Re: >memory problems
Marcus (giraffe.asd.sgi.com!sgi.sgi.com!uunet.uu.net!multigen!Marcus)
Wed, 31 May 1995 10:19:36 PST
Reply to: RE>>memory problems
>Message-Id: <199505302357.QAA23013++at++surreal.asd.sgi.com>
>To: robin++at++media.mit.edu
>Cc: info-performer++at++sgi.com
>Subject: Re: memory problems
>In-Reply-To: Your message of "Tue, 30 May 95 18:34:42 EDT."
> <9505302234.AA06095++at++media.mit.edu>
>Date: Tue, 30 May 95 16:57:13 -0700
>From: Jim Helman <jimh++at++surreal.asd.sgi.com>
>
>There are no fixed limits on database size in Performer,
>except for our default arena size of 256MB which should
>be adequate for most scenes.
>
>> why would it work when drawing 100 items, but not 200?
>
>It could be that something moves in memory so that a read of
>an unitialized varialble or a write to an incorrect location
>causes no problem with one database, but drops core with
>another.
>
>Check your code for unitialized variables and try running dbx
>on the resulting core dump to see where the problem occurs.
>
>-jim
This sounds very much like the Performer 1.2 pfHashTable
internal memory corruption bug from last summer.
A workaround Jim gave me was put into revision R14.1c,
and later, of the OpenFlight loader.
It was as follows:
{
extern void** _pfNameTable;
_pfNameTable[ 2 ] = ( void* ) 0; /* clear "numResets" */
}
This stub of code must be called fairly frequently while
building a large scene graph with pfNodeName's.
Regards,
Marcus Barnes, Member Technical Staff
MultiGen Inc., 1884 The Alameda, San Jose CA, 95126
PH: 1-408-261-4118 FX: 1-408-247-4329
EMAIL: multigen!marcus++at++uunet.UU.NET
This archive was generated by hypermail 2.0b2
on Mon Aug 10 1998 - 17:51:33 PDT