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

[Peter Zijlstra]

On Mon, Jan 23, 2023 at 05:44:31PM -0500, Steven Rostedt wrote:
> On Mon, 23 Jan 2023 21:59:18 +0100
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> 
> > 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
> > 
> 
> 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).

> I'm guessing that this is an issue because clang optimizes the static call
> site, right?

Specifically using Jcc (jne in this case) to tail-call the trampoline.

> > Teach the in-line static call text patching about this.
> > 
> > Notably, since there is no conditional-ret, in that caes patch the Jcc
> 
> 					      "in that case"

typing so hard.. :-)