00:05.53 | dcordes | ~ping Kevin2 |
00:05.54 | apt | pong Kevin2 |
00:21.55 | BabelO | ~pong revertive |
00:21.56 | apt | revertive: PONG!!!!!!!!!!!!!!!!!! |
00:22.21 | revertive | hi BabelO |
00:22.26 | BabelO | hi revertive |
00:22.39 | BabelO | no ts, i just finish a fresh 2.6.26-rc1 ) |
00:22.52 | BabelO | and mmc works well, i boot angstrom-qtopia on it directly |
00:22.52 | cr2 | wow, so many people do not sleep :) |
00:22.59 | revertive | 2.6.26 cool |
00:23.10 | BabelO | cr2: lol, yes |
00:23.11 | cr2 | BabelO: on ba ? ;) |
00:23.31 | BabelO | cr2: on artemis :) with new htc-egpio, but i think i fdo something wrong |
00:23.46 | BabelO | revertive: ts need to wait a little |
00:23.50 | cr2 | hmm. vibra works ? |
00:23.59 | BabelO | now all gpio / irq are working as expected |
00:24.11 | revertive | BabelO: thats ok |
00:24.17 | BabelO | cr2: no, black screen on boot when i use set_gpio_value on cpld |
00:24.44 | cr2 | BabelO: in machine_init() ? |
00:25.03 | BabelO | revertive: i comit in new branch at linuxtogo, so minimal change to 2.6.26 just omap850 is in |
00:25.15 | BabelO | cr2: no in lcd module, like i do in 2.6.21-hh |
00:25.34 | BabelO | cr2: you did not see my pastebin last night ;) |
00:25.43 | cr2 | BabelO: it does not work on athena too, which has an arcane 2 CPLD config, and the location of the second CPLD depends on the first CPLD bits. |
00:25.47 | miknix | cr2, dont know if you read it.. I finally finished the htc-i2c-cpld driver today. both dpad are supported, leds, backlight and vibrator |
00:26.08 | cr2 | BabelO: no, i'm busy doing other things, no time for hacking. |
00:26.10 | miknix | cr2, are you interested in testing it on other devices? herald is known to work too |
00:26.18 | BabelO | cr2: http://pastebin.ca/1014366 |
00:26.30 | BabelO | i just fix the serial detection today |
00:26.39 | cr2 | miknix: i don't have omap850 devices. pH5 has herald |
00:26.53 | cr2 | BabelO: gps works now ? |
00:27.11 | BabelO | cr2: i think yes i can cat /dev/ttyS1 but i miss the cpld power pin |
00:27.14 | miknix | cr2, pH5 has his herald b0rked.. it doesnt pass the bootloader screen |
00:27.25 | cr2 | ok |
00:27.47 | cr2 | BabelO: but if you'd add the cpld power pin, arte crashes too ? |
00:28.29 | BabelO | cr2: yes all cpld gpio access crash http://pastebin.ca/1014865 |
00:28.58 | BabelO | and http://pastebin.ca/1014872 for .h |
00:29.05 | cr2 | BabelO: then you are doing something wrong. |
00:29.14 | BabelO | cr2: i hope so |
00:29.23 | BabelO | looking at it right now |
00:29.30 | miknix | BabelO, what is the serial used for? irda? |
00:29.39 | BabelO | miknix: gps / BT |
00:29.43 | miknix | hum.. |
00:30.39 | miknix | cr2, on i2c chip 3, there is an address which I could found yet that turns off the mmc power |
00:30.43 | cr2 | BabelO: i'm not familiar with the new syntax. |
00:32.08 | BabelO | cr2: syntax is ok according to the magician syntax |
00:32.09 | cr2 | BabelO: do you have a direct link to the magician.c with new htc-egpio ? |
00:32.19 | BabelO | cr2: yes |
00:33.20 | BabelO | cr2: http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-pxa/magician.c;h=badba064dc0400326e74a3e874d9366aa6519c52;hb=2ddcca36c8bcfa251724fe342c8327451988be0d |
00:35.35 | BabelO | cr2: ah i think i know why :) |
00:38.22 | cr2 | BabelO: magician has 32bit reg offsets |
00:38.32 | cr2 | all others use 16 |
00:39.13 | BabelO | cr2: good found my bug, so i need only two beer, not more to found bugs :) |
00:39.30 | cr2 | lol |
00:40.07 | miknix | beer!! |
00:40.15 | miknix | BabelO, have you tried portuguese beer? |
00:40.44 | BabelO | miknix: german beer are better :) |
00:41.30 | miknix | BabelO, http://en.wikipedia.org/wiki/Super_Bock Super Bock has 15 consecutive medals in the "Monde selection" |
00:41.32 | BabelO | cr2: i ve setted .reg_with=8 and in gpio calculation i ve used *16 |
00:42.15 | BabelO | miknix: i ve to try :) |
00:42.31 | miknix | BabelO, :) |
00:42.54 | miknix | BabelO, I recommend the black roasted one.. |
00:43.06 | BabelO | anyway marti solveig music was good, but now did not heard musics or tv :( maybe the sound ... |
00:43.50 | BabelO | cr2: any possibility that i have a hidden usb ping on cpld ? |
00:43.53 | dcordes | aye cr2 |
00:44.20 | dcordes | found some time for kaiser mmutrace issue? |
00:44.23 | miknix | BabelO, is your cpld much different from wizard and herald? |
00:44.41 | miknix | BabelO, besides not being available from i2c |
00:45.38 | BabelO | miknix: i have direct acces to it over memory |
00:45.53 | miknix | BabelO, but the input logic is different? |
00:46.25 | BabelO | miknix: no, it is same logic, that is why pH5 tell you to try to do same logic as htc-egpio driver |
00:46.52 | BabelO | miknix: don't know if i have input on it |
00:48.19 | miknix | BabelO, migrating the htc-i2c-cpld driver to use gpios is fairly simple.. just change the cpld access from i2c to gpios |
00:49.03 | miknix | BabelO, the htc-i2c-cpld driver only acts as abstraction layer to other drivers (dpad, backlight, leds) |
00:49.38 | BabelO | miknix: i mean to drive the other side, to be more generic, because your driver, only works if device wired on cpld are always at same position |
00:50.33 | BabelO | anyway it is good if it works ;) |
00:50.40 | BabelO | good progress in few month |
00:51.26 | miknix | BabelO, how the egpio-driver works? |
00:51.40 | miknix | I dont have it on the tree |
00:52.02 | BabelO | miknix: transparent, you just have to use set_gpio_value or get_gpio_value |
00:52.58 | miknix | BabelO, and how I do chip and address selection? I need to implement some kind of i2c in it right? |
00:54.43 | BabelO | miknix: chip is a range of gpio, if you look here http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=include/asm-arm/arch-pxa/magician.h;h=169b374f9921968031b0b7dccb22673e3366256a;hb=2ddcca36c8bcfa251724fe342c8327451988be0d |
00:55.44 | BabelO | miknix: it use (MAGICIAN_EGPIO_BASE + 8*reg + bit) to found the gpio number you pass to set_gpio_value |
00:55.58 | BabelO | and those gpios are mapped after the processor gpios |
00:56.16 | miknix | hum.. |
00:56.34 | miknix | so each gpio will correspond to each address on the i2c bus? |
00:56.43 | miknix | something like that |
00:59.05 | miknix | BabelO, ok.. I'll migrate the driver to use it then |
00:59.40 | BabelO | miknix: yes it is exactly that |
00:59.59 | BabelO | each gpio correspond to a bit in a i2c chip |
01:00.12 | miknix | BabelO, then the driver I currently have is straightforward to migrate |
01:00.57 | miknix | BabelO, http://pastebin.com/d7d67fde |
01:01.15 | BabelO | miknix: it can be more generic, so you do not need to change code in driver |
01:01.54 | miknix | BabelO, I wont.. I'll only replace the i2c access functions with the 2gpio ones |
01:03.15 | miknix | BabelO, btw: I didnt found the power button address on i2c do you think its there but not available by i2c? |
01:06.04 | BabelO | miknix: don't know |
01:06.29 | BabelO | miknix: you can found it with haret, just log to a file |
01:09.32 | dcordes | miknix: super bock is a strange name for portuguese beer |
01:10.03 | miknix | dcordes, yeah.. the brand is quite old |
01:10.09 | dcordes | by the sound of it I would have said it's a brand from the area of the german language |
01:10.17 | miknix | maybe its that the point? |
01:10.53 | miknix | I bet that few of you know that portugal does a lot of beer production |
01:10.56 | dcordes | that the name actually is from german area? |
01:11.04 | dcordes | yes I didn't know this |
01:12.08 | miknix | dcordes, no, to give a name other than portuguese to get the brand name more more influencing |
01:12.49 | dcordes | It is really an appealing name -already the super is very appealing |
01:12.56 | miknix | dcordes, our second beer in portugal is Sagres which is the name of portuguese city |
01:13.11 | miknix | dcordes, lol, it has normal alcool percentage |
01:13.14 | dcordes | I wonder if I can get those cheap here |
01:13.44 | dcordes | according to wiki Abadia has 6,4 % alcohol |
01:14.09 | miknix | yeah.. its the one with most alcool by superbock |
01:14.29 | dcordes | I like wheat beers |
01:14.38 | dcordes | drinking one as I type |
01:14.51 | miknix | dcordes, http://en.wikipedia.org/wiki/Sagres_(beer) |
01:15.09 | miknix | dcordes, see Bohemia, its my favorite from sagres |
01:15.33 | miknix | 1835 is a limited edition beer |
01:21.33 | dcordes | http://youtube.com/watch?v=69XjWHh03Yo also james prefers bohemia |
01:23.53 | dcordes | miknix: do you know what the difference between mainline htc-egpio and hh.org htc-egpio is? |
01:26.03 | miknix | dcordes, lol, I remember that commercial |
01:26.25 | miknix | dcordes, hum.. no. but I'll also have to choose |
01:26.46 | dcordes | pH5 told me mainline is the choise. |
01:27.12 | dcordes | and he said that if we wanted to go mainline some day we would have to switch |
01:27.27 | dcordes | currently we have hh.org htc-egpio in linuxtogo htc-msm |
01:28.43 | miknix | dcordes, oh.. sorry. I tough you were asking the difference between egpio in mainline and hh.org branch |
01:29.09 | miknix | dcordes, yes.. mainline is the tree more closer to linus |
01:29.21 | miknix | dont know about htc-egpio |
01:31.49 | BabelO | miknix: htc-egpio is in mainline ;) |
01:32.19 | miknix | hum.. nice |
01:32.32 | dcordes | BabelO: can you help me with the question? |
01:32.32 | miknix | I'll choose that one then |
01:34.42 | BabelO | dcordes: which one ? :) |
01:34.53 | BabelO | dcordes: the difference ? |
01:36.02 | BabelO | dcordes: the difference is that htc-egpio in mainline is more flexible and use the genereic gpio api provided in mainline |
01:38.31 | dcordes | thanks. now I got the point in switching to mainline |
01:47.33 | miknix | BabelO, what about when we need to access address 0x11 on egpio? |
01:47.56 | BabelO | miknix: 0x11 ? |
01:47.57 | miknix | BabelO, we use gpio_write() for both GPIO bits? |
01:48.06 | BabelO | yes |
01:48.11 | BabelO | gpio0 and gpio8 |
01:48.13 | miknix | ah ok.. I'm seeing it |
01:48.38 | miknix | BabelO, and for getting a cpld value? |
01:48.58 | BabelO | miknix: just gpio_get_value() |
01:49.00 | BabelO | same |
01:49.38 | miknix | nice |
01:49.57 | BabelO | miknix: driver is here http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=tree;f=drivers/mfd;h=1ea88544f653135fe5da82d2a4c688ab4ef4194c;hb=2ddcca36c8bcfa251724fe342c8327451988be0d |
01:50.15 | miknix | I'm going to add a "testing" interface on sysfs to the htc-cpld driver |
01:50.28 | miknix | so we can access gpios by userspace |
01:52.42 | BabelO | good night |
01:52.50 | dcordes | nightey BabelO |
01:52.54 | miknix | night |
02:04.29 | dcordes | gnight miknix |
04:46.44 | Pryan | pira pa la kama, adios |
05:12.49 | *** join/#htc-linux dcordes (n=dcordes@unaffiliated/dcordes) |
05:26.15 | *** join/#htc-linux rmoravcik (n=rmoravci@ip-89-102-255-171.karneval.cz) |
05:37.42 | *** join/#htc-linux marbalon (n=marbalon@212.122.223.98) |
06:02.18 | *** join/#htc-linux goxboxlive (n=goxboxli@208.84-48-176.nextgentel.com) |
06:42.44 | *** join/#htc-linux rmoravcik (n=rmoravci@gtsgw.ttc.cz) |
07:16.18 | *** join/#htc-linux exco (n=excogita@e181098175.adsl.alicedsl.de) |
07:22.32 | *** join/#htc-linux rob_w (n=bob@X1246.x.pppool.de) |
08:06.06 | *** join/#htc-linux CVirus (n=GoD@196.205.192.185) |
09:18.01 | *** join/#htc-linux rob_w (n=bob@X1d25.x.pppool.de) |
09:19.42 | *** join/#htc-linux MHazem (n=GoD@196.205.192.185) |
09:46.54 | techsolo | Anybody here who knows if it is already possible to run android on the blueangel? I read some ghost story's about it. |
09:48.56 | *** join/#htc-linux dzo (n=dzo@121.98.128.127) |
10:06.31 | marbalon | SanMehat: hi |
10:09.07 | dzo | hi marbalon, how are things with kaiser now? |
10:09.24 | *** join/#htc-linux diogene31_ (n=rj@mur31-2-82-243-122-54.fbx.proxad.net) |
10:09.26 | marbalon | dzo: hi, yes |
10:09.49 | marbalon | correct me if I'm wrong but few days ago SanMehat said something about communication between Arm11 nad ARM9 in msm7x00 |
10:10.21 | dzo | I don't know, I haven't been watching the logs. |
10:10.45 | dzo | It's very different for them though (different version of AMSS) |
10:11.17 | dzo | Hvae you guys got smd working yet? |
10:12.07 | marbalon | no |
10:12.27 | marbalon | how did you find correct mapping for smd ? |
10:12.32 | dzo | it shouldn't be too hard, I sent dcordes some ideas the other day. |
10:13.10 | dzo | I can trace it on vogue, I don't know why you can't on kaiser. |
10:13.46 | *** join/#htc-linux patoche (n=Patrice@lns-bzn-24-82-64-163-99.adsl.proxad.net) |
10:13.49 | marbalon | trace using haret ? |
10:13.50 | dzo | probably all that is different is the A2M comms address. |
10:13.56 | dzo | yes, using haret. |
10:16.01 | marbalon | dzo: there is any doc like http://wiki.xda-developers.com/index.php?KaiserMemoryMap for vogue ? |
10:16.34 | dzo | It's the same as titan, there is a page for that. |
10:17.43 | dzo | can you trace any address at all with haret? |
10:18.57 | cr2 | hi |
10:19.10 | cr2 | dzo: i guess not. |
10:19.12 | marbalon | I'm not an haret expert |
10:19.17 | dzo | hi cr2 |
10:20.06 | cr2 | dzo: the mmu dump shows some extra bits used in the page table, but only Kevin2 deals with such wizardry :) |
10:20.32 | dzo | well, i bet kaiser is not very different, just try all possible values for MSM_A2M_INT in notify_other_smd |
10:21.19 | dzo | Thats the TEX bits, they are 010 on the vogue. |
10:21.19 | cr2 | dzo: the FIFOs are 8K in size |
10:22.01 | dzo | good, just find the correct interrupts and it will work it reckon. |
10:22.15 | cr2 | dzo: can you provide a direct link to the docs ? |
10:22.23 | dzo | which docs? |
10:22.54 | cr2 | on the mmu bits |
10:23.12 | cr2 | i don't have a kaiser myself |
10:23.26 | univac | which thc phone has the best support under linux? (wuth qwerty keyobard) |
10:23.33 | univac | htc* |
10:23.40 | cr2 | univac: universal |
10:23.54 | univac | ok, thanks. |
10:26.42 | dzo | cr2: http://infocenter.arm.com/help/topic/com.arm.doc.ddi0211j/Babifihd.html |
10:28.14 | cr2 | dzo: thanks |
10:29.11 | dzo | no problem, i'm off now, good night all. |
10:29.45 | marbalon | night dzo |
10:29.55 | univac | uh, those phones are big :> |
10:30.11 | marbalon | cr2: hmm. I have idea. Maybe I can write a driver with read SHMEM A11/A9 if any INT_A9_M2A_* occurs and display what data changed and offsets ? |
10:30.29 | marbalon | cr2: do you think it is good idea ? |
10:30.35 | cr2 | univac: qwerty has its price. i don't find the BA keyboard really usable. |
10:31.36 | cr2 | marbalon: no a driver, but a haret function which you call on the irq instead of just dumping the irq name |
10:31.45 | cr2 | s/no a/not a/ |
10:32.11 | dzo | marbalon: I trap all the M2A ints so that shouldn't be the problem |
10:32.31 | dzo | it will be the A2M writes that are different. |
10:33.08 | cr2 | you can allocate the 1MB buffer to compare things. |
10:35.24 | marbalon | cr2: do you think haret function implemented in haret or haret console ? |
10:36.00 | cr2 | marbalon: in haret. it#s easier, and more efficient. |
10:36.32 | *** join/#htc-linux skodde (n=skodde@unaffiliated/skodde) |
10:37.47 | marbalon | cr2: this CPU is on the HEAD ? or have own branch ? |
10:38.22 | cr2 | in haret ? haret does not have branches |
10:39.17 | marbalon | ok, have a copy on disk |
10:39.37 | marbalon | now need to setup a compiler |
10:46.13 | marbalon | cr2: I don't know how to use haret advanced functions but a need to implement a new function, strange way ;) |
11:44.24 | marbalon | cr2: cant find correct cross compile tarbal :( |
11:44.53 | marbalon | I downloaded cegcc-gcc430.tar.gz but there is no g++ |
12:17.46 | patoche | hi all |
12:17.55 | patoche | mov r1, #0x001f /* blue color */ <--- someone know how i can change color ? red for example ? |
12:18.07 | patoche | its for eol debugging :) |
12:22.58 | patoche | ok found something |
12:23.05 | cr2 | marbalon: i've used cegcc CVS |
12:23.15 | cr2 | patoche: it's the 565 color |
12:23.44 | cr2 | 5bits for R, 6 bits for G and 5bits for B |
12:23.53 | patoche | oh ok, thanks |
12:24.09 | cr2 | 0x1f=00011111 |
12:25.04 | cr2 | red will be 0xf800 |
12:25.51 | patoche | thx a lot |
12:35.22 | *** join/#htc-linux revertive (n=revertiv@83.119.75.151) |
12:48.43 | marbalon | dcordes: hi |
12:54.20 | marbalon | cr2: do you know why dcordes set AT send buff as 0x1018 not 0x2000 ? |
13:40.32 | dcordes | hi |
13:42.18 | dcordes | cr2: you were mentionning irq earlier in connection with kaiser smd. do we need irq changes? |
13:47.12 | *** join/#htc-linux ovod (n=ovod@wn1nat30.beelinegprs.ru) |
14:17.57 | *** join/#htc-linux miknix (n=miknix@81.193.88.108) |
14:36.15 | *** join/#htc-linux BabelO (n=Fabrice@lun34-2-82-238-28-28.fbx.proxad.net) |
14:43.45 | Kevin2 | marbalon: you need to get the cegcc-mingw32ce package. |
14:44.19 | cr2 | dcordes: no. |
14:44.44 | cr2 | Kevin2: have you seen the link by dzo about armv6 mmu bits ? |
14:45.10 | Kevin2 | cr2: No. |
14:45.32 | majorburn | so uh what do you think about those qualcomm cpu/chipsets |
14:46.45 | dcordes | I think they are nice but undocumented |
14:48.03 | dcordes | cr2: I'm doing a compile marathon with smd.c dzo told me to add a prntk that shows head and tail of at write buffer and with that tryout values 0-7 in function notify_other_smd |
14:48.50 | majorburn | hmm, undocumented but it seems to be better supported by linux than the arm/ati combo |
14:49.34 | cr2 | Kevin2: http://infocenter.arm.com/help/topic/com.arm.doc.ddi0211j/Babifihd.html |
14:51.08 | Kevin2 | cr2: Fun. I've had that pdf on my hard drive for 6 months, but never saw that section. |
14:51.08 | Kevin2 | So, this describes the page table bits? |
14:51.09 | cr2 | Kevin2: it's about the TEX bits. in addition to B and C |
14:51.09 | cr2 | and the S bit in the pagetable |
14:51.28 | cr2 | yes |
14:51.46 | cr2 | In addition to the TEX, C and B bits, certain page tables contain the Shared bit, S. This bit determines whether the memory region is shared (1), or non-shared (0). If not present the S bit is assumed to be 0 (non-shared). When the TRE flag is set the effect of the S bit can also be remapped |
14:56.58 | *** join/#htc-linux rob_w|laptop (n=rob@M8456.m.pppool.de) |
14:57.11 | Kevin2 | Anyone on with an arm v6 cpu? |
14:57.25 | cr2 | dcordes: still here ? |
14:57.59 | dcordes | cr2: yep |
14:58.12 | dcordes | need to finish the value tryouts |
14:58.13 | Kevin2 | dcordes: Can you run "dump cp(15)" on your device? |
15:00.36 | dcordes | in an instance |
15:01.28 | *** join/#htc-linux ltxda0 (n=ltxda@unaffiliated/ltxda) |
15:05.35 | cr2 | Kevin2: memory.h also does not support the APX bit ? |
15:06.03 | dcordes | k last value.. cr2 are you familiar with the smd.c? |
15:06.48 | cr2 | dcordes: no. |
15:06.59 | Kevin2 | cr2: APX is new to arm6 |
15:07.37 | cr2 | yes |
15:08.05 | dcordes | cr2: the AT buffer tail value should change each time I write something to the fifo, right? |
15:08.07 | cr2 | the .pdf manual says 'S' bit is deprecated, use APX instead |
15:08.19 | cr2 | dcordes: i think so. |
15:08.40 | cr2 | dcordes: you need to indicate where to start and stop |
15:09.42 | dcordes | cr2: well it doesn't. I tried 7 values for n in "writel(1, MSM_A2M_INT(n+ch));" in function static inline void notify_other_smd(int ch) |
15:10.05 | dcordes | each time the tail value was different, but stayed the same with each write to the device. only the head values changed |
15:10.17 | cr2 | ok. |
15:10.23 | dcordes | what does this mean? |
15:11.03 | cr2 | i don't really know how they operate the FIFO. |
15:11.23 | cr2 | we need a wince trace, and the .dll/wince reference address. |
15:12.26 | dcordes | I hope dzo will know since he suggested to check this. |
15:13.12 | dcordes | Kevin2: doing the dump now |
15:14.07 | Kevin2 | dcordes: Thanks, can you also post the output from "dump mmu"? |
15:14.33 | dcordes | of course |
15:16.22 | dcordes | Kevin2: is mmu memory management unit? |
15:17.20 | Kevin2 | dcordes: yes |
15:18.26 | dcordes | http://rafb.net/p/N9Aw5g91.html |
15:21.08 | dcordes | Kevin2: the mmu is too big for pastebin can I email it? |
15:21.51 | Kevin2 | dcordes: sure |
15:22.18 | dcordes | address? |
15:27.39 | *** join/#htc-linux CVirus (n=GoD@196.205.192.185) |
15:29.49 | Kevin2 | dcordes: Did you get the message I sent to you directly? |
15:30.11 | dcordes | Kevin2: query? no |
15:30.13 | *** join/#htc-linux ltxda (n=ltxda@unaffiliated/ltxda) |
15:30.32 | Kevin2 | Hrmm. Can you just split the "dump mmu" into two post? |
15:31.36 | *** join/#htc-linux ASUS-tek (i=microsta@221.128.205.158) |
15:32.26 | dcordes | Kevin2: here you go but you have to count the kittens http://rapidshare.com/files/114385705/haretlog-20080512_171454.log.html |
15:33.03 | *** part/#htc-linux ASUS-tek (i=microsta@221.128.205.158) |
15:33.40 | Kevin2 | dcordes: Thanks. |
15:34.00 | Kevin2 | dcordes: BTW, now that you have access to linuxtogo.org, you can ssh in, and make a public_html/ directory and put files there. |
15:34.12 | dcordes | oh cool |
15:34.21 | *** join/#htc-linux ltxda0 (n=ltxda@unaffiliated/ltxda) |
15:35.49 | dcordes | Kevin2: inside the directory I get after logging in? |
15:36.40 | Kevin2 | dcordes: Yeah. Run 'ls -l ~kevin/public_html' and compare it to http://linuxtogo.org/~kevin/ |
15:37.48 | dcordes | now that's easy |
15:38.54 | dcordes | Kevin2: will http://infocenter.arm.com/help/topic/com.arm.doc.ddi0211j/Babifihd.html help making mmutrace work on kaiser? |
15:46.41 | Kevin2 | dcordes: I thought mmu trace did work. |
15:46.58 | dcordes | Kevin2: no we have problems tracing the smem |
15:47.09 | Kevin2 | But you can trace some things? |
15:47.34 | dcordes | Kevin2: I'm not sure but I think so |
15:47.36 | Kevin2 | In any case, yes the documentation should help clarify many things. |
15:48.25 | dcordes | I only know I need mmutrace to find out the right values for AT and SMD1 fifo registers and cr2 said you could possibly help |
15:53.27 | Kevin2 | Bleh - the arm developers should be taken outside and shot. |
15:55.22 | cr2 | Kevin2: the 'shareable' areas can't be 'mmutraced' on kaiser for some reason. |
15:56.14 | Kevin2 | dcordes: Can you run "show cp(15, 0, 2, 0, 1)" and "show cp(15, 0, 2, 0, 2)" |
15:57.14 | Kevin2 | cr2: The mmu is all over the place... If the wind is blowing west and cp(15,2) is set, then use translation table 1, except if bit 4 on page table 7 is set, except if.... |
15:59.22 | cr2 | hehe. |
16:00.41 | majorburn | You just got armed. |
16:00.44 | majorburn | burn |
16:01.18 | miknix | hello Kevin2 |
16:01.24 | Kevin2 | Hi |
16:01.45 | miknix | Kevin2, I was able to get the dpad working over i2c |
16:01.46 | *** join/#htc-linux patoche_ (n=Patrice@lns-bzn-58-82-251-228-228.adsl.proxad.net) |
16:02.04 | Kevin2 | Cool! |
16:02.13 | miknix | Kevin2, but since your egpio driver is going mainstream, I'm switching to it |
16:02.54 | Kevin2 | miknix: The egpio and bbkeys drivers are different. |
16:03.26 | miknix | Kevin2, I know.. I'm only changing the cpld access interface in my driver |
16:03.35 | miknix | Kevin2, instead of using i2c, to use egpio |
16:04.09 | miknix | Kevin2, this is only the dpad driver which gets data from the htc-i2c-cpld driver: http://pastebin.com/d7c85c8aa |
16:13.20 | miknix | Kevin2, what do you think? I'm trying to create a generic htc-cpld driver which will work with everyone |
16:14.24 | Kevin2 | miknix: I don't really understand. I'm not familiar with the hardware you're trying to support. |
16:15.05 | miknix | Kevin2, I noticed that a lot of HTCs have the same cpld for controlling the leds, backlight, getting dpad buttons ... |
16:15.32 | Kevin2 | miknix: Sure, but that is a separate driver from egpio. |
16:15.59 | miknix | Kevin2, yes.. it is. I'm using egpio to access the cpld |
16:16.09 | miknix | instead of i2c |
16:16.43 | Kevin2 | How do you use egpio to access the cpld? |
16:16.55 | Kevin2 | Are you bit-banging on the egpio gpios? |
16:17.01 | miknix | Kevin2, yes |
16:17.26 | miknix | hum.. wait |
16:17.31 | miknix | now I'm confused |
16:18.07 | Kevin2 | On Apache and Hermes, the bbkeys stuff is connected to the main processor gpios. |
16:18.53 | miknix | Kevin2, on the wizard and herald and some others, the cpld is bitbanging on two gpio pins |
16:19.20 | miknix | Kevin2, or we can also use those pins to drive a i2c bus |
16:20.00 | miknix | Kevin2, I think that was you that told me this.. to use the second function of the i2c pins |
16:20.10 | Kevin2 | Hrmm. Using "cpld" is kind of confusing - the phones have several cpld chips doing different functions. |
16:20.30 | Kevin2 | Are you refferring the egpio chip (which uses cpld) or the front keypad chip (which may also be using a cpld). |
16:20.49 | miknix | Kevin2, yes.. I'm referring to them all |
16:21.11 | miknix | Kevin2, each cpld chip behaves very similar between htc devices |
16:21.45 | miknix | so why to implement that in each board if the majority of code works on all devices? |
16:22.43 | Kevin2 | I don't think they are that similar. The egpio chips are definitely similar to each other. |
16:23.06 | Kevin2 | The front keypad chips are not similar to egpio though. They have timers for led expiration and keypad debouncing stuff. |
16:25.02 | miknix | Kevin2, hum.. but I cannot access the front pad chips using egpio? |
16:26.00 | Kevin2 | miknix: I'm confused by the question. egpio uses a memory mapped set of 16bit registers. front keypad uses a 2 gpio bit banging mechanism. |
16:26.16 | Kevin2 | You definitely can't drive the front keypad using the egpio driver. |
16:26.29 | miknix | oops |
16:27.01 | miknix | I thought that the egpio was using the two bitganging pins to extended gpios |
16:28.02 | miknix | Kevin2, putting that in this way, what I was saying is ridiculous |
16:28.32 | miknix | s/extended/extend/ |
16:29.03 | Kevin2 | miknix: The egpio chip found on hermes, apache, and most other htc phones is driven by the memory controller. I've heard some people have found an egpio like device driven by i2c on some of the omap devices. But that chips is definitely distinct from the ones found in most htc phones. |
16:29.37 | miknix | Kevin2, hum.. ok then. so I guess I'll continue to use i2c |
16:29.58 | Kevin2 | The driver ph5 is pushing to the mainline kernel only supports the memory interface. |
16:31.07 | Kevin2 | miknix: Okay. So, what does the code look like when using i2c? I'm curious if it is similar to the bbkeys stuff. |
16:31.29 | miknix | Kevin2, its similar, but uses some abstraction provided by the i2c bus itself |
16:31.38 | miknix | Kevin2, it works with any i2c bus available |
16:31.52 | Kevin2 | It's been an open question for me if the "bbkeys chip" is implemented via cpld or is part of the lcd controller, or something else. |
16:32.56 | miknix | Kevin2, do you want to fetch from linwizard repo or prefer the files on pastebin? |
16:33.04 | Kevin2 | miknix: Is the i2c bus a bitbanging thing, or is it using an i2c control block on the main processor? |
16:33.18 | Kevin2 | miknix: Either is fine. |
16:33.36 | miknix | Kevin2, you can use both busses. it works with omap i2c bus or gpio bitbang i2c |
16:33.53 | miknix | Kevin2, omap i2c has some bugs but uses interrupts which is fine |
16:34.04 | miknix | Kevin2, gpio bitbang does polling |
16:34.15 | Kevin2 | miknix: On your particular phone, can you use the omap i2c bus? |
16:34.29 | miknix | Kevin2, yes |
16:34.50 | miknix | Kevin2, I had to modify the address shift (Babel0 fix) |
16:35.44 | *** join/#htc-linux noelcragg (n=noel@adsl-68-121-164-151.dsl.pltn13.pacbell.net) |
16:36.15 | *** part/#htc-linux noelcragg (n=noel@adsl-68-121-164-151.dsl.pltn13.pacbell.net) |
16:36.21 | miknix | Kevin2, git://polaris.lurian.net/linwizard-kernel.git |
16:36.30 | miknix | Kevin2, fetch the I2C branch |
16:36.53 | Kevin2 | miknix: Is git-web running on that server somewhere? |
16:37.00 | miknix | Kevin2, no. sorry |
16:37.25 | Kevin2 | miknix: Hrmm. Can you just pastebin the driver then? :-) |
16:37.36 | miknix | Kevin2, ok then |
16:38.55 | miknix | Kevin2, Main driver, it attaches as chip to the i2c bus: http://pastebin.com/d1c62ae7b |
16:39.44 | miknix | Kevin2, Main driver header: http://pastebin.com/d17ce6af6 |
16:40.23 | miknix | Kevin2, LEDs driver: http://pastebin.com/d3b741ec6 |
16:41.10 | miknix | Kevin2, Backlight driver: http://pastebin.com/d66abcabf |
16:41.52 | miknix | Kevin2, and the dpad buttons driver: http://pastebin.com/d3cfa8c51 |
16:52.31 | dcordes | Kevin2: |
16:52.36 | dcordes | HaRET(1)# show cp(15, 0, 2, 0, 1) |
16:52.37 | dcordes | 0x00000000 |
16:52.57 | dcordes | HaRET(2)# show cp(15, 0, 2, 0, 2) |
16:52.57 | dcordes | 0x00000000 |
16:59.21 | Kevin2 | miknix: It looks like your driver is very similar to bbkeys. Why can't you use bbkeys? |
16:59.26 | Kevin2 | dcordes: Thanks. |
17:00.03 | miknix | Kevin2, didnt try. I just wanted to get some hw abstraction, which the i2c bus can give me |
17:01.06 | Kevin2 | miknix: The bbkeys driver (if it works) should take care of everything. |
17:01.32 | miknix | Kevin2, yeah.. it is easy to replace the i2c instructions with the bbkeys hw access ones |
17:02.22 | Kevin2 | miknix: Not sure I understand. bbkeys provides a keyboard driver , and it turns the rest of the chip into gpio accesses. |
17:04.05 | miknix | Kevin2, how do you have your backlight driver working? |
17:04.25 | Kevin2 | miknix: By using the gpios.. |
17:05.00 | miknix | Kevin2, so you have a backlight driver using gpios somewhere.. in the board file? |
17:06.00 | Kevin2 | miknix: See http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-pxa/htcapache/htcapache.c;h=d44a5a34643d9caa8054243cb8bb7c106ccb29d8;hb=d6da66b97b9a4757e9c7b16fb461eb654a6fdf63#l157 |
17:07.35 | Kevin2 | Right above that (in the same file) is the definition for the rest of the bbkeys driver on Apache. |
17:07.37 | Kevin2 | http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-pxa/htcapache/htcapache.c;h=d44a5a34643d9caa8054243cb8bb7c106ccb29d8;hb=d6da66b97b9a4757e9c7b16fb461eb654a6fdf63#l102 |
17:08.10 | miknix | hum.. |
17:08.50 | miknix | Kevin2, but this cpld is similar for a lot of htc devices |
17:08.56 | Kevin2 | Later in the same file is where leds are defined - note all of Apache leds (including those not on bbkeys) are in the same section: |
17:08.59 | Kevin2 | http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-pxa/htcapache/htcapache.c;h=d44a5a34643d9caa8054243cb8bb7c106ccb29d8;hb=d6da66b97b9a4757e9c7b16fb461eb654a6fdf63#l332 |
17:09.02 | miknix | IMHO it would make sense a driver for all |
17:09.40 | Kevin2 | That wont work - the keys and registers are different from one phone to the next. |
17:09.56 | Kevin2 | The only thing that is probably the same is the lcd stuff - that could probably be abstracted out into the main driver. |
17:12.05 | BabelO | cr2: gps works :) |
17:13.57 | miknix | Kevin2, at least all the cpld registers are equal in htc wizard and htc herald and htc elf |
17:15.12 | Kevin2 | miknix: Compare to htc hermes at: http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-s3c2442/htchermes/htchermes.c;h=f8e21716b4c99976453a47dd683fd4e3684e3db1;hb=d6da66b97b9a4757e9c7b16fb461eb654a6fdf63#l106 |
17:18.36 | miknix | Kevin2, "hermesRight:alter", .gpio = BBKEYS_GPIO(2, 7) |
17:18.56 | miknix | chip 2 address 0x80? |
17:20.31 | *** join/#htc-linux JEEB (n=kanakana@a88-112-94-153.elisa-laajakaista.fi) |
17:20.46 | Kevin2 | miknix: Yes. On my phones, if you set a bit in causes the led to flash. |
17:20.52 | Kevin2 | (The "alter" bit.) |
17:21.16 | miknix | Kevin2, we also have an alter bit. but it only works sometimes |
17:22.04 | miknix | Kevin2, the chip behaviour changes if we booted linux with usb connected or wifi on |
17:22.27 | Kevin2 | Yeah - you can cause green to flash, blue to flash, but if you try to flash "light blue" it will cause blue / green to flash alternating. |
17:23.05 | miknix | Kevin2, err.. I'm talking to the usb hardware |
17:23.26 | miknix | Kevin2, if we boot linux and usb was previously connected on wm5 |
17:23.34 | Kevin2 | miknix: I haven't seen anything like that on my phones. |
17:24.04 | miknix | Kevin2, also in chip1 seems to be a bit to enable/disable the mmc power |
17:25.45 | Kevin2 | miknix: How do you "ack" your keypad presses? |
17:34.26 | *** join/#htc-linux TimRiker (n=timr@68-27-117-166.area1.spcsdns.net) |
17:37.31 | dcordes | Kevin2: further action required? |
17:38.18 | miknix | Kevin2, no ack needed |
17:38.23 | Kevin2 | dcordes: I'm still reading the docs - it's a real nightmare.. |
17:38.34 | Kevin2 | miknix: Do you get an irq when the keypad is hit? |
17:38.39 | miknix | Kevin2, yes |
17:38.59 | miknix | Kevin2, the dpad irq handler reads the i2c bus |
17:39.05 | Kevin2 | So, you don't need to ack the irq in order to get another irq? |
17:39.19 | Kevin2 | On Apache and hermes, we have to write a 0 to register 3 in order to get further irqs. |
17:39.29 | miknix | Kevin2, I return IRQ_HANDLED |
17:39.36 | miknix | Kevin2, no.. it works fine like this |
17:39.51 | *** join/#htc-linux exco (n=excogita@e181098175.adsl.alicedsl.de) |
17:40.00 | Kevin2 | miknix: Okay - so, the chips are a little different then. |
17:40.14 | miknix | Kevin2, yeah.. the LED bits are also different |
17:40.45 | miknix | Kevin2, but which interrupts are you handling, the i2c ones? |
17:41.07 | miknix | Kevin2, because I got 2 interrupts when I press the dpad, the dpad interrupt and the i2c one |
17:41.08 | Kevin2 | No, there is no i2c interrupt. |
17:41.53 | Kevin2 | On Apache and Hermes, there is only gpios. There is a clock gpio, a data gpio, and an irq gpio. |
17:42.15 | Kevin2 | Clock is always out, irq is always in, and data can go both ways. |
17:42.23 | miknix | Kevin2, same here |
17:47.50 | *** join/#htc-linux dortega (n=David@238.Red-88-16-218.dynamicIP.rima-tde.net) |
17:49.55 | Kevin2 | miknix: What did you mean earlier by "bitbanging driver uses polling"? |
17:50.58 | miknix | Kevin2, I think it scans the i2c adapter in a timed fashion |
17:51.21 | *** join/#htc-linux rmoravcik (n=rmoravci@ip-89-102-255-171.karneval.cz) |
17:51.32 | Kevin2 | miknix: I'm not sure what you mean. |
17:52.14 | miknix | Kevin2, sorry.. I dont have electronics background.. Its hard for me to call things with the right name |
17:52.35 | Kevin2 | miknix: Heh, I'm in the same boat. |
17:52.49 | miknix | Kevin2, :) |
17:54.44 | miknix | Kevin2, unlike omap bus, the bb gpio i2c bus doesnt handle interrupts. it just reads the i2c device in a polled interval |
17:55.48 | Kevin2 | Okay - that's different for Apache and Hermes. On these phones, we don't drive the i2c stuff at all until we want to change something or we get an irq. Then we read/write what we want. |
17:56.43 | miknix | Kevin2, yeah.. the omap i2c bus also works like that |
17:58.06 | miknix | but the omap i2c bus is conflicting interrupts with the dpad driver |
17:58.32 | miknix | didnt look much at it because the bb gpio bus works |
17:58.58 | miknix | but I would like to have the omap i2c bus working too with the dpad to minimize interrupts |
17:59.34 | *** join/#htc-linux JohnnyK (n=johnnyk@nfx-nat-229.pilsfree.net) |
18:04.47 | Kevin2 | miknix: I think you should see if you can get bbkeys working - that only raises an irq when one presses a key. |
18:05.15 | miknix | Kevin2, can you separate the dpad driver from it? |
18:05.36 | Kevin2 | cr2: It looks like dcordes device only sets "device non-shared" in the page tables. I don't see any "shared" mappings. |
18:05.50 | Kevin2 | miknix: Why not use the whole thing? |
18:06.32 | miknix | Kevin2, we dont have the same buttons/keymaps |
18:07.11 | BabelO | <PROTECTED> |
18:07.40 | Kevin2 | miknix: See the links I had above. They keyboard map is defined in the arch code - we don't need to have the same keys. |
18:08.09 | Kevin2 | http://git.linuxtogo.org/?p=groups/mobile-linux/kernel.git;a=blob;f=arch/arm/mach-s3c2442/htchermes/htchermes.c;h=f8e21716b4c99976453a47dd683fd4e3684e3db1;hb=d6da66b97b9a4757e9c7b16fb461eb654a6fdf63#l121 |
18:08.26 | miknix | Kevin2, hum.. ok. and the other bits are exported as gpios right? |
18:08.29 | Kevin2 | Note how hermes defines what registers are read on key event, and how those registers map to keys. |
18:08.42 | Kevin2 | miknix: yes. |
18:08.59 | miknix | seems fair to me |
18:15.21 | Kevin2 | cr2: Scratch that - I do see shared mappings. |
18:30.21 | *** join/#htc-linux exc1 (n=excogita@e181065251.adsl.alicedsl.de) |
18:38.49 | BabelO | Kevin2: still around ? :) |
18:47.17 | Kevin2 | BabelO: yes. |
18:47.40 | BabelO | Kevin2: i want to push my local htc-omap branch |
18:47.57 | BabelO | Kevin2: how can i proceed ? |
18:48.32 | BabelO | Kevin2: git push origin htc-omap:refs/heads/babelo |
18:48.34 | BabelO | ? |
18:48.37 | Kevin2 | BabelO: You have a branch in your local git, and want to push it to linuxtogo? |
18:48.48 | *** join/#htc-linux pikapika (n=pikapika@mar75-8-88-164-227-147.fbx.proxad.net) |
18:48.59 | pikapika | hi |
18:49.00 | BabelO | Kevin2: yes a clean branch based on mainline |
18:49.19 | BabelO | Kevin2: hi use git branch --track htc-omap origin/mainlinel |
18:49.22 | BabelO | to create it |
18:49.44 | cr2 | BabelO: hi |
18:49.54 | cr2 | BabelO: htc-egpio works on arte ? |
18:50.27 | BabelO | cr2: yes now it is ok, gps seem to works, i use cat /dev/ttyS1 and i see strange char incomming at same interval |
18:50.33 | BabelO | i think it is just a bad baudrate |
18:50.48 | BabelO | cr2: but bt need more work to init |
18:51.28 | Kevin2 | BabelO: I'm not really sure. I think you just push the branch you want to. |
18:52.15 | BabelO | Kevin2: ok then i prefer to wait pH5 for the right command :) this one looks good to me git push origin htc-omap:refs/heads/babelo |
18:53.31 | diogene31_ | BabelO: It looks good to me : git push <remote server> <localbranch>:<server branch> |
18:53.44 | cr2 | BabelO: check 4800 and 9600 |
18:53.56 | BabelO | diogene31_: the the remote rbranch should be git push origin htc-omap:refs/heads/htc-omap |
18:54.01 | BabelO | cr2: ok |
18:54.19 | cr2 | BabelO: if it will not help, then you can always run sirfmon. |
18:54.22 | BabelO | cr2: btw, my screen is not centered , i think i set wrong parameters |
18:54.42 | cr2 | are the lcd registers documented ? |
18:55.09 | BabelO | cr2: i ve some part yes |
18:55.28 | *** join/#htc-linux kiozen (n=kiozen@rgnb-4db1d8a4.pool.einsundeins.de) |
18:55.45 | kiozen | hi |
18:56.22 | BabelO | hi kiozen |
18:57.02 | kiozen | BabelO: got your two beers for decent programming? |
18:58.10 | BabelO | kiozen: yes when i come back from the "concert" |
18:58.25 | kiozen | oh another one today? |
18:59.03 | BabelO | kiozen: no the one from yesturday |
18:59.06 | cr2 | hi kiozen |
18:59.33 | kiozen | hi cr2, did you blow up the display already ;) |
18:59.36 | cr2 | kiozen: the resume does not work as expected, but the LCD is not toast. yet :) |
19:00.00 | kiozen | guess we have to call this a step forward :)) |
19:00.13 | cr2 | i'm checking the spiread_24bit() function. |
19:00.17 | diogene31_ | BabelO: I dont't understand. The remote branch is htc-omap in your case (refs/heads/htc-omap => remote branch htc-omap) |
19:00.40 | cr2 | kiozen: i mean the lcd resume/powerup, not the machine. |
19:00.56 | BabelO | diogene31_: yes i want to push my local branch named htc-omap to the remote server and name it htc-omap too |
19:01.00 | kiozen | cr2: I know |
19:01.18 | diogene31_ | BabelO: So that's the right command. |
19:01.40 | kiozen | never thought that there is so much business to do for a LCD during suspend/resume. |
19:01.56 | kiozen | thought you just start to fill the framebuffer again after resume |
19:03.03 | BabelO | diogene31_: ok,thx, but i think i need to do something on my side, for push permission |
19:06.47 | diogene31_ | BabelO: Probably. That depends on the git url you're using, and the git-server setup. |
19:07.23 | cr2 | kiozen: on ba6 is still does not work after 3 years ;) |
19:07.29 | BabelO | diogene31_: you are running your phone on 2.6.26 too ? |
19:07.47 | diogene31_ | BabelO: Yes I do. |
19:08.12 | BabelO | diogene31_: same here, i ll try later blueangel too |
19:08.27 | cr2 | BabelO: ba needs asic3_mmc |
19:08.52 | BabelO | cr2: yes i know, i see that there is already a ASIC3 driver in 2.6.26 |
19:08.53 | cr2 | BabelO: you may merge the n560 instead :) |
19:09.12 | cr2 | BabelO: link ? |
19:09.19 | BabelO | cr2: not the mmc part :) |
19:09.35 | kiozen | cr2: I always thought LCDs are pretty straight forward simple devices with just a framebuffer and a register for the backlight. I am just amazed that they need such a complicated setup during resume. |
19:10.06 | cr2 | BabelO: without mmc it's useless. |
19:10.23 | BabelO | cr2: yes :( |
19:11.02 | *** join/#htc-linux ltxda (n=ltxda@c-98-196-125-205.hsd1.tx.comcast.net) |
19:13.37 | cr2 | kiozen: it looks like that http://svnweb.openmoko.org/trunk/src/host/qemu-neo1973/hw/jbt6k74.c?rev=3443&view=markup |
19:24.13 | *** join/#htc-linux Napfton (n=Napfton@130.83.47.12) |
19:30.49 | Napfton | http://wiki.xda-developers.com/index.php?pagename=Ipaq6915 |
19:31.36 | Napfton | does this mean its close to be installable but not close enough to give any installation instructions for riskloving ppl? |
19:41.32 | cr2 | Napfton: the 240x240 will give some trouble with the gui elements. |
19:41.52 | cr2 | Napfton: you'd ask Marex about the phone status. |
19:42.20 | cr2 | the rest should work. (without the internal mdoc g4) |
19:45.48 | exc1 | hello |
20:25.24 | *** join/#htc-linux marbalon (n=balon@aegz72.neoplus.adsl.tpnet.pl) |
20:34.28 | *** join/#htc-linux patoche (n=Patrice@lns-bzn-58-82-251-228-228.adsl.proxad.net) |
20:40.35 | *** join/#htc-linux dcordes (n=dcordes@unaffiliated/dcordes) |
20:44.11 | marbalon | dcordes: hi |
20:44.15 | dcordes | hi marbalon |
20:44.47 | dcordes | you asked me about the fifo registers earlier. to be honest - I have no idea. see the comment above the code. dzo gave me that |
20:45.50 | marbalon | ok, I think it is a problem, I dump smem an have idea how to fix it, but how you test communication ? ppp ? |
20:46.05 | dcordes | marbalon: no it's easy |
20:47.03 | dcordes | cat /dev/smd0 & echo -e "ATV1\r" > /dev/smd0 |
20:47.11 | dcordes | the modem should reply something |
20:48.17 | dcordes | but before you can check if the data flow works at all |
20:48.26 | cr2 | dcordes: cu(1) |
20:49.16 | exco | can I reboot from OM without hardresetting to WinCE |
20:49.49 | dcordes | cr2: (1)? |
20:49.57 | majorburn | man page section 1 |
20:50.04 | dcordes | oic |
20:50.52 | dcordes | marbalon: in smd_write after the first check_for_smd_data, put "printk("head=%x, tail=%x\n",ch->send.buffer->head,ch->send.buffer->tail);" and see if tail changes when you write to /dev/smd0 a few times. |
20:51.03 | dcordes | (c) dzo |
20:51.42 | dcordes | If it doesn't, change the 3 in notify_other_smd with numbers from 0..7 until |
20:51.42 | dcordes | it does. |
20:51.58 | dcordes | but I tried all 8 values turns out none of them made tail change on writing |
20:52.04 | marbalon | ok I will try |
20:52.31 | marbalon | dcordes: I've push my last keyboard changes, please test it |
20:52.34 | dcordes | with 0 the tail value is 77656361, with 4 6a666a66, with 5 646a6a64, with 6 616c616c, with 7 660a7261 |
20:52.46 | dcordes | with 3,2,1 6c6f6c0a |
20:52.50 | dcordes | head changes but tail doesn't |
20:53.37 | dcordes | marbalon: ok will test later, brb |
20:57.40 | majorburn | # |
20:57.46 | majorburn | Chuck Norris can divide by zero |
20:58.32 | exco | :-) |
20:58.50 | *** join/#htc-linux LunohoD_ (n=alex@e180073211.adsl.alicedsl.de) |
20:58.58 | BabelO | majorburn: steven segal too :) |
21:03.59 | miknix | majorburn, and I heard that when Chuck Norris does division, there is no remainder |
21:07.22 | majorburn | nice one |
21:09.14 | exco | Chuck Norris counted to infinity - twice. |
21:10.01 | majorburn | <3 #htc-linux |
21:10.45 | miknix | Chuck Norris doesn't have blood. He is filled with magma. |
21:12.31 | exco | http://www.chucknorrisfacts.com/ |
21:12.50 | miknix | fortune -m "Chuck" |
21:15.59 | dcordes | ~Chuck Norris |
21:16.00 | apt | chuck norris is, like, and there is no chin under Chuck Norris' Beard. There is only another fist. |
21:16.08 | miknix | ahha |
21:16.14 | miknix | dcordes won! |
21:18.47 | dcordes | marbalon: building your keyboard changes. tried any smd stuff yet? |
21:23.57 | marbalon | dcordes: hmm when I run cat /dev/smd0 & a get a lot of buffered command |
21:25.03 | dcordes | marbalon: can you write some of them down? |
21:25.18 | dcordes | (I get only the printk, no commands) |
21:25.37 | marbalon | +CRES.... |
21:25.46 | marbalon | @HTCCSQ.... |
21:25.48 | dcordes | did you change anything in the code? |
21:26.00 | marbalon | yes |
21:26.03 | dcordes | cool |
21:26.05 | dcordes | what's your image? |
21:26.09 | marbalon | fix addresses |
21:26.28 | dcordes | that's great looks like we are one stead away from data connection |
21:26.41 | marbalon | what image ? initrd ? |
21:26.43 | dcordes | yew |
21:26.45 | cr2 | dcordes: these are "unsolicited responses" |
21:26.45 | dcordes | yep |
21:26.58 | marbalon | angstrom |
21:27.05 | cr2 | they are regularily sent by the phone. |
21:27.07 | dcordes | how do you make the device node? |
21:27.24 | cr2 | which one ? |
21:27.31 | dcordes | /dev/smd0 |
21:27.48 | dcordes | I make it with "mknod /dev/smd0 c 254 1" |
21:27.58 | marbalon | there is a smd 0 smd 1 snd smd127 |
21:28.22 | dcordes | odd. I get none of them. Can you show me your CMDLINE and which image do you use exactly? |
21:28.59 | marbalon | standard without ppp and password |
21:29.39 | cr2 | hmm. |
21:29.40 | dcordes | standard= you do not put anything in config, n'either in default.txt? |
21:30.05 | cr2 | 254 1 is used by sd/mmc p1 ? |
21:30.28 | marbalon | yes, without anything in config |
21:30.58 | dcordes | cr2: in proc there is a file that told me that major and minor for /dev/smd |
21:31.30 | marbalon | I need to go a sleep. i put my file temporary on my server |
21:31.30 | dcordes | well major 254 and minor 0-31 I think |
21:32.05 | dcordes | ok thanks |
21:32.30 | marbalon | http://baliniak.pl/android/kaiser-smd.c |
21:32.48 | marbalon | it is a clean copy 7x00-smd.c |
21:33.09 | marbalon | I only modify addresses |
21:33.13 | marbalon | night |
21:33.39 | dcordes | ok good night |
21:34.01 | dcordes | I wonder what will happen when I put those addresses in smd-kaiser.c which has some updates from dzo |
21:35.02 | exco | paulproteus, ping |
21:35.14 | paulproteus | exco, pong |
21:36.35 | paulproteus | Unfortunately I've not much to tell you, I've been very busy the past few days. |
21:36.49 | exco | I tried the kernel you built with the older image and it also doesn't work |
21:36.56 | paulproteus | How exciting. |
21:37.00 | exco | :-Ã |
21:37.05 | paulproteus | What sort of doesn't work, again? |
21:37.26 | exco | well should the new kernel still work with older images? |
21:37.42 | paulproteus | Assuming you put the modules directory in the right place, sure, generally speaking. |
21:37.50 | paulproteus | Same with vice versa. |
21:38.19 | exco | so we might not only have a "broken" new image but also a "broken" new kernel |
21:38.34 | paulproteus | Amazing. What was the brokenness again? I forgot. |
21:41.17 | exco | can't tell |
21:42.08 | paulproteus | What does it not do properly, I mean? |
21:42.43 | exco | It won't boot (plus I don't know if I'm the cause :-) ) |
21:43.34 | paulproteus | Hmm, I should try later too then. |
21:45.17 | exco | I can successfully change the kernel from hh17 to hh18 and hh20 with the image from ~htcpxa (Dec 07) but not to our built hh20 kernel |
21:45.41 | paulproteus | What about the image from us and the old hh20 from ~htcpxa? |
21:45.56 | exco | haven't gotten that to work either - so far |
21:47.44 | paulproteus | What error do you get when you try that? |
21:49.00 | exco | I will tell you soon |
21:49.06 | exco | (didn't document) |
21:49.10 | paulproteus | nods |
21:51.17 | dcordes | lol I just noticed I can hear the ocean in kaiser speaker after linux booted |
21:52.13 | paulproteus | Ouch. |
21:52.20 | exco | yeah |
21:52.35 | exco | just didn't replace the right zimage in /boot... |
21:52.49 | exco | ok, so the dec07 image boots with the new kernel |
21:53.10 | paulproteus | That's a relief. |
21:53.39 | BabelO | hmm no pH5 |
21:54.54 | BabelO | paulproteus: do you know what i need to setup to comit into git ? |
21:55.02 | paulproteus | For the linuxtogo git? |
21:55.14 | BabelO | paulproteus: yes |
21:55.24 | BabelO | paulproteus: i can already comit into qtopia :) |
21:55.30 | paulproteus | First of all, in git terminology, "commit" is a local action; "push" is an action that requires remote privileges. |
21:55.32 | BabelO | paulproteus: waiting for 4.4 to continue |
21:55.35 | paulproteus | nods |
21:55.42 | paulproteus | goxboxlive says he also waits for 4.4 |
21:55.43 | BabelO | paulproteus: yes push :) |
21:56.34 | paulproteus | I guess you need to give ph5 an SSH key. |
21:56.42 | paulproteus | And he has to put it somewhere. |
21:57.04 | BabelO | paulproteus: ah :( |
21:57.05 | dcordes | is there a thing in linux that let's me auto assign all the /dev files again? |
21:57.10 | dcordes | like it should be done at boot time? |
21:57.26 | paulproteus | dcordes, Well, these days udev is supposed to make them I think. |
21:57.56 | paulproteus | The other thing you can do, BabelO, is push your work and email ph5 and tell him to pull from you. That's not great, obviously. |
21:58.34 | BabelO | dcordes: what do you do to setup linuxtogo ? |
21:58.50 | cr2 | BabelO: can you have a look at the n560 and athena files and put them into git ? as a htc-egpio wizard :) |
21:58.59 | exco | is the order of modules lsmod gives important? |
21:59.07 | miknix | dcordes, mdev -a |
21:59.14 | miknix | dcordes, its from busybox |
21:59.40 | BabelO | cr2: ok, i look |
22:00.09 | dcordes | miknix: I have busybox but angstrom console image does not have mdev |
22:00.57 | miknix | dcordes, busybox is highly configurable at build time.. it might not be enabled |
22:01.28 | cr2 | BabelO: thanks. i'm fighting evil spi ;) |
22:01.56 | miknix | dcordes, http://linwizard.wiki.sourceforge.net/booting |
22:02.29 | miknix | dcordes, see the nfs boot.. it has automatic /dev population at boot |
22:02.39 | dcordes | cr2: I don't get the smd nodes :( |
22:02.59 | dcordes | nfs? I have no network |
22:03.50 | dcordes | /proc/devices says 254 smd |
22:04.18 | cr2 | it's in the kernel |
22:04.45 | cr2 | <PROTECTED> |
22:05.11 | dcordes | yes I know it is something with the image. I just wonder how to get the node right to reconstruct marbalon's findings |
22:05.41 | dcordes | cr2: already rebuilt with his changes to smd.c |
22:05.50 | miknix | dcordes, cat /sys/pathtoyourdevice/dev usually show you the device numbers |
22:05.52 | cr2 | dcordes: i've used 'ln -sf /dev/ttyS2 /dev/gps' in qlandkarteM startup script |
22:06.00 | miknix | s/device/node/ |
22:08.45 | dcordes | cr2: miknix: /sys/class/tty has smd0 smd1 smd27 |
22:09.33 | dcordes | does this mean the smd driver wants to use those? |
22:09.59 | exco | paulproteus: it boots |
22:10.27 | paulproteus | exco, woots! |
22:11.08 | exco | I removed the zimage symbolic link from /root - now it boots |
22:11.12 | *** join/#htc-linux dzo (n=dzo@121.98.128.127) |
22:11.59 | exco | to the calibration screen ... but I can't calibrate - it doesn't recognize me pressing the touchscreen |
22:12.41 | exco | I mean /boot - of course |
22:13.05 | dzo | dcordes: I just sent you some mail about smd, it should help. |
22:14.22 | dcordes | dzo: ok thx. did you see the recent log? marbalon said he was able to read data |
22:14.52 | dzo | no, i'll look at it. |
22:15.17 | dcordes | since he based his change on the bare msm7x00-smd.c which do not include your change I think I'm not sure which to take |
22:20.05 | dzo | his msm7x00 looks like an older version of my vogue-smd.c, he has the correct start and buffer values for smd0 though. |
22:20.30 | dcordes | dzo: I see. I will just put your values in the revision from the diff you sent me |
22:20.42 | dzo | and is using 3 for a2M so perhaps that is correct. |
22:21.14 | dzo | yes, try that and keep the 3 in notify-other-smd |
22:21.16 | dcordes | how is it his values look 'rounder' = |
22:21.25 | dcordes | ? |
22:22.10 | dzo | his smd1 values are wrong. |
22:22.16 | paulproteus | exco, Wait, the current OM image doesn't recognize your touching the touch screen at all? |
22:22.23 | dcordes | dzo: writel(1, MSM_A2M_INT(3+ch)); right? |
22:22.28 | dzo | yes |
22:22.52 | dcordes | k building |
22:23.24 | dzo | ok, got work to do, let me know when you get it working... |
22:23.40 | dcordes | dzo: ok |
22:26.20 | exco | right, paulproteus |
22:26.24 | exco | ssh works |
22:27.19 | exco | I can't even move the X that shows right before calibration |
22:28.08 | *** join/#htc-linux Marex (n=Marex@vasut.kolej.mff.cuni.cz) |
22:28.09 | paulproteus | That's very strange. |
22:28.18 | paulproteus | I don't know anything about the touchpad driver setup. |
22:29.14 | *** join/#htc-linux Marex (n=marex@vasut.kolej.mff.cuni.cz) |
22:29.25 | paulproteus | exco, You could try using x2x. |
22:29.32 | paulproteus | Then you can use your desktop mouse.... |
22:31.01 | Marex | cr2, hi, I think you wont get any patches from me in next two months :( |
22:34.07 | dcordes | dzo: ok with the regs from your email and a2M 3 I see a bunch of messages from the moem now when I cat /dev/smd0 |
22:34.22 | dcordes | s/moem/modem/ |
22:34.46 | cr2 | Marex: ? |
22:35.04 | Marex | cr2, well examination period started and my laptop died |
22:35.19 | Marex | isnt that cool, it dies when I need it most :E~ |
22:35.37 | dcordes | when I do an "echo -e fubar > /dev/smd0" repeatedly I see head but not tail changing (as before) |
22:35.54 | exco | ssh -XC xda x2x -east -to:0.0? |
22:37.09 | dzo | dcordes: what messages do you see from the modem? |
22:37.34 | dcordes | dzo: +CPBR ones eg |
22:37.50 | dcordes | with cosmic numbers |
22:38.07 | dzo | that's good, it means the receive buffer is working. looks like send isn't though. |
22:38.20 | dcordes | at the end there's a sequence of HTCCSQ ones |
22:38.39 | dzo | thats the modem reporting it's signal strength |
22:38.45 | dcordes | dzo: when I leave cat on a while, it does refres, ie the messages get reprint |
22:39.08 | dzo | yes, it will be set to report signal strength every few seconds. |
22:39.11 | dcordes | dzo: that's cool two weeks ago I didn't think we would have modem communication on kaiser at all anytime soon |
22:40.02 | dzo | perhaps the 3 in a2mmm is not correct, how far did marbalon get? could he send to the modem? |
22:40.07 | paulproteus | exco, -from :0 |
22:40.10 | paulproteus | I think. |
22:40.20 | paulproteus | exco, Wait, you're more confused than I thought. |
22:40.21 | dcordes | dzo: he just reported about the cat and response |
22:40.34 | dcordes | 23:23 < marbalon> dcordes: hmm when I run cat /dev/smd0 & a get a lot of buffered command |
22:40.36 | paulproteus | exco, ssh -XC xda x2x -east -from localhost:10 |
22:40.38 | paulproteus | That should do it. |
22:40.57 | paulproteus | Never mind- you had it right! |
22:40.59 | dcordes | 23:25 < marbalon> +CRES.... 23:25 < marbalon> @HTCCSQ... |
22:41.04 | paulproteus | exco, Ignore my comments - your way was okay. |
22:41.24 | dzo | OK, try changing the 3 then until the tail changes, you could just write to them all for a quick hack. |
22:41.28 | dcordes | dzo: and he used this smd.c: http://baliniak.pl/android/kaiser-smd.c |
22:41.34 | exco | sh: x2x: not found |
22:42.25 | dcordes | dzo: I learned how monotone a programmer's life can be today when I tried 0-7 today |
22:42.59 | dcordes | do you have the code for writing all 8 at a time? |
22:44.31 | dzo | dcordes: I saw that, he probably couldn't write to it either: try for(int i==0;i<7;i++) write(1,MSM_A2M_INT(i)) in notify-other-smd |
22:44.57 | dzo | s/i==/i= |
22:45.22 | *** join/#htc-linux roscomca_ (n=IceChat7@78.32.71.213) |
22:45.37 | dcordes | I don't think he wrote anything, otherwise he would have mentioned it I believe |
22:46.54 | dcordes | dzo: does that replace the old write line? |
22:47.26 | dzo | int i; for(i=0;i<8;i++) write(1,MSM_A2M_INT(i)); I don't know how many A2M addresses there are, 7 should do though. |
22:47.37 | *** join/#htc-linux JTRipper_ (n=JTRipper@p4FE96C38.dip.t-dialin.net) |
22:47.40 | dzo | yes |
22:48.52 | dcordes | so I put in "int i; for(i=0;i<8;i++) write(1,MSM_A2M_INT(i));" now? |
22:48.58 | dzo | yes |
22:49.49 | dcordes | dzo: 55: error: implicit declaration of function âwriteâ |
22:50.23 | dzo | sorry, should be writel |
22:50.35 | dcordes | ok works |
22:52.04 | dcordes | ..booting.. |
22:55.20 | dcordes | hrmm dzo did you ever experience those hangs on boot on vogue? |
22:55.28 | dzo | no, never. |
22:55.33 | dcordes | when it hangs in haret after 100% not saying gogogo |
22:56.10 | dzo | no, i don't know if it always says gogogo but it always boots. |
22:56.40 | dcordes | that's interesting I wander what kaiser has that creates that mess |
22:57.20 | dzo | could be a dma transfer or something that corrupts the image. |
22:58.01 | dcordes | now it hanged two times in a sequnce very annoying |
22:58.52 | dzo | could be that new code, it may do horrible things just writing to all the A2M addresses. |
22:59.39 | dcordes | hm I doubt it wrote anything - I need to create the node |
23:00.13 | dcordes | oh noes. tail still not changing |
23:00.25 | dzo | it could have because it calls stuff in the timer interrupt. |
23:00.44 | dcordes | cat still working, showing the modem status |
23:02.54 | dzo | OK, perhaps some bigger numbers then, try for(i=7;i<16;i++) |
23:04.15 | dzo | tis a pity you can't trace with haret, have you tried tracing MSM_CSR_BASE? |
23:06.10 | dzo | looking at android, they use 0 for notify-other-smd, you could try that. it could be that the other writes confused it. |
23:06.48 | dcordes | dzo: well I tried the 0 - but that was with the previous registers |
23:07.13 | dcordes | you got the mail history? |
23:07.18 | dzo | you were looking at the wrong tail then so wouldn't have known. |
23:07.33 | dcordes | ok |
23:07.44 | dcordes | shall I first try the for(i=7;i<16;i++) or rather 0 ? |
23:08.11 | dzo | try 7-16 first. |
23:08.18 | exco | isn't it weird I can't boot with the zimage symbolic link in place |
23:08.19 | exco | (it doesn't seem to be a problem with any of the other images)? |
23:08.47 | dcordes | symlink in wince? |
23:09.21 | dcordes | brb |
23:10.01 | exco | linux partition |
23:20.05 | dcordes | booting.. |
23:22.18 | dcordes | dzo: tail is stuck on dc1 now |
23:23.17 | dzo | OK, perhaps there is some other flag in the header to say that there is data in the buffer to write. |
23:23.43 | dcordes | which header? |
23:23.53 | dcordes | smd.h? |
23:23.57 | dzo | where the head and tail are. |
23:24.40 | dcordes | dzo: hm when the recv buffer works, can't we just look up there? |
23:24.45 | dcordes | I know that's stupid |
23:26.51 | dzo | no, receive buffer works because the arm9 tells us when things have changed with an M2A interrupt. We need to tell the A9 using the A2M int, you could print out the values before head and tail in the receive header to see if they change though. |
23:28.29 | dcordes | what's m2a again? |
23:30.57 | dzo | I'll think about it, I really have to do some work today though, see you later. M2A is an interrupt generated by the ARM9, A2M is a write to a memory location in MSM_CSR. A stands for Application Processor, M for Modem Processor, Arm11 and Arm9 respectively. |
23:55.08 | *** join/#htc-linux raph_ael (i=raphael@kikoolol.orbus.fr) |