* [PATCH v3 00/12] MMU related improvements for KVM RISC-V
@ 2025-06-18 11:35 Anup Patel
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
` (13 more replies)
0 siblings, 14 replies; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel
This series primarily has various MMU improvements for KVM RISC-V
and it also serves as a preparatory series for the upcoming nested
virtualization support.
PATCH1 to PATCH4: Few cosmetic improvements
PATCH5 to PATCH7: TLB maintenance and SBI NACL related improvements
PATCH8 to PATCH12: MMU related preparatory work for nested virtualization
These patches can also be found in the riscv_kvm_mmu_imp_v3 branch
at: https://github.com/avpatel/linux.git
Changes since v2:
- Rebased upon Linux-6.16-rc2 and latest riscv_kvm_queue branch
- Added Reviewed-by tags to appropriate patches
Changes since v1:
- Rebased upon Linux-6.16-rc1
- Dropped PATCH1 and PATCH2 of v1 series since these are queued
as fixes for Linux-6.16
- Addressed Atish's comment on PATCH1 in this series
- Added new PATCH7 in this series
Anup Patel (12):
RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize()
RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with
KVM_REQ_TLB_FLUSH
RISC-V: KVM: Don't flush TLB when PTE is unchanged
RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range()
RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect()
RISC-V: KVM: Factor-out MMU related declarations into separate headers
RISC-V: KVM: Introduce struct kvm_gstage_mapping
RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence
RISC-V: KVM: Factor-out g-stage page table management
RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs
arch/riscv/include/asm/kvm_aia.h | 2 +-
arch/riscv/include/asm/kvm_gstage.h | 72 ++++
arch/riscv/include/asm/kvm_host.h | 103 +-----
arch/riscv/include/asm/kvm_mmu.h | 21 ++
arch/riscv/include/asm/kvm_tlb.h | 84 +++++
arch/riscv/include/asm/kvm_vmid.h | 27 ++
arch/riscv/kvm/Makefile | 1 +
arch/riscv/kvm/aia_device.c | 6 +-
arch/riscv/kvm/aia_imsic.c | 12 +-
arch/riscv/kvm/gstage.c | 338 +++++++++++++++++++
arch/riscv/kvm/main.c | 3 +-
arch/riscv/kvm/mmu.c | 499 ++++++----------------------
arch/riscv/kvm/tlb.c | 110 +++---
arch/riscv/kvm/vcpu.c | 26 +-
arch/riscv/kvm/vcpu_exit.c | 20 +-
arch/riscv/kvm/vcpu_sbi_replace.c | 17 +-
arch/riscv/kvm/vcpu_sbi_v01.c | 25 +-
arch/riscv/kvm/vm.c | 7 +-
arch/riscv/kvm/vmid.c | 25 ++
19 files changed, 795 insertions(+), 603 deletions(-)
create mode 100644 arch/riscv/include/asm/kvm_gstage.h
create mode 100644 arch/riscv/include/asm/kvm_mmu.h
create mode 100644 arch/riscv/include/asm/kvm_tlb.h
create mode 100644 arch/riscv/include/asm/kvm_vmid.h
create mode 100644 arch/riscv/kvm/gstage.c
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-24 3:32 ` Atish Patra
2025-06-25 6:48 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init() Anup Patel
` (12 subsequent siblings)
13 siblings, 2 replies; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel
The kvm_riscv_vcpu_alloc_vector_context() does return an error code
upon failure so don't ignore this in kvm_arch_vcpu_create().
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/kvm/vcpu.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index 303aa0a8a5a1..b467dc1f4c7f 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -148,8 +148,9 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
spin_lock_init(&vcpu->arch.reset_state.lock);
- if (kvm_riscv_vcpu_alloc_vector_context(vcpu))
- return -ENOMEM;
+ rc = kvm_riscv_vcpu_alloc_vector_context(vcpu);
+ if (rc)
+ return rc;
/* Setup VCPU timer */
kvm_riscv_vcpu_timer_init(vcpu);
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-24 3:35 ` Atish Patra
2025-06-25 6:50 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 03/12] RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize() Anup Patel
` (11 subsequent siblings)
13 siblings, 2 replies; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Nutty Liu
The kvm_riscv_vcpu_aia_init() does not return any failure so drop
the return value which is always zero.
Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_aia.h | 2 +-
arch/riscv/kvm/aia_device.c | 6 ++----
arch/riscv/kvm/vcpu.c | 4 +---
3 files changed, 4 insertions(+), 8 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_aia.h b/arch/riscv/include/asm/kvm_aia.h
index 3b643b9efc07..0a0f12496f00 100644
--- a/arch/riscv/include/asm/kvm_aia.h
+++ b/arch/riscv/include/asm/kvm_aia.h
@@ -147,7 +147,7 @@ int kvm_riscv_vcpu_aia_rmw_ireg(struct kvm_vcpu *vcpu, unsigned int csr_num,
int kvm_riscv_vcpu_aia_update(struct kvm_vcpu *vcpu);
void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu);
-int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
+void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu);
int kvm_riscv_aia_inject_msi_by_id(struct kvm *kvm, u32 hart_index,
diff --git a/arch/riscv/kvm/aia_device.c b/arch/riscv/kvm/aia_device.c
index 806c41931cde..b195a93add1c 100644
--- a/arch/riscv/kvm/aia_device.c
+++ b/arch/riscv/kvm/aia_device.c
@@ -509,12 +509,12 @@ void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu)
kvm_riscv_vcpu_aia_imsic_reset(vcpu);
}
-int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
+void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
{
struct kvm_vcpu_aia *vaia = &vcpu->arch.aia_context;
if (!kvm_riscv_aia_available())
- return 0;
+ return;
/*
* We don't do any memory allocations over here because these
@@ -526,8 +526,6 @@ int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
/* Initialize default values in AIA vcpu context */
vaia->imsic_addr = KVM_RISCV_AIA_UNDEF_ADDR;
vaia->hart_index = vcpu->vcpu_idx;
-
- return 0;
}
void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu)
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index b467dc1f4c7f..f9fb3dbbe0c3 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -159,9 +159,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
kvm_riscv_vcpu_pmu_init(vcpu);
/* Setup VCPU AIA */
- rc = kvm_riscv_vcpu_aia_init(vcpu);
- if (rc)
- return rc;
+ kvm_riscv_vcpu_aia_init(vcpu);
/*
* Setup SBI extensions
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 03/12] RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize()
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
2025-06-18 11:35 ` [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init() Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-18 11:35 ` [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH Anup Patel
` (10 subsequent siblings)
13 siblings, 0 replies; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra, Nutty Liu
The kvm_riscv_local_tlb_sanitize() deals with sanitizing current
VMID related TLB mappings when a VCPU is moved from one host CPU
to another.
Let's move kvm_riscv_local_tlb_sanitize() to VMID management
sources and rename it to kvm_riscv_gstage_vmid_sanitize().
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_host.h | 3 +--
arch/riscv/kvm/tlb.c | 23 -----------------------
arch/riscv/kvm/vcpu.c | 4 ++--
arch/riscv/kvm/vmid.c | 23 +++++++++++++++++++++++
4 files changed, 26 insertions(+), 27 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index 9a617bf5363d..8aa705ac75a5 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -331,8 +331,6 @@ void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
unsigned long order);
void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
-void kvm_riscv_local_tlb_sanitize(struct kvm_vcpu *vcpu);
-
void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu);
void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
@@ -380,6 +378,7 @@ unsigned long kvm_riscv_gstage_vmid_bits(void);
int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
+void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu);
int kvm_riscv_setup_default_irq_routing(struct kvm *kvm, u32 lines);
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index 2f91ea5f8493..b3461bfd9756 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -156,29 +156,6 @@ void kvm_riscv_local_hfence_vvma_all(unsigned long vmid)
csr_write(CSR_HGATP, hgatp);
}
-void kvm_riscv_local_tlb_sanitize(struct kvm_vcpu *vcpu)
-{
- unsigned long vmid;
-
- if (!kvm_riscv_gstage_vmid_bits() ||
- vcpu->arch.last_exit_cpu == vcpu->cpu)
- return;
-
- /*
- * On RISC-V platforms with hardware VMID support, we share same
- * VMID for all VCPUs of a particular Guest/VM. This means we might
- * have stale G-stage TLB entries on the current Host CPU due to
- * some other VCPU of the same Guest which ran previously on the
- * current Host CPU.
- *
- * To cleanup stale TLB entries, we simply flush all G-stage TLB
- * entries by VMID whenever underlying Host CPU changes for a VCPU.
- */
-
- vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
- kvm_riscv_local_hfence_gvma_vmid_all(vmid);
-}
-
void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu)
{
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_FENCE_I_RCVD);
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index f9fb3dbbe0c3..a2dd4161e5a4 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -962,12 +962,12 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu)
}
/*
- * Cleanup stale TLB enteries
+ * Sanitize VMID mappings cached (TLB) on current CPU
*
* Note: This should be done after G-stage VMID has been
* updated using kvm_riscv_gstage_vmid_ver_changed()
*/
- kvm_riscv_local_tlb_sanitize(vcpu);
+ kvm_riscv_gstage_vmid_sanitize(vcpu);
trace_kvm_entry(vcpu);
diff --git a/arch/riscv/kvm/vmid.c b/arch/riscv/kvm/vmid.c
index ddc98714ce8e..92c01255f86f 100644
--- a/arch/riscv/kvm/vmid.c
+++ b/arch/riscv/kvm/vmid.c
@@ -122,3 +122,26 @@ void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu)
kvm_for_each_vcpu(i, v, vcpu->kvm)
kvm_make_request(KVM_REQ_UPDATE_HGATP, v);
}
+
+void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu)
+{
+ unsigned long vmid;
+
+ if (!kvm_riscv_gstage_vmid_bits() ||
+ vcpu->arch.last_exit_cpu == vcpu->cpu)
+ return;
+
+ /*
+ * On RISC-V platforms with hardware VMID support, we share same
+ * VMID for all VCPUs of a particular Guest/VM. This means we might
+ * have stale G-stage TLB entries on the current Host CPU due to
+ * some other VCPU of the same Guest which ran previously on the
+ * current Host CPU.
+ *
+ * To cleanup stale TLB entries, we simply flush all G-stage TLB
+ * entries by VMID whenever underlying Host CPU changes for a VCPU.
+ */
+
+ vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
+ kvm_riscv_local_hfence_gvma_vmid_all(vmid);
+}
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (2 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 03/12] RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize() Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:35 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged Anup Patel
` (9 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
The KVM_REQ_HFENCE_GVMA_VMID_ALL is same as KVM_REQ_TLB_FLUSH so
to avoid confusion let's replace KVM_REQ_HFENCE_GVMA_VMID_ALL with
KVM_REQ_TLB_FLUSH. Also, rename kvm_riscv_hfence_gvma_vmid_all_process()
to kvm_riscv_tlb_flush_process().
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_host.h | 4 ++--
arch/riscv/kvm/tlb.c | 8 ++++----
arch/riscv/kvm/vcpu.c | 8 ++------
3 files changed, 8 insertions(+), 12 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index 8aa705ac75a5..ff1f76d6f177 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -37,7 +37,6 @@
#define KVM_REQ_UPDATE_HGATP KVM_ARCH_REQ(2)
#define KVM_REQ_FENCE_I \
KVM_ARCH_REQ_FLAGS(3, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
-#define KVM_REQ_HFENCE_GVMA_VMID_ALL KVM_REQ_TLB_FLUSH
#define KVM_REQ_HFENCE_VVMA_ALL \
KVM_ARCH_REQ_FLAGS(4, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
#define KVM_REQ_HFENCE \
@@ -331,8 +330,9 @@ void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
unsigned long order);
void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
+void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
+
void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
-void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu);
void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index b3461bfd9756..da98ca801d31 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -162,7 +162,7 @@ void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu)
local_flush_icache_all();
}
-void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu)
+void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu)
{
struct kvm_vmid *v = &vcpu->kvm->arch.vmid;
unsigned long vmid = READ_ONCE(v->vmid);
@@ -342,14 +342,14 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
data.size = gpsz;
data.order = order;
make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
- KVM_REQ_HFENCE_GVMA_VMID_ALL, &data);
+ KVM_REQ_TLB_FLUSH, &data);
}
void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
unsigned long hbase, unsigned long hmask)
{
- make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_GVMA_VMID_ALL,
- KVM_REQ_HFENCE_GVMA_VMID_ALL, NULL);
+ make_xfence_request(kvm, hbase, hmask, KVM_REQ_TLB_FLUSH,
+ KVM_REQ_TLB_FLUSH, NULL);
}
void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index a2dd4161e5a4..6eb11c913b13 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -721,12 +721,8 @@ static void kvm_riscv_check_vcpu_requests(struct kvm_vcpu *vcpu)
if (kvm_check_request(KVM_REQ_FENCE_I, vcpu))
kvm_riscv_fence_i_process(vcpu);
- /*
- * The generic KVM_REQ_TLB_FLUSH is same as
- * KVM_REQ_HFENCE_GVMA_VMID_ALL
- */
- if (kvm_check_request(KVM_REQ_HFENCE_GVMA_VMID_ALL, vcpu))
- kvm_riscv_hfence_gvma_vmid_all_process(vcpu);
+ if (kvm_check_request(KVM_REQ_TLB_FLUSH, vcpu))
+ kvm_riscv_tlb_flush_process(vcpu);
if (kvm_check_request(KVM_REQ_HFENCE_VVMA_ALL, vcpu))
kvm_riscv_hfence_vvma_all_process(vcpu);
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (3 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:38 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range() Anup Patel
` (8 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
The gstage_set_pte() and gstage_op_pte() should flush TLB only when
a leaf PTE changes so that unnecessary TLB flushes can be avoided.
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/kvm/mmu.c | 14 +++++++++-----
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index 1087ea74567b..29f1bd853a66 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -167,9 +167,11 @@ static int gstage_set_pte(struct kvm *kvm, u32 level,
ptep = &next_ptep[gstage_pte_index(addr, current_level)];
}
- set_pte(ptep, *new_pte);
- if (gstage_pte_leaf(ptep))
- gstage_remote_tlb_flush(kvm, current_level, addr);
+ if (pte_val(*ptep) != pte_val(*new_pte)) {
+ set_pte(ptep, *new_pte);
+ if (gstage_pte_leaf(ptep))
+ gstage_remote_tlb_flush(kvm, current_level, addr);
+ }
return 0;
}
@@ -229,7 +231,7 @@ static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
pte_t *ptep, u32 ptep_level, enum gstage_op op)
{
int i, ret;
- pte_t *next_ptep;
+ pte_t old_pte, *next_ptep;
u32 next_ptep_level;
unsigned long next_page_size, page_size;
@@ -258,11 +260,13 @@ static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
if (op == GSTAGE_OP_CLEAR)
put_page(virt_to_page(next_ptep));
} else {
+ old_pte = *ptep;
if (op == GSTAGE_OP_CLEAR)
set_pte(ptep, __pte(0));
else if (op == GSTAGE_OP_WP)
set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
- gstage_remote_tlb_flush(kvm, ptep_level, addr);
+ if (pte_val(*ptep) != pte_val(old_pte))
+ gstage_remote_tlb_flush(kvm, ptep_level, addr);
}
}
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range()
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (4 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:39 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect() Anup Patel
` (7 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
The kvm_arch_flush_remote_tlbs_range() expected by KVM core can be
easily implemented for RISC-V using kvm_riscv_hfence_gvma_vmid_gpa()
hence provide it.
Also with kvm_arch_flush_remote_tlbs_range() available for RISC-V, the
mmu_wp_memory_region() can happily use kvm_flush_remote_tlbs_memslot()
instead of kvm_flush_remote_tlbs().
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_host.h | 2 ++
arch/riscv/kvm/mmu.c | 2 +-
arch/riscv/kvm/tlb.c | 8 ++++++++
3 files changed, 11 insertions(+), 1 deletion(-)
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index ff1f76d6f177..6162575e2177 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -43,6 +43,8 @@
KVM_ARCH_REQ_FLAGS(5, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
#define KVM_REQ_STEAL_UPDATE KVM_ARCH_REQ(6)
+#define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLBS_RANGE
+
#define KVM_HEDELEG_DEFAULT (BIT(EXC_INST_MISALIGNED) | \
BIT(EXC_BREAKPOINT) | \
BIT(EXC_SYSCALL) | \
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index 29f1bd853a66..a5387927a1c1 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -344,7 +344,7 @@ static void gstage_wp_memory_region(struct kvm *kvm, int slot)
spin_lock(&kvm->mmu_lock);
gstage_wp_range(kvm, start, end);
spin_unlock(&kvm->mmu_lock);
- kvm_flush_remote_tlbs(kvm);
+ kvm_flush_remote_tlbs_memslot(kvm, memslot);
}
int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index da98ca801d31..f46a27658c2e 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -403,3 +403,11 @@ void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_VVMA_ALL,
KVM_REQ_HFENCE_VVMA_ALL, NULL);
}
+
+int kvm_arch_flush_remote_tlbs_range(struct kvm *kvm, gfn_t gfn, u64 nr_pages)
+{
+ kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0,
+ gfn << PAGE_SHIFT, nr_pages << PAGE_SHIFT,
+ PAGE_SHIFT);
+ return 0;
+}
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect()
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (5 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range() Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:39 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers Anup Patel
` (6 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
The H-extension CSRs accessed by kvm_riscv_vcpu_trap_redirect() will
trap when KVM RISC-V is running as Guest/VM hence remove these traps
by using ncsr_xyz() instead of csr_xyz().
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/kvm/vcpu_exit.c | 13 +++++++------
1 file changed, 7 insertions(+), 6 deletions(-)
diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
index 6e0c18412795..85c43c83e3b9 100644
--- a/arch/riscv/kvm/vcpu_exit.c
+++ b/arch/riscv/kvm/vcpu_exit.c
@@ -9,6 +9,7 @@
#include <linux/kvm_host.h>
#include <asm/csr.h>
#include <asm/insn-def.h>
+#include <asm/kvm_nacl.h>
static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
struct kvm_cpu_trap *trap)
@@ -135,7 +136,7 @@ unsigned long kvm_riscv_vcpu_unpriv_read(struct kvm_vcpu *vcpu,
void kvm_riscv_vcpu_trap_redirect(struct kvm_vcpu *vcpu,
struct kvm_cpu_trap *trap)
{
- unsigned long vsstatus = csr_read(CSR_VSSTATUS);
+ unsigned long vsstatus = ncsr_read(CSR_VSSTATUS);
/* Change Guest SSTATUS.SPP bit */
vsstatus &= ~SR_SPP;
@@ -151,15 +152,15 @@ void kvm_riscv_vcpu_trap_redirect(struct kvm_vcpu *vcpu,
vsstatus &= ~SR_SIE;
/* Update Guest SSTATUS */
- csr_write(CSR_VSSTATUS, vsstatus);
+ ncsr_write(CSR_VSSTATUS, vsstatus);
/* Update Guest SCAUSE, STVAL, and SEPC */
- csr_write(CSR_VSCAUSE, trap->scause);
- csr_write(CSR_VSTVAL, trap->stval);
- csr_write(CSR_VSEPC, trap->sepc);
+ ncsr_write(CSR_VSCAUSE, trap->scause);
+ ncsr_write(CSR_VSTVAL, trap->stval);
+ ncsr_write(CSR_VSEPC, trap->sepc);
/* Set Guest PC to Guest exception vector */
- vcpu->arch.guest_context.sepc = csr_read(CSR_VSTVEC);
+ vcpu->arch.guest_context.sepc = ncsr_read(CSR_VSTVEC);
/* Set Guest privilege mode to supervisor */
vcpu->arch.guest_context.sstatus |= SR_SPP;
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (6 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect() Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:40 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping Anup Patel
` (5 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
The MMU, TLB, and VMID management for KVM RISC-V already exists as
seprate sources so create separate headers along these lines. This
further simplifies asm/kvm_host.h header.
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_host.h | 100 +-----------------------------
arch/riscv/include/asm/kvm_mmu.h | 26 ++++++++
arch/riscv/include/asm/kvm_tlb.h | 78 +++++++++++++++++++++++
arch/riscv/include/asm/kvm_vmid.h | 27 ++++++++
arch/riscv/kvm/aia_imsic.c | 1 +
arch/riscv/kvm/main.c | 1 +
arch/riscv/kvm/mmu.c | 1 +
arch/riscv/kvm/tlb.c | 2 +
arch/riscv/kvm/vcpu.c | 1 +
arch/riscv/kvm/vcpu_exit.c | 1 +
arch/riscv/kvm/vm.c | 1 +
arch/riscv/kvm/vmid.c | 2 +
12 files changed, 143 insertions(+), 98 deletions(-)
create mode 100644 arch/riscv/include/asm/kvm_mmu.h
create mode 100644 arch/riscv/include/asm/kvm_tlb.h
create mode 100644 arch/riscv/include/asm/kvm_vmid.h
diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
index 6162575e2177..bd5341efa127 100644
--- a/arch/riscv/include/asm/kvm_host.h
+++ b/arch/riscv/include/asm/kvm_host.h
@@ -16,6 +16,8 @@
#include <asm/hwcap.h>
#include <asm/kvm_aia.h>
#include <asm/ptrace.h>
+#include <asm/kvm_tlb.h>
+#include <asm/kvm_vmid.h>
#include <asm/kvm_vcpu_fp.h>
#include <asm/kvm_vcpu_insn.h>
#include <asm/kvm_vcpu_sbi.h>
@@ -56,24 +58,6 @@
BIT(IRQ_VS_TIMER) | \
BIT(IRQ_VS_EXT))
-enum kvm_riscv_hfence_type {
- KVM_RISCV_HFENCE_UNKNOWN = 0,
- KVM_RISCV_HFENCE_GVMA_VMID_GPA,
- KVM_RISCV_HFENCE_VVMA_ASID_GVA,
- KVM_RISCV_HFENCE_VVMA_ASID_ALL,
- KVM_RISCV_HFENCE_VVMA_GVA,
-};
-
-struct kvm_riscv_hfence {
- enum kvm_riscv_hfence_type type;
- unsigned long asid;
- unsigned long order;
- gpa_t addr;
- gpa_t size;
-};
-
-#define KVM_RISCV_VCPU_MAX_HFENCE 64
-
struct kvm_vm_stat {
struct kvm_vm_stat_generic generic;
};
@@ -99,15 +83,6 @@ struct kvm_vcpu_stat {
struct kvm_arch_memory_slot {
};
-struct kvm_vmid {
- /*
- * Writes to vmid_version and vmid happen with vmid_lock held
- * whereas reads happen without any lock held.
- */
- unsigned long vmid_version;
- unsigned long vmid;
-};
-
struct kvm_arch {
/* G-stage vmid */
struct kvm_vmid vmid;
@@ -311,77 +286,6 @@ static inline bool kvm_arch_pmi_in_guest(struct kvm_vcpu *vcpu)
return IS_ENABLED(CONFIG_GUEST_PERF_EVENTS) && !!vcpu;
}
-#define KVM_RISCV_GSTAGE_TLB_MIN_ORDER 12
-
-void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
- gpa_t gpa, gpa_t gpsz,
- unsigned long order);
-void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid);
-void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
- unsigned long order);
-void kvm_riscv_local_hfence_gvma_all(void);
-void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
- unsigned long asid,
- unsigned long gva,
- unsigned long gvsz,
- unsigned long order);
-void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
- unsigned long asid);
-void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
- unsigned long gva, unsigned long gvsz,
- unsigned long order);
-void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
-
-void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
-
-void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
-void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
-void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
-
-void kvm_riscv_fence_i(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask);
-void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask,
- gpa_t gpa, gpa_t gpsz,
- unsigned long order);
-void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask);
-void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask,
- unsigned long gva, unsigned long gvsz,
- unsigned long order, unsigned long asid);
-void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask,
- unsigned long asid);
-void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask,
- unsigned long gva, unsigned long gvsz,
- unsigned long order);
-void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask);
-
-int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
- phys_addr_t hpa, unsigned long size,
- bool writable, bool in_atomic);
-void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
- unsigned long size);
-int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
- struct kvm_memory_slot *memslot,
- gpa_t gpa, unsigned long hva, bool is_write);
-int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
-void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
-void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
-void __init kvm_riscv_gstage_mode_detect(void);
-unsigned long __init kvm_riscv_gstage_mode(void);
-int kvm_riscv_gstage_gpa_bits(void);
-
-void __init kvm_riscv_gstage_vmid_detect(void);
-unsigned long kvm_riscv_gstage_vmid_bits(void);
-int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
-bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
-void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
-void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu);
-
int kvm_riscv_setup_default_irq_routing(struct kvm *kvm, u32 lines);
void __kvm_riscv_unpriv_trap(void);
diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
new file mode 100644
index 000000000000..4e1654282ee4
--- /dev/null
+++ b/arch/riscv/include/asm/kvm_mmu.h
@@ -0,0 +1,26 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2025 Ventana Micro Systems Inc.
+ */
+
+#ifndef __RISCV_KVM_MMU_H_
+#define __RISCV_KVM_MMU_H_
+
+#include <linux/kvm_types.h>
+
+int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
+ phys_addr_t hpa, unsigned long size,
+ bool writable, bool in_atomic);
+void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
+ unsigned long size);
+int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
+ struct kvm_memory_slot *memslot,
+ gpa_t gpa, unsigned long hva, bool is_write);
+int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
+void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
+void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
+void kvm_riscv_gstage_mode_detect(void);
+unsigned long kvm_riscv_gstage_mode(void);
+int kvm_riscv_gstage_gpa_bits(void);
+
+#endif
diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
new file mode 100644
index 000000000000..cd00c9a46cb1
--- /dev/null
+++ b/arch/riscv/include/asm/kvm_tlb.h
@@ -0,0 +1,78 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2025 Ventana Micro Systems Inc.
+ */
+
+#ifndef __RISCV_KVM_TLB_H_
+#define __RISCV_KVM_TLB_H_
+
+#include <linux/kvm_types.h>
+
+enum kvm_riscv_hfence_type {
+ KVM_RISCV_HFENCE_UNKNOWN = 0,
+ KVM_RISCV_HFENCE_GVMA_VMID_GPA,
+ KVM_RISCV_HFENCE_VVMA_ASID_GVA,
+ KVM_RISCV_HFENCE_VVMA_ASID_ALL,
+ KVM_RISCV_HFENCE_VVMA_GVA,
+};
+
+struct kvm_riscv_hfence {
+ enum kvm_riscv_hfence_type type;
+ unsigned long asid;
+ unsigned long order;
+ gpa_t addr;
+ gpa_t size;
+};
+
+#define KVM_RISCV_VCPU_MAX_HFENCE 64
+
+#define KVM_RISCV_GSTAGE_TLB_MIN_ORDER 12
+
+void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
+ gpa_t gpa, gpa_t gpsz,
+ unsigned long order);
+void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid);
+void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
+ unsigned long order);
+void kvm_riscv_local_hfence_gvma_all(void);
+void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
+ unsigned long asid,
+ unsigned long gva,
+ unsigned long gvsz,
+ unsigned long order);
+void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
+ unsigned long asid);
+void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
+ unsigned long gva, unsigned long gvsz,
+ unsigned long order);
+void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
+
+void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
+
+void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
+void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
+void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
+
+void kvm_riscv_fence_i(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask);
+void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask,
+ gpa_t gpa, gpa_t gpsz,
+ unsigned long order);
+void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask);
+void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask,
+ unsigned long gva, unsigned long gvsz,
+ unsigned long order, unsigned long asid);
+void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask,
+ unsigned long asid);
+void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask,
+ unsigned long gva, unsigned long gvsz,
+ unsigned long order);
+void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
+ unsigned long hbase, unsigned long hmask);
+
+#endif
diff --git a/arch/riscv/include/asm/kvm_vmid.h b/arch/riscv/include/asm/kvm_vmid.h
new file mode 100644
index 000000000000..ab98e1434fb7
--- /dev/null
+++ b/arch/riscv/include/asm/kvm_vmid.h
@@ -0,0 +1,27 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (c) 2025 Ventana Micro Systems Inc.
+ */
+
+#ifndef __RISCV_KVM_VMID_H_
+#define __RISCV_KVM_VMID_H_
+
+#include <linux/kvm_types.h>
+
+struct kvm_vmid {
+ /*
+ * Writes to vmid_version and vmid happen with vmid_lock held
+ * whereas reads happen without any lock held.
+ */
+ unsigned long vmid_version;
+ unsigned long vmid;
+};
+
+void __init kvm_riscv_gstage_vmid_detect(void);
+unsigned long kvm_riscv_gstage_vmid_bits(void);
+int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
+bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
+void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
+void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu);
+
+#endif
diff --git a/arch/riscv/kvm/aia_imsic.c b/arch/riscv/kvm/aia_imsic.c
index 29ef9c2133a9..40b469c0a01f 100644
--- a/arch/riscv/kvm/aia_imsic.c
+++ b/arch/riscv/kvm/aia_imsic.c
@@ -16,6 +16,7 @@
#include <linux/swab.h>
#include <kvm/iodev.h>
#include <asm/csr.h>
+#include <asm/kvm_mmu.h>
#define IMSIC_MAX_EIX (IMSIC_MAX_ID / BITS_PER_TYPE(u64))
diff --git a/arch/riscv/kvm/main.c b/arch/riscv/kvm/main.c
index 4b24705dc63a..b861a5dd7bd9 100644
--- a/arch/riscv/kvm/main.c
+++ b/arch/riscv/kvm/main.c
@@ -11,6 +11,7 @@
#include <linux/module.h>
#include <linux/kvm_host.h>
#include <asm/cpufeature.h>
+#include <asm/kvm_mmu.h>
#include <asm/kvm_nacl.h>
#include <asm/sbi.h>
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index a5387927a1c1..c1a3eb076df3 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -15,6 +15,7 @@
#include <linux/vmalloc.h>
#include <linux/kvm_host.h>
#include <linux/sched/signal.h>
+#include <asm/kvm_mmu.h>
#include <asm/kvm_nacl.h>
#include <asm/page.h>
#include <asm/pgtable.h>
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index f46a27658c2e..6fc4361c3d75 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -15,6 +15,8 @@
#include <asm/cpufeature.h>
#include <asm/insn-def.h>
#include <asm/kvm_nacl.h>
+#include <asm/kvm_tlb.h>
+#include <asm/kvm_vmid.h>
#define has_svinval() riscv_has_extension_unlikely(RISCV_ISA_EXT_SVINVAL)
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index 6eb11c913b13..8ad7b31f5939 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -18,6 +18,7 @@
#include <linux/fs.h>
#include <linux/kvm_host.h>
#include <asm/cacheflush.h>
+#include <asm/kvm_mmu.h>
#include <asm/kvm_nacl.h>
#include <asm/kvm_vcpu_vector.h>
diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
index 85c43c83e3b9..965df528de90 100644
--- a/arch/riscv/kvm/vcpu_exit.c
+++ b/arch/riscv/kvm/vcpu_exit.c
@@ -9,6 +9,7 @@
#include <linux/kvm_host.h>
#include <asm/csr.h>
#include <asm/insn-def.h>
+#include <asm/kvm_mmu.h>
#include <asm/kvm_nacl.h>
static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c
index b27ec8f96697..8601cf29e5f8 100644
--- a/arch/riscv/kvm/vm.c
+++ b/arch/riscv/kvm/vm.c
@@ -11,6 +11,7 @@
#include <linux/module.h>
#include <linux/uaccess.h>
#include <linux/kvm_host.h>
+#include <asm/kvm_mmu.h>
const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
KVM_GENERIC_VM_STATS()
diff --git a/arch/riscv/kvm/vmid.c b/arch/riscv/kvm/vmid.c
index 92c01255f86f..3b426c800480 100644
--- a/arch/riscv/kvm/vmid.c
+++ b/arch/riscv/kvm/vmid.c
@@ -14,6 +14,8 @@
#include <linux/smp.h>
#include <linux/kvm_host.h>
#include <asm/csr.h>
+#include <asm/kvm_tlb.h>
+#include <asm/kvm_vmid.h>
static unsigned long vmid_version = 1;
static unsigned long vmid_next;
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (7 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:43 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence Anup Patel
` (4 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
Introduce struct kvm_gstage_mapping which represents a g-stage
mapping at a particular g-stage page table level. Also, update
the kvm_riscv_gstage_map() to return the g-stage mapping upon
success.
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_mmu.h | 9 ++++-
arch/riscv/kvm/mmu.c | 58 ++++++++++++++++++--------------
arch/riscv/kvm/vcpu_exit.c | 3 +-
3 files changed, 43 insertions(+), 27 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
index 4e1654282ee4..91c11e692dc7 100644
--- a/arch/riscv/include/asm/kvm_mmu.h
+++ b/arch/riscv/include/asm/kvm_mmu.h
@@ -8,6 +8,12 @@
#include <linux/kvm_types.h>
+struct kvm_gstage_mapping {
+ gpa_t addr;
+ pte_t pte;
+ u32 level;
+};
+
int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
phys_addr_t hpa, unsigned long size,
bool writable, bool in_atomic);
@@ -15,7 +21,8 @@ void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
unsigned long size);
int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
struct kvm_memory_slot *memslot,
- gpa_t gpa, unsigned long hva, bool is_write);
+ gpa_t gpa, unsigned long hva, bool is_write,
+ struct kvm_gstage_mapping *out_map);
int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index c1a3eb076df3..806614b3e46d 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -135,18 +135,18 @@ static void gstage_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0, addr, BIT(order), order);
}
-static int gstage_set_pte(struct kvm *kvm, u32 level,
- struct kvm_mmu_memory_cache *pcache,
- gpa_t addr, const pte_t *new_pte)
+static int gstage_set_pte(struct kvm *kvm,
+ struct kvm_mmu_memory_cache *pcache,
+ const struct kvm_gstage_mapping *map)
{
u32 current_level = gstage_pgd_levels - 1;
pte_t *next_ptep = (pte_t *)kvm->arch.pgd;
- pte_t *ptep = &next_ptep[gstage_pte_index(addr, current_level)];
+ pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
- if (current_level < level)
+ if (current_level < map->level)
return -EINVAL;
- while (current_level != level) {
+ while (current_level != map->level) {
if (gstage_pte_leaf(ptep))
return -EEXIST;
@@ -165,13 +165,13 @@ static int gstage_set_pte(struct kvm *kvm, u32 level,
}
current_level--;
- ptep = &next_ptep[gstage_pte_index(addr, current_level)];
+ ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
}
- if (pte_val(*ptep) != pte_val(*new_pte)) {
- set_pte(ptep, *new_pte);
+ if (pte_val(*ptep) != pte_val(map->pte)) {
+ set_pte(ptep, map->pte);
if (gstage_pte_leaf(ptep))
- gstage_remote_tlb_flush(kvm, current_level, addr);
+ gstage_remote_tlb_flush(kvm, current_level, map->addr);
}
return 0;
@@ -181,14 +181,16 @@ static int gstage_map_page(struct kvm *kvm,
struct kvm_mmu_memory_cache *pcache,
gpa_t gpa, phys_addr_t hpa,
unsigned long page_size,
- bool page_rdonly, bool page_exec)
+ bool page_rdonly, bool page_exec,
+ struct kvm_gstage_mapping *out_map)
{
- int ret;
- u32 level = 0;
- pte_t new_pte;
pgprot_t prot;
+ int ret;
- ret = gstage_page_size_to_level(page_size, &level);
+ out_map->addr = gpa;
+ out_map->level = 0;
+
+ ret = gstage_page_size_to_level(page_size, &out_map->level);
if (ret)
return ret;
@@ -216,10 +218,10 @@ static int gstage_map_page(struct kvm *kvm,
else
prot = PAGE_WRITE;
}
- new_pte = pfn_pte(PFN_DOWN(hpa), prot);
- new_pte = pte_mkdirty(new_pte);
+ out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
+ out_map->pte = pte_mkdirty(out_map->pte);
- return gstage_set_pte(kvm, level, pcache, gpa, &new_pte);
+ return gstage_set_pte(kvm, pcache, out_map);
}
enum gstage_op {
@@ -352,7 +354,6 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
phys_addr_t hpa, unsigned long size,
bool writable, bool in_atomic)
{
- pte_t pte;
int ret = 0;
unsigned long pfn;
phys_addr_t addr, end;
@@ -360,22 +361,25 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
.gfp_custom = (in_atomic) ? GFP_ATOMIC | __GFP_ACCOUNT : 0,
.gfp_zero = __GFP_ZERO,
};
+ struct kvm_gstage_mapping map;
end = (gpa + size + PAGE_SIZE - 1) & PAGE_MASK;
pfn = __phys_to_pfn(hpa);
for (addr = gpa; addr < end; addr += PAGE_SIZE) {
- pte = pfn_pte(pfn, PAGE_KERNEL_IO);
+ map.addr = addr;
+ map.pte = pfn_pte(pfn, PAGE_KERNEL_IO);
+ map.level = 0;
if (!writable)
- pte = pte_wrprotect(pte);
+ map.pte = pte_wrprotect(map.pte);
ret = kvm_mmu_topup_memory_cache(&pcache, gstage_pgd_levels);
if (ret)
goto out;
spin_lock(&kvm->mmu_lock);
- ret = gstage_set_pte(kvm, 0, &pcache, addr, &pte);
+ ret = gstage_set_pte(kvm, &pcache, &map);
spin_unlock(&kvm->mmu_lock);
if (ret)
goto out;
@@ -593,7 +597,8 @@ bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
struct kvm_memory_slot *memslot,
- gpa_t gpa, unsigned long hva, bool is_write)
+ gpa_t gpa, unsigned long hva, bool is_write,
+ struct kvm_gstage_mapping *out_map)
{
int ret;
kvm_pfn_t hfn;
@@ -608,6 +613,9 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
unsigned long vma_pagesize, mmu_seq;
struct page *page;
+ /* Setup initial state of output mapping */
+ memset(out_map, 0, sizeof(*out_map));
+
/* We need minimum second+third level pages */
ret = kvm_mmu_topup_memory_cache(pcache, gstage_pgd_levels);
if (ret) {
@@ -677,10 +685,10 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
if (writable) {
mark_page_dirty(kvm, gfn);
ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
- vma_pagesize, false, true);
+ vma_pagesize, false, true, out_map);
} else {
ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
- vma_pagesize, true, true);
+ vma_pagesize, true, true, out_map);
}
if (ret)
diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
index 965df528de90..6b4694bc07ea 100644
--- a/arch/riscv/kvm/vcpu_exit.c
+++ b/arch/riscv/kvm/vcpu_exit.c
@@ -15,6 +15,7 @@
static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
struct kvm_cpu_trap *trap)
{
+ struct kvm_gstage_mapping host_map;
struct kvm_memory_slot *memslot;
unsigned long hva, fault_addr;
bool writable;
@@ -43,7 +44,7 @@ static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
}
ret = kvm_riscv_gstage_map(vcpu, memslot, fault_addr, hva,
- (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false);
+ (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false, &host_map);
if (ret < 0)
return ret;
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (8 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:55 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management Anup Patel
` (3 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel,
Atish Patra
Currently, the struct kvm_riscv_hfence does not have vmid field
and various hfence processing functions always pick vmid assigned
to the guest/VM. This prevents us from doing hfence operation on
arbitrary vmid hence add vmid field to struct kvm_riscv_hfence
and use it wherever applicable.
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_tlb.h | 1 +
arch/riscv/kvm/tlb.c | 30 ++++++++++++++++--------------
2 files changed, 17 insertions(+), 14 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
index cd00c9a46cb1..f67e03edeaec 100644
--- a/arch/riscv/include/asm/kvm_tlb.h
+++ b/arch/riscv/include/asm/kvm_tlb.h
@@ -19,6 +19,7 @@ enum kvm_riscv_hfence_type {
struct kvm_riscv_hfence {
enum kvm_riscv_hfence_type type;
unsigned long asid;
+ unsigned long vmid;
unsigned long order;
gpa_t addr;
gpa_t size;
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index 6fc4361c3d75..349fcfc93f54 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -237,49 +237,43 @@ static bool vcpu_hfence_enqueue(struct kvm_vcpu *vcpu,
void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
{
- unsigned long vmid;
struct kvm_riscv_hfence d = { 0 };
- struct kvm_vmid *v = &vcpu->kvm->arch.vmid;
while (vcpu_hfence_dequeue(vcpu, &d)) {
switch (d.type) {
case KVM_RISCV_HFENCE_UNKNOWN:
break;
case KVM_RISCV_HFENCE_GVMA_VMID_GPA:
- vmid = READ_ONCE(v->vmid);
if (kvm_riscv_nacl_available())
- nacl_hfence_gvma_vmid(nacl_shmem(), vmid,
+ nacl_hfence_gvma_vmid(nacl_shmem(), d.vmid,
d.addr, d.size, d.order);
else
- kvm_riscv_local_hfence_gvma_vmid_gpa(vmid, d.addr,
+ kvm_riscv_local_hfence_gvma_vmid_gpa(d.vmid, d.addr,
d.size, d.order);
break;
case KVM_RISCV_HFENCE_VVMA_ASID_GVA:
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
- vmid = READ_ONCE(v->vmid);
if (kvm_riscv_nacl_available())
- nacl_hfence_vvma_asid(nacl_shmem(), vmid, d.asid,
+ nacl_hfence_vvma_asid(nacl_shmem(), d.vmid, d.asid,
d.addr, d.size, d.order);
else
- kvm_riscv_local_hfence_vvma_asid_gva(vmid, d.asid, d.addr,
+ kvm_riscv_local_hfence_vvma_asid_gva(d.vmid, d.asid, d.addr,
d.size, d.order);
break;
case KVM_RISCV_HFENCE_VVMA_ASID_ALL:
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
- vmid = READ_ONCE(v->vmid);
if (kvm_riscv_nacl_available())
- nacl_hfence_vvma_asid_all(nacl_shmem(), vmid, d.asid);
+ nacl_hfence_vvma_asid_all(nacl_shmem(), d.vmid, d.asid);
else
- kvm_riscv_local_hfence_vvma_asid_all(vmid, d.asid);
+ kvm_riscv_local_hfence_vvma_asid_all(d.vmid, d.asid);
break;
case KVM_RISCV_HFENCE_VVMA_GVA:
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_RCVD);
- vmid = READ_ONCE(v->vmid);
if (kvm_riscv_nacl_available())
- nacl_hfence_vvma(nacl_shmem(), vmid,
+ nacl_hfence_vvma(nacl_shmem(), d.vmid,
d.addr, d.size, d.order);
else
- kvm_riscv_local_hfence_vvma_gva(vmid, d.addr,
+ kvm_riscv_local_hfence_vvma_gva(d.vmid, d.addr,
d.size, d.order);
break;
default:
@@ -336,10 +330,12 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
gpa_t gpa, gpa_t gpsz,
unsigned long order)
{
+ struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_GVMA_VMID_GPA;
data.asid = 0;
+ data.vmid = READ_ONCE(v->vmid);
data.addr = gpa;
data.size = gpsz;
data.order = order;
@@ -359,10 +355,12 @@ void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
unsigned long gva, unsigned long gvsz,
unsigned long order, unsigned long asid)
{
+ struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_VVMA_ASID_GVA;
data.asid = asid;
+ data.vmid = READ_ONCE(v->vmid);
data.addr = gva;
data.size = gvsz;
data.order = order;
@@ -374,10 +372,12 @@ void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
unsigned long asid)
{
+ struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_VVMA_ASID_ALL;
data.asid = asid;
+ data.vmid = READ_ONCE(v->vmid);
data.addr = data.size = data.order = 0;
make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
KVM_REQ_HFENCE_VVMA_ALL, &data);
@@ -388,10 +388,12 @@ void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
unsigned long gva, unsigned long gvsz,
unsigned long order)
{
+ struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_VVMA_GVA;
data.asid = 0;
+ data.vmid = READ_ONCE(v->vmid);
data.addr = gva;
data.size = gvsz;
data.order = order;
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (9 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:57 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs Anup Patel
` (2 subsequent siblings)
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel
The upcoming nested virtualization can share g-stage page table
management with the current host g-stage implementation hence
factor-out g-stage page table management as separate sources
and also use "kvm_riscv_mmu_" prefix for host g-stage functions.
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_gstage.h | 72 ++++
arch/riscv/include/asm/kvm_mmu.h | 32 +-
arch/riscv/kvm/Makefile | 1 +
arch/riscv/kvm/aia_imsic.c | 11 +-
arch/riscv/kvm/gstage.c | 337 +++++++++++++++++++
arch/riscv/kvm/main.c | 2 +-
arch/riscv/kvm/mmu.c | 492 ++++++----------------------
arch/riscv/kvm/vcpu.c | 4 +-
arch/riscv/kvm/vcpu_exit.c | 5 +-
arch/riscv/kvm/vm.c | 6 +-
10 files changed, 530 insertions(+), 432 deletions(-)
create mode 100644 arch/riscv/include/asm/kvm_gstage.h
create mode 100644 arch/riscv/kvm/gstage.c
diff --git a/arch/riscv/include/asm/kvm_gstage.h b/arch/riscv/include/asm/kvm_gstage.h
new file mode 100644
index 000000000000..595e2183173e
--- /dev/null
+++ b/arch/riscv/include/asm/kvm_gstage.h
@@ -0,0 +1,72 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+/*
+ * Copyright (C) 2019 Western Digital Corporation or its affiliates.
+ * Copyright (c) 2025 Ventana Micro Systems Inc.
+ */
+
+#ifndef __RISCV_KVM_GSTAGE_H_
+#define __RISCV_KVM_GSTAGE_H_
+
+#include <linux/kvm_types.h>
+
+struct kvm_gstage {
+ struct kvm *kvm;
+ unsigned long flags;
+#define KVM_GSTAGE_FLAGS_LOCAL BIT(0)
+ unsigned long vmid;
+ pgd_t *pgd;
+};
+
+struct kvm_gstage_mapping {
+ gpa_t addr;
+ pte_t pte;
+ u32 level;
+};
+
+#ifdef CONFIG_64BIT
+#define kvm_riscv_gstage_index_bits 9
+#else
+#define kvm_riscv_gstage_index_bits 10
+#endif
+
+extern unsigned long kvm_riscv_gstage_mode;
+extern unsigned long kvm_riscv_gstage_pgd_levels;
+
+#define kvm_riscv_gstage_pgd_xbits 2
+#define kvm_riscv_gstage_pgd_size (1UL << (HGATP_PAGE_SHIFT + kvm_riscv_gstage_pgd_xbits))
+#define kvm_riscv_gstage_gpa_bits (HGATP_PAGE_SHIFT + \
+ (kvm_riscv_gstage_pgd_levels * \
+ kvm_riscv_gstage_index_bits) + \
+ kvm_riscv_gstage_pgd_xbits)
+#define kvm_riscv_gstage_gpa_size ((gpa_t)(1ULL << kvm_riscv_gstage_gpa_bits))
+
+bool kvm_riscv_gstage_get_leaf(struct kvm_gstage *gstage, gpa_t addr,
+ pte_t **ptepp, u32 *ptep_level);
+
+int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
+ struct kvm_mmu_memory_cache *pcache,
+ const struct kvm_gstage_mapping *map);
+
+int kvm_riscv_gstage_map_page(struct kvm_gstage *gstage,
+ struct kvm_mmu_memory_cache *pcache,
+ gpa_t gpa, phys_addr_t hpa, unsigned long page_size,
+ bool page_rdonly, bool page_exec,
+ struct kvm_gstage_mapping *out_map);
+
+enum kvm_riscv_gstage_op {
+ GSTAGE_OP_NOP = 0, /* Nothing */
+ GSTAGE_OP_CLEAR, /* Clear/Unmap */
+ GSTAGE_OP_WP, /* Write-protect */
+};
+
+void kvm_riscv_gstage_op_pte(struct kvm_gstage *gstage, gpa_t addr,
+ pte_t *ptep, u32 ptep_level, enum kvm_riscv_gstage_op op);
+
+void kvm_riscv_gstage_unmap_range(struct kvm_gstage *gstage,
+ gpa_t start, gpa_t size, bool may_block);
+
+void kvm_riscv_gstage_wp_range(struct kvm_gstage *gstage, gpa_t start, gpa_t end);
+
+void kvm_riscv_gstage_mode_detect(void);
+
+#endif
diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
index 91c11e692dc7..5439e76f0a96 100644
--- a/arch/riscv/include/asm/kvm_mmu.h
+++ b/arch/riscv/include/asm/kvm_mmu.h
@@ -6,28 +6,16 @@
#ifndef __RISCV_KVM_MMU_H_
#define __RISCV_KVM_MMU_H_
-#include <linux/kvm_types.h>
+#include <asm/kvm_gstage.h>
-struct kvm_gstage_mapping {
- gpa_t addr;
- pte_t pte;
- u32 level;
-};
-
-int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
- phys_addr_t hpa, unsigned long size,
- bool writable, bool in_atomic);
-void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
- unsigned long size);
-int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
- struct kvm_memory_slot *memslot,
- gpa_t gpa, unsigned long hva, bool is_write,
- struct kvm_gstage_mapping *out_map);
-int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
-void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
-void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
-void kvm_riscv_gstage_mode_detect(void);
-unsigned long kvm_riscv_gstage_mode(void);
-int kvm_riscv_gstage_gpa_bits(void);
+int kvm_riscv_mmu_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
+ unsigned long size, bool writable, bool in_atomic);
+void kvm_riscv_mmu_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size);
+int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
+ gpa_t gpa, unsigned long hva, bool is_write,
+ struct kvm_gstage_mapping *out_map);
+int kvm_riscv_mmu_alloc_pgd(struct kvm *kvm);
+void kvm_riscv_mmu_free_pgd(struct kvm *kvm);
+void kvm_riscv_mmu_update_hgatp(struct kvm_vcpu *vcpu);
#endif
diff --git a/arch/riscv/kvm/Makefile b/arch/riscv/kvm/Makefile
index 06e2d52a9b88..07197395750e 100644
--- a/arch/riscv/kvm/Makefile
+++ b/arch/riscv/kvm/Makefile
@@ -14,6 +14,7 @@ kvm-y += aia.o
kvm-y += aia_aplic.o
kvm-y += aia_device.o
kvm-y += aia_imsic.o
+kvm-y += gstage.o
kvm-y += main.o
kvm-y += mmu.o
kvm-y += nacl.o
diff --git a/arch/riscv/kvm/aia_imsic.c b/arch/riscv/kvm/aia_imsic.c
index 40b469c0a01f..ea1a36836d9c 100644
--- a/arch/riscv/kvm/aia_imsic.c
+++ b/arch/riscv/kvm/aia_imsic.c
@@ -704,9 +704,8 @@ void kvm_riscv_vcpu_aia_imsic_release(struct kvm_vcpu *vcpu)
*/
/* Purge the G-stage mapping */
- kvm_riscv_gstage_iounmap(vcpu->kvm,
- vcpu->arch.aia_context.imsic_addr,
- IMSIC_MMIO_PAGE_SZ);
+ kvm_riscv_mmu_iounmap(vcpu->kvm, vcpu->arch.aia_context.imsic_addr,
+ IMSIC_MMIO_PAGE_SZ);
/* TODO: Purge the IOMMU mapping ??? */
@@ -786,9 +785,9 @@ int kvm_riscv_vcpu_aia_imsic_update(struct kvm_vcpu *vcpu)
imsic_vsfile_local_clear(new_vsfile_hgei, imsic->nr_hw_eix);
/* Update G-stage mapping for the new IMSIC VS-file */
- ret = kvm_riscv_gstage_ioremap(kvm, vcpu->arch.aia_context.imsic_addr,
- new_vsfile_pa, IMSIC_MMIO_PAGE_SZ,
- true, true);
+ ret = kvm_riscv_mmu_ioremap(kvm, vcpu->arch.aia_context.imsic_addr,
+ new_vsfile_pa, IMSIC_MMIO_PAGE_SZ,
+ true, true);
if (ret)
goto fail_free_vsfile_hgei;
diff --git a/arch/riscv/kvm/gstage.c b/arch/riscv/kvm/gstage.c
new file mode 100644
index 000000000000..9c7c44f09b05
--- /dev/null
+++ b/arch/riscv/kvm/gstage.c
@@ -0,0 +1,337 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2019 Western Digital Corporation or its affiliates.
+ * Copyright (c) 2025 Ventana Micro Systems Inc.
+ */
+
+#include <linux/bitops.h>
+#include <linux/errno.h>
+#include <linux/kvm_host.h>
+#include <linux/module.h>
+#include <linux/pgtable.h>
+#include <asm/kvm_gstage.h>
+
+#ifdef CONFIG_64BIT
+unsigned long kvm_riscv_gstage_mode __ro_after_init = HGATP_MODE_SV39X4;
+unsigned long kvm_riscv_gstage_pgd_levels __ro_after_init = 3;
+#else
+unsigned long kvm_riscv_gstage_mode __ro_after_init = HGATP_MODE_SV32X4;
+unsigned long kvm_riscv_gstage_pgd_levels __ro_after_init = 2;
+#endif
+
+#define gstage_pte_leaf(__ptep) \
+ (pte_val(*(__ptep)) & (_PAGE_READ | _PAGE_WRITE | _PAGE_EXEC))
+
+static inline unsigned long gstage_pte_index(gpa_t addr, u32 level)
+{
+ unsigned long mask;
+ unsigned long shift = HGATP_PAGE_SHIFT + (kvm_riscv_gstage_index_bits * level);
+
+ if (level == (kvm_riscv_gstage_pgd_levels - 1))
+ mask = (PTRS_PER_PTE * (1UL << kvm_riscv_gstage_pgd_xbits)) - 1;
+ else
+ mask = PTRS_PER_PTE - 1;
+
+ return (addr >> shift) & mask;
+}
+
+static inline unsigned long gstage_pte_page_vaddr(pte_t pte)
+{
+ return (unsigned long)pfn_to_virt(__page_val_to_pfn(pte_val(pte)));
+}
+
+static int gstage_page_size_to_level(unsigned long page_size, u32 *out_level)
+{
+ u32 i;
+ unsigned long psz = 1UL << 12;
+
+ for (i = 0; i < kvm_riscv_gstage_pgd_levels; i++) {
+ if (page_size == (psz << (i * kvm_riscv_gstage_index_bits))) {
+ *out_level = i;
+ return 0;
+ }
+ }
+
+ return -EINVAL;
+}
+
+static int gstage_level_to_page_order(u32 level, unsigned long *out_pgorder)
+{
+ if (kvm_riscv_gstage_pgd_levels < level)
+ return -EINVAL;
+
+ *out_pgorder = 12 + (level * kvm_riscv_gstage_index_bits);
+ return 0;
+}
+
+static int gstage_level_to_page_size(u32 level, unsigned long *out_pgsize)
+{
+ int rc;
+ unsigned long page_order = PAGE_SHIFT;
+
+ rc = gstage_level_to_page_order(level, &page_order);
+ if (rc)
+ return rc;
+
+ *out_pgsize = BIT(page_order);
+ return 0;
+}
+
+bool kvm_riscv_gstage_get_leaf(struct kvm_gstage *gstage, gpa_t addr,
+ pte_t **ptepp, u32 *ptep_level)
+{
+ pte_t *ptep;
+ u32 current_level = kvm_riscv_gstage_pgd_levels - 1;
+
+ *ptep_level = current_level;
+ ptep = (pte_t *)gstage->pgd;
+ ptep = &ptep[gstage_pte_index(addr, current_level)];
+ while (ptep && pte_val(ptep_get(ptep))) {
+ if (gstage_pte_leaf(ptep)) {
+ *ptep_level = current_level;
+ *ptepp = ptep;
+ return true;
+ }
+
+ if (current_level) {
+ current_level--;
+ *ptep_level = current_level;
+ ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
+ ptep = &ptep[gstage_pte_index(addr, current_level)];
+ } else {
+ ptep = NULL;
+ }
+ }
+
+ return false;
+}
+
+static void gstage_tlb_flush(struct kvm_gstage *gstage, u32 level, gpa_t addr)
+{
+ unsigned long order = PAGE_SHIFT;
+
+ if (gstage_level_to_page_order(level, &order))
+ return;
+ addr &= ~(BIT(order) - 1);
+
+ if (gstage->flags & KVM_GSTAGE_FLAGS_LOCAL)
+ kvm_riscv_local_hfence_gvma_vmid_gpa(gstage->vmid, addr, BIT(order), order);
+ else
+ kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order);
+}
+
+int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
+ struct kvm_mmu_memory_cache *pcache,
+ const struct kvm_gstage_mapping *map)
+{
+ u32 current_level = kvm_riscv_gstage_pgd_levels - 1;
+ pte_t *next_ptep = (pte_t *)gstage->pgd;
+ pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
+
+ if (current_level < map->level)
+ return -EINVAL;
+
+ while (current_level != map->level) {
+ if (gstage_pte_leaf(ptep))
+ return -EEXIST;
+
+ if (!pte_val(ptep_get(ptep))) {
+ if (!pcache)
+ return -ENOMEM;
+ next_ptep = kvm_mmu_memory_cache_alloc(pcache);
+ if (!next_ptep)
+ return -ENOMEM;
+ set_pte(ptep, pfn_pte(PFN_DOWN(__pa(next_ptep)),
+ __pgprot(_PAGE_TABLE)));
+ } else {
+ if (gstage_pte_leaf(ptep))
+ return -EEXIST;
+ next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
+ }
+
+ current_level--;
+ ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
+ }
+
+ if (pte_val(*ptep) != pte_val(map->pte)) {
+ set_pte(ptep, map->pte);
+ if (gstage_pte_leaf(ptep))
+ gstage_tlb_flush(gstage, current_level, map->addr);
+ }
+
+ return 0;
+}
+
+int kvm_riscv_gstage_map_page(struct kvm_gstage *gstage,
+ struct kvm_mmu_memory_cache *pcache,
+ gpa_t gpa, phys_addr_t hpa, unsigned long page_size,
+ bool page_rdonly, bool page_exec,
+ struct kvm_gstage_mapping *out_map)
+{
+ pgprot_t prot;
+ int ret;
+
+ out_map->addr = gpa;
+ out_map->level = 0;
+
+ ret = gstage_page_size_to_level(page_size, &out_map->level);
+ if (ret)
+ return ret;
+
+ /*
+ * A RISC-V implementation can choose to either:
+ * 1) Update 'A' and 'D' PTE bits in hardware
+ * 2) Generate page fault when 'A' and/or 'D' bits are not set
+ * PTE so that software can update these bits.
+ *
+ * We support both options mentioned above. To achieve this, we
+ * always set 'A' and 'D' PTE bits at time of creating G-stage
+ * mapping. To support KVM dirty page logging with both options
+ * mentioned above, we will write-protect G-stage PTEs to track
+ * dirty pages.
+ */
+
+ if (page_exec) {
+ if (page_rdonly)
+ prot = PAGE_READ_EXEC;
+ else
+ prot = PAGE_WRITE_EXEC;
+ } else {
+ if (page_rdonly)
+ prot = PAGE_READ;
+ else
+ prot = PAGE_WRITE;
+ }
+ out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
+ out_map->pte = pte_mkdirty(out_map->pte);
+
+ return kvm_riscv_gstage_set_pte(gstage, pcache, out_map);
+}
+
+void kvm_riscv_gstage_op_pte(struct kvm_gstage *gstage, gpa_t addr,
+ pte_t *ptep, u32 ptep_level, enum kvm_riscv_gstage_op op)
+{
+ int i, ret;
+ pte_t old_pte, *next_ptep;
+ u32 next_ptep_level;
+ unsigned long next_page_size, page_size;
+
+ ret = gstage_level_to_page_size(ptep_level, &page_size);
+ if (ret)
+ return;
+
+ WARN_ON(addr & (page_size - 1));
+
+ if (!pte_val(ptep_get(ptep)))
+ return;
+
+ if (ptep_level && !gstage_pte_leaf(ptep)) {
+ next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
+ next_ptep_level = ptep_level - 1;
+ ret = gstage_level_to_page_size(next_ptep_level, &next_page_size);
+ if (ret)
+ return;
+
+ if (op == GSTAGE_OP_CLEAR)
+ set_pte(ptep, __pte(0));
+ for (i = 0; i < PTRS_PER_PTE; i++)
+ kvm_riscv_gstage_op_pte(gstage, addr + i * next_page_size,
+ &next_ptep[i], next_ptep_level, op);
+ if (op == GSTAGE_OP_CLEAR)
+ put_page(virt_to_page(next_ptep));
+ } else {
+ old_pte = *ptep;
+ if (op == GSTAGE_OP_CLEAR)
+ set_pte(ptep, __pte(0));
+ else if (op == GSTAGE_OP_WP)
+ set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
+ if (pte_val(*ptep) != pte_val(old_pte))
+ gstage_tlb_flush(gstage, ptep_level, addr);
+ }
+}
+
+void kvm_riscv_gstage_unmap_range(struct kvm_gstage *gstage,
+ gpa_t start, gpa_t size, bool may_block)
+{
+ int ret;
+ pte_t *ptep;
+ u32 ptep_level;
+ bool found_leaf;
+ unsigned long page_size;
+ gpa_t addr = start, end = start + size;
+
+ while (addr < end) {
+ found_leaf = kvm_riscv_gstage_get_leaf(gstage, addr, &ptep, &ptep_level);
+ ret = gstage_level_to_page_size(ptep_level, &page_size);
+ if (ret)
+ break;
+
+ if (!found_leaf)
+ goto next;
+
+ if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
+ kvm_riscv_gstage_op_pte(gstage, addr, ptep,
+ ptep_level, GSTAGE_OP_CLEAR);
+
+next:
+ addr += page_size;
+
+ /*
+ * If the range is too large, release the kvm->mmu_lock
+ * to prevent starvation and lockup detector warnings.
+ */
+ if (!(gstage->flags & KVM_GSTAGE_FLAGS_LOCAL) && may_block && addr < end)
+ cond_resched_lock(&gstage->kvm->mmu_lock);
+ }
+}
+
+void kvm_riscv_gstage_wp_range(struct kvm_gstage *gstage, gpa_t start, gpa_t end)
+{
+ int ret;
+ pte_t *ptep;
+ u32 ptep_level;
+ bool found_leaf;
+ gpa_t addr = start;
+ unsigned long page_size;
+
+ while (addr < end) {
+ found_leaf = kvm_riscv_gstage_get_leaf(gstage, addr, &ptep, &ptep_level);
+ ret = gstage_level_to_page_size(ptep_level, &page_size);
+ if (ret)
+ break;
+
+ if (!found_leaf)
+ goto next;
+
+ if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
+ kvm_riscv_gstage_op_pte(gstage, addr, ptep,
+ ptep_level, GSTAGE_OP_WP);
+
+next:
+ addr += page_size;
+ }
+}
+
+void __init kvm_riscv_gstage_mode_detect(void)
+{
+#ifdef CONFIG_64BIT
+ /* Try Sv57x4 G-stage mode */
+ csr_write(CSR_HGATP, HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
+ if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV57X4) {
+ kvm_riscv_gstage_mode = HGATP_MODE_SV57X4;
+ kvm_riscv_gstage_pgd_levels = 5;
+ goto skip_sv48x4_test;
+ }
+
+ /* Try Sv48x4 G-stage mode */
+ csr_write(CSR_HGATP, HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
+ if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV48X4) {
+ kvm_riscv_gstage_mode = HGATP_MODE_SV48X4;
+ kvm_riscv_gstage_pgd_levels = 4;
+ }
+skip_sv48x4_test:
+
+ csr_write(CSR_HGATP, 0);
+ kvm_riscv_local_hfence_gvma_all();
+#endif
+}
diff --git a/arch/riscv/kvm/main.c b/arch/riscv/kvm/main.c
index b861a5dd7bd9..67c876de74ef 100644
--- a/arch/riscv/kvm/main.c
+++ b/arch/riscv/kvm/main.c
@@ -135,7 +135,7 @@ static int __init riscv_kvm_init(void)
(rc) ? slist : "no features");
}
- switch (kvm_riscv_gstage_mode()) {
+ switch (kvm_riscv_gstage_mode) {
case HGATP_MODE_SV32X4:
str = "Sv32x4";
break;
diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
index 806614b3e46d..9f7dcd8cd741 100644
--- a/arch/riscv/kvm/mmu.c
+++ b/arch/riscv/kvm/mmu.c
@@ -6,9 +6,7 @@
* Anup Patel <anup.patel@wdc.com>
*/
-#include <linux/bitops.h>
#include <linux/errno.h>
-#include <linux/err.h>
#include <linux/hugetlb.h>
#include <linux/module.h>
#include <linux/uaccess.h>
@@ -17,342 +15,28 @@
#include <linux/sched/signal.h>
#include <asm/kvm_mmu.h>
#include <asm/kvm_nacl.h>
-#include <asm/page.h>
-#include <asm/pgtable.h>
-
-#ifdef CONFIG_64BIT
-static unsigned long gstage_mode __ro_after_init = (HGATP_MODE_SV39X4 << HGATP_MODE_SHIFT);
-static unsigned long gstage_pgd_levels __ro_after_init = 3;
-#define gstage_index_bits 9
-#else
-static unsigned long gstage_mode __ro_after_init = (HGATP_MODE_SV32X4 << HGATP_MODE_SHIFT);
-static unsigned long gstage_pgd_levels __ro_after_init = 2;
-#define gstage_index_bits 10
-#endif
-
-#define gstage_pgd_xbits 2
-#define gstage_pgd_size (1UL << (HGATP_PAGE_SHIFT + gstage_pgd_xbits))
-#define gstage_gpa_bits (HGATP_PAGE_SHIFT + \
- (gstage_pgd_levels * gstage_index_bits) + \
- gstage_pgd_xbits)
-#define gstage_gpa_size ((gpa_t)(1ULL << gstage_gpa_bits))
-
-#define gstage_pte_leaf(__ptep) \
- (pte_val(*(__ptep)) & (_PAGE_READ | _PAGE_WRITE | _PAGE_EXEC))
-
-static inline unsigned long gstage_pte_index(gpa_t addr, u32 level)
-{
- unsigned long mask;
- unsigned long shift = HGATP_PAGE_SHIFT + (gstage_index_bits * level);
-
- if (level == (gstage_pgd_levels - 1))
- mask = (PTRS_PER_PTE * (1UL << gstage_pgd_xbits)) - 1;
- else
- mask = PTRS_PER_PTE - 1;
-
- return (addr >> shift) & mask;
-}
-static inline unsigned long gstage_pte_page_vaddr(pte_t pte)
-{
- return (unsigned long)pfn_to_virt(__page_val_to_pfn(pte_val(pte)));
-}
-
-static int gstage_page_size_to_level(unsigned long page_size, u32 *out_level)
-{
- u32 i;
- unsigned long psz = 1UL << 12;
-
- for (i = 0; i < gstage_pgd_levels; i++) {
- if (page_size == (psz << (i * gstage_index_bits))) {
- *out_level = i;
- return 0;
- }
- }
-
- return -EINVAL;
-}
-
-static int gstage_level_to_page_order(u32 level, unsigned long *out_pgorder)
-{
- if (gstage_pgd_levels < level)
- return -EINVAL;
-
- *out_pgorder = 12 + (level * gstage_index_bits);
- return 0;
-}
-
-static int gstage_level_to_page_size(u32 level, unsigned long *out_pgsize)
-{
- int rc;
- unsigned long page_order = PAGE_SHIFT;
-
- rc = gstage_level_to_page_order(level, &page_order);
- if (rc)
- return rc;
-
- *out_pgsize = BIT(page_order);
- return 0;
-}
-
-static bool gstage_get_leaf_entry(struct kvm *kvm, gpa_t addr,
- pte_t **ptepp, u32 *ptep_level)
-{
- pte_t *ptep;
- u32 current_level = gstage_pgd_levels - 1;
-
- *ptep_level = current_level;
- ptep = (pte_t *)kvm->arch.pgd;
- ptep = &ptep[gstage_pte_index(addr, current_level)];
- while (ptep && pte_val(ptep_get(ptep))) {
- if (gstage_pte_leaf(ptep)) {
- *ptep_level = current_level;
- *ptepp = ptep;
- return true;
- }
-
- if (current_level) {
- current_level--;
- *ptep_level = current_level;
- ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
- ptep = &ptep[gstage_pte_index(addr, current_level)];
- } else {
- ptep = NULL;
- }
- }
-
- return false;
-}
-
-static void gstage_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
-{
- unsigned long order = PAGE_SHIFT;
-
- if (gstage_level_to_page_order(level, &order))
- return;
- addr &= ~(BIT(order) - 1);
-
- kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0, addr, BIT(order), order);
-}
-
-static int gstage_set_pte(struct kvm *kvm,
- struct kvm_mmu_memory_cache *pcache,
- const struct kvm_gstage_mapping *map)
-{
- u32 current_level = gstage_pgd_levels - 1;
- pte_t *next_ptep = (pte_t *)kvm->arch.pgd;
- pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
-
- if (current_level < map->level)
- return -EINVAL;
-
- while (current_level != map->level) {
- if (gstage_pte_leaf(ptep))
- return -EEXIST;
-
- if (!pte_val(ptep_get(ptep))) {
- if (!pcache)
- return -ENOMEM;
- next_ptep = kvm_mmu_memory_cache_alloc(pcache);
- if (!next_ptep)
- return -ENOMEM;
- set_pte(ptep, pfn_pte(PFN_DOWN(__pa(next_ptep)),
- __pgprot(_PAGE_TABLE)));
- } else {
- if (gstage_pte_leaf(ptep))
- return -EEXIST;
- next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
- }
-
- current_level--;
- ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
- }
-
- if (pte_val(*ptep) != pte_val(map->pte)) {
- set_pte(ptep, map->pte);
- if (gstage_pte_leaf(ptep))
- gstage_remote_tlb_flush(kvm, current_level, map->addr);
- }
-
- return 0;
-}
-
-static int gstage_map_page(struct kvm *kvm,
- struct kvm_mmu_memory_cache *pcache,
- gpa_t gpa, phys_addr_t hpa,
- unsigned long page_size,
- bool page_rdonly, bool page_exec,
- struct kvm_gstage_mapping *out_map)
-{
- pgprot_t prot;
- int ret;
-
- out_map->addr = gpa;
- out_map->level = 0;
-
- ret = gstage_page_size_to_level(page_size, &out_map->level);
- if (ret)
- return ret;
-
- /*
- * A RISC-V implementation can choose to either:
- * 1) Update 'A' and 'D' PTE bits in hardware
- * 2) Generate page fault when 'A' and/or 'D' bits are not set
- * PTE so that software can update these bits.
- *
- * We support both options mentioned above. To achieve this, we
- * always set 'A' and 'D' PTE bits at time of creating G-stage
- * mapping. To support KVM dirty page logging with both options
- * mentioned above, we will write-protect G-stage PTEs to track
- * dirty pages.
- */
-
- if (page_exec) {
- if (page_rdonly)
- prot = PAGE_READ_EXEC;
- else
- prot = PAGE_WRITE_EXEC;
- } else {
- if (page_rdonly)
- prot = PAGE_READ;
- else
- prot = PAGE_WRITE;
- }
- out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
- out_map->pte = pte_mkdirty(out_map->pte);
-
- return gstage_set_pte(kvm, pcache, out_map);
-}
-
-enum gstage_op {
- GSTAGE_OP_NOP = 0, /* Nothing */
- GSTAGE_OP_CLEAR, /* Clear/Unmap */
- GSTAGE_OP_WP, /* Write-protect */
-};
-
-static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
- pte_t *ptep, u32 ptep_level, enum gstage_op op)
-{
- int i, ret;
- pte_t old_pte, *next_ptep;
- u32 next_ptep_level;
- unsigned long next_page_size, page_size;
-
- ret = gstage_level_to_page_size(ptep_level, &page_size);
- if (ret)
- return;
-
- BUG_ON(addr & (page_size - 1));
-
- if (!pte_val(ptep_get(ptep)))
- return;
-
- if (ptep_level && !gstage_pte_leaf(ptep)) {
- next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
- next_ptep_level = ptep_level - 1;
- ret = gstage_level_to_page_size(next_ptep_level,
- &next_page_size);
- if (ret)
- return;
-
- if (op == GSTAGE_OP_CLEAR)
- set_pte(ptep, __pte(0));
- for (i = 0; i < PTRS_PER_PTE; i++)
- gstage_op_pte(kvm, addr + i * next_page_size,
- &next_ptep[i], next_ptep_level, op);
- if (op == GSTAGE_OP_CLEAR)
- put_page(virt_to_page(next_ptep));
- } else {
- old_pte = *ptep;
- if (op == GSTAGE_OP_CLEAR)
- set_pte(ptep, __pte(0));
- else if (op == GSTAGE_OP_WP)
- set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
- if (pte_val(*ptep) != pte_val(old_pte))
- gstage_remote_tlb_flush(kvm, ptep_level, addr);
- }
-}
-
-static void gstage_unmap_range(struct kvm *kvm, gpa_t start,
- gpa_t size, bool may_block)
-{
- int ret;
- pte_t *ptep;
- u32 ptep_level;
- bool found_leaf;
- unsigned long page_size;
- gpa_t addr = start, end = start + size;
-
- while (addr < end) {
- found_leaf = gstage_get_leaf_entry(kvm, addr,
- &ptep, &ptep_level);
- ret = gstage_level_to_page_size(ptep_level, &page_size);
- if (ret)
- break;
-
- if (!found_leaf)
- goto next;
-
- if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
- gstage_op_pte(kvm, addr, ptep,
- ptep_level, GSTAGE_OP_CLEAR);
-
-next:
- addr += page_size;
-
- /*
- * If the range is too large, release the kvm->mmu_lock
- * to prevent starvation and lockup detector warnings.
- */
- if (may_block && addr < end)
- cond_resched_lock(&kvm->mmu_lock);
- }
-}
-
-static void gstage_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
-{
- int ret;
- pte_t *ptep;
- u32 ptep_level;
- bool found_leaf;
- gpa_t addr = start;
- unsigned long page_size;
-
- while (addr < end) {
- found_leaf = gstage_get_leaf_entry(kvm, addr,
- &ptep, &ptep_level);
- ret = gstage_level_to_page_size(ptep_level, &page_size);
- if (ret)
- break;
-
- if (!found_leaf)
- goto next;
-
- if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
- gstage_op_pte(kvm, addr, ptep,
- ptep_level, GSTAGE_OP_WP);
-
-next:
- addr += page_size;
- }
-}
-
-static void gstage_wp_memory_region(struct kvm *kvm, int slot)
+static void mmu_wp_memory_region(struct kvm *kvm, int slot)
{
struct kvm_memslots *slots = kvm_memslots(kvm);
struct kvm_memory_slot *memslot = id_to_memslot(slots, slot);
phys_addr_t start = memslot->base_gfn << PAGE_SHIFT;
phys_addr_t end = (memslot->base_gfn + memslot->npages) << PAGE_SHIFT;
+ struct kvm_gstage gstage;
+
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
spin_lock(&kvm->mmu_lock);
- gstage_wp_range(kvm, start, end);
+ kvm_riscv_gstage_wp_range(&gstage, start, end);
spin_unlock(&kvm->mmu_lock);
kvm_flush_remote_tlbs_memslot(kvm, memslot);
}
-int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
- phys_addr_t hpa, unsigned long size,
- bool writable, bool in_atomic)
+int kvm_riscv_mmu_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
+ unsigned long size, bool writable, bool in_atomic)
{
int ret = 0;
unsigned long pfn;
@@ -362,6 +46,12 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
.gfp_zero = __GFP_ZERO,
};
struct kvm_gstage_mapping map;
+ struct kvm_gstage gstage;
+
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
end = (gpa + size + PAGE_SIZE - 1) & PAGE_MASK;
pfn = __phys_to_pfn(hpa);
@@ -374,12 +64,12 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
if (!writable)
map.pte = pte_wrprotect(map.pte);
- ret = kvm_mmu_topup_memory_cache(&pcache, gstage_pgd_levels);
+ ret = kvm_mmu_topup_memory_cache(&pcache, kvm_riscv_gstage_pgd_levels);
if (ret)
goto out;
spin_lock(&kvm->mmu_lock);
- ret = gstage_set_pte(kvm, &pcache, &map);
+ ret = kvm_riscv_gstage_set_pte(&gstage, &pcache, &map);
spin_unlock(&kvm->mmu_lock);
if (ret)
goto out;
@@ -392,10 +82,17 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
return ret;
}
-void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size)
+void kvm_riscv_mmu_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size)
{
+ struct kvm_gstage gstage;
+
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+
spin_lock(&kvm->mmu_lock);
- gstage_unmap_range(kvm, gpa, size, false);
+ kvm_riscv_gstage_unmap_range(&gstage, gpa, size, false);
spin_unlock(&kvm->mmu_lock);
}
@@ -407,8 +104,14 @@ void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
phys_addr_t base_gfn = slot->base_gfn + gfn_offset;
phys_addr_t start = (base_gfn + __ffs(mask)) << PAGE_SHIFT;
phys_addr_t end = (base_gfn + __fls(mask) + 1) << PAGE_SHIFT;
+ struct kvm_gstage gstage;
+
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
- gstage_wp_range(kvm, start, end);
+ kvm_riscv_gstage_wp_range(&gstage, start, end);
}
void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
@@ -425,7 +128,7 @@ void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen)
void kvm_arch_flush_shadow_all(struct kvm *kvm)
{
- kvm_riscv_gstage_free_pgd(kvm);
+ kvm_riscv_mmu_free_pgd(kvm);
}
void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
@@ -433,9 +136,15 @@ void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
{
gpa_t gpa = slot->base_gfn << PAGE_SHIFT;
phys_addr_t size = slot->npages << PAGE_SHIFT;
+ struct kvm_gstage gstage;
+
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
spin_lock(&kvm->mmu_lock);
- gstage_unmap_range(kvm, gpa, size, false);
+ kvm_riscv_gstage_unmap_range(&gstage, gpa, size, false);
spin_unlock(&kvm->mmu_lock);
}
@@ -450,7 +159,7 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
* the memory slot is write protected.
*/
if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES)
- gstage_wp_memory_region(kvm, new->id);
+ mmu_wp_memory_region(kvm, new->id);
}
int kvm_arch_prepare_memory_region(struct kvm *kvm,
@@ -472,7 +181,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
* space addressable by the KVM guest GPA space.
*/
if ((new->base_gfn + new->npages) >=
- (gstage_gpa_size >> PAGE_SHIFT))
+ (kvm_riscv_gstage_gpa_size >> PAGE_SHIFT))
return -EFAULT;
hva = new->userspace_addr;
@@ -528,9 +237,8 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
goto out;
}
- ret = kvm_riscv_gstage_ioremap(kvm, gpa, pa,
- vm_end - vm_start,
- writable, false);
+ ret = kvm_riscv_mmu_ioremap(kvm, gpa, pa, vm_end - vm_start,
+ writable, false);
if (ret)
break;
}
@@ -541,7 +249,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
goto out;
if (ret)
- kvm_riscv_gstage_iounmap(kvm, base_gpa, size);
+ kvm_riscv_mmu_iounmap(kvm, base_gpa, size);
out:
mmap_read_unlock(current->mm);
@@ -550,12 +258,18 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
{
+ struct kvm_gstage gstage;
+
if (!kvm->arch.pgd)
return false;
- gstage_unmap_range(kvm, range->start << PAGE_SHIFT,
- (range->end - range->start) << PAGE_SHIFT,
- range->may_block);
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+ kvm_riscv_gstage_unmap_range(&gstage, range->start << PAGE_SHIFT,
+ (range->end - range->start) << PAGE_SHIFT,
+ range->may_block);
return false;
}
@@ -564,14 +278,19 @@ bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
pte_t *ptep;
u32 ptep_level = 0;
u64 size = (range->end - range->start) << PAGE_SHIFT;
+ struct kvm_gstage gstage;
if (!kvm->arch.pgd)
return false;
WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
- if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
- &ptep, &ptep_level))
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+ if (!kvm_riscv_gstage_get_leaf(&gstage, range->start << PAGE_SHIFT,
+ &ptep, &ptep_level))
return false;
return ptep_test_and_clear_young(NULL, 0, ptep);
@@ -582,23 +301,27 @@ bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
pte_t *ptep;
u32 ptep_level = 0;
u64 size = (range->end - range->start) << PAGE_SHIFT;
+ struct kvm_gstage gstage;
if (!kvm->arch.pgd)
return false;
WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
- if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
- &ptep, &ptep_level))
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+ if (!kvm_riscv_gstage_get_leaf(&gstage, range->start << PAGE_SHIFT,
+ &ptep, &ptep_level))
return false;
return pte_young(ptep_get(ptep));
}
-int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
- struct kvm_memory_slot *memslot,
- gpa_t gpa, unsigned long hva, bool is_write,
- struct kvm_gstage_mapping *out_map)
+int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
+ gpa_t gpa, unsigned long hva, bool is_write,
+ struct kvm_gstage_mapping *out_map)
{
int ret;
kvm_pfn_t hfn;
@@ -611,13 +334,19 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
bool logging = (memslot->dirty_bitmap &&
!(memslot->flags & KVM_MEM_READONLY)) ? true : false;
unsigned long vma_pagesize, mmu_seq;
+ struct kvm_gstage gstage;
struct page *page;
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+
/* Setup initial state of output mapping */
memset(out_map, 0, sizeof(*out_map));
/* We need minimum second+third level pages */
- ret = kvm_mmu_topup_memory_cache(pcache, gstage_pgd_levels);
+ ret = kvm_mmu_topup_memory_cache(pcache, kvm_riscv_gstage_pgd_levels);
if (ret) {
kvm_err("Failed to topup G-stage cache\n");
return ret;
@@ -684,11 +413,11 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
if (writable) {
mark_page_dirty(kvm, gfn);
- ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
- vma_pagesize, false, true, out_map);
+ ret = kvm_riscv_gstage_map_page(&gstage, pcache, gpa, hfn << PAGE_SHIFT,
+ vma_pagesize, false, true, out_map);
} else {
- ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
- vma_pagesize, true, true, out_map);
+ ret = kvm_riscv_gstage_map_page(&gstage, pcache, gpa, hfn << PAGE_SHIFT,
+ vma_pagesize, true, true, out_map);
}
if (ret)
@@ -700,7 +429,7 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
return ret;
}
-int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
+int kvm_riscv_mmu_alloc_pgd(struct kvm *kvm)
{
struct page *pgd_page;
@@ -710,7 +439,7 @@ int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
}
pgd_page = alloc_pages(GFP_KERNEL | __GFP_ZERO,
- get_order(gstage_pgd_size));
+ get_order(kvm_riscv_gstage_pgd_size));
if (!pgd_page)
return -ENOMEM;
kvm->arch.pgd = page_to_virt(pgd_page);
@@ -719,13 +448,18 @@ int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
return 0;
}
-void kvm_riscv_gstage_free_pgd(struct kvm *kvm)
+void kvm_riscv_mmu_free_pgd(struct kvm *kvm)
{
+ struct kvm_gstage gstage;
void *pgd = NULL;
spin_lock(&kvm->mmu_lock);
if (kvm->arch.pgd) {
- gstage_unmap_range(kvm, 0UL, gstage_gpa_size, false);
+ gstage.kvm = kvm;
+ gstage.flags = 0;
+ gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
+ gstage.pgd = kvm->arch.pgd;
+ kvm_riscv_gstage_unmap_range(&gstage, 0UL, kvm_riscv_gstage_gpa_size, false);
pgd = READ_ONCE(kvm->arch.pgd);
kvm->arch.pgd = NULL;
kvm->arch.pgd_phys = 0;
@@ -733,12 +467,12 @@ void kvm_riscv_gstage_free_pgd(struct kvm *kvm)
spin_unlock(&kvm->mmu_lock);
if (pgd)
- free_pages((unsigned long)pgd, get_order(gstage_pgd_size));
+ free_pages((unsigned long)pgd, get_order(kvm_riscv_gstage_pgd_size));
}
-void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu)
+void kvm_riscv_mmu_update_hgatp(struct kvm_vcpu *vcpu)
{
- unsigned long hgatp = gstage_mode;
+ unsigned long hgatp = kvm_riscv_gstage_mode << HGATP_MODE_SHIFT;
struct kvm_arch *k = &vcpu->kvm->arch;
hgatp |= (READ_ONCE(k->vmid.vmid) << HGATP_VMID_SHIFT) & HGATP_VMID;
@@ -749,37 +483,3 @@ void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu)
if (!kvm_riscv_gstage_vmid_bits())
kvm_riscv_local_hfence_gvma_all();
}
-
-void __init kvm_riscv_gstage_mode_detect(void)
-{
-#ifdef CONFIG_64BIT
- /* Try Sv57x4 G-stage mode */
- csr_write(CSR_HGATP, HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
- if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV57X4) {
- gstage_mode = (HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
- gstage_pgd_levels = 5;
- goto skip_sv48x4_test;
- }
-
- /* Try Sv48x4 G-stage mode */
- csr_write(CSR_HGATP, HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
- if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV48X4) {
- gstage_mode = (HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
- gstage_pgd_levels = 4;
- }
-skip_sv48x4_test:
-
- csr_write(CSR_HGATP, 0);
- kvm_riscv_local_hfence_gvma_all();
-#endif
-}
-
-unsigned long __init kvm_riscv_gstage_mode(void)
-{
- return gstage_mode >> HGATP_MODE_SHIFT;
-}
-
-int kvm_riscv_gstage_gpa_bits(void)
-{
- return gstage_gpa_bits;
-}
diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
index 8ad7b31f5939..fe028b4274df 100644
--- a/arch/riscv/kvm/vcpu.c
+++ b/arch/riscv/kvm/vcpu.c
@@ -632,7 +632,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
}
}
- kvm_riscv_gstage_update_hgatp(vcpu);
+ kvm_riscv_mmu_update_hgatp(vcpu);
kvm_riscv_vcpu_timer_restore(vcpu);
@@ -717,7 +717,7 @@ static void kvm_riscv_check_vcpu_requests(struct kvm_vcpu *vcpu)
kvm_riscv_reset_vcpu(vcpu, true);
if (kvm_check_request(KVM_REQ_UPDATE_HGATP, vcpu))
- kvm_riscv_gstage_update_hgatp(vcpu);
+ kvm_riscv_mmu_update_hgatp(vcpu);
if (kvm_check_request(KVM_REQ_FENCE_I, vcpu))
kvm_riscv_fence_i_process(vcpu);
diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
index 6b4694bc07ea..0bb0c51e3c89 100644
--- a/arch/riscv/kvm/vcpu_exit.c
+++ b/arch/riscv/kvm/vcpu_exit.c
@@ -43,8 +43,9 @@ static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
};
}
- ret = kvm_riscv_gstage_map(vcpu, memslot, fault_addr, hva,
- (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false, &host_map);
+ ret = kvm_riscv_mmu_map(vcpu, memslot, fault_addr, hva,
+ (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false,
+ &host_map);
if (ret < 0)
return ret;
diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c
index 8601cf29e5f8..66d91ae6e9b2 100644
--- a/arch/riscv/kvm/vm.c
+++ b/arch/riscv/kvm/vm.c
@@ -32,13 +32,13 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
{
int r;
- r = kvm_riscv_gstage_alloc_pgd(kvm);
+ r = kvm_riscv_mmu_alloc_pgd(kvm);
if (r)
return r;
r = kvm_riscv_gstage_vmid_init(kvm);
if (r) {
- kvm_riscv_gstage_free_pgd(kvm);
+ kvm_riscv_mmu_free_pgd(kvm);
return r;
}
@@ -200,7 +200,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
r = KVM_USER_MEM_SLOTS;
break;
case KVM_CAP_VM_GPA_BITS:
- r = kvm_riscv_gstage_gpa_bits();
+ r = kvm_riscv_gstage_gpa_bits;
break;
default:
r = 0;
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (10 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management Anup Patel
@ 2025-06-18 11:35 ` Anup Patel
2025-06-25 7:58 ` Nutty Liu
2025-06-24 6:16 ` [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
2025-06-25 7:59 ` Atish Patra
13 siblings, 1 reply; 28+ messages in thread
From: Anup Patel @ 2025-06-18 11:35 UTC (permalink / raw)
To: Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Anup Patel
Currently, all kvm_riscv_hfence_xyz() APIs assume VMID to be the
host VMID of the Guest/VM which resticts use of these APIs only
for host TLB maintenance. Let's allow passing VMID as a parameter
to all kvm_riscv_hfence_xyz() APIs so that they can be re-used
for nested virtualization related TLB maintenance.
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
---
arch/riscv/include/asm/kvm_tlb.h | 17 ++++++---
arch/riscv/kvm/gstage.c | 3 +-
arch/riscv/kvm/tlb.c | 61 ++++++++++++++++++++-----------
arch/riscv/kvm/vcpu_sbi_replace.c | 17 +++++----
arch/riscv/kvm/vcpu_sbi_v01.c | 25 ++++++-------
5 files changed, 73 insertions(+), 50 deletions(-)
diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
index f67e03edeaec..38a2f933ad3a 100644
--- a/arch/riscv/include/asm/kvm_tlb.h
+++ b/arch/riscv/include/asm/kvm_tlb.h
@@ -11,9 +11,11 @@
enum kvm_riscv_hfence_type {
KVM_RISCV_HFENCE_UNKNOWN = 0,
KVM_RISCV_HFENCE_GVMA_VMID_GPA,
+ KVM_RISCV_HFENCE_GVMA_VMID_ALL,
KVM_RISCV_HFENCE_VVMA_ASID_GVA,
KVM_RISCV_HFENCE_VVMA_ASID_ALL,
KVM_RISCV_HFENCE_VVMA_GVA,
+ KVM_RISCV_HFENCE_VVMA_ALL
};
struct kvm_riscv_hfence {
@@ -59,21 +61,24 @@ void kvm_riscv_fence_i(struct kvm *kvm,
void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
gpa_t gpa, gpa_t gpsz,
- unsigned long order);
+ unsigned long order, unsigned long vmid);
void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask);
+ unsigned long hbase, unsigned long hmask,
+ unsigned long vmid);
void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
unsigned long gva, unsigned long gvsz,
- unsigned long order, unsigned long asid);
+ unsigned long order, unsigned long asid,
+ unsigned long vmid);
void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
- unsigned long asid);
+ unsigned long asid, unsigned long vmid);
void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
unsigned long gva, unsigned long gvsz,
- unsigned long order);
+ unsigned long order, unsigned long vmid);
void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask);
+ unsigned long hbase, unsigned long hmask,
+ unsigned long vmid);
#endif
diff --git a/arch/riscv/kvm/gstage.c b/arch/riscv/kvm/gstage.c
index 9c7c44f09b05..24c270d6d0e2 100644
--- a/arch/riscv/kvm/gstage.c
+++ b/arch/riscv/kvm/gstage.c
@@ -117,7 +117,8 @@ static void gstage_tlb_flush(struct kvm_gstage *gstage, u32 level, gpa_t addr)
if (gstage->flags & KVM_GSTAGE_FLAGS_LOCAL)
kvm_riscv_local_hfence_gvma_vmid_gpa(gstage->vmid, addr, BIT(order), order);
else
- kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order);
+ kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order,
+ gstage->vmid);
}
int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
index 349fcfc93f54..3c5a70a2b927 100644
--- a/arch/riscv/kvm/tlb.c
+++ b/arch/riscv/kvm/tlb.c
@@ -251,6 +251,12 @@ void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
kvm_riscv_local_hfence_gvma_vmid_gpa(d.vmid, d.addr,
d.size, d.order);
break;
+ case KVM_RISCV_HFENCE_GVMA_VMID_ALL:
+ if (kvm_riscv_nacl_available())
+ nacl_hfence_gvma_vmid_all(nacl_shmem(), d.vmid);
+ else
+ kvm_riscv_local_hfence_gvma_vmid_all(d.vmid);
+ break;
case KVM_RISCV_HFENCE_VVMA_ASID_GVA:
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
if (kvm_riscv_nacl_available())
@@ -276,6 +282,13 @@ void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
kvm_riscv_local_hfence_vvma_gva(d.vmid, d.addr,
d.size, d.order);
break;
+ case KVM_RISCV_HFENCE_VVMA_ALL:
+ kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_RCVD);
+ if (kvm_riscv_nacl_available())
+ nacl_hfence_vvma_all(nacl_shmem(), d.vmid);
+ else
+ kvm_riscv_local_hfence_vvma_all(d.vmid);
+ break;
default:
break;
}
@@ -328,14 +341,13 @@ void kvm_riscv_fence_i(struct kvm *kvm,
void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
gpa_t gpa, gpa_t gpsz,
- unsigned long order)
+ unsigned long order, unsigned long vmid)
{
- struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_GVMA_VMID_GPA;
data.asid = 0;
- data.vmid = READ_ONCE(v->vmid);
+ data.vmid = vmid;
data.addr = gpa;
data.size = gpsz;
data.order = order;
@@ -344,23 +356,28 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
}
void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask)
+ unsigned long hbase, unsigned long hmask,
+ unsigned long vmid)
{
- make_xfence_request(kvm, hbase, hmask, KVM_REQ_TLB_FLUSH,
- KVM_REQ_TLB_FLUSH, NULL);
+ struct kvm_riscv_hfence data = {0};
+
+ data.type = KVM_RISCV_HFENCE_GVMA_VMID_ALL;
+ data.vmid = vmid;
+ make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+ KVM_REQ_TLB_FLUSH, &data);
}
void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
unsigned long gva, unsigned long gvsz,
- unsigned long order, unsigned long asid)
+ unsigned long order, unsigned long asid,
+ unsigned long vmid)
{
- struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_VVMA_ASID_GVA;
data.asid = asid;
- data.vmid = READ_ONCE(v->vmid);
+ data.vmid = vmid;
data.addr = gva;
data.size = gvsz;
data.order = order;
@@ -370,15 +387,13 @@ void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
- unsigned long asid)
+ unsigned long asid, unsigned long vmid)
{
- struct kvm_vmid *v = &kvm->arch.vmid;
- struct kvm_riscv_hfence data;
+ struct kvm_riscv_hfence data = {0};
data.type = KVM_RISCV_HFENCE_VVMA_ASID_ALL;
data.asid = asid;
- data.vmid = READ_ONCE(v->vmid);
- data.addr = data.size = data.order = 0;
+ data.vmid = vmid;
make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
KVM_REQ_HFENCE_VVMA_ALL, &data);
}
@@ -386,14 +401,13 @@ void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
unsigned long hbase, unsigned long hmask,
unsigned long gva, unsigned long gvsz,
- unsigned long order)
+ unsigned long order, unsigned long vmid)
{
- struct kvm_vmid *v = &kvm->arch.vmid;
struct kvm_riscv_hfence data;
data.type = KVM_RISCV_HFENCE_VVMA_GVA;
data.asid = 0;
- data.vmid = READ_ONCE(v->vmid);
+ data.vmid = vmid;
data.addr = gva;
data.size = gvsz;
data.order = order;
@@ -402,16 +416,21 @@ void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
}
void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
- unsigned long hbase, unsigned long hmask)
+ unsigned long hbase, unsigned long hmask,
+ unsigned long vmid)
{
- make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_VVMA_ALL,
- KVM_REQ_HFENCE_VVMA_ALL, NULL);
+ struct kvm_riscv_hfence data = {0};
+
+ data.type = KVM_RISCV_HFENCE_VVMA_ALL;
+ data.vmid = vmid;
+ make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
+ KVM_REQ_HFENCE_VVMA_ALL, &data);
}
int kvm_arch_flush_remote_tlbs_range(struct kvm *kvm, gfn_t gfn, u64 nr_pages)
{
kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0,
gfn << PAGE_SHIFT, nr_pages << PAGE_SHIFT,
- PAGE_SHIFT);
+ PAGE_SHIFT, READ_ONCE(kvm->arch.vmid.vmid));
return 0;
}
diff --git a/arch/riscv/kvm/vcpu_sbi_replace.c b/arch/riscv/kvm/vcpu_sbi_replace.c
index b17fad091bab..b490ed1428a6 100644
--- a/arch/riscv/kvm/vcpu_sbi_replace.c
+++ b/arch/riscv/kvm/vcpu_sbi_replace.c
@@ -96,6 +96,7 @@ static int kvm_sbi_ext_rfence_handler(struct kvm_vcpu *vcpu, struct kvm_run *run
unsigned long hmask = cp->a0;
unsigned long hbase = cp->a1;
unsigned long funcid = cp->a6;
+ unsigned long vmid;
switch (funcid) {
case SBI_EXT_RFENCE_REMOTE_FENCE_I:
@@ -103,22 +104,22 @@ static int kvm_sbi_ext_rfence_handler(struct kvm_vcpu *vcpu, struct kvm_run *run
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_FENCE_I_SENT);
break;
case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA:
+ vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
if ((cp->a2 == 0 && cp->a3 == 0) || cp->a3 == -1UL)
- kvm_riscv_hfence_vvma_all(vcpu->kvm, hbase, hmask);
+ kvm_riscv_hfence_vvma_all(vcpu->kvm, hbase, hmask, vmid);
else
kvm_riscv_hfence_vvma_gva(vcpu->kvm, hbase, hmask,
- cp->a2, cp->a3, PAGE_SHIFT);
+ cp->a2, cp->a3, PAGE_SHIFT, vmid);
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_SENT);
break;
case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA_ASID:
+ vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
if ((cp->a2 == 0 && cp->a3 == 0) || cp->a3 == -1UL)
- kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
- hbase, hmask, cp->a4);
+ kvm_riscv_hfence_vvma_asid_all(vcpu->kvm, hbase, hmask,
+ cp->a4, vmid);
else
- kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
- hbase, hmask,
- cp->a2, cp->a3,
- PAGE_SHIFT, cp->a4);
+ kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm, hbase, hmask, cp->a2,
+ cp->a3, PAGE_SHIFT, cp->a4, vmid);
kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_SENT);
break;
case SBI_EXT_RFENCE_REMOTE_HFENCE_GVMA:
diff --git a/arch/riscv/kvm/vcpu_sbi_v01.c b/arch/riscv/kvm/vcpu_sbi_v01.c
index 8f4c4fa16227..368dfddd23d9 100644
--- a/arch/riscv/kvm/vcpu_sbi_v01.c
+++ b/arch/riscv/kvm/vcpu_sbi_v01.c
@@ -23,6 +23,7 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
struct kvm *kvm = vcpu->kvm;
struct kvm_cpu_context *cp = &vcpu->arch.guest_context;
struct kvm_cpu_trap *utrap = retdata->utrap;
+ unsigned long vmid;
switch (cp->a7) {
case SBI_EXT_0_1_CONSOLE_GETCHAR:
@@ -78,25 +79,21 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
if (cp->a7 == SBI_EXT_0_1_REMOTE_FENCE_I)
kvm_riscv_fence_i(vcpu->kvm, 0, hmask);
else if (cp->a7 == SBI_EXT_0_1_REMOTE_SFENCE_VMA) {
+ vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
if (cp->a1 == 0 && cp->a2 == 0)
- kvm_riscv_hfence_vvma_all(vcpu->kvm,
- 0, hmask);
+ kvm_riscv_hfence_vvma_all(vcpu->kvm, 0, hmask, vmid);
else
- kvm_riscv_hfence_vvma_gva(vcpu->kvm,
- 0, hmask,
- cp->a1, cp->a2,
- PAGE_SHIFT);
+ kvm_riscv_hfence_vvma_gva(vcpu->kvm, 0, hmask, cp->a1,
+ cp->a2, PAGE_SHIFT, vmid);
} else {
+ vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
if (cp->a1 == 0 && cp->a2 == 0)
- kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
- 0, hmask,
- cp->a3);
+ kvm_riscv_hfence_vvma_asid_all(vcpu->kvm, 0, hmask,
+ cp->a3, vmid);
else
- kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
- 0, hmask,
- cp->a1, cp->a2,
- PAGE_SHIFT,
- cp->a3);
+ kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm, 0, hmask,
+ cp->a1, cp->a2, PAGE_SHIFT,
+ cp->a3, vmid);
}
break;
default:
--
2.43.0
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply related [flat|nested] 28+ messages in thread
* Re: [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
@ 2025-06-24 3:32 ` Atish Patra
2025-06-25 6:48 ` Nutty Liu
1 sibling, 0 replies; 28+ messages in thread
From: Atish Patra @ 2025-06-24 3:32 UTC (permalink / raw)
To: Anup Patel
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/25 4:35 AM, Anup Patel wrote:
> The kvm_riscv_vcpu_alloc_vector_context() does return an error code
> upon failure so don't ignore this in kvm_arch_vcpu_create().
>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/kvm/vcpu.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index 303aa0a8a5a1..b467dc1f4c7f 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -148,8 +148,9 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
>
> spin_lock_init(&vcpu->arch.reset_state.lock);
>
> - if (kvm_riscv_vcpu_alloc_vector_context(vcpu))
> - return -ENOMEM;
> + rc = kvm_riscv_vcpu_alloc_vector_context(vcpu);
> + if (rc)
> + return rc;
>
> /* Setup VCPU timer */
> kvm_riscv_vcpu_timer_init(vcpu);
Reviewed-by: Atish Patra <atishp@rivosinc.com>
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
2025-06-18 11:35 ` [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init() Anup Patel
@ 2025-06-24 3:35 ` Atish Patra
2025-06-25 6:50 ` Nutty Liu
1 sibling, 0 replies; 28+ messages in thread
From: Atish Patra @ 2025-06-24 3:35 UTC (permalink / raw)
To: Anup Patel
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel, Nutty Liu
On 6/18/25 4:35 AM, Anup Patel wrote:
> The kvm_riscv_vcpu_aia_init() does not return any failure so drop
> the return value which is always zero.
>
> Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_aia.h | 2 +-
> arch/riscv/kvm/aia_device.c | 6 ++----
> arch/riscv/kvm/vcpu.c | 4 +---
> 3 files changed, 4 insertions(+), 8 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_aia.h b/arch/riscv/include/asm/kvm_aia.h
> index 3b643b9efc07..0a0f12496f00 100644
> --- a/arch/riscv/include/asm/kvm_aia.h
> +++ b/arch/riscv/include/asm/kvm_aia.h
> @@ -147,7 +147,7 @@ int kvm_riscv_vcpu_aia_rmw_ireg(struct kvm_vcpu *vcpu, unsigned int csr_num,
>
> int kvm_riscv_vcpu_aia_update(struct kvm_vcpu *vcpu);
> void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu);
> -int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
> +void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
> void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu);
>
> int kvm_riscv_aia_inject_msi_by_id(struct kvm *kvm, u32 hart_index,
> diff --git a/arch/riscv/kvm/aia_device.c b/arch/riscv/kvm/aia_device.c
> index 806c41931cde..b195a93add1c 100644
> --- a/arch/riscv/kvm/aia_device.c
> +++ b/arch/riscv/kvm/aia_device.c
> @@ -509,12 +509,12 @@ void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu)
> kvm_riscv_vcpu_aia_imsic_reset(vcpu);
> }
>
> -int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> +void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> {
> struct kvm_vcpu_aia *vaia = &vcpu->arch.aia_context;
>
> if (!kvm_riscv_aia_available())
> - return 0;
> + return;
>
> /*
> * We don't do any memory allocations over here because these
> @@ -526,8 +526,6 @@ int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> /* Initialize default values in AIA vcpu context */
> vaia->imsic_addr = KVM_RISCV_AIA_UNDEF_ADDR;
> vaia->hart_index = vcpu->vcpu_idx;
> -
> - return 0;
> }
>
> void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu)
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index b467dc1f4c7f..f9fb3dbbe0c3 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -159,9 +159,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
> kvm_riscv_vcpu_pmu_init(vcpu);
>
> /* Setup VCPU AIA */
> - rc = kvm_riscv_vcpu_aia_init(vcpu);
> - if (rc)
> - return rc;
> + kvm_riscv_vcpu_aia_init(vcpu);
>
> /*
> * Setup SBI extensions
Reviewed-by: Atish Patra <atishp@rivosinc.com>
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 00/12] MMU related improvements for KVM RISC-V
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (11 preceding siblings ...)
2025-06-18 11:35 ` [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs Anup Patel
@ 2025-06-24 6:16 ` Anup Patel
2025-06-25 7:59 ` Atish Patra
13 siblings, 0 replies; 28+ messages in thread
From: Anup Patel @ 2025-06-24 6:16 UTC (permalink / raw)
To: Anup Patel
Cc: Atish Patra, Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti,
Andrew Jones, kvm, kvm-riscv, linux-riscv, linux-kernel
On Wed, Jun 18, 2025 at 5:05 PM Anup Patel <apatel@ventanamicro.com> wrote:
>
> This series primarily has various MMU improvements for KVM RISC-V
> and it also serves as a preparatory series for the upcoming nested
> virtualization support.
>
> PATCH1 to PATCH4: Few cosmetic improvements
> PATCH5 to PATCH7: TLB maintenance and SBI NACL related improvements
> PATCH8 to PATCH12: MMU related preparatory work for nested virtualization
>
> These patches can also be found in the riscv_kvm_mmu_imp_v3 branch
> at: https://github.com/avpatel/linux.git
>
> Changes since v2:
> - Rebased upon Linux-6.16-rc2 and latest riscv_kvm_queue branch
> - Added Reviewed-by tags to appropriate patches
>
> Changes since v1:
> - Rebased upon Linux-6.16-rc1
> - Dropped PATCH1 and PATCH2 of v1 series since these are queued
> as fixes for Linux-6.16
> - Addressed Atish's comment on PATCH1 in this series
> - Added new PATCH7 in this series
>
> Anup Patel (12):
> RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
> RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
> RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize()
> RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with
> KVM_REQ_TLB_FLUSH
> RISC-V: KVM: Don't flush TLB when PTE is unchanged
> RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range()
> RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect()
> RISC-V: KVM: Factor-out MMU related declarations into separate headers
> RISC-V: KVM: Introduce struct kvm_gstage_mapping
> RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence
> RISC-V: KVM: Factor-out g-stage page table management
> RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs
Queued this series for Linux-6.17
Thanks,
Anup
>
> arch/riscv/include/asm/kvm_aia.h | 2 +-
> arch/riscv/include/asm/kvm_gstage.h | 72 ++++
> arch/riscv/include/asm/kvm_host.h | 103 +-----
> arch/riscv/include/asm/kvm_mmu.h | 21 ++
> arch/riscv/include/asm/kvm_tlb.h | 84 +++++
> arch/riscv/include/asm/kvm_vmid.h | 27 ++
> arch/riscv/kvm/Makefile | 1 +
> arch/riscv/kvm/aia_device.c | 6 +-
> arch/riscv/kvm/aia_imsic.c | 12 +-
> arch/riscv/kvm/gstage.c | 338 +++++++++++++++++++
> arch/riscv/kvm/main.c | 3 +-
> arch/riscv/kvm/mmu.c | 499 ++++++----------------------
> arch/riscv/kvm/tlb.c | 110 +++---
> arch/riscv/kvm/vcpu.c | 26 +-
> arch/riscv/kvm/vcpu_exit.c | 20 +-
> arch/riscv/kvm/vcpu_sbi_replace.c | 17 +-
> arch/riscv/kvm/vcpu_sbi_v01.c | 25 +-
> arch/riscv/kvm/vm.c | 7 +-
> arch/riscv/kvm/vmid.c | 25 ++
> 19 files changed, 795 insertions(+), 603 deletions(-)
> create mode 100644 arch/riscv/include/asm/kvm_gstage.h
> create mode 100644 arch/riscv/include/asm/kvm_mmu.h
> create mode 100644 arch/riscv/include/asm/kvm_tlb.h
> create mode 100644 arch/riscv/include/asm/kvm_vmid.h
> create mode 100644 arch/riscv/kvm/gstage.c
>
> --
> 2.43.0
>
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
2025-06-24 3:32 ` Atish Patra
@ 2025-06-25 6:48 ` Nutty Liu
1 sibling, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 6:48 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The kvm_riscv_vcpu_alloc_vector_context() does return an error code
> upon failure so don't ignore this in kvm_arch_vcpu_create().
>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/kvm/vcpu.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index 303aa0a8a5a1..b467dc1f4c7f 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -148,8 +148,9 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
>
> spin_lock_init(&vcpu->arch.reset_state.lock);
>
> - if (kvm_riscv_vcpu_alloc_vector_context(vcpu))
> - return -ENOMEM;
> + rc = kvm_riscv_vcpu_alloc_vector_context(vcpu);
> + if (rc)
> + return rc;
>
> /* Setup VCPU timer */
> kvm_riscv_vcpu_timer_init(vcpu);
Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
2025-06-18 11:35 ` [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init() Anup Patel
2025-06-24 3:35 ` Atish Patra
@ 2025-06-25 6:50 ` Nutty Liu
1 sibling, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 6:50 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The kvm_riscv_vcpu_aia_init() does not return any failure so drop
> the return value which is always zero.
>
> Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_aia.h | 2 +-
> arch/riscv/kvm/aia_device.c | 6 ++----
> arch/riscv/kvm/vcpu.c | 4 +---
> 3 files changed, 4 insertions(+), 8 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_aia.h b/arch/riscv/include/asm/kvm_aia.h
> index 3b643b9efc07..0a0f12496f00 100644
> --- a/arch/riscv/include/asm/kvm_aia.h
> +++ b/arch/riscv/include/asm/kvm_aia.h
> @@ -147,7 +147,7 @@ int kvm_riscv_vcpu_aia_rmw_ireg(struct kvm_vcpu *vcpu, unsigned int csr_num,
>
> int kvm_riscv_vcpu_aia_update(struct kvm_vcpu *vcpu);
> void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu);
> -int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
> +void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu);
> void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu);
>
> int kvm_riscv_aia_inject_msi_by_id(struct kvm *kvm, u32 hart_index,
> diff --git a/arch/riscv/kvm/aia_device.c b/arch/riscv/kvm/aia_device.c
> index 806c41931cde..b195a93add1c 100644
> --- a/arch/riscv/kvm/aia_device.c
> +++ b/arch/riscv/kvm/aia_device.c
> @@ -509,12 +509,12 @@ void kvm_riscv_vcpu_aia_reset(struct kvm_vcpu *vcpu)
> kvm_riscv_vcpu_aia_imsic_reset(vcpu);
> }
>
> -int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> +void kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> {
> struct kvm_vcpu_aia *vaia = &vcpu->arch.aia_context;
>
> if (!kvm_riscv_aia_available())
> - return 0;
> + return;
>
> /*
> * We don't do any memory allocations over here because these
> @@ -526,8 +526,6 @@ int kvm_riscv_vcpu_aia_init(struct kvm_vcpu *vcpu)
> /* Initialize default values in AIA vcpu context */
> vaia->imsic_addr = KVM_RISCV_AIA_UNDEF_ADDR;
> vaia->hart_index = vcpu->vcpu_idx;
> -
> - return 0;
> }
>
> void kvm_riscv_vcpu_aia_deinit(struct kvm_vcpu *vcpu)
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index b467dc1f4c7f..f9fb3dbbe0c3 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -159,9 +159,7 @@ int kvm_arch_vcpu_create(struct kvm_vcpu *vcpu)
> kvm_riscv_vcpu_pmu_init(vcpu);
>
> /* Setup VCPU AIA */
> - rc = kvm_riscv_vcpu_aia_init(vcpu);
> - if (rc)
> - return rc;
> + kvm_riscv_vcpu_aia_init(vcpu);
>
> /*
> * Setup SBI extensions
Reviewed-by: Nutty Liu<liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH
2025-06-18 11:35 ` [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH Anup Patel
@ 2025-06-25 7:35 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:35 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The KVM_REQ_HFENCE_GVMA_VMID_ALL is same as KVM_REQ_TLB_FLUSH so
> to avoid confusion let's replace KVM_REQ_HFENCE_GVMA_VMID_ALL with
> KVM_REQ_TLB_FLUSH. Also, rename kvm_riscv_hfence_gvma_vmid_all_process()
> to kvm_riscv_tlb_flush_process().
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_host.h | 4 ++--
> arch/riscv/kvm/tlb.c | 8 ++++----
> arch/riscv/kvm/vcpu.c | 8 ++------
> 3 files changed, 8 insertions(+), 12 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
> index 8aa705ac75a5..ff1f76d6f177 100644
> --- a/arch/riscv/include/asm/kvm_host.h
> +++ b/arch/riscv/include/asm/kvm_host.h
> @@ -37,7 +37,6 @@
> #define KVM_REQ_UPDATE_HGATP KVM_ARCH_REQ(2)
> #define KVM_REQ_FENCE_I \
> KVM_ARCH_REQ_FLAGS(3, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
> -#define KVM_REQ_HFENCE_GVMA_VMID_ALL KVM_REQ_TLB_FLUSH
> #define KVM_REQ_HFENCE_VVMA_ALL \
> KVM_ARCH_REQ_FLAGS(4, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
> #define KVM_REQ_HFENCE \
> @@ -331,8 +330,9 @@ void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
> unsigned long order);
> void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
>
> +void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
> +
> void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
> -void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu);
> void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
> void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
>
> diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
> index b3461bfd9756..da98ca801d31 100644
> --- a/arch/riscv/kvm/tlb.c
> +++ b/arch/riscv/kvm/tlb.c
> @@ -162,7 +162,7 @@ void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu)
> local_flush_icache_all();
> }
>
> -void kvm_riscv_hfence_gvma_vmid_all_process(struct kvm_vcpu *vcpu)
> +void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu)
> {
> struct kvm_vmid *v = &vcpu->kvm->arch.vmid;
> unsigned long vmid = READ_ONCE(v->vmid);
> @@ -342,14 +342,14 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> data.size = gpsz;
> data.order = order;
> make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
> - KVM_REQ_HFENCE_GVMA_VMID_ALL, &data);
> + KVM_REQ_TLB_FLUSH, &data);
> }
>
> void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask)
> {
> - make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_GVMA_VMID_ALL,
> - KVM_REQ_HFENCE_GVMA_VMID_ALL, NULL);
> + make_xfence_request(kvm, hbase, hmask, KVM_REQ_TLB_FLUSH,
> + KVM_REQ_TLB_FLUSH, NULL);
> }
>
> void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index a2dd4161e5a4..6eb11c913b13 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -721,12 +721,8 @@ static void kvm_riscv_check_vcpu_requests(struct kvm_vcpu *vcpu)
> if (kvm_check_request(KVM_REQ_FENCE_I, vcpu))
> kvm_riscv_fence_i_process(vcpu);
>
> - /*
> - * The generic KVM_REQ_TLB_FLUSH is same as
> - * KVM_REQ_HFENCE_GVMA_VMID_ALL
> - */
> - if (kvm_check_request(KVM_REQ_HFENCE_GVMA_VMID_ALL, vcpu))
> - kvm_riscv_hfence_gvma_vmid_all_process(vcpu);
> + if (kvm_check_request(KVM_REQ_TLB_FLUSH, vcpu))
> + kvm_riscv_tlb_flush_process(vcpu);
>
> if (kvm_check_request(KVM_REQ_HFENCE_VVMA_ALL, vcpu))
> kvm_riscv_hfence_vvma_all_process(vcpu);
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged
2025-06-18 11:35 ` [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged Anup Patel
@ 2025-06-25 7:38 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:38 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The gstage_set_pte() and gstage_op_pte() should flush TLB only when
> a leaf PTE changes so that unnecessary TLB flushes can be avoided.
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/kvm/mmu.c | 14 +++++++++-----
> 1 file changed, 9 insertions(+), 5 deletions(-)
>
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index 1087ea74567b..29f1bd853a66 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -167,9 +167,11 @@ static int gstage_set_pte(struct kvm *kvm, u32 level,
> ptep = &next_ptep[gstage_pte_index(addr, current_level)];
> }
>
> - set_pte(ptep, *new_pte);
> - if (gstage_pte_leaf(ptep))
> - gstage_remote_tlb_flush(kvm, current_level, addr);
> + if (pte_val(*ptep) != pte_val(*new_pte)) {
> + set_pte(ptep, *new_pte);
> + if (gstage_pte_leaf(ptep))
> + gstage_remote_tlb_flush(kvm, current_level, addr);
> + }
>
> return 0;
> }
> @@ -229,7 +231,7 @@ static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
> pte_t *ptep, u32 ptep_level, enum gstage_op op)
> {
> int i, ret;
> - pte_t *next_ptep;
> + pte_t old_pte, *next_ptep;
> u32 next_ptep_level;
> unsigned long next_page_size, page_size;
>
> @@ -258,11 +260,13 @@ static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
> if (op == GSTAGE_OP_CLEAR)
> put_page(virt_to_page(next_ptep));
> } else {
> + old_pte = *ptep;
> if (op == GSTAGE_OP_CLEAR)
> set_pte(ptep, __pte(0));
> else if (op == GSTAGE_OP_WP)
> set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
> - gstage_remote_tlb_flush(kvm, ptep_level, addr);
> + if (pte_val(*ptep) != pte_val(old_pte))
> + gstage_remote_tlb_flush(kvm, ptep_level, addr);
> }
> }
>
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range()
2025-06-18 11:35 ` [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range() Anup Patel
@ 2025-06-25 7:39 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:39 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The kvm_arch_flush_remote_tlbs_range() expected by KVM core can be
> easily implemented for RISC-V using kvm_riscv_hfence_gvma_vmid_gpa()
> hence provide it.
>
> Also with kvm_arch_flush_remote_tlbs_range() available for RISC-V, the
> mmu_wp_memory_region() can happily use kvm_flush_remote_tlbs_memslot()
> instead of kvm_flush_remote_tlbs().
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_host.h | 2 ++
> arch/riscv/kvm/mmu.c | 2 +-
> arch/riscv/kvm/tlb.c | 8 ++++++++
> 3 files changed, 11 insertions(+), 1 deletion(-)
>
> diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
> index ff1f76d6f177..6162575e2177 100644
> --- a/arch/riscv/include/asm/kvm_host.h
> +++ b/arch/riscv/include/asm/kvm_host.h
> @@ -43,6 +43,8 @@
> KVM_ARCH_REQ_FLAGS(5, KVM_REQUEST_WAIT | KVM_REQUEST_NO_WAKEUP)
> #define KVM_REQ_STEAL_UPDATE KVM_ARCH_REQ(6)
>
> +#define __KVM_HAVE_ARCH_FLUSH_REMOTE_TLBS_RANGE
> +
> #define KVM_HEDELEG_DEFAULT (BIT(EXC_INST_MISALIGNED) | \
> BIT(EXC_BREAKPOINT) | \
> BIT(EXC_SYSCALL) | \
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index 29f1bd853a66..a5387927a1c1 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -344,7 +344,7 @@ static void gstage_wp_memory_region(struct kvm *kvm, int slot)
> spin_lock(&kvm->mmu_lock);
> gstage_wp_range(kvm, start, end);
> spin_unlock(&kvm->mmu_lock);
> - kvm_flush_remote_tlbs(kvm);
> + kvm_flush_remote_tlbs_memslot(kvm, memslot);
> }
>
> int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
> index da98ca801d31..f46a27658c2e 100644
> --- a/arch/riscv/kvm/tlb.c
> +++ b/arch/riscv/kvm/tlb.c
> @@ -403,3 +403,11 @@ void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
> make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_VVMA_ALL,
> KVM_REQ_HFENCE_VVMA_ALL, NULL);
> }
> +
> +int kvm_arch_flush_remote_tlbs_range(struct kvm *kvm, gfn_t gfn, u64 nr_pages)
> +{
> + kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0,
> + gfn << PAGE_SHIFT, nr_pages << PAGE_SHIFT,
> + PAGE_SHIFT);
> + return 0;
> +}
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect()
2025-06-18 11:35 ` [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect() Anup Patel
@ 2025-06-25 7:39 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:39 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The H-extension CSRs accessed by kvm_riscv_vcpu_trap_redirect() will
> trap when KVM RISC-V is running as Guest/VM hence remove these traps
> by using ncsr_xyz() instead of csr_xyz().
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/kvm/vcpu_exit.c | 13 +++++++------
> 1 file changed, 7 insertions(+), 6 deletions(-)
>
> diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
> index 6e0c18412795..85c43c83e3b9 100644
> --- a/arch/riscv/kvm/vcpu_exit.c
> +++ b/arch/riscv/kvm/vcpu_exit.c
> @@ -9,6 +9,7 @@
> #include <linux/kvm_host.h>
> #include <asm/csr.h>
> #include <asm/insn-def.h>
> +#include <asm/kvm_nacl.h>
>
> static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
> struct kvm_cpu_trap *trap)
> @@ -135,7 +136,7 @@ unsigned long kvm_riscv_vcpu_unpriv_read(struct kvm_vcpu *vcpu,
> void kvm_riscv_vcpu_trap_redirect(struct kvm_vcpu *vcpu,
> struct kvm_cpu_trap *trap)
> {
> - unsigned long vsstatus = csr_read(CSR_VSSTATUS);
> + unsigned long vsstatus = ncsr_read(CSR_VSSTATUS);
>
> /* Change Guest SSTATUS.SPP bit */
> vsstatus &= ~SR_SPP;
> @@ -151,15 +152,15 @@ void kvm_riscv_vcpu_trap_redirect(struct kvm_vcpu *vcpu,
> vsstatus &= ~SR_SIE;
>
> /* Update Guest SSTATUS */
> - csr_write(CSR_VSSTATUS, vsstatus);
> + ncsr_write(CSR_VSSTATUS, vsstatus);
>
> /* Update Guest SCAUSE, STVAL, and SEPC */
> - csr_write(CSR_VSCAUSE, trap->scause);
> - csr_write(CSR_VSTVAL, trap->stval);
> - csr_write(CSR_VSEPC, trap->sepc);
> + ncsr_write(CSR_VSCAUSE, trap->scause);
> + ncsr_write(CSR_VSTVAL, trap->stval);
> + ncsr_write(CSR_VSEPC, trap->sepc);
>
> /* Set Guest PC to Guest exception vector */
> - vcpu->arch.guest_context.sepc = csr_read(CSR_VSTVEC);
> + vcpu->arch.guest_context.sepc = ncsr_read(CSR_VSTVEC);
>
> /* Set Guest privilege mode to supervisor */
> vcpu->arch.guest_context.sstatus |= SR_SPP;
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers
2025-06-18 11:35 ` [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers Anup Patel
@ 2025-06-25 7:40 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:40 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The MMU, TLB, and VMID management for KVM RISC-V already exists as
> seprate sources so create separate headers along these lines. This
> further simplifies asm/kvm_host.h header.
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_host.h | 100 +-----------------------------
> arch/riscv/include/asm/kvm_mmu.h | 26 ++++++++
> arch/riscv/include/asm/kvm_tlb.h | 78 +++++++++++++++++++++++
> arch/riscv/include/asm/kvm_vmid.h | 27 ++++++++
> arch/riscv/kvm/aia_imsic.c | 1 +
> arch/riscv/kvm/main.c | 1 +
> arch/riscv/kvm/mmu.c | 1 +
> arch/riscv/kvm/tlb.c | 2 +
> arch/riscv/kvm/vcpu.c | 1 +
> arch/riscv/kvm/vcpu_exit.c | 1 +
> arch/riscv/kvm/vm.c | 1 +
> arch/riscv/kvm/vmid.c | 2 +
> 12 files changed, 143 insertions(+), 98 deletions(-)
> create mode 100644 arch/riscv/include/asm/kvm_mmu.h
> create mode 100644 arch/riscv/include/asm/kvm_tlb.h
> create mode 100644 arch/riscv/include/asm/kvm_vmid.h
>
> diff --git a/arch/riscv/include/asm/kvm_host.h b/arch/riscv/include/asm/kvm_host.h
> index 6162575e2177..bd5341efa127 100644
> --- a/arch/riscv/include/asm/kvm_host.h
> +++ b/arch/riscv/include/asm/kvm_host.h
> @@ -16,6 +16,8 @@
> #include <asm/hwcap.h>
> #include <asm/kvm_aia.h>
> #include <asm/ptrace.h>
> +#include <asm/kvm_tlb.h>
> +#include <asm/kvm_vmid.h>
> #include <asm/kvm_vcpu_fp.h>
> #include <asm/kvm_vcpu_insn.h>
> #include <asm/kvm_vcpu_sbi.h>
> @@ -56,24 +58,6 @@
> BIT(IRQ_VS_TIMER) | \
> BIT(IRQ_VS_EXT))
>
> -enum kvm_riscv_hfence_type {
> - KVM_RISCV_HFENCE_UNKNOWN = 0,
> - KVM_RISCV_HFENCE_GVMA_VMID_GPA,
> - KVM_RISCV_HFENCE_VVMA_ASID_GVA,
> - KVM_RISCV_HFENCE_VVMA_ASID_ALL,
> - KVM_RISCV_HFENCE_VVMA_GVA,
> -};
> -
> -struct kvm_riscv_hfence {
> - enum kvm_riscv_hfence_type type;
> - unsigned long asid;
> - unsigned long order;
> - gpa_t addr;
> - gpa_t size;
> -};
> -
> -#define KVM_RISCV_VCPU_MAX_HFENCE 64
> -
> struct kvm_vm_stat {
> struct kvm_vm_stat_generic generic;
> };
> @@ -99,15 +83,6 @@ struct kvm_vcpu_stat {
> struct kvm_arch_memory_slot {
> };
>
> -struct kvm_vmid {
> - /*
> - * Writes to vmid_version and vmid happen with vmid_lock held
> - * whereas reads happen without any lock held.
> - */
> - unsigned long vmid_version;
> - unsigned long vmid;
> -};
> -
> struct kvm_arch {
> /* G-stage vmid */
> struct kvm_vmid vmid;
> @@ -311,77 +286,6 @@ static inline bool kvm_arch_pmi_in_guest(struct kvm_vcpu *vcpu)
> return IS_ENABLED(CONFIG_GUEST_PERF_EVENTS) && !!vcpu;
> }
>
> -#define KVM_RISCV_GSTAGE_TLB_MIN_ORDER 12
> -
> -void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
> - gpa_t gpa, gpa_t gpsz,
> - unsigned long order);
> -void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid);
> -void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
> - unsigned long order);
> -void kvm_riscv_local_hfence_gvma_all(void);
> -void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
> - unsigned long asid,
> - unsigned long gva,
> - unsigned long gvsz,
> - unsigned long order);
> -void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
> - unsigned long asid);
> -void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
> - unsigned long gva, unsigned long gvsz,
> - unsigned long order);
> -void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
> -
> -void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
> -
> -void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
> -void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
> -void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
> -
> -void kvm_riscv_fence_i(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask);
> -void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask,
> - gpa_t gpa, gpa_t gpsz,
> - unsigned long order);
> -void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask);
> -void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask,
> - unsigned long gva, unsigned long gvsz,
> - unsigned long order, unsigned long asid);
> -void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask,
> - unsigned long asid);
> -void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask,
> - unsigned long gva, unsigned long gvsz,
> - unsigned long order);
> -void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask);
> -
> -int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> - phys_addr_t hpa, unsigned long size,
> - bool writable, bool in_atomic);
> -void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
> - unsigned long size);
> -int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> - struct kvm_memory_slot *memslot,
> - gpa_t gpa, unsigned long hva, bool is_write);
> -int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
> -void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
> -void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
> -void __init kvm_riscv_gstage_mode_detect(void);
> -unsigned long __init kvm_riscv_gstage_mode(void);
> -int kvm_riscv_gstage_gpa_bits(void);
> -
> -void __init kvm_riscv_gstage_vmid_detect(void);
> -unsigned long kvm_riscv_gstage_vmid_bits(void);
> -int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
> -bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
> -void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
> -void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu);
> -
> int kvm_riscv_setup_default_irq_routing(struct kvm *kvm, u32 lines);
>
> void __kvm_riscv_unpriv_trap(void);
> diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
> new file mode 100644
> index 000000000000..4e1654282ee4
> --- /dev/null
> +++ b/arch/riscv/include/asm/kvm_mmu.h
> @@ -0,0 +1,26 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2025 Ventana Micro Systems Inc.
> + */
> +
> +#ifndef __RISCV_KVM_MMU_H_
> +#define __RISCV_KVM_MMU_H_
> +
> +#include <linux/kvm_types.h>
> +
> +int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> + phys_addr_t hpa, unsigned long size,
> + bool writable, bool in_atomic);
> +void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
> + unsigned long size);
> +int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> + struct kvm_memory_slot *memslot,
> + gpa_t gpa, unsigned long hva, bool is_write);
> +int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
> +void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
> +void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
> +void kvm_riscv_gstage_mode_detect(void);
> +unsigned long kvm_riscv_gstage_mode(void);
> +int kvm_riscv_gstage_gpa_bits(void);
> +
> +#endif
> diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
> new file mode 100644
> index 000000000000..cd00c9a46cb1
> --- /dev/null
> +++ b/arch/riscv/include/asm/kvm_tlb.h
> @@ -0,0 +1,78 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2025 Ventana Micro Systems Inc.
> + */
> +
> +#ifndef __RISCV_KVM_TLB_H_
> +#define __RISCV_KVM_TLB_H_
> +
> +#include <linux/kvm_types.h>
> +
> +enum kvm_riscv_hfence_type {
> + KVM_RISCV_HFENCE_UNKNOWN = 0,
> + KVM_RISCV_HFENCE_GVMA_VMID_GPA,
> + KVM_RISCV_HFENCE_VVMA_ASID_GVA,
> + KVM_RISCV_HFENCE_VVMA_ASID_ALL,
> + KVM_RISCV_HFENCE_VVMA_GVA,
> +};
> +
> +struct kvm_riscv_hfence {
> + enum kvm_riscv_hfence_type type;
> + unsigned long asid;
> + unsigned long order;
> + gpa_t addr;
> + gpa_t size;
> +};
> +
> +#define KVM_RISCV_VCPU_MAX_HFENCE 64
> +
> +#define KVM_RISCV_GSTAGE_TLB_MIN_ORDER 12
> +
> +void kvm_riscv_local_hfence_gvma_vmid_gpa(unsigned long vmid,
> + gpa_t gpa, gpa_t gpsz,
> + unsigned long order);
> +void kvm_riscv_local_hfence_gvma_vmid_all(unsigned long vmid);
> +void kvm_riscv_local_hfence_gvma_gpa(gpa_t gpa, gpa_t gpsz,
> + unsigned long order);
> +void kvm_riscv_local_hfence_gvma_all(void);
> +void kvm_riscv_local_hfence_vvma_asid_gva(unsigned long vmid,
> + unsigned long asid,
> + unsigned long gva,
> + unsigned long gvsz,
> + unsigned long order);
> +void kvm_riscv_local_hfence_vvma_asid_all(unsigned long vmid,
> + unsigned long asid);
> +void kvm_riscv_local_hfence_vvma_gva(unsigned long vmid,
> + unsigned long gva, unsigned long gvsz,
> + unsigned long order);
> +void kvm_riscv_local_hfence_vvma_all(unsigned long vmid);
> +
> +void kvm_riscv_tlb_flush_process(struct kvm_vcpu *vcpu);
> +
> +void kvm_riscv_fence_i_process(struct kvm_vcpu *vcpu);
> +void kvm_riscv_hfence_vvma_all_process(struct kvm_vcpu *vcpu);
> +void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu);
> +
> +void kvm_riscv_fence_i(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask);
> +void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask,
> + gpa_t gpa, gpa_t gpsz,
> + unsigned long order);
> +void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask);
> +void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask,
> + unsigned long gva, unsigned long gvsz,
> + unsigned long order, unsigned long asid);
> +void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask,
> + unsigned long asid);
> +void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask,
> + unsigned long gva, unsigned long gvsz,
> + unsigned long order);
> +void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
> + unsigned long hbase, unsigned long hmask);
> +
> +#endif
> diff --git a/arch/riscv/include/asm/kvm_vmid.h b/arch/riscv/include/asm/kvm_vmid.h
> new file mode 100644
> index 000000000000..ab98e1434fb7
> --- /dev/null
> +++ b/arch/riscv/include/asm/kvm_vmid.h
> @@ -0,0 +1,27 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (c) 2025 Ventana Micro Systems Inc.
> + */
> +
> +#ifndef __RISCV_KVM_VMID_H_
> +#define __RISCV_KVM_VMID_H_
> +
> +#include <linux/kvm_types.h>
> +
> +struct kvm_vmid {
> + /*
> + * Writes to vmid_version and vmid happen with vmid_lock held
> + * whereas reads happen without any lock held.
> + */
> + unsigned long vmid_version;
> + unsigned long vmid;
> +};
> +
> +void __init kvm_riscv_gstage_vmid_detect(void);
> +unsigned long kvm_riscv_gstage_vmid_bits(void);
> +int kvm_riscv_gstage_vmid_init(struct kvm *kvm);
> +bool kvm_riscv_gstage_vmid_ver_changed(struct kvm_vmid *vmid);
> +void kvm_riscv_gstage_vmid_update(struct kvm_vcpu *vcpu);
> +void kvm_riscv_gstage_vmid_sanitize(struct kvm_vcpu *vcpu);
> +
> +#endif
> diff --git a/arch/riscv/kvm/aia_imsic.c b/arch/riscv/kvm/aia_imsic.c
> index 29ef9c2133a9..40b469c0a01f 100644
> --- a/arch/riscv/kvm/aia_imsic.c
> +++ b/arch/riscv/kvm/aia_imsic.c
> @@ -16,6 +16,7 @@
> #include <linux/swab.h>
> #include <kvm/iodev.h>
> #include <asm/csr.h>
> +#include <asm/kvm_mmu.h>
>
> #define IMSIC_MAX_EIX (IMSIC_MAX_ID / BITS_PER_TYPE(u64))
>
> diff --git a/arch/riscv/kvm/main.c b/arch/riscv/kvm/main.c
> index 4b24705dc63a..b861a5dd7bd9 100644
> --- a/arch/riscv/kvm/main.c
> +++ b/arch/riscv/kvm/main.c
> @@ -11,6 +11,7 @@
> #include <linux/module.h>
> #include <linux/kvm_host.h>
> #include <asm/cpufeature.h>
> +#include <asm/kvm_mmu.h>
> #include <asm/kvm_nacl.h>
> #include <asm/sbi.h>
>
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index a5387927a1c1..c1a3eb076df3 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -15,6 +15,7 @@
> #include <linux/vmalloc.h>
> #include <linux/kvm_host.h>
> #include <linux/sched/signal.h>
> +#include <asm/kvm_mmu.h>
> #include <asm/kvm_nacl.h>
> #include <asm/page.h>
> #include <asm/pgtable.h>
> diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
> index f46a27658c2e..6fc4361c3d75 100644
> --- a/arch/riscv/kvm/tlb.c
> +++ b/arch/riscv/kvm/tlb.c
> @@ -15,6 +15,8 @@
> #include <asm/cpufeature.h>
> #include <asm/insn-def.h>
> #include <asm/kvm_nacl.h>
> +#include <asm/kvm_tlb.h>
> +#include <asm/kvm_vmid.h>
>
> #define has_svinval() riscv_has_extension_unlikely(RISCV_ISA_EXT_SVINVAL)
>
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index 6eb11c913b13..8ad7b31f5939 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -18,6 +18,7 @@
> #include <linux/fs.h>
> #include <linux/kvm_host.h>
> #include <asm/cacheflush.h>
> +#include <asm/kvm_mmu.h>
> #include <asm/kvm_nacl.h>
> #include <asm/kvm_vcpu_vector.h>
>
> diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
> index 85c43c83e3b9..965df528de90 100644
> --- a/arch/riscv/kvm/vcpu_exit.c
> +++ b/arch/riscv/kvm/vcpu_exit.c
> @@ -9,6 +9,7 @@
> #include <linux/kvm_host.h>
> #include <asm/csr.h>
> #include <asm/insn-def.h>
> +#include <asm/kvm_mmu.h>
> #include <asm/kvm_nacl.h>
>
> static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
> diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c
> index b27ec8f96697..8601cf29e5f8 100644
> --- a/arch/riscv/kvm/vm.c
> +++ b/arch/riscv/kvm/vm.c
> @@ -11,6 +11,7 @@
> #include <linux/module.h>
> #include <linux/uaccess.h>
> #include <linux/kvm_host.h>
> +#include <asm/kvm_mmu.h>
>
> const struct _kvm_stats_desc kvm_vm_stats_desc[] = {
> KVM_GENERIC_VM_STATS()
> diff --git a/arch/riscv/kvm/vmid.c b/arch/riscv/kvm/vmid.c
> index 92c01255f86f..3b426c800480 100644
> --- a/arch/riscv/kvm/vmid.c
> +++ b/arch/riscv/kvm/vmid.c
> @@ -14,6 +14,8 @@
> #include <linux/smp.h>
> #include <linux/kvm_host.h>
> #include <asm/csr.h>
> +#include <asm/kvm_tlb.h>
> +#include <asm/kvm_vmid.h>
>
> static unsigned long vmid_version = 1;
> static unsigned long vmid_next;
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping
2025-06-18 11:35 ` [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping Anup Patel
@ 2025-06-25 7:43 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:43 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> Introduce struct kvm_gstage_mapping which represents a g-stage
> mapping at a particular g-stage page table level. Also, update
> the kvm_riscv_gstage_map() to return the g-stage mapping upon
> success.
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_mmu.h | 9 ++++-
> arch/riscv/kvm/mmu.c | 58 ++++++++++++++++++--------------
> arch/riscv/kvm/vcpu_exit.c | 3 +-
> 3 files changed, 43 insertions(+), 27 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
> index 4e1654282ee4..91c11e692dc7 100644
> --- a/arch/riscv/include/asm/kvm_mmu.h
> +++ b/arch/riscv/include/asm/kvm_mmu.h
> @@ -8,6 +8,12 @@
>
> #include <linux/kvm_types.h>
>
> +struct kvm_gstage_mapping {
> + gpa_t addr;
> + pte_t pte;
> + u32 level;
> +};
> +
> int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> phys_addr_t hpa, unsigned long size,
> bool writable, bool in_atomic);
> @@ -15,7 +21,8 @@ void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
> unsigned long size);
> int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> struct kvm_memory_slot *memslot,
> - gpa_t gpa, unsigned long hva, bool is_write);
> + gpa_t gpa, unsigned long hva, bool is_write,
> + struct kvm_gstage_mapping *out_map);
> int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
> void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
> void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index c1a3eb076df3..806614b3e46d 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -135,18 +135,18 @@ static void gstage_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
> kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0, addr, BIT(order), order);
> }
>
> -static int gstage_set_pte(struct kvm *kvm, u32 level,
> - struct kvm_mmu_memory_cache *pcache,
> - gpa_t addr, const pte_t *new_pte)
> +static int gstage_set_pte(struct kvm *kvm,
> + struct kvm_mmu_memory_cache *pcache,
> + const struct kvm_gstage_mapping *map)
> {
> u32 current_level = gstage_pgd_levels - 1;
> pte_t *next_ptep = (pte_t *)kvm->arch.pgd;
> - pte_t *ptep = &next_ptep[gstage_pte_index(addr, current_level)];
> + pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
>
> - if (current_level < level)
> + if (current_level < map->level)
> return -EINVAL;
>
> - while (current_level != level) {
> + while (current_level != map->level) {
> if (gstage_pte_leaf(ptep))
> return -EEXIST;
>
> @@ -165,13 +165,13 @@ static int gstage_set_pte(struct kvm *kvm, u32 level,
> }
>
> current_level--;
> - ptep = &next_ptep[gstage_pte_index(addr, current_level)];
> + ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
> }
>
> - if (pte_val(*ptep) != pte_val(*new_pte)) {
> - set_pte(ptep, *new_pte);
> + if (pte_val(*ptep) != pte_val(map->pte)) {
> + set_pte(ptep, map->pte);
> if (gstage_pte_leaf(ptep))
> - gstage_remote_tlb_flush(kvm, current_level, addr);
> + gstage_remote_tlb_flush(kvm, current_level, map->addr);
> }
>
> return 0;
> @@ -181,14 +181,16 @@ static int gstage_map_page(struct kvm *kvm,
> struct kvm_mmu_memory_cache *pcache,
> gpa_t gpa, phys_addr_t hpa,
> unsigned long page_size,
> - bool page_rdonly, bool page_exec)
> + bool page_rdonly, bool page_exec,
> + struct kvm_gstage_mapping *out_map)
> {
> - int ret;
> - u32 level = 0;
> - pte_t new_pte;
> pgprot_t prot;
> + int ret;
>
> - ret = gstage_page_size_to_level(page_size, &level);
> + out_map->addr = gpa;
> + out_map->level = 0;
> +
> + ret = gstage_page_size_to_level(page_size, &out_map->level);
> if (ret)
> return ret;
>
> @@ -216,10 +218,10 @@ static int gstage_map_page(struct kvm *kvm,
> else
> prot = PAGE_WRITE;
> }
> - new_pte = pfn_pte(PFN_DOWN(hpa), prot);
> - new_pte = pte_mkdirty(new_pte);
> + out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
> + out_map->pte = pte_mkdirty(out_map->pte);
>
> - return gstage_set_pte(kvm, level, pcache, gpa, &new_pte);
> + return gstage_set_pte(kvm, pcache, out_map);
> }
>
> enum gstage_op {
> @@ -352,7 +354,6 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> phys_addr_t hpa, unsigned long size,
> bool writable, bool in_atomic)
> {
> - pte_t pte;
> int ret = 0;
> unsigned long pfn;
> phys_addr_t addr, end;
> @@ -360,22 +361,25 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> .gfp_custom = (in_atomic) ? GFP_ATOMIC | __GFP_ACCOUNT : 0,
> .gfp_zero = __GFP_ZERO,
> };
> + struct kvm_gstage_mapping map;
>
> end = (gpa + size + PAGE_SIZE - 1) & PAGE_MASK;
> pfn = __phys_to_pfn(hpa);
>
> for (addr = gpa; addr < end; addr += PAGE_SIZE) {
> - pte = pfn_pte(pfn, PAGE_KERNEL_IO);
> + map.addr = addr;
> + map.pte = pfn_pte(pfn, PAGE_KERNEL_IO);
> + map.level = 0;
>
> if (!writable)
> - pte = pte_wrprotect(pte);
> + map.pte = pte_wrprotect(map.pte);
>
> ret = kvm_mmu_topup_memory_cache(&pcache, gstage_pgd_levels);
> if (ret)
> goto out;
>
> spin_lock(&kvm->mmu_lock);
> - ret = gstage_set_pte(kvm, 0, &pcache, addr, &pte);
> + ret = gstage_set_pte(kvm, &pcache, &map);
> spin_unlock(&kvm->mmu_lock);
> if (ret)
> goto out;
> @@ -593,7 +597,8 @@ bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>
> int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> struct kvm_memory_slot *memslot,
> - gpa_t gpa, unsigned long hva, bool is_write)
> + gpa_t gpa, unsigned long hva, bool is_write,
> + struct kvm_gstage_mapping *out_map)
> {
> int ret;
> kvm_pfn_t hfn;
> @@ -608,6 +613,9 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> unsigned long vma_pagesize, mmu_seq;
> struct page *page;
>
> + /* Setup initial state of output mapping */
> + memset(out_map, 0, sizeof(*out_map));
> +
> /* We need minimum second+third level pages */
> ret = kvm_mmu_topup_memory_cache(pcache, gstage_pgd_levels);
> if (ret) {
> @@ -677,10 +685,10 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> if (writable) {
> mark_page_dirty(kvm, gfn);
> ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
> - vma_pagesize, false, true);
> + vma_pagesize, false, true, out_map);
> } else {
> ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
> - vma_pagesize, true, true);
> + vma_pagesize, true, true, out_map);
> }
>
> if (ret)
> diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
> index 965df528de90..6b4694bc07ea 100644
> --- a/arch/riscv/kvm/vcpu_exit.c
> +++ b/arch/riscv/kvm/vcpu_exit.c
> @@ -15,6 +15,7 @@
> static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
> struct kvm_cpu_trap *trap)
> {
> + struct kvm_gstage_mapping host_map;
> struct kvm_memory_slot *memslot;
> unsigned long hva, fault_addr;
> bool writable;
> @@ -43,7 +44,7 @@ static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
> }
>
> ret = kvm_riscv_gstage_map(vcpu, memslot, fault_addr, hva,
> - (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false);
> + (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false, &host_map);
> if (ret < 0)
> return ret;
>
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence
2025-06-18 11:35 ` [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence Anup Patel
@ 2025-06-25 7:55 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:55 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel,
Atish Patra
On 6/18/2025 7:35 PM, Anup Patel wrote:
> Currently, the struct kvm_riscv_hfence does not have vmid field
> and various hfence processing functions always pick vmid assigned
> to the guest/VM. This prevents us from doing hfence operation on
> arbitrary vmid hence add vmid field to struct kvm_riscv_hfence
> and use it wherever applicable.
>
> Reviewed-by: Atish Patra <atishp@rivosinc.com>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_tlb.h | 1 +
> arch/riscv/kvm/tlb.c | 30 ++++++++++++++++--------------
> 2 files changed, 17 insertions(+), 14 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
> index cd00c9a46cb1..f67e03edeaec 100644
> --- a/arch/riscv/include/asm/kvm_tlb.h
> +++ b/arch/riscv/include/asm/kvm_tlb.h
> @@ -19,6 +19,7 @@ enum kvm_riscv_hfence_type {
> struct kvm_riscv_hfence {
> enum kvm_riscv_hfence_type type;
> unsigned long asid;
> + unsigned long vmid;
> unsigned long order;
> gpa_t addr;
> gpa_t size;
> diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
> index 6fc4361c3d75..349fcfc93f54 100644
> --- a/arch/riscv/kvm/tlb.c
> +++ b/arch/riscv/kvm/tlb.c
> @@ -237,49 +237,43 @@ static bool vcpu_hfence_enqueue(struct kvm_vcpu *vcpu,
>
> void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
> {
> - unsigned long vmid;
> struct kvm_riscv_hfence d = { 0 };
> - struct kvm_vmid *v = &vcpu->kvm->arch.vmid;
>
> while (vcpu_hfence_dequeue(vcpu, &d)) {
> switch (d.type) {
> case KVM_RISCV_HFENCE_UNKNOWN:
> break;
> case KVM_RISCV_HFENCE_GVMA_VMID_GPA:
> - vmid = READ_ONCE(v->vmid);
> if (kvm_riscv_nacl_available())
> - nacl_hfence_gvma_vmid(nacl_shmem(), vmid,
> + nacl_hfence_gvma_vmid(nacl_shmem(), d.vmid,
> d.addr, d.size, d.order);
> else
> - kvm_riscv_local_hfence_gvma_vmid_gpa(vmid, d.addr,
> + kvm_riscv_local_hfence_gvma_vmid_gpa(d.vmid, d.addr,
> d.size, d.order);
> break;
> case KVM_RISCV_HFENCE_VVMA_ASID_GVA:
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
> - vmid = READ_ONCE(v->vmid);
> if (kvm_riscv_nacl_available())
> - nacl_hfence_vvma_asid(nacl_shmem(), vmid, d.asid,
> + nacl_hfence_vvma_asid(nacl_shmem(), d.vmid, d.asid,
> d.addr, d.size, d.order);
> else
> - kvm_riscv_local_hfence_vvma_asid_gva(vmid, d.asid, d.addr,
> + kvm_riscv_local_hfence_vvma_asid_gva(d.vmid, d.asid, d.addr,
> d.size, d.order);
> break;
> case KVM_RISCV_HFENCE_VVMA_ASID_ALL:
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
> - vmid = READ_ONCE(v->vmid);
> if (kvm_riscv_nacl_available())
> - nacl_hfence_vvma_asid_all(nacl_shmem(), vmid, d.asid);
> + nacl_hfence_vvma_asid_all(nacl_shmem(), d.vmid, d.asid);
> else
> - kvm_riscv_local_hfence_vvma_asid_all(vmid, d.asid);
> + kvm_riscv_local_hfence_vvma_asid_all(d.vmid, d.asid);
> break;
> case KVM_RISCV_HFENCE_VVMA_GVA:
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_RCVD);
> - vmid = READ_ONCE(v->vmid);
> if (kvm_riscv_nacl_available())
> - nacl_hfence_vvma(nacl_shmem(), vmid,
> + nacl_hfence_vvma(nacl_shmem(), d.vmid,
> d.addr, d.size, d.order);
> else
> - kvm_riscv_local_hfence_vvma_gva(vmid, d.addr,
> + kvm_riscv_local_hfence_vvma_gva(d.vmid, d.addr,
> d.size, d.order);
> break;
> default:
> @@ -336,10 +330,12 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> gpa_t gpa, gpa_t gpsz,
> unsigned long order)
> {
> + struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_GVMA_VMID_GPA;
> data.asid = 0;
> + data.vmid = READ_ONCE(v->vmid);
> data.addr = gpa;
> data.size = gpsz;
> data.order = order;
> @@ -359,10 +355,12 @@ void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> unsigned long gva, unsigned long gvsz,
> unsigned long order, unsigned long asid)
> {
> + struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_VVMA_ASID_GVA;
> data.asid = asid;
> + data.vmid = READ_ONCE(v->vmid);
> data.addr = gva;
> data.size = gvsz;
> data.order = order;
> @@ -374,10 +372,12 @@ void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> unsigned long asid)
> {
> + struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_VVMA_ASID_ALL;
> data.asid = asid;
> + data.vmid = READ_ONCE(v->vmid);
> data.addr = data.size = data.order = 0;
> make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
> KVM_REQ_HFENCE_VVMA_ALL, &data);
> @@ -388,10 +388,12 @@ void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> unsigned long gva, unsigned long gvsz,
> unsigned long order)
> {
> + struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_VVMA_GVA;
> data.asid = 0;
> + data.vmid = READ_ONCE(v->vmid);
> data.addr = gva;
> data.size = gvsz;
> data.order = order;
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management
2025-06-18 11:35 ` [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management Anup Patel
@ 2025-06-25 7:57 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:57 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/2025 7:35 PM, Anup Patel wrote:
> The upcoming nested virtualization can share g-stage page table
> management with the current host g-stage implementation hence
> factor-out g-stage page table management as separate sources
> and also use "kvm_riscv_mmu_" prefix for host g-stage functions.
>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_gstage.h | 72 ++++
> arch/riscv/include/asm/kvm_mmu.h | 32 +-
> arch/riscv/kvm/Makefile | 1 +
> arch/riscv/kvm/aia_imsic.c | 11 +-
> arch/riscv/kvm/gstage.c | 337 +++++++++++++++++++
> arch/riscv/kvm/main.c | 2 +-
> arch/riscv/kvm/mmu.c | 492 ++++++----------------------
> arch/riscv/kvm/vcpu.c | 4 +-
> arch/riscv/kvm/vcpu_exit.c | 5 +-
> arch/riscv/kvm/vm.c | 6 +-
> 10 files changed, 530 insertions(+), 432 deletions(-)
> create mode 100644 arch/riscv/include/asm/kvm_gstage.h
> create mode 100644 arch/riscv/kvm/gstage.c
>
> diff --git a/arch/riscv/include/asm/kvm_gstage.h b/arch/riscv/include/asm/kvm_gstage.h
> new file mode 100644
> index 000000000000..595e2183173e
> --- /dev/null
> +++ b/arch/riscv/include/asm/kvm_gstage.h
> @@ -0,0 +1,72 @@
> +/* SPDX-License-Identifier: GPL-2.0-only */
> +/*
> + * Copyright (C) 2019 Western Digital Corporation or its affiliates.
> + * Copyright (c) 2025 Ventana Micro Systems Inc.
> + */
> +
> +#ifndef __RISCV_KVM_GSTAGE_H_
> +#define __RISCV_KVM_GSTAGE_H_
> +
> +#include <linux/kvm_types.h>
> +
> +struct kvm_gstage {
> + struct kvm *kvm;
> + unsigned long flags;
> +#define KVM_GSTAGE_FLAGS_LOCAL BIT(0)
> + unsigned long vmid;
> + pgd_t *pgd;
> +};
> +
> +struct kvm_gstage_mapping {
> + gpa_t addr;
> + pte_t pte;
> + u32 level;
> +};
> +
> +#ifdef CONFIG_64BIT
> +#define kvm_riscv_gstage_index_bits 9
> +#else
> +#define kvm_riscv_gstage_index_bits 10
> +#endif
> +
> +extern unsigned long kvm_riscv_gstage_mode;
> +extern unsigned long kvm_riscv_gstage_pgd_levels;
> +
> +#define kvm_riscv_gstage_pgd_xbits 2
> +#define kvm_riscv_gstage_pgd_size (1UL << (HGATP_PAGE_SHIFT + kvm_riscv_gstage_pgd_xbits))
> +#define kvm_riscv_gstage_gpa_bits (HGATP_PAGE_SHIFT + \
> + (kvm_riscv_gstage_pgd_levels * \
> + kvm_riscv_gstage_index_bits) + \
> + kvm_riscv_gstage_pgd_xbits)
> +#define kvm_riscv_gstage_gpa_size ((gpa_t)(1ULL << kvm_riscv_gstage_gpa_bits))
> +
> +bool kvm_riscv_gstage_get_leaf(struct kvm_gstage *gstage, gpa_t addr,
> + pte_t **ptepp, u32 *ptep_level);
> +
> +int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
> + struct kvm_mmu_memory_cache *pcache,
> + const struct kvm_gstage_mapping *map);
> +
> +int kvm_riscv_gstage_map_page(struct kvm_gstage *gstage,
> + struct kvm_mmu_memory_cache *pcache,
> + gpa_t gpa, phys_addr_t hpa, unsigned long page_size,
> + bool page_rdonly, bool page_exec,
> + struct kvm_gstage_mapping *out_map);
> +
> +enum kvm_riscv_gstage_op {
> + GSTAGE_OP_NOP = 0, /* Nothing */
> + GSTAGE_OP_CLEAR, /* Clear/Unmap */
> + GSTAGE_OP_WP, /* Write-protect */
> +};
> +
> +void kvm_riscv_gstage_op_pte(struct kvm_gstage *gstage, gpa_t addr,
> + pte_t *ptep, u32 ptep_level, enum kvm_riscv_gstage_op op);
> +
> +void kvm_riscv_gstage_unmap_range(struct kvm_gstage *gstage,
> + gpa_t start, gpa_t size, bool may_block);
> +
> +void kvm_riscv_gstage_wp_range(struct kvm_gstage *gstage, gpa_t start, gpa_t end);
> +
> +void kvm_riscv_gstage_mode_detect(void);
> +
> +#endif
> diff --git a/arch/riscv/include/asm/kvm_mmu.h b/arch/riscv/include/asm/kvm_mmu.h
> index 91c11e692dc7..5439e76f0a96 100644
> --- a/arch/riscv/include/asm/kvm_mmu.h
> +++ b/arch/riscv/include/asm/kvm_mmu.h
> @@ -6,28 +6,16 @@
> #ifndef __RISCV_KVM_MMU_H_
> #define __RISCV_KVM_MMU_H_
>
> -#include <linux/kvm_types.h>
> +#include <asm/kvm_gstage.h>
>
> -struct kvm_gstage_mapping {
> - gpa_t addr;
> - pte_t pte;
> - u32 level;
> -};
> -
> -int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> - phys_addr_t hpa, unsigned long size,
> - bool writable, bool in_atomic);
> -void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa,
> - unsigned long size);
> -int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> - struct kvm_memory_slot *memslot,
> - gpa_t gpa, unsigned long hva, bool is_write,
> - struct kvm_gstage_mapping *out_map);
> -int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm);
> -void kvm_riscv_gstage_free_pgd(struct kvm *kvm);
> -void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu);
> -void kvm_riscv_gstage_mode_detect(void);
> -unsigned long kvm_riscv_gstage_mode(void);
> -int kvm_riscv_gstage_gpa_bits(void);
> +int kvm_riscv_mmu_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
> + unsigned long size, bool writable, bool in_atomic);
> +void kvm_riscv_mmu_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size);
> +int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
> + gpa_t gpa, unsigned long hva, bool is_write,
> + struct kvm_gstage_mapping *out_map);
> +int kvm_riscv_mmu_alloc_pgd(struct kvm *kvm);
> +void kvm_riscv_mmu_free_pgd(struct kvm *kvm);
> +void kvm_riscv_mmu_update_hgatp(struct kvm_vcpu *vcpu);
>
> #endif
> diff --git a/arch/riscv/kvm/Makefile b/arch/riscv/kvm/Makefile
> index 06e2d52a9b88..07197395750e 100644
> --- a/arch/riscv/kvm/Makefile
> +++ b/arch/riscv/kvm/Makefile
> @@ -14,6 +14,7 @@ kvm-y += aia.o
> kvm-y += aia_aplic.o
> kvm-y += aia_device.o
> kvm-y += aia_imsic.o
> +kvm-y += gstage.o
> kvm-y += main.o
> kvm-y += mmu.o
> kvm-y += nacl.o
> diff --git a/arch/riscv/kvm/aia_imsic.c b/arch/riscv/kvm/aia_imsic.c
> index 40b469c0a01f..ea1a36836d9c 100644
> --- a/arch/riscv/kvm/aia_imsic.c
> +++ b/arch/riscv/kvm/aia_imsic.c
> @@ -704,9 +704,8 @@ void kvm_riscv_vcpu_aia_imsic_release(struct kvm_vcpu *vcpu)
> */
>
> /* Purge the G-stage mapping */
> - kvm_riscv_gstage_iounmap(vcpu->kvm,
> - vcpu->arch.aia_context.imsic_addr,
> - IMSIC_MMIO_PAGE_SZ);
> + kvm_riscv_mmu_iounmap(vcpu->kvm, vcpu->arch.aia_context.imsic_addr,
> + IMSIC_MMIO_PAGE_SZ);
>
> /* TODO: Purge the IOMMU mapping ??? */
>
> @@ -786,9 +785,9 @@ int kvm_riscv_vcpu_aia_imsic_update(struct kvm_vcpu *vcpu)
> imsic_vsfile_local_clear(new_vsfile_hgei, imsic->nr_hw_eix);
>
> /* Update G-stage mapping for the new IMSIC VS-file */
> - ret = kvm_riscv_gstage_ioremap(kvm, vcpu->arch.aia_context.imsic_addr,
> - new_vsfile_pa, IMSIC_MMIO_PAGE_SZ,
> - true, true);
> + ret = kvm_riscv_mmu_ioremap(kvm, vcpu->arch.aia_context.imsic_addr,
> + new_vsfile_pa, IMSIC_MMIO_PAGE_SZ,
> + true, true);
> if (ret)
> goto fail_free_vsfile_hgei;
>
> diff --git a/arch/riscv/kvm/gstage.c b/arch/riscv/kvm/gstage.c
> new file mode 100644
> index 000000000000..9c7c44f09b05
> --- /dev/null
> +++ b/arch/riscv/kvm/gstage.c
> @@ -0,0 +1,337 @@
> +// SPDX-License-Identifier: GPL-2.0
> +/*
> + * Copyright (C) 2019 Western Digital Corporation or its affiliates.
> + * Copyright (c) 2025 Ventana Micro Systems Inc.
> + */
> +
> +#include <linux/bitops.h>
> +#include <linux/errno.h>
> +#include <linux/kvm_host.h>
> +#include <linux/module.h>
> +#include <linux/pgtable.h>
> +#include <asm/kvm_gstage.h>
> +
> +#ifdef CONFIG_64BIT
> +unsigned long kvm_riscv_gstage_mode __ro_after_init = HGATP_MODE_SV39X4;
> +unsigned long kvm_riscv_gstage_pgd_levels __ro_after_init = 3;
> +#else
> +unsigned long kvm_riscv_gstage_mode __ro_after_init = HGATP_MODE_SV32X4;
> +unsigned long kvm_riscv_gstage_pgd_levels __ro_after_init = 2;
> +#endif
> +
> +#define gstage_pte_leaf(__ptep) \
> + (pte_val(*(__ptep)) & (_PAGE_READ | _PAGE_WRITE | _PAGE_EXEC))
> +
> +static inline unsigned long gstage_pte_index(gpa_t addr, u32 level)
> +{
> + unsigned long mask;
> + unsigned long shift = HGATP_PAGE_SHIFT + (kvm_riscv_gstage_index_bits * level);
> +
> + if (level == (kvm_riscv_gstage_pgd_levels - 1))
> + mask = (PTRS_PER_PTE * (1UL << kvm_riscv_gstage_pgd_xbits)) - 1;
> + else
> + mask = PTRS_PER_PTE - 1;
> +
> + return (addr >> shift) & mask;
> +}
> +
> +static inline unsigned long gstage_pte_page_vaddr(pte_t pte)
> +{
> + return (unsigned long)pfn_to_virt(__page_val_to_pfn(pte_val(pte)));
> +}
> +
> +static int gstage_page_size_to_level(unsigned long page_size, u32 *out_level)
> +{
> + u32 i;
> + unsigned long psz = 1UL << 12;
> +
> + for (i = 0; i < kvm_riscv_gstage_pgd_levels; i++) {
> + if (page_size == (psz << (i * kvm_riscv_gstage_index_bits))) {
> + *out_level = i;
> + return 0;
> + }
> + }
> +
> + return -EINVAL;
> +}
> +
> +static int gstage_level_to_page_order(u32 level, unsigned long *out_pgorder)
> +{
> + if (kvm_riscv_gstage_pgd_levels < level)
> + return -EINVAL;
> +
> + *out_pgorder = 12 + (level * kvm_riscv_gstage_index_bits);
> + return 0;
> +}
> +
> +static int gstage_level_to_page_size(u32 level, unsigned long *out_pgsize)
> +{
> + int rc;
> + unsigned long page_order = PAGE_SHIFT;
> +
> + rc = gstage_level_to_page_order(level, &page_order);
> + if (rc)
> + return rc;
> +
> + *out_pgsize = BIT(page_order);
> + return 0;
> +}
> +
> +bool kvm_riscv_gstage_get_leaf(struct kvm_gstage *gstage, gpa_t addr,
> + pte_t **ptepp, u32 *ptep_level)
> +{
> + pte_t *ptep;
> + u32 current_level = kvm_riscv_gstage_pgd_levels - 1;
> +
> + *ptep_level = current_level;
> + ptep = (pte_t *)gstage->pgd;
> + ptep = &ptep[gstage_pte_index(addr, current_level)];
> + while (ptep && pte_val(ptep_get(ptep))) {
> + if (gstage_pte_leaf(ptep)) {
> + *ptep_level = current_level;
> + *ptepp = ptep;
> + return true;
> + }
> +
> + if (current_level) {
> + current_level--;
> + *ptep_level = current_level;
> + ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> + ptep = &ptep[gstage_pte_index(addr, current_level)];
> + } else {
> + ptep = NULL;
> + }
> + }
> +
> + return false;
> +}
> +
> +static void gstage_tlb_flush(struct kvm_gstage *gstage, u32 level, gpa_t addr)
> +{
> + unsigned long order = PAGE_SHIFT;
> +
> + if (gstage_level_to_page_order(level, &order))
> + return;
> + addr &= ~(BIT(order) - 1);
> +
> + if (gstage->flags & KVM_GSTAGE_FLAGS_LOCAL)
> + kvm_riscv_local_hfence_gvma_vmid_gpa(gstage->vmid, addr, BIT(order), order);
> + else
> + kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order);
> +}
> +
> +int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
> + struct kvm_mmu_memory_cache *pcache,
> + const struct kvm_gstage_mapping *map)
> +{
> + u32 current_level = kvm_riscv_gstage_pgd_levels - 1;
> + pte_t *next_ptep = (pte_t *)gstage->pgd;
> + pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
> +
> + if (current_level < map->level)
> + return -EINVAL;
> +
> + while (current_level != map->level) {
> + if (gstage_pte_leaf(ptep))
> + return -EEXIST;
> +
> + if (!pte_val(ptep_get(ptep))) {
> + if (!pcache)
> + return -ENOMEM;
> + next_ptep = kvm_mmu_memory_cache_alloc(pcache);
> + if (!next_ptep)
> + return -ENOMEM;
> + set_pte(ptep, pfn_pte(PFN_DOWN(__pa(next_ptep)),
> + __pgprot(_PAGE_TABLE)));
> + } else {
> + if (gstage_pte_leaf(ptep))
> + return -EEXIST;
> + next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> + }
> +
> + current_level--;
> + ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
> + }
> +
> + if (pte_val(*ptep) != pte_val(map->pte)) {
> + set_pte(ptep, map->pte);
> + if (gstage_pte_leaf(ptep))
> + gstage_tlb_flush(gstage, current_level, map->addr);
> + }
> +
> + return 0;
> +}
> +
> +int kvm_riscv_gstage_map_page(struct kvm_gstage *gstage,
> + struct kvm_mmu_memory_cache *pcache,
> + gpa_t gpa, phys_addr_t hpa, unsigned long page_size,
> + bool page_rdonly, bool page_exec,
> + struct kvm_gstage_mapping *out_map)
> +{
> + pgprot_t prot;
> + int ret;
> +
> + out_map->addr = gpa;
> + out_map->level = 0;
> +
> + ret = gstage_page_size_to_level(page_size, &out_map->level);
> + if (ret)
> + return ret;
> +
> + /*
> + * A RISC-V implementation can choose to either:
> + * 1) Update 'A' and 'D' PTE bits in hardware
> + * 2) Generate page fault when 'A' and/or 'D' bits are not set
> + * PTE so that software can update these bits.
> + *
> + * We support both options mentioned above. To achieve this, we
> + * always set 'A' and 'D' PTE bits at time of creating G-stage
> + * mapping. To support KVM dirty page logging with both options
> + * mentioned above, we will write-protect G-stage PTEs to track
> + * dirty pages.
> + */
> +
> + if (page_exec) {
> + if (page_rdonly)
> + prot = PAGE_READ_EXEC;
> + else
> + prot = PAGE_WRITE_EXEC;
> + } else {
> + if (page_rdonly)
> + prot = PAGE_READ;
> + else
> + prot = PAGE_WRITE;
> + }
> + out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
> + out_map->pte = pte_mkdirty(out_map->pte);
> +
> + return kvm_riscv_gstage_set_pte(gstage, pcache, out_map);
> +}
> +
> +void kvm_riscv_gstage_op_pte(struct kvm_gstage *gstage, gpa_t addr,
> + pte_t *ptep, u32 ptep_level, enum kvm_riscv_gstage_op op)
> +{
> + int i, ret;
> + pte_t old_pte, *next_ptep;
> + u32 next_ptep_level;
> + unsigned long next_page_size, page_size;
> +
> + ret = gstage_level_to_page_size(ptep_level, &page_size);
> + if (ret)
> + return;
> +
> + WARN_ON(addr & (page_size - 1));
> +
> + if (!pte_val(ptep_get(ptep)))
> + return;
> +
> + if (ptep_level && !gstage_pte_leaf(ptep)) {
> + next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> + next_ptep_level = ptep_level - 1;
> + ret = gstage_level_to_page_size(next_ptep_level, &next_page_size);
> + if (ret)
> + return;
> +
> + if (op == GSTAGE_OP_CLEAR)
> + set_pte(ptep, __pte(0));
> + for (i = 0; i < PTRS_PER_PTE; i++)
> + kvm_riscv_gstage_op_pte(gstage, addr + i * next_page_size,
> + &next_ptep[i], next_ptep_level, op);
> + if (op == GSTAGE_OP_CLEAR)
> + put_page(virt_to_page(next_ptep));
> + } else {
> + old_pte = *ptep;
> + if (op == GSTAGE_OP_CLEAR)
> + set_pte(ptep, __pte(0));
> + else if (op == GSTAGE_OP_WP)
> + set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
> + if (pte_val(*ptep) != pte_val(old_pte))
> + gstage_tlb_flush(gstage, ptep_level, addr);
> + }
> +}
> +
> +void kvm_riscv_gstage_unmap_range(struct kvm_gstage *gstage,
> + gpa_t start, gpa_t size, bool may_block)
> +{
> + int ret;
> + pte_t *ptep;
> + u32 ptep_level;
> + bool found_leaf;
> + unsigned long page_size;
> + gpa_t addr = start, end = start + size;
> +
> + while (addr < end) {
> + found_leaf = kvm_riscv_gstage_get_leaf(gstage, addr, &ptep, &ptep_level);
> + ret = gstage_level_to_page_size(ptep_level, &page_size);
> + if (ret)
> + break;
> +
> + if (!found_leaf)
> + goto next;
> +
> + if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> + kvm_riscv_gstage_op_pte(gstage, addr, ptep,
> + ptep_level, GSTAGE_OP_CLEAR);
> +
> +next:
> + addr += page_size;
> +
> + /*
> + * If the range is too large, release the kvm->mmu_lock
> + * to prevent starvation and lockup detector warnings.
> + */
> + if (!(gstage->flags & KVM_GSTAGE_FLAGS_LOCAL) && may_block && addr < end)
> + cond_resched_lock(&gstage->kvm->mmu_lock);
> + }
> +}
> +
> +void kvm_riscv_gstage_wp_range(struct kvm_gstage *gstage, gpa_t start, gpa_t end)
> +{
> + int ret;
> + pte_t *ptep;
> + u32 ptep_level;
> + bool found_leaf;
> + gpa_t addr = start;
> + unsigned long page_size;
> +
> + while (addr < end) {
> + found_leaf = kvm_riscv_gstage_get_leaf(gstage, addr, &ptep, &ptep_level);
> + ret = gstage_level_to_page_size(ptep_level, &page_size);
> + if (ret)
> + break;
> +
> + if (!found_leaf)
> + goto next;
> +
> + if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> + kvm_riscv_gstage_op_pte(gstage, addr, ptep,
> + ptep_level, GSTAGE_OP_WP);
> +
> +next:
> + addr += page_size;
> + }
> +}
> +
> +void __init kvm_riscv_gstage_mode_detect(void)
> +{
> +#ifdef CONFIG_64BIT
> + /* Try Sv57x4 G-stage mode */
> + csr_write(CSR_HGATP, HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
> + if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV57X4) {
> + kvm_riscv_gstage_mode = HGATP_MODE_SV57X4;
> + kvm_riscv_gstage_pgd_levels = 5;
> + goto skip_sv48x4_test;
> + }
> +
> + /* Try Sv48x4 G-stage mode */
> + csr_write(CSR_HGATP, HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
> + if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV48X4) {
> + kvm_riscv_gstage_mode = HGATP_MODE_SV48X4;
> + kvm_riscv_gstage_pgd_levels = 4;
> + }
> +skip_sv48x4_test:
> +
> + csr_write(CSR_HGATP, 0);
> + kvm_riscv_local_hfence_gvma_all();
> +#endif
> +}
> diff --git a/arch/riscv/kvm/main.c b/arch/riscv/kvm/main.c
> index b861a5dd7bd9..67c876de74ef 100644
> --- a/arch/riscv/kvm/main.c
> +++ b/arch/riscv/kvm/main.c
> @@ -135,7 +135,7 @@ static int __init riscv_kvm_init(void)
> (rc) ? slist : "no features");
> }
>
> - switch (kvm_riscv_gstage_mode()) {
> + switch (kvm_riscv_gstage_mode) {
> case HGATP_MODE_SV32X4:
> str = "Sv32x4";
> break;
> diff --git a/arch/riscv/kvm/mmu.c b/arch/riscv/kvm/mmu.c
> index 806614b3e46d..9f7dcd8cd741 100644
> --- a/arch/riscv/kvm/mmu.c
> +++ b/arch/riscv/kvm/mmu.c
> @@ -6,9 +6,7 @@
> * Anup Patel <anup.patel@wdc.com>
> */
>
> -#include <linux/bitops.h>
> #include <linux/errno.h>
> -#include <linux/err.h>
> #include <linux/hugetlb.h>
> #include <linux/module.h>
> #include <linux/uaccess.h>
> @@ -17,342 +15,28 @@
> #include <linux/sched/signal.h>
> #include <asm/kvm_mmu.h>
> #include <asm/kvm_nacl.h>
> -#include <asm/page.h>
> -#include <asm/pgtable.h>
> -
> -#ifdef CONFIG_64BIT
> -static unsigned long gstage_mode __ro_after_init = (HGATP_MODE_SV39X4 << HGATP_MODE_SHIFT);
> -static unsigned long gstage_pgd_levels __ro_after_init = 3;
> -#define gstage_index_bits 9
> -#else
> -static unsigned long gstage_mode __ro_after_init = (HGATP_MODE_SV32X4 << HGATP_MODE_SHIFT);
> -static unsigned long gstage_pgd_levels __ro_after_init = 2;
> -#define gstage_index_bits 10
> -#endif
> -
> -#define gstage_pgd_xbits 2
> -#define gstage_pgd_size (1UL << (HGATP_PAGE_SHIFT + gstage_pgd_xbits))
> -#define gstage_gpa_bits (HGATP_PAGE_SHIFT + \
> - (gstage_pgd_levels * gstage_index_bits) + \
> - gstage_pgd_xbits)
> -#define gstage_gpa_size ((gpa_t)(1ULL << gstage_gpa_bits))
> -
> -#define gstage_pte_leaf(__ptep) \
> - (pte_val(*(__ptep)) & (_PAGE_READ | _PAGE_WRITE | _PAGE_EXEC))
> -
> -static inline unsigned long gstage_pte_index(gpa_t addr, u32 level)
> -{
> - unsigned long mask;
> - unsigned long shift = HGATP_PAGE_SHIFT + (gstage_index_bits * level);
> -
> - if (level == (gstage_pgd_levels - 1))
> - mask = (PTRS_PER_PTE * (1UL << gstage_pgd_xbits)) - 1;
> - else
> - mask = PTRS_PER_PTE - 1;
> -
> - return (addr >> shift) & mask;
> -}
>
> -static inline unsigned long gstage_pte_page_vaddr(pte_t pte)
> -{
> - return (unsigned long)pfn_to_virt(__page_val_to_pfn(pte_val(pte)));
> -}
> -
> -static int gstage_page_size_to_level(unsigned long page_size, u32 *out_level)
> -{
> - u32 i;
> - unsigned long psz = 1UL << 12;
> -
> - for (i = 0; i < gstage_pgd_levels; i++) {
> - if (page_size == (psz << (i * gstage_index_bits))) {
> - *out_level = i;
> - return 0;
> - }
> - }
> -
> - return -EINVAL;
> -}
> -
> -static int gstage_level_to_page_order(u32 level, unsigned long *out_pgorder)
> -{
> - if (gstage_pgd_levels < level)
> - return -EINVAL;
> -
> - *out_pgorder = 12 + (level * gstage_index_bits);
> - return 0;
> -}
> -
> -static int gstage_level_to_page_size(u32 level, unsigned long *out_pgsize)
> -{
> - int rc;
> - unsigned long page_order = PAGE_SHIFT;
> -
> - rc = gstage_level_to_page_order(level, &page_order);
> - if (rc)
> - return rc;
> -
> - *out_pgsize = BIT(page_order);
> - return 0;
> -}
> -
> -static bool gstage_get_leaf_entry(struct kvm *kvm, gpa_t addr,
> - pte_t **ptepp, u32 *ptep_level)
> -{
> - pte_t *ptep;
> - u32 current_level = gstage_pgd_levels - 1;
> -
> - *ptep_level = current_level;
> - ptep = (pte_t *)kvm->arch.pgd;
> - ptep = &ptep[gstage_pte_index(addr, current_level)];
> - while (ptep && pte_val(ptep_get(ptep))) {
> - if (gstage_pte_leaf(ptep)) {
> - *ptep_level = current_level;
> - *ptepp = ptep;
> - return true;
> - }
> -
> - if (current_level) {
> - current_level--;
> - *ptep_level = current_level;
> - ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> - ptep = &ptep[gstage_pte_index(addr, current_level)];
> - } else {
> - ptep = NULL;
> - }
> - }
> -
> - return false;
> -}
> -
> -static void gstage_remote_tlb_flush(struct kvm *kvm, u32 level, gpa_t addr)
> -{
> - unsigned long order = PAGE_SHIFT;
> -
> - if (gstage_level_to_page_order(level, &order))
> - return;
> - addr &= ~(BIT(order) - 1);
> -
> - kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0, addr, BIT(order), order);
> -}
> -
> -static int gstage_set_pte(struct kvm *kvm,
> - struct kvm_mmu_memory_cache *pcache,
> - const struct kvm_gstage_mapping *map)
> -{
> - u32 current_level = gstage_pgd_levels - 1;
> - pte_t *next_ptep = (pte_t *)kvm->arch.pgd;
> - pte_t *ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
> -
> - if (current_level < map->level)
> - return -EINVAL;
> -
> - while (current_level != map->level) {
> - if (gstage_pte_leaf(ptep))
> - return -EEXIST;
> -
> - if (!pte_val(ptep_get(ptep))) {
> - if (!pcache)
> - return -ENOMEM;
> - next_ptep = kvm_mmu_memory_cache_alloc(pcache);
> - if (!next_ptep)
> - return -ENOMEM;
> - set_pte(ptep, pfn_pte(PFN_DOWN(__pa(next_ptep)),
> - __pgprot(_PAGE_TABLE)));
> - } else {
> - if (gstage_pte_leaf(ptep))
> - return -EEXIST;
> - next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> - }
> -
> - current_level--;
> - ptep = &next_ptep[gstage_pte_index(map->addr, current_level)];
> - }
> -
> - if (pte_val(*ptep) != pte_val(map->pte)) {
> - set_pte(ptep, map->pte);
> - if (gstage_pte_leaf(ptep))
> - gstage_remote_tlb_flush(kvm, current_level, map->addr);
> - }
> -
> - return 0;
> -}
> -
> -static int gstage_map_page(struct kvm *kvm,
> - struct kvm_mmu_memory_cache *pcache,
> - gpa_t gpa, phys_addr_t hpa,
> - unsigned long page_size,
> - bool page_rdonly, bool page_exec,
> - struct kvm_gstage_mapping *out_map)
> -{
> - pgprot_t prot;
> - int ret;
> -
> - out_map->addr = gpa;
> - out_map->level = 0;
> -
> - ret = gstage_page_size_to_level(page_size, &out_map->level);
> - if (ret)
> - return ret;
> -
> - /*
> - * A RISC-V implementation can choose to either:
> - * 1) Update 'A' and 'D' PTE bits in hardware
> - * 2) Generate page fault when 'A' and/or 'D' bits are not set
> - * PTE so that software can update these bits.
> - *
> - * We support both options mentioned above. To achieve this, we
> - * always set 'A' and 'D' PTE bits at time of creating G-stage
> - * mapping. To support KVM dirty page logging with both options
> - * mentioned above, we will write-protect G-stage PTEs to track
> - * dirty pages.
> - */
> -
> - if (page_exec) {
> - if (page_rdonly)
> - prot = PAGE_READ_EXEC;
> - else
> - prot = PAGE_WRITE_EXEC;
> - } else {
> - if (page_rdonly)
> - prot = PAGE_READ;
> - else
> - prot = PAGE_WRITE;
> - }
> - out_map->pte = pfn_pte(PFN_DOWN(hpa), prot);
> - out_map->pte = pte_mkdirty(out_map->pte);
> -
> - return gstage_set_pte(kvm, pcache, out_map);
> -}
> -
> -enum gstage_op {
> - GSTAGE_OP_NOP = 0, /* Nothing */
> - GSTAGE_OP_CLEAR, /* Clear/Unmap */
> - GSTAGE_OP_WP, /* Write-protect */
> -};
> -
> -static void gstage_op_pte(struct kvm *kvm, gpa_t addr,
> - pte_t *ptep, u32 ptep_level, enum gstage_op op)
> -{
> - int i, ret;
> - pte_t old_pte, *next_ptep;
> - u32 next_ptep_level;
> - unsigned long next_page_size, page_size;
> -
> - ret = gstage_level_to_page_size(ptep_level, &page_size);
> - if (ret)
> - return;
> -
> - BUG_ON(addr & (page_size - 1));
> -
> - if (!pte_val(ptep_get(ptep)))
> - return;
> -
> - if (ptep_level && !gstage_pte_leaf(ptep)) {
> - next_ptep = (pte_t *)gstage_pte_page_vaddr(ptep_get(ptep));
> - next_ptep_level = ptep_level - 1;
> - ret = gstage_level_to_page_size(next_ptep_level,
> - &next_page_size);
> - if (ret)
> - return;
> -
> - if (op == GSTAGE_OP_CLEAR)
> - set_pte(ptep, __pte(0));
> - for (i = 0; i < PTRS_PER_PTE; i++)
> - gstage_op_pte(kvm, addr + i * next_page_size,
> - &next_ptep[i], next_ptep_level, op);
> - if (op == GSTAGE_OP_CLEAR)
> - put_page(virt_to_page(next_ptep));
> - } else {
> - old_pte = *ptep;
> - if (op == GSTAGE_OP_CLEAR)
> - set_pte(ptep, __pte(0));
> - else if (op == GSTAGE_OP_WP)
> - set_pte(ptep, __pte(pte_val(ptep_get(ptep)) & ~_PAGE_WRITE));
> - if (pte_val(*ptep) != pte_val(old_pte))
> - gstage_remote_tlb_flush(kvm, ptep_level, addr);
> - }
> -}
> -
> -static void gstage_unmap_range(struct kvm *kvm, gpa_t start,
> - gpa_t size, bool may_block)
> -{
> - int ret;
> - pte_t *ptep;
> - u32 ptep_level;
> - bool found_leaf;
> - unsigned long page_size;
> - gpa_t addr = start, end = start + size;
> -
> - while (addr < end) {
> - found_leaf = gstage_get_leaf_entry(kvm, addr,
> - &ptep, &ptep_level);
> - ret = gstage_level_to_page_size(ptep_level, &page_size);
> - if (ret)
> - break;
> -
> - if (!found_leaf)
> - goto next;
> -
> - if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> - gstage_op_pte(kvm, addr, ptep,
> - ptep_level, GSTAGE_OP_CLEAR);
> -
> -next:
> - addr += page_size;
> -
> - /*
> - * If the range is too large, release the kvm->mmu_lock
> - * to prevent starvation and lockup detector warnings.
> - */
> - if (may_block && addr < end)
> - cond_resched_lock(&kvm->mmu_lock);
> - }
> -}
> -
> -static void gstage_wp_range(struct kvm *kvm, gpa_t start, gpa_t end)
> -{
> - int ret;
> - pte_t *ptep;
> - u32 ptep_level;
> - bool found_leaf;
> - gpa_t addr = start;
> - unsigned long page_size;
> -
> - while (addr < end) {
> - found_leaf = gstage_get_leaf_entry(kvm, addr,
> - &ptep, &ptep_level);
> - ret = gstage_level_to_page_size(ptep_level, &page_size);
> - if (ret)
> - break;
> -
> - if (!found_leaf)
> - goto next;
> -
> - if (!(addr & (page_size - 1)) && ((end - addr) >= page_size))
> - gstage_op_pte(kvm, addr, ptep,
> - ptep_level, GSTAGE_OP_WP);
> -
> -next:
> - addr += page_size;
> - }
> -}
> -
> -static void gstage_wp_memory_region(struct kvm *kvm, int slot)
> +static void mmu_wp_memory_region(struct kvm *kvm, int slot)
> {
> struct kvm_memslots *slots = kvm_memslots(kvm);
> struct kvm_memory_slot *memslot = id_to_memslot(slots, slot);
> phys_addr_t start = memslot->base_gfn << PAGE_SHIFT;
> phys_addr_t end = (memslot->base_gfn + memslot->npages) << PAGE_SHIFT;
> + struct kvm_gstage gstage;
> +
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
>
> spin_lock(&kvm->mmu_lock);
> - gstage_wp_range(kvm, start, end);
> + kvm_riscv_gstage_wp_range(&gstage, start, end);
> spin_unlock(&kvm->mmu_lock);
> kvm_flush_remote_tlbs_memslot(kvm, memslot);
> }
>
> -int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> - phys_addr_t hpa, unsigned long size,
> - bool writable, bool in_atomic)
> +int kvm_riscv_mmu_ioremap(struct kvm *kvm, gpa_t gpa, phys_addr_t hpa,
> + unsigned long size, bool writable, bool in_atomic)
> {
> int ret = 0;
> unsigned long pfn;
> @@ -362,6 +46,12 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> .gfp_zero = __GFP_ZERO,
> };
> struct kvm_gstage_mapping map;
> + struct kvm_gstage gstage;
> +
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
>
> end = (gpa + size + PAGE_SIZE - 1) & PAGE_MASK;
> pfn = __phys_to_pfn(hpa);
> @@ -374,12 +64,12 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> if (!writable)
> map.pte = pte_wrprotect(map.pte);
>
> - ret = kvm_mmu_topup_memory_cache(&pcache, gstage_pgd_levels);
> + ret = kvm_mmu_topup_memory_cache(&pcache, kvm_riscv_gstage_pgd_levels);
> if (ret)
> goto out;
>
> spin_lock(&kvm->mmu_lock);
> - ret = gstage_set_pte(kvm, &pcache, &map);
> + ret = kvm_riscv_gstage_set_pte(&gstage, &pcache, &map);
> spin_unlock(&kvm->mmu_lock);
> if (ret)
> goto out;
> @@ -392,10 +82,17 @@ int kvm_riscv_gstage_ioremap(struct kvm *kvm, gpa_t gpa,
> return ret;
> }
>
> -void kvm_riscv_gstage_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size)
> +void kvm_riscv_mmu_iounmap(struct kvm *kvm, gpa_t gpa, unsigned long size)
> {
> + struct kvm_gstage gstage;
> +
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> +
> spin_lock(&kvm->mmu_lock);
> - gstage_unmap_range(kvm, gpa, size, false);
> + kvm_riscv_gstage_unmap_range(&gstage, gpa, size, false);
> spin_unlock(&kvm->mmu_lock);
> }
>
> @@ -407,8 +104,14 @@ void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
> phys_addr_t base_gfn = slot->base_gfn + gfn_offset;
> phys_addr_t start = (base_gfn + __ffs(mask)) << PAGE_SHIFT;
> phys_addr_t end = (base_gfn + __fls(mask) + 1) << PAGE_SHIFT;
> + struct kvm_gstage gstage;
> +
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
>
> - gstage_wp_range(kvm, start, end);
> + kvm_riscv_gstage_wp_range(&gstage, start, end);
> }
>
> void kvm_arch_sync_dirty_log(struct kvm *kvm, struct kvm_memory_slot *memslot)
> @@ -425,7 +128,7 @@ void kvm_arch_memslots_updated(struct kvm *kvm, u64 gen)
>
> void kvm_arch_flush_shadow_all(struct kvm *kvm)
> {
> - kvm_riscv_gstage_free_pgd(kvm);
> + kvm_riscv_mmu_free_pgd(kvm);
> }
>
> void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
> @@ -433,9 +136,15 @@ void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
> {
> gpa_t gpa = slot->base_gfn << PAGE_SHIFT;
> phys_addr_t size = slot->npages << PAGE_SHIFT;
> + struct kvm_gstage gstage;
> +
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
>
> spin_lock(&kvm->mmu_lock);
> - gstage_unmap_range(kvm, gpa, size, false);
> + kvm_riscv_gstage_unmap_range(&gstage, gpa, size, false);
> spin_unlock(&kvm->mmu_lock);
> }
>
> @@ -450,7 +159,7 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
> * the memory slot is write protected.
> */
> if (change != KVM_MR_DELETE && new->flags & KVM_MEM_LOG_DIRTY_PAGES)
> - gstage_wp_memory_region(kvm, new->id);
> + mmu_wp_memory_region(kvm, new->id);
> }
>
> int kvm_arch_prepare_memory_region(struct kvm *kvm,
> @@ -472,7 +181,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
> * space addressable by the KVM guest GPA space.
> */
> if ((new->base_gfn + new->npages) >=
> - (gstage_gpa_size >> PAGE_SHIFT))
> + (kvm_riscv_gstage_gpa_size >> PAGE_SHIFT))
> return -EFAULT;
>
> hva = new->userspace_addr;
> @@ -528,9 +237,8 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
> goto out;
> }
>
> - ret = kvm_riscv_gstage_ioremap(kvm, gpa, pa,
> - vm_end - vm_start,
> - writable, false);
> + ret = kvm_riscv_mmu_ioremap(kvm, gpa, pa, vm_end - vm_start,
> + writable, false);
> if (ret)
> break;
> }
> @@ -541,7 +249,7 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
> goto out;
>
> if (ret)
> - kvm_riscv_gstage_iounmap(kvm, base_gpa, size);
> + kvm_riscv_mmu_iounmap(kvm, base_gpa, size);
>
> out:
> mmap_read_unlock(current->mm);
> @@ -550,12 +258,18 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
>
> bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
> {
> + struct kvm_gstage gstage;
> +
> if (!kvm->arch.pgd)
> return false;
>
> - gstage_unmap_range(kvm, range->start << PAGE_SHIFT,
> - (range->end - range->start) << PAGE_SHIFT,
> - range->may_block);
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> + kvm_riscv_gstage_unmap_range(&gstage, range->start << PAGE_SHIFT,
> + (range->end - range->start) << PAGE_SHIFT,
> + range->may_block);
> return false;
> }
>
> @@ -564,14 +278,19 @@ bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
> pte_t *ptep;
> u32 ptep_level = 0;
> u64 size = (range->end - range->start) << PAGE_SHIFT;
> + struct kvm_gstage gstage;
>
> if (!kvm->arch.pgd)
> return false;
>
> WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
>
> - if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
> - &ptep, &ptep_level))
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> + if (!kvm_riscv_gstage_get_leaf(&gstage, range->start << PAGE_SHIFT,
> + &ptep, &ptep_level))
> return false;
>
> return ptep_test_and_clear_young(NULL, 0, ptep);
> @@ -582,23 +301,27 @@ bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
> pte_t *ptep;
> u32 ptep_level = 0;
> u64 size = (range->end - range->start) << PAGE_SHIFT;
> + struct kvm_gstage gstage;
>
> if (!kvm->arch.pgd)
> return false;
>
> WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
>
> - if (!gstage_get_leaf_entry(kvm, range->start << PAGE_SHIFT,
> - &ptep, &ptep_level))
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> + if (!kvm_riscv_gstage_get_leaf(&gstage, range->start << PAGE_SHIFT,
> + &ptep, &ptep_level))
> return false;
>
> return pte_young(ptep_get(ptep));
> }
>
> -int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> - struct kvm_memory_slot *memslot,
> - gpa_t gpa, unsigned long hva, bool is_write,
> - struct kvm_gstage_mapping *out_map)
> +int kvm_riscv_mmu_map(struct kvm_vcpu *vcpu, struct kvm_memory_slot *memslot,
> + gpa_t gpa, unsigned long hva, bool is_write,
> + struct kvm_gstage_mapping *out_map)
> {
> int ret;
> kvm_pfn_t hfn;
> @@ -611,13 +334,19 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> bool logging = (memslot->dirty_bitmap &&
> !(memslot->flags & KVM_MEM_READONLY)) ? true : false;
> unsigned long vma_pagesize, mmu_seq;
> + struct kvm_gstage gstage;
> struct page *page;
>
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> +
> /* Setup initial state of output mapping */
> memset(out_map, 0, sizeof(*out_map));
>
> /* We need minimum second+third level pages */
> - ret = kvm_mmu_topup_memory_cache(pcache, gstage_pgd_levels);
> + ret = kvm_mmu_topup_memory_cache(pcache, kvm_riscv_gstage_pgd_levels);
> if (ret) {
> kvm_err("Failed to topup G-stage cache\n");
> return ret;
> @@ -684,11 +413,11 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
>
> if (writable) {
> mark_page_dirty(kvm, gfn);
> - ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
> - vma_pagesize, false, true, out_map);
> + ret = kvm_riscv_gstage_map_page(&gstage, pcache, gpa, hfn << PAGE_SHIFT,
> + vma_pagesize, false, true, out_map);
> } else {
> - ret = gstage_map_page(kvm, pcache, gpa, hfn << PAGE_SHIFT,
> - vma_pagesize, true, true, out_map);
> + ret = kvm_riscv_gstage_map_page(&gstage, pcache, gpa, hfn << PAGE_SHIFT,
> + vma_pagesize, true, true, out_map);
> }
>
> if (ret)
> @@ -700,7 +429,7 @@ int kvm_riscv_gstage_map(struct kvm_vcpu *vcpu,
> return ret;
> }
>
> -int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
> +int kvm_riscv_mmu_alloc_pgd(struct kvm *kvm)
> {
> struct page *pgd_page;
>
> @@ -710,7 +439,7 @@ int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
> }
>
> pgd_page = alloc_pages(GFP_KERNEL | __GFP_ZERO,
> - get_order(gstage_pgd_size));
> + get_order(kvm_riscv_gstage_pgd_size));
> if (!pgd_page)
> return -ENOMEM;
> kvm->arch.pgd = page_to_virt(pgd_page);
> @@ -719,13 +448,18 @@ int kvm_riscv_gstage_alloc_pgd(struct kvm *kvm)
> return 0;
> }
>
> -void kvm_riscv_gstage_free_pgd(struct kvm *kvm)
> +void kvm_riscv_mmu_free_pgd(struct kvm *kvm)
> {
> + struct kvm_gstage gstage;
> void *pgd = NULL;
>
> spin_lock(&kvm->mmu_lock);
> if (kvm->arch.pgd) {
> - gstage_unmap_range(kvm, 0UL, gstage_gpa_size, false);
> + gstage.kvm = kvm;
> + gstage.flags = 0;
> + gstage.vmid = READ_ONCE(kvm->arch.vmid.vmid);
> + gstage.pgd = kvm->arch.pgd;
> + kvm_riscv_gstage_unmap_range(&gstage, 0UL, kvm_riscv_gstage_gpa_size, false);
> pgd = READ_ONCE(kvm->arch.pgd);
> kvm->arch.pgd = NULL;
> kvm->arch.pgd_phys = 0;
> @@ -733,12 +467,12 @@ void kvm_riscv_gstage_free_pgd(struct kvm *kvm)
> spin_unlock(&kvm->mmu_lock);
>
> if (pgd)
> - free_pages((unsigned long)pgd, get_order(gstage_pgd_size));
> + free_pages((unsigned long)pgd, get_order(kvm_riscv_gstage_pgd_size));
> }
>
> -void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu)
> +void kvm_riscv_mmu_update_hgatp(struct kvm_vcpu *vcpu)
> {
> - unsigned long hgatp = gstage_mode;
> + unsigned long hgatp = kvm_riscv_gstage_mode << HGATP_MODE_SHIFT;
> struct kvm_arch *k = &vcpu->kvm->arch;
>
> hgatp |= (READ_ONCE(k->vmid.vmid) << HGATP_VMID_SHIFT) & HGATP_VMID;
> @@ -749,37 +483,3 @@ void kvm_riscv_gstage_update_hgatp(struct kvm_vcpu *vcpu)
> if (!kvm_riscv_gstage_vmid_bits())
> kvm_riscv_local_hfence_gvma_all();
> }
> -
> -void __init kvm_riscv_gstage_mode_detect(void)
> -{
> -#ifdef CONFIG_64BIT
> - /* Try Sv57x4 G-stage mode */
> - csr_write(CSR_HGATP, HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
> - if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV57X4) {
> - gstage_mode = (HGATP_MODE_SV57X4 << HGATP_MODE_SHIFT);
> - gstage_pgd_levels = 5;
> - goto skip_sv48x4_test;
> - }
> -
> - /* Try Sv48x4 G-stage mode */
> - csr_write(CSR_HGATP, HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
> - if ((csr_read(CSR_HGATP) >> HGATP_MODE_SHIFT) == HGATP_MODE_SV48X4) {
> - gstage_mode = (HGATP_MODE_SV48X4 << HGATP_MODE_SHIFT);
> - gstage_pgd_levels = 4;
> - }
> -skip_sv48x4_test:
> -
> - csr_write(CSR_HGATP, 0);
> - kvm_riscv_local_hfence_gvma_all();
> -#endif
> -}
> -
> -unsigned long __init kvm_riscv_gstage_mode(void)
> -{
> - return gstage_mode >> HGATP_MODE_SHIFT;
> -}
> -
> -int kvm_riscv_gstage_gpa_bits(void)
> -{
> - return gstage_gpa_bits;
> -}
> diff --git a/arch/riscv/kvm/vcpu.c b/arch/riscv/kvm/vcpu.c
> index 8ad7b31f5939..fe028b4274df 100644
> --- a/arch/riscv/kvm/vcpu.c
> +++ b/arch/riscv/kvm/vcpu.c
> @@ -632,7 +632,7 @@ void kvm_arch_vcpu_load(struct kvm_vcpu *vcpu, int cpu)
> }
> }
>
> - kvm_riscv_gstage_update_hgatp(vcpu);
> + kvm_riscv_mmu_update_hgatp(vcpu);
>
> kvm_riscv_vcpu_timer_restore(vcpu);
>
> @@ -717,7 +717,7 @@ static void kvm_riscv_check_vcpu_requests(struct kvm_vcpu *vcpu)
> kvm_riscv_reset_vcpu(vcpu, true);
>
> if (kvm_check_request(KVM_REQ_UPDATE_HGATP, vcpu))
> - kvm_riscv_gstage_update_hgatp(vcpu);
> + kvm_riscv_mmu_update_hgatp(vcpu);
>
> if (kvm_check_request(KVM_REQ_FENCE_I, vcpu))
> kvm_riscv_fence_i_process(vcpu);
> diff --git a/arch/riscv/kvm/vcpu_exit.c b/arch/riscv/kvm/vcpu_exit.c
> index 6b4694bc07ea..0bb0c51e3c89 100644
> --- a/arch/riscv/kvm/vcpu_exit.c
> +++ b/arch/riscv/kvm/vcpu_exit.c
> @@ -43,8 +43,9 @@ static int gstage_page_fault(struct kvm_vcpu *vcpu, struct kvm_run *run,
> };
> }
>
> - ret = kvm_riscv_gstage_map(vcpu, memslot, fault_addr, hva,
> - (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false, &host_map);
> + ret = kvm_riscv_mmu_map(vcpu, memslot, fault_addr, hva,
> + (trap->scause == EXC_STORE_GUEST_PAGE_FAULT) ? true : false,
> + &host_map);
> if (ret < 0)
> return ret;
>
> diff --git a/arch/riscv/kvm/vm.c b/arch/riscv/kvm/vm.c
> index 8601cf29e5f8..66d91ae6e9b2 100644
> --- a/arch/riscv/kvm/vm.c
> +++ b/arch/riscv/kvm/vm.c
> @@ -32,13 +32,13 @@ int kvm_arch_init_vm(struct kvm *kvm, unsigned long type)
> {
> int r;
>
> - r = kvm_riscv_gstage_alloc_pgd(kvm);
> + r = kvm_riscv_mmu_alloc_pgd(kvm);
> if (r)
> return r;
>
> r = kvm_riscv_gstage_vmid_init(kvm);
> if (r) {
> - kvm_riscv_gstage_free_pgd(kvm);
> + kvm_riscv_mmu_free_pgd(kvm);
> return r;
> }
>
> @@ -200,7 +200,7 @@ int kvm_vm_ioctl_check_extension(struct kvm *kvm, long ext)
> r = KVM_USER_MEM_SLOTS;
> break;
> case KVM_CAP_VM_GPA_BITS:
> - r = kvm_riscv_gstage_gpa_bits();
> + r = kvm_riscv_gstage_gpa_bits;
> break;
> default:
> r = 0;
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs
2025-06-18 11:35 ` [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs Anup Patel
@ 2025-06-25 7:58 ` Nutty Liu
0 siblings, 0 replies; 28+ messages in thread
From: Nutty Liu @ 2025-06-25 7:58 UTC (permalink / raw)
To: Anup Patel, Atish Patra
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/2025 7:35 PM, Anup Patel wrote:
> Currently, all kvm_riscv_hfence_xyz() APIs assume VMID to be the
> host VMID of the Guest/VM which resticts use of these APIs only
> for host TLB maintenance. Let's allow passing VMID as a parameter
> to all kvm_riscv_hfence_xyz() APIs so that they can be re-used
> for nested virtualization related TLB maintenance.
>
> Signed-off-by: Anup Patel <apatel@ventanamicro.com>
> ---
> arch/riscv/include/asm/kvm_tlb.h | 17 ++++++---
> arch/riscv/kvm/gstage.c | 3 +-
> arch/riscv/kvm/tlb.c | 61 ++++++++++++++++++++-----------
> arch/riscv/kvm/vcpu_sbi_replace.c | 17 +++++----
> arch/riscv/kvm/vcpu_sbi_v01.c | 25 ++++++-------
> 5 files changed, 73 insertions(+), 50 deletions(-)
>
> diff --git a/arch/riscv/include/asm/kvm_tlb.h b/arch/riscv/include/asm/kvm_tlb.h
> index f67e03edeaec..38a2f933ad3a 100644
> --- a/arch/riscv/include/asm/kvm_tlb.h
> +++ b/arch/riscv/include/asm/kvm_tlb.h
> @@ -11,9 +11,11 @@
> enum kvm_riscv_hfence_type {
> KVM_RISCV_HFENCE_UNKNOWN = 0,
> KVM_RISCV_HFENCE_GVMA_VMID_GPA,
> + KVM_RISCV_HFENCE_GVMA_VMID_ALL,
> KVM_RISCV_HFENCE_VVMA_ASID_GVA,
> KVM_RISCV_HFENCE_VVMA_ASID_ALL,
> KVM_RISCV_HFENCE_VVMA_GVA,
> + KVM_RISCV_HFENCE_VVMA_ALL
> };
>
> struct kvm_riscv_hfence {
> @@ -59,21 +61,24 @@ void kvm_riscv_fence_i(struct kvm *kvm,
> void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> gpa_t gpa, gpa_t gpsz,
> - unsigned long order);
> + unsigned long order, unsigned long vmid);
> void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask);
> + unsigned long hbase, unsigned long hmask,
> + unsigned long vmid);
> void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> unsigned long gva, unsigned long gvsz,
> - unsigned long order, unsigned long asid);
> + unsigned long order, unsigned long asid,
> + unsigned long vmid);
> void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> - unsigned long asid);
> + unsigned long asid, unsigned long vmid);
> void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> unsigned long gva, unsigned long gvsz,
> - unsigned long order);
> + unsigned long order, unsigned long vmid);
> void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask);
> + unsigned long hbase, unsigned long hmask,
> + unsigned long vmid);
>
> #endif
> diff --git a/arch/riscv/kvm/gstage.c b/arch/riscv/kvm/gstage.c
> index 9c7c44f09b05..24c270d6d0e2 100644
> --- a/arch/riscv/kvm/gstage.c
> +++ b/arch/riscv/kvm/gstage.c
> @@ -117,7 +117,8 @@ static void gstage_tlb_flush(struct kvm_gstage *gstage, u32 level, gpa_t addr)
> if (gstage->flags & KVM_GSTAGE_FLAGS_LOCAL)
> kvm_riscv_local_hfence_gvma_vmid_gpa(gstage->vmid, addr, BIT(order), order);
> else
> - kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order);
> + kvm_riscv_hfence_gvma_vmid_gpa(gstage->kvm, -1UL, 0, addr, BIT(order), order,
> + gstage->vmid);
> }
>
> int kvm_riscv_gstage_set_pte(struct kvm_gstage *gstage,
> diff --git a/arch/riscv/kvm/tlb.c b/arch/riscv/kvm/tlb.c
> index 349fcfc93f54..3c5a70a2b927 100644
> --- a/arch/riscv/kvm/tlb.c
> +++ b/arch/riscv/kvm/tlb.c
> @@ -251,6 +251,12 @@ void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
> kvm_riscv_local_hfence_gvma_vmid_gpa(d.vmid, d.addr,
> d.size, d.order);
> break;
> + case KVM_RISCV_HFENCE_GVMA_VMID_ALL:
> + if (kvm_riscv_nacl_available())
> + nacl_hfence_gvma_vmid_all(nacl_shmem(), d.vmid);
> + else
> + kvm_riscv_local_hfence_gvma_vmid_all(d.vmid);
> + break;
> case KVM_RISCV_HFENCE_VVMA_ASID_GVA:
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_RCVD);
> if (kvm_riscv_nacl_available())
> @@ -276,6 +282,13 @@ void kvm_riscv_hfence_process(struct kvm_vcpu *vcpu)
> kvm_riscv_local_hfence_vvma_gva(d.vmid, d.addr,
> d.size, d.order);
> break;
> + case KVM_RISCV_HFENCE_VVMA_ALL:
> + kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_RCVD);
> + if (kvm_riscv_nacl_available())
> + nacl_hfence_vvma_all(nacl_shmem(), d.vmid);
> + else
> + kvm_riscv_local_hfence_vvma_all(d.vmid);
> + break;
> default:
> break;
> }
> @@ -328,14 +341,13 @@ void kvm_riscv_fence_i(struct kvm *kvm,
> void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> gpa_t gpa, gpa_t gpsz,
> - unsigned long order)
> + unsigned long order, unsigned long vmid)
> {
> - struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_GVMA_VMID_GPA;
> data.asid = 0;
> - data.vmid = READ_ONCE(v->vmid);
> + data.vmid = vmid;
> data.addr = gpa;
> data.size = gpsz;
> data.order = order;
> @@ -344,23 +356,28 @@ void kvm_riscv_hfence_gvma_vmid_gpa(struct kvm *kvm,
> }
>
> void kvm_riscv_hfence_gvma_vmid_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask)
> + unsigned long hbase, unsigned long hmask,
> + unsigned long vmid)
> {
> - make_xfence_request(kvm, hbase, hmask, KVM_REQ_TLB_FLUSH,
> - KVM_REQ_TLB_FLUSH, NULL);
> + struct kvm_riscv_hfence data = {0};
> +
> + data.type = KVM_RISCV_HFENCE_GVMA_VMID_ALL;
> + data.vmid = vmid;
> + make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
> + KVM_REQ_TLB_FLUSH, &data);
> }
>
> void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> unsigned long gva, unsigned long gvsz,
> - unsigned long order, unsigned long asid)
> + unsigned long order, unsigned long asid,
> + unsigned long vmid)
> {
> - struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_VVMA_ASID_GVA;
> data.asid = asid;
> - data.vmid = READ_ONCE(v->vmid);
> + data.vmid = vmid;
> data.addr = gva;
> data.size = gvsz;
> data.order = order;
> @@ -370,15 +387,13 @@ void kvm_riscv_hfence_vvma_asid_gva(struct kvm *kvm,
>
> void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> - unsigned long asid)
> + unsigned long asid, unsigned long vmid)
> {
> - struct kvm_vmid *v = &kvm->arch.vmid;
> - struct kvm_riscv_hfence data;
> + struct kvm_riscv_hfence data = {0};
>
> data.type = KVM_RISCV_HFENCE_VVMA_ASID_ALL;
> data.asid = asid;
> - data.vmid = READ_ONCE(v->vmid);
> - data.addr = data.size = data.order = 0;
> + data.vmid = vmid;
> make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
> KVM_REQ_HFENCE_VVMA_ALL, &data);
> }
> @@ -386,14 +401,13 @@ void kvm_riscv_hfence_vvma_asid_all(struct kvm *kvm,
> void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> unsigned long hbase, unsigned long hmask,
> unsigned long gva, unsigned long gvsz,
> - unsigned long order)
> + unsigned long order, unsigned long vmid)
> {
> - struct kvm_vmid *v = &kvm->arch.vmid;
> struct kvm_riscv_hfence data;
>
> data.type = KVM_RISCV_HFENCE_VVMA_GVA;
> data.asid = 0;
> - data.vmid = READ_ONCE(v->vmid);
> + data.vmid = vmid;
> data.addr = gva;
> data.size = gvsz;
> data.order = order;
> @@ -402,16 +416,21 @@ void kvm_riscv_hfence_vvma_gva(struct kvm *kvm,
> }
>
> void kvm_riscv_hfence_vvma_all(struct kvm *kvm,
> - unsigned long hbase, unsigned long hmask)
> + unsigned long hbase, unsigned long hmask,
> + unsigned long vmid)
> {
> - make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE_VVMA_ALL,
> - KVM_REQ_HFENCE_VVMA_ALL, NULL);
> + struct kvm_riscv_hfence data = {0};
> +
> + data.type = KVM_RISCV_HFENCE_VVMA_ALL;
> + data.vmid = vmid;
> + make_xfence_request(kvm, hbase, hmask, KVM_REQ_HFENCE,
> + KVM_REQ_HFENCE_VVMA_ALL, &data);
> }
>
> int kvm_arch_flush_remote_tlbs_range(struct kvm *kvm, gfn_t gfn, u64 nr_pages)
> {
> kvm_riscv_hfence_gvma_vmid_gpa(kvm, -1UL, 0,
> gfn << PAGE_SHIFT, nr_pages << PAGE_SHIFT,
> - PAGE_SHIFT);
> + PAGE_SHIFT, READ_ONCE(kvm->arch.vmid.vmid));
> return 0;
> }
> diff --git a/arch/riscv/kvm/vcpu_sbi_replace.c b/arch/riscv/kvm/vcpu_sbi_replace.c
> index b17fad091bab..b490ed1428a6 100644
> --- a/arch/riscv/kvm/vcpu_sbi_replace.c
> +++ b/arch/riscv/kvm/vcpu_sbi_replace.c
> @@ -96,6 +96,7 @@ static int kvm_sbi_ext_rfence_handler(struct kvm_vcpu *vcpu, struct kvm_run *run
> unsigned long hmask = cp->a0;
> unsigned long hbase = cp->a1;
> unsigned long funcid = cp->a6;
> + unsigned long vmid;
>
> switch (funcid) {
> case SBI_EXT_RFENCE_REMOTE_FENCE_I:
> @@ -103,22 +104,22 @@ static int kvm_sbi_ext_rfence_handler(struct kvm_vcpu *vcpu, struct kvm_run *run
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_FENCE_I_SENT);
> break;
> case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA:
> + vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
> if ((cp->a2 == 0 && cp->a3 == 0) || cp->a3 == -1UL)
> - kvm_riscv_hfence_vvma_all(vcpu->kvm, hbase, hmask);
> + kvm_riscv_hfence_vvma_all(vcpu->kvm, hbase, hmask, vmid);
> else
> kvm_riscv_hfence_vvma_gva(vcpu->kvm, hbase, hmask,
> - cp->a2, cp->a3, PAGE_SHIFT);
> + cp->a2, cp->a3, PAGE_SHIFT, vmid);
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_SENT);
> break;
> case SBI_EXT_RFENCE_REMOTE_SFENCE_VMA_ASID:
> + vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
> if ((cp->a2 == 0 && cp->a3 == 0) || cp->a3 == -1UL)
> - kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
> - hbase, hmask, cp->a4);
> + kvm_riscv_hfence_vvma_asid_all(vcpu->kvm, hbase, hmask,
> + cp->a4, vmid);
> else
> - kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
> - hbase, hmask,
> - cp->a2, cp->a3,
> - PAGE_SHIFT, cp->a4);
> + kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm, hbase, hmask, cp->a2,
> + cp->a3, PAGE_SHIFT, cp->a4, vmid);
> kvm_riscv_vcpu_pmu_incr_fw(vcpu, SBI_PMU_FW_HFENCE_VVMA_ASID_SENT);
> break;
> case SBI_EXT_RFENCE_REMOTE_HFENCE_GVMA:
> diff --git a/arch/riscv/kvm/vcpu_sbi_v01.c b/arch/riscv/kvm/vcpu_sbi_v01.c
> index 8f4c4fa16227..368dfddd23d9 100644
> --- a/arch/riscv/kvm/vcpu_sbi_v01.c
> +++ b/arch/riscv/kvm/vcpu_sbi_v01.c
> @@ -23,6 +23,7 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
> struct kvm *kvm = vcpu->kvm;
> struct kvm_cpu_context *cp = &vcpu->arch.guest_context;
> struct kvm_cpu_trap *utrap = retdata->utrap;
> + unsigned long vmid;
>
> switch (cp->a7) {
> case SBI_EXT_0_1_CONSOLE_GETCHAR:
> @@ -78,25 +79,21 @@ static int kvm_sbi_ext_v01_handler(struct kvm_vcpu *vcpu, struct kvm_run *run,
> if (cp->a7 == SBI_EXT_0_1_REMOTE_FENCE_I)
> kvm_riscv_fence_i(vcpu->kvm, 0, hmask);
> else if (cp->a7 == SBI_EXT_0_1_REMOTE_SFENCE_VMA) {
> + vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
> if (cp->a1 == 0 && cp->a2 == 0)
> - kvm_riscv_hfence_vvma_all(vcpu->kvm,
> - 0, hmask);
> + kvm_riscv_hfence_vvma_all(vcpu->kvm, 0, hmask, vmid);
> else
> - kvm_riscv_hfence_vvma_gva(vcpu->kvm,
> - 0, hmask,
> - cp->a1, cp->a2,
> - PAGE_SHIFT);
> + kvm_riscv_hfence_vvma_gva(vcpu->kvm, 0, hmask, cp->a1,
> + cp->a2, PAGE_SHIFT, vmid);
> } else {
> + vmid = READ_ONCE(vcpu->kvm->arch.vmid.vmid);
> if (cp->a1 == 0 && cp->a2 == 0)
> - kvm_riscv_hfence_vvma_asid_all(vcpu->kvm,
> - 0, hmask,
> - cp->a3);
> + kvm_riscv_hfence_vvma_asid_all(vcpu->kvm, 0, hmask,
> + cp->a3, vmid);
> else
> - kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm,
> - 0, hmask,
> - cp->a1, cp->a2,
> - PAGE_SHIFT,
> - cp->a3);
> + kvm_riscv_hfence_vvma_asid_gva(vcpu->kvm, 0, hmask,
> + cp->a1, cp->a2, PAGE_SHIFT,
> + cp->a3, vmid);
> }
> break;
> default:
Reviewed-by: Nutty Liu <liujingqi@lanxincomputing.com>
Thanks,
Nutty
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH v3 00/12] MMU related improvements for KVM RISC-V
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
` (12 preceding siblings ...)
2025-06-24 6:16 ` [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
@ 2025-06-25 7:59 ` Atish Patra
13 siblings, 0 replies; 28+ messages in thread
From: Atish Patra @ 2025-06-25 7:59 UTC (permalink / raw)
To: Anup Patel
Cc: Palmer Dabbelt, Paul Walmsley, Alexandre Ghiti, Andrew Jones,
Anup Patel, kvm, kvm-riscv, linux-riscv, linux-kernel
On 6/18/25 4:35 AM, Anup Patel wrote:
> This series primarily has various MMU improvements for KVM RISC-V
> and it also serves as a preparatory series for the upcoming nested
> virtualization support.
>
> PATCH1 to PATCH4: Few cosmetic improvements
> PATCH5 to PATCH7: TLB maintenance and SBI NACL related improvements
> PATCH8 to PATCH12: MMU related preparatory work for nested virtualization
For the entire series:
Tested-by: Atish Patra <atishp@rivosinc.com>
> These patches can also be found in the riscv_kvm_mmu_imp_v3 branch
> at: https://github.com/avpatel/linux.git
>
> Changes since v2:
> - Rebased upon Linux-6.16-rc2 and latest riscv_kvm_queue branch
> - Added Reviewed-by tags to appropriate patches
>
> Changes since v1:
> - Rebased upon Linux-6.16-rc1
> - Dropped PATCH1 and PATCH2 of v1 series since these are queued
> as fixes for Linux-6.16
> - Addressed Atish's comment on PATCH1 in this series
> - Added new PATCH7 in this series
>
> Anup Patel (12):
> RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value
> RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init()
> RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize()
> RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with
> KVM_REQ_TLB_FLUSH
> RISC-V: KVM: Don't flush TLB when PTE is unchanged
> RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range()
> RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect()
> RISC-V: KVM: Factor-out MMU related declarations into separate headers
> RISC-V: KVM: Introduce struct kvm_gstage_mapping
> RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence
> RISC-V: KVM: Factor-out g-stage page table management
> RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs
>
> arch/riscv/include/asm/kvm_aia.h | 2 +-
> arch/riscv/include/asm/kvm_gstage.h | 72 ++++
> arch/riscv/include/asm/kvm_host.h | 103 +-----
> arch/riscv/include/asm/kvm_mmu.h | 21 ++
> arch/riscv/include/asm/kvm_tlb.h | 84 +++++
> arch/riscv/include/asm/kvm_vmid.h | 27 ++
> arch/riscv/kvm/Makefile | 1 +
> arch/riscv/kvm/aia_device.c | 6 +-
> arch/riscv/kvm/aia_imsic.c | 12 +-
> arch/riscv/kvm/gstage.c | 338 +++++++++++++++++++
> arch/riscv/kvm/main.c | 3 +-
> arch/riscv/kvm/mmu.c | 499 ++++++----------------------
> arch/riscv/kvm/tlb.c | 110 +++---
> arch/riscv/kvm/vcpu.c | 26 +-
> arch/riscv/kvm/vcpu_exit.c | 20 +-
> arch/riscv/kvm/vcpu_sbi_replace.c | 17 +-
> arch/riscv/kvm/vcpu_sbi_v01.c | 25 +-
> arch/riscv/kvm/vm.c | 7 +-
> arch/riscv/kvm/vmid.c | 25 ++
> 19 files changed, 795 insertions(+), 603 deletions(-)
> create mode 100644 arch/riscv/include/asm/kvm_gstage.h
> create mode 100644 arch/riscv/include/asm/kvm_mmu.h
> create mode 100644 arch/riscv/include/asm/kvm_tlb.h
> create mode 100644 arch/riscv/include/asm/kvm_vmid.h
> create mode 100644 arch/riscv/kvm/gstage.c
>
--
kvm-riscv mailing list
kvm-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/kvm-riscv
^ permalink raw reply [flat|nested] 28+ messages in thread
end of thread, other threads:[~2025-06-25 9:53 UTC | newest]
Thread overview: 28+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-06-18 11:35 [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
2025-06-18 11:35 ` [PATCH v3 01/12] RISC-V: KVM: Check kvm_riscv_vcpu_alloc_vector_context() return value Anup Patel
2025-06-24 3:32 ` Atish Patra
2025-06-25 6:48 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 02/12] RISC-V: KVM: Drop the return value of kvm_riscv_vcpu_aia_init() Anup Patel
2025-06-24 3:35 ` Atish Patra
2025-06-25 6:50 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 03/12] RISC-V: KVM: Rename and move kvm_riscv_local_tlb_sanitize() Anup Patel
2025-06-18 11:35 ` [PATCH v3 04/12] RISC-V: KVM: Replace KVM_REQ_HFENCE_GVMA_VMID_ALL with KVM_REQ_TLB_FLUSH Anup Patel
2025-06-25 7:35 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 05/12] RISC-V: KVM: Don't flush TLB when PTE is unchanged Anup Patel
2025-06-25 7:38 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 06/12] RISC-V: KVM: Implement kvm_arch_flush_remote_tlbs_range() Anup Patel
2025-06-25 7:39 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 07/12] RISC-V: KVM: Use ncsr_xyz() in kvm_riscv_vcpu_trap_redirect() Anup Patel
2025-06-25 7:39 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 08/12] RISC-V: KVM: Factor-out MMU related declarations into separate headers Anup Patel
2025-06-25 7:40 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 09/12] RISC-V: KVM: Introduce struct kvm_gstage_mapping Anup Patel
2025-06-25 7:43 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 10/12] RISC-V: KVM: Add vmid field to struct kvm_riscv_hfence Anup Patel
2025-06-25 7:55 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 11/12] RISC-V: KVM: Factor-out g-stage page table management Anup Patel
2025-06-25 7:57 ` Nutty Liu
2025-06-18 11:35 ` [PATCH v3 12/12] RISC-V: KVM: Pass VMID as parameter to kvm_riscv_hfence_xyz() APIs Anup Patel
2025-06-25 7:58 ` Nutty Liu
2025-06-24 6:16 ` [PATCH v3 00/12] MMU related improvements for KVM RISC-V Anup Patel
2025-06-25 7:59 ` Atish Patra
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).