The SBI IPI handler walks vCPUs with kvm_for_each_vcpu(), which
iterates by vcpu_idx (creation order) rather than vcpu_id order.
Since vcpu_id can be assigned out of order by userspace, a vCPU whose
hart_bit falls outside the XLEN-bit hart_mask range may be reached
before vCPUs the mask actually targets. In that case the handler jumps
to "done" and stops sending IPIs, leaving valid target vCPUs without
an interrupt.

Replace the early "goto done" with "continue" so vCPUs outside the
hart_mask range are skipped without aborting the loop.

Reproduced with a minimal userspace VMM driving KVM inside a QEMU
(RISC-V virt) guest: three vCPUs are created with ids 100, 0, 1 (in
creation order) and sbi_send_ipi(hart_mask=bit0, hbase=0) is called
from vcpu_id 1. Before this change the IPI is dropped and the ecall
returns SBI_ERR_INVALID_PARAM; with this change the IPI is delivered
and the ecall returns SBI_SUCCESS.

Fixes: 0611f78f83c9 ("riscv: KVM: Fix SBI IPI error generation")
Cc: stable@vger.kernel.org
Suggested-by: Andrew Jones <ajones@ventanamicro.com>
Signed-off-by: Xiaofeng Yuan <yuanxiaofeng@eswincomputing.com>
---
 arch/riscv/kvm/vcpu_sbi_replace.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

diff --git a/arch/riscv/kvm/vcpu_sbi_replace.c b/arch/riscv/kvm/vcpu_sbi_replace.c
index 506a510b6..c2f1f7f26 100644
--- a/arch/riscv/kvm/vcpu_sbi_replace.c
+++ b/arch/riscv/kvm/vcpu_sbi_replace.c
@@ -64,8 +64,17 @@ static int kvm_sbi_ext_ipi_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
 			if (tmp->vcpu_id < hbase)
 				continue;
 			hart_bit = tmp->vcpu_id - hbase;
+			/*
+			 * kvm_for_each_vcpu() walks kvm->vcpus[] by
+			 * vcpu_idx, i.e. the creation order, which has
+			 * nothing to do with the vcpu_id (hart id) space
+			 * that the SBI IPI operates on. vcpu_ids need not
+			 * increase along the iteration, so harts outside
+			 * the hart_mask window must be skipped instead of
+			 * aborting the loop.
+			 */
 			if (hart_bit >= __riscv_xlen)
-				goto done;
+				continue;
 			if (!(hmask & (1UL << hart_bit)))
 				continue;
 		}
@@ -76,7 +85,6 @@ static int kvm_sbi_ext_ipi_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
 		kvm_riscv_vcpu_pmu_incr_fw(tmp, SBI_PMU_FW_IPI_RCVD);
 	}
 
-done:
 	if (hbase != -1UL && (hmask ^ sentmask))
 		retdata->err_val = SBI_ERR_INVALID_PARAM;
 
-- 
2.34.1

On Thu Sep 17, 2026 at 8:20 PM +08, Xiaofeng Yuan wrote:
> The SBI IPI handler walks vCPUs with kvm_for_each_vcpu(), which
> iterates by vcpu_idx (creation order) rather than vcpu_id order.
> Since vcpu_id can be assigned out of order by userspace, a vCPU whose
> hart_bit falls outside the XLEN-bit hart_mask range may be reached
> before vCPUs the mask actually targets. In that case the handler jumps
> to "done" and stops sending IPIs, leaving valid target vCPUs without
> an interrupt.
>
> Replace the early "goto done" with "continue" so vCPUs outside the
> hart_mask range are skipped without aborting the loop.
>
> Reproduced with a minimal userspace VMM driving KVM inside a QEMU
> (RISC-V virt) guest: three vCPUs are created with ids 100, 0, 1 (in
> creation order) and sbi_send_ipi(hart_mask=bit0, hbase=0) is called
> from vcpu_id 1. Before this change the IPI is dropped and the ecall
> returns SBI_ERR_INVALID_PARAM; with this change the IPI is delivered
> and the ecall returns SBI_SUCCESS.
>
> Fixes: 0611f78f83c9 ("riscv: KVM: Fix SBI IPI error generation")
> Cc: stable@vger.kernel.org
> Suggested-by: Andrew Jones <ajones@ventanamicro.com>
> Signed-off-by: Xiaofeng Yuan <yuanxiaofeng@eswincomputing.com>
Reviewed-by: Troy Mitchell <troy.mitchell@linux.spacemit.com>

-- 
Troy Mitchell