qemu-riscv
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v5 5/5] target/riscv: add support for svpbmt extension


From: Weiwei Li
Subject: Re: [PATCH v5 5/5] target/riscv: add support for svpbmt extension
Date: Tue, 18 Jan 2022 19:21:32 +0800
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.14.0


在 2022/1/18 下午7:04, Anup Patel 写道:
On Tue, Jan 18, 2022 at 2:40 PM Weiwei Li <liweiwei@iscas.ac.cn> wrote:

在 2022/1/18 上午11:35, Anup Patel 写道:
On Tue, Jan 18, 2022 at 6:47 AM Weiwei Li <liweiwei@iscas.ac.cn> wrote:
- add PTE_PBMT bits: It uses two PTE bits, but otherwise has no effect on QEMU, 
since QEMU is sequentially consistent and doesn't model PMAs currently
- add PTE_PBMT bit check for inner PTE

Signed-off-by: Weiwei Li <liweiwei@iscas.ac.cn>
Signed-off-by: Junqiang Wang <wangjunqiang@iscas.ac.cn>
Cc: Heiko Stuebner <heiko@sntech.de>
Cc: Anup Patel <anup@brainfault.org>
---
   target/riscv/cpu.c        | 1 +
   target/riscv/cpu.h        | 1 +
   target/riscv/cpu_bits.h   | 2 ++
   target/riscv/cpu_helper.c | 4 +++-
   4 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
index 45ac98e06b..4f82bd00a3 100644
--- a/target/riscv/cpu.c
+++ b/target/riscv/cpu.c
@@ -670,6 +670,7 @@ static Property riscv_cpu_properties[] = {

       DEFINE_PROP_BOOL("svinval", RISCVCPU, cfg.ext_svinval, false),
       DEFINE_PROP_BOOL("svnapot", RISCVCPU, cfg.ext_svnapot, false),
+    DEFINE_PROP_BOOL("svpbmt", RISCVCPU, cfg.ext_svpbmt, false),

       DEFINE_PROP_BOOL("zba", RISCVCPU, cfg.ext_zba, true),
       DEFINE_PROP_BOOL("zbb", RISCVCPU, cfg.ext_zbb, true),
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index c3d1845ca1..53f314c752 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -329,6 +329,7 @@ struct RISCVCPU {
           bool ext_icsr;
           bool ext_svinval;
           bool ext_svnapot;
+        bool ext_svpbmt;
           bool ext_zfh;
           bool ext_zfhmin;

diff --git a/target/riscv/cpu_bits.h b/target/riscv/cpu_bits.h
index 5501e9698b..24b7eb2b1f 100644
--- a/target/riscv/cpu_bits.h
+++ b/target/riscv/cpu_bits.h
@@ -486,7 +486,9 @@ typedef enum {
   #define PTE_A               0x040 /* Accessed */
   #define PTE_D               0x080 /* Dirty */
   #define PTE_SOFT            0x300 /* Reserved for Software */
+#define PTE_PBMT            0x6000000000000000 /* Page-based memory types */
   #define PTE_N               0x8000000000000000 /* NAPOT translation */
+#define PTE_ATTR            (PTE_N | PTE_PBMT) /* All attributes bits */

   /* Page table PPN shift amount */
   #define PTE_PPN_SHIFT       10
diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
index c276760c7f..9fffaccffb 100644
--- a/target/riscv/cpu_helper.c
+++ b/target/riscv/cpu_helper.c
@@ -625,9 +625,11 @@ restart:
           if (!(pte & PTE_V)) {
               /* Invalid PTE */
               return TRANSLATE_FAIL;
+        } else if (!cpu->cfg.ext_svpbmt && (pte & (target_ulong)PTE_PBMT)) {
Rather than, type-casting defines here you can simply define
ULL constants. E.g.
#define PTE_PBMT            0x6000000000000000ULL
Sorry, I'm wonder why add ULL can replace the function of type-casting.

The type-casting here is to compatible with RV32 for possible strict
type check warnings since pte is 32 bits and PTE_PBMT is 64 bits in RV32.
If adding ULL does not help for RV32 target then no need to change.

OK. Thanks a lot.

Regards,

Weiwei Li

If I add ULL in PTE_PBMT, it seems have no change to PTE_PBMT. It's
still 64 bits in RV32.
Reviewed-by: Anup Patel <anup@brainfault.org>

Regards,
Anup

Regards,

Weiwei Li

+            return TRANSLATE_FAIL;
           } else if (!(pte & (PTE_R | PTE_W | PTE_X))) {
               /* Inner PTE, continue walking */
-            if (pte & (target_ulong)(PTE_D | PTE_A | PTE_U | PTE_N)) {
+            if (pte & (target_ulong)(PTE_D | PTE_A | PTE_U | PTE_ATTR)) {
                   return TRANSLATE_FAIL;
               }
               base = ppn << PGSHIFT;
--
2.17.1

Regards,
Anup





reply via email to

[Prev in Thread] Current Thread [Next in Thread]