| Message ID | 20190802145017.42543-1-steven.price@arm.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | arm64: Stolen time support | expand |
On Fri, 2 Aug 2019 15:50:08 +0100 Steven Price <steven.price@arm.com> wrote: Hi Steven, > This series add support for paravirtualized time for arm64 guests and > KVM hosts following the specification in Arm's document DEN 0057A: > > https://developer.arm.com/docs/den0057/a > > It implements support for stolen time, allowing the guest to > identify time when it is forcibly not executing. > > It doesn't implement support for Live Physical Time (LPT) as there are > some concerns about the overheads and approach in the above > specification, and I expect an updated version of the specification to > be released soon with just the stolen time parts. Thanks for posting this. My current concern with this series is around the fact that we allocate memory from the kernel on behalf of the guest. It is the first example of such thing in the ARM port, and I can't really say I'm fond of it. x86 seems to get away with it by having the memory allocated from userspace, why I tend to like more. Yes, put_user is more expensive than a straight store, but this isn't done too often either. What is the rational for your current approach? Thanks, M.
On 03/08/2019 19:05, Marc Zyngier wrote: > On Fri, 2 Aug 2019 15:50:08 +0100 > Steven Price <steven.price@arm.com> wrote: > > Hi Steven, > >> This series add support for paravirtualized time for arm64 guests and >> KVM hosts following the specification in Arm's document DEN 0057A: >> >> https://developer.arm.com/docs/den0057/a >> >> It implements support for stolen time, allowing the guest to >> identify time when it is forcibly not executing. >> >> It doesn't implement support for Live Physical Time (LPT) as there are >> some concerns about the overheads and approach in the above >> specification, and I expect an updated version of the specification to >> be released soon with just the stolen time parts. > > Thanks for posting this. > > My current concern with this series is around the fact that we allocate > memory from the kernel on behalf of the guest. It is the first example > of such thing in the ARM port, and I can't really say I'm fond of it. > > x86 seems to get away with it by having the memory allocated from > userspace, why I tend to like more. Yes, put_user is more > expensive than a straight store, but this isn't done too often either. > > What is the rational for your current approach? As I see it there are 3 approaches that can be taken here: 1. Hypervisor allocates memory and adds it to the virtual machine. This means that everything to do with the 'device' is encapsulated behind the KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the stolen time structure to be fast it cannot be a trapping region and has to be backed by real memory - in this case allocated by the host kernel. 2. Host user space allocates memory. Similar to above, but this time user space needs to manage the memory region as well as the usual KVM_CREATE_DEVICE dance. I've no objection to this, but it means kvmtool/QEMU needs to be much more aware of what is going on (e.g. how to size the memory region). 3. Guest kernel "donates" the memory to the hypervisor for the structure. As far as I'm aware this is what x86 does. The problems I see this approach are: a) kexec becomes much more tricky - there needs to be a disabling mechanism for the guest to stop the hypervisor scribbling on memory before starting the new kernel. b) If there is more than one entity that is interested in the information (e.g. firmware and kernel) then this requires some form of arbitration in the guest because the hypervisor doesn't want to have to track an arbitrary number of regions to update. c) Performance can suffer if the host kernel doesn't have a suitably aligned/sized area to use. As you say - put_user() is more expensive. The structure is updated on every return to the VM. Of course x86 does prove the third approach can work, but I'm not sure which is actually better. Avoid the kexec cancellation requirements was the main driver of the current approach. Although many of the conversations about this were also tied up with Live Physical Time which adds its own complications. Steve
On 05/08/2019 14:06, Steven Price wrote: > On 03/08/2019 19:05, Marc Zyngier wrote: >> On Fri, 2 Aug 2019 15:50:08 +0100 >> Steven Price <steven.price@arm.com> wrote: >> >> Hi Steven, >> >>> This series add support for paravirtualized time for arm64 guests and >>> KVM hosts following the specification in Arm's document DEN 0057A: >>> >>> https://developer.arm.com/docs/den0057/a >>> >>> It implements support for stolen time, allowing the guest to >>> identify time when it is forcibly not executing. >>> >>> It doesn't implement support for Live Physical Time (LPT) as there are >>> some concerns about the overheads and approach in the above >>> specification, and I expect an updated version of the specification to >>> be released soon with just the stolen time parts. >> >> Thanks for posting this. >> >> My current concern with this series is around the fact that we allocate >> memory from the kernel on behalf of the guest. It is the first example >> of such thing in the ARM port, and I can't really say I'm fond of it. >> >> x86 seems to get away with it by having the memory allocated from >> userspace, why I tend to like more. Yes, put_user is more >> expensive than a straight store, but this isn't done too often either. >> >> What is the rational for your current approach? > > As I see it there are 3 approaches that can be taken here: > > 1. Hypervisor allocates memory and adds it to the virtual machine. This > means that everything to do with the 'device' is encapsulated behind the > KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the > stolen time structure to be fast it cannot be a trapping region and has > to be backed by real memory - in this case allocated by the host kernel. > > 2. Host user space allocates memory. Similar to above, but this time > user space needs to manage the memory region as well as the usual > KVM_CREATE_DEVICE dance. I've no objection to this, but it means > kvmtool/QEMU needs to be much more aware of what is going on (e.g. how > to size the memory region). > > 3. Guest kernel "donates" the memory to the hypervisor for the > structure. As far as I'm aware this is what x86 does. The problems I see > this approach are: > > a) kexec becomes much more tricky - there needs to be a disabling > mechanism for the guest to stop the hypervisor scribbling on memory > before starting the new kernel. > > b) If there is more than one entity that is interested in the > information (e.g. firmware and kernel) then this requires some form of > arbitration in the guest because the hypervisor doesn't want to have to > track an arbitrary number of regions to update. > > c) Performance can suffer if the host kernel doesn't have a suitably > aligned/sized area to use. As you say - put_user() is more expensive. > The structure is updated on every return to the VM. > > > Of course x86 does prove the third approach can work, but I'm not sure > which is actually better. Avoid the kexec cancellation requirements was > the main driver of the current approach. Although many of the > conversations about this were also tied up with Live Physical Time which > adds its own complications. My current train of thoughts is around (2): - We don't need a new mechanism to track pages or deal with overlapping IPA ranges - We can get rid of the save/restore interface The drawback is that the amount of memory required per vcpu becomes ABI. I don't think that's a huge deal, as the hypervisor has the same contract with the guest. We also take a small hit with put_user(), but this is only done as a consequence of vcpu_load() (and not on every entry as you suggest above). It'd be worth quantifying this overhead before making any decision one way or another. Thanks, M.
On 05.08.19 15:06, Steven Price wrote: > On 03/08/2019 19:05, Marc Zyngier wrote: >> On Fri, 2 Aug 2019 15:50:08 +0100 >> Steven Price <steven.price@arm.com> wrote: >> >> Hi Steven, >> >>> This series add support for paravirtualized time for arm64 guests and >>> KVM hosts following the specification in Arm's document DEN 0057A: >>> >>> https://developer.arm.com/docs/den0057/a >>> >>> It implements support for stolen time, allowing the guest to >>> identify time when it is forcibly not executing. >>> >>> It doesn't implement support for Live Physical Time (LPT) as there are >>> some concerns about the overheads and approach in the above >>> specification, and I expect an updated version of the specification to >>> be released soon with just the stolen time parts. >> >> Thanks for posting this. >> >> My current concern with this series is around the fact that we allocate >> memory from the kernel on behalf of the guest. It is the first example >> of such thing in the ARM port, and I can't really say I'm fond of it. >> >> x86 seems to get away with it by having the memory allocated from >> userspace, why I tend to like more. Yes, put_user is more >> expensive than a straight store, but this isn't done too often either. >> >> What is the rational for your current approach? > > As I see it there are 3 approaches that can be taken here: > > 1. Hypervisor allocates memory and adds it to the virtual machine. This > means that everything to do with the 'device' is encapsulated behind the > KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the > stolen time structure to be fast it cannot be a trapping region and has > to be backed by real memory - in this case allocated by the host kernel. > > 2. Host user space allocates memory. Similar to above, but this time > user space needs to manage the memory region as well as the usual > KVM_CREATE_DEVICE dance. I've no objection to this, but it means > kvmtool/QEMU needs to be much more aware of what is going on (e.g. how > to size the memory region). You ideally want to get the host overhead for a VM to as little as you can. I'm not terribly fond of the idea of reserving a full page just because we're too afraid of having the guest donate memory. > > 3. Guest kernel "donates" the memory to the hypervisor for the > structure. As far as I'm aware this is what x86 does. The problems I see > this approach are: > > a) kexec becomes much more tricky - there needs to be a disabling > mechanism for the guest to stop the hypervisor scribbling on memory > before starting the new kernel. I wouldn't call "quiesce a device" much more tricky. We have to do that for other devices as well today. > b) If there is more than one entity that is interested in the > information (e.g. firmware and kernel) then this requires some form of > arbitration in the guest because the hypervisor doesn't want to have to > track an arbitrary number of regions to update. Why would FW care? > c) Performance can suffer if the host kernel doesn't have a suitably > aligned/sized area to use. As you say - put_user() is more expensive. Just define the interface to always require natural alignment when donating a memory location? > The structure is updated on every return to the VM. If you really do suffer from put_user(), there are alternatives. You could just map the page on the registration hcall and then leave it pinned until the vcpu gets destroyed again. > Of course x86 does prove the third approach can work, but I'm not sure > which is actually better. Avoid the kexec cancellation requirements was > the main driver of the current approach. Although many of the I really don't understand the problem with kexec cancellation. Worst case, let guest FW set it up for you and propagate only the address down via ACPI/DT. That way you can mark the respective memory as reserved too. But even with a Linux only mechanism, just take a look at arch/x86/kernel/kvmclock.c. All they do to remove the map is to hook into machine_crash_shutdown() and machine_shutdown(). Alex > conversations about this were also tied up with Live Physical Time which > adds its own complications. > > Steve > _______________________________________________ > kvmarm mailing list > kvmarm@lists.cs.columbia.edu > https://lists.cs.columbia.edu/mailman/listinfo/kvmarm >
On Wed, 14 Aug 2019 14:02:25 +0100, Alexander Graf <graf@amazon.com> wrote: > > > > On 05.08.19 15:06, Steven Price wrote: > > On 03/08/2019 19:05, Marc Zyngier wrote: > >> On Fri, 2 Aug 2019 15:50:08 +0100 > >> Steven Price <steven.price@arm.com> wrote: > >> > >> Hi Steven, > >> > >>> This series add support for paravirtualized time for arm64 guests and > >>> KVM hosts following the specification in Arm's document DEN 0057A: > >>> > >>> https://developer.arm.com/docs/den0057/a > >>> > >>> It implements support for stolen time, allowing the guest to > >>> identify time when it is forcibly not executing. > >>> > >>> It doesn't implement support for Live Physical Time (LPT) as there are > >>> some concerns about the overheads and approach in the above > >>> specification, and I expect an updated version of the specification to > >>> be released soon with just the stolen time parts. > >> > >> Thanks for posting this. > >> > >> My current concern with this series is around the fact that we allocate > >> memory from the kernel on behalf of the guest. It is the first example > >> of such thing in the ARM port, and I can't really say I'm fond of it. > >> > >> x86 seems to get away with it by having the memory allocated from > >> userspace, why I tend to like more. Yes, put_user is more > >> expensive than a straight store, but this isn't done too often either. > >> > >> What is the rational for your current approach? > > > > As I see it there are 3 approaches that can be taken here: > > > > 1. Hypervisor allocates memory and adds it to the virtual machine. This > > means that everything to do with the 'device' is encapsulated behind the > > KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the > > stolen time structure to be fast it cannot be a trapping region and has > > to be backed by real memory - in this case allocated by the host kernel. > > > > 2. Host user space allocates memory. Similar to above, but this time > > user space needs to manage the memory region as well as the usual > > KVM_CREATE_DEVICE dance. I've no objection to this, but it means > > kvmtool/QEMU needs to be much more aware of what is going on (e.g. how > > to size the memory region). > > You ideally want to get the host overhead for a VM to as little as you > can. I'm not terribly fond of the idea of reserving a full page just > because we're too afraid of having the guest donate memory. Well, reduce the amount of memory you give to the guest by one page, and allocate that page to the stolen time device. Problem solved! Seriously, if you're worried about the allocation of a single page, you should first look at how many holes we have in the vcpu structure, for example (even better, with the 8.4 NV patches applied). Just fixing that would give you that page back *per vcpu*. > > 3. Guest kernel "donates" the memory to the hypervisor for the > > structure. As far as I'm aware this is what x86 does. The problems I see > > this approach are: > > > > a) kexec becomes much more tricky - there needs to be a disabling > > mechanism for the guest to stop the hypervisor scribbling on memory > > before starting the new kernel. > > I wouldn't call "quiesce a device" much more tricky. We have to do > that for other devices as well today. And since there is no standard way of doing it, we keep inventing weird and wonderful ways of doing so -- cue the terrible GICv3 LPI situation, and all the various hacks to keep existing IOMMU mappings around across firmware/kernel handovers as well as kexec. > > > b) If there is more than one entity that is interested in the > > information (e.g. firmware and kernel) then this requires some form of > > arbitration in the guest because the hypervisor doesn't want to have to > > track an arbitrary number of regions to update. > > Why would FW care? Exactly. It doesn't care. Not caring means it doesn't know about the page the guest has allocated for stolen time, and starts using it for its own purposes. Hello, memory corruption. Same thing goes if you reboot into a non stolen time aware kernel. > > > c) Performance can suffer if the host kernel doesn't have a suitably > > aligned/sized area to use. As you say - put_user() is more expensive. > > Just define the interface to always require natural alignment when > donating a memory location? > > > The structure is updated on every return to the VM. > > If you really do suffer from put_user(), there are alternatives. You > could just map the page on the registration hcall and then leave it > pinned until the vcpu gets destroyed again. put_user() should be cheap enough. It is one of the things we tend to optimise anyway. And yes, worse case, we pin the page. > > > Of course x86 does prove the third approach can work, but I'm not sure > > which is actually better. Avoid the kexec cancellation requirements was > > the main driver of the current approach. Although many of the > > I really don't understand the problem with kexec cancellation. Worst > case, let guest FW set it up for you and propagate only the address > down via ACPI/DT. That way you can mark the respective memory as > reserved too. We already went down that road with the LPI hack. I'm not going there again if we can avoid it. And it turn out that we can. Just allocate the stolen time page as a separate memblock, give it to KVM for that purpose. Your suggestion of letting the guest firmware set something up only works if whatever you're booting after that understands it. If it doesn't, you're screwed. > But even with a Linux only mechanism, just take a look at > arch/x86/kernel/kvmclock.c. All they do to remove the map is to hook > into machine_crash_shutdown() and machine_shutdown(). I'm not going to take something that is Linux specific. It has to work for all guests, at all times, whether they know about the hypervisor service or not. M. > > > Alex > > > conversations about this were also tied up with Live Physical Time which > > adds its own complications. > > > > Steve > > _______________________________________________ > > kvmarm mailing list > > kvmarm@lists.cs.columbia.edu > > https://lists.cs.columbia.edu/mailman/listinfo/kvmarm > >
On 14.08.19 16:19, Marc Zyngier wrote: > On Wed, 14 Aug 2019 14:02:25 +0100, > Alexander Graf <graf@amazon.com> wrote: >> >> >> >> On 05.08.19 15:06, Steven Price wrote: >>> On 03/08/2019 19:05, Marc Zyngier wrote: >>>> On Fri, 2 Aug 2019 15:50:08 +0100 >>>> Steven Price <steven.price@arm.com> wrote: >>>> >>>> Hi Steven, >>>> >>>>> This series add support for paravirtualized time for arm64 guests and >>>>> KVM hosts following the specification in Arm's document DEN 0057A: >>>>> >>>>> https://developer.arm.com/docs/den0057/a >>>>> >>>>> It implements support for stolen time, allowing the guest to >>>>> identify time when it is forcibly not executing. >>>>> >>>>> It doesn't implement support for Live Physical Time (LPT) as there are >>>>> some concerns about the overheads and approach in the above >>>>> specification, and I expect an updated version of the specification to >>>>> be released soon with just the stolen time parts. >>>> >>>> Thanks for posting this. >>>> >>>> My current concern with this series is around the fact that we allocate >>>> memory from the kernel on behalf of the guest. It is the first example >>>> of such thing in the ARM port, and I can't really say I'm fond of it. >>>> >>>> x86 seems to get away with it by having the memory allocated from >>>> userspace, why I tend to like more. Yes, put_user is more >>>> expensive than a straight store, but this isn't done too often either. >>>> >>>> What is the rational for your current approach? >>> >>> As I see it there are 3 approaches that can be taken here: >>> >>> 1. Hypervisor allocates memory and adds it to the virtual machine. This >>> means that everything to do with the 'device' is encapsulated behind the >>> KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want the >>> stolen time structure to be fast it cannot be a trapping region and has >>> to be backed by real memory - in this case allocated by the host kernel. >>> >>> 2. Host user space allocates memory. Similar to above, but this time >>> user space needs to manage the memory region as well as the usual >>> KVM_CREATE_DEVICE dance. I've no objection to this, but it means >>> kvmtool/QEMU needs to be much more aware of what is going on (e.g. how >>> to size the memory region). >> >> You ideally want to get the host overhead for a VM to as little as you >> can. I'm not terribly fond of the idea of reserving a full page just >> because we're too afraid of having the guest donate memory. > > Well, reduce the amount of memory you give to the guest by one page, > and allocate that page to the stolen time device. Problem solved! > > Seriously, if you're worried about the allocation of a single page, > you should first look at how many holes we have in the vcpu structure, > for example (even better, with the 8.4 NV patches applied). Just > fixing that would give you that page back *per vcpu*. I'm worried about additional memory slots, about fragmenting the cachable guest memory regions, about avoidable HV taxes. I think we need to distinguish here between the KVM implementation and the hypervisor/guest interface. Just because in KVM we can save overhead today doesn't mean that the HV interface should be built around the assumption that "memory is free". > >>> 3. Guest kernel "donates" the memory to the hypervisor for the >>> structure. As far as I'm aware this is what x86 does. The problems I see >>> this approach are: >>> >>> a) kexec becomes much more tricky - there needs to be a disabling >>> mechanism for the guest to stop the hypervisor scribbling on memory >>> before starting the new kernel. >> >> I wouldn't call "quiesce a device" much more tricky. We have to do >> that for other devices as well today. > > And since there is no standard way of doing it, we keep inventing > weird and wonderful ways of doing so -- cue the terrible GICv3 LPI > situation, and all the various hacks to keep existing IOMMU mappings > around across firmware/kernel handovers as well as kexec. Well, the good news here is that we don't have to keep it around ;). > >> >>> b) If there is more than one entity that is interested in the >>> information (e.g. firmware and kernel) then this requires some form of >>> arbitration in the guest because the hypervisor doesn't want to have to >>> track an arbitrary number of regions to update. >> >> Why would FW care? > > Exactly. It doesn't care. Not caring means it doesn't know about the > page the guest has allocated for stolen time, and starts using it for > its own purposes. Hello, memory corruption. Same thing goes if you > reboot into a non stolen time aware kernel. If you reboot, you go via the vcpu reset path which clears the map, no? Same goes for FW entry. If you enter firmware that does not set up the map, you never see it. > >> >>> c) Performance can suffer if the host kernel doesn't have a suitably >>> aligned/sized area to use. As you say - put_user() is more expensive. >> >> Just define the interface to always require natural alignment when >> donating a memory location? >> >>> The structure is updated on every return to the VM. >> >> If you really do suffer from put_user(), there are alternatives. You >> could just map the page on the registration hcall and then leave it >> pinned until the vcpu gets destroyed again. > > put_user() should be cheap enough. It is one of the things we tend to > optimise anyway. And yes, worse case, we pin the page. > >> >>> Of course x86 does prove the third approach can work, but I'm not sure >>> which is actually better. Avoid the kexec cancellation requirements was >>> the main driver of the current approach. Although many of the >> >> I really don't understand the problem with kexec cancellation. Worst >> case, let guest FW set it up for you and propagate only the address >> down via ACPI/DT. That way you can mark the respective memory as >> reserved too. > > We already went down that road with the LPI hack. I'm not going there > again if we can avoid it. And it turn out that we can. Just allocate > the stolen time page as a separate memblock, give it to KVM for that > purpose. > > Your suggestion of letting the guest firmware set something up only > works if whatever you're booting after that understands it. If it > doesn't, you're screwed. Why? For UEFI, mark the region as reserved in the memory map. For DT, just mark it straight on reserved. That said, I'm not advocating for doing it in the FW. I think this can be solved really easily with a simple guest driver to enable and a vcpu reset hook to disable the map. > >> But even with a Linux only mechanism, just take a look at >> arch/x86/kernel/kvmclock.c. All they do to remove the map is to hook >> into machine_crash_shutdown() and machine_shutdown(). > > I'm not going to take something that is Linux specific. It has to work > for all guests, at all times, whether they know about the hypervisor > service or not. If they don't know about the HV service, they don't register the writer, so they don't see corruption. If they know about the HV service and they don't support kexec, they don't have to worry because a vcpu reset should also clear the map. If they do support kexec, they already have a mechanism to quiesce devices. So I don't understand how this is Linux specific? The question was Linux specific, so I answered with precedence to show that disabling on kexec is not all that hard :). Alex
On 14/08/2019 15:52, Alexander Graf wrote: > > > On 14.08.19 16:19, Marc Zyngier wrote: >> On Wed, 14 Aug 2019 14:02:25 +0100, >> Alexander Graf <graf@amazon.com> wrote: >>> >>> >>> >>> On 05.08.19 15:06, Steven Price wrote: >>>> On 03/08/2019 19:05, Marc Zyngier wrote: >>>>> On Fri, 2 Aug 2019 15:50:08 +0100 >>>>> Steven Price <steven.price@arm.com> wrote: >>>>> >>>>> Hi Steven, >>>>> >>>>>> This series add support for paravirtualized time for arm64 guests and >>>>>> KVM hosts following the specification in Arm's document DEN 0057A: >>>>>> >>>>>> https://developer.arm.com/docs/den0057/a >>>>>> >>>>>> It implements support for stolen time, allowing the guest to >>>>>> identify time when it is forcibly not executing. >>>>>> >>>>>> It doesn't implement support for Live Physical Time (LPT) as there >>>>>> are >>>>>> some concerns about the overheads and approach in the above >>>>>> specification, and I expect an updated version of the >>>>>> specification to >>>>>> be released soon with just the stolen time parts. >>>>> >>>>> Thanks for posting this. >>>>> >>>>> My current concern with this series is around the fact that we >>>>> allocate >>>>> memory from the kernel on behalf of the guest. It is the first example >>>>> of such thing in the ARM port, and I can't really say I'm fond of it. >>>>> >>>>> x86 seems to get away with it by having the memory allocated from >>>>> userspace, why I tend to like more. Yes, put_user is more >>>>> expensive than a straight store, but this isn't done too often either. >>>>> >>>>> What is the rational for your current approach? >>>> >>>> As I see it there are 3 approaches that can be taken here: >>>> >>>> 1. Hypervisor allocates memory and adds it to the virtual machine. This >>>> means that everything to do with the 'device' is encapsulated behind >>>> the >>>> KVM_CREATE_DEVICE / KVM_[GS]ET_DEVICE_ATTR ioctls. But since we want >>>> the >>>> stolen time structure to be fast it cannot be a trapping region and has >>>> to be backed by real memory - in this case allocated by the host >>>> kernel. >>>> >>>> 2. Host user space allocates memory. Similar to above, but this time >>>> user space needs to manage the memory region as well as the usual >>>> KVM_CREATE_DEVICE dance. I've no objection to this, but it means >>>> kvmtool/QEMU needs to be much more aware of what is going on (e.g. how >>>> to size the memory region). >>> >>> You ideally want to get the host overhead for a VM to as little as you >>> can. I'm not terribly fond of the idea of reserving a full page just >>> because we're too afraid of having the guest donate memory. >> >> Well, reduce the amount of memory you give to the guest by one page, >> and allocate that page to the stolen time device. Problem solved! >> >> Seriously, if you're worried about the allocation of a single page, >> you should first look at how many holes we have in the vcpu structure, >> for example (even better, with the 8.4 NV patches applied). Just >> fixing that would give you that page back *per vcpu*. > > I'm worried about additional memory slots, about fragmenting the > cachable guest memory regions, about avoidable HV taxes. > > I think we need to distinguish here between the KVM implementation and > the hypervisor/guest interface. Just because in KVM we can save overhead > today doesn't mean that the HV interface should be built around the > assumption that "memory is free". The HV interface just requires that the host provides some memory for the structures to live in. The memory can be adjacent (or even within) the normal memory of the guest. The only requirement is that the guest isn't told to use this memory for normal allocations (i.e. it should either be explicitly reserved or just not contained within the normal memory block). >> >>>> 3. Guest kernel "donates" the memory to the hypervisor for the >>>> structure. As far as I'm aware this is what x86 does. The problems I >>>> see >>>> this approach are: >>>> >>>> a) kexec becomes much more tricky - there needs to be a disabling >>>> mechanism for the guest to stop the hypervisor scribbling on memory >>>> before starting the new kernel. >>> >>> I wouldn't call "quiesce a device" much more tricky. We have to do >>> that for other devices as well today. >> >> And since there is no standard way of doing it, we keep inventing >> weird and wonderful ways of doing so -- cue the terrible GICv3 LPI >> situation, and all the various hacks to keep existing IOMMU mappings >> around across firmware/kernel handovers as well as kexec. > > Well, the good news here is that we don't have to keep it around ;). > >> >>> >>>> b) If there is more than one entity that is interested in the >>>> information (e.g. firmware and kernel) then this requires some form of >>>> arbitration in the guest because the hypervisor doesn't want to have to >>>> track an arbitrary number of regions to update. >>> >>> Why would FW care? >> >> Exactly. It doesn't care. Not caring means it doesn't know about the >> page the guest has allocated for stolen time, and starts using it for >> its own purposes. Hello, memory corruption. Same thing goes if you >> reboot into a non stolen time aware kernel. > > If you reboot, you go via the vcpu reset path which clears the map, no? > Same goes for FW entry. If you enter firmware that does not set up the > map, you never see it. Doing this per-vcpu implies you are probably going to have to allocate an entire page per vcpu. Because it's entirely possible for a guest to reset an individual vcpu. Or at the least there's some messy reference counting going on here. Having a region of memory provided by the host means the structures can be packed and there's nothing to be done in the reset path. >> >>> >>>> c) Performance can suffer if the host kernel doesn't have a suitably >>>> aligned/sized area to use. As you say - put_user() is more expensive. >>> >>> Just define the interface to always require natural alignment when >>> donating a memory location? >>> >>>> The structure is updated on every return to the VM. >>> >>> If you really do suffer from put_user(), there are alternatives. You >>> could just map the page on the registration hcall and then leave it >>> pinned until the vcpu gets destroyed again. >> >> put_user() should be cheap enough. It is one of the things we tend to >> optimise anyway. And yes, worse case, we pin the page. >> >>> >>>> Of course x86 does prove the third approach can work, but I'm not sure >>>> which is actually better. Avoid the kexec cancellation requirements was >>>> the main driver of the current approach. Although many of the >>> >>> I really don't understand the problem with kexec cancellation. Worst >>> case, let guest FW set it up for you and propagate only the address >>> down via ACPI/DT. That way you can mark the respective memory as >>> reserved too. >> >> We already went down that road with the LPI hack. I'm not going there >> again if we can avoid it. And it turn out that we can. Just allocate >> the stolen time page as a separate memblock, give it to KVM for that >> purpose. >> >> Your suggestion of letting the guest firmware set something up only >> works if whatever you're booting after that understands it. If it >> doesn't, you're screwed. > > Why? For UEFI, mark the region as reserved in the memory map. For DT, > just mark it straight on reserved. > > That said, I'm not advocating for doing it in the FW. I think this can > be solved really easily with a simple guest driver to enable and a vcpu > reset hook to disable the map. > >> >>> But even with a Linux only mechanism, just take a look at >>> arch/x86/kernel/kvmclock.c. All they do to remove the map is to hook >>> into machine_crash_shutdown() and machine_shutdown(). >> >> I'm not going to take something that is Linux specific. It has to work >> for all guests, at all times, whether they know about the hypervisor >> service or not. > > If they don't know about the HV service, they don't register the writer, > so they don't see corruption. > > If they know about the HV service and they don't support kexec, they > don't have to worry because a vcpu reset should also clear the map. > > If they do support kexec, they already have a mechanism to quiesce devices. > > So I don't understand how this is Linux specific? The question was Linux > specific, so I answered with precedence to show that disabling on kexec > is not all that hard :). My concern is more around a something like Jailhouse as a guest-hypervisor. There Linux gives up CPUs to run another OS. This hand-off of a CPU is much easier if there's just a structure in memory somewhere which doesn't move. The kexec case like you say can be handled as a device to quiesce. I don't think either scheme is unworkable, but I do think getting the host to provide the memory is easier for both guest and host. Marc had a good point that getting user space to allocate the memory is probably preferable to getting the host kernel to do so, so I'm reworking the code to do that. Steve
Hi Steven, On 2019/8/2 22:50, Steven Price wrote: > This series add support for paravirtualized time for arm64 guests and > KVM hosts following the specification in Arm's document DEN 0057A: > > https://developer.arm.com/docs/den0057/a > > It implements support for stolen time, allowing the guest to > identify time when it is forcibly not executing. > > It doesn't implement support for Live Physical Time (LPT) as there are > some concerns about the overheads and approach in the above Do you plan to pick up LPT support? As there is demand of cross-frequency migration (from older platform to newer platform). I am not clear about the overheads and approach problem here, could you please give some detail information? Maybe we can work together to solve these concerns. :-) Thanks, Keqian > specification, and I expect an updated version of the specification to > be released soon with just the stolen time parts. > > I previously posted a series including LPT (as well as stolen time): > https://lore.kernel.org/kvmarm/20181212150226.38051-1-steven.price@arm.com/ > > Patches 2, 5, 7 and 8 are cleanup patches and could be taken separately. > > Christoffer Dall (1): > KVM: arm/arm64: Factor out hypercall handling from PSCI code > > Steven Price (8): > KVM: arm64: Document PV-time interface > KVM: arm64: Implement PV_FEATURES call > KVM: arm64: Support stolen time reporting via shared structure > KVM: Allow kvm_device_ops to be const > KVM: arm64: Provide a PV_TIME device to user space > arm/arm64: Provide a wrapper for SMCCC 1.1 calls > arm/arm64: Make use of the SMCCC 1.1 wrapper > arm64: Retrieve stolen time as paravirtualized guest > > Documentation/virtual/kvm/arm/pvtime.txt | 107 +++++++++++++ > arch/arm/kvm/Makefile | 2 +- > arch/arm/kvm/handle_exit.c | 2 +- > arch/arm/mm/proc-v7-bugs.c | 13 +- > arch/arm64/include/asm/kvm_host.h | 13 +- > arch/arm64/include/asm/kvm_mmu.h | 2 + > arch/arm64/include/asm/pvclock-abi.h | 20 +++ > arch/arm64/include/uapi/asm/kvm.h | 6 + > arch/arm64/kernel/Makefile | 1 + > arch/arm64/kernel/cpu_errata.c | 80 ++++------ > arch/arm64/kernel/kvm.c | 155 ++++++++++++++++++ > arch/arm64/kvm/Kconfig | 1 + > arch/arm64/kvm/Makefile | 2 + > arch/arm64/kvm/handle_exit.c | 4 +- > include/kvm/arm_hypercalls.h | 44 ++++++ > include/kvm/arm_psci.h | 2 +- > include/linux/arm-smccc.h | 58 +++++++ > include/linux/cpuhotplug.h | 1 + > include/linux/kvm_host.h | 4 +- > include/linux/kvm_types.h | 2 + > include/uapi/linux/kvm.h | 2 + > virt/kvm/arm/arm.c | 18 +++ > virt/kvm/arm/hypercalls.c | 138 ++++++++++++++++ > virt/kvm/arm/mmu.c | 44 ++++++ > virt/kvm/arm/psci.c | 84 +--------- > virt/kvm/arm/pvtime.c | 190 +++++++++++++++++++++++ > virt/kvm/kvm_main.c | 6 +- > 27 files changed, 848 insertions(+), 153 deletions(-) > create mode 100644 Documentation/virtual/kvm/arm/pvtime.txt > create mode 100644 arch/arm64/include/asm/pvclock-abi.h > create mode 100644 arch/arm64/kernel/kvm.c > create mode 100644 include/kvm/arm_hypercalls.h > create mode 100644 virt/kvm/arm/hypercalls.c > create mode 100644 virt/kvm/arm/pvtime.c >
On 21/07/2020 04:26, zhukeqian wrote: > Hi Steven, Hi Keqian, > On 2019/8/2 22:50, Steven Price wrote: >> This series add support for paravirtualized time for arm64 guests and >> KVM hosts following the specification in Arm's document DEN 0057A: >> >> https://developer.arm.com/docs/den0057/a >> >> It implements support for stolen time, allowing the guest to >> identify time when it is forcibly not executing. >> >> It doesn't implement support for Live Physical Time (LPT) as there are >> some concerns about the overheads and approach in the above > Do you plan to pick up LPT support? As there is demand of cross-frequency migration > (from older platform to newer platform). I don't have any plans to pick up the LPT support at the moment - feel free to pick it up! ;) > I am not clear about the overheads and approach problem here, could you please > give some detail information? Maybe we can work together to solve these concerns. :-) Fundamentally the issue here is that LPT only solves one small part of migration between different hosts. To successfully migrate between hosts with different CPU implementations it is also necessary to be able to virtualise various ID registers (e.g. MIDR_EL1, REVIDR_EL1, AIDR_EL1) which we have no support for currently. The problem with just virtualising the registers is how you handle errata. The guest will currently use those (and other) ID registers to decide whether to enable specific errata workarounds. But what errata should be enabled for a guest which might migrate to another host? What we ideally need is a mechanism to communicate to the guest what workarounds are required to successfully run on any of the hosts that the guest may be migrated to. You may also have the situation where the workarounds required for two hosts are mutually incompatible - something needs to understand this and do the "right thing" (most likely just reject this situation, i.e. prevent the migration). There are various options here: e.g. a para-virtualised interface to describe the workarounds (but this is hard to do in an OS-agnostic way), or virtual-ID registers describing an idealised environment where no workarounds are required (and only hosts that have no errata affecting a guest would be able to provide this). Given the above complexity and the fact that Armv8.6-A standardises the frequency to 1GHz this didn't seem worth continuing with. So LPT was dropped from the spec and patches to avoid holding up the stolen time support. However, if you have a use case which doesn't require such a generic migration (e.g. perhaps old and new platforms are based on the same IP) then it might be worth looking at bring this back. But to make the problem solvable it either needs to be restricted to platforms which are substantially the same (so the errata list will be identical), or there's work to be done in preparation to deal with migrating a guest successfully between hosts with potentially different errata requirements. Can you share more details about the hosts that you are interested in migrating between? Thanks, Steve
Hi Steven, On 2020/7/27 18:48, Steven Price wrote: > On 21/07/2020 04:26, zhukeqian wrote: >> Hi Steven, > > Hi Keqian, > >> On 2019/8/2 22:50, Steven Price wrote: >>> This series add support for paravirtualized time for arm64 guests and >>> KVM hosts following the specification in Arm's document DEN 0057A: >>> >>> https://developer.arm.com/docs/den0057/a >>> >>> It implements support for stolen time, allowing the guest to >>> identify time when it is forcibly not executing. >>> >>> It doesn't implement support for Live Physical Time (LPT) as there are >>> some concerns about the overheads and approach in the above >> Do you plan to pick up LPT support? As there is demand of cross-frequency migration >> (from older platform to newer platform). > > I don't have any plans to pick up the LPT support at the moment - feel free to pick it up! ;) > >> I am not clear about the overheads and approach problem here, could you please >> give some detail information? Maybe we can work together to solve these concerns. :-) > > Fundamentally the issue here is that LPT only solves one small part of migration between different hosts. To successfully migrate between hosts with different CPU implementations it is also necessary to be able to virtualise various ID registers (e.g. MIDR_EL1, REVIDR_EL1, AIDR_EL1) which we have no support for currently. > Yeah, currently we are trying to do both timer freq virtualization and CPU feature virtualization. > The problem with just virtualising the registers is how you handle errata. The guest will currently use those (and other) ID registers to decide whether to enable specific errata workarounds. But what errata should be enabled for a guest which might migrate to another host? > Thanks for pointing this out. I think the most important thing is that we should introduce a concept named CPU baseline which represents a standard platform. If we bring up a guest with a specific CPU baseline, then this guest can only run on a platform that is compatible with this CPU baseline. So "baseline" and "compatible" are the key point to promise successful cross-platform migration. > What we ideally need is a mechanism to communicate to the guest what workarounds are required to successfully run on any of the hosts that the guest may be migrated to. You may also have the situation where the workarounds required for two hosts are mutually incompatible - something needs to understand this and do the "right thing" (most likely just reject this situation, i.e. prevent the migration). > > There are various options here: e.g. a para-virtualised interface to describe the workarounds (but this is hard to do in an OS-agnostic way), or virtual-ID registers describing an idealised environment where no workarounds are required (and only hosts that have no errata affecting a guest would be able to provide this). > My idea is similar with the "idealised environment", but errata workaround still exists. We do not provide para-virtualised interface, and migration is restricted between platforms that are compatible with baseline. Baseline should has two aspects: CPU feature and errata. These platforms that are compatible with a specific baseline should have the corresponding CPU feature and errata. > Given the above complexity and the fact that Armv8.6-A standardises the frequency to 1GHz this didn't seem worth continuing with. So LPT was dropped from the spec and patches to avoid holding up the stolen time support. > > However, if you have a use case which doesn't require such a generic migration (e.g. perhaps old and new platforms are based on the same IP) then it might be worth looking at bring this back. But to make the problem solvable it either needs to be restricted to platforms which are substantially the same (so the errata list will be identical), or there's work to be done in preparation to deal with migrating a guest successfully between hosts with potentially different errata requirements. > > Can you share more details about the hosts that you are interested in migrating between? Here we have new platform with 1GHz timer, and old platform is 100MHZ, so we want to solve the cross-platform migration firstly. Thanks, Keqian > > Thanks, > > Steve > . >