x86: explicitly use edx in const delay function.

For x86_64, we can't just use %0, as it would
generate a mul against rdx, which is not really what we
want (note the ">> 32" in x86_64 version).

Using a u64 variable with a shift in i386 generates bad code,
so the solution is to explicitly use %%edx in inline assembly
for both.

Signed-off-by: Glauber Costa <gcosta@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

2 files changed, 10 insertions, 3 deletions

diff --git a/arch/x86/lib/delay_32.c b/arch/x86/lib/delay_32.c
index bf6de05445ba..0b659a320b1e 100644
--- a/arch/x86/lib/delay_32.c
+++ b/arch/x86/lib/delay_32.c
@@ -114,7 +114,7 @@ inline void __const_udelay(unsigned long xloops)
         int d0;
 
         xloops *= 4;
-        __asm__("mull %0"
+        __asm__("mull %%edx"
                 :"=d" (xloops), "=&a" (d0)
                 :"1" (xloops), "0"
                 (cpu_data(raw_smp_processor_id()).loops_per_jiffy * (HZ/4)));
diff --git a/arch/x86/lib/delay_64.c b/arch/x86/lib/delay_64.c
index d0326d07c845..ff3dfecdb6f9 100644
--- a/arch/x86/lib/delay_64.c
+++ b/arch/x86/lib/delay_64.c
@@ -103,9 +103,16 @@ EXPORT_SYMBOL(__delay);
 
 inline void __const_udelay(unsigned long xloops)
 {
-        __delay(((xloops * HZ *
-                cpu_data(raw_smp_processor_id()).loops_per_jiffy) >> 32) + 1);
+        int d0;
+        xloops *= 4;
+        __asm__("mull %%edx"
+                :"=d" (xloops), "=&a" (d0)
+                :"1" (xloops), "0"
+                (cpu_data(raw_smp_processor_id()).loops_per_jiffy * (HZ/4)));
+
+        __delay(++xloops);
 }
+
 EXPORT_SYMBOL(__const_udelay);
 
 void __udelay(unsigned long usecs)