kernel/linux.git/arch/x86, branch v6.1.177

KVM: SVM: Fix page overflow in sev_dbg_crypt() for ENCRYPT path

2026-07-04T11:41:32+00:00

commit 78ee2d50185a037b3d2452a97f3dad69c3f7f389 upstream. In sev_dbg_crypt(), the per-iteration transfer length is bounded by the source page offset (PAGE_SIZE - s_off) but not by the destination page offset (PAGE_SIZE - d_off). When d_off > s_off, the encrypt path (__sev_dbg_encrypt_user) performs a read-modify-write using a single-page intermediate buffer (dst_tpage): 1. __sev_dbg_decrypt() expands the size to round_up(len + (d_off & 15), 16) before issuing the PSP command. If len + (d_off & 15) > PAGE_SIZE, the PSP writes beyond the end of the 4096-byte dst_tpage allocation. 2. The subsequent memcpy()/copy_from_user() into page_address(dst_tpage) + (d_off & 15) of 'len' bytes overflows by up to 15 bytes under the same condition. Trigger example: s_off = 0, d_off = 1, debug.len = PAGE_SIZE - the PSP is instructed to write round_up(4097, 16) = 4112 bytes to a 4096-byte buffer. Fix by also bounding len by (PAGE_SIZE - d_off), the same check that sev_send_update_data() already performs for its single-page guest region. ================================================================== BUG: KASAN: slab-use-after-free in sev_dbg_crypt+0x993/0xd10 [kvm_amd] Write of size 4095 at addr ff110062293bb009 by task sev_dbg_test/228214 CPU: 96 UID: 0 PID: 228214 Comm: sev_dbg_test Tainted: G U W 7.0.0-smp--5ce9b0c48211-dbg #156 PREEMPTLAZY Tainted: [U]=USER, [W]=WARN Hardware name: Google Astoria/astoria, BIOS 0.20250817.1-0 08/25/2025 Call Trace: dump_stack_lvl+0x54/0x70 print_report+0xbc/0x260 kasan_report+0xa2/0xd0 kasan_check_range+0x25f/0x2c0 __asan_memcpy+0x40/0x70 sev_dbg_crypt+0x993/0xd10 [kvm_amd] sev_mem_enc_ioctl+0x33c/0x450 [kvm_amd] kvm_vm_ioctl+0x65d/0x6d0 [kvm] __se_sys_ioctl+0xb2/0x100 do_syscall_64+0xe8/0x870 entry_SYSCALL_64_after_hwframe+0x4b/0x53 The buggy address belongs to the physical page: page: refcount:1 mapcount:0 mapping:0000000000000000 index:0x7fe72b6a0 pfn:0x62293bb memcg:ff11000112827d82 flags: 0x1400000000000000(node=1|zone=1) raw: 1400000000000000 0000000000000000 dead000000000122 0000000000000000 raw: 00000007fe72b6a0 0000000000000000 00000001ffffffff ff11000112827d82 page dumped because: kasan: bad access detected Memory state around the buggy address: ff110062293bbf00: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ff110062293bbf80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 >ff110062293bc000: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc ^ ff110062293bc080: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc ff110062293bc100: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc ================================================================== Disabling lock debugging due to kernel taint Fixes: 24f41fb23a39 ("KVM: SVM: Add support for SEV DEBUG_DECRYPT command") Fixes: 7d1594f5d94b ("KVM: SVM: Add support for SEV DEBUG_ENCRYPT command") Cc: stable@vger.kernel.org Signed-off-by: Ashutosh Desai [sean: add sample KASAN splat, Fixes, and stable@] Link: https://patch.msgid.link/20260501203537.2120074-2-seanjc@google.com Signed-off-by: Sean Christopherson Signed-off-by: Greg Kroah-Hartman

KVM: x86/mmu: Ensure hugepage is in by slot before checking max mapping level

2026-07-04T11:41:29+00:00

commit ef057cbf825e03b63f6edf5980f96abf3c53089d upstream. When recovering hugepages in the shadow MMU, verify that the base gfn of the shadow page is actually contained within the target memslot, *before* querying the max mapping level given the shadow page's gfn. Failure to pre-check the validity of the gfn can lead to an out-of-bounds access to the slot's lpage_info (which typically manifests as a host #PF because the lpage_info is vmalloc'd) if the guest creates a hugepage mapping (in its PTEs) that extends "below" the bounds of a memslot. When faulting in memory for a guest, and the size of the guest mapping is greater than KVM's (current) max mapping, then KVM will create a "direct" shadow page (direct in that there are no gPTEs to shadow, and so the target gfn is a direct calculation given the base gfn of the shadow page). The hugepage recovery flow looks for such direct shadow pages, as forcing 4KiB mappings when dirty logging generates the guest > host mapping size case. When the 4KiB restriction is lifted, then KVM can replace the shadow page with a hugepage. But if KVM originally used a smaller mapping than the guest because the range of memory covered by the guest hugepage exceeds the bounds of a memslot, then KVM will link a direct shadow page with a gfn that is outside the bounds of the memslot being used to fault in memory. The rmap entry added for the leaf mapping is correct and within bounds, but the gfn of the leaf SPTE's parent shadow page will be out of bounds. BUG: unable to handle page fault for address: ffffc90000806ffc #PF: supervisor read access in kernel mode #PF: error_code(0x0000) - not-present page PGD 100000067 P4D 100000067 PUD 1002a7067 PMD 10612f067 PTE 0 Oops: Oops: 0000 [#1] SMP CPU: 13 UID: 1000 PID: 757 Comm: mmu_stress_test Not tainted 7.1.0-rc1-48ce1e26eace-x86_pir_to_irr_comments-vm #341 PREEMPT Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 RIP: 0010:kvm_mmu_max_mapping_level+0x79/0x2b0 [kvm] Call Trace: kvm_mmu_recover_huge_pages+0x21b/0x320 [kvm] kvm_set_memslot+0x1ee/0x590 [kvm] kvm_set_memory_region.part.0+0x3a1/0x4d0 [kvm] kvm_vm_ioctl+0x9bf/0x15d0 [kvm] __x64_sys_ioctl+0x8a/0xd0 do_syscall_64+0xb7/0xbb0 entry_SYSCALL_64_after_hwframe+0x4b/0x53 RIP: 0033:0x7f21c0f1a9bf Don't bother pre-checking the bounds of the potential hugepage, i.e. don't check that e.g. sp->gfn + KVM_PAGES_PER_HPAGE(sp->role.level + 1) is also within the memslot, as the checks performed by kvm_mmu_max_mapping_level() are a superset of the basic bounds checks. I.e. pre-checking the full range would be a dubious micro-optimization. Fixes: 9eba50f8d7fc ("KVM: x86/mmu: Consult max mapping level when zapping collapsible SPTEs") Cc: stable@vger.kernel.org Cc: David Matlack Cc: James Houghton Cc: Alexander Bulekov Cc: Fred Griffoul Cc: Alexander Graf Cc: David Woodhouse Cc: Filippo Sironi Cc: Ivan Orlov Signed-off-by: Sean Christopherson Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

KVM: x86: Fix shadow paging use-after-free due to unexpected role

2026-07-04T11:41:29+00:00

commit 81ccda30b4e83d8f5cc4fd50503c44e3a33abfeb upstream. Commit 0cb2af2ea66ad ("KVM: x86: Fix shadow paging use-after-free due to unexpected GFN") fixed a shadow paging mismatch between stored and computed GFNs; the bug could be triggered by changing a PDE mapping from outside the guest, and then deleting a memslot. The rmap_remove() call would miss entries created after the PDE change because the GFN of the leaf SPTE does not match the GFN of the struct kvm_mmu_page. A similar hole however remains if the modified PDE points to a non-leaf page. In this case the gfn can be made to match, but the role does not match: the original large 2MB page creates a kvm_mmu_page with direct=1, while the new 4KB needs a kvm_mmu_page with direct=0. However, kvm_mmu_get_child_sp() does not compare the role, and therefore reuses the page. The next step is installing a leaf (4KB) SPTE on the new path which records an rmap entry under the gfn resolved by the walk. But when that child is zapped its parent kvm_mmu_page has direct=1 and kvm_mmu_page_get_gfn() computes the gfn for the 4KB page as sp->gfn + index instead of using sp->shadowed_translation[] (or sp->gfns[] in older kernels). It therefore fails to remove the recorded entry. When the memslot is dropped the shadow page is freed but the rmap entry survives, as in the scenario that was already fixed. Code that later walks that gfn (dirty logging, MMU notifier invalidation, and so on) dereferences an sptep that lies in the freed page, causing the use-after-free. Fixes: 2032a93d66fa ("KVM: MMU: Don't allocate gfns page for direct mmu pages") Reported-by: Hyunwoo Kim Signed-off-by: Paolo Bonzini Signed-off-by: Sasha Levin

KVM: nVMX: Fold requested virtual interrupt check into has_nested_events()

2026-07-04T11:41:27+00:00

commit 321ef62b0c5f6f57bb8500a2ca5986052675abbf upstream. Check for a Requested Virtual Interrupt, i.e. a virtual interrupt that is pending delivery, in vmx_has_nested_events() and drop the one-off kvm_x86_ops.guest_apic_has_interrupt() hook. In addition to dropping a superfluous hook, this fixes a bug where KVM would incorrectly treat virtual interrupts _for L2_ as always enabled due to kvm_arch_interrupt_allowed(), by way of vmx_interrupt_blocked(), treating IRQs as enabled if L2 is active and vmcs12 is configured to exit on IRQs, i.e. KVM would treat a virtual interrupt for L2 as a valid wake event based on L1's IRQ blocking status. Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240607172609.3205077-6-seanjc@google.com Signed-off-by: Sean Christopherson [ Nicholas Dudar: backport to 6.1.y. 6.1.y predates the vmx main.c / x86_ops.h split, so drop .guest_apic_has_interrupt from vmx_x86_ops in vmx.c rather than vt_x86_ops in main.c. The function is static in vmx.c, so upstream's x86_ops.h prototype removal does not apply. 6.1.y keeps the current hwapic_isr_update signature. ] Signed-off-by: Nicholas Dudar Signed-off-by: Sasha Levin

KVM: nVMX: Check for pending posted interrupts when looking for nested events

2026-07-04T11:41:27+00:00

commit 27c4fa42b11af780d49ce704f7fa67b3c2544df4 upstream. Check for pending (and notified!) posted interrupts when checking if L2 has a pending wake event, as fully posted/notified virtual interrupt is a valid wake event for HLT. Note that KVM must check vmx->nested.pi_pending to avoid prematurely waking L2, e.g. even if KVM sees a non-zero PID.PIR and PID.0N=1, the virtual interrupt won't actually be recognized until a notification IRQ is received by the vCPU or the vCPU does (nested) VM-Enter. Fixes: 26844fee6ade ("KVM: x86: never write to memory from kvm_vcpu_check_block()") Cc: stable@vger.kernel.org Cc: Maxim Levitsky Reported-by: Jim Mattson Closes: https://lore.kernel.org/all/20231207010302.2240506-1-jmattson@google.com Link: https://lore.kernel.org/r/20240607172609.3205077-5-seanjc@google.com Signed-off-by: Sean Christopherson [ Nicholas Dudar: backport to 6.1.y. Prerequisite for the next patch, which folds its check into the vmx_has_nested_events() body this patch builds. Applies cleanly. The for_injection path still returns preemption_timer || mtf, as the previous 6.1.y body did. ] Signed-off-by: Nicholas Dudar Signed-off-by: Sasha Levin

KVM: nVMX: Add a helper to get highest pending from Posted Interrupt vector

2026-07-04T11:41:27+00:00

commit d83c36d822be44db4bad0c43bea99c8908f54117 upstream. Add a helper to retrieve the highest pending vector given a Posted Interrupt descriptor. While the actual operation is straightforward, it's surprisingly easy to mess up, e.g. if one tries to reuse lapic.c's find_highest_vector(), which doesn't work with PID.PIR due to the APIC's IRR and ISR component registers being physically discontiguous (they're 4-byte registers aligned at 16-byte intervals). To make PIR handling more consistent with respect to IRR and ISR handling, return -1 to indicate "no interrupt pending". Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240607172609.3205077-2-seanjc@google.com Signed-off-by: Sean Christopherson [ Nicholas Dudar: backport to 6.1.y. 6.1.y defines struct pi_desc in posted_intr.h and predates the move to , so the helper and the include go in posted_intr.h. ] Signed-off-by: Nicholas Dudar Signed-off-by: Sasha Levin

KVM: VMX: Make vmread_error_trampoline() uncallable from C code

2026-07-04T11:41:27+00:00

[ Upstream commit 0b5e7a16a0a79a3742f0df9e45bca46f01b40e6a ] Declare vmread_error_trampoline() as an opaque symbol so that it cannot be called from C code, at least not without some serious fudging. The trampoline always passes parameters on the stack so that the inline VMREAD sequence doesn't need to clobber registers. regparm(0) was originally added to document the stack behavior, but it ended up being confusing because regparm(0) is a nop for 64-bit targets. Opportunustically wrap the trampoline and its declaration in #ifdeffery to make it even harder to invoke incorrectly, to document why it exists, and so that it's not left behind if/when CONFIG_CC_HAS_ASM_GOTO_OUTPUT is true for all supported toolchains. No functional change intended. Cc: Uros Bizjak Signed-off-by: Sean Christopherson Link: https://lore.kernel.org/r/20220928232015.745948-1-seanjc@google.com (cherry picked from commit 0b5e7a16a0a79a3742f0df9e45bca46f01b40e6a) Signed-off-by: Hanne-Lotta Mäenpää Signed-off-by: Sasha Levin

x86/CPU/AMD: Move the Zen3 BTC_NO detection to the Zen3 init function

2026-06-19T11:37:32+00:00

commit affc66cb96f865b3763a8e18add52e133d864f04 upstream. No functional changes. Signed-off-by: Borislav Petkov (AMD) Reviewed-by: Nikolay Borisov Link: http://lore.kernel.org/r/20231120104152.13740-4-bp@alien8.de Stable-dep-of: 7c81ad8e8bc2 ("x86/CPU/AMD: Rename init_amd_zn() to init_amd_zen_common()") [bwh: Adjusted to apply after backports of the above commit which actually depended on this] Signed-off-by: Ben Hutchings Signed-off-by: Greg Kroah-Hartman

Revert "x86/vdso: Fix output operand size of RDPID"

2026-06-01T15:39:34+00:00

This reverts commit 757a9e78a1c5b824d0a2b7de14c3cd8d841dfbee. Signed-off-by: Sasha Levin

KVM: x86: Fix Xen hypercall tracepoint argument assignment

2026-06-01T15:39:32+00:00

commit 2b72f1674e427c56e3772c5ccf785fdda2138820 upstream. TRACE_EVENT(kvm_xen_hypercall) stores a5 in __entry->a4 instead of __entry->a5. That overwrites the recorded a4 argument and leaves a5 unset in the trace entry. Fix the typo so both arguments are captured correctly. Signed-off-by: Qiang Ma Link: https://patch.msgid.link/20260512015313.1685784-1-maqianga@uniontech.com/ Cc: stable@vger.kernel.org Signed-off-by: Paolo Bonzini Signed-off-by: Greg Kroah-Hartman