From 164bf7e56c5a73f2f819c39ba7e0f20e0f97dc7b Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Tue, 5 Feb 2019 13:01:18 -0800 Subject: KVM: Move the memslot update in-progress flag to bit 63 ...now that KVM won't explode by moving it out of bit 0. Using bit 63 eliminates the need to jump over bit 0, e.g. when calculating a new memslots generation or when propagating the memslots generation to an MMIO spte. Signed-off-by: Sean Christopherson Signed-off-by: Paolo Bonzini --- Documentation/virtual/kvm/mmu.txt | 13 ++++++++----- 1 file changed, 8 insertions(+), 5 deletions(-) (limited to 'Documentation/virtual/kvm') diff --git a/Documentation/virtual/kvm/mmu.txt b/Documentation/virtual/kvm/mmu.txt index e507a9e0421e..367a952f50ab 100644 --- a/Documentation/virtual/kvm/mmu.txt +++ b/Documentation/virtual/kvm/mmu.txt @@ -452,13 +452,16 @@ stored into the MMIO spte. Thus, the MMIO spte might be created based on out-of-date information, but with an up-to-date generation number. To avoid this, the generation number is incremented again after synchronize_srcu -returns; thus, the low bit of kvm_memslots(kvm)->generation is only 1 during a +returns; thus, bit 63 of kvm_memslots(kvm)->generation set to 1 only during a memslot update, while some SRCU readers might be using the old copy. We do not want to use an MMIO sptes created with an odd generation number, and we can do -this without losing a bit in the MMIO spte. The low bit of the generation -is not stored in MMIO spte, and presumed zero when it is extracted out of the -spte. If KVM is unlucky and creates an MMIO spte while the low bit is 1, -the next access to the spte will always be a cache miss. +this without losing a bit in the MMIO spte. The "update in-progress" bit of the +generation is not stored in MMIO spte, and is so is implicitly zero when the +generation is extracted out of the spte. If KVM is unlucky and creates an MMIO +spte while an update is in-progress, the next access to the spte will always be +a cache miss. For example, a subsequent access during the update window will +miss due to the in-progress flag diverging, while an access after the update +window closes will have a higher generation number (as compared to the spte). Further reading -- cgit v1.2.3 From a592a3b8fc62af25a6e76aebde97a5d5f6f13e0f Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Tue, 5 Feb 2019 13:01:22 -0800 Subject: Revert "KVM: MMU: document fast invalidate all pages" Remove x86 KVM's fast invalidate mechanism, i.e. revert all patches from the original series[1]. Though not explicitly stated, for all intents and purposes the fast invalidate mechanism was added to speed up the scenario where removing a memslot, e.g. as part of accessing reading PCI ROM, caused KVM to flush all shadow entries[1]. Now that the memslot case flushes only shadow entries belonging to the memslot, i.e. doesn't use the fast invalidate mechanism, the only remaining usage of the mechanism are when the VM is being destroyed and when the MMIO generation rolls over. When a VM is being destroyed, either there are no active vcpus, i.e. there's no lock contention, or the VM has ungracefully terminated, in which case we want to reclaim its pages as quickly as possible, i.e. not release the MMU lock if there are still CPUs executing in the VM. The MMIO generation scenario is almost literally a one-in-a-million occurrence, i.e. is not a performance sensitive scenario. Given that lock-breaking is not desirable (VM teardown) or irrelevant (MMIO generation overflow), remove the fast invalidate mechanism to simplify the code (a small amount) and to discourage future code from zapping all pages as using such a big hammer should be a last restort. This reverts commit f6f8adeef542a18b1cb26a0b772c9781a10bb477. [1] https://lkml.kernel.org/r/1369960590-14138-1-git-send-email-xiaoguangrong@linux.vnet.ibm.com Cc: Xiao Guangrong Signed-off-by: Sean Christopherson Signed-off-by: Paolo Bonzini --- Documentation/virtual/kvm/mmu.txt | 28 +--------------------------- arch/x86/include/asm/kvm_host.h | 3 --- 2 files changed, 1 insertion(+), 30 deletions(-) (limited to 'Documentation/virtual/kvm') diff --git a/Documentation/virtual/kvm/mmu.txt b/Documentation/virtual/kvm/mmu.txt index 367a952f50ab..f365102c80f5 100644 --- a/Documentation/virtual/kvm/mmu.txt +++ b/Documentation/virtual/kvm/mmu.txt @@ -224,10 +224,6 @@ Shadow pages contain the following information: A bitmap indicating which sptes in spt point (directly or indirectly) at pages that may be unsynchronized. Used to quickly locate all unsychronized pages reachable from a given page. - mmu_valid_gen: - Generation number of the page. It is compared with kvm->arch.mmu_valid_gen - during hash table lookup, and used to skip invalidated shadow pages (see - "Zapping all pages" below.) clear_spte_count: Only present on 32-bit hosts, where a 64-bit spte cannot be written atomically. The reader uses this while running out of the MMU lock @@ -402,27 +398,6 @@ causes its disallow_lpage to be incremented, thus preventing instantiation of a large spte. The frames at the end of an unaligned memory slot have artificially inflated ->disallow_lpages so they can never be instantiated. -Zapping all pages (page generation count) -========================================= - -For the large memory guests, walking and zapping all pages is really slow -(because there are a lot of pages), and also blocks memory accesses of -all VCPUs because it needs to hold the MMU lock. - -To make it be more scalable, kvm maintains a global generation number -which is stored in kvm->arch.mmu_valid_gen. Every shadow page stores -the current global generation-number into sp->mmu_valid_gen when it -is created. Pages with a mismatching generation number are "obsolete". - -When KVM need zap all shadow pages sptes, it just simply increases the global -generation-number then reload root shadow pages on all vcpus. As the VCPUs -create new shadow page tables, the old pages are not used because of the -mismatching generation number. - -KVM then walks through all pages and zaps obsolete pages. While the zap -operation needs to take the MMU lock, the lock can be released periodically -so that the VCPUs can make progress. - Fast invalidation of MMIO sptes =============================== @@ -435,8 +410,7 @@ shadow pages, and is made more scalable with a similar technique. MMIO sptes have a few spare bits, which are used to store a generation number. The global generation number is stored in kvm_memslots(kvm)->generation, and increased whenever guest memory info -changes. This generation number is distinct from the one described in -the previous section. +changes. When KVM finds an MMIO spte, it checks the generation number of the spte. If the generation number of the spte does not equal the global generation diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index c4758e1a8843..69daa57b08a7 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -333,10 +333,7 @@ struct kvm_mmu_page { int root_count; /* Currently serving as active root */ unsigned int unsync_children; struct kvm_rmap_head parent_ptes; /* rmap pointers to parent sptes */ - - /* The page is obsolete if mmu_valid_gen != kvm->arch.mmu_valid_gen. */ unsigned long mmu_valid_gen; - DECLARE_BITMAP(unsync_child_bitmap, 512); #ifdef CONFIG_X86_32 -- cgit v1.2.3 From 49113d360bdeb4dd916fb6bffbcc3e157422b6fd Mon Sep 17 00:00:00 2001 From: Nir Weiner Date: Sun, 27 Jan 2019 12:17:15 +0200 Subject: KVM: Expose the initial start value in grow_halt_poll_ns() as a module parameter The hard-coded value 10000 in grow_halt_poll_ns() stands for the initial start value when raising up vcpu->halt_poll_ns. It actually sets the first timeout to the first polling session. This value has significant effect on how tolerant we are to outliers. On the standard case, higher value is better - we will spend more time in the polling busyloop, handle events/interrupts faster and result in better performance. But on outliers it puts us in a busy loop that does nothing. Even if the shrink factor is zero, we will still waste time on the first iteration. The optimal value changes between different workloads. It depends on outliers rate and polling sessions length. As this value has significant effect on the dynamic halt-polling algorithm, it should be configurable and exposed. Reviewed-by: Boris Ostrovsky Reviewed-by: Liran Alon Signed-off-by: Nir Weiner Signed-off-by: Paolo Bonzini --- Documentation/virtual/kvm/halt-polling.txt | 37 +++++++++++++++++++----------- arch/powerpc/kvm/book3s_hv.c | 3 +-- include/linux/kvm_host.h | 1 + virt/kvm/kvm_main.c | 8 +++++-- 4 files changed, 31 insertions(+), 18 deletions(-) (limited to 'Documentation/virtual/kvm') diff --git a/Documentation/virtual/kvm/halt-polling.txt b/Documentation/virtual/kvm/halt-polling.txt index 4a8418318769..4f791b128dd2 100644 --- a/Documentation/virtual/kvm/halt-polling.txt +++ b/Documentation/virtual/kvm/halt-polling.txt @@ -53,7 +53,8 @@ the global max polling interval then the polling interval can be increased in the hope that next time during the longer polling interval the wake up source will be received while the host is polling and the latency benefits will be received. The polling interval is grown in the function grow_halt_poll_ns() and -is multiplied by the module parameter halt_poll_ns_grow. +is multiplied by the module parameters halt_poll_ns_grow and +halt_poll_ns_grow_start. In the event that the total block time was greater than the global max polling interval then the host will never poll for long enough (limited by the global @@ -80,22 +81,30 @@ shrunk. These variables are defined in include/linux/kvm_host.h and as module parameters in virt/kvm/kvm_main.c, or arch/powerpc/kvm/book3s_hv.c in the powerpc kvm-hv case. -Module Parameter | Description | Default Value +Module Parameter | Description | Default Value -------------------------------------------------------------------------------- -halt_poll_ns | The global max polling interval | KVM_HALT_POLL_NS_DEFAULT - | which defines the ceiling value | - | of the polling interval for | (per arch value) - | each vcpu. | +halt_poll_ns | The global max polling | KVM_HALT_POLL_NS_DEFAULT + | interval which defines | + | the ceiling value of the | + | polling interval for | (per arch value) + | each vcpu. | -------------------------------------------------------------------------------- -halt_poll_ns_grow | The value by which the halt | 2 - | polling interval is multiplied | - | in the grow_halt_poll_ns() | - | function. | +halt_poll_ns_grow | The value by which the | 2 + | halt polling interval is | + | multiplied in the | + | grow_halt_poll_ns() | + | function. | -------------------------------------------------------------------------------- -halt_poll_ns_shrink | The value by which the halt | 0 - | polling interval is divided in | - | the shrink_halt_poll_ns() | - | function. | +halt_poll_ns_grow_start | The initial value to grow | 10000 + | to from zero in the | + | grow_halt_poll_ns() | + | function. | +-------------------------------------------------------------------------------- +halt_poll_ns_shrink | The value by which the | 0 + | halt polling interval is | + | divided in the | + | shrink_halt_poll_ns() | + | function. | -------------------------------------------------------------------------------- These module parameters can be set from the debugfs files in: diff --git a/arch/powerpc/kvm/book3s_hv.c b/arch/powerpc/kvm/book3s_hv.c index e316a2ddb70b..29ffc99bd79b 100644 --- a/arch/powerpc/kvm/book3s_hv.c +++ b/arch/powerpc/kvm/book3s_hv.c @@ -3634,9 +3634,8 @@ static void grow_halt_poll_ns(struct kvmppc_vcore *vc) if (!halt_poll_ns_grow) return; - /* 10us base */ if (vc->halt_poll_ns == 0) - vc->halt_poll_ns = 10000; + vc->halt_poll_ns = halt_poll_ns_grow_start; else vc->halt_poll_ns *= halt_poll_ns_grow; } diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h index 85c0c00d5159..9d55c63db09b 100644 --- a/include/linux/kvm_host.h +++ b/include/linux/kvm_host.h @@ -1203,6 +1203,7 @@ extern bool kvm_rebooting; extern unsigned int halt_poll_ns; extern unsigned int halt_poll_ns_grow; +extern unsigned int halt_poll_ns_grow_start; extern unsigned int halt_poll_ns_shrink; struct kvm_device { diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 9c8a8bf6e686..ae818d27a1a4 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -81,6 +81,11 @@ unsigned int halt_poll_ns_grow = 2; module_param(halt_poll_ns_grow, uint, 0644); EXPORT_SYMBOL_GPL(halt_poll_ns_grow); +/* The start value to grow halt_poll_ns from */ +unsigned int halt_poll_ns_grow_start = 10000; /* 10us */ +module_param(halt_poll_ns_grow_start, uint, 0644); +EXPORT_SYMBOL_GPL(halt_poll_ns_grow_start); + /* Default resets per-vcpu halt_poll_ns . */ unsigned int halt_poll_ns_shrink; module_param(halt_poll_ns_shrink, uint, 0644); @@ -2191,9 +2196,8 @@ static void grow_halt_poll_ns(struct kvm_vcpu *vcpu) if (!grow) goto out; - /* 10us base */ if (val == 0) - val = 10000; + val = halt_poll_ns_grow_start; else val *= grow; -- cgit v1.2.3 From eca6be566d47029f945a5f8e1c94d374e31df2ca Mon Sep 17 00:00:00 2001 From: Sean Christopherson Date: Fri, 15 Feb 2019 12:48:40 -0800 Subject: KVM: doc: Document the life cycle of a VM and its resources The series to add memcg accounting to KVM allocations[1] states: There are many KVM kernel memory allocations which are tied to the life of the VM process and should be charged to the VM process's cgroup. While it is correct to account KVM kernel allocations to the cgroup of the process that created the VM, it's technically incorrect to state that the KVM kernel memory allocations are tied to the life of the VM process. This is because the VM itself, i.e. struct kvm, is not tied to the life of the process which created it, rather it is tied to the life of its associated file descriptor. In other words, kvm_destroy_vm() is not invoked until fput() decrements its associated file's refcount to zero. A simple example is to fork() in Qemu and have the child sleep indefinitely; kvm_destroy_vm() isn't called until Qemu closes its file descriptor *and* the rogue child is killed. The allocations are guaranteed to be *accounted* to the process which created the VM, but only because KVM's per-{VM,vCPU} ioctls reject the ioctl() with -EIO if kvm->mm != current->mm. I.e. the child can keep the VM "alive" but can't do anything useful with its reference. Note that because 'struct kvm' also holds a reference to the mm_struct of its owner, the above behavior also applies to userspace allocations. Given that mucking with a VM's file descriptor can lead to subtle and undesirable behavior, e.g. memcg charges persisting after a VM is shut down, explicitly document a VM's lifecycle and its impact on the VM's resources. Alternatively, KVM could aggressively free resources when the creating process exits, e.g. via mmu_notifier->release(). However, mmu_notifier isn't guaranteed to be available, and freeing resources when the creator exits is likely to be error prone and fragile as KVM would need to ensure that it only freed resources that are truly out of reach. In practice, the existing behavior shouldn't be problematic as a properly configured system will prevent a child process from being moved out of the appropriate cgroup hierarchy, i.e. prevent hiding the process from the OOM killer, and will prevent an unprivileged user from being able to to hold a reference to struct kvm via another method, e.g. debugfs. [1]https://patchwork.kernel.org/patch/10806707/ Signed-off-by: Sean Christopherson Signed-off-by: Paolo Bonzini --- Documentation/virtual/kvm/api.txt | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) (limited to 'Documentation/virtual/kvm') diff --git a/Documentation/virtual/kvm/api.txt b/Documentation/virtual/kvm/api.txt index 356156f5c52d..7de9eee73fcd 100644 --- a/Documentation/virtual/kvm/api.txt +++ b/Documentation/virtual/kvm/api.txt @@ -45,6 +45,23 @@ the API. The only supported use is one virtual machine per process, and one vcpu per thread. +It is important to note that althought VM ioctls may only be issued from +the process that created the VM, a VM's lifecycle is associated with its +file descriptor, not its creator (process). In other words, the VM and +its resources, *including the associated address space*, are not freed +until the last reference to the VM's file descriptor has been released. +For example, if fork() is issued after ioctl(KVM_CREATE_VM), the VM will +not be freed until both the parent (original) process and its child have +put their references to the VM's file descriptor. + +Because a VM's resources are not freed until the last reference to its +file descriptor is released, creating additional references to a VM via +via fork(), dup(), etc... without careful consideration is strongly +discouraged and may have unwanted side effects, e.g. memory allocated +by and on behalf of the VM's process may not be freed/unaccounted when +the VM is shut down. + + 3. Extensions ------------- -- cgit v1.2.3