The register is not defined and reading it can result in a UBSAN out-of-bounds array access error, specifically when the srcreg field value is 31.
Cc: Kristina Martsenko kristina.martsenko@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Mark Rutland mark.rutland@arm.com Cc: Will Deacon will@kernel.org Cc: Marc Zyngier maz@kernel.org Cc: stable@vger.kernel.org Signed-off-by: Keir Fraser keirf@google.com --- arch/arm64/include/asm/traps.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/include/asm/traps.h b/arch/arm64/include/asm/traps.h index d780d1bd2eac..82cf1f879c61 100644 --- a/arch/arm64/include/asm/traps.h +++ b/arch/arm64/include/asm/traps.h @@ -109,10 +109,9 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon int dstreg = ESR_ELx_MOPS_ISS_DESTREG(esr); int srcreg = ESR_ELx_MOPS_ISS_SRCREG(esr); int sizereg = ESR_ELx_MOPS_ISS_SIZEREG(esr); - unsigned long dst, src, size; + unsigned long dst, size;
dst = regs->regs[dstreg]; - src = regs->regs[srcreg]; size = regs->regs[sizereg];
/* @@ -129,6 +128,7 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon } } else { /* CPY* instruction */ + unsigned long src = regs->regs[srcreg]; if (!(option_a ^ wrong_option)) { /* Format is from Option B */ if (regs->pstate & PSR_N_BIT) {
On Wed, 26 Mar 2025 07:02:55 +0000, Keir Fraser keirf@google.com wrote:
The register is not defined and reading it can result in a UBSAN out-of-bounds array access error, specifically when the srcreg field value is 31.
Gah, XZR/SP encoding strikes back...
Cc: Kristina Martsenko kristina.martsenko@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Mark Rutland mark.rutland@arm.com Cc: Will Deacon will@kernel.org Cc: Marc Zyngier maz@kernel.org Cc: stable@vger.kernel.org Signed-off-by: Keir Fraser keirf@google.com
arch/arm64/include/asm/traps.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/include/asm/traps.h b/arch/arm64/include/asm/traps.h index d780d1bd2eac..82cf1f879c61 100644 --- a/arch/arm64/include/asm/traps.h +++ b/arch/arm64/include/asm/traps.h @@ -109,10 +109,9 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon int dstreg = ESR_ELx_MOPS_ISS_DESTREG(esr); int srcreg = ESR_ELx_MOPS_ISS_SRCREG(esr); int sizereg = ESR_ELx_MOPS_ISS_SIZEREG(esr);
- unsigned long dst, src, size;
- unsigned long dst, size;
dst = regs->regs[dstreg];
- src = regs->regs[srcreg]; size = regs->regs[sizereg];
/* @@ -129,6 +128,7 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon } } else { /* CPY* instruction */
unsigned long src = regs->regs[srcreg];
Reviewed-by: Marc Zyngier maz@kernel.org
M.
On Wed, Mar 26, 2025 at 07:02:55AM +0000, Keir Fraser wrote:
The register is not defined and reading it can result in a UBSAN out-of-bounds array access error, specifically when the srcreg field value is 31.
I'm assuming this is for a MOPS exception taken from a SET* sequence with XZR as the source?
It'd be nice to say that explicitly, as this is the only case where any of the src/dst/size fields in the ESR can be reported as 31. In all other cases where a CPY* or SET* instruction takes register 31 as an argument, the behaviour is CONSTRAINED UNPREDICTABLE and cannot generate a MOPS exception.
Note that in ARM DDI 0487 L.a there's a bug where:
* The prose says that SET* taking XZR as a src is CONSTRAINED UNPREDICTABLE, per K1.2.17.1.1 linked from C6.2.332.
The title for the K1.2.17.1.1 is "Memory Copy and Memory Set CPY*", which looks like an editing error.
* The pseudocode is explicit that the CONSTRAINED UNPREDICTABLE behaviours differ for CPY* and SET* per J1.1.3.121 CheckCPYConstrainedUnpredictable() and J1.1.3.125 CheckSETConstrainedUnpredictable().
... and I'll go file a ticket about that soon if someone doesn't beat me to it.
Cc: Kristina Martsenko kristina.martsenko@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Mark Rutland mark.rutland@arm.com Cc: Will Deacon will@kernel.org Cc: Marc Zyngier maz@kernel.org Cc: stable@vger.kernel.org
Looks like this should have:
Fixes: 2de451a329cf662b ("KVM: arm64: Add handler for MOPS exceptions")
Prior to that, the code in do_el0_mops() was benign as the use of pt_regs_read_reg() prevented the out-of-bounds access. It'd also be nice to note that in the commit message.
Mark.
Signed-off-by: Keir Fraser keirf@google.com
arch/arm64/include/asm/traps.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/include/asm/traps.h b/arch/arm64/include/asm/traps.h index d780d1bd2eac..82cf1f879c61 100644 --- a/arch/arm64/include/asm/traps.h +++ b/arch/arm64/include/asm/traps.h @@ -109,10 +109,9 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon int dstreg = ESR_ELx_MOPS_ISS_DESTREG(esr); int srcreg = ESR_ELx_MOPS_ISS_SRCREG(esr); int sizereg = ESR_ELx_MOPS_ISS_SIZEREG(esr);
- unsigned long dst, src, size;
- unsigned long dst, size;
dst = regs->regs[dstreg];
- src = regs->regs[srcreg]; size = regs->regs[sizereg];
/* @@ -129,6 +128,7 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon } } else { /* CPY* instruction */
unsigned long src = regs->regs[srcreg];-- 2.49.0.395.g12beb8f557-goog
On Wed, Mar 26, 2025 at 10:26:47AM +0000, Mark Rutland wrote:
On Wed, Mar 26, 2025 at 07:02:55AM +0000, Keir Fraser wrote:
The register is not defined and reading it can result in a UBSAN out-of-bounds array access error, specifically when the srcreg field value is 31.
I'm assuming this is for a MOPS exception taken from a SET* sequence with XZR as the source?
Yes.
It'd be nice to say that explicitly, as this is the only case where any of the src/dst/size fields in the ESR can be reported as 31. In all other cases where a CPY* or SET* instruction takes register 31 as an argument, the behaviour is CONSTRAINED UNPREDICTABLE and cannot generate a MOPS exception.
Okay, will do.
Note that in ARM DDI 0487 L.a there's a bug where:
The prose says that SET* taking XZR as a src is CONSTRAINED UNPREDICTABLE, per K1.2.17.1.1 linked from C6.2.332.
The title for the K1.2.17.1.1 is "Memory Copy and Memory Set CPY*", which looks like an editing error.
The pseudocode is explicit that the CONSTRAINED UNPREDICTABLE behaviours differ for CPY* and SET* per J1.1.3.121 CheckCPYConstrainedUnpredictable() and J1.1.3.125 CheckSETConstrainedUnpredictable().
... and I'll go file a ticket about that soon if someone doesn't beat me to it.
Cc: Kristina Martsenko kristina.martsenko@arm.com Cc: Catalin Marinas catalin.marinas@arm.com Cc: Mark Rutland mark.rutland@arm.com Cc: Will Deacon will@kernel.org Cc: Marc Zyngier maz@kernel.org Cc: stable@vger.kernel.org
Looks like this should have:
Fixes: 2de451a329cf662b ("KVM: arm64: Add handler for MOPS exceptions")
Prior to that, the code in do_el0_mops() was benign as the use of pt_regs_read_reg() prevented the out-of-bounds access. It'd also be nice to note that in the commit message.
I will add this too. And Marc's reviewed-by. And re-send as v2. Thanks!
Keir
Mark.
Signed-off-by: Keir Fraser keirf@google.com
arch/arm64/include/asm/traps.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/include/asm/traps.h b/arch/arm64/include/asm/traps.h index d780d1bd2eac..82cf1f879c61 100644 --- a/arch/arm64/include/asm/traps.h +++ b/arch/arm64/include/asm/traps.h @@ -109,10 +109,9 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon int dstreg = ESR_ELx_MOPS_ISS_DESTREG(esr); int srcreg = ESR_ELx_MOPS_ISS_SRCREG(esr); int sizereg = ESR_ELx_MOPS_ISS_SIZEREG(esr);
- unsigned long dst, src, size;
- unsigned long dst, size;
dst = regs->regs[dstreg];
- src = regs->regs[srcreg]; size = regs->regs[sizereg];
/* @@ -129,6 +128,7 @@ static inline void arm64_mops_reset_regs(struct user_pt_regs *regs, unsigned lon } } else { /* CPY* instruction */
unsigned long src = regs->regs[srcreg];-- 2.49.0.395.g12beb8f557-goog
linux-stable-mirror@lists.linaro.org
-
Keir Fraser -
Marc Zyngier -
Mark Rutland