[Bug middle-end/42501] New: Code bloating on operations with bit fields

[Bug middle-end/42501] New: Code bloating on operations with bit fields

[sliao at google dot com]

Alexvod tried pr31849-patch from
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31849. It changed result size from
72 to 68 below, but still didn't eliminated the loop counter below.

See the following code

// compilation options: -march=armv5te -mthumb -Os
struct tree_block
{
 unsigned handler_block_flag:2;
 unsigned block_num:30;
};
static int next_block_index = 2;
void number_blocks (int n_blocks, struct tree_block **block_vector)
{
 int i;
 for (i = 0; i < n_blocks; ++i)
   ((block_vector[i])->block_num) = next_block_index++;
}

is compiled by gcc 4.4.0 in very inefficient way. gcc 4.2.1 compiles it to 48
bytes, and gcc 4.4.0 to 72 bytes (1.5 times bigger).

Analysis of assembly files shows the following problems:

1) operations with bitfields are done inefficiently. gcc-4.2.1 sets block_num
by LSLing the value and ORRing it.gcc-4.4.0 loads an extra constant 0x3fffffff
from memory and does AND in addition to that LSL and ORR.

2) gcc-4.4.0 doesn't eliminate loop counter i. It increments both block_vector
and i. Instead gcc-4.2.1 computes the end of the loop and increments only
block_vector

3) register allocation performs badly, it spills some registers to stack, which
causes extra LDR, STR operations in the loop body

The code was taken from GCC SPEC benchmark.


-- 
           Summary: Code bloating on operations with bit fields
           Product: gcc
           Version: 4.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: sliao 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=42501

[Bug middle-end/42501] [4.4 only] Code bloating on operations with bit fields

[ramana at gcc dot gnu dot org]

------- Comment #1 from ramana at gcc dot gnu dot org  2009-12-30 00:19 -------
This appears to be a 4.4 only size regression and this  appears to be fixed on
trunk.

With -Os -march=armv5te -mthumb size is 44 bytes on trunk as of today. gcc
version 4.5.0 20091229 (experimental) (GCC) 


00000000 <number_blocks>:
   0:   b5f0            push    {r4, r5, r6, r7, lr}
   2:   4b09            ldr     r3, [pc, #36]   ; (28 <number_blocks+0x28>)
   4:   2200            movs    r2, #0
   6:   681b            ldr     r3, [r3, #0]
   8:   2503            movs    r5, #3
   a:   e007            b.n     1c <number_blocks+0x1c>
   c:   c910            ldmia   r1!, {r4}
   e:   009f            lsls    r7, r3, #2
  10:   6826            ldr     r6, [r4, #0]
  12:   3201            adds    r2, #1
  14:   402e            ands    r6, r5
  16:   433e            orrs    r6, r7
  18:   3301            adds    r3, #1
  1a:   6026            str     r6, [r4, #0]
  1c:   4282            cmp     r2, r0
  1e:   dbf5            blt.n   c <number_blocks+0xc>
  20:   4a01            ldr     r2, [pc, #4]    ; (28 <number_blocks+0x28>)
  22:   6013            str     r3, [r2, #0]
  24:   bdf0            pop     {r4, r5, r6, r7, pc}
  26:   46c0            nop                     ; (mov r8, r8)
  28:   00000000        .word   0x00000000


-- 

ramana at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2009-12-30 00:19:26
               date|                            |
            Summary|Code bloating on operations |[4.4 only] Code bloating on
                   |with bit fields             |operations with bit fields


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42501

[Bug middle-end/42501] [4.4 only] Code bloating on operations with bit fields

[rguenth at gcc dot gnu dot org]

------- Comment #2 from rguenth at gcc dot gnu dot org  2009-12-31 15:28 -------
Fixed in 4.5.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
      Known to work|                            |4.5.0
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42501

[Bug middle-end/42501] [4.4 only] Code bloating on operations with bit fields

[sliao at google dot com]

------- Comment #3 from sliao at google dot com  2010-01-08 00:29 -------
Yes, I generated the code from the trunk and got the same code as Ramana got,
except for the last line below. (Sorry that I filed this bug right before the
Christmas flight and couldn't check the trunk at that time.)

815c:   b5f0            push    {r4, r5, r6, r7, lr}
815e:   4b09            ldr     r3, [pc, #36]   ; (8184 <main+0x28>)
8160:   2200            movs    r2, #0
8162:   681b            ldr     r3, [r3, #0]
8164:   2503            movs    r5, #3
8166:   e007            b.n     8178 <main+0x1c>
8168:   c910            ldmia   r1!, {r4}
816a:   009f            lsls    r7, r3, #2
816c:   6826            ldr     r6, [r4, #0]
816e:   3201            adds    r2, #1
8170:   402e            ands    r6, r5
8172:   433e            orrs    r6, r7
8174:   3301            adds    r3, #1
8176:   6026            str     r6, [r4, #0]
8178:   4282            cmp     r2, r0
817a:   dbf5            blt.n   8168 <main+0xc>
817c:   4a01            ldr     r2, [pc, #4]    ; (8184 <main+0x28>)
817e:   6013            str     r3, [r2, #0]
8180:   bdf0            pop     {r4, r5, r6, r7, pc}
8182:   46c0            nop                     ; (mov r8, r8)
8184:   00011b04        .word   0x00011b04


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42501