Date: Wed, 11 Feb 2015 13:50:10 +0200 From: Beeblebrox <zaphod@berentweb.com> To: freebsd-current@freebsd.org Cc: Hans Petter Selasky <hps@selasky.org>, Poul-Henning Kamp <phk@phk.freebsd.dk> Subject: Re: Some unresolved but important X.org problems Message-ID: <20150211135010.0626605d@rsbsd.rsb> In-Reply-To: <20907.1423654172@critter.freebsd.dk> References: <20150211132039.42665673@rsbsd.rsb> <20907.1423654172@critter.freebsd.dk>
next in thread | previous in thread | raw e-mail | index | archive | help
Hello. > The drm2.ko graphics driver has some hard spinning loops, which may > be the cause of the problems you are seeing and not > actually /dev/ums0. Are any error messages printed in "dmesg" ? I have all debug disabled in my kernel (no WITNESS, no GDB/DDB), so no mess= ages. If you think it's worth wile, I'll build a fresh world and a full DEB= UG enabled kernel or I could build graphics/drm with the debug option? > evince crashes for me about 1 out of two but files with purely > images seems to fare slightly better than others. > people have pointed fingers in the general direction of gtk/cairo I was one of those finger pointers as well, since all apps would display an= y/all text as little squares then immediately crash; unless cairo was speci= fically built with the WITH_DEBUG=3Dyes option. IDK why that worked really,= it just did. =20 > I'm running firefox 35 on current and that seems to be working OK. I get same results as I described for Seamonkey. =20 > My guess is that your problems are specific to your graphics > hardware, and therefore it could be valuable if you could try, > just as an experiment, to move your diskdrive to different hardware, > just to see if the problems comes along for the ride. =20 I have diskless environment set up and serving from a jail (NFS not jailed = obviously), so I try your suggestion with any pxe_boot capable machine. Unf= ortunately the available clients are either RS780/RS880 family, have no GPU= , or i386. I'll have to get creative to test your idea. However, I have cor= responded with JS Pedron several times before, so he is aware of the issue(= s) and he did not seem to think the problem was with Radeon drivers. I also collected "loud debug" output in the past from the Kernel, without g= etting much of a result. Which makes me think the debug should be set for t= he application level and not the Kernel level. The problem is, how to log a= ll that output from 5-10 different apps all at the same time? (plus would i= t be worth the exercise?) Regards. --=20 FreeBSD_amd64_11-Current_RadeonKMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150211135010.0626605d>