Re: [PATCH bpf-next 2/3] bpf, x86: Introduce JIT emission context

[Jiri Olsa <olsajiri@gmail.com>]

On Mon, Aug 31, 2026 at 12:54:35PM +0000, bot+bpf-ci@kernel.org wrote:
> > diff --git a/arch/x86/net/bpf_jit_comp.c b/arch/x86/net/bpf_jit_comp.c
> > index 5922bb4728aad..eff0523161c3e 100644
> > --- a/arch/x86/net/bpf_jit_comp.c
> > +++ b/arch/x86/net/bpf_jit_comp.c
> 
> [ ... ]
> 
> > @@ -947,10 +898,9 @@ static void bpf_tail_call_direct_fixup(struct bpf_prog *prog)
> >  	}
> >  }
> >
> > -static void emit_mov_imm32(u8 **pprog, bool sign_propagate,
> > +static void emit_mov_imm32(struct jit_emit_context *jit, bool sign_propagate,
> >  			   u32 dst_reg, const u32 imm32)
> >  {
> > -	u8 *prog = *pprog;
> >  	u8 b1, b2, b3;
> >
> >  	/*
> > @@ -984,14 +934,12 @@ static void emit_mov_imm32(u8 **pprog, bool sign_propagate,
> >  		EMIT1(add_1mod(0x40, dst_reg));
> >  	EMIT1_off32(add_1reg(0xB8, dst_reg), imm32);
> >  done:
> > -	*pprog = prog;
> >  }
> 
> Does this break the build? Removing the trailing '*pprog = prog;' leaves
> the 'done:' label as the last thing in the function body, with no statement
> after it. C11 and C17 require a labeled-statement to be 'identifier ':'
> statement', where a closing brace is not a statement. Labels at the end of
> a compound statement were only legalized in C23 (N2508). The kernel builds
> with '-std=gnu11' (per the top-level Makefile), so GCC will reject this with
> 'error: label at end of compound statement' and Clang with 'error: expected
> statement' when using any currently supported minimum toolchain. Only GCC
> >= 11 silently accepts it (downgrading to a pedwarn that's off without
> -Wpedantic), which is why the breakage can hide on a modern build host.
> 
> The function has two 'goto done;' statements earlier (the sign-propagate
> path and the imm32 == 0 path), so the label cannot simply be dropped. Could
> the label be given a statement, such as 'done:\n\treturn;' or 'done:\n\t;',
> or could the two gotos be replaced with 'return;' and the label removed?

yea, I overlooked that, we can simple replace goto with return

jirka