| 00:00.17 | cr2 | Can't open serial port . |
| 00:00.24 | cr2 | it's probably the bluetooth. |
| 00:00.53 | BabelO | cr2: there is only getty on this image |
| 00:01.00 | cr2 | BabelO: ok. |
| 00:01.10 | BabelO | cr2: yes desactivate /etc/init.d/local script |
| 00:01.55 | cr2 | getty is /sbin/tinylogin |
| 00:05.44 | cr2 | BabelO: does not work at all |
| 00:06.18 | cr2 | it seems i need to use the image on goxboxlive. |
| 00:06.22 | BabelO | i just start again this image on ba right ow .. |
| 00:06.53 | cr2 | how do i copy install roadmap there ? |
| 00:08.53 | BabelO | cr2: you need bin/qtroadmap qtroadfgps buildus and some additionnal librarie popt |
| 00:09.04 | BabelO | cr2: i try to make a package for you |
| 00:09.25 | cr2 | ok, thanks. |
| 00:09.33 | BabelO | cr2: ipk does not work onroadmap cvs ? |
| 00:14.14 | cr2 | ipk ? |
| 00:14.31 | cr2 | i think it's broken. |
| 00:14.56 | cr2 | tar.gz is better. i'm not going to unistall it :) |
| 00:16.21 | BabelO | cr2: i got this error on roadmap build now /tmp/ccA93Kz8.s:621: Error: no such instruction: `swpb %eax,%dl,[%ecx]' |
| 00:17.20 | cr2 | bizarre. |
| 00:17.28 | cr2 | it's x86 asm |
| 00:18.31 | BabelO | togh :) bad compiler settings ;) |
| 00:19.36 | BabelO | i need again the magic command line .... do you have it ? |
| 00:20.51 | cr2 | which one ? |
| 00:23.33 | BabelO | something like CXX=arm-linux-gcc DESKTOP=QPE4 QTDIR=..... make |
| 00:23.51 | cr2 | hmm. i'll have a look. |
| 00:24.55 | cr2 | hehe. the logs say |
| 00:24.59 | cr2 | LANG=C make CROSS_COMPILE=~/stuff/tmp/cross/bin/arm-angstrom-linux-gnueabi- DESKTOP=QPE4 QTOPIADIR=~/qtopia-opensource-4.3.0/qtopiacore/qt/src/ SHAPEFILES=NO EXPAT=NO RANLIB=~/stuff/tmp/cross/bin/arm-angstrom-linux-gnueabi-ranlib LDFLAGS='-L ~/stuff/tmp/staging/arm-angstrom-linux-gnueabi/lib' |
| 00:25.14 | BabelO | ok good ;) |
| 00:25.24 | cr2 | EXPAT=NO will kill gpx tracks. |
| 00:27.09 | cr2 | but it's not an issue for now. |
| 00:49.15 | cr2 | kbluetooth crashed ;) |
| 00:59.35 | BabelO | cr2: good night, too late to understand why my compile fail |
| 01:04.49 | cr2 | ok. good night |
| 01:49.44 | *** join/#htc-linux ellisway (n=ellis@homealone.good1.com) |
| 02:56.11 | dcordes | moin |
| 03:09.00 | *** join/#htc-linux the_sys0p (i=the_sys0@gateway/tor/x-8aa0bb32f325fdf4) |
| 03:24.55 | *** join/#htc-linux dcordes_ (n=user0815@f048033146.adsl.alicedsl.de) |
| 04:10.13 | *** join/#htc-linux Zoolooc (n=fredsiba@p5495087A.dip0.t-ipconnect.de) |
| 04:57.49 | *** join/#htc-linux ellisway (n=ellis@homealone.good1.com) |
| 06:05.19 | *** join/#htc-linux jgw (n=jgw@unaffiliated/jgw) |
| 07:41.01 | *** join/#htc-linux rob_w (n=bob@X0b9d.x.pppool.de) |
| 08:50.33 | *** join/#htc-linux pH5 (n=ph5@e178215117.adsl.alicedsl.de) |
| 09:07.01 | *** join/#htc-linux the_sys0p (i=the_sys0@gateway/tor/x-9571e190ae484040) |
| 09:09.02 | *** join/#htc-linux goxboxlive (n=goxboxli@81.80-202-132.nextgentel.com) |
| 09:16.19 | *** join/#htc-linux |dion| (n=dion@inhex.net) |
| 09:18.26 | *** join/#htc-linux AC_Mast (i=s@gateway/tor/x-8c05652f70383718) |
| 09:19.01 | *** join/#htc-linux TeringTu1y (n=maarten@195-241-125-243.ip.telfort.nl) |
| 09:26.53 | goxboxlive | morning |
| 09:27.00 | pH5 | morning goxboxlive |
| 09:27.24 | *** join/#htc-linux pleemans (n=peter@d51A5E76A.access.telenet.be) |
| 10:21.48 | *** join/#htc-linux goxboxlive (n=goxboxli@81.80-202-132.nextgentel.com) |
| 10:38.44 | *** join/#htc-linux goxboxlive (n=goxboxli@81.80-202-132.nextgentel.com) |
| 10:41.31 | *** join/#htc-linux Robwoerle (n=bob@X03ea.x.pppool.de) |
| 11:02.29 | goxboxlive | cr2: awake |
| 12:30.33 | dcordes_ | hi chan, Zoolooc |
| 12:32.28 | Zoolooc | hello |
| 12:33.38 | dcordes_ | could you help me with the msm kernel hacks? |
| 13:34.47 | dcordes_ | Zoolooc: ping |
| 13:35.22 | Zoolooc | dcordes_: pong |
| 13:46.54 | *** join/#htc-linux the_sys0p (i=the_sys0@gateway/tor/x-762eebfe943d90e6) |
| 13:53.01 | dcordes_ | Zoolooc: is there anything I have to do before make menuconfig for the new objects to appear? |
| 14:51.17 | dcordes_ | does somebody know any documentation on how to compile msm kernel? Which toolchain do I need and where can I get it? |
| 14:54.41 | *** join/#htc-linux TeringTuby (n=maarten@195-241-125-243.ip.telfort.nl) |
| 15:03.47 | Kevin2 | Hi |
| 15:04.32 | pH5 | hi Kevin2 |
| 15:04.44 | pH5 | have you seen what I've done to your htc-egpio driver yesterday? |
| 15:04.49 | Kevin2 | Yes. |
| 15:04.58 | pH5 | good? bad? both? |
| 15:05.05 | Kevin2 | Is gpiolib in the kernel now? |
| 15:05.31 | pH5 | Kevin2: it is going to be. basic gpiolib is in 2.6.23-rc3-mm1 and PXA gpiolib support will follow soon. |
| 15:06.01 | Kevin2 | I don't understand the gpio chip having a max of 32 gios. Why is that? |
| 15:07.04 | pH5 | because that is the common size of a gpio bank on 32bit architecture |
| 15:07.57 | Kevin2 | Well, I preferred psolovosky's 256 per chip. :-) |
| 15:09.29 | Kevin2 | In your modifications - what happens to egpio_to_irq? |
| 15:13.13 | Kevin2 | I think it would be worthwhile to introduce two new inline functions - egpio_readw and egpio_writew - that abstract out the bus shift stuff. (I had that in my local tree but didn't get around to pushing it up.) |
| 15:13.55 | pH5 | Kevin2: I don't mind at all how it is implemented. I just want it upstream some time :) |
| 15:14.09 | pH5 | the gpio_to_irq mapping is not implemented in gpiolib at all, |
| 15:14.35 | pH5 | so in the case of PXA arch, where the mapping is currently IRQ=GPIO+64 this would have to be taken care of by the board init code. |
| 15:15.26 | Kevin2 | Well, that seems to be a pretty serious limitation. How will things like gpio_keys work? |
| 15:16.35 | pH5 | Kevin2: as it is now, we'd have to align the IRQ numbers manually, leaving holes in the irq_desc array. |
| 15:17.16 | *** join/#htc-linux |dion| (n=dion@inhex.net) |
| 15:18.08 | Kevin2 | The code in drivers/input/keyboard/gpio_keys.c today calls gpio_to_irq -- that means we wouldn't be able to use it with anything but arch gpios. |
| 15:18.13 | pH5 | egpio_readw/writew sounds like a sane thing to do. |
| 15:18.46 | pH5 | Kevin2: I have the same problem on magician with the headphone insert irq |
| 15:19.13 | Kevin2 | As for your transformations - I don't have any objections. |
| 15:19.18 | pH5 | I'd try to shuffle the gpios so that the input gpios come in the beginning and leave some irqs empty. |
| 15:19.51 | pH5 | Kevin2: should we take that discussion to lkml or lak somehow? |
| 15:19.58 | Kevin2 | I think this gpiolib stuff is a bit silly. Why don't we just come up with a real interface? It can't be that hard to convert the users over to something sane. |
| 15:20.16 | pH5 | I had the same reservations as you, but for my one single GPIO/IRQ it really wasn't worth thinking about that much. |
| 15:22.26 | Kevin2 | I agree we should take the discussion to lkml. (And I see other people objecting to the interface.) |
| 15:22.39 | Kevin2 | s/I see/I did see/ |
| 15:28.52 | goxboxlive | Kevin2: Read the 5th section: http://forum.xda-developers.com/showpost.php?p=1690028&postcount=1741 |
| 15:34.28 | Kevin2 | goxboxlive: Yeah - I've been waiting for a repository at linuxtogo. I want to be able to check the scripts in somewhere. |
| 15:34.42 | goxboxlive | Kevin2: ok. |
| 15:35.09 | goxboxlive | Kevin2: talk to florian . |
| 15:43.46 | cr2 | hi |
| 15:44.50 | cr2 | Kevin2: i have compared the decoded SD/SDIO traces on asic3 and ati. while the registers look the same, the operation modes are different. |
| 15:46.05 | cr2 | ATI works in a 100% irq driven environment. asic3 is much more complex, and has very little amount of external interrupts. |
| 15:47.27 | *** join/#htc-linux BabelO (n=Fabrice@lun34-2-82-238-28-28.fbx.proxad.net) |
| 15:49.37 | BabelO | hi |
| 15:49.59 | Kevin2 | cr2: Okay. Does that me we use two different drivers? |
| 15:56.39 | goxboxlive | cr2: fiphone and gps pyroute: http://scap.linuxtogo.org/ |
| 15:56.50 | goxboxlive | hi BabelO |
| 15:59.42 | cr2 | Kevin2: it will be necessary to make some substantial changes to asic3_mmc. |
| 15:59.59 | cr2 | goxboxlive: is it a raster-based program ? |
| 16:00.13 | goxboxlive | cr2: i have no idea, i just saw it |
| 16:00.36 | cr2 | goxboxlive: to be honest, i dislike the tomtom-style interface. |
| 16:01.39 | pH5 | wow osm berlin coverage sure improved since I looked at it the last time! |
| 16:01.40 | cr2 | goxboxlive: it's unlikely that they can render such maps in realtime |
| 16:02.10 | pH5 | heh, I just killed a magician |
| 16:02.33 | cr2 | goxboxlive: the tiles look like being rendered with osmarender, also a web2.0 technology ;) |
| 16:02.36 | cr2 | pH5: ? |
| 16:03.07 | pH5 | flashing unfinished u-boot over the htc bootloader instead of after it |
| 16:03.16 | BabelO | cr2: , goxboxlive navit picture ;) and roadmap on page http://scap.linuxtogo.org/index.php?page=8 |
| 16:03.32 | cr2 | goxboxlive: it's osm->svg->png renderer. takes some time even on quad-xeon... |
| 16:04.34 | cr2 | goxboxlive: looks like the map of Munich . |
| 16:05.19 | cr2 | it may be interesting to create the navit skin too. |
| 16:05.42 | cr2 | pH5: are the jtag pins known for the magician ? |
| 16:06.01 | cr2 | pH5: afaik they are known for the universal. |
| 16:07.29 | cr2 | Kevin2: is it possible to trace the CPLD pins over suspend ? some of them are used only during suspend. |
| 16:10.39 | Kevin2 | cr2: Give it a shot. |
| 16:10.57 | Kevin2 | You can also query the physical addresses by adding them to the RESUMETRACES list. |
| 16:11.18 | pH5 | cr2: afaik jtag pinout on mag. is not known. maybe this is an opportunity here :) |
| 16:11.34 | Kevin2 | But note - on apache - the egpio address wasn't even registered with the pxa controller at resume time. |
| 16:11.53 | pH5 | I need to buy a new one, or better two. Does anybody here have experience with baking PCBs? |
| 16:13.03 | cr2 | Kevin2: on resume they are not used, only during suspend. can you point me to an example for RESUMETRACES ? |
| 16:13.07 | pH5 | maybe buying a defect one and baking that is a better idea. |
| 16:13.24 | cr2 | pH5: they say 230°C is needed |
| 16:13.39 | cr2 | pH5: yes, with a broken LCD probably. |
| 16:13.44 | Kevin2 | Oh, if you're looking for the suspend actions, then just trace them as you normally would. |
| 16:14.14 | cr2 | Kevin2: is the log buffered in RAM ? |
| 16:15.08 | cr2 | pH5: i can send you the BA if you want :) |
| 16:16.12 | Kevin2 | cr2: Yes. Haret installs a "irq section" chunk of code into physical memory. Part of that memory is a "trace buffer". The trace buffer collects info on what happens, and periodically the main haret process pulls out the entries from the trace buffer and prints them. |
| 16:16.51 | Kevin2 | When you suspend/resume, it just means the haret process goes a longer time to polling the trace buffer. No big deal. |
| 16:17.32 | pH5 | Kevin2: btw, did you have an idea why the msm imageon register readouts are failing? |
| 16:17.47 | Kevin2 | addlist resumetraces cp 15 0 1 0 0 ; addlist resumetraces cp 15 0 2 0 0 ; addlist resumetraces cpsr |
| 16:18.08 | Kevin2 | are all traces i've done. Tracing physical memory also works - let me see if I can find some examples. |
| 16:18.56 | Kevin2 | addlist resumetraces 0x0a000000 0 16 none |
| 16:19.06 | Kevin2 | addlist resumetraces 0x48000008 0 32 none |
| 16:20.01 | cr2 | Kevin2: thanks, i will try. |
| 16:20.25 | Kevin2 | pH5: No. We did see that the mmu tables have some undefined bits which could cause the exceptions. However, I have to wonder if the addresses were valid. |
| 16:21.14 | cr2 | pH5: have you seen that kaiser uses SDRAM for the framebuffer ? i suspect that the 7200A android stuff does not apply in full to 7200. |
| 16:22.18 | Kevin2 | cr2: Oh, to use resumetraces you must define the RESUMEADDR variable. |
| 16:23.52 | pH5 | cr2: I was already thinking about buying a used BA to better be able to test the overlapping bits (pasic2/3+ds1wm, calypso, etc.) |
| 16:23.53 | pH5 | Kevin2: ok. |
| 16:26.02 | cr2 | pH5: should i send it to you? i spend most of the time researching ATI, and have no time for BA. |
| 16:30.31 | pH5 | cr2: yes, that'd be great. |
| 16:32.57 | cr2 | ok, write me an e-mail. |
| 16:38.12 | cr2 | goxboxlive: if i plug the usb keyboard, then qtopia goes crazy. i guess the same will happen if i'll plug the standard spi keyboard. |
| 16:39.21 | cr2 | goxboxlive: qtopia on athena is amazingly fast. roadmap will bog it down soon :) |
| 16:44.36 | *** join/#htc-linux |dion| (n=dion@inhex.net) |
| 16:49.53 | pH5 | ~lart evolution, can't handle the gmail imap account |
| 16:49.53 | apt | nabs the moon and broadsides evolution, can't handle the gmail imap account with the sea of tranquility |
| 16:50.09 | pH5 | cr2: mail should be sent, if evo didn't screw up |
| 17:03.01 | cr2 | pH5: yes, i've got it. will go to the post tomorrow :) |
| 17:03.14 | cr2 | bbl |
| 17:09.06 | Kevin2 | bbl |
| 17:51.54 | dcordes_ | where can I get arm-2007q3-51-arm-none-linux-gnueabi-i686-pc-linux-gnu.tar.bz2 = |
| 18:00.07 | pH5 | 2007q3 sounds like the codesourcery toolchain |
| 18:00.12 | pH5 | try http://www.codesourcery.com/gnu_toolchains/arm/download.html |
| 18:07.18 | dcordes_ | pH5: that's just what they http://titanux.lwp.ca/wiki/index.php/Compiling_Kernel use to compile msm kernel. Shall I use a different toolchain? |
| 18:09.13 | pH5 | dcordes_: if it works for them and is easy to install, why not use it. I always use the OpenEmbedded integrated toolchains, but there you'd need to wait for it to build the compiler itself before you can use it. |
| 18:09.51 | dcordes_ | I have a set OE for zaurus akita (armv5te) |
| 18:10.08 | dcordes_ | but no idea how to compile a kernel which is not in the present metadata |
| 18:10.19 | dcordes_ | or how to add it |
| 18:14.35 | pH5 | dcordes_: you need to tell make about the location and prefix of the cross compilation toolchain |
| 18:15.06 | pH5 | for me it is |
| 18:15.09 | pH5 | export CROSS_COMPILE=/home/ph5/src/oe/build/angstrom/cross/bin/arm-angstrom-linux-gnueabi- |
| 18:15.21 | dcordes_ | pH5: make ARCH="arm" CROSS_COMPILE="/bin/arm-none-linux-gnueabi-" zImage |
| 18:15.35 | dcordes_ | that's taken from the howto on titanux page |
| 18:15.51 | pH5 | dcordes_: yes, that worked for me so far for any kernel or u-boot compilation |
| 18:16.14 | pH5 | just the location has to be adapted to the actual location of your oe toolchain |
| 18:17.51 | dcordes_ | where is it in normal case? |
| 18:18.48 | dcordes_ | ah sorry didn't notice the example you gave |
| 18:24.07 | dcordes_ | pH5: when I "export ARCH=arm; export CROSS_COMPILE=pathtomyOEtoolchain" before I can just hit make zImage in the same console? |
| 18:25.11 | pH5 | dcordes_: yes |
| 18:27.00 | dcordes_ | pH5: another ting: I tried to add the hermes' vsfb driver ( http://handhelds.org/cgi-bin/cvsweb.cgi/linux/kernel26/arch/arm/mach-s3c2442/htchermes/htchermes.c?only_with_tag=K2-6-21-hh9 ) to http://git.android.com/?p=linux-msm.git;a=tree;f=arch/arm/mach-msm;h=141a71859d0b5493b35e4fa79e30b7921b099636;hb=6230922501b6053e99b42edc7b20ea4f356be67e |
| 18:28.43 | dcordes_ | I replaced all hermes names with kaiser and copied the refering entries from Makefile and Kconfig from the handheld tree to the according Kconfig and Makefile inside the msm folder above. Is that the right way to do it? |
| 18:31.26 | pH5 | dcordes_: I think so, yes. The important thing is that you adapt the values in the fb_fix_screeninfo structure |
| 18:31.41 | pH5 | (smem_start, smem_len, line_length, etc.) |
| 18:32.21 | pH5 | hm, width/height should be the same. but frame buffer location is certainly different, right? |
| 18:48.52 | dcordes_ | pH5: I only altered .smem_start = 0x10000000, |
| 18:48.57 | dcordes_ | cr2: told me to |
| 18:52.42 | goxboxlive | cr2: I like the tomtom style, it's clean and easy. |
| 18:53.19 | goxboxlive | pH5: oh no, did you brick your magician. I guess it will slow down the bootloader then :-( |
| 19:02.35 | pH5 | goxboxlive: yes, I feel stupid now :/ |
| 19:02.45 | pH5 | dcordes_: did you already try to compile the kernel? |
| 19:04.54 | *** join/#htc-linux pleemans (n=peter@d51A5E76A.access.telenet.be) |
| 19:08.20 | *** join/#htc-linux kiozen (n=oeichler@rgnb-4db08037.pool.einsundeins.de) |
| 19:14.18 | kiozen | cr2 ? |
| 19:19.38 | cr2 | hi kiozen |
| 19:19.45 | kiozen | hi, |
| 19:19.49 | cr2 | goxboxlive: do you run wince2005se ? |
| 19:19.56 | kiozen | still struggling with gdal |
| 19:20.13 | cr2 | kiozen: for arm ? |
| 19:20.18 | kiozen | yes |
| 19:20.30 | kiozen | second .. |
| 19:21.29 | kiozen | http://pastebin.ca/797095 |
| 19:21.49 | kiozen | output tells me that it can't download tar.gz |
| 19:22.09 | kiozen | no wonder if bitbake appends the name by "-1.0" |
| 19:22.33 | kiozen | I tried to overide the name in the *bb |
| 19:22.33 | cr2 | yes, which local.conf are you using ? |
| 19:22.42 | kiozen | second.. |
| 19:23.18 | cr2 | it's hopeless to override something in oe :) either it works or not. |
| 19:23.27 | kiozen | http://pastebin.ca/797100 |
| 19:23.35 | kiozen | yes I know now, too :) |
| 19:23.52 | cr2 | i'll compare with mine. |
| 19:24.20 | kiozen | everything else is working fine |
| 19:24.49 | kiozen | kernel, qtopia & co compiled with the same toolchain from oe |
| 19:25.29 | kiozen | it's like a small distro to start development on |
| 19:26.04 | kiozen | hope to see a raster map soon :) |
| 19:26.58 | cr2 | looks good. |
| 19:27.10 | cr2 | i'll update to the today's version and check. |
| 19:27.16 | kiozen | ok |
| 19:27.53 | cr2 | 24 revs since yesterday. |
| 19:32.09 | kiozen | hm, "mtn update" says nothing to do |
| 19:33.24 | kiozen | you simply moved the *bb from 1.3.2 to 1.4.2, didn't you? |
| 19:37.24 | dcordes_ | pH5: sorry I was getting food. You still there? I took the prepared defconfig for halibut in the msm tree and copied it to the main kernel dir as .config. Then I ran make xconfig with the right vars set. I can find my kaiser lcd thing there with the search option but can't find it to activate |
| 19:39.43 | cr2 | kiozen: yes. don't see any changes for gdal. |
| 19:39.43 | pH5 | dcordes_: is LCD_CLASS_DEVICE enabled? HTCHERMES_LCD depends on that. |
| 19:40.07 | dcordes_ | I renamed HTCHERMES_LCD HTCKAISER_LCD is that a problem? |
| 19:40.49 | dcordes_ | I can find that with the search funktion but not in the "menu" do enable it. Under which should I look for it? |
| 19:41.48 | pH5 | dcordes_: no, that's ok. but you have to enable the backlight & lcd class device options in drivers/video/backlight first |
| 19:41.57 | dcordes_ | ok |
| 19:42.06 | pH5 | grep LCD_CLASS_DEVICE .config |
| 19:43.36 | dcordes_ | I found Device Drivers/ Graphics support/ |
| 19:43.41 | cr2 | kiozen: gcc needs to be recompiled. |
| 19:43.51 | dcordes_ | Backlight & LCD device support and Display device support |
| 19:44.31 | pH5 | dcordes_: that's it |
| 19:45.03 | kiozen | cr2: but that can't be the reason why it appends the archive name by "-1.0" |
| 19:45.26 | dcordes_ | pH5: ok enabled all that. Where do I find my htckaiser_lcd now? |
| 19:45.49 | dcordes_ | I have the file in /arch/mach-msm/htckaiser_lcd.c |
| 19:46.17 | pH5 | dcordes_: where do you find the rest of the options in arch/mach-msm/Kconfig ? |
| 19:46.27 | dcordes_ | let me check |
| 19:47.04 | dcordes_ | it's arch/arm/mach-msm, sorry |
| 19:48.13 | dcordes_ | they are in System Type |
| 19:49.00 | dcordes_ | hcan't find the htc_kaiser_lcd there |
| 19:50.18 | dcordes_ | pH5: when I look for kaiser with xconfig's search function the result is strange: it shows under Option the entry "HTC Kaiser VSFB LCD support" but it is grey like it would be disabled, and there is a line in front of it that points to nothing |
| 19:50.43 | goxboxlive | cr2: if that is wince6 then yes |
| 19:50.55 | dcordes_ | pH5: and when I doubble click the entry, it doesn't jump to the position |
| 19:51.05 | dcordes_ | I fear I did something wrong in the refering Kconfig. |
| 19:51.22 | pH5 | dcordes_: try to enable it manually in .config, or pastebin the Kconfig in that folder. I didn't download the msm kernel yet. |
| 19:51.32 | cr2 | goxboxlive: can you make 'dump asic3' and 'dump asic3gpio' on it ? |
| 19:51.33 | dcordes_ | one second. I pastebin |
| 19:52.14 | goxboxlive | cr2: in linux or with Haret? |
| 19:52.35 | dcordes_ | pH5: I tihnk I found an error. I will try to fix it myself |
| 19:54.39 | dcordes_ | pH5: I have a seperate directory for the htckaiser_lcd.c and its Kconfig and Makefile |
| 19:54.48 | dcordes_ | inside linux-msm/linux-msm/arch/arm/mach-msm |
| 19:54.55 | cr2 | goxboxlive: in haret |
| 19:55.36 | dcordes_ | pH5: shall I leave it that way or better put the htckaiser_lcd.c into /linux-msm/arch/arm/mach-msm instead of /linux-msm/arch/arm/mach-msm/htckaiser? |
| 19:55.39 | pH5 | dcordes_: two choices. a) don't do that, b) include that directory in the Kconfig and Makefile one layer above |
| 19:55.54 | dcordes_ | that's what I tried |
| 19:55.57 | pH5 | dcordes_: or just take the code from _lcd.c and put it in the main file. |
| 19:55.58 | dcordes_ | (b) |
| 19:56.20 | dcordes_ | I think I try a) put it directly into linux-msm/arch/arm/mach-msm |
| 19:56.44 | goxboxlive | cr2: what shoudl be the content of the default.txt? |
| 19:57.20 | cr2 | goxboxlive: hm. they are not in haret. |
| 19:58.06 | dcordes_ | pH5: http://pastebin.ca/797132 that's htckaiser_lcd.c now inside linux-msm/arch/arm/mach-msm |
| 19:59.39 | pH5 | dcordes_: did you put the HTCKAISER_LCD option in linux-msm/arch/arm/mach-msm/Kconfig now? |
| 19:59.49 | dcordes_ | yes about to paste it |
| 20:00.09 | dcordes_ | pH5: http://pastebin.ca/797136 |
| 20:00.46 | pH5 | dcordes_: but where is MACH_KAISER? |
| 20:00.56 | dcordes_ | pH5: and the "Makefile" http://pastebin.ca/797137 . All three are inside linux-msm/arch/arm/mach-msm |
| 20:01.37 | dcordes_ | pH5: ahhh that will be the htchermes.c I removed.... |
| 20:02.50 | dcordes_ | I will try to fix that |
| 20:06.16 | dcordes_ | pH5: http://pastebin.ca/797144 here. I noticed all those references to s3c24xx. How do I get around this? |
| 20:07.02 | *** join/#htc-linux Marex-notebook (n=marex@vasut.kolej.mff.cuni.cz) |
| 20:08.40 | dcordes_ | pH5: I mean there is no linux/kernel26/arch/arm/mach-s3c2442/ in msm tree.. |
| 20:09.46 | dcordes_ | pH5: the original Kconfig entry for htchermes.c says select CPU_S3C2440; select CPU_S3C2442 |
| 20:09.50 | pH5 | dcordes_: you should not copy htchermes.c, hermes has a s3c2442 cpu |
| 20:10.06 | pH5 | can't you start from the halibut.c file? |
| 20:10.27 | dcordes_ | pH5: oh ok. But what about the MACH_KAISER thing? |
| 20:11.03 | pH5 | MACH_KAISER is the option to enable the board init code (htckaiser.c) |
| 20:11.17 | pH5 | just copy the MACH_HALIBUT option in Kconfig that is already there |
| 20:11.30 | pH5 | and use it in Makefile to enable compilation of htckaiser.c |
| 20:11.56 | dcordes_ | hm let me get this sorted |
| 20:13.31 | dcordes_ | pH5: http://pastebin.ca/797136 in line 56 I replace MACH_KAISER (because I don't have this) with MACH_HALIBUT? |
| 20:14.52 | pH5 | dcordes_: no, I suggest you add a new MACH_KAISER block after line 11 (copy the MACH_HALIBUT block) |
| 20:15.22 | pH5 | it should depend on ARCH_MSM and have a nice title and description |
| 20:16.13 | pH5 | then you add a line "obj-$(CONFIG_MACH_KAISER) += htckaiser.o" line to the Makefile |
| 20:16.30 | pH5 | and let KAISER_LCD depend on MACH_KAISER |
| 20:16.40 | dcordes_ | alright |
| 20:18.13 | goxboxlive | cr2: ? |
| 20:19.36 | goxboxlive | pH5: is there anything i could test with uboot? Now we know how to overwrite the bootloader wich dont want to do, do we know how to not overwrite it? |
| 20:23.54 | dcordes_ | pH5: Kconfig plus Makefile http://pastebin.ca/797163 |
| 20:25.53 | pH5 | dcordes_: looks good |
| 20:26.17 | dcordes_ | ok I recopied halibut-defconfig to maindir/.config |
| 20:26.23 | dcordes_ | set the vars and now I run xconfig |
| 20:26.44 | dcordes_ | pH5: export CROSS_COMPILE=/media/archiv/OE/build/tmp/angstrom/cross/bin/arm-angstrom-linux-gnueabi- ok? |
| 20:27.44 | pH5 | you could rename htckaiser.c to board-htckaiser.c and call the config option MACH_HTCKAISER to better fit in. |
| 20:28.09 | pH5 | can you now select kaiser and the lcd in xconfigß |
| 20:28.12 | dcordes_ | I have no htckaiser.c . I though I should get rid of it because of the wrong cpu thing? |
| 20:28.14 | pH5 | s/ß/?/ |
| 20:29.01 | dcordes_ | or do you mean htckaiser_lcd.c? |
| 20:29.11 | pH5 | dcordes_: no, you need that htckaiser.c. it is the basic board initialization file and includes the list of drivers to be loaded. |
| 20:29.29 | pH5 | I just said that you maybe shouldn't copy htchermes.c, but maybe halibut.c instead |
| 20:29.43 | pH5 | but you'll have to remove board specific stuff that doesn't apply to kaiser in any case |
| 20:30.00 | dcordes_ | ok that replies my next question.. |
| 20:30.26 | pH5 | I'd copy halibut.c and add in the vsfb platform_device from htchermes.c |
| 20:30.49 | dcordes_ | there is /arch/arm/mach-msm/board-halibut.c |
| 20:31.19 | pH5 | dcordes_: aye, that's the one I mean. copy that to board-htckaiser.c (and update Makefile accordingly) |
| 20:31.53 | dcordes_ | http://pastebin.ca/797170 that's board-halibut.c |
| 20:31.57 | pH5 | and then edit it, remove anything that looks like it doesn't belong there ;) and add vsfb |
| 20:32.22 | pH5 | the smc9x stuff for example has to go |
| 20:32.27 | pH5 | *91x |
| 20:32.50 | pH5 | and the halibut_keypad_device, too, probably |
| 20:33.13 | pH5 | and then add the vsfb from htchermes.c to the devices[] array |
| 20:33.31 | dcordes_ | Makefile: obj-$(CONFIG_MACH_KAISER) += board_htckaiser.o |
| 20:34.34 | pH5 | cr2: are msm_init_gpio/irq and msm_pmrr_config safe to call? |
| 20:35.10 | pH5 | dcordes_: actually, you could comment out all the platform_devices and only add vsfb for now. |
| 20:35.38 | pH5 | don't use an underscore _ , but a hyphen - |
| 20:35.39 | dcordes_ | pH5: ok before that could you take a look at the Makefile again? http://pastebin.ca/797173 |
| 20:36.15 | dcordes_ | in which place? |
| 20:36.35 | pH5 | and I'd use MACH_HTCKAISER (well, I'd prefer MACH_KAISER, but HTCKAISER is what is registered in the machine registry already) and put it right after the MACH_HALIBUT line |
| 20:37.06 | pH5 | like this: http://pastebin.ca/797175 |
| 20:37.42 | pH5 | (renaming it to HTCKAISER in Kconfig too) |
| 20:39.15 | dcordes_ | ok I saved that and put config MACH_HTCKAISER and MACH_HTCKAISER as KAISER_LCD depend |
| 20:39.19 | dcordes_ | no try my luck again? |
| 20:39.24 | pH5 | dcordes_: yes :) |
| 20:40.28 | dcordes_ | ok I have a checkboc HTC Kaiser phone aka HTC TyTN II now in System Type |
| 20:40.54 | dcordes_ | BUT cannot find HAC Kaiser VSFB LCD support now |
| 20:41.07 | dcordes_ | although it is in the search results, again grey |
| 20:41.48 | pH5 | and LCD_CLASS_DEVICE is enabled? |
| 20:42.48 | dcordes_ | nope wasn't and now I can see and check vsfb :D :D |
| 20:43.31 | dcordes_ | I didn't edit board-htckaiser.c yet |
| 20:43.46 | pH5 | dcordes_: now the fun begins |
| 20:43.52 | pH5 | a) edit board-htckaiser.c |
| 20:44.03 | pH5 | b) make zImage (fails) |
| 20:44.10 | pH5 | --> back to a) |
| 20:44.14 | pH5 | and so on :) |
| 20:44.16 | dcordes_ | ok |
| 20:46.09 | dcordes_ | editing kernels is fun! |
| 20:46.54 | dcordes_ | you told me get rid of platform_device entries. Now many lines contain this |
| 20:47.12 | dcordes_ | I guess I'm not to remove those lines but all blobs that have them? |
| 20:47.28 | dcordes_ | and also #include <linux/platform_device.h> ? |
| 20:48.02 | pH5 | dcordes_: no. |
| 20:48.21 | pH5 | you should remove complete blobs, but only the smc91x and keypad stuff |
| 20:48.56 | pH5 | then /* comment out */ the remaining lines in the devices[] array and add vsfb instead, like it is done in htchermes.c |
| 20:48.56 | dcordes_ | from static struct resource smc91.. until the first }; after it? |
| 20:49.20 | pH5 | dcordes_: yes |
| 20:49.58 | dcordes_ | what's the difference between extern and static struct? |
| 20:51.10 | pH5 | "static" means a variable is for this .c file only, the optimizer can throw away its symbol |
| 20:51.25 | pH5 | without static means, the local variable could be referenced elsewhere, so the symbol is kept |
| 20:51.44 | pH5 | "extern" means that the variable is not local, but in some other .c file |
| 20:51.45 | dcordes_ | ah ok |
| 20:51.52 | pH5 | so there it has to be declared without static. |
| 20:53.28 | dcordes_ | do you have the file local? |
| 20:53.48 | dcordes_ | I commented out 43-63 |
| 20:59.42 | dcordes_ | pH5: and I added the static struct platform_device htckaiser_lcd = { blob from the htchermes.c file from hh.org tree! |
| 20:59.47 | dcordes_ | is that right? |
| 21:03.29 | dcordes_ | pH5: in xconfig, do I select both Halibut Board and HTC Kaiser phone? |
| 21:09.47 | dcordes_ | ok I got around the board-htckaiser.c errors by comenting out everything unnecessary |
| 21:10.06 | dcordes_ | pH5: now it mounrs about htckaiser_lcd.c |
| 21:11.02 | dcordes_ | msm tree lacks vsfb |
| 21:12.02 | *** join/#htc-linux rmoravcik (n=rmoravci@ip-89-102-255-171.karneval.cz) |
| 21:16.29 | pH5 | dcordes_: I'll fetch the kernel tomorrow. |
| 21:16.29 | *** join/#htc-linux BabelO (n=Fabrice@lun34-2-82-238-28-28.fbx.proxad.net) |
| 21:16.30 | pH5 | you only have to select HTC Kaiser |
| 21:16.47 | pH5 | and vsfb has to be extracted from the hh.org tree somewhere |
| 21:17.04 | dcordes_ | I'm trying to figure how to fetch it right now to find somewhere |
| 21:17.11 | dcordes_ | cvs looks complicated |
| 21:17.46 | pH5 | (or from the "e-series framebuffer support" mail by ian molton to lkml or lak) |
| 21:19.45 | dcordes_ | 23 Nov 13:13ian |
| 21:21.17 | dcordes_ | pH5: can't I just apply that whole diff file to my local msm tree? |
| 21:22.34 | pH5 | dcordes_: sure |
| 21:27.00 | dcordes_ | pH5: http://pastebin.ca/797256 would that be the patch I want to apply? |
| 21:28.31 | pH5 | dcordes_: yes, that looks like the one |
| 21:29.58 | dcordes_ | I put it as vsfb.diff into linux-msm |
| 21:30.00 | dcordes_ | how to apply it? |
| 21:32.00 | dcordes_ | pH5: |
| 21:34.29 | dcordes_ | pH5: oh no- does this mess up anything? http://pastebin.ca/797269 |
| 21:34.43 | pH5 | dcordes_: git-apply for example |
| 21:34.55 | dcordes_ | I did the thing in the pastebin.. |
| 21:36.02 | dcordes_ | Shall I restore the old drivers/video/Makefile? |
| 21:36.15 | pH5 | oh, look at the pastebin again |
| 21:36.27 | pH5 | there are some lines broken, probably by the mailer |
| 21:36.53 | pH5 | 102, 123, 125, 134, 140, 146, 148 |
| 21:37.18 | pH5 | those have to be appended to the previous lines again |
| 21:37.22 | dcordes_ | wow how did you figure that so quick? |
| 21:37.42 | dcordes_ | ok there's each time an absatz too much? |
| 21:37.45 | pH5 | dcordes_: it's not the first time I try to apply a patch from a mail :) |
| 21:38.15 | pH5 | the "malformed patch" message in most cases indicated that there are either broken lines or incomplete patches. |
| 21:39.18 | pH5 | the Makefile was patched just fine |
| 21:39.24 | pH5 | but Kconfig failed |
| 21:39.51 | pH5 | open Kconfig and Kconfig.rej in an editor and try to do the changes manually |
| 21:40.09 | dcordes_ | has it changed my Kconfig yet? |
| 21:40.18 | dcordes_ | (until the broken spot) ? |
| 21:41.20 | pH5 | dcordes_: no, the patch contains only one hunk for Kconfig |
| 21:41.35 | pH5 | (the CONFIG_VFSB block) |
| 21:41.42 | dcordes_ | ok |
| 21:41.46 | pH5 | should be easy to add |
| 21:43.48 | dcordes_ | I fixed the diff |
| 21:44.45 | dcordes_ | it failed again |
| 21:44.54 | dcordes_ | I restore the original msm file and try it manually |
| 21:49.17 | dcordes_ | pH5: I don't understand the whole thing. Could you add it for me quick? |
| 21:53.28 | dcordes_ | think I understood the whole thing now |
| 21:53.50 | dcordes_ | pH5: but the patch adds a line source "drivers/video/omap/Kconfig" |
| 21:54.30 | pH5 | dcordes_: no, it does not |
| 21:54.38 | pH5 | only the lines beginning with "+" are to be added |
| 21:54.59 | dcordes_ | the others identify where to put it? |
| 21:55.06 | pH5 | dcordes_: exactly |
| 21:55.09 | dcordes_ | grewat |
| 21:55.12 | pH5 | and that is why a patch like that can fail |
| 21:55.25 | dcordes_ | I noticed the files are differnt, |
| 21:55.33 | pH5 | if something was changed so that it doesn't find the three lines above and below anymore, it will bail out |
| 21:56.05 | dcordes_ | now what about drivers/video/vsfb.c and include/linux/vsfb.h ? |
| 21:56.15 | dcordes_ | ok |
| 21:56.25 | pH5 | those should just be added if the patch is fixed again |
| 21:56.34 | pH5 | no conflicts possible |
| 21:56.39 | dcordes_ | didn't fix the patch |
| 21:56.45 | dcordes_ | I added it manually now |
| 21:56.48 | dcordes_ | should I? |
| 21:57.36 | pH5 | either way is okay |
| 21:57.47 | pH5 | you are working in the git tree, right |
| 21:57.55 | pH5 | can you pastebin the git-diff output? |
| 21:58.03 | dcordes_ | yes |
| 21:58.07 | dcordes_ | I fix patch so I learn |
| 21:59.34 | dcordes_ | pH5: I added up the laste three lines after the + area as they are in the right Kconfig |
| 21:59.46 | dcordes_ | now the patch mourns about + |
| 21:59.47 | dcordes_ | if VT |
| 21:59.56 | dcordes_ | that's the first line after the + area |
| 22:00.59 | dcordes_ | says malformed patch |
| 22:01.30 | pH5 | dcordes_: gotta go now, sorry. |
| 22:01.43 | dcordes_ | ok thank you very much for all the help and valuable information |
| 22:02.03 | dcordes_ | and time. I will give you the git-diff tomorrow |
| 22:02.39 | pH5 | dcordes_: ok, I'll download the msm kernel tomorrow so it'll be easier to make sure we're talking about the same things :) |
| 22:02.43 | pH5 | bye |
| 22:02.50 | dcordes_ | bye |
| 22:09.48 | *** join/#htc-linux Dinde342 (n=kayser@m133.net81-65-172.noos.fr) |
| 22:13.51 | *** join/#htc-linux Mineque (n=Mineque@195.60.65.50) |
| 22:28.18 | *** join/#htc-linux Babel (n=Fabrice@lun34-2-82-238-28-28.fbx.proxad.net) |
| 22:28.39 | dcordes_ | yeay kaiser vsfb built |
| 22:28.41 | dcordes_ | now testing... |
| 22:34.34 | dcordes_ | who would have guessed: Makes kaiser freeze immedeately |
| 23:10.53 | dcordes_ | gnite all |