* [PATCH v2] x86/sev: Skip DR7 write during kexec when it would trigger an unserviceable #VC
@ 2026-11-16 23:03 Ashish Kalra
0 siblings, 0 replies; only message in thread
From: Ashish Kalra @ 2026-11-16 23:03 UTC (permalink / raw)
To: tglx, mingo, bp, dave.hansen, x86, hpa, thomas.lendacky, nikunj,
Srikanth.Aithal
Cc: kvm, linux-coco, linux-kernel
From: Ashish Kalra <ashish.kalra@amd.com>
machine_kexec() calls hw_breakpoint_disable(), which writes DR7. On
SEV-ES/SNP guests that do not have DebugSwap enabled, DR7 accesses are
intercepted and delivered to the guest as a #VC exception, which must be
serviced through the GHCB.
By the time machine_kexec() runs, snp_kexec_finish() (called from
native_machine_shutdown()) has already converted the per-CPU GHCBs and
the boot GHCB back to private and set boot_ghcb to NULL, tearing down the
GHCB infrastructure. The DR7 write therefore raises a #VC that
dereferences a NULL GHCB pointer, causing a page fault and the kexec to
fail.
The NULL boot_ghcb pointer is a result of commit 3645eb7e3915 ("x86/fred:
Fix early boot failures on SEV-ES/SNP guests"): that change makes
__sev_get_ghcb() return the now-NULL boot_ghcb in this window, turning
the previously-benign teardown-window #VC (a bogus VMGEXIT the host
rejected but survived) into a NULL dereference.
Skip the DR7 write once the GHCBs have been torn down; the kexec'ed
kernel re-initializes DR7.
Suggested-by: Tom Lendacky <thomas.lendacky@amd.com>
Fixes: 3074152e56c9 ("x86/sev: Convert shared memory back to private on kexec")
Cc: stable@vger.kernel.org
Signed-off-by: Ashish Kalra <ashish.kalra@amd.com>
---
v2:
- Move the check into hw_breakpoint_disable() through a new
snp_hw_breakpoint_disable() helper, so the SNP-specific logic lives in
SEV code (Borislav Petkov).
- Drop CC_ATTR_GUEST_DEBUG_VIRT and read DebugSwap from sev_status.
- Skip only once the GHCBs have been torn down, so use_temporary_mm()
keeps disabling breakpoints at runtime.
- Trim the commit message (Borislav Petkov).
arch/x86/coco/sev/core.c | 16 ++++++++++++++++
arch/x86/include/asm/debugreg.h | 4 ++++
arch/x86/include/asm/sev.h | 2 ++
3 files changed, 22 insertions(+)
diff --git a/arch/x86/coco/sev/core.c b/arch/x86/coco/sev/core.c
index eb2e853e950a..64a5a037a9c8 100644
--- a/arch/x86/coco/sev/core.c
+++ b/arch/x86/coco/sev/core.c
@@ -723,6 +723,22 @@ void snp_kexec_finish(void)
}
}
+/*
+ * Without DebugSwap, a DR7 write raises a #VC, which can no longer be serviced
+ * once snp_kexec_finish() has torn down the GHCBs. The kexec'ed kernel
+ * re-initializes DR7, so skip the write in that case.
+ */
+bool snp_hw_breakpoint_disable(void)
+{
+ if (!cc_platform_has(CC_ATTR_GUEST_STATE_ENCRYPT))
+ return false;
+
+ if (sev_status & MSR_AMD64_SNP_DEBUG_SWAP)
+ return false;
+
+ return !sev_cfg.ghcbs_initialized;
+}
+
#define __ATTR_BASE (SVM_SELECTOR_P_MASK | SVM_SELECTOR_S_MASK)
#define INIT_CS_ATTRIBS (__ATTR_BASE | SVM_SELECTOR_READ_MASK | SVM_SELECTOR_CODE_MASK)
#define INIT_DS_ATTRIBS (__ATTR_BASE | SVM_SELECTOR_WRITE_MASK)
diff --git a/arch/x86/include/asm/debugreg.h b/arch/x86/include/asm/debugreg.h
index 854d82b88ff4..af15e2c928db 100644
--- a/arch/x86/include/asm/debugreg.h
+++ b/arch/x86/include/asm/debugreg.h
@@ -8,6 +8,7 @@
#include <asm/cpufeature.h>
#include <asm/msr.h>
+#include <asm/sev.h>
/*
* Define bits that are always set to 1 in DR7, only bit 10 is
@@ -108,6 +109,9 @@ static __always_inline void native_set_debugreg(int regno, unsigned long value)
static inline void hw_breakpoint_disable(void)
{
+ if (snp_hw_breakpoint_disable())
+ return;
+
/* Reset the control register for HW Breakpoint */
set_debugreg(DR7_FIXED_1, 7);
diff --git a/arch/x86/include/asm/sev.h b/arch/x86/include/asm/sev.h
index c7ea9845351e..f38bb521987d 100644
--- a/arch/x86/include/asm/sev.h
+++ b/arch/x86/include/asm/sev.h
@@ -522,6 +522,7 @@ int prepare_pte_enc(struct pte_enc_desc *d);
void set_pte_enc_mask(pte_t *kpte, unsigned long pfn, pgprot_t new_prot);
void snp_kexec_finish(void);
void snp_kexec_begin(void);
+bool snp_hw_breakpoint_disable(void);
int snp_msg_init(struct snp_msg_desc *mdesc, int vmpck_id);
struct snp_msg_desc *snp_msg_alloc(void);
@@ -632,6 +633,7 @@ static inline int prepare_pte_enc(struct pte_enc_desc *d) { return 0; }
static inline void set_pte_enc_mask(pte_t *kpte, unsigned long pfn, pgprot_t new_prot) { }
static inline void snp_kexec_finish(void) { }
static inline void snp_kexec_begin(void) { }
+static inline bool snp_hw_breakpoint_disable(void) { return false; }
static inline int snp_msg_init(struct snp_msg_desc *mdesc, int vmpck_id) { return -1; }
static inline struct snp_msg_desc *snp_msg_alloc(void) { return NULL; }
static inline void snp_msg_free(struct snp_msg_desc *mdesc) { }
base-commit: 9f74092fda91ef93c2b07585785c1dc83cd7971f
--
2.43.0
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2026-11-16 23:03 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-16 23:03 [PATCH v2] x86/sev: Skip DR7 write during kexec when it would trigger an unserviceable #VC Ashish Kalra
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®