Skip to content
  1. Feb 09, 2016
  3. Jan 16, 2016
    • Dan Williams's avatar
      kvm: rename pfn_t to kvm_pfn_t · ba049e93
      Dan Williams authored 
      To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace).  This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o.  It allows userspace to coordinate
DMA/RDMA from/to persistent memory.

The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver.  The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.

The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag.  Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.

Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array.  Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory.  The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.

This patch (of 18):

The core has developed a need for a "pfn_t" type [1].  Move the existing
pfn_t in KVM to kvm_pfn_t [2].

[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html



Signed-off-by: default avatarDan Williams <dan.j.williams@intel.com>
Acked-by: default avatarChristoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: default avatarAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@linux-foundation.org>
      ba049e93
  5. Jan 12, 2016
    • Huaitong Han's avatar
      kvm: x86: Fix vmwrite to SECONDARY_VM_EXEC_CONTROL · 45bdbcfd
      Huaitong Han authored 
      

vmx_cpuid_tries to update SECONDARY_VM_EXEC_CONTROL in the VMCS, but
it will cause a vmwrite error on older CPUs because the code does not
check for the presence of CPU_BASED_ACTIVATE_SECONDARY_CONTROLS.

This will get rid of the following trace on e.g. Core2 6600:

vmwrite error: reg 401e value 10 (err 12)
Call Trace:
[<ffffffff8116e2b9>] dump_stack+0x40/0x57
[<ffffffffa020b88d>] vmx_cpuid_update+0x5d/0x150 [kvm_intel]
[<ffffffffa01d8fdc>] kvm_vcpu_ioctl_set_cpuid2+0x4c/0x70 [kvm]
[<ffffffffa01b8363>] kvm_arch_vcpu_ioctl+0x903/0xfa0 [kvm]

Fixes: feda805f
Cc: stable@vger.kernel.org
Reported-by: default avatarZdenek Kaspar <zkaspar82@gmail.com>
Signed-off-by: default avatarHuaitong Han <huaitong.han@intel.com>
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      45bdbcfd
  7. Jan 08, 2016
  9. Jan 07, 2016
  11. Dec 23, 2015
  13. Dec 22, 2015
  15. Dec 18, 2015
  17. Dec 16, 2015
    • Andrey Smetanin's avatar
      kvm/x86: Remove Hyper-V SynIC timer stopping · 481d2bcc
      Andrey Smetanin authored 
      

It's possible that guest send us Hyper-V EOM at the middle
of Hyper-V SynIC timer running, so we start processing of Hyper-V
SynIC timers in vcpu context and stop the Hyper-V SynIC timer
unconditionally:

    host                                       guest
    ------------------------------------------------------------------------------
                                           start periodic stimer
    start periodic timer
    timer expires after 15ms
    send expiration message into guest
    restart periodic timer
    timer expires again after 15 ms
    msg slot is still not cleared so
    setup ->msg_pending
(1) restart periodic timer
                                           process timer msg and clear slot
                                           ->msg_pending was set:
                                               send EOM into host
    received EOM
      kvm_make_request(KVM_REQ_HV_STIMER)

    kvm_hv_process_stimers():
        ...
        stimer_stop()
        if (time_now >= stimer->exp_time)
                stimer_expiration(stimer);

Because the timer was rearmed at (1), time_now < stimer->exp_time
and stimer_expiration is not called.  The timer then never fires.

The patch fixes such situation by not stopping Hyper-V SynIC timer
at all, because it's safe to restart it without stop in vcpu context
and timer callback always returns HRTIMER_NORESTART.

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      481d2bcc
    • Paolo Bonzini's avatar
      KVM: vmx: detect mismatched size in VMCS read/write · 8a86aea9
      Paolo Bonzini authored 
      

Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
---
	I am sending this as RFC because the error messages it produces are
	very ugly.  Because of inlining, the original line is lost.  The
	alternative is to change vmcs_read/write/checkXX into macros, but
	then you need to have a single huge BUILD_BUG_ON or BUILD_BUG_ON_MSG
	because multiple BUILD_BUG_ON* with the same __LINE__ are not
	supported well.
      8a86aea9
    • Paolo Bonzini's avatar
      KVM: VMX: fix read/write sizes of VMCS fields in dump_vmcs · 845c5b40
      Paolo Bonzini authored 
      

This was not printing the high parts of several 64-bit fields on
32-bit kernels.  Separate from the previous one to make the patches
easier to review.

Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      845c5b40
    • Paolo Bonzini's avatar
      KVM: VMX: fix read/write sizes of VMCS fields · f3531054
      Paolo Bonzini authored 
      

In theory this should have broken EPT on 32-bit kernels (due to
reading the high part of natural-width field GUEST_CR3).  Not sure
if no one noticed or the processor behaves differently from the
documentation.

Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      f3531054
    • Li RongQing's avatar
      KVM: VMX: fix the writing POSTED_INTR_NV · 0bcf261c
      Li RongQing authored 
      

POSTED_INTR_NV is 16bit, should not use 64bit write function

[ 5311.676074] vmwrite error: reg 3 value 0 (err 12)
  [ 5311.680001] CPU: 49 PID: 4240 Comm: qemu-system-i38 Tainted: G I 4.1.13-WR8.0.0.0_standard #1
  [ 5311.689343] Hardware name: Intel Corporation S2600WT2/S2600WT2, BIOS SE5C610.86B.01.01.0008.021120151325 02/11/2015
  [ 5311.699550] 00000000 00000000 e69a7e1c c1950de1 00000000 e69a7e38 fafcff45 fafebd24
  [ 5311.706924] 00000003 00000000 0000000c b6a06dfa e69a7e40 fafcff79 e69a7eb0 fafd5f57
  [ 5311.714296] e69a7ec0 c1080600 00000000 00000001 c0e18018 000001be 00000000 00000b43
  [ 5311.721651] Call Trace:
  [ 5311.722942] [<c1950de1>] dump_stack+0x4b/0x75
  [ 5311.726467] [<fafcff45>] vmwrite_error+0x35/0x40 [kvm_intel]
  [ 5311.731444] [<fafcff79>] vmcs_writel+0x29/0x30 [kvm_intel]
  [ 5311.736228] [<fafd5f57>] vmx_create_vcpu+0x337/0xb90 [kvm_intel]
  [ 5311.741600] [<c1080600>] ? dequeue_task_fair+0x2e0/0xf60
  [ 5311.746197] [<faf3b9ca>] kvm_arch_vcpu_create+0x3a/0x70 [kvm]
  [ 5311.751278] [<faf29e9d>] kvm_vm_ioctl+0x14d/0x640 [kvm]
  [ 5311.755771] [<c1129d44>] ? free_pages_prepare+0x1a4/0x2d0
  [ 5311.760455] [<c13e2842>] ? debug_smp_processor_id+0x12/0x20
  [ 5311.765333] [<c10793be>] ? sched_move_task+0xbe/0x170
  [ 5311.769621] [<c11752b3>] ? kmem_cache_free+0x213/0x230
  [ 5311.774016] [<faf29d50>] ? kvm_set_memory_region+0x60/0x60 [kvm]
  [ 5311.779379] [<c1199fa2>] do_vfs_ioctl+0x2e2/0x500
  [ 5311.783285] [<c11752b3>] ? kmem_cache_free+0x213/0x230
  [ 5311.787677] [<c104dc73>] ? __mmdrop+0x63/0xd0
  [ 5311.791196] [<c104dc73>] ? __mmdrop+0x63/0xd0
  [ 5311.794712] [<c104dc73>] ? __mmdrop+0x63/0xd0
  [ 5311.798234] [<c11a2ed7>] ? __fget+0x57/0x90
  [ 5311.801559] [<c11a2f72>] ? __fget_light+0x22/0x50
  [ 5311.805464] [<c119a240>] SyS_ioctl+0x80/0x90
  [ 5311.808885] [<c1957d30>] sysenter_do_call+0x12/0x12
  [ 5312.059280] kvm: zapping shadow pages for mmio generation wraparound
  [ 5313.678415] kvm [4231]: vcpu0 disabled perfctr wrmsr: 0xc2 data 0xffff
  [ 5313.726518] kvm [4231]: vcpu0 unhandled rdmsr: 0x570

Signed-off-by: default avatarLi RongQing <roy.qing.li@gmail.com>
Cc: Yang Zhang <yang.z.zhang@Intel.com>
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      0bcf261c
    • Andrey Smetanin's avatar
      kvm/x86: Hyper-V SynIC timers · 1f4b34f8
      Andrey Smetanin authored 
      

Per Hyper-V specification (and as required by Hyper-V-aware guests),
SynIC provides 4 per-vCPU timers.  Each timer is programmed via a pair
of MSRs, and signals expiration by delivering a special format message
to the configured SynIC message slot and triggering the corresponding
synthetic interrupt.

Note: as implemented by this patch, all periodic timers are "lazy"
(i.e. if the vCPU wasn't scheduled for more than the timer period the
timer events are lost), regardless of the corresponding configuration
MSR.  If deemed necessary, the "catch up" mode (the timer period is
shortened until the timer catches up) will be implemented later.

Changes v2:
* Use remainder to calculate periodic timer expiration time

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: default avatarRoman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Vitaly Kuznetsov <vkuznets@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      1f4b34f8
    • Andrey Smetanin's avatar
      kvm/x86: Hyper-V SynIC message slot pending clearing at SINT ack · 765eaa0f
      Andrey Smetanin authored 
      

The SynIC message protocol mandates that the message slot is claimed
by atomically setting message type to something other than HVMSG_NONE.
If another message is to be delivered while the slot is still busy,
message pending flag is asserted to indicate to the guest that the
hypervisor wants to be notified when the slot is released.

To make sure the protocol works regardless of where the message
sources are (kernel or userspace), clear the pending flag on SINT ACK
notification, and let the message sources compete for the slot again.

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: default avatarRoman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Vitaly Kuznetsov <vkuznets@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      765eaa0f
    • Andrey Smetanin's avatar
      kvm/x86: Hyper-V internal helper to read MSR HV_X64_MSR_TIME_REF_COUNT · 93bf4172
      Andrey Smetanin authored 
      

This helper will be used also in Hyper-V SynIC timers implementation.

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: default avatarRoman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Vitaly Kuznetsov <vkuznets@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      93bf4172
    • Andrey Smetanin's avatar
      kvm/x86: Added Hyper-V vcpu_to_hv_vcpu()/hv_vcpu_to_vcpu() helpers · 0ae80384
      Andrey Smetanin authored 
      

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: default avatarRoman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Vitaly Kuznetsov <vkuznets@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      0ae80384
    • Andrey Smetanin's avatar
      kvm/x86: Rearrange func's declarations inside Hyper-V header · e18eaeed
      Andrey Smetanin authored 
      

This rearrangement places functions declarations together
according to their functionality, so future additions
will be simplier.

Signed-off-by: default avatarAndrey Smetanin <asmetanin@virtuozzo.com>
Reviewed-by: default avatarRoman Kagan <rkagan@virtuozzo.com>
CC: Gleb Natapov <gleb@kernel.org>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: "K. Y. Srinivasan" <kys@microsoft.com>
CC: Haiyang Zhang <haiyangz@microsoft.com>
CC: Vitaly Kuznetsov <vkuznets@redhat.com>
CC: Roman Kagan <rkagan@virtuozzo.com>
CC: Denis V. Lunev <den@openvz.org>
CC: qemu-devel@nongnu.org
Signed-off-by: default avatarPaolo Bonzini <pbonzini@redhat.com>
      e18eaeed
  19. Dec 14, 2015
  21. Dec 11, 2015