As of today's "fix up gcc 3.3 bits" patch [1], x86 math emulation
is now even more broken, since this latest patch has added some
mismatched quotes while still failing to address the actual
problems preventing this code from compiling with gcc 3.3.
This patch, first sent to you on Jun 21st, fixes the missing
semicolons and missing quotes in the x86 math-emu code, allowing
it to compile with gcc 3.3. I have updated things to also fix
the mismatched quotes that were added today. Unlike the patch
you applied earlier today, my patch is actually tested...
Please apply,
-Erik
[1] http://www.kernel.org/diff/diffview.cgi?file=/pub/linux/kernel/v2.4/testing/cset/[email protected]|ChangeSet|20030701183359|14011.txt
--
Erik B. Andersen http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--
--- linux/arch/i386/math-emu/poly.h.orig 2003-07-01 20:39:49.000000000 -0600
+++ linux/arch/i386/math-emu/poly.h 2003-07-01 20:39:57.000000000 -0600
@@ -64,7 +64,7 @@
const unsigned long arg2)
{
int retval;
- asm volatile ("mull %2; movl %%edx,%%eax" \
+ asm volatile ("mull %2; movl %%edx,%%eax; " \
:"=a" (retval) \
:"0" (arg1), "g" (arg2) \
:"dx");
@@ -75,11 +75,11 @@
/* Add the 12 byte Xsig x2 to Xsig dest, with no checks for overflow. */
static inline void add_Xsig_Xsig(Xsig *dest, const Xsig *x2)
{
- asm volatile ("movl %1,%%edi; movl %2,%%esi;" \
- movl (%%esi),%%eax; addl %%eax,(%%edi);" \
- movl 4(%%esi),%%eax; adcl %%eax,4(%%edi);" \
- movl 8(%%esi),%%eax; adcl %%eax,8(%%edi);"
- :"=g" (*dest):"g" (dest), "g" (x2)
+ asm volatile ("movl %1,%%edi; movl %2,%%esi; " \
+ "movl (%%esi),%%eax; addl %%eax,(%%edi); " \
+ "movl 4(%%esi),%%eax; adcl %%eax,4(%%edi); " \
+ "movl 8(%%esi),%%eax; adcl %%eax,8(%%edi); " \
+ :"=g" (*dest):"g" (dest), "g" (x2) \
:"ax","si","di");
}
@@ -90,19 +90,19 @@
problem, but keep fingers crossed! */
static inline void add_two_Xsig(Xsig *dest, const Xsig *x2, long int *exp)
{
- asm volatile ("movl %2,%%ecx; movl %3,%%esi;
- movl (%%esi),%%eax; addl %%eax,(%%ecx);
- movl 4(%%esi),%%eax; adcl %%eax,4(%%ecx);
- movl 8(%%esi),%%eax; adcl %%eax,8(%%ecx);
- jnc 0f;
- rcrl 8(%%ecx); rcrl 4(%%ecx); rcrl (%%ecx)
- movl %4,%%ecx; incl (%%ecx)
- movl $1,%%eax; jmp 1f;
- 0: xorl %%eax,%%eax;
- 1:"
- :"=g" (*exp), "=g" (*dest)
- :"g" (dest), "g" (x2), "g" (exp)
- :"cx","si","ax");
+ asm volatile ("movl %2,%%ecx; movl %3,%%esi; " \
+ "movl (%%esi),%%eax; addl %%eax,(%%ecx); " \
+ "movl 4(%%esi),%%eax; adcl %%eax,4(%%ecx); " \
+ "movl 8(%%esi),%%eax; adcl %%eax,8(%%ecx); " \
+ "jnc 0f; " \
+ "rcrl 8(%%ecx); rcrl 4(%%ecx); rcrl (%%ecx); " \
+ "movl %4,%%ecx; incl (%%ecx); " \
+ "movl $1,%%eax; jmp 1f; " \
+ "0: xorl %%eax,%%eax; " \
+ "1: " \
+ :"=g" (*exp), "=g" (*dest) \
+ :"g" (dest), "g" (x2), "g" (exp) \
+ :"cx","si","ax");
}
@@ -110,11 +110,11 @@
/* This is faster in a loop on my 386 than using the "neg" instruction. */
static inline void negate_Xsig(Xsig *x)
{
- asm volatile("movl %1,%%esi; "
- "xorl %%ecx,%%ecx; "
- "movl %%ecx,%%eax; subl (%%esi),%%eax; movl %%eax,(%%esi); "
- "movl %%ecx,%%eax; sbbl 4(%%esi),%%eax; movl %%eax,4(%%esi); "
- "movl %%ecx,%%eax; sbbl 8(%%esi),%%eax; movl %%eax,8(%%esi); "
+ asm volatile("movl %1,%%esi; " \
+ "xorl %%ecx,%%ecx; " \
+ "movl %%ecx,%%eax; subl (%%esi),%%eax; movl %%eax,(%%esi); " \
+ "movl %%ecx,%%eax; sbbl 4(%%esi),%%eax; movl %%eax,4(%%esi); " \
+ "movl %%ecx,%%eax; sbbl 8(%%esi),%%eax; movl %%eax,8(%%esi); " \
:"=g" (*x):"g" (x):"si","ax","cx");
}
I'm no GCC nor asm guru, so, Alan?
On Tue, 1 Jul 2003, Erik Andersen wrote:
> As of today's "fix up gcc 3.3 bits" patch [1], x86 math emulation
> is now even more broken, since this latest patch has added some
> mismatched quotes while still failing to address the actual
> problems preventing this code from compiling with gcc 3.3.
>
> This patch, first sent to you on Jun 21st, fixes the missing
> semicolons and missing quotes in the x86 math-emu code, allowing
> it to compile with gcc 3.3. I have updated things to also fix
> the mismatched quotes that were added today. Unlike the patch
> you applied earlier today, my patch is actually tested...
>
> Please apply,
>
> -Erik
>
> [1] http://www.kernel.org/diff/diffview.cgi?file=/pub/linux/kernel/v2.4/testing/cset/[email protected]|ChangeSet|20030701183359|14011.txt
>
> --
> Erik B. Andersen http://codepoet-consulting.com/
> --This message was written using 73% post-consumer electrons--
>
>
> --- linux/arch/i386/math-emu/poly.h.orig 2003-07-01 20:39:49.000000000 -0600
> +++ linux/arch/i386/math-emu/poly.h 2003-07-01 20:39:57.000000000 -0600
> @@ -64,7 +64,7 @@
> const unsigned long arg2)
> {
> int retval;
> - asm volatile ("mull %2; movl %%edx,%%eax" \
> + asm volatile ("mull %2; movl %%edx,%%eax; " \
> :"=a" (retval) \
> :"0" (arg1), "g" (arg2) \
> :"dx");
> @@ -75,11 +75,11 @@
> /* Add the 12 byte Xsig x2 to Xsig dest, with no checks for overflow. */
> static inline void add_Xsig_Xsig(Xsig *dest, const Xsig *x2)
> {
> - asm volatile ("movl %1,%%edi; movl %2,%%esi;" \
> - movl (%%esi),%%eax; addl %%eax,(%%edi);" \
> - movl 4(%%esi),%%eax; adcl %%eax,4(%%edi);" \
> - movl 8(%%esi),%%eax; adcl %%eax,8(%%edi);"
> - :"=g" (*dest):"g" (dest), "g" (x2)
> + asm volatile ("movl %1,%%edi; movl %2,%%esi; " \
> + "movl (%%esi),%%eax; addl %%eax,(%%edi); " \
> + "movl 4(%%esi),%%eax; adcl %%eax,4(%%edi); " \
> + "movl 8(%%esi),%%eax; adcl %%eax,8(%%edi); " \
> + :"=g" (*dest):"g" (dest), "g" (x2) \
> :"ax","si","di");
> }
>
> @@ -90,19 +90,19 @@
> problem, but keep fingers crossed! */
> static inline void add_two_Xsig(Xsig *dest, const Xsig *x2, long int *exp)
> {
> - asm volatile ("movl %2,%%ecx; movl %3,%%esi;
> - movl (%%esi),%%eax; addl %%eax,(%%ecx);
> - movl 4(%%esi),%%eax; adcl %%eax,4(%%ecx);
> - movl 8(%%esi),%%eax; adcl %%eax,8(%%ecx);
> - jnc 0f;
> - rcrl 8(%%ecx); rcrl 4(%%ecx); rcrl (%%ecx)
> - movl %4,%%ecx; incl (%%ecx)
> - movl $1,%%eax; jmp 1f;
> - 0: xorl %%eax,%%eax;
> - 1:"
> - :"=g" (*exp), "=g" (*dest)
> - :"g" (dest), "g" (x2), "g" (exp)
> - :"cx","si","ax");
> + asm volatile ("movl %2,%%ecx; movl %3,%%esi; " \
> + "movl (%%esi),%%eax; addl %%eax,(%%ecx); " \
> + "movl 4(%%esi),%%eax; adcl %%eax,4(%%ecx); " \
> + "movl 8(%%esi),%%eax; adcl %%eax,8(%%ecx); " \
> + "jnc 0f; " \
> + "rcrl 8(%%ecx); rcrl 4(%%ecx); rcrl (%%ecx); " \
> + "movl %4,%%ecx; incl (%%ecx); " \
> + "movl $1,%%eax; jmp 1f; " \
> + "0: xorl %%eax,%%eax; " \
> + "1: " \
> + :"=g" (*exp), "=g" (*dest) \
> + :"g" (dest), "g" (x2), "g" (exp) \
> + :"cx","si","ax");
> }
>
>
> @@ -110,11 +110,11 @@
> /* This is faster in a loop on my 386 than using the "neg" instruction. */
> static inline void negate_Xsig(Xsig *x)
> {
> - asm volatile("movl %1,%%esi; "
> - "xorl %%ecx,%%ecx; "
> - "movl %%ecx,%%eax; subl (%%esi),%%eax; movl %%eax,(%%esi); "
> - "movl %%ecx,%%eax; sbbl 4(%%esi),%%eax; movl %%eax,4(%%esi); "
> - "movl %%ecx,%%eax; sbbl 8(%%esi),%%eax; movl %%eax,8(%%esi); "
> + asm volatile("movl %1,%%esi; " \
> + "xorl %%ecx,%%ecx; " \
> + "movl %%ecx,%%eax; subl (%%esi),%%eax; movl %%eax,(%%esi); " \
> + "movl %%ecx,%%eax; sbbl 4(%%esi),%%eax; movl %%eax,4(%%esi); " \
> + "movl %%ecx,%%eax; sbbl 8(%%esi),%%eax; movl %%eax,8(%%esi); " \
> :"=g" (*x):"g" (x):"si","ax","cx");
> }
>
>
On Wed Jul 02, 2003 at 10:52:02AM -0300, Marcelo Tosatti wrote:
>
>
> I'm no GCC nor asm guru, so, Alan?
This does not change even one asm instruction. This merely
changes the quoting and semicolons...
-Erik
--
Erik B. Andersen http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--
On Mer, 2003-07-02 at 17:22, Erik Andersen wrote:
> On Wed Jul 02, 2003 at 10:52:02AM -0300, Marcelo Tosatti wrote:
> >
> >
> > I'm no GCC nor asm guru, so, Alan?
>
> This does not change even one asm instruction. This merely
> changes the quoting and semicolons...
Erik is right - I sent you a buggy fix, then a correct fix, but
his cleanup is nicer