kernel/linux.git/include/linux/kvm_host.h, branch v5.10.258

KVM: x86: Use scratch field in MMIO fragment to hold small write values

2026-06-01T15:29:12+00:00

commit 0b16e69d17d8c35c5c9d5918bf596c75a44655d3 upstream. When exiting to userspace to service an emulated MMIO write, copy the to-be-written value to a scratch field in the MMIO fragment if the size of the data payload is 8 bytes or less, i.e. can fit in a single chunk, instead of pointing the fragment directly at the source value. This fixes a class of use-after-free bugs that occur when the emulator initiates a write using an on-stack, local variable as the source, the write splits a page boundary, *and* both pages are MMIO pages. Because KVM's ABI only allows for physically contiguous MMIO requests, accesses that split MMIO pages are separated into two fragments, and are sent to userspace one at a time. When KVM attempts to complete userspace MMIO in response to KVM_RUN after the first fragment, KVM will detect the second fragment and generate a second userspace exit, and reference the on-stack variable. The issue is most visible if the second KVM_RUN is performed by a separate task, in which case the stack of the initiating task can show up as truly freed data. ================================================================== BUG: KASAN: use-after-free in complete_emulated_mmio+0x305/0x420 Read of size 1 at addr ffff888009c378d1 by task syz-executor417/984 CPU: 1 PID: 984 Comm: syz-executor417 Not tainted 5.10.0-182.0.0.95.h2627.eulerosv2r13.x86_64 #3 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014 Call Trace: dump_stack+0xbe/0xfd print_address_description.constprop.0+0x19/0x170 __kasan_report.cold+0x6c/0x84 kasan_report+0x3a/0x50 check_memory_region+0xfd/0x1f0 memcpy+0x20/0x60 complete_emulated_mmio+0x305/0x420 kvm_arch_vcpu_ioctl_run+0x63f/0x6d0 kvm_vcpu_ioctl+0x413/0xb20 __se_sys_ioctl+0x111/0x160 do_syscall_64+0x30/0x40 entry_SYSCALL_64_after_hwframe+0x67/0xd1 RIP: 0033:0x42477d Code: <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48 RSP: 002b:00007faa8e6890e8 EFLAGS: 00000246 ORIG_RAX: 0000000000000010 RAX: ffffffffffffffda RBX: 00000000004d7338 RCX: 000000000042477d RDX: 0000000000000000 RSI: 000000000000ae80 RDI: 0000000000000005 RBP: 00000000004d7330 R08: 00007fff28d546df R09: 0000000000000000 R10: 0000000000000000 R11: 0000000000000246 R12: 00000000004d733c R13: 0000000000000000 R14: 000000000040a200 R15: 00007fff28d54720 The buggy address belongs to the page: page:0000000029f6a428 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x9c37 flags: 0xfffffc0000000(node=0|zone=1|lastcpupid=0x1fffff) raw: 000fffffc0000000 0000000000000000 ffffea0000270dc8 0000000000000000 raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000 page dumped because: kasan: bad access detected Memory state around the buggy address: ffff888009c37780: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ffff888009c37800: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff >ffff888009c37880: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ^ ffff888009c37900: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ffff888009c37980: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ================================================================== The bug can also be reproduced with a targeted KVM-Unit-Test by hacking KVM to fill a large on-stack variable in complete_emulated_mmio(), i.e. by overwrite the data value with garbage. Limit the use of the scratch fields to 8-byte or smaller accesses, and to just writes, as larger accesses and reads are not affected thanks to implementation details in the emulator, but add a sanity check to ensure those details don't change in the future. Specifically, KVM never uses on-stack variables for accesses larger that 8 bytes, e.g. uses an operand in the emulator context, and *all* reads are buffered through the mem_read cache. Note! Using the scratch field for reads is not only unnecessary, it's also extremely difficult to handle correctly. As above, KVM buffers all reads through the mem_read cache, and heavily relies on that behavior when re-emulating the instruction after a userspace MMIO read exit. If a read splits a page, the first page is NOT an MMIO page, and the second page IS an MMIO page, then the MMIO fragment needs to point at _just_ the second chunk of the destination, i.e. its position in the mem_read cache. Taking the "obvious" approach of copying the fragment value into the destination when re-emulating the instruction would clobber the first chunk of the destination, i.e. would clobber the data that was read from guest memory. Fixes: f78146b0f923 ("KVM: Fix page-crossing MMIO") Suggested-by: Yashu Zhang Reported-by: Yashu Zhang Closes: https://lore.kernel.org/all/369eaaa2b3c1425c85e8477066391bc7@huawei.com Cc: stable@vger.kernel.org Tested-by: Tom Lendacky Tested-by: Rick Edgecombe Link: https://patch.msgid.link/20260225012049.920665-2-seanjc@google.com Signed-off-by: Sean Christopherson Signed-off-by: Greg Kroah-Hartman

KVM: Explicitly verify target vCPU is online in kvm_get_vcpu()

2025-03-13T11:47:06+00:00

commit 1e7381f3617d14b3c11da80ff5f8a93ab14cfc46 upstream. Explicitly verify the target vCPU is fully online _prior_ to clamping the index in kvm_get_vcpu(). If the index is "bad", the nospec clamping will generate '0', i.e. KVM will return vCPU0 instead of NULL. In practice, the bug is unlikely to cause problems, as it will only come into play if userspace or the guest is buggy or misbehaving, e.g. KVM may send interrupts to vCPU0 instead of dropping them on the floor. However, returning vCPU0 when it shouldn't exist per online_vcpus is problematic now that KVM uses an xarray for the vCPUs array, as KVM needs to insert into the xarray before publishing the vCPU to userspace (see commit c5b077549136 ("KVM: Convert the kvm->vcpus array to a xarray")), i.e. before vCPU creation is guaranteed to succeed. As a result, incorrectly providing access to vCPU0 will trigger a use-after-free if vCPU0 is dereferenced and kvm_vm_ioctl_create_vcpu() bails out of vCPU creation due to an error and frees vCPU0. Commit afb2acb2e3a3 ("KVM: Fix vcpu_array[0] races") papered over that issue, but in doing so introduced an unsolvable teardown conundrum. Preventing accesses to vCPU0 before it's fully online will allow reverting commit afb2acb2e3a3, without re-introducing the vcpu_array[0] UAF race. Fixes: 1d487e9bf8ba ("KVM: fix spectrev1 gadgets") Cc: stable@vger.kernel.org Cc: Will Deacon Cc: Michal Luczaj Reviewed-by: Pankaj Gupta Acked-by: Will Deacon Link: https://lore.kernel.org/r/20241009150455.1057573-2-seanjc@google.com Signed-off-by: Sean Christopherson Signed-off-by: Greg Kroah-Hartman

kvm: Add support for arch compat vm ioctls

2022-10-30T08:41:15+00:00

commit ed51862f2f57cbce6fed2d4278cfe70a490899fd upstream. We will introduce the first architecture specific compat vm ioctl in the next patch. Add all necessary boilerplate to allow architectures to override compat vm ioctls when necessary. Signed-off-by: Alexander Graf Message-Id: <20221017184541.2658-2-graf@amazon.com> Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman

KVM: SEV: add cache flush to solve SEV cache incoherency issues

2022-09-28T09:10:28+00:00

commit 683412ccf61294d727ead4a73d97397396e69a6b upstream. Flush the CPU caches when memory is reclaimed from an SEV guest (where reclaim also includes it being unmapped from KVM's memslots). Due to lack of coherency for SEV encrypted memory, failure to flush results in silent data corruption if userspace is malicious/broken and doesn't ensure SEV guest memory is properly pinned and unpinned. Cache coherency is not enforced across the VM boundary in SEV (AMD APM vol.2 Section 15.34.7). Confidential cachelines, generated by confidential VM guests have to be explicitly flushed on the host side. If a memory page containing dirty confidential cachelines was released by VM and reallocated to another user, the cachelines may corrupt the new user at a later time. KVM takes a shortcut by assuming all confidential memory remain pinned until the end of VM lifetime. Therefore, KVM does not flush cache at mmu_notifier invalidation events. Because of this incorrect assumption and the lack of cache flushing, malicous userspace can crash the host kernel: creating a malicious VM and continuously allocates/releases unpinned confidential memory pages when the VM is running. Add cache flush operations to mmu_notifier operations to ensure that any physical memory leaving the guest VM get flushed. In particular, hook mmu_notifier_invalidate_range_start and mmu_notifier_release events and flush cache accordingly. The hook after releasing the mmu lock to avoid contention with other vCPUs. Cc: stable@vger.kernel.org Suggested-by: Sean Christpherson Reported-by: Mingwei Zhang Signed-off-by: Mingwei Zhang Message-Id: <20220421031407.2516575-4-mizhang@google.com> Signed-off-by: Paolo Bonzini [OP: applied kvm_arch_guest_memory_reclaimed() calls in kvm_set_memslot() and kvm_mmu_notifier_invalidate_range_start(); OP: adjusted kvm_arch_guest_memory_reclaimed() to not use static_call_cond()] Signed-off-by: Ovidiu Panait Reviewed-by: Liam Merwick Signed-off-by: Greg Kroah-Hartman

KVM: Add infrastructure and macro to mark VM as bugged

2022-08-21T13:16:24+00:00

commit 0b8f11737cffc1a406d1134b58687abc29d76b52 upstream Signed-off-by: Sean Christopherson Signed-off-by: Isaku Yamahata Reviewed-by: Paolo Bonzini Message-Id: <3a0998645c328bf0895f1290e61821b70f048549.1625186503.git.isaku.yamahata@intel.com> Signed-off-by: Paolo Bonzini [SG: Adjusted context for kernel version 5.10] Signed-off-by: Stefan Ghinea Signed-off-by: Greg Kroah-Hartman

x86/kvm/vmx: Make noinstr clean

2022-07-25T09:26:34+00:00

commit 742ab6df974ae8384a2dd213db1a3a06cf6d8936 upstream. The recent mmio_stale_data fixes broke the noinstr constraints: vmlinux.o: warning: objtool: vmx_vcpu_enter_exit+0x15b: call to wrmsrl.constprop.0() leaves .noinstr.text section vmlinux.o: warning: objtool: vmx_vcpu_enter_exit+0x1bf: call to kvm_arch_has_assigned_device() leaves .noinstr.text section make it all happy again. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Borislav Petkov Signed-off-by: Thadeu Lima de Souza Cascardo Signed-off-by: Ben Hutchings Signed-off-by: Greg Kroah-Hartman

kvm: fix previous commit for 32-bit builds

2021-06-16T10:01:46+00:00

commit 4422829e8053068e0225e4d0ef42dc41ea7c9ef5 upstream. array_index_nospec does not work for uint64_t on 32-bit builds. However, the size of a memory slot must be less than 20 bits wide on those system, since the memory slot must fit in the user address space. So just store it in an unsigned long. Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman

kvm: avoid speculation-based attacks from out-of-range memslot accesses

2021-06-16T10:01:40+00:00

commit da27a83fd6cc7780fea190e1f5c19e87019da65c upstream. KVM's mechanism for accessing guest memory translates a guest physical address (gpa) to a host virtual address using the right-shifted gpa (also known as gfn) and a struct kvm_memory_slot. The translation is performed in __gfn_to_hva_memslot using the following formula: hva = slot->userspace_addr + (gfn - slot->base_gfn) * PAGE_SIZE It is expected that gfn falls within the boundaries of the guest's physical memory. However, a guest can access invalid physical addresses in such a way that the gfn is invalid. __gfn_to_hva_memslot is called from kvm_vcpu_gfn_to_hva_prot, which first retrieves a memslot through __gfn_to_memslot. While __gfn_to_memslot does check that the gfn falls within the boundaries of the guest's physical memory or not, a CPU can speculate the result of the check and continue execution speculatively using an illegal gfn. The speculation can result in calculating an out-of-bounds hva. If the resulting host virtual address is used to load another guest physical address, this is effectively a Spectre gadget consisting of two consecutive reads, the second of which is data dependent on the first. Right now it's not clear if there are any cases in which this is exploitable. One interesting case was reported by the original author of this patch, and involves visiting guest page tables on x86. Right now these are not vulnerable because the hva read goes through get_user(), which contains an LFENCE speculation barrier. However, there are patches in progress for x86 uaccess.h to mask kernel addresses instead of using LFENCE; once these land, a guest could use speculation to read from the VMM's ring 3 address space. Other architectures such as ARM already use the address masking method, and would be susceptible to this same kind of data-dependent access gadgets. Therefore, this patch proactively protects from these attacks by masking out-of-bounds gfns in __gfn_to_hva_memslot, which blocks speculation of invalid hvas. Sean Christopherson noted that this patch does not cover kvm_read_guest_offset_cached. This however is limited to a few bytes past the end of the cache, and therefore it is unlikely to be useful in the context of building a chain of data dependent accesses. Reported-by: Artemiy Margaritov Co-developed-by: Artemiy Margaritov Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman

KVM: Stop looking for coalesced MMIO zones if the bus is destroyed

2021-05-14T07:50:04+00:00

commit 5d3c4c79384af06e3c8e25b7770b6247496b4417 upstream. Abort the walk of coalesced MMIO zones if kvm_io_bus_unregister_dev() fails to allocate memory for the new instance of the bus. If it can't instantiate a new bus, unregister_dev() destroys all devices _except_ the target device. But, it doesn't tell the caller that it obliterated the bus and invoked the destructor for all devices that were on the bus. In the coalesced MMIO case, this can result in a deleted list entry dereference due to attempting to continue iterating on coalesced_zones after future entries (in the walk) have been deleted. Opportunistically add curly braces to the for-loop, which encompasses many lines but sneaks by without braces due to the guts being a single if statement. Fixes: f65886606c2d ("KVM: fix memory leak in kvm_io_bus_unregister_dev()") Cc: stable@vger.kernel.org Reported-by: Hao Sun Signed-off-by: Sean Christopherson Message-Id: <20210412222050.876100-3-seanjc@google.com> Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman

kvm: x86/mmu: Support dirty logging for the TDP MMU

2020-10-23T07:42:13+00:00

Dirty logging is a key feature of the KVM MMU and must be supported by the TDP MMU. Add support for both the write protection and PML dirty logging modes. Tested by running kvm-unit-tests and KVM selftests on an Intel Haswell machine. This series introduced no new failures. This series can be viewed in Gerrit at: https://linux-review.googlesource.com/c/virt/kvm/kvm/+/2538 Signed-off-by: Ben Gardon Message-Id: <20201014182700.2888246-16-bgardon@google.com> Signed-off-by: Paolo Bonzini