[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

--- Comment #10 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Still happens on the trunk.
Note armv7-a produces reasonable code though as it uses ubfx which is new for
armv7-a:
        push    {r4, lr}
        mov     r4, r0
        movs    r0, #8
        lsrs    r1, r4, #24
        bl      func
        ubfx    r1, r4, #16, #8
        movs    r0, #8
        bl      func
        ubfx    r1, r4, #8, #8
        movs    r0, #8
        pop     {r4, lr}
        b       func

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[rearnsha at arm dot com]

------- Comment #9 from rearnsha at arm dot com  2010-02-08 16:31 -------
Subject: Re:  [arm] Combine cannot do its job because
        immediate operand is used instead of register


On Mon, 2010-02-08 at 16:30 +0000, rearnsha at arm dot com wrote:
>         mov     r1, r5, r4, lsr #8
> 

Should, of course, be AND.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[rearnsha at arm dot com]

------- Comment #8 from rearnsha at arm dot com  2010-02-08 16:30 -------
Subject: Re:  [arm] Combine cannot do its job because
        immediate operand is used instead of register


On Mon, 2010-02-08 at 16:11 +0000, steven at gcc dot gnu dot org wrote:
> Can someone please explain what the expected code is?

Something like

        stmfd   sp!, {r3, r4, r5, lr}
        mov     r4, r0
        mov     r5, #255
        mov     r1, r4, lsr #24
        mov     r0, #8
        bl      func
        and     r1, r5, r4, lsr #16
        mov     r0, #8
        bl      func
        mov     r1, r5, r4, lsr #8
        mov     r0, #8
        ldmfd   sp!, {r3, r4, r5, lr}
        b       func

By putting 255 into a register we add one instruction, but remove 2; so
save one overall.  This is one of the rare cases where an immediate is
actually more expensive on ARM than using a register.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[steven at gcc dot gnu dot org]

------- Comment #7 from steven at gcc dot gnu dot org  2010-02-08 16:11 -------
New test case is with func defined extern, as already mentioned in comment #5:

extern void func(char c, int t);

void foo(int u)
{
  func ( 8, (u >> 24) & 0xffL );
  func ( 8, (u >> 16) & 0xffL );
  func ( 8, (u >> 8) & 0xffL );
}

Trunk today (r156595) produces this:

        stmfd   sp!, {r4, lr}
        mov     r4, r0
        mov     r1, r4, lsr #24
        mov     r0, #8
        bl      func
        mov     r1, r4, lsr #16
        mov     r0, #8
        and     r1, r1, #255
        bl      func
        mov     r1, r4, lsr #8
        mov     r0, #8
        and     r1, r1, #255
        ldmfd   sp!, {r4, lr}
        b       func

Can someone please explain what the expected code is?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[ramana dot r at gmail dot com]

------- Comment #6 from ramana dot r at gmail dot com  2009-03-12 18:45 -------
With Mainline today gcc produces : 

        stmfd   sp!, {r4, lr}
        mov     r1, r0, lsr #24
        mov     r4, r0
        mov     r0, #8
        bl      func
        mov     r1, r4, lsr #16
        and     r1, r1, #255
        mov     r0, #8
        bl      func
        mov     r1, r4, lsr #8
        and     r1, r1, #255
        mov     r0, #8
        ldmfd   sp!, {r4, lr}
        b       func

The problem still exists. This can't be a problem with the combiner because the
combine would stop at function call boundaries. 


-- 

ramana dot r at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rearnsha at arm dot com,
                   |                            |ramana dot r at gmail dot
                   |                            |com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-20 06:42 -------
Note to reproduce this on the mainline you have remove the definition of the function func because 
func is really a pure function and we remove the call to the function and the rest of the foo is really 
dead code.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2004-05-21 03:44:10         |2004-08-20 06:42:58
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760


pinskia at gcc dot gnu dot org changed:

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

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[pinskia@physics.uc.edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760


pinskia@physics.uc.edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-06-02 01:18:46
               date|                            |


------- Additional Comments From pinskia@physics.uc.edu  2003-06-02 01:18 -------
According to the submitter the problem still exists.



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[pinskia@physics.uc.edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760


pinskia@physics.uc.edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


------- Additional Comments From pinskia@physics.uc.edu  2003-05-26 20:09 -------
See Dara's question.



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug target/9760] [arm] Combine cannot do its job because immediate operand is used instead of register

[dhazeghi@yahoo.com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=9760



------- Additional Comments From dhazeghi@yahoo.com  2003-05-26 19:49 -------
Hello,

with gcc 3.3 I get the following code:

        .file   "01.i"
        .text
        .align  2
        .global func
        .type   func, %function
func:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        @ lr needed for prologue
        mov     pc, lr
        .size   func, .-func
        .align  2
        .global foo
        .type   foo, %function
foo:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 1, uses_anonymous_args = 0
        mov     ip, sp
        stmfd   sp!, {r4, fp, ip, lr, pc}
        mov     r1, r0, lsr #24
        sub     fp, ip, #4
        mov     r4, r0
        mov     r0, #8
        bl      func
        mov     r1, r4, asr #16
        and     r1, r1, #255
        mov     r0, #8
        bl      func
        mov     r4, r4, asr #8
        and     r4, r4, #255
        mov     r1, r4
        mov     r0, #8
        ldmea   fp, {r4, fp, sp, lr}
        b       func
        .size   foo, .-foo
        .ident  "GCC: (GNU) 3.3 20030508 (prerelease)"

with mainline, I get this slightly different code:

        .file   "01.i"
        .text
        .align  2
        .global func
        .type   func, %function
func:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        @ lr needed for prologue
        mov     pc, lr
        .size   func, .-func
        .align  2
        .global foo
        .type   foo, %function
foo:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 1, uses_anonymous_args = 0
        mov     ip, sp
        stmfd   sp!, {r4, fp, ip, lr, pc}
        mov     r1, r0, lsr #24
        sub     fp, ip, #4
        mov     r4, r0
        mov     r0, #8
        bl      func
        mov     r1, r4, lsr #16
        and     r1, r1, #255
        mov     r0, #8
        bl      func
        mov     r4, r4, lsr #8
        and     r1, r4, #255
        mov     r0, #8
        ldmea   fp, {r4, fp, sp, lr}
        b       func
        .size   foo, .-foo
        .ident  "GCC: (GNU) 3.4 20030508 (experimental)"

Is this an improvement from the code you saw (sorry, I don't know arm asm, and there are too 
many changes from the .s file you provided just to do a diff)? Thanks,

Dara



------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.