[PATCH] KVM: SEV: Allocate full pages for {DE,EN}CRYPT ops on SNP-enabled hosts

[PATCH] KVM: SEV: Allocate full pages for {DE,EN}CRYPT ops on SNP-enabled hosts

[Sean Christopherson]

When {de,en}crypting memory of an SEV or SEV-ES guest on an SNP-enabled
host via a temporary buffer, allocate a full 4KiB page for the buffer to
ensure the page containing the buffer is wholly owned by KVM, i.e. won't
be concurrently allocated and accessed by other kernel code while KVM is
using the buffer to {de,en}crypt memory.  On SNP-enabled platforms, when
sending SEV/SEV-ES commands that trigger firmware writes to memory, the
to-be-written page(s) must be (temporarily) assigned to Firmware (as
required by the SNP architecture, to guard against using such commands as
gadgets to attack SNP guests).  See snp_map_cmd_buf_desc() and friends.

Unfortunately, transferring ownership of a page to Firmware makes the page
inaccessible to software, and thus writes generate RMP #PF violations.  If
KVM uses a sub-page allocation for its temporary buffer, some other actor
in the kernel can allocate and use the other portions of the page, and thus
trigger unexpected (and seemingly spurious) RMP #PF violations due to
software attempting to access a Firmware-owned page.

  BUG: unable to handle page fault for address: ffff906ae30f0300
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x80000003) - RMP violation
  PGD 6b1b80d067 P4D 6b1b80d067 PUD 100231e2063 PMD 10055a88063 PTE 80000100630f0163
  SEV-SNP: PFN 0x100630f0 unassigned, dumping non-zero entries in 2M PFN region: [0x10063000 - 0x10063200]
  Oops: Oops: 0003 [#1] SMP
  CPU: 70 UID: 0 PID: 10658 Comm: svw_WaiterThrea Tainted: G     U  W  O        7.1.0-smp--c22293789940-seanjc-next #1 PREEMPTLAZY
  Tainted: [U]=USER, [W]=WARN, [O]=OOT_MODULE
  Hardware name: Google, Inc.                                                       Arcadia_IT_80/Arcadia_IT_80, BIOS 34.86.0-102 01/25/2026
  RIP: 0010:memset+0xf/0x20
  Call Trace:
   <TASK>
   __kvmalloc_node_noprof+0x2a4/0x710
   do_getxattr+0x4e/0x130
   path_getxattrat+0x125/0x1b0
   do_syscall_64+0x10a/0x480
   entry_SYSCALL_64_after_hwframe+0x4b/0x53
  RIP: 0033:0x7f3a22cb6daa
   </TASK>
  Modules linked in: kvm_amd kvm irqbypass vfat fat ccp k10temp sha3 libsha3 i2c_piix4 gq(O) cdc_acm xhci_pci xhci_hcd
  gsmi: Log Shutdown Reason 0x03
  CR2: ffff906ae30f0300
  ---[ end trace 0000000000000000 ]---
  RIP: 0010:memset+0xf/0x20
  Kernel panic - not syncing: Fatal exception
  Kernel Offset: 0x39e00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
  gsmi: Log Shutdown Reason 0x02

Fixes: 4c735bf1bc22 ("KVM: SEV: Allocate only as many bytes as needed for temp crypt buffers")
Cc: stable@vger.kernel.org
Cc: Michael Roth <michael.roth@amd.com>
Debugged-by: Michael Roth <michael.roth@amd.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
---
 arch/x86/kvm/svm/sev.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
index 74fb15551e83..e3733ed9d588 100644
--- a/arch/x86/kvm/svm/sev.c
+++ b/arch/x86/kvm/svm/sev.c
@@ -1280,6 +1280,17 @@ static void *sev_dbg_crypt_slow_alloc(struct page *page, unsigned long __va,
 	if (WARN_ON_ONCE((*pa & PAGE_MASK) != ((*pa + *nr_bytes - 1) & PAGE_MASK)))
 		return NULL;
 
+	/*
+	 * If SNP is enabled, i.e. the RMP is active, allocate a full page to
+	 * prevent concurrent accesses to the page.  As required by firmware,
+	 * the PSP driver updates the RMP to temporarily transfer ownership of
+	 * the page to Firmware while the {DE,EN}CRYPT operation is in-progress,
+	 * and so concurrent software accesses to the page will encounter
+	 * seemingly spurious RMP #PF violations
+	 */
+	if (cc_platform_has(CC_ATTR_HOST_SEV_SNP))
+		return kmalloc(PAGE_SIZE, GFP_KERNEL);
+
 	return kmalloc(*nr_bytes, GFP_KERNEL);
 }
 

base-commit: a204badd8432f93b7e862e7dac6db0fe3d65f370
-- 
2.55.0.229.g6434b31f56-goog

Re: [PATCH] KVM: SEV: Allocate full pages for {DE,EN}CRYPT ops on SNP-enabled hosts

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

New issues:
- [High] kmalloc(PAGE_SIZE) does not guarantee an isolated, page-aligned physical page under KASAN/SLUB debug, leading to Host RMP #PF crashes.

Pre-existing issues:
- [High] The PSP driver miscalculates the number of pages to transition to firmware-owned state for guest-owned buffers, causing firmware RMP violations.
--

--- Patch [1]: [PATCH] KVM: SEV: Allocate full pages for {DE,EN}CRYPT ops on SNP-enabled hosts ---
commit d5055bec1548a3803ac93e35043a743044a123b7
Author: Sean Christopherson <seanjc@google.com>

KVM: SEV: Allocate full pages for {DE,EN}CRYPT ops on SNP-enabled hosts

When {de,en}crypting memory of an SEV or SEV-ES guest on an SNP-enabled host
via a temporary buffer, this commit allocates a full 4KiB page for the buffer.
This ensures the page containing the buffer is wholly owned by KVM. It intends
to prevent concurrent kernel accesses from triggering spurious RMP #PF
violations when firmware temporarily takes ownership of the page.

> diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
> index 74fb15551e83f..e3733ed9d5886 100644
> --- a/arch/x86/kvm/svm/sev.c
> +++ b/arch/x86/kvm/svm/sev.c
> @@ -1280,6 +1280,17 @@ static void *sev_dbg_crypt_slow_alloc(struct page *page, unsigned long __va,
>  	if (WARN_ON_ONCE((*pa & PAGE_MASK) != ((*pa + *nr_bytes - 1) & PAGE_MASK)))
>  		return NULL;
>  
> +	/*
> +	 * If SNP is enabled, i.e. the RMP is active, allocate a full page to
> +	 * prevent concurrent accesses to the page.  As required by firmware,
> +	 * the PSP driver updates the RMP to temporarily transfer ownership of
> +	 * the page to Firmware while the {DE,EN}CRYPT operation is in-progress,
> +	 * and so concurrent software accesses to the page will encounter
> +	 * seemingly spurious RMP #PF violations
> +	 */
> +	if (cc_platform_has(CC_ATTR_HOST_SEV_SNP))
> +		return kmalloc(PAGE_SIZE, GFP_KERNEL);

[Severity: High]
Does kmalloc(PAGE_SIZE, GFP_KERNEL) guarantee an isolated, page-aligned
physical page when SLUB debugging (e.g., SLAB_RED_ZONE) or KASAN is enabled?

When debugging is enabled, kmalloc(PAGE_SIZE) requires extra metadata, pushing
the allocation into a higher-order slab cache (like kmalloc-8k). The resulting
buffer is not guaranteed to be page-aligned and shares its underlying physical
pages with adjacent slab objects, KASAN redzones, and SLUB metadata.

When sev_issue_dbg_cmd() passes this buffer as dst_addr, the PSP driver in
snp_map_cmd_buf_desc() transitions the page to firmware-owned. Concurrent
host kernel accesses to the shared page (e.g., KASAN background checks or
another CPU accessing adjacent objects) will trigger a fatal Host RMP #PF and
crash the host kernel.

Should this use a page allocation API like __get_free_page(GFP_KERNEL) or
alloc_page() instead to guarantee an exclusive physical page?

[Severity: High]
This is a pre-existing issue, but the PSP driver miscalculates the number of
pages to transition to firmware-owned state for guest-owned buffers in
snp_map_cmd_buf_desc():

drivers/crypto/ccp/sev-dev.c:snp_map_cmd_buf_desc() {
    ...
    npages = PAGE_ALIGN(desc->len) >> PAGE_SHIFT;
    if (rmp_mark_pages_firmware(*desc->paddr_ptr, npages, true)) {
    ...
}

Because this calculation ignores the physical address offset
(desc->paddr_ptr & ~PAGE_MASK), an unaligned buffer that crosses a page
boundary but has a total length <= PAGE_SIZE evaluates to 1 page instead of 2.

If the unaligned buffer from the KASAN-enabled kmalloc() above crosses a page
boundary, rmp_mark_pages_firmware() will only transition the first physical
page.

When the SEV firmware accesses the tail of the buffer on the untransitioned
second page, will this trigger a firmware RMP violation and cause the command
to fail?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260723000715.1235773-1-seanjc@google.com?part=1