Issue 22932 - compiler optimisation runtime errors
Summary: compiler optimisation runtime errors
Status: CLOSED FIXED
Alias: None
Product: Calc
Classification: Application
Component: code (show other issues)
Version: OOo 1.1 RC5
Hardware: Sun Linux, all
: P3 Trivial (vote)
Target Milestone: ---
Assignee: ooo
QA Contact: issues@sc
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2003-11-27 10:18 UTC by sparcmoz
Modified: 2013-08-07 15:13 UTC (History)
1 user (show)

See Also:
Issue Type: DEFECT
Latest Confirmation in: ---
Developer Difficulty: ---

Attachments
fix runtime errors (1.21 KB, patch)
2003-11-27 10:19 UTC, sparcmoz 		no flags Details | Diff
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this issue.
Description sparcmoz 2003-11-27 10:18:45 UTC 
Attached are patches for calc for linux sparc.

sc/source/core/data/conditio.cxx
sc/source/filter/excel/xiescher.cxx

Runtime errors caused by compiling with optimisation.

These errors were found using qatesttool see issue 19750

patch made using cvs diff -u
in module sc
patch -p0 < sc.diff

Also removes a comment that was inserted by a previous patch.
Comment 1 sparcmoz 2003-11-27 10:19:53 UTC 
Created attachment 11583 [details]
fix runtime errors
Comment 2 frank 2003-11-27 10:39:46 UTC 
Hi Niklas,

one for you.

Frank
Comment 3 oc 2003-12-12 09:06:28 UTC 
Adjusting target milestone
Comment 4 niklas.nebel 2003-12-12 15:48:15 UTC 
Reassigning to Eike.
Not sure if there are really that many independent problems, or if there's a
common cause that should be found and fixed instead.
Comment 5 sparcmoz 2003-12-12 22:44:59 UTC 
i have a plan, but it is slow because i do not have much time. the debian
package for linux sparc is nearly ready and then we might get some more helpers
since i think debian is where most linux sparc users are lurking.
(a) these problems go away if i compile with no optimisation.
(b) -O1 is enough to cause these problems.
(c) i compile with -O3 hoping that will force more problems to be revealed.
(d) that might help find a common factor, 
(e) next step is for porting (me) to try and isolate the code where the problems
exist in each cases, i will try and do this in my xmas holiday
(f) meanwhile the current patch is needed to get calc through qatesttool
Comment 6 ooo 2003-12-15 16:08:15 UTC 
Patch accepted. However, as Niklas already stated, I also think there probably 
will be some common cause. The whole thing resembles much the
-fno-strict-aliasing problem we encountered (by enabling optimizations without
specifying that switch, which crashed in accessing the "casted offsets" of
ScAddress).

Btw: As all the other Linux SPARC patches went into some OOo1.1.1 branch,
shouldn't this one also go there? Currently it's targeted to OOo2.0
Comment 7 sparcmoz 2003-12-18 00:44:05 UTC 
jim@sun:~$ locale -a
C
en_US
en_US.iso88591
POSIX
jim@sun:~$
Comment 8 sparcmoz 2003-12-18 00:45:12 UTC 
sorry ignore that
Comment 9 ooo 2003-12-18 10:42:53 UTC 
Hi Jim,

Will do ;-)

However, what about my comment if OOo1.1.x wouldn't be the proper target instead
of OOo2.0?

Eike
Comment 10 sparcmoz 2003-12-18 10:59:46 UTC 
I would be happy to see this done in 1.1.x thank you. 
There is one more of these (ouch) at issue 23229, i am not sure the correct
steps to send it to you.
Comment 11 ooo 2003-12-18 12:07:19 UTC 
Align target with other Linux SPARC patches, OOo1.1.1
Comment 12 ooo 2003-12-18 12:24:56 UTC 
On branch cws_fix645_ooo111fix2
sc/source/core/data/makefile.mk  1.6.260.1.38.1.2.1
sc/source/filter/excel/makefile.mk  1.18.80.1.2.1.2.1
Comment 13 sparcmoz 2004-04-18 03:31:23 UTC 
verified in 112fix1, i will close this one and make a new issue some day to
investigate the optimisation problem.
Comment 14 sparcmoz 2004-04-18 03:31:49 UTC 
closed
Comment 15 misbirdcemu 2010-11-10 17:36:09 UTC 
Created attachment 73530