[Bug target/23250] [4.0/4.1 regression] 4.0.1 switch -mcallee-super-interworking generating wrong code for ARM7TDMI

[th dot r dot klein at web dot de]

------- Additional Comments From th dot r dot klein at web dot de  2005-08-12 11:54 -------

The patch did not compiles with keyword MASK_INTERWORK
so using keyword ARM_FLAG_INTERWORK instead

--- arm.c.orig	Mon May  9 14:53:19 2005
+++ arm.c	Fri Aug 12 13:16:23 2005
@@ -944,6 +944,11 @@
       target_flags &= ~ARM_FLAG_APCS_FRAME;
     }
 
+  /* Callee super interworking implies thumb interworking.  Adding
+     this to the flags here simplifies the logic elsewhere.  */
+  if (TARGET_THUMB && TARGET_CALLEE_INTERWORKING)
+      target_flags |= ARM_FLAG_INTERWORK;
+
   /* TARGET_BACKTRACE calls leaf_function_p, which causes a crash if done
      from here where no function is being compiled currently.  */
   if ((target_flags & (THUMB_FLAG_LEAF_BACKTRACE | THUMB_FLAG_BACKTRACE))

I've got now the result

arm-elf-gcc-3.4.4 -g -Os -fPIC -mthumb -mlittle-endian -mapcs \
      -march=armv4t -mcallee-super-interworking \
      -c progWord.c -o 3_4_4_thumb.o

3_4_4_thumb.o:     file format elf32-littlearm

Disassembly of section .text:

00000000 <progWord>:
   0:	e38fc001 	orr	ip, pc, #1	; 0x1
   4:	e12fff1c 	bx	ip

00000008 <.real_start_ofprogWord>:
   8:	b570      	push	{r4, r5, r6, lr}
   .
   register r10 (sl) never been used
   .
  52:	2000      	movs	r0, #0
  54:	bc70      	pop	{r4, r5, r6}
  56:	bc02      	pop	{r1}
  58:	4708      	bx	r1

arm-elf-gcc-4.0.1 -g -Os -fPIC -mthumb -mlittle-endian -mapcs \
      -march=armv4t -mcallee-super-interworking \
      -c progWord.c -o 4_0_1_thumb.o

4_0_1_thumb.o:     file format elf32-littlearm

Disassembly of section .text:

00000000 <progWord>:
   0:	e38fc001 	orr	ip, pc, #1	; 0x1
   4:	e12fff1c 	bx	ip

00000008 <.real_start_ofprogWord>:
   8:	b570      	push	{r4, r5, r6, lr}
   a:	4656      	mov	r6, sl
   c:	b440      	push	{r6}
   .
   register r10 (sl) never been used
   .
  56:	2000      	movs	r0, #0
  58:	bc04      	pop	{r2}
  5a:	4692      	mov	sl, r2
  5c:	bc70      	pop	{r4, r5, r6}
  5e:	bc02      	pop	{r1}
  60:	4708      	bx	r1

This is a working version. 
But it is still not optimized.
Should I open the missing otimization within an other bug entry?


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=23250