From: H. Peter Anvin <hpa@zytor.com>
Date: Thu, 25 Dec 2008 18:39:01 +0000 (-0800)
Subject: x86: unify the implementation of FPU traps
X-Git-Tag: v2.6.29-rc1~4^2~33^2
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=a73ad3331fdbf4191cf99b83ea9ac7082b6757ba;p=platform%2Fupstream%2Fkernel-adaptation-pc.git

x86: unify the implementation of FPU traps

On 32 bits, we may suffer IRQ 13, or supposedly we might have a buggy
implementation which gives spurious trap 16.  We did not check for
this on 64 bits, but there is no reason we can't make the code the
same in both cases.  Furthermore, this is presumably rare, so do the
spurious check last, instead of first.

Signed-off-by: H. Peter Anvin <hpa@zytor.com>
---

diff --git a/arch/x86/kernel/traps.c b/arch/x86/kernel/traps.c
index f37cee7..f5a640b 100644
--- a/arch/x86/kernel/traps.c
+++ b/arch/x86/kernel/traps.c
@@ -689,12 +689,7 @@ void math_error(void __user *ip)
 	cwd = get_fpu_cwd(task);
 	swd = get_fpu_swd(task);
 
-	err = swd & ~cwd & 0x3f;
-
-#ifdef CONFIG_X86_32
-	if (!err)
-		return;
-#endif
+	err = swd & ~cwd;
 
 	if (err & 0x001) {	/* Invalid op */
 		/*
@@ -712,7 +707,9 @@ void math_error(void __user *ip)
 	} else if (err & 0x020) { /* Precision */
 		info.si_code = FPE_FLTRES;
 	} else {
-		info.si_code = __SI_FAULT|SI_KERNEL; /* WTF? */
+		/* If we're using IRQ 13, or supposedly even some trap 16
+		   implementations, it's possible we get a spurious trap... */
+		return;		/* Spurious trap, no error */
 	}
 	force_sig_info(SIGFPE, &info, task);
 }