Re: [PATCH v1.1 3/3] x86/static_call: Add support for Jcc tail-calls

From: Nick Desaulniers <ndesaulniers@google.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>,
	mingo@kernel.org, tglx@linutronix.de, bp@alien8.de,
	dave.hansen@linux.intel.com, x86@kernel.org, hpa@zytor.com,
	jpoimboe@kernel.org, jbaron@akamai.com, ardb@kernel.org,
	linux-kernel@vger.kernel.org, erhard_f@mailbox.org,
	mhiramat@kernel.org, sandipan.das@amd.com
Subject: Re: [PATCH v1.1 3/3] x86/static_call: Add support for Jcc tail-calls
Date: Thu, 26 Jan 2023 10:14:49 -0800	[thread overview]
Message-ID: <CAKwvOdmE4bHdOr4TCm0EKrqKjMWk52NAkEOOd9JgsbUFaknOrw@mail.gmail.com> (raw)
In-Reply-To: <Y9Kdg9QjHkr9G5b5@hirez.programming.kicks-ass.net>

On Thu, Jan 26, 2023 at 7:34 AM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Tue, Jan 24, 2023 at 10:07:53AM -0500, Steven Rostedt wrote:
> > On Tue, 24 Jan 2023 14:06:49 +0100
> > Peter Zijlstra <peterz@infradead.org> wrote:
> >
> > > > Just to confirm, as it's not clear if this is the static call site or one
> > > > of the functions that is being called.
> > >
> > > Ah, you've not looked at enough asm then? ;-) Yes this is the static
> > > call site, see the __SCT_ target (instruction at 0x35d).
> >
> > Yeah, could you specify it a bit more in the change log such that those
> > looking back at this don't have to have that requirement of staring at
> > enough asm ;-)
>
> How's this then?
>
> ---
> Subject: x86/static_call: Add support for Jcc tail-calls
> From: Peter Zijlstra <peterz@infradead.org>
> Date: Fri Jan 20 16:40:33 CET 2023
>
> Clang likes to create conditional tail calls like:
>
> 0000000000000350 <amd_pmu_add_event>:
> 350:       0f 1f 44 00 00          nopl   0x0(%rax,%rax,1) 351: R_X86_64_NONE      __fentry__-0x4
> 355:       48 83 bf 20 01 00 00 00         cmpq   $0x0,0x120(%rdi)
> 35d:       0f 85 00 00 00 00       jne    363 <amd_pmu_add_event+0x13>     35f: R_X86_64_PLT32     __SCT__amd_pmu_branch_add-0x4
> 363:       e9 00 00 00 00          jmp    368 <amd_pmu_add_event+0x18>     364: R_X86_64_PLT32     __x86_return_thunk-0x4

If it helps reduce the line width in the commit message to focus on
just the instructions, lately I've been using
`--no-addresses --no-show-raw-insn` flags to llvm-objdump.

In my ~/.zshrc, I have this function:
dis () {
  file=$1
  func=$2
  llvm-objdump -Dr --no-addresses --no-show-raw-insn
--disassemble-symbols=$func $file
}
Which let me run `dis vmlinux amd_pmu_add_event` or `dis foo.o func`.
Sometimes I don't want `-r` or `--no-show-raw-insn or
`--no-addresses`, but most of the time I do.

-- 
Thanks,
~Nick Desaulniers