00:01.01 | cr2 | ok. i don't need = before the label. then it looks much better. |
00:01.35 | cr2 | so i did: |
00:01.45 | cr2 | ldr r7. myfb |
00:01.47 | cr2 | ... |
00:02.09 | cr2 | myfb: .long 0xa3faa000 |
00:04.18 | awelux_ | hihi |
00:04.29 | awelux_ | I get into the preloader ! |
00:04.43 | cr2 | great! |
00:04.44 | awelux_ | that took how many days ? |
00:04.46 | awelux_ | 10? |
00:04.51 | cr2 | :) |
00:08.15 | cr2 | assembler programming is fun. |
00:08.53 | cr2 | now i need to understand why sti() and icmr=0 break my fb refresh. |
00:12.03 | awelux_ | And I need to check if I get past the MMU switch off |
00:12.24 | cr2 | it should be easy ? |
00:12.44 | awelux_ | yes. easy |
00:12.49 | awelux_ | njust takes time |
00:12.52 | awelux_ | ... |
00:12.56 | awelux_ | and a cold reset |
00:13.13 | awelux_ | I guess that is a good sign |
00:13.25 | cr2 | + for wince2005. |
00:14.00 | awelux_ | wince2005 needs no cold reset? |
00:14.42 | cr2 | no. |
00:16.53 | awelux_ | :/ |
00:17.04 | awelux_ | the mmu switch off fails it seams |
00:17.42 | cr2 | that is difficult. it is probably the most hardcore arm stuff. |
00:18.59 | awelux_ | I will continue tomorrow. cu cr2 |
00:19.06 | cr2 | good night. |
00:19.21 | awelux_ | all these cold resets are taking to long for tonight ^^ |
04:24.03 | *** join/#htc-blueangel pigeon (n=pigeon@adsl-60-232.swiftdsl.com.au) |
07:52.27 | *** join/#htc-blueangel rob_w|mis (n=rob_w@p549BA905.dip0.t-ipconnect.de) |
08:02.33 | *** join/#htc-blueangel geers_ (n=DevzMan@c51008F31.inet.catch.no) |
08:24.10 | *** join/#htc-blueangel awelux__ (n=awelux__@dslb-084-058-174-244.pools.arcor-ip.net) |
12:04.04 | *** join/#htc-blueangel dullard (n=jim@adsl-static-1-30.uklinux.net) |
14:39.06 | mndctrl | I'm working on a common gsm library based upon gomunicator among others... Since there already is a libgsm (wich deals with gsm audio or something) under OE, what should I call it? |
14:39.11 | mndctrl | libgsmdial ? |
14:53.04 | *** join/#htc-blueangel rob_w (n=bob@p213.54.200.190.tisdip.tiscali.de) |
14:53.04 | *** mode/#htc-blueangel [+o rob_w] by ChanServ |
19:38.56 | awelux__ | what is the scope of this library? All gsm realted stuff or just managing connections? |
19:47.37 | cr2 | hi awelux |
20:01.37 | cr2 | found one more dumb mistake. if the interrupts are disabled you can't call sleep() ;) |
20:05.56 | awelux__ | eh. yes. I noticed that a while ago ;) |
20:06.51 | cr2 | writing to the fb works well. now i need to check if the mov pc, r3 is passed. |
20:09.38 | awelux__ | I found that I get past the MMU code |
20:09.49 | awelux__ | but not the lines before that invalidate the cache |
20:11.00 | awelux__ | And I downloaded the xscale manuals |
20:11.01 | cr2 | the tlb ? |
20:11.37 | awelux__ | the xscale core developers manual |
20:12.07 | awelux__ | on the intel ftp in downloads/xscale is some interesting stuff |
20:13.32 | awelux__ | I'm still searching the xscale architecture reference manual though |
20:13.44 | cr2 | ok. i will think about writing to the asic3 gpio reg. |
20:13.53 | awelux__ | oh |
20:14.01 | awelux__ | Ah good ;) |
20:14.14 | awelux__ | I mean the arm architecture reference manual |
20:14.42 | cr2 | it's probably on the arm site. |
20:44.41 | cr2 | ok. i can boot the kernel with the IR console now. the kernel oopses, but it is probably a config problem. the ir output is messed up, and the pxafb is not initialized. hmm. |
20:45.04 | awelux__ | still nice |
20:45.27 | cr2 | have you found the ir enable gpio ? |
20:45.58 | awelux__ | I don't have an ir port on my pc anyway |
20:46.16 | awelux__ | I've a small ir adapter for the serial port though |
20:46.37 | awelux__ | It was for using remote controls. |
20:46.44 | awelux__ | I can use it with irda |
20:46.56 | awelux__ | not sure if that would work as console |
20:47.18 | awelux__ | But for now that is not important |
20:47.25 | awelux__ | I still don't reach the kerneƶ |
20:47.33 | cr2 | ok. i think that booting is not a problem now. pxafb is a problem. |
20:47.41 | cr2 | try my version. |
20:49.35 | awelux__ | upload. |
20:49.54 | awelux__ | the source |
20:49.59 | cr2 | i have commented out the non-portable stuff. |
20:50.04 | awelux__ | ok |
20:50.08 | awelux__ | good |
20:52.29 | cr2 | ok. uploaded. |
20:53.08 | cr2 | run a diff to be sure. |
20:58.12 | awelux__ | http://www.mnementh.co.uk/docs/ |
20:58.22 | awelux__ | the ARM reference manual |
20:58.24 | awelux__ | 800 pages |
21:48.42 | cr2 | awelux, i have usb eth gadget running, ping -f works :) can't login, probably the dropbear in the ramdisk is confused. ir output is not really readable. |
21:53.19 | *** join/#htc-blueangel ChanServ (ChanServ@services.) |
21:53.19 | *** mode/#htc-blueangel [+o ChanServ] by irc.freenode.net |
22:21.35 | cr2 | wow! i am impressed with my own work. |
22:21.53 | cr2 | i can login over usb-rndis |
22:21.59 | cr2 | and mount the sd card |
22:22.29 | cr2 | and write to the sd card, so the asic3 sd driver works. |
22:23.37 | cr2 | if the pxafb init problem will be solved, then we can talk about gpe/opie :) |
22:32.01 | *** join/#htc-blueangel frank__ (n=user@84.92.70.37) |
23:01.20 | awelux__ | wow |
23:01.25 | awelux__ | nthat is verty nice |
23:01.32 | awelux__ | <-- was playing bzflag again |
23:01.56 | cr2 | :) |
23:02.21 | cr2 | i think you will have the same problems (and solutions) now. |
23:03.01 | awelux__ | didok. |
23:03.22 | awelux__ | mcrp15, 0, r0, c7, c10, 4@ Drain write buffer |
23:03.22 | awelux__ | mcrp15, 0, r0, c7, c7, 0@ invalidate I+D & BTB |
23:03.29 | awelux__ | these still seam to cause problems |
23:03.38 | cr2 | hmm. |
23:04.04 | cr2 | i have thought that you can boot my kernel on magician. |
23:04.15 | awelux__ | I didn't try that yet. |
23:04.19 | awelux__ | Will do |
23:04.21 | cr2 | pxafb params are different, but is does not work anyway. |
23:04.36 | cr2 | you can also comment out the asic3 stuff. |
23:04.47 | cr2 | usb is the same and should work. |
23:05.03 | cr2 | my ramdisk is from blueangel. |
23:05.14 | awelux__ | do I have to select a certain port? |
23:05.32 | cr2 | with the "universal" modules added |
23:05.33 | awelux__ | my magician is using 192.168.131.201 right now |
23:05.40 | cr2 | -un0 branch. |
23:06.08 | cr2 | 192.168.0.205 - host , 206 - phone. exactly as the blueangel. |
23:06.25 | awelux__ | What is the IoRemap doing ? |
23:06.44 | cr2 | you don't need it. |
23:06.49 | awelux__ | I should try to reconfigure my firewall for these ports then |
23:06.58 | cr2 | it is a dirty hack. |
23:08.29 | cr2 | compile the pxa-mci modules. |
23:09.06 | cr2 | with the correct irq you should have a working mmc (maybe sd too). |
23:09.23 | awelux__ | wow |
23:09.38 | awelux__ | sounds quite good. what haret to use? your new one ? |
23:09.51 | cr2 | yes. |
23:10.07 | *** part/#htc-blueangel frank__ (n=user@84.92.70.37) |
23:10.32 | *** join/#htc-blueangel frank__ (n=user@84.92.70.37) |
23:10.40 | cr2 | it works very stable now. |
23:12.05 | cr2 | maybe needs a patch to work with bigger ramdisks. but not really necessary to help bloatware writers ;) |
23:14.41 | awelux__ | bloatware. Image: |
23:15.15 | awelux__ | currently when writing software you don't reinvent everything but take other modules when they exist |
23:15.38 | awelux__ | now these modules usually have dependencys on their own. |
23:16.31 | awelux__ | nif the each module needs more than one other module on average to run the amount of modules needed goes against infinity. |
23:16.54 | cr2 | i have 900K kernel + 5.3MB ramdisk. the tests say it will work up to 7.5MB |
23:16.56 | awelux__ | that's why you can easily install 1-2 CD's of software just to get a webpage running |
23:17.09 | cr2 | there is an X server on the ramdisk!!! |
23:17.16 | awelux__ | wow |
23:17.19 | cr2 | +850K |
23:18.28 | cr2 | but you can't remove it with just "rm" :) |
23:22.43 | awelux__ | wow |
23:22.53 | awelux__ | the blue line on the sceen just got red |
23:23.01 | cr2 | yes. |
23:23.17 | cr2 | wait a bit. |
23:23.33 | awelux__ | I added an led off command. |
23:23.54 | cr2 | then connect the usb cable and do "lsusb" |
23:24.55 | awelux__ | that no change there |
23:25.21 | awelux__ | what I don't understand: |
23:25.34 | awelux__ | The part that draws red pixels in the asm is commented out |
23:26.32 | cr2 | there is c code before linux_start |
23:26.54 | awelux__ | ah. good |
23:27.30 | awelux__ | I understand |
23:28.46 | awelux__ | you used that instead of switching of led's or so |
23:28.54 | awelux__ | to see where you can gt |
23:28.57 | awelux__ | *get |
23:29.37 | cr2 | black,blue,green,red. should be a function. |
23:30.13 | awelux__ | what was the problem with the old code ? |
23:30.36 | awelux__ | you changed the way how memory for the kernel is allocated |
23:31.11 | cr2 | the ptable entries were sometimes =0xffffffff. it is never tested. |
23:32.43 | cr2 | allocating memory with malloc() is not very wince-physmem-alloc-style. malloc is LocalAlloc. stupid micro$oft invention. |
23:33.11 | cr2 | what a braindamage. |
23:34.18 | awelux__ | if writing braindamadged stuff helps them that no sane person would want to emulate their API they reached theri goal ;) |
23:36.14 | cr2 | instead of the linux ioremap they have such an obfuscated code api with obscure parameters, that their own programmerz made multiple mistakes in the inplementation! |
23:38.04 | cr2 | mmiospace was leaking memory, and it is _necessary_ to use virtualfree(,mem_decommit) before virtualfree(,mem_release). the m$ docs don't say that. it is "empirical knowledge". |
23:38.36 | awelux__ | lol |
23:40.24 | *** join/#htc-blueangel dullard (n=jim@adsl-static-1-30.uklinux.net) |
23:41.21 | cr2 | it is actually virtualfree(ptr,size,*) but the "size" _must_ be zero on "physical" allocations, otherwise you will leak memory :) and it was ms own code that leaked memory ! |
23:42.03 | cr2 | high quality programming. |
23:42.24 | cr2 | linux is so simple compared to that shit. |
23:43.33 | awelux__ | yes. But you have to give them credit that it's convenient for the end user. Not sophisticated or great but convenient, |
23:45.57 | cr2 | if somebody has ported the coredll from wine to wince and has written the coff loader, wince will be really toast. and wince-wine will be much easier than "real" wine. |
23:46.51 | cr2 | ms made it really easy. if you will look at the coredll disassembly, it is a jump table. |
23:48.07 | awelux__ | hmm. good plan ^^ |
23:48.45 | awelux__ | nyes. if users could ru their wince apps on any handheld they wouldn't care if it's running wince or linux |
23:50.02 | awelux__ | gee. how can you live with a new gnuharet name on every recompile ? |
23:50.29 | cr2 | i remember running solitair in 1995 on my 386dx25 linux box with 4MB ram. that was really cool. would like to test that on wince :) |
23:50.51 | cr2 | it was intentional. to upload the binaries to sf.net |
23:51.07 | awelux__ | ic. |
23:52.11 | cr2 | make inst and make instz |
23:52.26 | awelux__ | 1995 is early. I converted from win to linux from 1997-1998 |
23:53.07 | awelux__ | I remember I installed several win95 betas but had no interest in updating to win98 |
23:53.32 | cr2 | i converted in late summer 1994 :) that's why i am so much anti-ms zealot :) |
23:53.33 | awelux__ | Actually I deleted the win98 just last year. |
23:53.39 | cr2 | :) |
23:53.44 | awelux__ | 1994 is freaking early |
23:54.43 | cr2 | i worked with x11 on a vax before even seeing windows 3.0. win3.1 was very lame. no network display, no tcpip. |
23:54.53 | cr2 | :) |
23:56.27 | cr2 | and idiotic turbo pascal and m$ fortran with 64K segments for code and data. when there as 16MB ram on the vax, and good DEC compilers that really used it. |
23:57.01 | awelux__ | ah. yes compared to such compuers the early pc's must have been real toys. |
23:57.43 | awelux__ | I've an old slackware CD here with kernel 1.2.13 |
23:58.09 | awelux__ | I rmember we installed it once and started X |
23:58.21 | awelux__ | But we were not impressed. We didn't start a WM ;) |
23:59.34 | cr2 | then there was alpha with 64bit cpu already in 1993. DEC was so great in engineering... |