diff mbox series

RISC-V: KVM: Fix out-of-bounds by 1

Message ID 20260228152226.2116895-1-ethantidmore06@gmail.com
State Not Applicable
Headers show
Series RISC-V: KVM: Fix out-of-bounds by 1 | expand

Commit Message

Ethan Tidmore Feb. 28, 2026, 3:22 p.m. UTC 
The array kvpmu->pmc is defined as:

struct kvm_pmc pmc[RISCV_KVM_MAX_COUNTERS];

So, accessing it with index RISCV_KVM_MAX_COUNTERS would be
out-of-bounds by 1.

Change index check from > to >=.

Detected by Smatch:
arch/riscv/kvm/vcpu_pmu.c:528 kvm_riscv_vcpu_pmu_ctr_info() error:
buffer overflow 'kvpmu->pmc' 64 <= 64

Fixes: 8f0153ecd3bf1 ("RISC-V: KVM: Add skeleton support for perf")
Signed-off-by: Ethan Tidmore <ethantidmore06@gmail.com>
---
 arch/riscv/kvm/vcpu_pmu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Anup Patel March 2, 2026, 1:13 p.m. UTC | #1 
On Sat, Feb 28, 2026 at 8:52 PM Ethan Tidmore <ethantidmore06@gmail.com> wrote:
>
> The array kvpmu->pmc is defined as:
>
> struct kvm_pmc pmc[RISCV_KVM_MAX_COUNTERS];
>
> So, accessing it with index RISCV_KVM_MAX_COUNTERS would be
> out-of-bounds by 1.
>
> Change index check from > to >=.
>
> Detected by Smatch:
> arch/riscv/kvm/vcpu_pmu.c:528 kvm_riscv_vcpu_pmu_ctr_info() error:
> buffer overflow 'kvpmu->pmc' 64 <= 64
>
> Fixes: 8f0153ecd3bf1 ("RISC-V: KVM: Add skeleton support for perf")
> Signed-off-by: Ethan Tidmore <ethantidmore06@gmail.com>

Radim had already sent a similar which was merged.
Refer, https://lore.kernel.org/r/20260227134617.23378-1-radim.krcmar@oss.qualcomm.com

Regards,
Anup

> ---
>  arch/riscv/kvm/vcpu_pmu.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c
> index 3a4d54aa96d8..51a12f90fb30 100644
> --- a/arch/riscv/kvm/vcpu_pmu.c
> +++ b/arch/riscv/kvm/vcpu_pmu.c
> @@ -520,7 +520,7 @@ int kvm_riscv_vcpu_pmu_ctr_info(struct kvm_vcpu *vcpu, unsigned long cidx,
>  {
>         struct kvm_pmu *kvpmu = vcpu_to_pmu(vcpu);
>
> -       if (cidx > RISCV_KVM_MAX_COUNTERS || cidx == 1) {
> +       if (cidx >= RISCV_KVM_MAX_COUNTERS || cidx == 1) {
>                 retdata->err_val = SBI_ERR_INVALID_PARAM;
>                 return 0;
>         }
> --
> 2.53.0
>
Ethan Tidmore March 18, 2026, 9:19 p.m. UTC | #2 
On Sat Feb 28, 2026 at 9:22 AM CST, Ethan Tidmore wrote:
> The array kvpmu->pmc is defined as:
>
> struct kvm_pmc pmc[RISCV_KVM_MAX_COUNTERS];
>
> So, accessing it with index RISCV_KVM_MAX_COUNTERS would be
> out-of-bounds by 1.
>
> Change index check from > to >=.
>
> Detected by Smatch:
> arch/riscv/kvm/vcpu_pmu.c:528 kvm_riscv_vcpu_pmu_ctr_info() error:
> buffer overflow 'kvpmu->pmc' 64 <= 64
>
> Fixes: 8f0153ecd3bf1 ("RISC-V: KVM: Add skeleton support for perf")
> Signed-off-by: Ethan Tidmore <ethantidmore06@gmail.com>
> ---

Friendly ping.

Thanks,

ET
patchwork-bot+linux-riscv@kernel.org March 24, 2026, 6:07 a.m. UTC | #3 
Hello:

This patch was applied to riscv/linux.git (for-next)
by Anup Patel <anup@brainfault.org>:

On Sat, 28 Feb 2026 09:22:26 -0600 you wrote:
> The array kvpmu->pmc is defined as:
> 
> struct kvm_pmc pmc[RISCV_KVM_MAX_COUNTERS];
> 
> So, accessing it with index RISCV_KVM_MAX_COUNTERS would be
> out-of-bounds by 1.
> 
> [...]

Here is the summary with links:
  - RISC-V: KVM: Fix out-of-bounds by 1
    https://git.kernel.org/riscv/c/5c1bb0787111

You are awesome, thank you!
diff mbox series

Patch

diff --git a/arch/riscv/kvm/vcpu_pmu.c b/arch/riscv/kvm/vcpu_pmu.c
index 3a4d54aa96d8..51a12f90fb30 100644
--- a/arch/riscv/kvm/vcpu_pmu.c
+++ b/arch/riscv/kvm/vcpu_pmu.c
@@ -520,7 +520,7 @@  int kvm_riscv_vcpu_pmu_ctr_info(struct kvm_vcpu *vcpu, unsigned long cidx,
 {
 	struct kvm_pmu *kvpmu = vcpu_to_pmu(vcpu);
 
-	if (cidx > RISCV_KVM_MAX_COUNTERS || cidx == 1) {
+	if (cidx >= RISCV_KVM_MAX_COUNTERS || cidx == 1) {
 		retdata->err_val = SBI_ERR_INVALID_PARAM;
 		return 0;
 	}