01:24:04 | Sammy | morning everyone ... |
01:31:55 | Fare | Sammy: can you help with debugging blob? |
01:32:02 | Fare | (a new port of it) |
01:36:36 | prpplague | russ__: hows things? |
01:37:23 | Sammy | Fare: what's new port ? |
01:37:33 | Sammy | hello prpplague :D |
01:37:42 | prpplague | Sammy: knee how |
01:38:38 | Sammy | how's you going today ? |
01:39:20 | prpplague | Sammy: fine, just getting a bunch of office politics out of the way today |
01:43:50 | Sammy | prpplague: |o| |
01:46:47 | Sammy | go get more coffee ... |
01:50:24 | Fare | Sammy: jornada 820 |
01:50:30 | Fare | a nice machine |
01:50:42 | Fare | see ftp://samaris.tunes.org/jornada820/ |
07:12:30 | mmatten__ | morning |
07:15:02 | mmatten | akira|work :Nickname is already in use. |
07:15:04 | mmatten | !!!!!! |
07:15:24 | mmatten | oops |
07:15:47 | mmatten | wrong channel and even server.... |
09:15:00 | seletz | hi |
09:15:43 | seletz | erikm: hi. Was yesterday's porting session a success? |
09:22:55 | erikm | sort of |
09:23:22 | erikm | blob booted, but we think it died after a IRQ or FIQ |
09:23:34 | erikm | my guess was that the system wasn't in SVC mode |
09:23:47 | erikm | I don't know what Fare did to fix that |
09:48:10 | seletz | erikm: no LEDs? |
10:14:28 | erikm | seletz: no |
10:17:31 | seletz | erikm: Doh. I thought I'm the only one with no LEDs on his board. |
10:17:31 | seletz | erikm: LEDS help soooo mutch. grrr. |
10:17:31 | seletz | anyway, SIGLUNCH :) |
10:23:52 | Sammy | evening all |
10:25:34 | erikm | afternoon, sammy :) |
11:34:05 | seletz | Sammy: hi |
12:00:32 | Fare | Gakuk! |
12:00:37 | Fare | erikm: are you around? |
12:01:15 | Fare | erikm: while you were away, I actually regressed! |
12:01:31 | Fare | If you go away on this summer day, then you might as well take the sun away |
12:04:49 | erikm | Fare: hi |
12:05:17 | Fare | hi. |
12:05:33 | Fare | Now I can't even get any output on the serial console |
12:05:57 | Fare | (not from blob, anyway - I realized I could get lots from hpcboot) |
12:06:16 | Fare | also, in its logs, hpcboot reports CPSR :600000df which seems just WRONG |
12:06:18 | erikm | Fare: try to blink a LED from blob. or blink the LCD backlight |
12:06:42 | Fare | what user mode is 11111 ? or can hpcboot report a false CPSR? |
12:06:55 | Fare | how can I determine where the LCD backlight is? |
12:08:01 | Fare | And why does blob shut down the screen, whereas the hpcboot-misbooted vmlinux would crash with the framebuffer mapped and binary visible on it? |
12:08:44 | Fare | considers crashing blob early on to see if the framebuffer is still mapped - maybe we can put the framebuffer to good use if we don't disable it |
12:08:51 | erikm | Fare: blob does not switch down the screen. it assumes the framebuffer is not switched on |
12:09:14 | Fare | what's a good way to crash blob early on? |
12:09:48 | erikm | Fare: the best way is to run it in something other than SVC mode |
12:10:18 | erikm | Fare: another good way is to have interrupts enabled and bogus IRQ/FIQ vector at 0x00000000 |
12:10:21 | Fare | I mean, what could I put in a .S to crash it, if possible w/o modifying framebuffer settings |
12:10:45 | Fare | isn't 0x00000000 in ROM while the MMU is deactivated? |
12:10:52 | erikm | yes |
12:11:46 | Fare | code that would basically "disable interrupts, blink a few pixels, flush buffers, go back to blink" would be great |
12:12:11 | Fare | (and why could I get output before and no more?) |
12:16:24 | Sammy | ho ho , go weekend ... |
12:16:48 | Sammy | later all ... |
12:18:30 | erikm | Fare: yes, such code would be nice |
12:18:45 | erikm | Fare: and why no more output? I don't know what you changed... |
12:21:24 | Fare | I don't understand what I changed either. |
12:21:46 | Fare | The change set is rather small, and I uncommented any special thing I did |
12:22:12 | Fare | at one time, I got mixed up with the link command, before I settled it in a way that looks like it works |
12:23:19 | Fare | I can send you my current patch |
12:27:44 | Fare | could it play a role that I'm running on external power supply, without batteries? |
12:30:34 | Fare | doesn't seem to make any difference. |
12:30:43 | Fare | That CPSR:600000df bugs me |
12:31:01 | Fare | does the SA1100 have a running mode 11111 ? |
12:33:44 | erikm | no idea, I don't have the documentation over here |
12:34:21 | Fare | The SA1100 docs doesn't say anything. It tells me to look for a separate architecture document. |
12:34:42 | Fare | The only architecture document I have is the ARMv7.pdf which doesn't have a 11111 mode |
12:35:11 | Fare | (SVC is 10011, user is 10000, other ones are all 10xxx) |
12:42:08 | Fare | hum. I inserted a halt: b halt just after disabling interrupts in start.S, and the computer crashed with black screen immediately |
12:42:09 | Fare | WAIT! |
12:42:55 | Fare | if RAM is mapped at 0x00000000 as I believe, then loading at 0xc000000 could cause severe lossage, since it would overwrite the vectors! |
12:57:47 | Fare | grrrrr/ |
12:57:58 | Fare | I don't understand what happens. |
13:01:43 | Fare | hum. How come, when I touch start.S and make blob-chain, nothing happens? |
13:03:07 | erikm | because blob-chain doesn't depend on start.S |
13:03:23 | erikm | RAM is not mapped at 0x00000000 |
13:03:42 | erikm | physical address 0x00000000 is *always* some kind of ROM |
13:04:00 | erikm | (and yes, that's on every ARM CPU) |
13:06:55 | Fare | what's the first file in blob-chain? chain.S? trampoline.S ? |
13:11:09 | Fare | puts halt: b halt on the beginning of chain.S... |
13:12:29 | Fare | well, at least that works: |
13:12:45 | Fare | it hangs the computer w/o blanking the screen |
13:13:20 | Fare | Is there a simple way to detect where the framebuffer is mapped? |
13:15:12 | erikm | chain.S |
13:15:27 | erikm | the Makefile.am shows the dependencies |
13:16:51 | erikm | if the framebuffer is on, it's mapped |
13:21:11 | seletz | erikm: marcus has a problem: We've updated our cramfs libs to libc2.2.3, now init won't start anymore. I've change the libc.so file to point to the right libs (its a text file). Is there something obvious i'm missing? (the libs are from your tool chain). |
13:26:20 | erikm | seletz: yes, ld.so has a hardcoded search path |
13:26:38 | erikm | seletz: the path is set to where your configure --prefix is |
13:26:54 | mmatten | mmm how to change ? |
13:27:01 | mmatten | (without recompile) |
13:27:19 | erikm | either binary edit ld.so, or make the correct path+symlink on the cramfs |
13:29:24 | mmatten | erikm: ok thanks, i#ll have a look :) |
13:38:20 | seletz | man, this is bad design. |
13:38:27 | seletz | IMHO, that is. |
13:45:46 | Fare | ok, so I can debug, now that I can distinguish between on and off |
13:45:53 | Fare | I have an output bit per run. |
13:46:13 | Fare | let's work towards increasing the throughput. |
13:46:41 | Fare | the goal is: identifying where the framebuffer is mapped. |
13:47:21 | Fare | any hint? |
13:51:30 | Fare | hum. |
13:52:37 | Fare | let me get things right: if hpcboot did its job correctly, we'd be in SVC mode with MMU off, so that 0x00000000 would be ROM, isn't it? |
14:06:17 | erikm | *nod* |
14:06:53 | erikm | frame buffer map can be figured out from the frame buffer registers, see SA1100 manual |
14:10:29 | prpplague | morning all |
14:10:53 | erikm | hi prpplague |
14:11:12 | prpplague | erikm: whats cookin today? |
14:28:53 | seletz | moin |
14:29:21 | Fare | so when hpcboot says the framebuffer is mapped at 0x0000000, it might as well be a failed access to the register, due to lack of privileges? |
14:29:48 | Fare | can you recommend a hacking environment under wince? |
14:39:42 | Fare | if we aren't in SVC, how come we could go as far as outputing a few characters on the serial line? |
14:44:48 | prpplague | seletz: hows work on the cpu core? |
14:47:40 | erikm | Fare: you probably get an exception before the serial output buffer is slushed. |
14:47:44 | erikm | s/slushed/flushed/ |
14:47:56 | seletz | prpplague: fine so far. Have the specs and BOM mailed to a HW manufactorer. Next week on Friday we'll have a meeting about layout and design issues. Then i'll get a price offer. |
14:48:43 | seletz | prpplague: They are fine with our "order xx per month" idea. |
14:49:04 | Fare | hum. If I was writing zeros over the palette, it might have explained the black screen. |
14:49:04 | seletz | prpplague: so we dont have to order 100 units first. |
14:49:46 | prpplague | seletz: cool |
14:49:55 | Fare | erikm: how come I could output more characters when the serial was at higher speed? isn't the output buffer fixed-size? |
14:50:02 | seletz | prpplague: erm, s/100/1000 |
14:50:15 | prpplague | seletz: we've started plans to move our thin clients from x86+crt to SA+lcd |
14:51:26 | seletz | prpplague: cool. Next week i'll meet a guy who's interested in embedded on PC104. I hope i'll get him to use ARM. He wants LCD+Touchscreen. Fits fine with our core.... |
14:52:11 | seletz | prpplague: for now i'll show him my board from PT and claim its my own ond :^} |
14:52:18 | seletz | prpplague: s/ond/one/ |
14:52:41 | prpplague | seletz: lol |
14:52:41 | Fare | seletz: how much costs a custom-built ARM board + LCD + touchscreen ? |
14:52:51 | seletz | Fare: lol |
14:53:00 | prpplague | seletz: i ws thinking about ordering one of these - http://www.thinplanet.com/Products/list.asp?field=PRdesclng&Value=risc&Sort=Manufacturer |
14:53:57 | seletz | uargh, thats cheap |
14:54:12 | seletz | lol |
14:54:18 | | | Server O/S Windows 2000 Terminal Service, |
14:54:18 | | | WindowsNT 4.0,(with Terminal Service edition) |
14:55:42 | seletz | Fare: depends on what _exactly_ you want |
14:57:12 | | | Texas School District Chooses ThinSoft |
14:57:12 | | | The Poth Independent School District joins a growing list of US schools and universities implementing the BeTwin solution to meet educational computing requirements. |
14:57:33 | seletz | prpplague: what do you think? |
14:59:39 | prpplague | seletz: whats that from? |
15:00:05 | seletz | prpplague: your url |
15:00:15 | prpplague | hmm |
15:00:23 | seletz | prpplague: just followed some links |
15:00:38 | seletz | prpplague: the no where say that they support linux |
15:01:47 | prpplague | seletz: ohh, for the thinclient, ya, i don't even bother looking for linux support these days, i just find the hardware and make linux work |
15:03:36 | prpplague | seletz: we remove the "made for winblows" or "powered by windows ce" logo off 75% of the hardware we resale |
15:04:22 | seletz | prpplague: lol, cool |
15:04:24 | mmatten | erikm: neither link nor replacing the path in the ld.so seem to work, i still receive:Kernel panic: Attempted to kill init! |
15:04:24 | mmatten | oot: erro r while loading shared libraries: cannot read file data: cannot load shared object file: Error 21 |
15:05:07 | mmatten | error 21 is supposed to mean: hey i'm a directory not a file |
15:05:12 | mmatten | but that makes no sense to me |
15:05:57 | prpplague | seletz: that arm based thin client is supposed to be $351, but without all the licensed software, its only $289 |
15:06:12 | seletz | prpplague: thats really cheap |
15:08:47 | seletz | prpplague: i wonder what the TCO with citrix server licenses and win2k licenses is ... |
15:09:20 | prpplague | seletz: alot |
15:09:42 | erikm | mmatten: try to link init statically |
15:09:50 | prpplague | seletz: we go up against other accounting systems that use citrix and win2k stuff, and they are usually 3x our costs |
15:18:33 | Fare | seletz: I want an (interactive if possible) (free software if possible) low-level development environment to help debug a bootloader for linux |
15:18:42 | Fare | (and/or another OS) |
15:23:00 | Fare | hum. look like blitting a VGA screen at 25Hz requires 3 to 4 MHz of memory access bandwidth |
15:24:33 | prpplague | erikm: what do you think, think i could get blob working on that thin client? |
15:25:21 | erikm | prpplague: should be doable. does it have flash and a serial port? |
15:25:55 | prpplague | erikm: ya, and the jtag is clear marked on the pcb |
15:26:33 | erikm | prpplague: ok, should be doable. I wonder about their video specs, though |
15:26:47 | erikm | prpplague: but that's not necessary to get blob working on it :) |
15:28:53 | erikm | goes home |
15:49:17 | mmatten | goes home too |
15:49:55 | Russ | morning prpplague |
15:51:06 | seletz | going home |
15:51:11 | seletz | Russ: hi |
15:51:24 | seletz | see ya all .... |
15:53:35 | prpplague | Russ: morning |
15:53:45 | prpplague | Russ: hows the camion hacking? |
16:08:52 | Russ | good |
16:09:07 | Russ | tracing the audio paths and some remaining BCR bits and gpio's |
16:10:31 | prpplague | Russ: sounds like you've done nice job |
16:10:51 | prpplague | Russ: maybe i can send you that thin client to play with next, lol |