Re: lib/59391: unnecessary __PIC__ conditionals clutter .S files

["Taylor R Campbell via gnats" <gnats-admin%NetBSD.org@localhost>]

The following reply was made to PR lib/59391; it has been noted by GNATS.

From: Taylor R Campbell <riastradh%NetBSD.org@localhost>
To: Valery Ushakov <uwe%stderr.spb.ru@localhost>
Cc: gnats-bugs%NetBSD.org@localhost, netbsd-bugs%NetBSD.org@localhost
Subject: Re: lib/59391: unnecessary __PIC__ conditionals clutter .S files
Date: Sat, 3 May 2025 14:58:48 +0000

 > Date: Sat, 3 May 2025 17:07:24 +0300
 > From: Valery Ushakov <uwe%stderr.spb.ru@localhost>
 >=20
 > > Sometimes entire files are conditionalized on __PIC__, such as:
 > >
 > > https://nxr.netbsd.org/xref/src/lib/libc/arch/sparc/gen/sigsetjmp.S?r=
 =3D1.7
 >=20
 > Which, in this particular case, is, IMHO, the best way to write it.
 > [...]
 > But trying to cram this into one text for static and PIC with macro
 > magic will be a bad idea.
 
 So I see, the control flow is nontrivially different, I guess because
 there's no PLT here (and no conditional branch to register)?  Fair
 enough!  (Though why not use PLT here?)
 
 But the x86 ones are just silly: we already have a macro PIC_PLT to
 conditionally expand to x@PLT or x depending on __PIC__, so
 
 #ifdef __PIC__
 	call PIC_PLT(x)
 #else
 	call x
 #endif
 
 redundantly makes this PIC code an ATM machine.