On Tue, Nov 16, 2004 at 05:44:30PM -0500, Clint Adams wrote: > > Okay. I've removed application of this patch from the TODO list; take > > all the time you need to work on this, and thanks a lot for go after it > > with vigor. > > Slow progress. > > Here is a diff of the registers I know about on a fresh boot versus > after a Ctrl-Alt-Backspace. Not sure which are relevant. I'm afraid most of these are graphics driver jargon that I don't know much about. > -(II) GLINT(0): SCISSOR_MODE: 2 > +(II) GLINT(0): SCISSOR_MODE: 0 I think the "scissor mode" has something do with what were called hardware sprites back in the old days. > -(II) GLINT(1): PMFramebufferWriteMask: 90420 > +(II) GLINT(1): PMFramebufferWriteMask: ffffffff No idea. > -(II) GLINT(1): FIFO_CONTROL: 0 > +(II) GLINT(1): FIFO_CONTROL: 10000 This might be some sort of threshold for controlling how many acceleration commands are pushed to the video card. A value of zero probably means either "disabled" or "unlimited". "Unlimited" could be bad, if the host processor is outpacing the video card. > -(II) GLINT(1): LB_READ_MODE: 20000 > -(II) GLINT(1): LB_SOURCE_OFFSET: 40 > +(II) GLINT(1): LB_READ_MODE: 12d > +(II) GLINT(1): LB_SOURCE_OFFSET: 0 I don't even know what "LB" stands for here. > -(II) GLINT(1): WINDOW_ORIGIN: d220404 > -(II) GLINT(1): LB_WINDOW_BASE: 200 > +(II) GLINT(1): WINDOW_ORIGIN: 0 > +(II) GLINT(1): LB_WINDOW_BASE: 0 No idea. > -(II) GLINT(1): FB_SOFT_WRITE_MASK: 1800304 > -(II) GLINT(1): FB_HARD_WRITE_MASK: 2800100c > +(II) GLINT(1): FB_SOFT_WRITE_MASK: ffffffff > +(II) GLINT(1): FB_HARD_WRITE_MASK: ffffffff I know what a mask is, but I don't know how it's being applied here. > -(II) GLINT(1): DITHER_MODE: 403 > +(II) GLINT(1): DITHER_MODE: 0 Again, I know the term but not the applicaton. > -(II) GLINT(1): STENCIL_MODE: a00 > +(II) GLINT(1): STENCIL_MODE: 0 I've heard of "stencil modes", nothing more. Maybe it's used for accelerated line draws? > -(II) GLINT(1): COLOR_DDA_MODE: 1 > +(II) GLINT(1): COLOR_DDA_MODE: 0 DDA: EACRONYM > -(II) GLINT(1): ALPHA_BLEND_MODE: 7084 > +(II) GLINT(1): ALPHA_BLEND_MODE: 0 Know the term, not the application (i.e., what the value means). > -(II) GLINT(1): STATISTICS_MODE: 1 > -(II) GLINT(1): SCISSOR_MODE: 1 > +(II) GLINT(1): STATISTICS_MODE: 0 > +(II) GLINT(1): SCISSOR_MODE: 0 Interesting that SCISSOR_MODE gets reported *again*, and that its value changed during the register dump? Oh wait, the first one was GLINT(0), the first head. I'm conjecturing that the card has a way of dumping some sort of profile about how its accleration engine has been used, and that might be what "STATISTICS_MODE" is for. I'm sorry I can't be of much help here. Maybe this message will provoke those who do know into saying something. -- G. Branden Robinson | I had thought very carefully about Debian GNU/Linux | committing hara-kiri over this, but branden@debian.org | I overslept this morning. http://people.debian.org/~branden/ | -- Toshio Yamaguchi
Attachment:
signature.asc
Description: Digital signature