On Tue, Jan 20, 2009 at 3:50 PM, Belisko Marek <[email protected]> wrote: > Hi, > > > On Mon, Jan 19, 2009 at 12:59 PM, pHilipp Zabel <[email protected]> > wrote: >> On Mon, Jan 19, 2009 at 12:39 PM, Belisko Marek <[email protected]> >> wrote: >>> Hi, >>> >>> On Mon, Jan 19, 2009 at 12:10 PM, pHilipp Zabel <[email protected]> >>> wrote: >>>> Hi Marek, >>>> >>>> On Mon, Jan 19, 2009 at 11:34 AM, Belisko Marek <[email protected]> >>>> wrote: >>>>> Hi, >>>>> >>>>> thanks for help. Ialreadyread a whole documentation but still ca't >>>>> find a clue how to get pins for TFT display. >>>>> In attachment you can find my log. Could you please lookon this and >>>>> give some proposal how to understand. >>>> >>>> What pins are you looking for, exactly? SPV C100 is HTC Oxygen = >>>> omap850 based, right? >>>> I think all those omap850 phones have the TFT connected to the >>>> internal LCD controller, so I'd propose to have a look at the LCDC >>>> register settings (see drivers/video/omap/lcdc.c for the physical >>>> register addresses) and try to recreate them in a custom struct >>>> lcd_panel for omapfb. IIRC no GPIOs are involved. >>> >>> After looking to kernel code seems no GPIOS for LCD are used in code >>> (they're only defined - using source for typhoon) soyou're right. >> >> Or maybe not - looking at the code for htcartemis, there are some >> GPIOs involved in turning on/off power to the LCD, but there they are >> on the GPIO extender chip: >> http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-omap1/htcartemis/htcartemis_lcd.c;h=9e5241e329c18e5102de0ddd408c79287821e797;hb=ce3552ca024ed17381fead3a7451a31737d6e33f >> If you're already seeing strange colours, the LCD is still turned on >> from wince - you should be fine for now... >> >>> I'm trying to fill lcd_panel struct but there'is missing some infos. >>> E.g:how fill values for : pixel clock, horizontal synchronization >>> pulse width, horizontal front porch, horizontal back porch, vertical >>> synchronization pulse width, vertical front porch, vertical back >>> porch, ac-bias pin frequency. >> >> Look at how setup_regs() in lcdc.c fills the OMAP_LCDC_* registers. If >> you dump the LCDC registers with haret, you should be able to recreate >> the settings that result in those register values when run through >> setup_regs(). >> Or, if you're impatient, you could go quick&dirty and directly >> overwrite the registers at the end of setup_regs() with the values >> that you got from haret (just to test). >> >>> It would be possible to get it from Haret? >> >> Something like >> pd 0xfffce000 0x20 >> to dump the LCDC regs? > With magic command pd I got address values and according setup_regs
No magic! Just read the haret documentation and kernel source ;) > parse it back > to values for lcd_pabel struct. What I'm still missing (not set in > setup_regs) is pixel_clock. It's somehow derived > from system clock I think but no clue how to get it. One way would be to just set pcd directly for now and afterwards obtain the pixel_clock value from dmesg (see lines 536-548 in lcdc.c). Or you can try to figure out lcd_ck rate manually. The divider can be taken from ARM_CKCTL (2 bits at CKCTL_LCDDIV_OFFSET), lcd_ck is derived from ck_dpll1, which is based on the 12 MHz ck_ref, controlled by the DPLL_CTL register. See arch/hardware.h for phys register addresses, you can dump them, too, with HaRET's pd. The rate calculations can be found in arch-omap1/clock.c. > Yesterday I also disassembly my Oxygen and find out that display has > type TD020 (some china producer). > Try to contact producer for specs but no response :(. > >> >>> Values from typhoon isn't working (display shows only strange colours). >>> >>> If you have some spec about omap 850 would be helpful also :) >> >> I wish. There is the OMAP850 TRM floating around, but it only shows >> the differences between OMAP730 and OMAP850. As OMAP850 is very >> similar to the older one, it's not that much. If anybody has access to >> the OMAP730 TRM, I'd be severely interested. A lot of useful register >> information can already be found in the kernel source, though. >> >> regards >> Philipp >> > > thanks, > > Marek > > -- > as simple as primitive as possible > ---------------------------------------------- > Marek Beliško - open-nandra > Ruská Nová Ves 219 > 08005 Prešov > Slovakia > http://open-nandra.com > cheers Philipp _______________________________________________ Haret mailing list [email protected] https://handhelds.org/mailman/listinfo/haret
