working no-op dma ops for all of the struct devices that provide virtio interfaces. On a related note, shouldn't virtio be doing something to provide dma ops to the virtio device and any of its children? I don't know how it would even try to do this, given how architecture-dependent this code currently is. Calling dma_map_single on the virtio device (as opposed to its parent) is currently likely to crash on x86. Fortunately, nothing does this. --Andy ___ Virtualization mailing list Virtualization@lists.linux-foundation.org https://lists.linuxfoundation.org/mailman/listinfo/virtualization