From: Anup Patel <anup.patel@oss.qualcomm.com>
To: Paolo Bonzini <pbonzini@redhat.com>, Atish Patra <atish.patra@linux.dev>
Cc: Palmer Dabbelt <palmer@dabbelt.com>,
Paul Walmsley <pjw@kernel.org>, Alexandre Ghiti <alex@ghiti.fr>,
Shuah Khan <shuah@kernel.org>, Anup Patel <anup@brainfault.org>,
Andrew Jones <andrew.jones@oss.qualcomm.com>,
kvm-riscv@lists.infradead.org, kvm@vger.kernel.org,
linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org,
linux-kselftest@vger.kernel.org,
Anup Patel <anup.patel@oss.qualcomm.com>
Subject: [PATCH 18/27] RISC-V: KVM: Trap-n-emulate SRET for Guest HS-mode
Date: Tue, 20 Jan 2026 13:30:04 +0530 [thread overview]
Message-ID: <20260120080013.2153519-19-anup.patel@oss.qualcomm.com> (raw)
In-Reply-To: <20260120080013.2153519-1-anup.patel@oss.qualcomm.com>
The guest HS-mode (aka L1/guest hypervisor) can enter Guest VS/VU-mode
(aka L2/nested guest) using the SRET instruction. To achieve this, host
host hypervisor must trap-n-emulate SRET instruction for guest HS-mode
(aka L1/guest hypervisor) using host hstatus.VTSR bit.
Trapping all SRET instructions executed by guest HS-mode (aka L1/guest
hypervisor) will impact performance so host hypervisor will only set
hstatus.VTSR bit when guest sets HSTATUS.SPV bit and hstatus.VTSR bit
is cleared by SRET emulation upon entry into guest VS/VU-mode.
Signed-off-by: Anup Patel <anup.patel@oss.qualcomm.com>
---
arch/riscv/include/asm/insn.h | 3 ++
arch/riscv/include/asm/kvm_vcpu_nested.h | 2 +
arch/riscv/kvm/Makefile | 1 +
arch/riscv/kvm/vcpu_insn.c | 6 +++
arch/riscv/kvm/vcpu_nested_insn.c | 54 ++++++++++++++++++++++++
5 files changed, 66 insertions(+)
create mode 100644 arch/riscv/kvm/vcpu_nested_insn.c
diff --git a/arch/riscv/include/asm/insn.h b/arch/riscv/include/asm/insn.h
index c3005573e8c9..24a8abb3283c 100644
--- a/arch/riscv/include/asm/insn.h
+++ b/arch/riscv/include/asm/insn.h
@@ -331,6 +331,9 @@ static __always_inline bool riscv_insn_is_c_jalr(u32 code)
#define INSN_OPCODE_SHIFT 2
#define INSN_OPCODE_SYSTEM 28
+#define INSN_MASK_SRET 0xffffffff
+#define INSN_MATCH_SRET 0x10200073
+
#define INSN_MASK_WFI 0xffffffff
#define INSN_MATCH_WFI 0x10500073
diff --git a/arch/riscv/include/asm/kvm_vcpu_nested.h b/arch/riscv/include/asm/kvm_vcpu_nested.h
index 6d9d252a378c..665c60f09ee6 100644
--- a/arch/riscv/include/asm/kvm_vcpu_nested.h
+++ b/arch/riscv/include/asm/kvm_vcpu_nested.h
@@ -63,6 +63,8 @@ struct kvm_vcpu_nested {
#define kvm_riscv_vcpu_nested_virt(__vcpu) ((__vcpu)->arch.nested.virt)
+int kvm_riscv_vcpu_nested_insn_sret(struct kvm_vcpu *vcpu, struct kvm_run *run, ulong insn);
+
int kvm_riscv_vcpu_nested_swtlb_xlate(struct kvm_vcpu *vcpu,
const struct kvm_cpu_trap *trap,
struct kvm_gstage_mapping *out_map,
diff --git a/arch/riscv/kvm/Makefile b/arch/riscv/kvm/Makefile
index a8806b69205f..c0534d4a469e 100644
--- a/arch/riscv/kvm/Makefile
+++ b/arch/riscv/kvm/Makefile
@@ -26,6 +26,7 @@ kvm-y += vcpu_exit.o
kvm-y += vcpu_fp.o
kvm-y += vcpu_insn.o
kvm-y += vcpu_nested.o
+kvm-y += vcpu_nested_insn.o
kvm-y += vcpu_nested_swtlb.o
kvm-y += vcpu_onereg.o
kvm-$(CONFIG_RISCV_PMU_SBI) += vcpu_pmu.o
diff --git a/arch/riscv/kvm/vcpu_insn.c b/arch/riscv/kvm/vcpu_insn.c
index 4d89b94128ae..745cd654df94 100644
--- a/arch/riscv/kvm/vcpu_insn.c
+++ b/arch/riscv/kvm/vcpu_insn.c
@@ -9,6 +9,7 @@
#include <asm/cpufeature.h>
#include <asm/insn.h>
+#include <asm/kvm_vcpu_nested.h>
struct insn_func {
unsigned long mask;
@@ -257,6 +258,11 @@ static const struct insn_func system_opcode_funcs[] = {
.match = INSN_MATCH_CSRRCI,
.func = csr_insn,
},
+ {
+ .mask = INSN_MASK_SRET,
+ .match = INSN_MATCH_SRET,
+ .func = kvm_riscv_vcpu_nested_insn_sret,
+ },
{
.mask = INSN_MASK_WFI,
.match = INSN_MATCH_WFI,
diff --git a/arch/riscv/kvm/vcpu_nested_insn.c b/arch/riscv/kvm/vcpu_nested_insn.c
new file mode 100644
index 000000000000..8f5b2992dbb9
--- /dev/null
+++ b/arch/riscv/kvm/vcpu_nested_insn.c
@@ -0,0 +1,54 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (c) 2026 Qualcomm Technologies, Inc.
+ */
+
+#include <linux/kvm_host.h>
+#include <asm/kvm_nacl.h>
+#include <asm/kvm_vcpu_insn.h>
+
+int kvm_riscv_vcpu_nested_insn_sret(struct kvm_vcpu *vcpu, struct kvm_run *run, ulong insn)
+{
+ unsigned long vsstatus, next_sepc, next_spp;
+ bool next_virt;
+
+ /*
+ * Trap from virtual-VS/VU modes should be forwarded to
+ * virtual-HS mode as a virtual instruction trap.
+ */
+ if (kvm_riscv_vcpu_nested_virt(vcpu))
+ return KVM_INSN_VIRTUAL_TRAP;
+
+ /*
+ * Trap from virtual-U mode should be forwarded to
+ * virtual-HS mode as illegal instruction trap.
+ */
+ if (!(vcpu->arch.guest_context.hstatus & HSTATUS_SPVP))
+ return KVM_INSN_ILLEGAL_TRAP;
+
+ vsstatus = ncsr_read(CSR_VSSTATUS);
+
+ /*
+ * Find next nested virtualization mode, next privilege mode,
+ * and next sepc
+ */
+ next_virt = (vcpu->arch.nested.csr.hstatus & HSTATUS_SPV) ? true : false;
+ next_sepc = ncsr_read(CSR_VSEPC);
+ next_spp = vsstatus & SR_SPP;
+
+ /* Update Guest sstatus.sie */
+ vsstatus &= ~SR_SIE;
+ vsstatus |= (vsstatus & SR_SPIE) ? SR_SIE : 0;
+ ncsr_write(CSR_VSSTATUS, vsstatus);
+
+ /* Update return address and return privilege mode*/
+ vcpu->arch.guest_context.sepc = next_sepc;
+ vcpu->arch.guest_context.sstatus &= ~SR_SPP;
+ vcpu->arch.guest_context.sstatus |= next_spp;
+
+ /* Set nested virtualization state based on guest hstatus.SPV */
+ kvm_riscv_vcpu_nested_set_virt(vcpu, NESTED_SET_VIRT_EVENT_SRET,
+ next_virt, false, false);
+
+ return KVM_INSN_CONTINUE_SAME_SEPC;
+}
--
2.43.0
next prev parent reply other threads:[~2026-01-20 8:00 UTC|newest]
Thread overview: 38+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-01-20 7:59 [PATCH 00/27] Nested virtualization for KVM RISC-V Anup Patel
2026-01-20 7:59 ` [PATCH 01/27] RISC-V: KVM: Fix error code returned for Smstateen ONE_REG Anup Patel
2026-03-06 7:04 ` Anup Patel
2026-01-20 7:59 ` [PATCH 02/27] RISC-V: KVM: Fix error code returned for Ssaia ONE_REG Anup Patel
2026-03-06 7:04 ` Anup Patel
2026-01-20 7:59 ` [PATCH 03/27] RISC-V: KVM: Check host Ssaia extension when creating AIA irqchip Anup Patel
2026-03-06 7:04 ` Anup Patel
2026-01-20 7:59 ` [PATCH 04/27] RISC-V: KVM: Introduce common kvm_riscv_isa_check_host() Anup Patel
2026-03-13 14:22 ` Radim Krčmář
2026-01-20 7:59 ` [PATCH 05/27] RISC-V: KVM: Factor-out ISA checks into separate sources Anup Patel
2026-03-13 14:14 ` Radim Krčmář
2026-04-03 12:34 ` Anup Patel
2026-01-20 7:59 ` [PATCH 06/27] RISC-V: KVM: Move timer state defines closer to struct in UAPI header Anup Patel
2026-01-20 7:59 ` [PATCH 07/27] RISC-V: KVM: Add hideleg to struct kvm_vcpu_config Anup Patel
2026-03-13 13:49 ` Radim Krčmář
2026-01-20 7:59 ` [PATCH 08/27] RISC-V: KVM: Factor-out VCPU config into separate sources Anup Patel
2026-03-13 13:46 ` Radim Krčmář
2026-01-20 7:59 ` [PATCH 09/27] RISC-V: KVM: Don't check hstateen0 when updating sstateen0 CSR Anup Patel
2026-03-13 13:27 ` Radim Krčmář
2026-01-20 7:59 ` [PATCH 10/27] RISC-V: KVM: Initial skeletal nested virtualization support Anup Patel
2026-01-20 7:59 ` [PATCH 11/27] RISC-V: KVM: Use half VMID space for nested guest Anup Patel
2026-01-20 7:59 ` [PATCH 12/27] RISC-V: KVM: Extend kvm_riscv_mmu_update_hgatp() for nested virtualization Anup Patel
2026-01-20 7:59 ` [PATCH 13/27] RISC-V: KVM: Extend kvm_riscv_vcpu_config_load() " Anup Patel
2026-01-20 8:00 ` [PATCH 14/27] RISC-V: KVM: Extend kvm_riscv_vcpu_update_timedelta() for nested virt Anup Patel
2026-01-20 8:00 ` [PATCH 15/27] RISC-V: KVM: Extend trap redirection for nested virtualization Anup Patel
2026-01-20 8:00 ` [PATCH 16/27] RISC-V: KVM: Check and inject nested virtual interrupts Anup Patel
2026-01-20 8:00 ` [PATCH 17/27] RISC-V: KVM: Extend kvm_riscv_isa_check_host() for nested virt Anup Patel
2026-01-20 8:00 ` Anup Patel [this message]
2026-01-20 8:00 ` [PATCH 19/27] RISC-V: KVM: Redirect nested supervisor ecall and breakpoint traps Anup Patel
2026-01-20 8:00 ` [PATCH 20/27] RISC-V: KVM: Redirect nested WFI and WRS traps Anup Patel
2026-01-20 8:00 ` [PATCH 21/27] RISC-V: KVM: Implement remote HFENCE SBI calls for guest Anup Patel
2026-01-20 8:00 ` [PATCH 22/27] RISC-V: KVM: Add CSR emulation for nested virtualization Anup Patel
2026-01-20 8:00 ` [PATCH 23/27] RISC-V: KVM: Add HFENCE " Anup Patel
2026-01-20 8:00 ` [PATCH 24/27] RISC-V: KVM: Add ONE_REG interface for nested virtualization state Anup Patel
2026-01-20 8:00 ` [PATCH 25/27] RISC-V: KVM: selftests: Add nested virt state to get-reg-list test Anup Patel
2026-01-20 8:00 ` [PATCH 26/27] RISC-V: KVM: Add ONE_REG interface for nested virtualization CSRs Anup Patel
2026-01-20 8:00 ` [PATCH 27/27] RISC-V: KVM: selftests: Add nested virt CSRs to get-reg-list test Anup Patel
2026-04-03 12:36 ` [PATCH 00/27] Nested virtualization for KVM RISC-V Anup Patel
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260120080013.2153519-19-anup.patel@oss.qualcomm.com \
--to=anup.patel@oss.qualcomm.com \
--cc=alex@ghiti.fr \
--cc=andrew.jones@oss.qualcomm.com \
--cc=anup@brainfault.org \
--cc=atish.patra@linux.dev \
--cc=kvm-riscv@lists.infradead.org \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-kselftest@vger.kernel.org \
--cc=linux-riscv@lists.infradead.org \
--cc=palmer@dabbelt.com \
--cc=pbonzini@redhat.com \
--cc=pjw@kernel.org \
--cc=shuah@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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®