* [PATCH] target/riscv: Flush the TLB entry for the specified address
@ 2025-09-30 9:48 wang.yechao255
2025-09-30 10:04 ` Philippe Mathieu-Daudé
0 siblings, 1 reply; 3+ messages in thread
From: wang.yechao255 @ 2025-09-30 9:48 UTC (permalink / raw)
To: palmer, alistair.francis, liwei1518, dbarboza, zhiwei_liu
Cc: qemu-riscv, qemu-devel
From: yechao-w <wang.yechao255@zte.com.cn>
When the guest executes sfence/svinval with an address parameter,
invalidating only the specific TLB entry for that address rather
than the entire TLB enhances TCG performance.
Signed-off-by: yechao-w <wang.yechao255@zte.com.cn>
---
target/riscv/helper.h | 2 +-
target/riscv/insn_trans/trans_privileged.c.inc | 4 +++-
target/riscv/insn_trans/trans_svinval.c.inc | 4 +++-
target/riscv/op_helper.c | 8 ++++++--
4 files changed, 13 insertions(+), 5 deletions(-)
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index f712b1c368..de3757fa75 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -135,7 +135,7 @@ DEF_HELPER_1(mnret, tl, env)
DEF_HELPER_1(ctr_clear, void, env)
DEF_HELPER_1(wfi, void, env)
DEF_HELPER_1(wrs_nto, void, env)
-DEF_HELPER_1(tlb_flush, void, env)
+DEF_HELPER_2(tlb_flush, void, env, tl)
DEF_HELPER_1(tlb_flush_all, void, env)
DEF_HELPER_4(ctr_add_entry, void, env, tl, tl, tl)
/* Native Debug */
diff --git a/target/riscv/insn_trans/trans_privileged.c.inc b/target/riscv/insn_trans/trans_privileged.c.inc
index 8a62b4cfcd..b6865f51b2 100644
--- a/target/riscv/insn_trans/trans_privileged.c.inc
+++ b/target/riscv/insn_trans/trans_privileged.c.inc
@@ -154,8 +154,10 @@ static bool trans_wfi(DisasContext *ctx, arg_wfi *a)
static bool trans_sfence_vma(DisasContext *ctx, arg_sfence_vma *a)
{
#ifndef CONFIG_USER_ONLY
+ TCGv addr = get_gpr(ctx, a->rs1, EXT_NONE);
+
decode_save_opc(ctx, 0);
- gen_helper_tlb_flush(tcg_env);
+ gen_helper_tlb_flush(tcg_env, addr);
return true;
#endif
return false;
diff --git a/target/riscv/insn_trans/trans_svinval.c.inc b/target/riscv/insn_trans/trans_svinval.c.inc
index a06c3b214f..1b1b932ac6 100644
--- a/target/riscv/insn_trans/trans_svinval.c.inc
+++ b/target/riscv/insn_trans/trans_svinval.c.inc
@@ -24,12 +24,14 @@
static bool trans_sinval_vma(DisasContext *ctx, arg_sinval_vma *a)
{
+ TCGv addr = get_gpr(ctx, a->rs1, EXT_NONE);
+
REQUIRE_SVINVAL(ctx);
/* Do the same as sfence.vma currently */
REQUIRE_EXT(ctx, RVS);
#ifndef CONFIG_USER_ONLY
decode_save_opc(ctx, 0);
- gen_helper_tlb_flush(tcg_env);
+ gen_helper_tlb_flush(tcg_env, addr);
return true;
#endif
return false;
diff --git a/target/riscv/op_helper.c b/target/riscv/op_helper.c
index 110292e84d..0fe5fcb3ac 100644
--- a/target/riscv/op_helper.c
+++ b/target/riscv/op_helper.c
@@ -555,7 +555,7 @@ void helper_wrs_nto(CPURISCVState *env)
}
}
-void helper_tlb_flush(CPURISCVState *env)
+void helper_tlb_flush(CPURISCVState *env, target_ulong addr)
{
CPUState *cs = env_cpu(env);
if (!env->virt_enabled &&
@@ -566,7 +566,11 @@ void helper_tlb_flush(CPURISCVState *env)
(env->priv == PRV_U || get_field(env->hstatus, HSTATUS_VTVM))) {
riscv_raise_exception(env, RISCV_EXCP_VIRT_INSTRUCTION_FAULT, GETPC());
} else {
- tlb_flush(cs);
+ if (addr) {
+ tlb_flush_page(cs, addr);
+ } else {
+ tlb_flush(cs);
+ }
}
}
--
2.27.0
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] target/riscv: Flush the TLB entry for the specified address
2025-09-30 9:48 [PATCH] target/riscv: Flush the TLB entry for the specified address wang.yechao255
@ 2025-09-30 10:04 ` Philippe Mathieu-Daudé
2025-10-09 2:49 ` wang.yechao255
0 siblings, 1 reply; 3+ messages in thread
From: Philippe Mathieu-Daudé @ 2025-09-30 10:04 UTC (permalink / raw)
To: wang.yechao255, palmer, alistair.francis, liwei1518, dbarboza,
zhiwei_liu
Cc: qemu-riscv, qemu-devel
Hi,
On 30/9/25 11:48, wang.yechao255@zte.com.cn wrote:
> From: yechao-w <wang.yechao255@zte.com.cn>
>
> When the guest executes sfence/svinval with an address parameter,
> invalidating only the specific TLB entry for that address rather
> than the entire TLB enhances TCG performance.
>
> Signed-off-by: yechao-w <wang.yechao255@zte.com.cn>
> ---
> target/riscv/helper.h | 2 +-
> target/riscv/insn_trans/trans_privileged.c.inc | 4 +++-
> target/riscv/insn_trans/trans_svinval.c.inc | 4 +++-
> target/riscv/op_helper.c | 8 ++++++--
> 4 files changed, 13 insertions(+), 5 deletions(-)
> diff --git a/target/riscv/op_helper.c b/target/riscv/op_helper.c
> index 110292e84d..0fe5fcb3ac 100644
> --- a/target/riscv/op_helper.c
> +++ b/target/riscv/op_helper.c
> @@ -555,7 +555,7 @@ void helper_wrs_nto(CPURISCVState *env)
> }
> }
>
> -void helper_tlb_flush(CPURISCVState *env)
> +void helper_tlb_flush(CPURISCVState *env, target_ulong addr)
> {
> CPUState *cs = env_cpu(env);
> if (!env->virt_enabled &&
> @@ -566,7 +566,11 @@ void helper_tlb_flush(CPURISCVState *env)
> (env->priv == PRV_U || get_field(env->hstatus, HSTATUS_VTVM))) {
> riscv_raise_exception(env, RISCV_EXCP_VIRT_INSTRUCTION_FAULT, GETPC());
> } else {
> - tlb_flush(cs);
> + if (addr) {
> + tlb_flush_page(cs, addr);
I suspect tlb_flush_page(cs, addr) is the right thing to do,
even for addr=0.
Can you point at the doc?
> + } else {
> + tlb_flush(cs);
> + }
> }
> }
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] target/riscv: Flush the TLB entry for the specified address
2025-09-30 10:04 ` Philippe Mathieu-Daudé
@ 2025-10-09 2:49 ` wang.yechao255
0 siblings, 0 replies; 3+ messages in thread
From: wang.yechao255 @ 2025-10-09 2:49 UTC (permalink / raw)
To: philmd
Cc: palmer, alistair.francis, liwei1518, dbarboza, zhiwei_liu,
qemu-riscv, qemu-devel
[-- Attachment #1.1.1: Type: text/plain, Size: 895 bytes --]
Hi Phil,
> > -void helper_tlb_flush(CPURISCVState *env)> > +void helper_tlb_flush(CPURISCVState *env, target_ulong addr)> > {> > CPUState *cs = env_cpu(env);> > if (!env->virt_enabled &&> > @@ -566,7 +566,11 @@ void helper_tlb_flush(CPURISCVState *env)> > (env->priv == PRV_U || get_field(env->hstatus, HSTATUS_VTVM))) {> > riscv_raise_exception(env, RISCV_EXCP_VIRT_INSTRUCTION_FAULT, GETPC());> > } else {> > - tlb_flush(cs);> > + if (addr) {> > + tlb_flush_page(cs, addr);> I suspect tlb_flush_page(cs, addr) is the right thing to do,> even for addr=0.> > Can you point at the doc?
When addr=0, the current tlb_flush_page(cs, addr) does not flush all TLBs.
Should tlb_flush_page(cs, addr) be extended to flush all TLBs when addr=0?
By the way, which document are you referring to here?
---
Best wishes
Yechao
[-- Attachment #1.1.2: Type: text/html , Size: 11715 bytes --]
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2025-10-09 3:32 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-09-30 9:48 [PATCH] target/riscv: Flush the TLB entry for the specified address wang.yechao255
2025-09-30 10:04 ` Philippe Mathieu-Daudé
2025-10-09 2:49 ` wang.yechao255
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).