Stephan Assmus <superstippi@xxxxxx> wrote: > Hi Zenja, > > Zenja Solaja wrote: > > Morning / Afternoon / Evening. > > > > I just came back from a month away (Thailand is beautiful), and noticed a > > weird quirk with what may be Vesa related. When booting on real > > hardware, just before the desktop appears (after all boot-screen icons > > finish loading) there is a new delay which lasts approximately a full > > minute. During this delay, the actual PC speaker beeped 3 times, with a > > frequency of one beep every 20 seconds or so. After the 3rd beep, the > > screen went blank, and 20 seconds later the desktop appeared. > > > > When changing screen resolutions, the same thing will happen again. The > > PC will appear as if its locked up (no mouse movement, no hard disk > > activity), and beep once every 20 seconds. On the 3rd beep, the screen > > will go blank. > > After a final 20 seconds, the screen resolution will change, with the > > good > > old timer to accept new screen resolution or cancel. Yep, thats right, > > even the timer got suspended for a full minute while waiting for the > > video card to respond. > > > > I have an nVidia 8800GT, Vesa driver. Before I went on a vacation a > > month ago, this behaviour wasn't happening. Latest revision displays > > this weirdness. I have noticed that some work has been done with the > > Vesa driver to allow dynamic updates without reboots. > > > > Has anyone seen this before? Just for the record... there's already a ticket for the problem: #2337. Looks like quite a few people seem to experience the problem. :-( > I know that Ingo is experiencing the same behavior. On my system, I just > get "General System Error" when I try to change resolution (VESA also) > during runtime. But I think if I have a vesa kernel settings file with > anything different than the native resolution, my system does not boot. I > am not sure, but I thought for the behavior you describe, there is a bug > report, but I could be wrong. > > Jan, any ideas? Unfortunately not really. I suspect that there is an error in the instruction emulation part of the vm86 stuff which confuses the BIOS. Just a wild guess. I was not able yet to reproduce it on any of my systems here. What makes it really interesting and what is new to me is that you get a "General System Error". Would you mind to enable TRACE_VM86 in src/system/kernel/arch/x86/vm86.cpp and capture the output? Maybe I can spot something there because the traces I've got so far did not contain anything suspicious... /Jan