[Bug c++/22380] New: [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased

[Bug c++/22380] New: [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased

[jsm28 at gcc dot gnu dot org]

On hppa2.0w-hp-hpux11.11, on mainline between 20050627 and 20050630 both at
07:00 UTC, the failure

FAIL: g++.dg/opt/inline4.C (test for excess errors)

appeared, the error being compile-time stack overflow.

http://gcc.gnu.org/ml/gcc-regression/2005-06/msg00121.html
shows the same failure appearing on Darwin.  I can reproduce the problem on
i686-pc-linux-gnu if I use "ulimit -s 4096", while a 4.0 branch compiler works
on the same testcase with "ulimit -s 256".  I don't know how much stack the test
needed just before the failure appeared.

-- 
           Summary: [4.1 Regression] stack usage of g++.dg/opt/inline4.C
                    increased
           Product: gcc
           Version: 4.1.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jsm28 at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22380

[Bug middle-end/22380] [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased (GC with blocks)

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-08 22:46 -------
Confirmed,  the stack is overflowing when doing GC, marking the BLOCKs.  This test regressed when 
the early inliner came in.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
          Component|c++                         |middle-end
     Ever Confirmed|                            |1
           Keywords|                            |ice-on-valid-code
   Last reconfirmed|0000-00-00 00:00:00         |2005-07-08 22:46:37
               date|                            |
            Summary|[4.1 Regression] stack usage|[4.1 Regression] stack usage
                   |of g++.dg/opt/inline4.C     |of g++.dg/opt/inline4.C
                   |increased                   |increased (GC with blocks)
   Target Milestone|---                         |4.1.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22380

[Bug middle-end/22380] [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased (GC with blocks)

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-31 00:11 -------
Hmm, we leak all the BLOCKs created by the tree-inline:
tree-inline.c:421 (remap_block)                           0: 0.0%          0: 0.0%    3166956:74.8%          0: 0.0%      
60903
tree-inline.c:2027 (expand_call_inline)                   0: 0.0%          0: 0.0%      10452: 0.2%          0: 0.0%        
201


that is with 200 instead of 250 so we don't get a seg fault in GC when collecting it for the last time.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22380

[Bug middle-end/22380] [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased (GC with blocks)

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-07-31 00:26 -------
I was wrong in comment #1 for saying this was caused by the early inliner.
This was caused by: "Copy debug info properly in the inliner":
http://gcc.gnu.org/ml/gcc-patches/2005-06/msg02022.html

What is happening is that we are getting more BLOCKS which seems to link to each other which is 
weird.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22380

[Bug middle-end/22380] [4.1 Regression] stack usage of g++.dg/opt/inline4.C increased (GC with blocks)

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2005-10-02 15:49 -------
Fixed.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22380