* src/powerpc/aix_closure.S: Reorganize 64-bit code to match

        linux64_closure.S.

git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@155001 138bc75d-0d04-0410-961f-82ee72b054a4

diff --git a/libffi/ChangeLog b/libffi/ChangeLog
index 7c877a1..6ea5043 100644 (file)
--- a/libffi/ChangeLog
+++ b/libffi/ChangeLog
@@ -1,3 +1,8 @@
+2009-12-04  David Edelsohn  <edelsohn@gnu.org>
+
+       * src/powerpc/aix_closure.S: Reorganize 64-bit code to match
+        linux64_closure.S.
+       
 2009-12-04  Uros Bizjak  <ubizjak@gmail.com>
 
        PR libffi/41908

diff --git a/libffi/src/powerpc/aix_closure.S b/libffi/src/powerpc/aix_closure.S
index 7045618..3ba7aae 100644 (file)
--- a/libffi/src/powerpc/aix_closure.S
+++ b/libffi/src/powerpc/aix_closure.S
@@ -99,9 +99,23 @@ ffi_closure_ASM:
        .llong .ffi_closure_ASM, TOC[tc0], 0
        .csect .text[PR]
 .ffi_closure_ASM:
+/* we want to build up an area for the parameters passed */
+/* in registers (both floating point and integer) */
+
+       /* we store gpr 3 to gpr 10 (aligned to 4)
+       in the parents outgoing area  */
+       std   r3, 48+(0*8)(r1)
+       std   r4, 48+(1*8)(r1)
+       std   r5, 48+(2*8)(r1)
+       std   r6, 48+(3*8)(r1)
+       mflr  r0
+
+       std   r7, 48+(4*8)(r1)
+       std   r8, 48+(5*8)(r1)
+       std   r9, 48+(6*8)(r1)
+       std   r10, 48+(7*8)(r1)
+       std   r0, 16(r1)        /* save the return address */
 
-       mflr r0                 /* extract return address */
-       std r0,16(r1)           /* save the return address */
 
        /* 48  Bytes (Linkage Area) */
        /* 64  Bytes (params) */
@@ -110,23 +124,9 @@ ffi_closure_ASM:
        /* 8   Bytes (alignment) */
        /* 240 Bytes */
 
-       stdu r1,-240(r1)        /* skip over caller save area
+       stdu  r1, -240(r1)      /* skip over caller save area
                                   keep stack aligned to 16  */
 
-/* we want to build up an area for the parameters passed */
-/* in registers (both floating point and integer) */
-
-       /* we store gpr 3 to gpr 10 (aligned to 4)
-       in the parents outgoing area  */
-       std   r3, 288+(0*8)(r1)
-       std   r4, 288+(1*8)(r1)
-       std   r5, 288+(2*8)(r1)
-       std   r6, 288+(3*8)(r1)
-       std   r7, 288+(4*8)(r1)
-       std   r8, 288+(5*8)(r1)
-       std   r9, 288+(6*8)(r1)
-       std   r10, 288+(7*8)(r1)
-
        /* next save fpr 1 to fpr 13 (aligned to 8) */
        stfd  f1, 128+(0*8)(r1)
        stfd  f2, 128+(1*8)(r1)
@@ -144,16 +144,16 @@ ffi_closure_ASM:
 
        /* set up registers for the routine that actually does the work */
        /* get the context pointer from the trampoline */
-       mr r3,r11
+       mr r3, r11
 
        /* now load up the pointer to the result storage */
-       addi r4,r1,112
+       addi r4, r1, 112
 
        /* now load up the pointer to the saved gpr registers */
-       addi r5,r1,288
+       addi r5, r1, 288
 
        /* now load up the pointer to the saved fpr registers */
-       addi r6,r1,128
+       addi r6, r1, 128
 
        /* make the call */
        bl .ffi_closure_helper_DARWIN
@@ -165,88 +165,107 @@ ffi_closure_ASM:
 
        /* look up the proper starting point in table  */
        /* by using return type as offset */
-       addi r5,r1,112          /* get pointer to results area */
-       ld r4,LC..60(2)         /* get address of jump table */
-       sldi r3,r3,2            /* now multiply return type by 4 */
-       lwzx r3,r4,r3           /* get the contents of that table value */
+       ld r4, LC..60(2)                /* get address of jump table */
+       sldi r3, r3, 4          /* now multiply return type by 16 */
+       ld r0, 240+16(r1)       /* load return address */
        add r3,r3,r4            /* add contents of table to table address */
        mtctr r3
        bctr                    /* jump to it */
 
+/* Each fragment must be exactly 16 bytes long (4 instructions).
+   Align to 16 byte boundary for cache and dispatch efficiency.  */
+       .align 4
+
 L..60:
-       .long L..44-L..60    /* FFI_TYPE_VOID */
-       .long L..51-L..60    /* FFI_TYPE_INT */
-       .long L..47-L..60    /* FFI_TYPE_FLOAT */
-       .long L..46-L..60    /* FFI_TYPE_DOUBLE */
-       .long L..45-L..60    /* FFI_TYPE_LONGDOUBLE */
-       .long L..56-L..60    /* FFI_TYPE_UINT8 */
-       .long L..55-L..60    /* FFI_TYPE_SINT8 */
-       .long L..58-L..60    /* FFI_TYPE_UINT16 */
-       .long L..57-L..60    /* FFI_TYPE_SINT16 */
-       .long L..50-L..60    /* FFI_TYPE_UINT32 */
-       .long L..51-L..60    /* FFI_TYPE_SINT32 */
-       .long L..48-L..60    /* FFI_TYPE_UINT64 */
-       .long L..48-L..60    /* FFI_TYPE_SINT64 */
-       .long L..44-L..60    /* FFI_TYPE_STRUCT */
-       .long L..48-L..60    /* FFI_TYPE_POINTER */
+/* case FFI_TYPE_VOID */
+       mtlr r0
+       addi r1, r1, 240
+       blr
+       nop
 
+/* case FFI_TYPE_INT */
+       lwa r3, 112+4(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case long double */
-L..45:
-       lfd f1,0(r5)
-       lfd f2,8(r5)
-       b L..44
+/* case FFI_TYPE_FLOAT */
+       lfs f1, 112+0(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case double */
-L..46:
-       lfd f1,0(r5)
-       b L..44
+/* case FFI_TYPE_DOUBLE */
+       lfd f1, 112+0(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case float */
-L..47:
-       lfs f1,0(r5)
-       b L..44
+/* case FFI_TYPE_LONGDOUBLE */
+       lfd f1, 112+0(r1)
+       mtlr r0
+       lfd f2, 112+8(r1)
+       b L..finish
 
-/* case long long / pointer */
-L..48:
-       ld r3,0(r5)
-       b L..44
+/* case FFI_TYPE_UINT8 */
+       lbz r3, 112+7(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case uint32 */
-L..50:
-       lwz r3,4(r5)
-       b L..44
+/* case FFI_TYPE_SINT8 */
+       lbz r3, 112+7(r1)
+       mtlr r0
+       extsb r3, r3
+       b L..finish
+
+/* case FFI_TYPE_UINT16 */
+       lhz r3, 112+6(r1)
+       mtlr r0
+L..finish:
+       addi r1, r1, 240
+       blr
 
-/* case int / sint32 */
-L..51:
-       lwa r3,4(r5)
-       b L..44
+/* case FFI_TYPE_SINT16 */
+       lha r3, 112+6(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case signed int8 */
-L..55:
-       lbz r3,7(r5)
-       extsb r3,r3
-       b L..44
+/* case FFI_TYPE_UINT32 */
+       lwz r3, 112+4(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case unsigned int8 */
-L..56:
-       lbz r3,7(r5)
-       b L..44
+/* case FFI_TYPE_SINT32 */
+       lwa r3, 112+4(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case signed int16 */
-L..57:
-       lha r3,6(r5)
-       b L..44
+/* case FFI_TYPE_UINT64 */
+       ld r3, 112+0(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case unsigned int16 */
-L..58:
-       lhz r3,6(r5)
+/* case FFI_TYPE_SINT64 */
+       ld r3, 112+0(r1)
+       mtlr r0
+       addi r1, r1, 240
+       blr
 
-/* case void / done     */
-L..44:
-       addi r1,r1,240          /* restore stack pointer */
-       ld r0,16(r1)            /* get return address */
-       mtlr r0                 /* reset link register */
+/* case FFI_TYPE_STRUCT */
+       mtlr r0
+       addi r1, r1, 240
+       blr
+       nop
+
+/* case FFI_TYPE_POINTER */
+       ld r3, 112+0(r1)
+       mtlr r0
+       addi r1, r1, 240
        blr
 
 #else /* ! __64BIT__ */