[Bug target/40657] New: allocate local variables with fewer instructions

[Bug target/40657] New: allocate local variables with fewer instructions

[carrot at google dot com]

Compile following code with options -Os -mthumb -march=armv5te

extern void bar(int*);
int foo()
{
  int x;
  bar(&x);
  return x;
}

Gcc generates:

        push    {lr}
        sub     sp, sp, #12
        add     r0, sp, #4
        bl      bar
        ldr     r0, [sp, #4]
        add     sp, sp, #12
        @ sp needed for prologue
        pop     {pc}

A better code sequence could be:

        push    {r1-r3,lr}
        add     r0, sp, #4
        bl      bar
        ldr     r0, [sp, #4]
        @ sp needed for prologue
        pop     {r1-r3, pc}

The local variable allocation and deallocation can be merged into the push/pop
instruction, so we can avoid the extra sub/add instructions and reduce two
instructions.


-- 
           Summary: allocate local variables with fewer instructions
           Product: gcc
           Version: 4.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: carrot at google dot com
 GCC build triplet: i686-linux
  GCC host triplet: i686-linux
GCC target triplet: arm-eabi


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[carrot at google dot com]

------- Comment #1 from carrot at google dot com  2009-07-06 08:16 -------
Created an attachment (id=18140)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18140&action=view)
test case


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[steven at gcc dot gnu dot org]

------- Comment #2 from steven at gcc dot gnu dot org  2009-07-06 09:13 -------
IIRC push-multiple and pop-multiple are not supported yet. Richard E.?


-- 

steven at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rearnsha at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2009-07-06 09:13:55
               date|                            |
            Summary|allocate local variables    |allocate local variables
                   |with fewer instructions     |with fewer instructions


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[ramana at gcc dot gnu dot org]

------- Comment #3 from ramana at gcc dot gnu dot org  2009-07-06 09:54 -------
(In reply to comment #2)
> IIRC push-multiple and pop-multiple are not supported yet. Richard E.?
> 

I am not sure what you mean here. push and pop multiple are supported to the
best of the cases that GCC can detect them. since load and store multiple are
not first class citizens, the ARM backend attempts to detect them in the form
of peephole2's and peephole's.

However this case isn't to do with push or pop multiple but to do with saving
space in the thumb prologue and epilogue by emitting push and pop multiple
instructions instead of explicit instructions to manipulate the stack. Notice
that r2 and r3 are not used in the code that Carrot is proposing in comment #1.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[steven at gcc dot gnu dot org]

------- Comment #4 from steven at gcc dot gnu dot org  2009-07-06 10:43 -------
Ah, heh, so you're saying that pushing/popping registers you don't have to save
may be a size optimization? That's an interesting idea.

But how to do this in GCC...  The "push {lr}" is never even in the RTL.  Output
with same options + "-dAP":

foo:
        push    {lr}
        @ basic block 2
@(insn/f 20 2 21 t.c:3 (set (reg/f:SI 13 sp)
@        (plus:SI (reg/f:SI 13 sp)
@            (const_int -12 [0xfffffffffffffff4]))) 5 {*thumb1_addsi3} (nil))
@ 0x0000
        sub     sp, sp, #12     @ 20    *thumb1_addsi3/7        [length = 2]
@(insn 6 21 7 t.c:5 (set (reg:SI 0 r0)
@        (plus:SI (reg/f:SI 13 sp)
@            (const_int 4 [0x4]))) 5 {*thumb1_addsi3} (expr_list:REG_EQUAL
(plus:SI (reg/f:SI 13 sp)
@            (const_int 4 [0x4]))
@        (nil)))
@ 0x0002
        add     r0, sp, #4      @ 6     *thumb1_addsi3/6        [length = 2]
@(call_insn 7 6 13 t.c:5 (parallel [
@            (call (mem:SI (symbol_ref:SI ("bar") [flags 0x41] <function_decl
0x2000000000545200 bar>) [0 S4 A32])
@                (const_int 0 [0x0]))
@            (use (const_int 0 [0x0]))
@            (clobber (reg:SI 14 lr))
@        ]) 253 {*call_insn} (expr_list:REG_DEAD (reg:SI 0 r0)
@        (nil))
@    (expr_list:REG_DEP_TRUE (use (reg:SI 0 r0))
@        (nil)))
@ 0x0004
        bl      bar     @ 7     *call_insn      [length = 4]
@(insn 13 7 22 t.c:7 (set (reg/i:SI 0 r0)
@        (mem/c/i:SI (plus:SI (reg/f:SI 13 sp)
@                (const_int 4 [0x4])) [2 x+0 S4 A32])) 167 {*thumb1_movsi_insn}
(nil))
@ 0x0008
        ldr     r0, [sp, #4]    @ 13    *thumb1_movsi_insn/7    [length = 2]
@(insn 23 22 16 t.c:7 (set (reg/f:SI 13 sp)
@        (plus:SI (reg/f:SI 13 sp)
@            (const_int 12 [0xc]))) 5 {*thumb1_addsi3} (nil))
@ 0x000a
        add     sp, sp, #12     @ 23    *thumb1_addsi3/7        [length = 2]
@(insn 24 16 25 t.c:7 (unspec:SI [
@            (reg/f:SI 13 sp)
@        ] 6) 339 {prologue_use} (nil))
@ 0x000c
        @ sp needed for prologue        @ 24    prologue_use    [length = 0]
@(jump_insn 25 24 26 t.c:7 (unspec_volatile [
@            (return)
@        ] 1) 321 {*epilogue_insns} (nil))
@ 0x000c
        pop     {pc}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[carrot at google dot com]

------- Comment #5 from carrot at google dot com  2009-07-07 06:44 -------
Could we do the optimization in function thumb1_expand_prologue? If we find
this opportunity in function thumb1_expand_prologue, we can remove the sp
manipulations from prologue and epilogue. We also should add extra registers to
push/pop operands.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[rearnsha at arm dot com]

------- Comment #6 from rearnsha at arm dot com  2009-07-13 14:22 -------
Subject: Re:  allocate local variables with fewer
        instructions

On Mon, 2009-07-06 at 10:43 +0000, steven at gcc dot gnu dot org wrote:
> But how to do this in GCC...  The "push {lr}" is never even in the
> RTL.  Output with same options + "-dAP":

We already do this for ARM state.  In thumb the split between epilogue
code in RTL and text makes it slightly more complex, but not massively
so.

R.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[ramana at gcc dot gnu dot org]

-- 

ramana at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[bernds at codesourcery dot com]

------- Comment #7 from bernds at codesourcery dot com  2010-03-31 21:42 -------
A patch is at http://gcc.gnu.org/ml/gcc-patches/2010-03/msg01274.html


-- 

bernds at codesourcery dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bernds at codesourcery dot
                   |                            |com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[bernds at gcc dot gnu dot org]

------- Comment #8 from bernds at gcc dot gnu dot org  2010-04-27 09:34 -------
Subject: Bug 40657

Author: bernds
Date: Tue Apr 27 09:34:08 2010
New Revision: 158771

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=158771
Log:
        PR target/40657
        * config/arm/arm.c (thumb1_extra_regs_pushed): New function.
        (thumb1_expand_prologue, thumb1_output_function_prologue): Call it
        here to determine which regs to push and how much stack to reserve.

        PR target/40657
        * gcc.target/arm/thumb-stackframe.c: New test.


Added:
    trunk/gcc/testsuite/gcc.target/arm/thumb-stackframe.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/arm/arm.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[bernds at gcc dot gnu dot org]

------- Comment #9 from bernds at gcc dot gnu dot org  2010-04-27 09:35 -------
Fixed.


-- 

bernds at gcc dot gnu dot org changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[carrot at google dot com]

------- Comment #10 from carrot at google dot com  2010-07-02 17:41 -------
The patch optimizes the prologue part only, the epilogue can also be enhanced.
Now compile the attached test case, I get

        push    {r0, r1, r2, lr}
        add     r0, sp, #4
        bl      bar
        ldr     r0, [sp, #4]
        add     sp, sp, #12
        pop     {pc}

The last two instructions can be merged as 

        pop     {r1-r3, pc}


-- 

carrot at google dot com changed:

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[bernds at gcc dot gnu dot org]

------- Comment #11 from bernds at gcc dot gnu dot org  2010-07-09 09:03 -------
Subject: Bug 40657

Author: bernds
Date: Fri Jul  9 09:03:22 2010
New Revision: 161988

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=161988
Log:
        PR target/40657
        * config/arm/arm.c (thumb1_extra_regs_pushed): New arg FOR_PROLOGUE.
        All callers changed.
        Handle the case when we're called for the epilogue.
        (thumb_unexpanded_epilogue): Use it.
        (thumb1_expand_epilogue): Likewise.

testsuite/
        PR target/40657
        * gcc.target/arm/pr40657-1.c: New test.
        * gcc.target/arm/pr40657-2.c: New test.
        * gcc.c-torture/execute/pr40657.c: New test.


Added:
    trunk/gcc/testsuite/gcc.c-torture/execute/pr40657.c
    trunk/gcc/testsuite/gcc.target/arm/pr40657-1.c
    trunk/gcc/testsuite/gcc.target/arm/pr40657-2.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/config/arm/arm.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

[Bug target/40657] allocate local variables with fewer instructions

[aldot at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

Bernhard Reutner-Fischer <aldot at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |aldot at gcc dot gnu.org

--- Comment #12 from Bernhard Reutner-Fischer <aldot at gcc dot gnu.org> 2012-06-20 18:24:29 UTC ---
Fixed?


$ arm-oe-linux-uclibceabi-gcc -Os -mthumb -march=armv5te -S pr40657.c -o -
    .arch armv5te
    .fpu softvfp
    .eabi_attribute 20, 1
    .eabi_attribute 21, 1
    .eabi_attribute 23, 3
    .eabi_attribute 24, 1
    .eabi_attribute 25, 1
    .eabi_attribute 26, 2
    .eabi_attribute 30, 4
    .eabi_attribute 34, 0
    .eabi_attribute 18, 4
    .code    16
    .file    "pr40657.c"
    .text
    .align    1
    .global    foo
    .code    16
    .thumb_func
    .type    foo, %function
foo:
    push    {r0, r1, r2, lr}
    add    r0, sp, #4
    bl    bar
    @ sp needed for prologue
    ldr    r0, [sp, #4]
    pop    {r1, r2, r3, pc}
    .size    foo, .-foo
    .ident    "GCC: (GNU) 4.7.1 20120421 (prerelease)"
    .section    .note.GNU-stack,"",%progbits

[Bug target/40657] allocate local variables with fewer instructions

[bernds at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40657

Bernd Schmidt <bernds at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
                 CC|                            |bernds at gcc dot gnu.org
         Resolution|                            |FIXED

--- Comment #13 from Bernd Schmidt <bernds at gcc dot gnu.org> 2012-06-20 18:30:54 UTC ---
Yeah.