--- Comment #16 from hosking at cs dot purdue dot edu 2006-08-31 05:37
---
It turns out there were problems with aliasing in the trees generated by the
Modula-3 front end. There are multiple places where aliasing is used to
loophole values from one type to another, and this was
--- Comment #15 from hosking at cs dot purdue dot edu 2006-08-29 15:52
---
Sorry, my information was in error last time. It seems I got the src/dest
information wrong for the edge being split. The edge is between blocks 32 and
33 in the dump below:
;; Function TrestleGoo__PutProp
--- Comment #13 from hosking at cs dot purdue dot edu 2006-08-24 15:27
---
Is this enough?
Here is the dump output, followed by stack traces at the resize and remove
points (the remove goes on to fail). The edge being split is:
(gdb) p *e
$9 = {
src = 0x425cc8d0,
dest
--- Comment #11 from hosking at cs dot purdue dot edu 2006-08-24 00:57
---
(In reply to comment #9)
> Does 4.1 have the check for EDGE_CRITICAL_P in insert_aux?
Yes:
/* This can happen in the very weird case
that our f
--- Comment #8 from hosking at cs dot purdue dot edu 2006-08-23 23:43
---
I can send whatever traces you might need for diagnosis.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28798
--- Comment #7 from hosking at cs dot purdue dot edu 2006-08-23 22:29
---
This is with the Modula-3 backend. I am porting it to 4.1.1 and encountered
this problem with -O3 turned on.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=28798
--- Comment #4 from hosking at cs dot purdue dot edu 2006-08-23 13:40
---
Here is a stack trace showing call to resize_phi_node from execute_pre.
#0 resize_phi_node (phi=0x42580028, len=9) at
../../gcc/gcc/tree-phinodes.c:271
#1 0x001415e8 in reserve_phi_args_for_new_edge (bb
Priority: P3
Component: tree-optimization
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: hosking at cs dot purdue dot edu
GCC build triplet: i386-apple-darwin8.7.1
GCC host triplet: i386-apple-darwin8.7.1
GCC target triplet: i386-apple-darwin8.7
