Martin Bochnig wrote: > On Fri, Feb 13, 2009 at 3:55 PM, John Martin <John.M.Martin at sun.com> wrote: > >> Martin Bochnig wrote: >> >>> Are there any chances of seeing pfb or m64 /dev/fb drivers being >>> published? >>> >> This would require permission from AMD/ATI since the code was written >> using docs under NDA from ATI. >> > > > Ok, then I cannot expect it to see how the libpciaccess performance > can be increased on those SPARC bridges where it takes so long. > Did somebody already talk to the kernel specialists? > > We know that at least one problem is that in current form on Sparc, its probing all busses (00..ff) and all devices per bus. However, there are severe problems with libpciaccess on x86 which are being attended to first, as they are so much more visible. On x86, when it probes higher than real bus numbers (usually 0x3f), it gets junk including echo of lower buses. This may be tripping the perceived problems with multiple graphics cards and causing crash during screeninit, "as best we can guess". Please don't quote me on that, its just a guess. The point is that Sparc takes it differently- it succeeds after 140 seconds of scanning.
The appropriate lib/kernel experts are working with us, I'm hoping to get a test article any hour now. I hope the x86 solution helps the sparc behaviour, but enough change is going on. Meanwhile, we have Install-time hangs in 107 and 108. <50% of the time, the Xorg server is implicated, but I'm leaning towards more kernel issues. There's just a lot of moving parts right now. Stuart
