2003-08-11 14:06:38

by Dave Jones

[permalink] [raw]
Subject: [PATCH]

>From Dave Richards ([email protected])

While diagnosing an MMX/FPU problem I found a minor problem in the code
which diagnoses and generates signals for FPU exceptions. On x86 Stack
Fault Exception are a subclass of Invalid Operation. Thus, the FPU status
register will have both the SF and IF bits set when a stack fault occurs.
The code which turns FPU exceptions into signals was assuming IF would be
clear:

diff -urpN --exclude-from=/home/davej/.exclude bk-linus/arch/i386/kernel/traps.c linux-2.5/arch/i386/kernel/traps.c
--- bk-linus/arch/i386/kernel/traps.c 2003-06-26 19:49:20.000000000 +0100
+++ linux-2.5/arch/i386/kernel/traps.c 2003-06-26 20:11:29.000000000 +0100
@@ -629,9 +629,10 @@ void math_error(void *eip)
default:
break;
case 0x001: /* Invalid Op */
- case 0x040: /* Stack Fault */
- case 0x240: /* Stack Fault | Direction */
+ case 0x041: /* Stack Fault */
+ case 0x241: /* Stack Fault | Direction */
info.si_code = FPE_FLTINV;
+ /* Should we clear the SF or let user space do it ???? */
break;
case 0x002: /* Denormalize */
case 0x010: /* Underflow */