x86-32, fpu: Fix DNA exception during check_fpu()
Suresh Siddha [Fri, 1 Jul 2011 00:19:32 +0000 (17:19 -0700)]
Before check_fpu() is called, we have cr0.TS bit set and hence the floating
point code to check the FDIV bug was generating a DNA exception.

Use kernel_fpu_begin()/kernel_fpu_end() around the floating point
code to avoid this unnecessary device not available exception during
boot.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Link: http://lkml.kernel.org/r/1309479572.2665.1372.camel@sbsiddha-MOBL3.sc.intel.com
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>

arch/x86/kernel/cpu/bugs.c

index 525514c..46674fb 100644 (file)
@@ -62,6 +62,8 @@ static void __init check_fpu(void)
                return;
        }
 
+       kernel_fpu_begin();
+
        /*
         * trap_init() enabled FXSR and company _before_ testing for FP
         * problems here.
@@ -80,6 +82,8 @@ static void __init check_fpu(void)
                : "=m" (*&fdiv_bug)
                : "m" (*&x), "m" (*&y));
 
+       kernel_fpu_end();
+
        boot_cpu_data.fdiv_bug = fdiv_bug;
        if (boot_cpu_data.fdiv_bug)
                printk(KERN_WARNING "Hmm, FPU with FDIV bug.\n");