bug 2975, eh and renumber_insns

[Nathan Sidwell]

Hi,
bug 2975 is a high priority -O3 ice. In this analysis I'm refering to an
i686-pc-linux-gnu build of the 3.0 CVS taken on 20010527

The problematic function is MainWindow::MainWindow(bool) ending at line 15041.

This happens at loop.c:2556 (find_and_verify_loops).
here we have a label with insn uid 31032, whereas earlier on at loop.c:391
(loop_optimize), get_max_uid returned 12805. This lower bound is set at
emit-rtl.c:2139 (renumber_insns). However, the label with uid 31032
is never renumbered, because it is not on the instruction chain.
The label is generated in except.c (dw2_build_landing_pads)

My brain if now full, and I'm out of my depth, I wish to be excused.

Richard, I have no idea whether the new eh scheme introduced this or not, but
I guess you're the most appropriate to look at this -- please forward elsewhere
if not.

nathan
-- 
Dr Nathan Sidwell   ::   http://www.codesourcery.com   ::   CodeSourcery LLC
         'But that's a lie.' - 'Yes it is. What's your point?'
nathan@codesourcery.com : http://www.cs.bris.ac.uk/~nathan/ : nathan@acm.org
2975.ii.bz2

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 2975.ii.bz2
Type: application/x-bzip2
Size: 81191 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-bugs/attachments/20010528/737a9507/attachment.bz2>