Skip to contentSkip to conversation

target/riscv: Bugfix reserved bits in PTE for RV64

PATCH Superseded [email protected] 3 participants 6 messages Sep 24, 2019 – Sep 25, 2019 RISC-V

v2 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 target/riscv: Bugfix reserved bits in PTE for RV64

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

---
 target/riscv/cpu_bits.h   | 3 +++
 target/riscv/cpu_helper.c | 3 ++-
 2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +3 -0
index e998348..ae8aa0f 100644
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
target/riscv/cpu_helper.c +2 -1
index 87dd6a6..7a540cc 100644
@@ -258,10 +258,11 @@ restart:
         }
 #if defined(TARGET_RISCV32)
         target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 
         if (!(pte & PTE_V)) {
             /* Invalid PTE */
Signature · 1 line
2.7.4
author
#1 lore raw

I only tested it on qemu-3.1.0, pls have a try before merge.

Quoted 56 lines · On Tue, Sep 24, 2019 at 4:00 PM <[email protected]> wrote:

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

---
 target/riscv/cpu_bits.h   | 3 +++
 target/riscv/cpu_helper.c | 3 ++-
 2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +3 -0
index e998348..ae8aa0f 100644
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
target/riscv/cpu_helper.c +2 -2
index 87dd6a6..7a540cc 100644
@@ -258,10 +258,11 @@ restart:
         }
 #if defined(TARGET_RISCV32)
         target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 
         if (!(pte & PTE_V)) {
             /* Invalid PTE */
--

2.7.4

Signature · 4 lines
Best Regards
 Guo Ren

ML: https://lore.kernel.org/linux-csky/
#2 lore raw
In reply to [email protected]
Quoted 9 lines · On Tue, Sep 24, 2019 at 12:58 AM <[email protected]> wrote:

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Thanks for the patch!

The spec says "must be zeroed by software for forward compatibility" so I don't think it's correct for QEMU to zero out the bits.

Does this fix a bug you are seeing?

Quoted 4 lines

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

The changelog shouldn't be in the commit, it should be kept under the line line below.

Quoted 4 lines

---

The change log should go here.

Quoted 9 lines
 target/riscv/cpu_bits.h   | 3 +++
 target/riscv/cpu_helper.c | 3 ++-
 2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +0 -0
index e998348..ae8aa0f 100644

target/riscv/cpu_bits.h line 474

Quoted 5 lines
target/riscv/cpu_bits.h +2 -0
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)

I think it's just easier to define this as 0xFFC0000000000000ULL and remove the cast.

Quoted 9 lines
target/riscv/cpu_bits.h +1 -0
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
target/riscv/cpu_helper.c +0 -0
index 87dd6a6..7a540cc 100644

target/riscv/cpu_helper.c restart: · line 264

Quoted 9 lines
target/riscv/cpu_helper.c +2 -1
@@ -258,10 +258,11 @@ restart:
         }
 #if defined(TARGET_RISCV32)
         target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;

You don't have to move this shift

Quoted 6 lines
target/riscv/cpu_helper.c +0 -1
         if (!(pte & PTE_V)) {
             /* Invalid PTE */
--

2.7.4

#3 lore raw
Quoted 10 lines · 在 2019年9月25日,上午7:33,Alistair Francis <[email protected]> 写道:
Quoted 6 lines · On Tue, Sep 24, 2019 at 12:58 AM <[email protected]> wrote:

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

Quoted 8 lines
Quoted 3 lines

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Thanks for the patch!

The spec says "must be zeroed by software for forward compatibility" so I don't think it's correct for QEMU to zero out the bits.

QEMU don’t zero out the bits, QEMU just ignore the bits for ppn.

Quoted 2 lines

Does this fix a bug you are seeing?

Yes, because we try to reuse these bits as attributes.

Quoted 8 lines
Quoted 5 lines

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

The changelog shouldn't be in the commit, it should be kept under the line line below.

I just prefer to save them in commit.

Quoted 7 lines
Quoted 5 lines

---

The change log should go here.

OK, but git am we’ll lose them.

Quoted 13 lines
Quoted 13 lines

target/riscv/cpu_bits.h | 3 +++
target/riscv/cpu_helper.c | 3 ++-

2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +0 -0
index e998348..ae8aa0f 100644
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
Quoted 5 lines

target/riscv/cpu_bits.h line 474

Quoted 2 lines
target/riscv/cpu_bits.h +2 -0
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)

I think it's just easier to define this as 0xFFC0000000000000ULL and remove the cast.

OK follow your rule, but I still prefer prior.

Quoted 5 lines
Quoted 5 lines
target/riscv/cpu_bits.h +1 -0
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
Quoted 16 lines
Quoted 5 lines
target/riscv/cpu_helper.c +0 -0
index 87dd6a6..7a540cc 100644

target/riscv/cpu_helper.c restart: · line 264

Quoted 9 lines
target/riscv/cpu_helper.c +2 -1
@@ -258,10 +258,11 @@ restart:
        }
 #if defined(TARGET_RISCV32)
        target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
        target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;

You don't have to move this shift

En… Do you want this: ?

#if defined(TARGET_RISCV32)
        target_ulong pte = ldl_phys(cs->as, pte_addr);
+      hwaddr ppn = pte;
#elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);

+ hwaddr ppn = (pte & ~PTE_RESERVED);
#endif
+ ppn = ppn >> PTE_PPN_SHIFT;

The pte couldn’t be destroyed, just ppn ignore the RESERVED bits.

#4 lore raw
Quoted 15 lines · Guo Ren
Quoted 15 lines · 在 2019年9月25日,上午7:33,Alistair Francis <[email protected]> 写道:
Quoted 9 lines · On Tue, Sep 24, 2019 at 12:58 AM <[email protected]> wrote:

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Quoted 5 lines
Quoted 4 lines

Thanks for the patch!

The spec says "must be zeroed by software for forward compatibility" so I don't think it's correct for QEMU to zero out the bits.

QEMU don’t zero out the bits, QEMU just ignore the bits for ppn.

Yes, from reading the spec that seems to be the correct behaviour.

Quoted 4 lines
Quoted 3 lines

Does this fix a bug you are seeing?

Yes, because we try to reuse these bits as attributes.

That isn't really a bug then, the spec says not to do that.

Quoted 10 lines
Quoted 9 lines
Quoted 6 lines

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

The changelog shouldn't be in the commit, it should be kept under the line line below.

I just prefer to save them in commit.

Fair, but in QEMU we don't commit the change log in the commit.

Quoted 53 lines
Quoted 8 lines
Quoted 6 lines

---

The change log should go here.

OK, but git am we’ll lose them.

Quoted 13 lines
Quoted 13 lines

target/riscv/cpu_bits.h | 3 +++
target/riscv/cpu_helper.c | 3 ++-

2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +0 -0
index e998348..ae8aa0f 100644
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
Quoted 5 lines

target/riscv/cpu_bits.h line 474

Quoted 2 lines
target/riscv/cpu_bits.h +2 -0
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)

I think it's just easier to define this as 0xFFC0000000000000ULL and remove the cast.

OK follow your rule, but I still prefer prior.

Quoted 5 lines
Quoted 5 lines
target/riscv/cpu_bits.h +1 -0
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
Quoted 16 lines
Quoted 5 lines
target/riscv/cpu_helper.c +0 -0
index 87dd6a6..7a540cc 100644

target/riscv/cpu_helper.c restart: · line 264

Quoted 9 lines
target/riscv/cpu_helper.c +2 -1
@@ -258,10 +258,11 @@ restart:
        }
 #if defined(TARGET_RISCV32)
        target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
        target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;

You don't have to move this shift

En… Do you want this: ?

Quoted 9 lines
#if defined(TARGET_RISCV32)
        target_ulong pte = ldl_phys(cs->as, pte_addr);
+      hwaddr ppn = pte;
#elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);

+ hwaddr ppn = (pte & ~PTE_RESERVED);
#endif
+ ppn = ppn >> PTE_PPN_SHIFT;

Yeah, it seems a little cleaner.

Alistair

Quoted 3 lines

The pte couldn’t be destroyed, just ppn ignore the RESERVED bits.

#5 lore raw
Quoted 16 lines · Alistair Francis
Quoted 15 lines · Guo Ren
Quoted 15 lines · 在 2019年9月25日,上午7:33,Alistair Francis <[email protected]> 写道:
Quoted 9 lines · On Tue, Sep 24, 2019 at 12:58 AM <[email protected]> wrote:

From: Guo Ren <[email protected]>

Highest 10 bits of PTE are reserved in riscv-privileged, ref: [1], so we need to ignore them. They can not be a part of ppn.

1: The RISC-V Instruction Set Manual, Volume II: Privileged Architecture
4.4 Sv39: Page-Based 39-bit Virtual-Memory System
4.5 Sv48: Page-Based 48-bit Virtual-Memory System

Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

Thanks for the patch!

The spec says "must be zeroed by software for forward compatibility" so I don't think it's correct for QEMU to zero out the bits.

QEMU don’t zero out the bits, QEMU just ignore the bits for ppn.

Yes, from reading the spec that seems to be the correct behaviour.

Thank you very much :)

Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

Does this fix a bug you are seeing?

Yes, because we try to reuse these bits as attributes.

That isn't really a bug then, the spec says not to do that.

Yes, I'll change the tile that "Ignore reserved bits in PTE for RV64"

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines

Changelog V2:
- Bugfix pte destroyed cause boot fail
- Change to AND with a mask instead of shifting both directions

Quoted 5 lines
Quoted 3 lines
Quoted 2 lines

The changelog shouldn't be in the commit, it should be kept under the line line below.

I just prefer to save them in commit.

Fair, but in QEMU we don't commit the change log in the commit.

Ok.

Quoted 64 lines
Quoted 54 lines
Quoted 9 lines
Quoted 7 lines

---

The change log should go here.

OK, but git am we’ll lose them.

Quoted 13 lines
Quoted 13 lines

target/riscv/cpu_bits.h | 3 +++
target/riscv/cpu_helper.c | 3 ++-

2 files changed, 5 insertions(+), 1 deletion(-)
target/riscv/cpu_bits.h +0 -0
index e998348..ae8aa0f 100644
@@ -470,6 +470,9 @@
 #define PTE_D               0x080 /* Dirty */
 #define PTE_SOFT            0x300 /* Reserved for Software */
 
Quoted 5 lines

target/riscv/cpu_bits.h line 474

Quoted 2 lines
target/riscv/cpu_bits.h +2 -0
+/* Reserved highest 10 bits in PTE */
+#define PTE_RESERVED        ((target_ulong)0x3ff << 54)

I think it's just easier to define this as 0xFFC0000000000000ULL and remove the cast.

OK follow your rule, but I still prefer prior.

Quoted 5 lines
Quoted 5 lines
target/riscv/cpu_bits.h +1 -0
+
 /* Page table PPN shift amount */
 #define PTE_PPN_SHIFT       10
 
Quoted 16 lines
Quoted 5 lines
target/riscv/cpu_helper.c +0 -0
index 87dd6a6..7a540cc 100644

target/riscv/cpu_helper.c restart: · line 264

Quoted 9 lines
target/riscv/cpu_helper.c +2 -1
@@ -258,10 +258,11 @@ restart:
         }
 #if defined(TARGET_RISCV32)
         target_ulong pte = ldl_phys(cs->as, pte_addr);
+        hwaddr ppn = pte >> PTE_PPN_SHIFT;
 #elif defined(TARGET_RISCV64)
         target_ulong pte = ldq_phys(cs->as, pte_addr);
+        hwaddr ppn = (pte & ~PTE_RESERVED) >> PTE_PPN_SHIFT;
 #endif
-        hwaddr ppn = pte >> PTE_PPN_SHIFT;

You don't have to move this shift

En… Do you want this: ?

Quoted 9 lines
#if defined(TARGET_RISCV32)
         target_ulong pte = ldl_phys(cs->as, pte_addr);
+      hwaddr ppn = pte;
#elif defined(TARGET_RISCV64)
          target_ulong pte = ldq_phys(cs->as, pte_addr);

+ hwaddr ppn = (pte & ~PTE_RESERVED);
#endif
+ ppn = ppn >> PTE_PPN_SHIFT;

Quoted 1 line

Yeah, it seems a little cleaner.

:)

Best Regards
Guo Ren