Re: [Nasm-devel] Fwd: Re: NASM Bug 3392348

[Cyrill G. <gor...@gm...>]

Peter?

On Tue, May 10, 2016 at 03:08:31PM +0300, Zenith432 wrote:
> -------- Forwarded Message --------
> Subject: Re: NASM Bug 3392348
> Date: Tue, 10 May 2016 03:29:58 -0700
> From: H. Peter Anvin <hp...@zy...>
> To: Zenith432 <zen...@us...>
> 
> On May 10, 2016 2:25:17 AM PDT, Zenith432 <zen...@us...>
> wrote:
> > I'm sending this by email, because the nasm bugzilla does not provide
> > option for reopening a bug, and I'd rather not open a new one.
> > 
> > The commit has a modification that doesn't work.
> > 
> > =====
> > --- output/outmacho.c
> > +++ output/outmacho.c
> > @@ -465,7 +465,7 @@ static int64_t add_reloc(struct section *sect,
> > int32_t section,
> > 	    r->snum = raa_read(extsyms, section);
> > 	    if (reltype == RL_BRANCH)
> > 		r->type = X86_64_RELOC_BRANCH;
> > -	    else if (r->type == GENERIC_RELOC_VANILLA)
> > +	    else if (reltype == GENERIC_RELOC_VANILLA)
> > 		adjust = -sect->size;
> > 	} else {
> > 	    /* local */
> > =====
> > 
> > At that point in the code, reltype has one of the values
> > RL_REL (1) for i386 or x86_64
> > RL_BRANCH (3) for x86_64
> > 
> > But r->type has one of the values
> > GENERIC_RELOC_VANILLA (0) for i386
> > X86_64_RELOC_SIGNED (1) for x86_64
> > X86_64_RELOC_BRANCH (2) for x86_64
> > 
> > so comparing reltype to 0 nullifies the fix.
> > The fix in negative adjustment should be for external, r->pcrel,
> > r->type == GENERIC_RELOC_VANILLA (i386 only).
> > 
> > There is also a negative adjustment for local, but that was ok before.
> > 
> > Thanks and Regards.