From owner-freebsd-current Fri Apr 3 11:53:11 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id LAA05905 for freebsd-current-outgoing; Fri, 3 Apr 1998 11:53:11 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from critter.freebsd.dk (critter.freebsd.dk [195.8.129.14]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id LAA05890 for ; Fri, 3 Apr 1998 11:53:04 -0800 (PST) (envelope-from phk@critter.freebsd.dk) Received: from critter.freebsd.dk (localhost [127.0.0.1]) by critter.freebsd.dk (8.8.7/8.8.5) with ESMTP id VAA03269; Fri, 3 Apr 1998 21:52:12 +0200 (CEST) To: =?koi8-r?B?4c7E0sXKIP7F0s7P1w==?= cc: "Alok K. Dhir" , current@FreeBSD.ORG, adkin003@tc.umn.edu Subject: Re: Working patch *with* splhigh() (Was Re: More info RE: X slowdown in -current) In-reply-to: Your message of "Fri, 03 Apr 1998 23:32:42 +0400." <19980403233242.31917@nagual.pp.ru> Date: Fri, 03 Apr 1998 21:52:12 +0200 Message-ID: <3267.891633132@critter.freebsd.dk> From: Poul-Henning Kamp Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG >> So what exactly is the problem you're seing (any URLs I should try ?) > >You need no URLs, when Netscape4-communicator starts it draws its icon >bar. With the bug it draws it at 1icon per 2secs rate. Not for me :-( >Next example is xterm scrolling, usually showed with 'ls -R'. Not for me :-( >When you move your mouse intensively, it helps to draw faster since >generate more events. I use /dev/sysmouse (syscons mouse). My xserver is remote, but it still doesn't make any difference :-( >Something the picture locks completely until you touch mouse. not here. >> Does "systat :vm" show abnormally many syscalls ? > >I not check. please do. >> Does ktrace on your netscape show anything weird ? >No. > >> Could it maybe be the Xserver which is having the trouble, not the >> netscape process ? > >No, when I run Netscape remotely there is no slowness. Dang. I can't reproduce it here. I know the problem you describe, and I specifically tested with netscape before I committed 1.35 :-( The problem is that time must somehow become zero when we reach tsleep() where it should have been non-zero. I'd be damned if I can see how that can happen. The code looks like this: s = splhigh(); if (term && term <= ticks) { splx(s); goto done; } timo = term ? term - ticks : 0; The splhigh location is important btw, to protect against just this problem (ie, ticks changing between the two uses of it.) Anybody else see this problem with version 1.35 ? BTW: The current code is not really what I plan on, but I'm hashing some ideas out right now with Bruce, which might be even better than the code I have running here right now... -- Poul-Henning Kamp FreeBSD coreteam member phk@FreeBSD.ORG "Real hackers run -current on their laptop." "Drink MONO-tonic, it goes down but it will NEVER come back up!" To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message