irclog2html for #htc-blueangel on 20060216

06:46.42*** join/#htc-blueangel rob_w (n=bob@p213.54.209.156.tisdip.tiscali.de)
06:46.42*** mode/#htc-blueangel [+o rob_w] by ChanServ
07:06.12*** join/#htc-blueangel rob_w|mis (n=rob_w@p213.54.209.156.tisdip.tiscali.de)
07:06.38*** join/#htc-blueangel rob_w-mib (n=rob_w@p213.54.209.156.tisdip.tiscali.de)
08:23.59*** join/#htc-blueangel awelux_ (n=awelux__@dslb-084-058-175-019.pools.arcor-ip.net)
11:57.32*** join/#htc-blueangel rob__w (n=bob@p213.54.211.206.tisdip.tiscali.de)
11:57.34*** join/#htc-blueangel rob_w-mib (n=rob_w@p213.54.211.206.tisdip.tiscali.de)
21:58.16*** join/#htc-blueangel _frank (n=user@84.92.70.37)
22:20.54*** join/#htc-blueangel Olli_G (n=bbird@port-195-158-178-185.dynamic.qsc.de)
22:47.15awelux_woho.  After using the pocket pc for a while and starting haret the memor for the kernel is very fragmented.
22:47.47cr2hi awelux
22:48.05awelux_hi.
22:48.22awelux_Somehow the code seams to crash at an odd place
22:48.35cr2i have broken something in haret.
22:48.55awelux_I'm drawing some pixels to the screen and suddenly the code stops.
22:48.57awelux_oh ?
22:49.02awelux_tell.
22:49.09cr2rewritten the memory allocation code, but now it breaks later.
22:49.43cr2the addresses given by "ioremap" are very strange.
22:50.16awelux_ah. currently I'm trying to write the content of the physical location of the preloader to the FB.
22:50.23cr2the memVirtToPhys fails here and there.
22:50.34awelux_I can see the f30f8
22:50.38awelux_and then it stops.
22:50.42awelux_all the rime
22:51.16cr2i can't pass through the icmr=0 because it is mapped to some weird address.
22:51.35cr2strange.
22:51.55awelux_I've  commented out the shutdown peripherals
22:52.04awelux_so I don't fall about that
22:52.17awelux_what should the icmr=0 do ?
22:52.27cr2yes. it was a bug introduced by me. it is a null pointer.
22:52.46cr2icmr masks the interrupts.
22:52.50awelux_ah.
22:52.57awelux_that would explain something
22:53.57awelux_Because the device shouldn't reac to to keys or do anything after the bootloader started.
22:54.12awelux_But sometimes it jumps back to wince or reacts to ekys
22:55.42cr2i am thinking about writing a sane replacement for memPhysMap. i have already written something but it crashes ;)
22:56.07awelux_I didn't understand memPhysMap
22:56.25awelux_It sounded like it tries to allocate the virtual memory for a physical location
22:56.42cr2it is very lowlevel.
22:58.07cr2i think it should disable interrupts when it works
22:58.28awelux_memphysmap?
22:58.33awelux_or icmr?
22:59.01cr2memphysmap. because it walks through the page table while wince is running.
22:59.54awelux_disabling irq's can't hurt there
23:00.09awelux_the dump mmu fails every time  for me too
23:00.33awelux_but I get a lot of the addresses already
23:01.22awelux_yea. I wrote to a different address of the FB and now I got the full mem content :)
23:01.44cr2i will try the second memphysmap version. if it will fail too, i will be thinking about rewriting it properly.
23:01.45awelux_I've grouped bit's into blocks of four and gave them different colors
23:01.46cr2?
23:02.30awelux_now I can count the dot's and reassemble the mmemory content.
23:02.54awelux_I'm doing this after the mmu 1:1 mapping
23:03.24cr2i use allocphysmem now. makes the code very simple, but will fail with big ramdisks.
23:03.55cr2after 1:1 ? i have written some asm to put the pixles, but it has the fbaddr compiled in.
23:04.21cr2maybe it should be a sixth parameter.
23:04.44cr2then it will be portable.
23:05.28awelux_ops
23:06.17awelux_xchat got very disappointing lately
23:06.37awelux_All the features got removed
23:06.44awelux_and could be added as user scripts.
23:07.01awelux_But of course there is no easy way to add them
23:07.19cr2hmm.
23:07.30cr2before bne nextpage
23:07.43cr2ldr r7,[r3]
23:08.09awelux_ok.
23:08.11cr2hmm. too painful on this linuxppc machine ;)
23:08.22cr2i can't cut and paste.
23:08.35awelux_what should that line do?
23:09.17cr2<PROTECTED>
23:09.19awelux_Ah. X style cut and paste is a good thing.
23:09.21cr2<PROTECTED>
23:09.25cr2<PROTECTED>
23:09.29cr2<PROTECTED>
23:09.33cr2<PROTECTED>
23:09.37cr2<PROTECTED>
23:09.38cr2<PROTECTED>
23:09.38cr2<PROTECTED>
23:09.38cr2<PROTECTED>
23:09.38cr2:)))
23:09.40awelux_oh. ic
23:09.54awelux_nice idea.
23:10.08cr2the 0xa3faa000 is my FBaddr.
23:10.11awelux_But I think my code doesn't reach that yet.
23:10.20awelux_nit should vibrate if it does..
23:13.10awelux_hmm
23:13.25cr2hehe. this one does not work at all. crashes very aerly.
23:13.37awelux_First I check if the bootloader is at the expected virtual address. It is.
23:14.20awelux_Then I add 1:1 mapping and can find the preloader code using the physicalPA address.
23:14.29cr2ok.
23:14.46cr2is yout "page table
23:14.49awelux_So I should be fine when I jump to that code with the movpc, r3 but I'm not ?!?
23:14.53cr2" corerect ?
23:15.23cr2the ptable
23:15.32cr2[] array.
23:15.37awelux_Right after the jump I wswitch on the vibra in the code.
23:15.41awelux_but nothing happens
23:15.56awelux_So I don' get to the pagetable part at all
23:16.30cr2i mean earlier. when the pixels are written.
23:16.38cr2into fb.
23:17.48awelux_VRAM virtual: 0xa8500020 w=240 h=320 b=16
23:17.48awelux_ptable: page=0 pfnn=12513 0xa30e1000
23:17.48awelux_ptable: page=1 pfnn=12514 0xa30e2000
23:17.48awelux_ptable: page=2 pfnn=12515 0xa30e3000
23:17.48awelux_p
23:17.52awelux_and so on
23:17.57cr2ok.
23:18.02awelux_so I think that's fine
23:18.17cr2i am asking because some pages were wrong on my universal.
23:18.40cr2so i have rewritten the memory allocation part.
23:18.59cr2now it works, but i have problems later.
23:19.11awelux_They seam to be fine. I've not really checked that part.
23:19.32awelux_Maybe we just need to switch off all interrupts.
23:19.33cr2i had 0xffffffff as an address.
23:19.57awelux_all adresses are continous right now
23:26.38cr2try to disassemble the asmstuff.o
23:26.49cr2does it make sense ?
23:28.41awelux_It looks like I'm doing a random jumkp or so
23:29.20cr2i don"t see anything written to the fb after  icmr=0. maybe i should enable the LCD interrupt ?
23:29.40awelux_oh?
23:29.57cr2i will try :)
23:30.12awelux_did you try without the icmr=0?
23:30.33cr21<<17 first.
23:33.39cr2does not help.
23:34.17awelux_I noticed that I can write to some pixels only
23:35.30cr2?
23:37.08cr2maybe it is necessary to ioremap() the fbram to your own address space ?
23:39.17awelux_so for you curpix only works before switching off interupts?
23:39.33cr2yes.
23:39.43awelux_do you reach the code after that at all?
23:39.51awelux_After switching of irqs?
23:39.56cr2how do i know ?
23:40.12awelux_search an led out gpio ;)
23:40.46awelux_check with haret what gpio changes when the backlight turns off.
23:40.47cr2it is on asic3. hmm. i will think about it.
23:40.53awelux_oh
23:41.06awelux_the universal has an asic3 again?
23:41.17awelux_or are you working on ba?
23:41.31cr2backlight is also on asic3. don't forget that magician is the simpliest htc device :)
23:41.40cr2universal has asic3.
23:42.01awelux_ok. well I'm not really using backlight. I use the key led's
23:42.09cr2like ba,hima,alpine,beetles,etc.
23:42.11awelux_which are on a normal gpio
23:42.35awelux_Not sure how the universal handels it's key led's
23:42.48cr2asic3 :)
23:43.41awelux_ah.
23:43.45awelux_tricky
23:44.26awelux_yopu have added quite a lot of universal info already
23:44.45cr2everything "visible" is on asic3. it should not be a problem, but i have never tried.
23:45.07cr2the only thing that i don't know is where is the flashlight.
23:45.14cr2and some sound stuff.
23:46.28cr2universal is better researched than ba. it is just more complex in some places.
23:46.46awelux_I think I found a problem
23:47.07cr2where?
23:47.17awelux_ldr     r0, =0xaee00018
23:47.38awelux_I do that after the jump to get some feedback
23:47.43cr2put ldr r0,
23:47.45cr2[r0
23:47.47cr2]
23:47.51awelux_put ?
23:47.52cr2after it.
23:48.00cr2ldr r0,
23:48.17cr2ldr r0,[r0]
23:48.24awelux_I think I know what you mean.
23:48.48awelux_In the code the =0xaee00018 isn't part of the bootloader block
23:48.57awelux_but part of a data block
23:48.58cr2damn apple keyboard
23:49.06awelux_which doesn't get copied,
23:49.35cr2hmm. sounds reasonable.
23:49.38awelux_When I'm rich I want to buy a logitech nova media keyboard. One of these flat ones ;)
23:49.51cr2but it is just behind the .text in my .o
23:50.32awelux_So the address copied with ldr is a relative address to the machine command ldr
23:50.37cr2put the constant in the text segment and reference it indirectly.
23:50.41awelux_and we just relocated that one
23:50.55cr2yes.
23:50.57awelux_I will copy the address into a register before the jump
23:51.46cr2everything between two labels is copied, so you need your constant be there.
23:52.23awelux_do you know how to define a constant in gnu asm ?
23:52.33cr2.long
23:53.05cr2and then use a lable. look into the kernel .S files.
23:54.25awelux_so just before the linux_preloader_end I can write .long GPSR0 0xaee00018 ?
23:55.00cr2gpsr0: .long 0xaee00018
23:55.04cr2i think.
23:55.39awelux_ah. thx.
23:58.37cr2no. it does not help. weird.
23:59.04awelux_strange.
23:59.17cr2it is put in the right place, but is referenced in a very strange manner.
23:59.36cr2i am looking at the objdump output.

Generated by irclog2html.pl by Jeff Waugh - find it at freshmeat.net! Modified by Tim Riker to work with blootbot logs, split per channel, etc.