Bug 22392 - Huge memory usage and infinite(?) loop with -fno-enforce-eh-specs
Summary: Huge memory usage and infinite(?) loop with -fno-enforce-eh-specs
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: rtl-optimization (show other bugs)
Version: 4.0.2
: P2 normal
Target Milestone: ---
Assignee: Not yet assigned to anyone
URL:
Keywords: compile-time-hog, memory-hog
Depends on:
Blocks: 20826
  Show dependency tree / graph
 
Reported: 2005-07-10 12:00 UTC by Debian GCC Maintainers
Modified: 2007-04-28 14:21 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work:
Known to fail:
Last reconfirmed:

Attachments
Test case (200.21 KB, application/octet-stream)
2005-07-10 12:01 UTC, Debian GCC Maintainers 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Debian GCC Maintainers 2005-07-10 12:00:41 UTC 
[ forwarded from http://bugs.debian.org/317563 ]

The test case chews up 789m of memory and then doesn't seem to progress
for at least 10 minutes with -O2 -fno-enforce-eh-specs. Functions showing up
prominently in a profile are:

samples  %        symbol name
34188    22.3480  bitmap_set_bit
18170    11.8774  free_deps
17475    11.4231  sched_analyze_insn
14018     9.1633  alloc_INSN_LIST
10173     6.6499  free_list
6100      3.9874  free_INSN_LIST_list
3946      2.5794  bitmap_bit_p
2442      1.5963  build_tree_conflict_graph

With just -O2, it finishes in 1:09 minute using 341m. With g++ 4.1, it finishes
in 1:21 using 148m even with -fno-enforce-eh-specs. g++-3.4 doesn't like this
test case.

This is on alphaev68-linux. The original report was on powerpc.
Comment 1 Debian GCC Maintainers 2005-07-10 12:01:40 UTC 
Created attachment 9237 [details]
Test case
Comment 2 Andrew Pinski 2005-07-10 18:12:28 UTC 
This is the scheduler messing up which is normal with big BBs.
Comment 3 Andrew Pinski 2005-07-25 01:53:03 UTC 
I think this is the same bug as PR 20826.
Comment 4 Falk Hueffner 2007-04-28 14:21:07 UTC 
Not a problem in 4.1 anymore, so let's close it.