Re: toolchain/48630: sun4v support causes sparc/sparc64 build failure on netbsd-5/i386 host

[christos%zoulas.com@localhost (Christos Zoulas)]

On Mar 2,  2:00pm, martin%duskware.de@localhost (Martin Husemann) wrote:
-- Subject: Re: toolchain/48630: sun4v support causes sparc/sparc64 build fai

| The following reply was made to PR toolchain/48630; it has been noted by 
GNATS.
| 
| From: Martin Husemann <martin%duskware.de@localhost>
| To: gnats-bugs%NetBSD.org@localhost
| Cc: palle%NetBSD.org@localhost
| Subject: Re: toolchain/48630: sun4v support causes sparc/sparc64 build 
failure on netbsd-5/i386 host
| Date: Sun, 2 Mar 2014 14:56:08 +0100
| 
|  On Sun, Mar 02, 2014 at 01:40:00PM +0000, jdbaker%mylinuxisp.com@localhost 
wrote:
|  > As the nightly builds are succeeding and the autobuild hosts are
|  > running NetBSD/amd64-6.1.x, I would assume that tools are being mis-
|  > configured on a NetBSD-5 host.
|  
|  This is a nice one. The problem is not netbsd-6 as build host vs. netbsd-5,
|  but a 64bit build host vs. a 32bit one.
|  
|  Gennassym creates the constant in question as:
|  
|  assym.h:#define A_SUN4V_TLB_TSB_LOCK 1152921504606846976
|  
|  while the original definition is:
|  
|  pte.h:#define SUN4V_TLB_TSB_LOCK        0x1000000000000000UL
|  
|  
|  I don't remember right now wether there was some special asm syntax to make
|  gas on 32bit hosts understand 64bit constants (this all sounds like a
|  gas bug to me).

On 32 bits you need ull...

christos