[Bug tree-optimization/16049] New: -ftree-points-to=andersen moans about "Inconsistent system of constraints"

[Bug tree-optimization/16049] New: -ftree-points-to=andersen moans about "Inconsistent system of constraints"

[rguenth at tat dot physik dot uni-tuebingen dot de]

Compiling the tramp3d-v3 testcase with -O[12] -ftree-points-to=andersen results in

> ../gcc-3.5-180604/bin/g++ -O1 -o tramp3d-v3.leafify.tune -static
../src/tramp3d-v3.cpp -ftree-points-to=andersen
Inconsistent system of constraints

g++ (GCC) 3.5.0 20040617 (experimental)

(I don't have a recent ia32 build, but maybe someone can check if the problem is
architecture dependend).

The tramp3d-v3.cpp testcase can be found (as usual) at
http://www.tat.physik.uni-tuebingen.de/~rguenth/gcc/tramp3d-v3.cpp.gz

It looks like the failure is in the compiler-generated copy-constructor
CopyPatchEvent::CopyPatchEvent(const CopyPatchEvent&) from -v output:
 [...]
 void DynamicLayoutData::copy(constIndirectionList<IndirectionList<int> >&,
constIndirectionList<int>&, int, bool)
 CopyPatchEvent::CopyPatchEvent(constCopyPatchEvent&)
 CopyPatchEvent::CopyPatchEvent(constCopyPatchEvent&)
 CopyPatchEvent::CopyPatchEvent(constCopyPatchEvent&)
Inconsistent system of constraints

As usual, debugging on ia64 is not very helpful:

Breakpoint 2, failure (message=0x4000000000995bf8 "Inconsistent system of
constraints\n")
    at ../../../src/gcc-3.5/libbanshee/engine/util.c:81
81	  fprintf(stderr,message);
(gdb) bt
#0  failure (message=0x4000000000995bf8 "Inconsistent system of constraints\n")
    at ../../../src/gcc-3.5/libbanshee/engine/util.c:81
#1  0x40000000007bf590 in setif_inclusion (con_match=@0x400000000099ac30:
0x40000000007b63f0 <label_term_con_match>, 
    res_proj=@0x400000000099a940: 0x40000000007b63e0 <label_term_res_proj>,
e1=0x200000000735f510, e2=0x200000000735aa90)
    at ../../../src/gcc-3.5/libbanshee/engine/setif-sort.c:472
#2  0x40000000007b5f60 in label_term_inclusion (e1=0x200000000735f510,
e2=0x200000000735aa90)
    at ../../../src/gcc-3.5/libbanshee/points-to/andersen_terms.c:222
#3  0x40000000007bb3f0 in aterm_con_match (arg1=0x200000000735f528,
arg2=0x200000000735aaa8)
    at ../../../src/gcc-3.5/libbanshee/points-to/andersen_terms.c:1265
#4  0x40000000007bfc90 in setif_inclusion (con_match=@0x400000000099b650:
0x40000000007bb1f0 <aterm_con_match>, 
    res_proj=@0x400000000099c390: 0x40000000007baed0 <aterm_res_proj>,
e1=0x200000000735f528, e2=0x200000000735aaa8)
    at ../../../src/gcc-3.5/libbanshee/engine/setif-sort.c:571
#5  0x40000000007bf150 in update_upper_bound (v=0x200000000735fa70,
e=0x200000000735aaa8)
    at ../../../src/gcc-3.5/libbanshee/engine/setif-sort.c:455
#6  0x0000000000000000 in ?? ()
Previous frame identical to this frame (corrupt stack?)
(gdb)

-- 
           Summary: -ftree-points-to=andersen moans about "Inconsistent
                    system of constraints"
           Product: gcc
           Version: 3.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rguenth at tat dot physik dot uni-tuebingen dot de
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: ia64-unknown-linux-gnu
  GCC host triplet: ia64-unknown-linux-gnu
GCC target triplet: ia64-unknown-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16049

[Bug tree-optimization/16049] -ftree-points-to=andersen moans about "Inconsistent system of constraints"

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2004-06-18 10:26 -------
I'm seeing this, too, on Alpha, with lots of test cases (basically anything).
So it seems it's either a 64 bit bug, or a problem with nested functions, which
are extensively used in the libbanshee code, including thunk generation 
(engine/flowrow-sort.c:flowrow_extract_field).

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16049

[Bug tree-optimization/16049] -ftree-points-to=andersen moans about "Inconsistent system of constraints"

[rguenth at tat dot physik dot uni-tuebingen dot de]

------- Additional Comments From rguenth at tat dot physik dot uni-tuebingen dot de  2004-09-14 13:25 -------
points-to=andersen was removed, so closing as "INVALID".

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16049