From: Fei Wu <fei2.wu@intel.com>
Cc: Fei Wu <fei2.wu@intel.com>,
LIU Zhiwei <zhiwei_liu@linux.alibaba.com>,
Weiwei Li <liweiwei@iscas.ac.cn>,
Palmer Dabbelt <palmer@dabbelt.com>,
Alistair Francis <alistair.francis@wdc.com>,
Bin Meng <bin.meng@windriver.com>,
Daniel Henrique Barboza <dbarboza@ventanamicro.com>,
qemu-riscv@nongnu.org (open list:RISC-V TCG CPUs),
qemu-devel@nongnu.org (open list:All patches CC here)
Subject: [PATCH v2] target/riscv: reduce overhead of MSTATUS_SUM change
Date: Tue, 21 Mar 2023 21:40:09 +0800 [thread overview]
Message-ID: <20230321134009.156071-1-fei2.wu@intel.com> (raw)
Kernel needs to access user mode memory e.g. during syscalls, the window
is usually opened up for a very limited time through MSTATUS.SUM, the
overhead is too much if tlb_flush() gets called for every SUM change.
This patch saves addresses accessed when SUM=1, and flushs only these
pages when SUM changes to 0. If the buffer is not large enough to save
all the pages during SUM=1, it will fall back to tlb_flush when
necessary.
The buffer size is set to 4 since in this MSTATUS.SUM open-up window,
most of the time kernel accesses 1 or 2 pages, it's very rare to see
more than 4 pages accessed.
It's not necessary to save/restore these new added status, as
tlb_flush() is always called after restore.
Result of 'pipe 10' from unixbench boosts from 223656 to 1327407. Many
other syscalls benefit a lot from this one too.
Signed-off-by: Fei Wu <fei2.wu@intel.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Weiwei Li <liweiwei@iscas.ac.cn>
---
target/riscv/cpu.h | 4 ++++
target/riscv/cpu_helper.c | 7 +++++++
target/riscv/csr.c | 14 +++++++++++++-
3 files changed, 24 insertions(+), 1 deletion(-)
diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
index 638e47c75a..926dbce59f 100644
--- a/target/riscv/cpu.h
+++ b/target/riscv/cpu.h
@@ -383,6 +383,10 @@ struct CPUArchState {
uint64_t kvm_timer_compare;
uint64_t kvm_timer_state;
uint64_t kvm_timer_frequency;
+
+#define MAX_CACHED_SUM_U_ADDR_NUM 4
+ uint64_t sum_u_count;
+ uint64_t sum_u_addr[MAX_CACHED_SUM_U_ADDR_NUM];
};
OBJECT_DECLARE_CPU_TYPE(RISCVCPU, RISCVCPUClass, RISCV_CPU)
diff --git a/target/riscv/cpu_helper.c b/target/riscv/cpu_helper.c
index f88c503cf4..d701017a60 100644
--- a/target/riscv/cpu_helper.c
+++ b/target/riscv/cpu_helper.c
@@ -1068,6 +1068,13 @@ restart:
(access_type == MMU_DATA_STORE || (pte & PTE_D))) {
*prot |= PAGE_WRITE;
}
+ if ((pte & PTE_U) && (mode == PRV_S) &&
+ get_field(env->mstatus, MSTATUS_SUM)) {
+ if (env->sum_u_count < MAX_CACHED_SUM_U_ADDR_NUM) {
+ env->sum_u_addr[env->sum_u_count] = addr;
+ }
+ ++env->sum_u_count;
+ }
return TRANSLATE_SUCCESS;
}
}
diff --git a/target/riscv/csr.c b/target/riscv/csr.c
index ab566639e5..e7dfdc6a93 100644
--- a/target/riscv/csr.c
+++ b/target/riscv/csr.c
@@ -1246,9 +1246,21 @@ static RISCVException write_mstatus(CPURISCVState *env, int csrno,
/* flush tlb on mstatus fields that affect VM */
if ((val ^ mstatus) & (MSTATUS_MXR | MSTATUS_MPP | MSTATUS_MPV |
- MSTATUS_MPRV | MSTATUS_SUM)) {
+ MSTATUS_MPRV)) {
tlb_flush(env_cpu(env));
+ env->sum_u_count = 0;
+ } else if ((mstatus & MSTATUS_SUM) && !(val & MSTATUS_SUM)) {
+ if (env->sum_u_count > MAX_CACHED_SUM_U_ADDR_NUM) {
+ tlb_flush_by_mmuidx(env_cpu(env), 1 << PRV_S | 1 << PRV_M);
+ } else {
+ for (int i = 0; i < env->sum_u_count; ++i) {
+ tlb_flush_page_by_mmuidx(env_cpu(env), env->sum_u_addr[i],
+ 1 << PRV_S | 1 << PRV_M);
+ }
+ }
+ env->sum_u_count = 0;
}
+
mask = MSTATUS_SIE | MSTATUS_SPIE | MSTATUS_MIE | MSTATUS_MPIE |
MSTATUS_SPP | MSTATUS_MPRV | MSTATUS_SUM |
MSTATUS_MPP | MSTATUS_MXR | MSTATUS_TVM | MSTATUS_TSR |
--
2.25.1
reply other threads:[~2023-03-21 13:39 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20230321134009.156071-1-fei2.wu@intel.com \
--to=fei2.wu@intel.com \
--cc=alistair.francis@wdc.com \
--cc=bin.meng@windriver.com \
--cc=dbarboza@ventanamicro.com \
--cc=liweiwei@iscas.ac.cn \
--cc=palmer@dabbelt.com \
--cc=qemu-devel@nongnu.org \
--cc=qemu-riscv@nongnu.org \
--cc=zhiwei_liu@linux.alibaba.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
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).