ali@polya.STANFORD.EDU (Ali Ozer) (03/08/88)
[] If I have the following two statements in my startup-sequence: popcli setfont pearl 9 -s -r And I let my machine sit after rebooting until popcli blanks the screen, my chip memory is massively fragmented. We're talking: BEFORE: Max Chip 397K, largest 396K AFTER: Max Chip SAME, largest 1300 bytes This problem was first occuring with a full startup-sequence (with Conman .99B, Facc II with 512K of buffers, 1 Meg VD0:, and various other stuff), and it took me a while to localize it to the above two statements. The problem will occur even if the above two are the only two statements in the startup-sequence file and I do absolutely nothing else. This is on 1000 with 2.5 Megs of memory. By the way, I should note that the problem did *not* occur with Topaz 11, but did occur with ruby 8... And it did occur with many different configs --- VD0: mounted, filled, no VD0:, with most of fast mem filled, etc. By the way, sometimes the fragmentation will not occur for a long while (if the PopCLI doesn't get the chance to blank out the screen). But, even hours later, as soon as PopCLI does it's job, >blam<, you're left with tiny chunks of memory. I believe setfont is the latest version (2.0?), and PopCLI is version III. I haven't hit this problem without PopCLI (but I didn't really try very hard). I was going to try Tom's PopCLI ("Mackie"), except all my copies were at the office for some reason (Murphy's Law). I should also mention that while I was playing around with (large) fonts a couple of months ago (for TeXF, which never appeared on comp.sources by the way, sigh), I often encountered massive CHIP memory fragmentation. I always blamed it on my program (which was still buggy). But it seems like there's something wrong with OpenDiskFont(). Any words of wisdom will be appreciated. I will post more info if I discover more specifics. Ali Ozer, ali@polya.stanford.edu
page@swan.ulowell.edu (Bob Page) (03/12/88)
ali@polya.UUCP (Ali Ozer) wrote: >popcli >setfont pearl 9 -s -r >my chip memory is massively fragmented. We're talking: Interesting ... I've seen the same problem, but had not tracked it down. I also use setfont (pearl2, w/o options) but not popcli, and my fragmentation was appearing BEFORE setfont. Turns out (as near as I can determine) that is was caused by gamma 1 of the FastFileSystem. Once I went to a later version it went away. My startup-sequence was something like: Mount wb: (wb: is a 2MB partition running ffs) wb:s/DefDisk (reassign logicals) execute s:Startup-Sequence And wb:s/startup-sequence was: mount dh0: [etc] I couldn't mount dh0: because memory was too fragmented. The largest fragment I had was about 440 bytes (out of 1MB fast mem). Maybe you can run SNOOP and find out who's allocating memory. ..Bob -- Bob Page, U of Lowell CS Dept. page@swan.ulowell.edu ulowell!page "I don't know such stuff. I just do eyes." -- from 'Blade Runner'