irclog2html for #htc-blueangel on 20060330

02:43.22*** join/#htc-blueangel angom_h (n=angom@red-corp-200.79.154.31.telnor.net)
05:27.18ablisshey, anybody know how cr2 managed to get usbfs to mount on his universal?
07:19.14*** join/#htc-blueangel awelux_ (n=awelux__@dslb-084-058-153-195.pools.arcor-ip.net)
08:24.33mndctrlabliss: did you try booting it while in the cradle? worked for me on the blueangel...
14:41.55*** join/#htc-blueangel jeanseb (n=jeanseb@lns-bzn-47f-62-147-133-94.adsl.proxad.net)
15:28.59*** join/#htc-blueangel jeanseb (n=jeanseb@lns-bzn-47f-62-147-133-94.adsl.proxad.net)
16:49.05*** join/#htc-blueangel geers_ (n=DevzMan@c51008F31.inet.catch.no)
17:45.30*** join/#htc-blueangel rob_w (n=bob@p213.54.206.219.tisdip.tiscali.de)
17:45.30*** mode/#htc-blueangel [+o rob_w] by ChanServ
17:45.51rob_wgreetings all
18:36.23*** join/#htc-blueangel BabelOued (n=BabelOue@lun34-2-82-238-28-28.fbx.proxad.net)
19:09.58*** join/#htc-blueangel tuks (n=tuks@cust-02-5286af4e.adsl.scarlet.nl)
19:10.05tukshi there!
19:12.48tukscan i ask a question about linux on the htc-bleuangel?
19:12.59*** join/#htc-blueangel JuNuKN (n=JuNuKN@Mc5b1.m.pppool.de)
19:13.45tuksive got a problem booting linux on the htc-bleuangel
19:18.38awelux_ok
19:19.13tuksno :) hehe.. isnt ok :)
19:19.45awelux_I mean it's ok to ask the question.
19:20.16tuksif it boots.... i only get a command-line with a login.
19:20.38awelux_People don't look in here every minute but you might get an answer after a while.
19:20.59tuksive got time :)
19:21.31awelux_tuks, that's not that bad. What should happen after you booted linux (which obviously works)?
19:22.08tuksuhm... how can I login?
19:22.14awelux_oh
19:22.21awelux_the key's don't work ?
19:22.26tuksyes
19:22.57awelux_But you don't have the password. ic.
19:23.16tukscan not find it on the website
19:24.42awelux_http://wiki.xda-developers.com/index.php?pagename=BlueangelKernel
19:24.47awelux_http://wiki.xda-developers.com/index.php?pagename=BlueangelKernel
19:24.53awelux_Login via terminal as root, password is rootme
19:25.03tuksI also get the folowing: getty: i0ct1() YIOCSPGRP call failed: Inappropriate i0ct1 for device
19:25.39awelux_no idea about that one.
19:28.28tuksand this one: "sh: can't access tty; job control turned off"
19:28.43awelux_this sounds bad
19:29.01awelux_but then again
19:29.15awelux_maybe it will just mean that jobs, fg and bg doesn't woirk
19:30.50tuksdoes that has something to do with te current firmware?
19:45.34*** part/#htc-blueangel tuks (n=tuks@cust-02-5286af4e.adsl.scarlet.nl)
20:09.28awelux_JuNuKN, I got a lot further to a working magician.
20:10.00awelux_It seams I've reached the point where I'm loading the compiled in kernel modules.
20:11.30JuNuKNHey awelux_, thanks for the info.
20:11.56JuNuKNAre you fare away from framebuffer ?
20:13.57JuNuKNAnd what where the last mistakes? A not matching driver from universal, which overwrites Magicians memory
20:14.51awelux_no. I think that's still ahead
20:15.27awelux_The last "mistake" was that I didnt get any feedback afte the pagetable got reinitialized.
20:15.50awelux_I tried accessing the FB or a GPIO pin later but the address wasn't mapped.
20:16.06awelux_So now I map the FB again and I can see how far I get.
20:16.25JuNuKNWhat means pagetable exactly ?
20:17.20awelux_the pagetable for virtual address to physical address mapping of the MMU
20:19.06JuNuKN? setting up the mmu to map special memory adresses to other physical memory adresses!? ... but the GPIO's are internal Adresses of the processor !?
20:19.54JuNuKNSo they can also be remapped ?
20:20.14awelux_JuNuKN, setting up the mmu is to map _all_ addresses to physical addresses, Including gpio.
20:20.46awelux_the gpio will remain at the same physical address. but you use another address to access them
20:20.51JuNuKNAhh .... OK, I understand. And the setup needs some assembler!?
20:21.36awelux_Haret always uses memPhysMap a lot to map a physical address of something to a useable virtual one.
20:21.49JuNuKNSo if the mapping matches, it should be easy to write into the memory of the LCD-Adresses / fb ?
20:21.51awelux_JuNuKN, no. That's done in C already
20:22.12awelux_yes. I've written a 1:1 mapping for the LCD
20:22.33awelux_So I can find it at 0x5c0... for now
20:23.34JuNuKNFor what reaso is such a mapping needed? Why not use the memory areas on the original places ?
20:24.39awelux_it's very usefull to have it for some purposes.
20:25.17JuNuKN:o)   ... yes but for which?
20:25.53awelux_For example you could make it transparent to an application if it's writing to the FB or a memory location by selecting a propper mapping
20:26.11awelux_Or writing an application in general becomes easier.
20:26.13JuNuKNone example, where linux has advantages on it?
20:26.39awelux_Since you and usually you use the MMU to shield memory locations from an application
20:26.40JuNuKNOK
20:26.45awelux_you just remove the mapping
20:27.13awelux_Almost everything uses the MMU.
20:27.23awelux_Including window
20:27.33JuNuKNI see, so you can also create bigger virtual memory areas, when some hardware is interrupting areas within it
20:27.45awelux_exactly
20:28.13awelux_this is used for swap space.
20:28.29awelux_The application accesses an invalid memory location.
20:28.35JuNuKNAnd to be able to manage biger memory areas, than the cpu can address ?
20:28.43awelux_And the page fault is trapped, the page loaded from disk
20:28.48awelux_and the operation resumed
20:29.12awelux_JuNuKN, no. the virtual memory is limited to what the cpu can address.
20:29.14JuNuKNWhat means page fault? How is it detected?
20:29.26awelux_so a 32bit cpu can have 4GB of virtual memory
20:29.48awelux_But in theoriy you could provide 4GB to each process.
20:30.05awelux_a page fault is generated by the MMU
20:30.12awelux_Memory Management 'Unit
20:30.28awelux_It's detected in hardware
20:30.33JuNuKN... was clear to me for the 32bit processors. I thought on my good old 8 bit processors  ;o)
20:31.17awelux_Usually 8bit processors have a bigger address space than 8 bit
20:31.35awelux_which is a bit tricky to access
20:35.55JuNuKNSo your current problem seems, - if I understood correct - , that If you change the MMU settings your aplication crashes
20:36.12JuNuKNTherefore it is no more possible to load the modules!?
20:37.48awelux_no
20:38.08awelux_I don't know the current problem. I expect I'm loading a bad module.
20:38.40awelux_It's possible that the magician would boot if you select the right modules with a default kernel and avoid zImage
20:41.43JuNuKNSo the main-kernel will only manage memory handling and starting child processes, like configured ... and later loading module after module, which extends to hardwre access ... like seriell line and fb!?
20:42.04awelux_Basically yes
20:45.23JuNuKNSo I would expect, if a standard-kernel is used and the first module to load (maybe the fb-driver) was compiled with the physical adressmappings, the running kernel schould be visible?
20:45.41JuNuKNHave you tested this in the past?
20:46.28awelux_no. But that is very much what I wnat to do.
20:46.45awelux_To figure out how to push up the fb initialization
20:50.06JuNuKNIsn't it the definition of the adreeses, where the memory is started, - maybe the bits per pixel and maybe wdith and height?
20:50.39JuNuKNI see, - and maybe a virtuall adress-mapping to have one byte per pixel, following one after each other ;o)
20:52.35awelux_the virtual-> physical mapping happens in "pages" which are of varying size depending on the MMU.
20:52.47JuNuKNMaybe I have to read the book linux from scratch to understand it
20:52.54awelux_the arm supports 1k, 4k, 64k and 1MB I think
20:55.22JuNuKNThanks for your explanations. Helps much to understand ... a little bit more of the neccessary steps.
20:55.52JuNuKNI wish you good luck for this night ...
20:56.08awelux_gn8
20:56.26JuNuKNAre you still working on it, while you are in chat ?
20:57.06JuNuKNJa, gute Nacht ;o) Danke
20:59.21awelux_yes, I am
20:59.32awelux_thx
21:58.07cr2hi awelux
21:58.13awelux_hi
21:58.28cr2no more time for hacking :(
21:58.35awelux_oh?
21:58.41awelux_what happened?
21:59.00cr2i must prepare for my trip to Vienna.
21:59.15cr2next week i'm there.
21:59.18awelux_ah. how long will you be there ?
21:59.37cr2the whole week.
22:00.04awelux_well. then enjoy your stay.#
22:00.22awelux_I've made much progress
22:00.57awelux_processes are up.
22:00.58cr2can you boot now ?
22:01.07cr2pxafb ?
22:01.10awelux_I think I'm reaching the driver loading.
22:01.20awelux_No the universal pxa module isn't reached
22:01.34awelux_I should try the pxafb though
22:01.44cr2can you put the source into CVS ?
22:01.45awelux_Do you know if I can get fb console somehoe ?
22:01.51awelux_somehow?
22:02.02awelux_I had no real change to the kernel so far.
22:02.10cr2you should not use the 0x5c* at first.
22:02.13awelux_I guess an unmodified one will reach just as far
22:02.23cr2let the pxafb allocate the DRAM buffe.r
22:02.49awelux_removing the debug dot's from the code is hopeless now ;)
22:03.09cr2you can try to reuse the htcuniversalfb
22:03.25cr2just set the 0x5c*20 there.
22:04.06awelux_that's what I did
22:04.42awelux_I just discovered that I compiled fb console as a module...
22:04.56cr2hehe :)
22:06.09awelux_the next minute could be interesting
22:06.29awelux_or not
22:06.49awelux_I've to understand how the fb console stuff works
22:07.46awelux_But right now things are moving fast. I've more ideas than time.
22:08.22awelux_wednesday I found where the kernel rewrote the pagetable
22:08.31awelux_And added the fb mapping there
22:08.42awelux_So I was able to trace execution much further
23:10.00awelux_ohhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
23:10.53cr2?
23:11.04awelux_There is a little penguin in the top left corner of the screen now
23:11.12awelux_And the screen is cluttered with text
23:11.43awelux_I removed all modules with the exception of the fb.
23:11.56awelux_It want's a ramdisk
23:12.01cr2!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
23:12.14cr2glorious leader :)))
23:12.19awelux_"Unable to mount root fs ..."
23:12.30awelux_hehe. Feels good.
23:12.38cr2which modules did you compile ?
23:12.51awelux_I will try to make a few pictures over the weekend.
23:12.57cr2yes.
23:13.06awelux_I used the pxafb module and your htcuniversal_lcd one
23:13.23cr2both work ?
23:14.16awelux_no. I mean I used the lcd module in the system type section
23:14.28cr2try the ba ramdisk.
23:14.33cr2and ssh.
23:14.44awelux_I removed the network modules
23:14.46cr2i guess you will need some pxa-mci patches.
23:15.05cr2pxa27x_udc and gether
23:15.06awelux_I used the pxa fb and not the htc universal fb
23:15.10cr2is all you need.
23:15.19cr2ok.
23:15.20awelux_I have to continue tomorrow though
23:15.34cr2still, it is great!
23:15.36awelux_the lowlevel backlight and lowlevel lcd support is compiled in
23:15.52cr2they do not work anyway.
23:15.53awelux_Yes. A lot of progress in a few days.
23:16.17awelux_The dot's on the screen helped to see when I was moving forward
23:16.43cr2can you make a screenshot ?
23:17.10awelux_I will get a camera this weekend and make a few screenshots
23:17.18cr2ok.
23:18.15cr2i will post to xda-dev linux thread.
23:18.35awelux_oh let me do that ;)
23:18.55awelux_when I have the screenies
23:19.28cr2to the famous thread of JuNuKN: "Are there NO news on the Linux developement ?"
23:19.49awelux_hrhr
23:20.22cr2when he first posted it, the situation was very bad...
23:21.11cr2Posted:  Thu Apr 28, 2005 13:57
23:21.40awelux_I can't find it that fast.
23:22.45cr2http://forum.xda-developers.com/viewtopic.php?t=20670&start=0&postdays=0&postorder=asc&highlight=
23:25.26cr2hehe.
23:25.29cr2Subject:   Linux on Universal and Magician booting
23:27.29cr2now you can debug the touchscreen driver :)
23:28.00awelux_<PROTECTED>
23:29.06cr2i guess that you phone will work earlier than mine.
23:29.49cr2but your bt is not that easy. just like my phone.
23:31.41cr2you can upload your kernel, so the alpine guys will try it too.
23:42.01awelux_http://home.arcor.de/awelux/pic/Magician01.png#
23:42.03awelux_http://home.arcor.de/awelux/pic/Magician01.png
23:42.06awelux_horrible picture
23:45.30cr2yeah.
23:45.58cr2but still you can see a penguin and the text.
23:47.46awelux_http://forum.xda-developers.com/viewtopic.php?p=259214#259214
23:48.05awelux_right
23:49.30awelux_scary. Didn't expect this to work today.
23:49.33awelux_gn8
23:49.35cr2great.
23:49.39cr2good night.
23:52.10awelux_un Dec 18, 2005 01:59
23:52.10awelux_Subject: Linux on Universal and Magician booting
23:52.32awelux_did you change that just now or was someone there before ?
23:54.44cr2there was such subtitle.
23:55.01cr2probably someboday has asked if it works.
23:55.31awelux_ah.
23:56.35cr2http://www.handhelds.org/moin/moin.cgi/SupportedHandheldSummary
23:56.46cr2you can also change this page :)
23:56.55awelux_Next thing is a ramdisk and ssh.
23:57.09cr2and the touchscreen driver :)
23:57.53cr2you know the sd/mcc irq, so you only need to find the powerup gpio.
23:58.05cr2s/mcc/mmc/
23:58.53cr2then you will get opie/gpe running.
23:59.10awelux_I'm not allowed to edit that page

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.