Date: Mon, 03 May 2010 09:56:15 -0500 From: Robert Noland <rnoland@FreeBSD.org> To: Peter Jeremy <peterjeremy@acm.org> Cc: x11@FreeBSD.org, Martin Wilke <miwi@FreeBSD.org> Subject: Re: [HEADS UP] Xorg 7.5 merge comming tomorrow. Message-ID: <4BDEE40F.9080503@FreeBSD.org> In-Reply-To: <20100503061344.GA98887@server.vk2pj.dyndns.org> References: <20100430183522.GD64008@bsdcrew.de> <20100503061344.GA98887@server.vk2pj.dyndns.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Peter Jeremy wrote: > [Cc list reduced to x11- only] > > On 2010-Apr-30 20:35:23 +0200, Martin Wilke <miwi@FreeBSD.org> wrote: >> Now the facts. Xorg 7.5 is completely finished last issues >> were fixed by fluffy@, and the exp-run was also fine, I expect >> the merge of xorg 7.5 tomorrow evening UTC. I think the KDE > > I have upgraded my dual-screen testbox with a up-to-date 8-stable and > ports tree and built xorg from scratch and (as I feared), it's still > broken: There is no change from the problems I reported regarding the > test release in mid-March. > > To re-iterate, on a RV380 with dual screens laid out side-by-side, > moving a window into the RH screen causes corruption of both screens. > The RH screen updating is also far slower then the LH screen (which > seems fairly normal). Is all of this configured using randr? With randr, both displays are a single framebuffer, so from a performance perspective, there is no difference in the displays. The other part of your issue sounds like possibly an exceeded texture limit, or mismapping somehow. > I've done some more experimenting with "interesting" results: > > With the screens vertically stacked, the top ~1/4 of the bottom screen > behaves normally, but anything below that is corrupted (looks like > noise). When I turn off RenderAccel, the corruption turns into > alternating black and white horizontal bands about 5px high. In either > case, the update rate appears normal. You are using EXA, right? robert. > If I set NoAccel, I get > (EE) RADEON(0): Acceleration required for rotation > (which is surprising since I haven't specified acceleration anywhere). > > I had a quick look into the memory allocation since the reported > mappings and sizes overlapped - but it turns out this is just bugs > in the printf()'s code. > > At this stage, I am stymied. I'm not even sure where to start > looking at the code. >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BDEE40F.9080503>