[Bug middle-end/46555] New: PHI RTL expansion leads to CSiBE regression

[Bug middle-end/46555] New: PHI RTL expansion leads to CSiBE regression

[hubicka at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

           Summary: PHI RTL expansion leads to CSiBE regression
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: middle-end
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: hubicka@gcc.gnu.org


Created attachment 22452
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22452
testcase (OpenTCP-1.0.4/icmp.c)

Hi,
the problem here seems to be worse regalloc and also
  # D.4060_6 = PHI <-1(2), -1(9), -1(11), -1(14), 0(15), -1(10)>
used to be optimized into since set of var to -1 (4 bytes), while now we
produce 3 different copies. 

Crossjumping would unify it, but very late in the game. The problem is that
ifcvt actually moves the set before conditoinal guarding the BB in question, so
the individual sets are drifted earlier to different places in the program.

Doing so might also complicate the regalloc.

Michael, perhaps we can tell out-of-ssa to unify such cases?  They are not that
infrequent (and I think old tree based out-of-ssa did that?)

[Bug middle-end/46555] PHI RTL expansion leads to CSiBE regression

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2010.11.19 10:55:11
     Ever Confirmed|0                           |1

--- Comment #1 from Richard Guenther <rguenth at gcc dot gnu.org> 2010-11-19 10:55:11 UTC ---
Yep.  That's one optimization that was removed (out-of-SSA did that) and
we thought of doing this reverse mergephi optimization as a separate pass
before expansion.

[Bug middle-end/46555] [9/10/11/12 Regression] PHI RTL expansion leads to CSiBE regression

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|8.5                         |9.4

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 8 branch is being closed.

[Bug middle-end/46555] [9/10/11/12 Regression] PHI RTL expansion leads to CSiBE regression

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|9.4                         |9.5

--- Comment #13 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 9.4 is being released, retargeting bugs to GCC 9.5.

[Bug middle-end/46555] [9/10/11/12 Regression] PHI RTL expansion leads to CSiBE regression

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Last reconfirmed|2016-01-05 00:00:00         |2021-9-5

--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Still happening on the trunk:
(code_label 164 163 165 24 2 (nil) [5 uses])
(note 165 164 11 24 [bb 24] NOTE_INSN_BASIC_BLOCK)
(insn 11 165 180 24 (set (reg:HI 110 [ <retval> ])
        (const_int -1 [0xffffffffffffffff]))
"/home/jan/Downloads/CSiBE/src/./OpenTCP-1.0.4/icmp.c":112:9 -1
     (nil))
(jump_insn 180 11 181 24 (set (pc)
        (label_ref:SI 170)) 842 {jump}
     (nil)
 -> 170)
(barrier 181 180 183)
(code_label 183 181 182 25 14 (nil) [1 uses])
(note 182 183 12 25 [bb 25] NOTE_INSN_BASIC_BLOCK)
(insn 12 182 170 25 (set (reg:HI 110 [ <retval> ])
        (const_int -1 [0xffffffffffffffff]))
"/home/jan/Downloads/CSiBE/src/./OpenTCP-1.0.4/icmp.c":155:2 -1
     (nil))

[Bug middle-end/46555] [10/11/12/13 Regression] PHI RTL expansion leads to CSiBE regression

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|9.5                         |10.4

--- Comment #15 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 9 branch is being closed

[Bug middle-end/46555] [10/11/12/13 Regression] PHI RTL expansion leads to CSiBE regression

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|10.4                        |10.5

--- Comment #16 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 10.4 is being released, retargeting bugs to GCC 10.5.

[Bug middle-end/46555] [10/11/12/13 Regression] PHI RTL expansion leads to CSiBE regression

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

--- Comment #17 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Looks like this was fixed in GCC 12:
  <bb 20> [local count: 344921666]:
  # _30 = PHI <-1(12), 0(17)>
  tbuf ={v} {CLOBBER(eol)};
  return _30;

[Bug middle-end/46555] [10/11/12/13/14 Regression] PHI RTL expansion leads to CSiBE regression

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

--- Comment #18 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #1)
> Yep.  That's one optimization that was removed (out-of-SSA did that) and
> we thought of doing this reverse mergephi optimization as a separate pass
> before expansion.

make_forwarders_with_degenerate_phis in tree-ssa-dce.cc does basically (if I
read the code correctly) that which was added by r12-5301-g045206450386 for PR
102880 which is why I am seeing it fixed for GCC 12 at -O2. And the reason why
-O1 I still see the old PHI behavior.

[Bug middle-end/46555] [11/12/13/14 Regression] PHI RTL expansion leads to CSiBE regression

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=46555

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|10.5                        |11.5

--- Comment #19 from Richard Biener <rguenth at gcc dot gnu.org> ---
GCC 10 branch is being closed.