qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Philippe Mathieu-Daudé" <philmd@linaro.org>
To: Paolo Bonzini <pbonzini@redhat.com>
Cc: qemu-devel@nongnu.org, Peter Maydell <peter.maydell@linaro.org>,
	Eduardo Habkost <eduardo@habkost.net>,
	Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>,
	Richard Henderson <richard.henderson@linaro.org>,
	qemu-s390x@nongnu.org, qemu-ppc@nongnu.org,
	qemu-riscv@nongnu.org, qemu-arm@nongnu.org
Subject: Re: [PATCH 1/2] scripts/coccinelle: Add cpu_env.cocci_template script
Date: Fri, 26 Jan 2024 13:34:56 +0100	[thread overview]
Message-ID: <d739d778-c84e-4c72-bf55-21d4aa562d96@linaro.org> (raw)
In-Reply-To: <CABgObfZuPwzVdPjgeLOOtjmNqUMGcK=ExsvVQOzLqnUABmmX-A@mail.gmail.com>

On 26/1/24 12:24, Paolo Bonzini wrote:
> On Fri, Jan 26, 2024 at 11:38 AM Philippe Mathieu-Daudé
> <philmd@linaro.org> wrote:
>>
>> On 25/1/24 17:56, Philippe Mathieu-Daudé wrote:
>>> Add a Coccinelle script to convert the following slow path
>>> (due to the QOM cast macro):
>>>
>>>     &ARCH_CPU(..)->env
>>>
>>> to the following fast path:
>>>
>>>     cpu_env(..)
>>>
>>> Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
>>> ---
>>>    MAINTAINERS                               |  1 +
>>>    scripts/coccinelle/cpu_env.cocci_template | 60 +++++++++++++++++++++++
>>>    2 files changed, 61 insertions(+)
>>>    create mode 100644 scripts/coccinelle/cpu_env.cocci_template
>>
>>
>>> diff --git a/scripts/coccinelle/cpu_env.cocci_template b/scripts/coccinelle/cpu_env.cocci_template
>>> new file mode 100644
>>> index 0000000000..53aa3a1fea
>>> --- /dev/null
>>> +++ b/scripts/coccinelle/cpu_env.cocci_template
>>> @@ -0,0 +1,60 @@
>>> +/*
>>> +
>>> + Convert &ARCH_CPU(..)->env to use cpu_env(..).
>>> +
>>> + Rationale: ARCH_CPU() might be slow, being a QOM cast macro.
>>> +            cpu_env() is its fast equivalent.
>>> +
>>> + SPDX-License-Identifier: GPL-2.0-or-later
>>> + SPDX-FileCopyrightText: Linaro Ltd 2024
>>> + SPDX-FileContributor: Philippe Mathieu-Daudé
>>> +
>>> + Usage as of v8.2.0:
>>> +
>>> + $ for targetdir in target/*; do test -d $targetdir || continue; \
>>> +       export target=${targetdir:7}; \
>>> +       sed \
>>> +           -e "s/__CPUArchState__/$( \
>>> +               git grep -h --no-line-number '@env: #CPU.*State' \
>>> +                   target/$target/cpu.h \
>>> +               | sed -n -e 's/.*\(CPU.*State\).\?/\1/p')/g" \
>>> +           -e "s/__ARCHCPU__/$( \
>>> +               git grep -h --no-line-number OBJECT_DECLARE_CPU_TYPE.*CPU \
>>> +                   target/$target/cpu-qom.h \
>>> +               | sed -n -e 's/.*(\(.*\), .*, .*)/\1/p')/g" \
>>> +           -e "s/__ARCH_CPU__/$( \
>>> +               git grep -h --no-line-number OBJECT_DECLARE_CPU_TYPE.*CPU \
>>> +                   target/$target/cpu-qom.h \
>>> +               | sed -n -e 's/.*(.*, .*, \(.*\))/\1/p')/g" \
>>> +       < scripts/coccinelle/cpu_env.cocci_template \
>>> +       > $TMPDIR/cpu_env_$target.cocci; \
>>> +       for dir in hw target/$target; do \
>>> +           spatch --macro-file scripts/cocci-macro-file.h \
>>> +                  --sp-file $TMPDIR/cpu_env_$target.cocci \
>>> +                  --keep-comments \
>>> +                  --dir $dir \
>>> +                  --in-place; \
>>> +       done; \
>>> +   done
>>> +
>>> +*/
>>> +
>>> +@ CPUState_arg_used @
>>> +CPUState *cs;
>>> +identifier cpu;
>>> +identifier env;
>>> +@@
>>> +-   __ARCHCPU__ *cpu = __ARCH_CPU__(cs);
>>
>> Here we remove ARCH_CPU(), ...
>>
>>> +-   __CPUArchState__ *env = &cpu->env;
>>> ++   __CPUArchState__ *env = cpu_env(cs);
>>> +    ... when != cpu
>>> +
>>> +@ depends on never CPUState_arg_used @
>>> +identifier obj;
>>> +identifier cpu;
>>> +identifier env;
>>> +@@
>>> +-   __ARCHCPU__ *cpu = __ARCH_CPU__(obj);
>>> +-   __CPUArchState__ *env = &cpu->env;
>>> ++   __CPUArchState__ *env = cpu_env(CPU(obj));
>>
>> ... but here we just change it by a CPU() QOM call.
>> So this 2nd change is just style cleanup.
> 
> Can you also add a hunk that is
> 
> CPUState *cs;
> @@
> -  CPU(cs)
> + cs
> 
> to clean up on the second?  cpu_env(CPU(current_cpu)) is suboptimal
> and also a bit ugly.

These case should be cleaned because this is already a CPUState*:

+    CPUX86State *env = cpu_env(CPU(current_cpu));

+    CPUPPCState *env = cpu_env(CPU(first_cpu));

But these (instance_init and QOM visitors) can't, the argument
isn't a CPUState*:

  static void ev4_cpu_initfn(Object *obj)
  {
-    AlphaCPU *cpu = ALPHA_CPU(obj);
-    CPUAlphaState *env = &cpu->env;
+    CPUAlphaState *env = cpu_env(CPU(obj));

@@ -5186,8 +5179,7 @@ static char *x86_cpuid_get_vendor(Object *obj, 
Error **errp)
  static void x86_cpuid_set_vendor(Object *obj, const char *value,
                                   Error **errp)
  {
-    X86CPU *cpu = X86_CPU(obj);
-    CPUX86State *env = &cpu->env;
+    CPUX86State *env = cpu_env(CPU(obj));

That said, these visitors take a Object* param because they implement
the generic QOM visitor API, but we know the visitor are registered
on classes/objects implementing CPUState, so QOM cast macro is
redundant.


  reply	other threads:[~2024-01-26 12:35 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-01-25 16:56 [PATCH 0/2] hw, target: Prefer fast cpu_env() over slower CPU QOM cast macro Philippe Mathieu-Daudé
2024-01-25 16:56 ` [PATCH 1/2] scripts/coccinelle: Add cpu_env.cocci_template script Philippe Mathieu-Daudé
2024-01-26 10:38   ` Philippe Mathieu-Daudé
2024-01-26 11:24     ` Paolo Bonzini
2024-01-26 12:34       ` Philippe Mathieu-Daudé [this message]
2024-01-26 13:50         ` Philippe Mathieu-Daudé
2024-01-25 16:56 ` [PATCH 2/2] bulk: Prefer fast cpu_env() over slower CPU QOM cast macro Philippe Mathieu-Daudé
2024-01-26 17:09   ` Thomas Huth
2024-01-26 21:11     ` Philippe Mathieu-Daudé

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=d739d778-c84e-4c72-bf55-21d4aa562d96@linaro.org \
    --to=philmd@linaro.org \
    --cc=eduardo@habkost.net \
    --cc=pbonzini@redhat.com \
    --cc=peter.maydell@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-ppc@nongnu.org \
    --cc=qemu-riscv@nongnu.org \
    --cc=qemu-s390x@nongnu.org \
    --cc=richard.henderson@linaro.org \
    --cc=vsementsov@virtuozzo.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).