Skip to contentSkip to conversation

Fix hgatp mode settings within kvm_riscv_gstage_vmid_detect

PATCH In review [email protected] 4 participants 8 messages Aug 20, 2025 – Aug 21, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v3 cover Fix hgatp mode settings within kvm_riscv_gstage_vmid_detect

From: Fangyu Yu <[email protected]>

According to the RISC-V Privileged Architecture Spec, when MODE=Bare is selected,software must write zero to the remaining fields of hgatp.

So,use a valid mode when detecting vmids of the HW supported,meanwhile remove unnecessary HGATP csr_read.

---
Changes in v3:
- Add "Fixes" tag.
- Involve("RISC-V KVM: Remove unnecessary HGATP csr_read"), which depends on patch 1.

Changes in v2:
- Fixed build error since kvm_riscv_gstage_mode() has been modified.
---

Fangyu Yu (1):
RISC-V: KVM: Write hgatp register with valid mode bits

Guo Ren (Alibaba DAMO Academy) (1):
RISC-V KVM: Remove unnecessary HGATP csr_read

 arch/riscv/kvm/vmid.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
Signature · 1 line
2.49.0
#1 lore raw

PATCH v3 1/2 RISC-V: KVM: Write hgatp register with valid mode bits

From: Fangyu Yu <[email protected]>

According to the RISC-V Privileged Architecture Spec, when MODE=Bare is selected,software must write zero to the remaining fields of hgatp.

We have detected the valid mode supported by the HW before, So using a valid mode to detect how many vmid bits are supported.

Fixes fd7bb4a251df ("RISC-V: KVM: Implement VMID allocator")
---
 arch/riscv/kvm/vmid.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/riscv/kvm/vmid.c +2 -1
index 3b426c800480..5f33625f4070 100644
@@ -14,6 +14,7 @@
 #include <linux/smp.h>
 #include <linux/kvm_host.h>
 #include <asm/csr.h>
+#include <asm/kvm_mmu.h>
 #include <asm/kvm_tlb.h>
 #include <asm/kvm_vmid.h>
 
@@ -28,7 +29,7 @@ void __init kvm_riscv_gstage_vmid_detect(void)
 
 	/* Figure-out number of VMID bits in HW */
 	old = csr_read(CSR_HGATP);
-	csr_write(CSR_HGATP, old | HGATP_VMID);
+	csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
 	vmid_bits = csr_read(CSR_HGATP);
 	vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
 	vmid_bits = fls_long(vmid_bits);
Signature · 1 line
2.49.0
#2 lore raw

PATCH v3 2/2 RISC-V KVM: Remove unnecessary HGATP csr_read

In reply to [email protected]

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
 arch/riscv/kvm/vmid.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +1 -4
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
 void __init kvm_riscv_gstage_vmid_detect(void)
 {
-	unsigned long old;
-
 	/* Figure-out number of VMID bits in HW */
-	old = csr_read(CSR_HGATP);
 	csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
 	vmid_bits = csr_read(CSR_HGATP);
 	vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
 	vmid_bits = fls_long(vmid_bits);
-	csr_write(CSR_HGATP, old);
+	csr_write(CSR_HGATP, 0);
 
 	/* We polluted local TLB so flush all guest TLB */
 	kvm_riscv_local_hfence_gvma_all();
Signature · 1 line
2.49.0
#3 lore raw

Re: [PATCH v3 2/2] RISC-V KVM: Remove unnecessary HGATP csr_read

Quoted 20 lines · On Wed, Aug 20, 2025 at 08:59:52PM +0800, [email protected] wrote:

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
 arch/riscv/kvm/vmid.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +0 -0
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
 void __init kvm_riscv_gstage_vmid_detect(void)
 {

arch/riscv/kvm/vmid.c static DEFINE_SPINLOCK(vmid_lock); · line 33

Quoted 10 lines
arch/riscv/kvm/vmid.c +1 -4
-	unsigned long old;
-
 	/* Figure-out number of VMID bits in HW */
-	old = csr_read(CSR_HGATP);
 	csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
 	vmid_bits = csr_read(CSR_HGATP);
 	vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
 	vmid_bits = fls_long(vmid_bits);
-	csr_write(CSR_HGATP, old);
+	csr_write(CSR_HGATP, 0);

Is setting HGATP to 0 in gstage_mode_detect meaningless now? If so, it might be better to drop it and just keep the one here.

- Troy

Quoted 11 lines
arch/riscv/kvm/vmid.c +0 -0
 	/* We polluted local TLB so flush all guest TLB */
 	kvm_riscv_local_hfence_gvma_all();
Signature · 7 lines
2.49.0


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
#4 lore raw

Re: [PATCH v3 2/2] RISC-V KVM: Remove unnecessary HGATP csr_read

Quoted 20 lines · Troy Mitchell
Quoted 18 lines · On Wed, Aug 20, 2025 at 08:59:52PM +0800, [email protected] wrote:

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
 arch/riscv/kvm/vmid.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +0 -0
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
Quoted 13 lines

arch/riscv/kvm/vmid.c static DEFINE_SPINLOCK(vmid_lock); · line 33

Quoted 12 lines
arch/riscv/kvm/vmid.c +1 -4
 void __init kvm_riscv_gstage_vmid_detect(void)
 {
-     unsigned long old;
-
      /* Figure-out number of VMID bits in HW */
-     old = csr_read(CSR_HGATP);
      csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
      vmid_bits = csr_read(CSR_HGATP);
      vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
      vmid_bits = fls_long(vmid_bits);
-     csr_write(CSR_HGATP, old);
+     csr_write(CSR_HGATP, 0);

Is setting HGATP to 0 in gstage_mode_detect meaningless now?

It's not meaningless, it means keep hgatp off. CSR_HGATP is set to 0 by gstage_mode_detect(), but that's another independent function whose coincidence is set to 0. And in that function, we use "csr_write(CSR_HGATP, 0)", but here we change to confusing save_old & restore_old coding convention?

So, keep kvm_riscv_gstage_vmid_detect() coding convention clear. We should follow the same approach as gstage_mode_detect(), that is, "csr_write(CSR_HGATP, 0)".

- The first patch is focused on fixing meaning, so keep the minimum
modifications, no coding convention cleanup.
- The second one is about coding convention, which is no need to Cc:
[email protected]

Hope the above explanation makes sense.

Quoted 15 lines

If so, it might be better to drop it and just keep the one here.

- Troy

Quoted 11 lines
arch/riscv/kvm/vmid.c +0 -1
      /* We polluted local TLB so flush all guest TLB */
      kvm_riscv_local_hfence_gvma_all();
--

2.49.0

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

Signature · 2 lines
Best Regards
 Guo Ren
#5 lore raw

Re: [PATCH v3 2/2] RISC-V KVM: Remove unnecessary HGATP csr_read

In reply to Troy Mitchell
Quoted 20 lines · Troy Mitchell
Quoted 18 lines · On Wed, Aug 20, 2025 at 08:59:52PM +0800, [email protected] wrote:

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
 arch/riscv/kvm/vmid.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +0 -0
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
Quoted 14 lines

arch/riscv/kvm/vmid.c static DEFINE_SPINLOCK(vmid_lock); · line 33

Quoted 12 lines
arch/riscv/kvm/vmid.c +1 -4
 void __init kvm_riscv_gstage_vmid_detect(void)
 {
-     unsigned long old;
-
      /* Figure-out number of VMID bits in HW */
-     old = csr_read(CSR_HGATP);
      csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
      vmid_bits = csr_read(CSR_HGATP);
      vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
      vmid_bits = fls_long(vmid_bits);
-     csr_write(CSR_HGATP, old);
+     csr_write(CSR_HGATP, 0);

Is setting HGATP to 0 in gstage_mode_detect meaningless now? If so, it might be better to drop it and just keep the one here.

Sorry, I misunderstood here.

1. kvm_riscv_gstage_vmid_detect() & gstage_mode_detect() are indepent function, so keep csr_write(CSR_HGATP, 0) is considerable.

2. But your idea is good, because csr_write(CSR_HGATP, 0) would cause TLB flush in some micro-arch, which reduces the IPC. So, removing unnecessary CSR_write (CSR_HGATP, 0) is also considerable.

I would update V4 to merge kvm_riscv_gstage_vmid_detect() & gstage_mode_detect() into a single function, inspired by your idea.

Quoted 14 lines

- Troy

Quoted 11 lines
arch/riscv/kvm/vmid.c +0 -1
      /* We polluted local TLB so flush all guest TLB */
      kvm_riscv_local_hfence_gvma_all();
--

2.49.0

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

Signature · 2 lines
Best Regards
 Guo Ren
#6 lore raw

Re: [PATCH v3 2/2] RISC-V KVM: Remove unnecessary HGATP csr_read

Quoted 41 lines · Guo Ren
Quoted 20 lines · Troy Mitchell
Quoted 18 lines · On Wed, Aug 20, 2025 at 08:59:52PM +0800, [email protected] wrote:

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
 arch/riscv/kvm/vmid.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +0 -0
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
Quoted 14 lines

arch/riscv/kvm/vmid.c static DEFINE_SPINLOCK(vmid_lock); · line 33

Quoted 12 lines
arch/riscv/kvm/vmid.c +1 -4
 void __init kvm_riscv_gstage_vmid_detect(void)
 {
-     unsigned long old;
-
      /* Figure-out number of VMID bits in HW */
-     old = csr_read(CSR_HGATP);
      csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
      vmid_bits = csr_read(CSR_HGATP);
      vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
      vmid_bits = fls_long(vmid_bits);
-     csr_write(CSR_HGATP, old);
+     csr_write(CSR_HGATP, 0);

Is setting HGATP to 0 in gstage_mode_detect meaningless now? If so, it might be better to drop it and just keep the one here.

Sorry, I misunderstood here.

1. kvm_riscv_gstage_vmid_detect() & gstage_mode_detect() are indepent function, so keep csr_write(CSR_HGATP, 0) is considerable.

Quoted 6 lines

2. But your idea is good, because csr_write(CSR_HGATP, 0) would cause TLB flush in some micro-arch, which reduces the IPC. So, removing unnecessary CSR_write (CSR_HGATP, 0) is also considerable.

I would update V4 to merge kvm_riscv_gstage_vmid_detect() & gstage_mode_detect() into a single function, inspired by your idea.

I found we can't remove "csr_write(CSR_HGATP, 0)" in gstage_mode_detect(), because this patch needs to reset the HGATP when mode check failure:
https://lore.kernel.org/linux-riscv/[email protected]/

I would update to v4, which collects these related patches together for review convenience.

Quoted 21 lines
Quoted 15 lines

- Troy

Quoted 11 lines
arch/riscv/kvm/vmid.c +0 -1
      /* We polluted local TLB so flush all guest TLB */
      kvm_riscv_local_hfence_gvma_all();
--

2.49.0

_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv

--
Best Regards
Guo Ren

Signature · 2 lines
Best Regards
 Guo Ren
Reviewed-by on v3 2/2
#7 lore raw

Re: [PATCH v3 2/2] RISC-V KVM: Remove unnecessary HGATP csr_read

In reply to [email protected]
Quoted 20 lines · On 8/20/2025 8:59 PM, [email protected] wrote:

From: "Guo Ren (Alibaba DAMO Academy)" <[email protected]>

The HGATP has been set to zero in gstage_mode_detect(), so there is no need to save the old context. Unify the code convention with gstage_mode_detect().

---
  arch/riscv/kvm/vmid.c | 5 +----
  1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kvm/vmid.c +0 -0
index 5f33625f4070..abb1c2bf2542 100644
@@ -25,15 +25,12 @@ static DEFINE_SPINLOCK(vmid_lock);
 
  void __init kvm_riscv_gstage_vmid_detect(void)
  {

arch/riscv/kvm/vmid.c static DEFINE_SPINLOCK(vmid_lock); · line 33

Quoted 10 lines
arch/riscv/kvm/vmid.c +1 -4
-	unsigned long old;
-
  	/* Figure-out number of VMID bits in HW */
-	old = csr_read(CSR_HGATP);
  	csr_write(CSR_HGATP, (kvm_riscv_gstage_mode << HGATP_MODE_SHIFT) | HGATP_VMID);
  	vmid_bits = csr_read(CSR_HGATP);
  	vmid_bits = (vmid_bits & HGATP_VMID) >> HGATP_VMID_SHIFT;
  	vmid_bits = fls_long(vmid_bits);
-	csr_write(CSR_HGATP, old);
+	csr_write(CSR_HGATP, 0);

Seems there is no need to set 'CSR_HGATP' as 0.
Otherwise,

Thanks,
Nutty

Quoted 3 lines
arch/riscv/kvm/vmid.c +0 -0
  	/* We polluted local TLB so flush all guest TLB */
  	kvm_riscv_local_hfence_gvma_all();