[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH v2 10/23] target/i386: Prefer fast cpu_env() over slower CPU QOM cast macro
- To: Zhao Liu <zhao1.liu@xxxxxxxxx>
- From: Philippe Mathieu-Daudé <philmd@xxxxxxxxxx>
- Date: Sun, 28 Jan 2024 17:16:19 +0100
- Cc: qemu-devel@xxxxxxxxxx, qemu-arm@xxxxxxxxxx, Thomas Huth <thuth@xxxxxxxxxx>, qemu-s390x@xxxxxxxxxx, qemu-riscv@xxxxxxxxxx, Eduardo Habkost <eduardo@xxxxxxxxxxx>, kvm@xxxxxxxxxxxxxxx, qemu-ppc@xxxxxxxxxx, Richard Henderson <richard.henderson@xxxxxxxxxx>, Vladimir Sementsov-Ogievskiy <vsementsov@xxxxxxxxxxxxxx>, Paolo Bonzini <pbonzini@xxxxxxxxxx>, "Michael S. Tsirkin" <mst@xxxxxxxxxx>, Marcel Apfelbaum <marcel.apfelbaum@xxxxxxxxx>, Stefano Stabellini <sstabellini@xxxxxxxxxx>, Anthony Perard <anthony.perard@xxxxxxxxxx>, Paul Durrant <paul@xxxxxxx>, Cameron Esfahani <dirty@xxxxxxxxx>, Roman Bolshakov <rbolshakov@xxxxxxx>, Marcelo Tosatti <mtosatti@xxxxxxxxxx>, David Woodhouse <dwmw2@xxxxxxxxxxxxx>, xen-devel@xxxxxxxxxxxxxxxxxxxx
- Delivery-date: Sun, 28 Jan 2024 16:16:43 +0000
- List-id: Xen developer discussion <xen-devel.lists.xenproject.org>
On 27/1/24 13:21, Zhao Liu wrote:
Hi Philippe,
On Fri, Jan 26, 2024 at 11:03:52PM +0100, Philippe Mathieu-Daudé wrote:
Date: Fri, 26 Jan 2024 23:03:52 +0100
From: Philippe Mathieu-Daudé <philmd@xxxxxxxxxx>
Subject: [PATCH v2 10/23] target/i386: Prefer fast cpu_env() over slower
CPU QOM cast macro
X-Mailer: git-send-email 2.41.0
Mechanical patch produced running the command documented
in scripts/coccinelle/cpu_env.cocci_template header.
Signed-off-by: Philippe Mathieu-Daudé <philmd@xxxxxxxxxx>
---
target/i386/hvf/vmx.h | 13 +++-------
hw/i386/vmmouse.c | 6 ++---
hw/i386/xen/xen-hvm.c | 3 +--
target/i386/arch_memory_mapping.c | 3 +--
target/i386/cpu-dump.c | 3 +--
target/i386/cpu.c | 37 +++++++++------------------
target/i386/helper.c | 39 ++++++++---------------------
target/i386/hvf/hvf.c | 8 ++----
target/i386/hvf/x86.c | 4 +--
target/i386/hvf/x86_emu.c | 6 ++---
target/i386/hvf/x86_task.c | 10 +++-----
target/i386/hvf/x86hvf.c | 6 ++---
target/i386/kvm/kvm.c | 6 ++---
target/i386/kvm/xen-emu.c | 32 ++++++++---------------
target/i386/tcg/sysemu/bpt_helper.c | 3 +--
target/i386/tcg/tcg-cpu.c | 14 +++--------
target/i386/tcg/user/excp_helper.c | 3 +--
target/i386/tcg/user/seg_helper.c | 3 +--
18 files changed, 59 insertions(+), 140 deletions(-)
[snip]
diff --git a/target/i386/hvf/x86hvf.c b/target/i386/hvf/x86hvf.c
index 3b1ef5f49a..1e7fd587fe 100644
--- a/target/i386/hvf/x86hvf.c
+++ b/target/i386/hvf/x86hvf.c
@@ -238,8 +238,7 @@ void hvf_get_msrs(CPUState *cs)
int hvf_put_registers(CPUState *cs)
{
- X86CPU *x86cpu = X86_CPU(cs);
- CPUX86State *env = &x86cpu->env;
+ CPUX86State *env = cpu_env(cs);
wreg(cs->accel->fd, HV_X86_RAX, env->regs[R_EAX]);
wreg(cs->accel->fd, HV_X86_RBX, env->regs[R_EBX]);
@@ -282,8 +281,7 @@ int hvf_put_registers(CPUState *cs)
int hvf_get_registers(CPUState *cs)
{
- X86CPU *x86cpu = X86_CPU(cs);
- CPUX86State *env = &x86cpu->env;
+ CPUX86State *env = cpu_env(cs);
env->regs[R_EAX] = rreg(cs->accel->fd, HV_X86_RAX);
env->regs[R_EBX] = rreg(cs->accel->fd, HV_X86_RBX);
In this file, there's another corner case:
diff --git a/target/i386/hvf/x86hvf.c b/target/i386/hvf/x86hvf.c
index 3b1ef5f49a8a..9a145aa5aa4f 100644
--- a/target/i386/hvf/x86hvf.c
+++ b/target/i386/hvf/x86hvf.c
@@ -342,8 +342,7 @@ void vmx_clear_int_window_exiting(CPUState *cs)
bool hvf_inject_interrupts(CPUState *cs)
{
- X86CPU *x86cpu = X86_CPU(cs);
- CPUX86State *env = &x86cpu->env;
+ CPUX86State *env = cpu_env(cs);
uint8_t vector;
uint64_t intr_type;
@@ -408,7 +407,7 @@ bool hvf_inject_interrupts(CPUState *cs)
if (!(env->hflags & HF_INHIBIT_IRQ_MASK) &&
(cs->interrupt_request & CPU_INTERRUPT_HARD) &&
(env->eflags & IF_MASK) && !(info & VMCS_INTR_VALID)) {
- int line = cpu_get_pic_interrupt(&x86cpu->env);
+ int line = cpu_get_pic_interrupt(env);
cs->interrupt_request &= ~CPU_INTERRUPT_HARD;
if (line >= 0) {
wvmcs(cs->accel->fd, VMCS_ENTRY_INTR_INFO, line |
For this special case, I'm not sure if the script can cover it as well,
otherwise maybe it's OK to be cleaned up manually ;-).
BTW I forgot to mention I had to skip target/i386/tcg/translate.c
(7100 LoC) because it is too complex for Coccinelle.
|