[Bug target/15184] [3.4/4.0/4.1 Regression] Direct access to byte inside word not working with -march=pentiumpro

[Bug target/15184] [3.4/4.0/4.1 Regression] Direct access to byte inside word not working with -march=pentiumpro

[wilson at gcc dot gnu dot org]

------- Comment #9 from wilson at gcc dot gnu dot org  2005-10-14 17:44 -------
The cause of this problem is the following two lines in the i386.c file
const int x86_himode_math = ~(m_PPRO);
const int x86_promote_hi_regs = m_PPRO;

They were added here:
    http://gcc.gnu.org/ml/gcc-patches/2000-02/msg00890.html

The reason for this is, as a previous comment mentioned, that HImode
instructions are slow on the pentiumpro and should be avoided.  Doing this
gives better performance in general, but unfortunately, for this particular
testcase, it causes us to miss an optimization.

The issue in this case is a combiner limit.  If you compile for Pentium, you
get
  (set (reg:HI 61) (and:HI (mem/c/i:HI (symbol_ref:SI ("y")))
                           (const_int -256)))
  (set (reg:HI 63) (ior:HI (reg:HI 61)
                           (reg:HI 62)))
  (set (mem/c/i:HI (symbol_ref:SI ("y"))
       (reg:HI 63))
The combiner combines these 3 instructions to get
  (set (mem/c/i:QI (const:SI (plus:SI (symbol_ref:SI ("x"))
                                      (const_int 1 [0x1]))))
       (subreg:QI (reg:SI 59 [ c ])))

However, for pentium pro, we end up with 4 instructions due to the HImode
promotion.
  (set (reg:HI 61 [ y ])
       (mem/c/i:HI (symbol_ref:SI ("y"))))
  (set (reg:SI 62) (and:SI (subreg:SI (reg:HI 61 [ y ]) 0)
                           (const_int -256 [0xffffffffffffff00])))
  (set (reg:SI 65) (ior:SI (reg:SI 62)
                           (subreg:SI (reg:HI 63 [ c ]) 0)))                    
  (set (mem/c/i:HI (symbol_ref:SI ("y"))
        (subreg:HI (reg:SI 65) 0))

The combiner combines at most 3 instructions, to avoid combinatorial explosion,
and hence we are not able to optimize this.

I'll look at this a bit more, but at the moment, I'm skeptical that there is
any easy solution.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [3.4/4.0/4.1 Regression] Direct access to byte inside word not working with -march=pentiumpro

[mmitchel at gcc dot gnu dot org]

------- Comment #10 from mmitchel at gcc dot gnu dot org  2005-10-30 22:26 -------
Downgraded to P5.

I don't think this missed-optimization on the Pentium Pro (now a relatively old
CPU) is release-critical, especially since we think that the change that caused
the problem was generally beneficial.


-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P2                          |P5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [3.4/4.0/4.1/4.2 Regression] Direct access to byte inside word not working with -march=pentiumpro

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.3                       |4.0.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.0/4.1/4.2 Regression] Direct access to byte inside word not working with -march=pentiumpro

[bonzini at gnu dot org]

------- Comment #11 from bonzini at gnu dot org  2006-06-08 08:36 -------
I would note, however, that Pentium Pro also means Pentium 2/3/M, Core, etc. 
In practice every Intel chip after the Pentium Pro, except the P4 and Nocona,
is based on that pipeline.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.0/4.1/4.2/4.3 Regression] Direct access to byte inside word not working with -march=pentiumpro

[gdr at gcc dot gnu dot org]

------- Comment #12 from gdr at gcc dot gnu dot org  2007-01-18 02:55 -------
Won't fix for GCC-4.0.x


-- 

gdr at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.0.4                       |---


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.0/4.1/4.2/4.3 Regression] Direct access to byte inside word not working with -march=pentiumpro

[pinskia at gcc dot gnu dot org]

-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.0/4.1/4.2/4.3 Regression] Direct access to byte inside word not working with -march=pentiumpro

[mmitchel at gcc dot gnu dot org]

-- 

mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.1.2                       |4.1.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.0/4.1/4.2/4.3 Regression] Direct access to byte inside word not working with -march=pentiumpro

[rask at sygehus dot dk]

------- Comment #13 from rask at sygehus dot dk  2007-07-01 19:52 -------
   This has regressed quite badly. Here's what we get with
-O2 -fomit-frame-pointer -m32 -march=pentium4:

fl:
        movzbl  %al, %eax       # 21    *zero_extendqisi2_movzbw        [length
= 3]
        movl    x, %edx # 8     *movsi_1/1      [length = 6]
        xorb    %dl, %dl        # 27    *movstrictqi_xor        [length = 2]
        orl     %edx, %eax      # 10    *iorsi_1/1      [length = 2]
        movl    %eax, x # 11    *movsi_1/2      [length = 5]
        ret     # 25    return_internal [length = 1]
fu:
        movzbl  %al, %eax       # 22    *zero_extendqisi2_movzbw        [length
= 3]
        sall    $8, %eax        # 8     *ashlsi3_1/1    [length = 3]
        movl    x, %edx # 9     *movsi_1/1      [length = 6]
        xorb    %dh, %dh        # 23    *xorqi_ext_2    [length = 2]
        orl     %edx, %eax      # 11    *iorsi_1/1      [length = 2]
        movl    %eax, x # 12    *movsi_1/2      [length = 5]
        ret     # 26    return_internal [length = 1]
gl:
        movzbl  %al, %eax       # 21    *zero_extendqihi2_movzbl        [length
= 3]
        movzwl  y, %edx # 8     *movhi_1/3      [length = 7]
        xorb    %dl, %dl        # 28    *movstrictqi_xor        [length = 2]
        orl     %edx, %eax      # 23    *iorsi_1/1      [length = 2]
        movw    %ax, y  # 11    *movhi_1/4      [length = 6]
        ret     # 26    return_internal [length = 1]
gu:
        sall    $8, %eax        # 22    *ashlsi3_1/1    [length = 3]
        movzbl  y, %edx # 23    *zero_extendqihi2_movzbl        [length = 7]
        orl     %edx, %eax      # 24    *iorsi_1/1      [length = 2]
        movw    %ax, y  # 12    *movhi_1/4      [length = 6]
        ret     # 27    return_internal [length = 1]

Likewise for -march=pentium:

fl:
        movl    x, %edx # 8     *movsi_1/1      [length = 6]
        andl    $255, %eax      # 21    *andsi_1/1      [length = 6]
        xorb    %dl, %dl        # 27    *movstrictqi_xor        [length = 2]
        orl     %edx, %eax      # 10    *iorsi_1/1      [length = 2]
        movl    %eax, x # 11    *movsi_1/2      [length = 5]
        ret     # 25    return_internal [length = 1]
fu:
        movl    x, %edx # 9     *movsi_1/1      [length = 6]
        andl    $255, %eax      # 22    *andsi_1/1      [length = 6]
        sall    $8, %eax        # 8     *ashlsi3_1/1    [length = 3]
        xorb    %dh, %dh        # 23    *xorqi_ext_2    [length = 2]
        orl     %edx, %eax      # 11    *iorsi_1/1      [length = 2]
        movl    %eax, x # 12    *movsi_1/2      [length = 5]
        ret     # 26    return_internal [length = 1]
gl:
        movw    y, %dx  # 8     *movhi_1/3      [length = 7]
        andl    $255, %eax      # 22    *andsi_1/1      [length = 6]
        xorb    %dl, %dl        # 29    *movstrictqi_xor        [length = 2]
        orl     %edx, %eax      # 24    *iorsi_1/1      [length = 2]
        movw    %ax, y  # 11    *movhi_1/4      [length = 6]
        ret     # 27    return_internal [length = 1]
gu:
        movb    %al, y+1        # 12    *movqi_1/7      [length = 5]
        ret     # 24    return_internal [length = 1]

        .ident  "GCC: (GNU) 4.3.0 20070628 (experimental)"


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.2/4.3/4.4 Regression] Direct access to byte inside word not working with -march=pentiumpro

[jsm28 at gcc dot gnu dot org]

------- Comment #14 from jsm28 at gcc dot gnu dot org  2008-07-04 16:31 -------
Closing 4.1 branch.


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.1/4.2/4.3/4.4 Regression]|[4.2/4.3/4.4 Regression]
                   |Direct access to byte inside|Direct access to byte inside
                   |word not working with -     |word not working with -
                   |march=pentiumpro            |march=pentiumpro
   Target Milestone|4.1.3                       |4.2.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.3/4.4/4.5 Regression] Direct access to byte inside word not working with -march=pentiumpro

[jsm28 at gcc dot gnu dot org]

------- Comment #15 from jsm28 at gcc dot gnu dot org  2009-03-31 16:17 -------
Closing 4.2 branch.


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.2/4.3/4.4/4.5 Regression]|[4.3/4.4/4.5 Regression]
                   |Direct access to byte inside|Direct access to byte inside
                   |word not working with -     |word not working with -
                   |march=pentiumpro            |march=pentiumpro
   Target Milestone|4.2.5                       |4.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.3/4.4/4.5 Regression] Direct access to byte inside word not working with -march=pentiumpro

[rguenth at gcc dot gnu dot org]

------- Comment #16 from rguenth at gcc dot gnu dot org  2009-08-04 12:26 -------
GCC 4.3.4 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.4                       |4.3.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184

[Bug target/15184] [4.3/4.4/4.5/4.6 Regression] Direct access to byte inside word not working with -march=pentiumpro

[rguenth at gcc dot gnu dot org]

------- Comment #17 from rguenth at gcc dot gnu dot org  2010-05-22 18:10 -------
GCC 4.3.5 is being released, adjusting target milestone.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.3.5                       |4.3.6


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15184