Aren
/
NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

can't use fnop directly now, explain why.

This commit is contained in:
christos 2018-03-05 02:39:06 +00:00
parent 7e8fa9f58b
commit 46327c26d3
1 changed files with 14 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
sys/arch/amiga/amiga/machdep.c
View File

@ -1,4 +1,4 @@
/* $NetBSD: machdep.c,v 1.248 2017/08/15 09:51:43 maya Exp $ */
/* $NetBSD: machdep.c,v 1.249 2018/03/05 02:39:06 christos Exp $ */
/*
* Copyright (c) 1988 University of Utah.
@ -50,7 +50,7 @@
#include "empm.h"
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.248 2017/08/15 09:51:43 maya Exp $");
__KERNEL_RCSID(0, "$NetBSD: machdep.c,v 1.249 2018/03/05 02:39:06 christos Exp $");
#include <sys/param.h>
#include <sys/systm.h>
@ -721,8 +721,20 @@ initcpu(void)
/*
* in this case, we're about to switch the FPU off;
* do a FNOP to avoid stray FP traps later
*
* But we can't use fnop directly anymore as of
* gcc-6 because it passes -mno-float to the assembler
* because of -msoft-float and the assembler refuses
* to assemble the instruction; adding -Wa,-mfloat
* does not work either because the assembler then
* complains about feature being turned off and on
* so we just put in the opcode directly.
*/
#if 0
__asm("fnop");
#else
__asm(".word 0xf280,0x0000");
#endif
/* ... and mark FPU as absent for identifyfpu() */
machineid &= ~(AMIGA_FPU40|AMIGA_68882|AMIGA_68881);
}