kernel/git/stable/linux.git/arch/x86/hyperv, branch linux-5.7.y

x86/hyperv: Properly suspend/resume reenlightenment notifications

2020-05-13T15:02:03+00:00

Errors during hibernation with reenlightenment notifications enabled were reported: [ 51.730435] PM: hibernation entry [ 51.737435] PM: Syncing filesystems ... ... [ 54.102216] Disabling non-boot CPUs ... [ 54.106633] smpboot: CPU 1 is now offline [ 54.110006] unchecked MSR access error: WRMSR to 0x40000106 (tried to write 0x47c72780000100ee) at rIP: 0xffffffff90062f24 native_write_msr+0x4/0x20) [ 54.110006] Call Trace: [ 54.110006] hv_cpu_die+0xd9/0xf0 ... Normally, hv_cpu_die() just reassigns reenlightenment notifications to some other CPU when the CPU receiving them goes offline. Upon hibernation, there is no other CPU which is still online so cpumask_any_but(cpu_online_mask) returns >= nr_cpu_ids and using it as hv_vp_index index is incorrect. Disable the feature when cpumask_any_but() fails. Also, as we now disable reenlightenment notifications upon hibernation we need to restore them on resume. Check if hv_reenlightenment_cb was previously set and restore from hv_resume(). Signed-off-by: Vitaly Kuznetsov Reviewed-by: Dexuan Cui Reviewed-by: Tianyu Lan Link: https://lore.kernel.org/r/20200512160153.134467-1-vkuznets@redhat.com Signed-off-by: Wei Liu

x86/hyperv: Suspend/resume the VP assist page for hibernation

2020-04-21T09:03:19+00:00

Unlike the other CPUs, CPU0 is never offlined during hibernation, so in the resume path, the "new" kernel's VP assist page is not suspended (i.e. not disabled), and later when we jump to the "old" kernel, the page is not properly re-enabled for CPU0 with the allocated page from the old kernel. So far, the VP assist page is used by hv_apic_eoi_write(), and is also used in the case of nested virtualization (running KVM atop Hyper-V). For hv_apic_eoi_write(), when the page is not properly re-enabled, hvp->apic_assist is always 0, so the HV_X64_MSR_EOI MSR is always written. This is not ideal with respect to performance, but Hyper-V can still correctly handle this according to the Hyper-V spec; nevertheless, Linux still must update the Hyper-V hypervisor with the correct VP assist page to prevent Hyper-V from writing to the stale page, which causes guest memory corruption and consequently may have caused the hangs and triple faults seen during non-boot CPUs resume. Fix the issue by calling hv_cpu_die()/hv_cpu_init() in the syscore ops. Without the fix, hibernation can fail at a rate of 1/300 ~ 1/500. With the fix, hibernation can pass a long-haul test of 2000 runs. In the case of nested virtualization, disabling/reenabling the assist page upon hibernation may be unsafe if there are active L2 guests. It looks KVM should be enhanced to abort the hibernation request if there is any active L2 guest. Fixes: 05bd330a7fd8 ("x86/hyperv: Suspend/resume the hypercall page for hibernation") Cc: stable@vger.kernel.org Signed-off-by: Dexuan Cui Link: https://lore.kernel.org/r/1587437171-2472-1-git-send-email-decui@microsoft.com Signed-off-by: Wei Liu

x86/Hyper-V: Report crash data in die() when panic_on_oops is set

2020-04-11T16:19:07+00:00

When oops happens with panic_on_oops unset, the oops thread is killed by die() and system continues to run. In such case, guest should not report crash register data to host since system still runs. Check panic_on_oops and return directly in hyperv_report_panic() when the function is called in the die() and panic_on_oops is unset. Fix it. Fixes: 7ed4325a44ea ("Drivers: hv: vmbus: Make panic reporting to be more useful") Signed-off-by: Tianyu Lan Reviewed-by: Michael Kelley Link: https://lore.kernel.org/r/20200406155331.2105-7-Tianyu.Lan@microsoft.com Signed-off-by: Wei Liu

x86/hyperv: Suspend/resume the hypercall page for hibernation

2020-02-01T08:41:16+00:00

For hibernation the hypercall page must be disabled before the hibernation image is created so that subsequent hypercall operations fail safely. On resume the hypercall page has to be restored and reenabled to ensure proper operation of the resumed kernel. Implement the necessary suspend/resume callbacks. [ tglx: Decrypted changelog ] Signed-off-by: Dexuan Cui Signed-off-by: Thomas Gleixner Reviewed-by: Michael Kelley Link: https://lore.kernel.org/r/1578350559-130275-1-git-send-email-decui@microsoft.com

Merge tag 'hyperv-next-signed' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux

2019-11-30T22:50:51+00:00

Pull Hyper-V updates from Sasha Levin: - support for new VMBus protocols (Andrea Parri) - hibernation support (Dexuan Cui) - latency testing framework (Branden Bonaby) - decoupling Hyper-V page size from guest page size (Himadri Pandya) * tag 'hyperv-next-signed' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux: (22 commits) Drivers: hv: vmbus: Fix crash handler reset of Hyper-V synic drivers/hv: Replace binary semaphore with mutex drivers: iommu: hyperv: Make HYPERV_IOMMU only available on x86 HID: hyperv: Add the support of hibernation hv_balloon: Add the support of hibernation x86/hyperv: Implement hv_is_hibernation_supported() Drivers: hv: balloon: Remove dependencies on guest page size Drivers: hv: vmbus: Remove dependencies on guest page size x86: hv: Add function to allocate zeroed page for Hyper-V Drivers: hv: util: Specify ring buffer size using Hyper-V page size Drivers: hv: Specify receive buffer size using Hyper-V page size tools: hv: add vmbus testing tool drivers: hv: vmbus: Introduce latency testing video: hyperv: hyperv_fb: Support deferred IO for Hyper-V frame buffer driver video: hyperv: hyperv_fb: Obtain screen resolution from Hyper-V host hv_netvsc: Add the support of hibernation hv_sock: Add the support of hibernation video: hyperv_fb: Add the support of hibernation scsi: storvsc: Add the support of hibernation Drivers: hv: vmbus: Add module parameter to cap the VMBus version ...

x86/hyperv: Implement hv_is_hibernation_supported()

2019-11-22T01:10:45+00:00

The API will be used by the hv_balloon and hv_vmbus drivers. Balloon up/down and hot-add of memory must not be active if the user wants the Linux VM to support hibernation, because they are incompatible with hibernation according to Hyper-V team, e.g. upon suspend the balloon VSP doesn't save any info about the ballooned-out pages (if any); so, after Linux resumes, Linux balloon VSC expects that the VSP will return the pages if Linux is under memory pressure, but the VSP will never do that, since the VSP thinks it never stole the pages from the VM. So, if the user wants Linux VM to support hibernation, Linux must forbid balloon up/down and hot-add, and the only functionality of the balloon VSC driver is reporting the VM's memory pressure to the host. Ideally, when Linux detects that the user wants it to support hibernation, the balloon VSC should tell the VSP that it does not support ballooning and hot-add. However, the current version of the VSP requires the VSC should support these capabilities, otherwise the capability negotiation fails and the VSC can not load at all, so with the later changes to the VSC driver, Linux VM still reports to the VSP that the VSC supports these capabilities, but the VSC ignores the VSP's requests of balloon up/down and hot add, and reports an error to the VSP, when applicable. BTW, in the future the balloon VSP driver will allow the VSC to not support the capabilities of balloon up/down and hot add. The ACPI S4 state is not a must for hibernation to work, because Linux is able to hibernate as long as the system can shut down. However in practice we decide to artificially use the presence of the virtual ACPI S4 state as an indicator of the user's intent of using hibernation, because Linux VM must find a way to know if the user wants to use the hibernation feature or not. By default, Hyper-V does not enable the virtual ACPI S4 state; on recent Hyper-V hosts (e.g. RS5, 19H1), the administrator is able to enable the state for a VM by WMI commands. Once all the vmbus and VSC patches for the hibernation feature are accepted, an extra patch will be submitted to forbid hibernation if the virtual ACPI S4 state is absent, i.e. hv_is_hibernation_supported() is false. Signed-off-by: Dexuan Cui Reviewed-by: Michael Kelley Acked-by: Thomas Gleixner Signed-off-by: Sasha Levin

x86: hv: Add function to allocate zeroed page for Hyper-V

2019-11-22T01:10:45+00:00

Hyper-V assumes page size to be 4K. While this assumption holds true on x86 architecture, it might not be true for ARM64 architecture. Hence define hyper-v specific function to allocate a zeroed page which can have a different implementation on ARM64 architecture to handle the conflict between hyper-v's assumed page size and actual guest page size. Signed-off-by: Himadri Pandya Reviewed-by: Michael Kelley Signed-off-by: Sasha Levin

x86/hyperv: Initialize clockevents earlier in CPU onlining

2019-11-15T09:33:49+00:00

Hyper-V has historically initialized stimer-based clockevents late in the process of onlining a CPU because clockevents depend on stimer interrupts. In the original Hyper-V design, stimer interrupts generate a VMbus message, so the VMbus machinery must be running first, and VMbus can't be initialized until relatively late. On x86/64, LAPIC timer based clockevents are used during early initialization before VMbus and stimer-based clockevents are ready, and again during CPU offlining after the stimer clockevents have been shut down. Unfortunately, this design creates problems when offlining CPUs for hibernation or other purposes. stimer-based clockevents are shut down relatively early in the offlining process, so clockevents_unbind_device() must be used to fallback to the LAPIC-based clockevents for the remainder of the offlining process. Furthermore, the late initialization and early shutdown of stimer-based clockevents doesn't work well on ARM64 since there is no other timer like the LAPIC to fallback to. So CPU onlining and offlining doesn't work properly. Fix this by recognizing that stimer Direct Mode is the normal path for newer versions of Hyper-V on x86/64, and the only path on other architectures. With stimer Direct Mode, stimer interrupts don't require any VMbus machinery. stimer clockevents can be initialized and shut down consistent with how it is done for other clockevent devices. While the old VMbus-based stimer interrupts must still be supported for backward compatibility on x86, that mode of operation can be treated as legacy. So add a new Hyper-V stimer entry in the CPU hotplug state list, and use that new state when in Direct Mode. Update the Hyper-V clocksource driver to allocate and initialize stimer clockevents earlier during boot. Update Hyper-V initialization and the VMbus driver to use this new design. As a result, the LAPIC timer is no longer used during boot or CPU onlining/offlining and clockevents_unbind_device() is not called. But retain the old design as a legacy implementation for older versions of Hyper-V that don't support Direct Mode. Signed-off-by: Michael Kelley Signed-off-by: Thomas Gleixner Tested-by: Dexuan Cui Reviewed-by: Dexuan Cui Link: https://lkml.kernel.org/r/1573607467-9456-1-git-send-email-mikelley@microsoft.com

Merge branch 'linus' into x86/hyperv

2019-11-15T09:30:50+00:00

Pick up upstream fixes to avoid conflicts.

x86/hyperv: Micro-optimize send_ipi_one()

2019-11-12T10:44:20+00:00

When sending an IPI to a single CPU there is no need to deal with cpumasks. With 2 CPU guest on WS2019 a minor (like 3%, 8043 -> 7761 CPU cycles) improvement with smp_call_function_single() loop benchmark can be seeb. The optimization, however, is tiny and straitforward. Also, send_ipi_one() is important for PV spinlock kick. Switching to the regular APIC IPI send for CPU > 64 case does not make sense as it is twice as expesive (12650 CPU cycles for __send_ipi_mask_ex() call, 26000 for orig_apic.send_IPI(cpu, vector)). Signed-off-by: Vitaly Kuznetsov Signed-off-by: Thomas Gleixner Reviewed-by: Michael Kelley Reviewed-by: Roman Kagan Link: https://lkml.kernel.org/r/20191027151938.7296-1-vkuznets@redhat.com