Date: Sat, 03 Nov 2012 11:14:55 +0100 From: Niclas Zeising <zeising@freebsd.org> To: "Ronald F. Guilmette" <rfg@tristatelogic.com> Cc: freebsd-x11@freebsd.org Subject: Re: r300-based boards... Are they now officially a lost cause? Message-ID: <5094EE9F.70806@freebsd.org> In-Reply-To: <92688.1351935814@tristatelogic.com> References: <92688.1351935814@tristatelogic.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 11/03/12 10:43, Ronald F. Guilmette wrote: > > > Greetings, > > I filed the following bugreport way back in mid-March.... ohhhh... > the Ides of March in fact! (Maybe THAT was the problem. :-) > > http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/166163 > > Anyway, as far as I can tell, nobody has even looked at this bug since > it was filed, but it does seem rather serious to me. > > Back then, when I filed the PR, I was on 8.2-RELEASE and the dri port > was at version 7.4.4. > > Now however, I have one of my machines running 9.1-RC2 and the current > (installed) version of dri on that machine is now dri-7.6.1_2,2. So > guess what... I'm still getting what looks to be the exact same SIGSEGV > at what looks like it is probably the exactly same place in r300_state.c. > (It is a little hard to tell if it is the exact same place because there > is now apparently some new brokenness in gdb that doesn't allow full > tracing into explicitly loaded dynamic libraries.) > > Based on this, you know, I am inclined to ask: Has support for r300 > based cards been abandoned, either officially or informally? > > I don't mean to get in anybody face about this, but from where I am > sitting, it sure does look like FreeBSD support for r300 based cards > has gone the way of the dinosaur. I sure would like to know the > real story however, if someone would be kind enough to clue me in. > > Also, could someone (anyone?) please explain all of this "Gallium" > stuff to me? I don't know squat about any of it. (When it comes > to all of this graphics stuff, I am just Joe Blow end-luser.) I just > now cd'd into the top of the work/ directory for the current dri port > and then did: > > find . -name r300\* -print > > What I found as a result of this sure looks to me like we've got two > whole and separate different flavors of r300 drivers in the DRI port > now. Is that correct? If so, which one am I using? If so, which one > _should_ I be using? And how do I go about selecting one or the other? > (Is either one less likley to SIGSEGV than the one one I'm currently > using?) > > OK, three final questions: > > 1) Would it really be just a pointless waste of time (for everybody, > including me) if I were to file a new PR for this same old bug, but now > specifying that the bug (still) exists in 9.1-RC2 + dri-7.6.1_2,2 ? > > 2) I get the impression that the "r300" drivers are used not just for > r300/x600 cards like the one I have, but also for r420/X800 cards. Is > that correct? (If so, I guuess that it would be a waste of my time and > money to try to solve this problem by simply purchasing an X800 card off > eBay to replace my existing X600 card, right? I need to stay at X800 or > below because the machine in question is setup to dual boot to Win2k > and the ATI-supplied Windoze drivers only support up to X800 on Win2k.) > > 3) Is there any lack of activity related to the r300 driver which might > possibly stem from a lack of hardware on the part of some specific developer > or developers? I ask because if that is the only problem, I'll be happy > to donate an X600 card to the cause and to the common good. > > > Regards, > rfg > > > P.S. The problems I've experienced with my X600 card are not limited to > just the gthumb SIGSEGV crash. Another problem I've seen (on 9.1-RC2 with > the laters DRI port) is that in XBMC when running the Aeon skin, the > background image goes absolutely bonkers at times. It's hard to describe, > but basically it looks like the colorized version of what would happen > on my family's old 1950's vintage Philco TV when my idiot brother would > try to adjust the antenna. > > As I say, it is hard to describe, but this is the closest I could find > to a picture of what I'm seeing (only colorized): > > http://image.shutterstock.com/display_pic_with_logo/511447/511447,1264449432,5/stock-photo-genuine-s-television-interference-and-ghosting-retrieved-from-a-inch-reel-from-that-time-45270595.jpg > > Note that the standard XBMC Confluence skin does now produce any such > anomolies. Hi! To try to answer some of your questions... The FreeBSD x11 team is currently only two members that work in it regularly. It is a big beast, and it grows more and more Linux-isms every day, making porting increasingly harder every day. The Linux people have a tendency to break APIs and KPIs every now and again, just to make porting hard. The reason that mesa hasn't been updated to the 8.0 branch in the ports tree is that this would kill support for numerous legacy graphics cards, and this would probably raise an outcry from many, at least in the past that has been the case when something in x11 land changes or breaks. There is also a distinct lack of hardware, at least on my part, so I can't test every graphics card under the sun to see if it still works. With that said, have you tried the newer xorg distribution, by setting WITH_NEW_XORG=yes in /etc/make.conf and recompile all xorg related ports? Does it work? Even better, can you also try the experimental ports repository, and see if that works? The latter will also give you mesa 8.0. More information can be found here: http://wiki.freebsd.org/Xorg With regards to gallium, I suggest you use google to find more information. Perhaps http://en.wikipedia.org/wiki/Gallium3D will give you some clues? Regards! -- Niclas Zeising
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5094EE9F.70806>