[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.4.5                       |4.4.6

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com
      Known to fail|                            |

--- Comment #7 from Jeffrey A. Law <law at redhat dot com> 2011-01-19 20:15:32 UTC ---
Looking at it

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot       |law at redhat dot com
                   |gnu.org                     |

--- Comment #8 from Jeffrey A. Law <law at redhat dot com> 2011-01-20 00:49:46 UTC ---

This problem only exposes itself when there are uses of uninitialized variables
and -fno-ira-share-save-slots is enabled.   In this circumstance the reg_info
structure may claim that a pseudo is not live across a  call.  However, if the
pseudo is used after the call in an initialized state, then the reload chains
will claim the pseudo is live across the call.

The inconsistency in those data structures ultimately leads to an inconsistency
in setup_save_areas and save_call_clobbered_regs which triggers the ICE.

I think the way to go is use the same code regardless of
-fno-ira-share-save-slots in setup_save_areas to determine which hard regs need
saving.  A quick test shows that works.

I'm going to have to clean up that preliminary patch and do some more rigorous
testing, including regtesting with -fno-ira-share-save-slots.

Jeff

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

--- Comment #9 from Jeffrey A. Law <law at redhat dot com> 2011-01-21 16:45:36 UTC ---
http://gcc.gnu.org/ml/gcc-patches/2011-01/msg01424.html

After some brief discussion with rth (concerns about global_regs) in IRC, rth
approved the patch.

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[law at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

--- Comment #10 from Jeffrey A. Law <law at gcc dot gnu.org> 2011-01-21 16:49:34 UTC ---
Author: law
Date: Fri Jan 21 16:49:31 2011
New Revision: 169095

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=169095
Log:

    PR rtl-optimization/41619
    * caller-save.c (setup_save_areas): Break out code to determine
    which hard regs are live across calls by examining the reload chains
    so that it is always used.
    Eliminate code which checked REG_N_CALLS_CROSSED.

    PR rtl-optimization/41619
    * gcc.dg/pr41619.c: New.


Added:
    trunk/gcc/testsuite/gcc.dg/pr41619.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/caller-save.c
    trunk/gcc/testsuite/ChangeLog

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[law at redhat dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|4.4.6                       |4.6.0

--- Comment #11 from Jeffrey A. Law <law at redhat dot com> 2011-01-21 16:50:09 UTC ---
Fixed

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[ebotcazou at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

--- Comment #12 from Eric Botcazou <ebotcazou at gcc dot gnu.org> 2011-05-26 12:43:10 UTC ---
Author: ebotcazou
Date: Thu May 26 12:43:06 2011
New Revision: 174278

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=174278
Log:
    Backport from mainline
    2011-01-21  Jeff Law  <law@redhat.com>

    PR rtl-optimization/41619
    * caller-save.c (setup_save_areas): Break out code to determine
    which hard regs are live across calls by examining the reload chains
    so that it is always used.
    Eliminate code which checked REG_N_CALLS_CROSSED.

Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/caller-save.c

[Bug c++/41619] New: ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[kenneth dot hoste at elis dot ugent dot be]

While doing a bunch of experiments with randomly constructed sets of
optimization flags, I ran into the following compilation error while trying to
build SPEC CPU2000's 252.eon:

internal compiler error: in insert_save, at caller-save.c:1122

The problem occurs during compilation of various of the benchmark's C++ source
files, mainly for  mrSpotAreaXYDiskLuminaire.cc and mrGrid.cc, while using any
of the following combination of flags:

-O1 -fno-ira-share-save-slots -fcaller-saves 
-O2 -fno-ira-share-save-slots 
-O3 -fno-ira-share-save-slots 
-Os -fno-ira-share-save-slots 

I've created a significantly reduced testcase using delta (attached).

Reproducing the error should be as simple as the line below, using g++ 4.4.1:

g++ -O2 -fno-ira-share-saved-spill-slots testcase.ii


-- 
           Summary: ICE in insert_save (caller-save.c) for SPEC CPU2000's
                    252.eon
           Product: gcc
           Version: 4.4.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: kenneth dot hoste at elis dot ugent dot be
  GCC host triplet: x86_64-unknown-linux-gnu
GCC target triplet: x86_64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619

[Bug rtl-optimization/41619] [4.4/4.5/4.6 regression] ICE in insert_save (caller-save.c) for SPEC CPU2000's 252.eon

[jakub at gcc dot gnu dot org]

-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.4.4                       |4.4.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=41619