04:39:47 | Russ | anyone who has amd flash and jtag, bang on that if you would |
08:52:47 | mmatten | morning |
09:31:34 | Fare | ok, I have the *proof* that some DMA activity happens, thanks to md5chk |
09:56:48 | Fare | thanks to DAC, I could locate the DMA buffer at 0xc005c080 |
09:57:28 | Fare | and it looks like it is USB |
10:12:24 | Fare | and the length of the DMA buffer seems to be 4 bytes (!) |
11:50:35 | seletz | hi * |
11:51:40 | seletz | Russ: are you here? |
11:55:42 | hugendubl | hi all! |
11:56:22 | seletz | hugendubl: russ is not here as it seems. So no amd flash answers today :( |
11:56:44 | seletz | hugendubl: have a fax in your office? I could fax you the offer from tecnotron. |
11:58:29 | seletz | hugendubl: ping .... |
12:02:32 | hugendubl | sorry. Yes... |
12:02:35 | hugendubl | ? |
12:02:47 | seletz | k |
12:03:34 | hugendubl | nick hugendubl_afk |
12:03:42 | seletz | :) |
12:12:46 | Fare | if any sf.net blob maintainer is here, you can delete the old version of my patch |
12:13:42 | Fare | is any blob expert in here? |
12:23:09 | seletz | yup |
12:23:26 | seletz | Fare: have you posted it to sf.net? |
12:23:37 | seletz | checks out |
12:24:56 | seletz | Fare: patch nr 570014 to delete, ok? |
12:27:32 | Fare | seletz: don't delete the patch |
12:27:37 | Fare | delete the older file attachment |
12:27:38 | seletz | :) |
12:27:57 | Fare | darn -- shouldn't have given both the same name |
12:28:10 | Fare | but if you delete the wrong one, then delete both and I'll resubmit |
12:28:19 | seletz | Fare: ok, the patch on sf.net is ok, then? |
12:28:23 | Fare | you can check by size |
12:28:29 | Fare | I attached a new file |
12:28:32 | Fare | the old one is ba |
12:28:34 | Fare | bad |
12:28:37 | Fare | the new one is ok |
12:29:03 | seletz | hmm. |
12:29:38 | Fare | the right one has size 37772 |
12:29:44 | seletz | ok, i delete both because i cant differ them (the delete combo box does not show the file size) |
12:29:50 | Fare | delete the one below |
12:30:16 | Fare | the listing further down gives files in newer-to-older order |
12:30:26 | seletz | k |
12:30:30 | seletz | done |
12:30:42 | Fare | (when I submitted, I clicked on "delete" for the old one, but I got a "permission denied") |
12:31:00 | seletz | doh |
12:31:19 | seletz | ok, should be ok now. Can you double check, please? |
12:31:32 | Fare | the changelog says you deleted the right one |
12:31:41 | seletz | Fare: is it to be integrated? |
12:31:54 | seletz | Fare: or do you work on it? |
12:32:38 | Fare | I downloaded it - it's ok |
12:32:43 | Fare | it is to be integrated |
12:33:10 | seletz | Fare: well, ok. I'll take a look and integrate it as time allows. (a week at max). |
12:33:35 | Fare | I still have to fix an issue, though: some DMA device (I think) is constantly modifying word at 0xc005c080 |
12:34:07 | Fare | Are you familiar with the SA1100 ? |
12:34:14 | Fare | what device could that be? |
12:34:28 | Fare | (Or could it be a bug in the RAM chip?) |
12:34:38 | seletz | use diag to check |
12:34:42 | Fare | (doesn't look like a typical RAM bug - plus WinCE runs fine) |
12:34:45 | seletz | it has a memchk routine |
12:34:59 | Fare | I *know* the value at that place changes constantly |
12:35:14 | Fare | every time I peek there, it's different, but < 65536 |
12:35:14 | seletz | hmm |
12:35:24 | Fare | using md5chk, I could locate the place |
12:35:36 | Fare | no other word seems to change that way -- not in the first 1MB, at least |
12:35:49 | Fare | (and I have no problem anywhere else yet) |
12:36:01 | seletz | have had a look at your LCCR dma registers? |
12:36:06 | Fare | it took me quite a while to understand what was going on |
12:36:08 | Fare | LCCR ? |
12:36:12 | Fare | no |
12:36:13 | seletz | LCD |
12:36:25 | seletz | the only thing i know of using DMA |
12:36:37 | Fare | the LCD framebuffer is mapped at 0xc0002e00 (including 0x200 bytes for palette) |
12:37:08 | Fare | I don't think the LCD uses DMA to *write* anything, only to read. |
12:37:16 | seletz | k |
12:37:19 | seletz | agreed |
12:37:19 | Fare | here, obviously, someone is *writing* |
12:37:41 | seletz | erik mentioned some serial io could also do DMA |
12:37:42 | Fare | (being mapped at 0xc0002e00 has the interesting side-effect that you can *see* the kernel getting loaded) |
12:37:50 | seletz | cool |
12:37:52 | seletz | :) |
12:38:09 | seletz | looks weird i guess |
12:38:26 | Fare | you can make the difference between code, data, and compressed stuff. |
12:39:06 | Fare | you can even see the place the logo is stored :) |
12:39:07 | seletz | reminds me of core wars .... |
12:39:15 | Fare | yeah, me too |
12:39:37 | seletz | write a LCD kernel tracer. |
12:39:47 | seletz | (just kidding) |
12:39:52 | Fare | especially when I saw words being written where they shouldn't be written (by blanking the screen first), during a few crashes |
12:40:16 | Fare | "hey! a core war program is invading the screen!" |
12:40:24 | seletz | hmm. actually a quite smart way to debug. |
12:40:51 | seletz | You cant zoom in, can you? |
12:40:54 | Fare | well, first thing I did when the kernel was loaded was to try to blink the logo in-core |
12:41:02 | Fare | but the routine was never called |
12:41:06 | Fare | no, I can't |
12:41:09 | seletz | :) |
12:41:21 | Fare | problem is: you need the 512 bytes of palette first |
12:41:27 | Fare | so you can't do that to a live program |
12:41:52 | seletz | skimming through diff |
12:41:54 | Fare | unless you make sure that you can write those 512 bytes with something you like |
12:42:00 | Fare | which diff? mine? |
12:42:04 | seletz | Fare: yup. |
12:42:08 | Fare | oh, thanks :) |
12:42:38 | Fare | (might be a nice idea: every 300kb, reserve 512 bytes for the palette) |
12:44:18 | Fare | seletz: if you delete some of my debugging stuff, it'd be nice to put it in CVS first and remove it afterwards, so there's a trace left in CVS |
12:44:39 | seletz | Fare: didn't you use the dontdiff file? |
12:45:06 | seletz | Fare: sorry, yes you did. |
12:45:45 | Fare | dontdiff ? |
12:46:16 | seletz | Fare: false alarm, dont mind :) |
12:46:20 | Fare | I used cvs diff, and piped it into a script to transform it into something acceptable by patch -p0 |
12:46:46 | Fare | cvs diff only diffs registered files; I manually added a diff for the 4 added files |
12:47:27 | Fare | the serial console is neat - I wrote some debugging help in lisp that I could cut&paste into the blob console |
12:47:42 | seletz | ugh |
12:48:07 | Fare | like, outputing the md5chk commands necessary to locate that changing memory location |
12:50:02 | Fare | (including an semi-automated dichotomy search, where I answered "yes" or "no") |
12:55:36 | seletz | Fare: lots of debug code still in. |
12:55:52 | seletz | Fare: and also lots of asm stuff i dont quite understand |
12:56:20 | seletz | Fare: i have to test it wether or not it breaks anything |
12:56:46 | seletz | Fare: did you test whether or not the other platforms still compile? |
12:57:06 | seletz | Fare: pity, i havent a sa1100 board around. |
12:57:37 | Fare | no, I haven't tested on any other platform, but it should still compile |
12:57:51 | Fare | I made no change that could modify other platforms, except in chain.S |
12:58:01 | Fare | (that was buggy in my previous patch, but now fixed) |
12:58:08 | seletz | Fare: yup, thats it. chain.S |
12:58:20 | Fare | basically, it used to move 64KB always |
12:58:28 | seletz | Fare: there are tools to compile on all platforms in tools/ |
12:58:31 | Fare | I changed that to move exactly the size of the rest |
12:58:51 | Fare | (as computed by rest-ld-script and stored in trampoline.S) |
13:01:08 | seletz | Fare: you basically used different registers and a different copy loop |
13:01:18 | Fare | yes |
13:01:24 | seletz | Fare: what was wrong with the old version? |
13:01:31 | Fare | and different initialization of the copy loop |
13:01:40 | Fare | it moved exactly 64KB |
13:01:58 | Fare | didn't look for the actual size of blob |
13:02:11 | seletz | ah, ok |
13:02:14 | seletz | i understand |
13:02:23 | Fare | I changed registers because it's easier to count r0-r7 than r3-r10 |
13:02:39 | seletz | this might actually solve some strange problems i have with blob |
13:03:05 | seletz | Fare: my blob is quite, erm, bloated as erik would put it :) |
13:03:07 | Fare | :) |
13:03:36 | Fare | I recommend you put a forth in it |
13:03:36 | seletz | Fare: if i get over a certain size blob wont start anymore |
13:03:47 | seletz | running away in horror |
13:03:47 | Fare | now it will :) |
13:04:03 | Fare | why horror? |
13:04:13 | seletz | Fare: you mean the same stuff like on macs? |
13:04:27 | Fare | mac actually took it from sun |
13:04:50 | seletz | Fare: PPC OpenYetAnotherName SuperDuperBIOS |
13:04:57 | Fare | and sun uses it because they had a 10 man team failing to develop their boot prom in C, whereas a lone man in another team made it work superbly in FORTH. |
13:05:03 | Fare | in his spare time. |
13:05:14 | Fare | OpenBoot |
13:05:34 | Fare | IEEE 1275 |
13:05:37 | seletz | i never understood how to actually use it :( |
13:06:01 | Fare | http://playground.sun.com/1275/home.html |
13:06:47 | Fare | I have a 32-bit i386 forth in 12KB |
13:07:16 | seletz | checking out |
13:08:49 | seletz | uargh |
13:09:35 | seletz | huge |
13:09:51 | Fare | what's huge? |
13:09:58 | Fare | the 1275 site? |
13:10:02 | seletz | ya |
13:10:10 | seletz | lots of PDF |
13:10:12 | Fare | well, you're not forced to abide by 1275 so as to use FORTH |
13:10:43 | Fare | actually, lots of FORTHers think that a small roll-your-own FORTH is worth much more than a bloated ANS-compliant FORTH. |
13:10:55 | Fare | (bloated *and* misdesigned at times) |
13:11:15 | seletz | Fare: so whats that FORTH language like, anyway? |
13:11:35 | seletz | and why have a own language in a bios? |
13:11:41 | seletz | or even a boot loader? |
13:11:46 | Fare | it's, together with commonlisp, the only turing-equivalent language around |
13:12:00 | Fare | forth is a dynamic language |
13:12:12 | Fare | you can interactively debug your program |
13:12:32 | Fare | cuts the compile-crash cycle to a read-eval-print cycle |
13:12:57 | Fare | between two attempts to fix bug/add feature, you have a few seconds, not a few minutes |
13:12:58 | seletz | give me an example |
13:13:28 | Fare | well, for instance, I could have written a small program at the prompt to locate that memory location |
13:13:32 | Fare | without having to reboot |
13:13:40 | Fare | or recompile or anything |
13:13:55 | Fare | I could develop a serial driver right there at the prompt |
13:13:59 | Fare | tweak it |
13:14:07 | Fare | or change one routine which I think is buggy |
13:14:17 | seletz | hmm |
13:14:20 | Fare | or apply a bug fix on the running systm |
13:14:27 | Fare | or do various experiments |
13:14:28 | seletz | w/o a serial device you wont see an output |
13:14:42 | Fare | depends which is your primary console |
13:14:48 | seletz | k |
13:14:50 | Fare | or write a console device from the serial |
13:14:54 | Fare | if you prefer |
13:15:08 | Fare | and see it as you type it |
13:15:15 | seletz | but all this would be limited to pre-kernel stuff |
13:15:21 | Fare | hehe |
13:15:28 | Fare | not if I get it my ways :) |
13:15:36 | seletz | ? |
13:15:55 | seletz | in-kernel fortzh module ? eek! |
13:16:27 | seletz | dammit 30+ deg. C. |
13:16:31 | Fare | actually, more like rewriting the kernel in lisp and forth |
13:16:39 | seletz | needs a drink |
13:16:41 | Fare | where are you? |
13:16:43 | seletz | aaaaaaahhhh |
13:16:49 | seletz | Fare: southern germany |
13:17:28 | Fare | I know there is a strong forth group in vienna, and many forth groups in germany |
13:18:01 | Fare | but I don't know enough of german geography to tell where a town is by its name, except for a few towns, most of them in the north of .de |
13:18:29 | Fare | the only towns in the south that I could name are Munich and Regensburg |
13:18:41 | seletz | Fare: sorry, are you saing you want to re-write the kernel in forth?? |
13:19:04 | Fare | more of it in lisp, actually |
13:19:08 | seletz | Fare: 2 hrs off Munich, to the south-west. Near the lake of constance. |
13:19:18 | seletz | Ravensburg |
13:19:38 | seletz | Fare: but why? Another virtual machine? |
13:19:50 | Fare | ever heard of genera? |
13:19:54 | seletz | no |
13:19:58 | Fare | http://fare.tunes.org/LispM.html |
13:20:14 | seletz | checking out |
13:20:33 | Fare | the interesting stuff is not my page, but those it points to |
13:20:54 | Fare | my page is just silly details. |
13:26:30 | seletz | Fare: you spent lots of money |
13:28:23 | Fare | not more than a PIII would have cost at the time |
13:28:36 | Fare | and cheaper than an Alpha with a pirated (oh!) copy of Genera |
13:29:40 | Fare | nice to see a 1991 single-user (at a time) workstation with X, NFS, an integrated debugger, and more. |
13:30:20 | Fare | and an interactive environment, integrated hypertext documentation, source for most everything, etc. |
13:39:57 | seletz | just reading your LAK thread ... |
13:44:40 | Fare | LAK ? |
13:46:12 | Fare | I hesitate between 1) loading the kernel higher and not bothering or 2) trying to find which device is mapped at 0xc005c080 |
13:46:16 | Fare | not today, though |
14:26:26 | Fare | seletz: so, what did you decide about the patch? |
17:55:55 | Fare | hum |
18:04:11 | Fare | seletz: I have a version that cleans up the mess in main.c and moves it to jornada820.c only. |
18:14:11 | Russ | I'm here |
18:33:41 | Russ | Fare: selectz is gone tho |
18:46:15 | Fare | oh |
18:47:03 | Fare | Anyway. I won't resend a patch just for a small cleanup, unless the patch is rejected -- the overhead for the code review is not worth it |
18:47:33 | Fare | but the cleanup will be there in next patch if/when I solve the current issue with DMA from within blob rather than in the kernel. |
18:49:24 | Fare | are you familiar with SA-1100 ? |
19:24:55 | Russ|werk | ya |
19:29:04 | Russ|werk | why not turn off the DMA transfer by turning off the appropriate register bit |
19:29:48 | Russ|werk | because even when you mask all interrupts, DMA transfers will still occur |
19:49:32 | Fare | Russ: there remains to identify the proper register |
19:49:38 | Fare | Russ: do you know which it might be? |
19:50:08 | Russ|werk | read the sa-1100 manual |
19:50:20 | Fare | yeah, that's on my TODO list. |
19:50:26 | Fare | too bad there isn't a grep for PDF |
19:50:34 | Fare | the manual is HUGE |
19:50:46 | prpplague | Russ|werk: you got a sec for question on framebuffer? |
19:51:13 | Russ|werk | Fare: go to the udc section |
19:51:20 | Russ|werk | Fare: it has a table of contents |
19:51:21 | Fare | udc section? |
19:51:37 | Russ|werk | ya, its an acronym for usb client or something |
19:51:39 | Russ|werk | prpplague: ya |
19:56:21 | prpplague | Russ|werk: i'm kinda new on the fb stuff, whats the difference in compiling fonts in the kernel and not? |
19:56:51 | Russ|werk | I don't know about the fonts |
19:56:56 | Russ|werk | I just let the kernel do its thing |
19:57:24 | prpplague | Russ|werk: hmm so you don't compile in the low level drivers and the 8x8 and 8x16 fonts? |
19:58:04 | Russ|werk | I just go with the defaults |
19:58:10 | prpplague | hmmmm |
19:58:39 | prpplague | Russ|werk: thats what i did, but now it seems that the ansi escape sequence for flashing text does not work |
19:58:53 | Fare | btw, has anyone ever gotten the 6x11 font to work on anything else but a mac? |
19:58:56 | Fare | it'd be nice |
19:59:25 | prpplague | Fare: what about you? you compile in fonts on fb? |
20:01:09 | Fare | prpp: the current video drivers take special measures so it won't work but with the mac driver |
20:01:27 | Fare | sad but true |
20:01:48 | Fare | I will do it, but it's low on my hot list |
20:41:43 | Fare | section 11.8 of the SA-1100 mentions that the UDC can use DMA, but none of the registers described explain how DMA happens :( |
20:42:15 | Fare | I guess I lost my time and should have been reading section 11.6 instead |
21:06:13 | Fare | ok, grokked 11.6 |
21:06:19 | Fare | now for implementation time. |
21:06:22 | Fare | My! |
21:06:25 | Fare | hardware SUCKS |
21:06:28 | Fare | yet it ROCKS |
21:07:51 | prpplague | Fare: hmm, SUCKS ROCKS? |
21:08:15 | prpplague | Fare: sounds kink ick to me |
21:08:23 | prpplague | s/kink/kinda |
21:22:56 | Fare | well, it's amazing what hardware does for you. But it sucks having to write the grunt code to tame it. |
21:24:22 | prpplague | ya i'm in the same boat, try to work with a National Sh#$^h^h^h Semiconductor geodegx1 system |
21:25:13 | prpplague | erikm: ! |
21:25:17 | prpplague | erikm: up late? |
21:25:29 | erikm | prpplague: just the guy I was looking for! |
21:25:41 | prpplague | erikm: i didn't do it, you can't prove a thing |
21:25:53 | erikm | hah! |
21:26:06 | erikm | prpplague: nah, I have info about my flight |
21:26:25 | erikm | prpplague: I'll be flying to NY tomorrow afternoon |
21:26:39 | prpplague | erikm: tomorrow? |
21:26:53 | prpplague | erikm: i won;t be arriving in ottawa till tuesday the 25th |
21:26:57 | erikm | prpplague: *nod* staying a couple of days with a friend who works with IBM watson |
21:27:04 | prpplague | erikm: ok cool |
21:28:01 | erikm | I'll be flying to ottawa at the 25th |
21:28:50 | erikm | my flight is continental co4082 from newark, arriving at 4:50pm in ottawa |
21:29:03 | prpplague | copies to gedit |
21:29:45 | prpplague | ok, i don't arrive unit 5:15pm |
21:29:54 | prpplague | aa1828 |
21:30:23 | prpplague | erikm: i double checked the hotel, we have a two room suite with 3 beds |
21:31:38 | erikm | I just bought you four bottles of beer |
21:32:02 | prpplague | coooool |
21:34:52 | erikm | I'll try to wear some linux related t-shirt. penguin shirt or so. |
21:35:18 | Fare | erikm: hi! |
21:35:25 | erikm | hi |
21:35:42 | prpplague | erikm: ok cool, i'll probably be wearing a blue dress-shirt with a tux penguin logo |
21:36:06 | Fare | now that I read section 11.6 (DMA Controller) of the SA-1100 manual, I'm armed to make linux decompress at the right place |
21:36:55 | erikm | cool |
21:37:43 | Russ|werk | erikm: do you have any devices with amd flash? |
21:38:02 | erikm | Russ|werk: only the tux, but you have the same |
21:38:40 | Russ|werk | k, the old flash driver can destroy the bootloader even if you are flashing other partitions |
21:38:56 | Russ|werk | so its very important that the new one gets testest and out there asap |
21:38:58 | Russ|werk | er |
21:38:59 | Russ|werk | tested |
21:39:11 | Fare | it's so *slow* to work with hardware |
21:39:28 | Fare | especially when you're isolated and halfwise reverse-engineering it. |
21:39:38 | Fare | yet I admit the SA-1100 is a very clean design. |
21:39:40 | Russ|werk | erikm: also, why is cpu-sa1110.c so much simpler than cpu-sa1100.c |
21:39:50 | Fare | It's easy to learn, because it's well designed. |
21:40:17 | erikm | Russ|werk: hmm, IIRC somebody on tuxscreen.net had exactly that problem |
21:40:26 | Fare | very much unlike the 1980 crapola that still exists in the PC world. |
21:40:46 | Russ|werk | erikm: I know |
21:41:11 | Fare | erikm: should I define registers in the jornada820.h, or should I read definitions from the linux headers? |
21:41:13 | erikm | Russ|werk: cpu-sa1100 needs some more cleanup. seletz and chris hoover cleaned up cpu-sa1110 some more |
21:41:33 | Russ|werk | Fare: I'm finished up my reverse engineering of this: http://russ.dhs.org:8000/back.jpg |
21:41:35 | BZFlag | Russ|werk: do you think you found the cause of the bootloader killing? |
21:41:48 | erikm | Fare: *nod* ARM design in general is very clean |
21:41:51 | Russ|werk | BZFlag: ya, the new flash driver is basically a complete rewrite |
21:42:26 | BZFlag | but did you determine what was wrong with the old method? |
21:42:38 | Fare | erikm: shouldn't disabling of DMA be part of the standard blob procedure? |
21:42:41 | erikm | Fare: what registers do you need? |
21:42:58 | Fare | well, I suppose enough to disable DMA registers |
21:43:02 | Russ|werk | I think it was putting it in the mode where it would be able to write multiple bytes at once, without doing the unlock commands again |
21:43:07 | Russ|werk | but then not taking it out of that mode |
21:43:15 | BZFlag | ahh |
21:43:38 | erikm | Fare: no. blob expects to be the first program to run after reset, in which case DMA is disabled. the chainloader is a hack and might need more work |
21:43:38 | Fare | DCSR{0-5} |
21:43:53 | erikm | Fare: just include blob/sa1100.h and you have the register definitions |
21:43:59 | Fare | ok |
21:44:02 | prpplague | Russ|werk: you do anything with that tcsx-1 board i sent to ya? |
21:44:08 | erikm | Russ|werk: what kind of device is that? |
21:44:15 | prpplague | Russ|werk: not that it matters, just curious |
21:44:57 | erikm | Russ|werk: (some kind of handheld, judging from the form factor) |
21:44:58 | Russ|werk | prpplague: I have a DoC for it, as well as a 30G pcmcia drive |
21:45:04 | Russ|werk | I need to get a cable for the drive though |
21:45:14 | erikm | 30G pcmcia drive??? |
21:45:18 | Fare | ok, so I'll improve my jornada820 patch, and update it on the sf.net page |
21:45:30 | Fare | 30G pcmcia drive sounds nice |
21:45:43 | Fare | too bad I can't have BOTH it and the network card |
21:45:57 | Fare | hum. is it a type 1 (thin) pcmcia card? |
21:46:09 | Fare | maybe not... |
21:46:40 | Russ|werk | erikm: the removal header does say majordomo@lart.tudelft.nl and not majordomo@lart.tudelft.edu, right? |
21:46:52 | erikm | yes |
21:47:03 | Russ|werk | this rob biggs is a real dumbass |
21:47:13 | erikm | tudelft.edu doesn't exist |
21:48:13 | prpplague | Russ|werk: lol, i hope thats not how you guys talk about me when i'm not on the channe |
21:48:13 | erikm | last time I checked, that is.... /me rechecks |
21:48:19 | Fare | hum. |
21:48:43 | Fare | Ok. And what shall I do first thing after getting the kernel to uncompress? |
21:49:18 | Russ|werk | mtd? |
21:49:34 | Fare | how do I disable devices I don't want to debug yet, and enable devices I want to debug (console) ? |
21:49:36 | erikm | Fare: take a beer? |
21:49:38 | erikm | ;) |
21:49:44 | Fare | erikm: yay :) |
21:49:52 | erikm | Fare: I'm sure prpplague can recommend you a nice victory beer :) |
21:50:12 | prpplague | did i hear someone say beer? |
21:50:23 | Russ|werk | Fare: kernel config |
21:51:47 | Fare | also, which devices should I try first? I suppose serial, then lcd, then pcmcia (and with it network and IDE), then USB, then internal mouse and keyboard. |
21:52:36 | erikm | Fare: serial is first, but will probably work out of the box. ram disk is nice so you can actually test if ls etc. work |
21:53:16 | erikm | Fare: LCD should be easy as well, the only problem you could encounter is how to enable the backlight |
21:53:48 | erikm | Fare: LCD is easy to debug: once you get a penguin at boot with the correct colors you know you did it right |
21:54:25 | Fare | LCD seems pretty straightforward, but it isn't clear how the parameters map to hardware registers |
21:54:43 | Fare | ideally, I should be reading the registers from the current setup and feeding them back to linux |
21:55:10 | Fare | only it's not clear what parameter should go where... |
21:55:22 | Fare | if you know someone who can tip me, that'd be great. |
21:56:10 | Fare | because I just don't have the data to make up the values - I *must* get them from the WinCE setup. |
21:56:26 | Russ|werk | look at the relevant portion of the sa-1100 manual, the header file, and the spec sheet for the LCD panel |
21:56:57 | erikm | Fare: you don't have to know how the parameters map to the registers, the current driver calculates the hardware settings for you |
21:57:10 | Russ|werk | as well as some of the sa1100 source code |
21:59:02 | Fare | Russ: I do NOT have the spec sheet for the LCD panel. |
21:59:37 | Fare | I do not even know what LCD panel it is. |
21:59:56 | Fare | and I'm not proficient enough to rip the hardware apart then put it together again. |
22:00:45 | Russ|werk | ah, I had to pull apart my lcd panel/ts unit to figure out who made my panel |
22:00:50 | erikm | Fare: does the LCD work if you start blob? |
22:01:56 | Russ|werk | erikm: so will the new cpu-sa1100 revision just have the basic timings like cpu-sa1110? |
22:02:16 | erikm | Russ|werk: that's the plan, yes |
22:02:58 | Russ|werk | good |
22:03:17 | Fare | erikm: it sure works |
22:03:20 | Russ|werk | I'm also working on the tuxscreen ssp bus |
22:03:36 | Fare | and since the wince fb is at 0xc0003000, I can see the kernel loading at 0xc0008000 :) |
22:04:00 | Fare | and I replaced the "led" with writing to the framebuffer. |
22:04:04 | erikm | Fare: heh |
22:04:13 | Russ|werk | its using the motorola frame format, and reads work by writing a frame of zero's, and reading back simultaniously |
22:04:17 | Russ|werk | I don't want to wait around while this is happening, I just want an IRQ when its done |
22:04:35 | erikm | Fare: anyway, you can let blob dump the framebuffer registers |
22:04:42 | Fare | that's the easy part |
22:04:49 | erikm | Russ|werk: you want an irq in blob? |
22:04:56 | Russ|werk | reading the docs, it looks like I can't get an IRQ by just having one byte in the receive buffer |
22:04:57 | Fare | the difficult part is to retrofit the register values into linux parameters |
22:05:01 | Russ|werk | erikm: no, this is arm-kernel |
22:05:29 | Fare | (or else, I can have a stupid #ifdef JORNADA820 to prevent modifying anything but the DBAR1) |
22:05:47 | erikm | Russ|werk: ah. I have been thinking about a way to deliver irqs to blob. should help abraham and his broken watchdog |
22:07:08 | Fare | btw, I think I convinced seletz to integrate my patch when I explained him how piggy.c and my chain.S changes could help him produce bloated blob variants and get away with it. |
22:07:22 | Fare | :) |
22:07:56 | Russ|werk | erikm: ya, I don't need the phone part of the tuxscreen to work while its in blob |
22:08:09 | erikm | hah! seletz has to be careful: I'll see him at OLS and I *will* carry a LART! :) |
22:08:50 | Fare | erikm: don't be caught at the airport checkpoint with that bomb of yours. |
22:08:59 | Fare | "open your bag, sir" |
22:09:08 | Russ|werk | erikm: did you have anything to do with woody/disks-arm/current/lart? |
22:09:34 | erikm | Russ|werk: does that exist? |
22:09:35 | Fare | "now, tell us what this electronic controller is doing in the same case as this suspect-looking shampoo" |
22:09:53 | Russ|werk | erikm: ya |
22:10:07 | Russ|werk | erikm: there is a linux.bin, a root.tar.gz, and a driver.tgz |
22:10:09 | erikm | Russ|werk: I think it's wookey's work. afaik he's debian developer |
22:10:47 | Russ|werk | congrats to wookey then |
22:10:54 | Fare | oh, btw, for the ramdisk, should I extend piggy.c and stuff so as to have a load_ramdisk just like we have a load_kernel ? |
22:11:32 | Russ|werk | never thought about it that way... |
22:23:17 | erikm | BZFlag: when will you be arriving at OLS? |
22:25:42 | BZFlag | Tuesday night. staying at the westin |
22:27:53 | BZFlag | is there a place for posting travel plans and arranging BOFs etc? |
22:28:16 | erikm | BZFlag: here is your travel plan poster board: _rmk_! |
22:28:19 | BZFlag | ie: when is the blob/arm/tux/bzflag ;-) BOF at OLS? |
22:28:43 | erikm | BZFlag: don't forget the beer, or otherwise prpplague won't join :) |
22:28:45 | BZFlag | ponders DLing and printing _rmk_ |
22:29:07 | _rmk_ | bz: unfortunately, 24/25 are the kernel summit, which starts at 9am through to 10 or 11pm, so you may not see me until OLS. |
22:29:37 | BZFlag | open to party crashers? ;-) |
22:29:45 | _rmk_ | unfortunately not |
22:30:02 | BZFlag | not a spectator sport then. hmm. |
22:30:53 | erikm | BZFlag: bit like the selection of a new pope: white smoke == ok |
22:30:56 | prpplague | BZFlag: you going to get to attend OLS? |
22:31:01 | prpplague | BZFlag: cool |
22:31:02 | _rmk_ | Ted Ts'o can and will probably act as a bouncer. He's built well enough for that. 8) |
22:31:53 | Russ|werk | hey _rmk_ |
22:32:04 | _rmk_ | Hi Russ, you're missing out on OLS I hear. |
22:33:04 | Russ|werk | ya, I don't have the funds |
22:33:19 | Russ|werk | maybe next time |
22:33:38 | BZFlag | prpplague: yep. |
22:33:52 | prpplague | Russ|werk: should have said something, we could have gotten you sleep in the bath tub |
22:34:01 | erikm | really wonders if he should bring a suspicious looking LART this time |
22:34:17 | BZFlag | _rmk_: hmm. /me wonders if having met/talked to the bouncer is good or bad. ;-) |
22:34:22 | prpplague | BZFlag: great, i'm planning to spring for a couple rounds of brew for the arm BOF |
22:34:27 | Russ|werk | mainly its the air fare |
22:34:47 | Fare | I'm not made just out of (hot) air! |
22:35:19 | prpplague | lol |
22:36:00 | BZFlag | I'm just wondering how/where everyone is getting together for the various BOF/geekfests. |
22:36:16 | BZFlag | trying to maximize the geeking possibilities. ;-) |
22:36:18 | Fare | I'm going to switzerland :) |
22:36:24 | Fare | meeting geeks there |
22:36:33 | prpplague | BZFlag: lol, i'm torturing erikm and making him stay in the extra room |
22:36:39 | Fare | oops. Reminds me I have to contact more geeks who live there... |
22:37:23 | erikm | shudders |
22:37:35 | prpplague | erikm: did you get stopped at the airport last time? |
22:37:53 | erikm | prpplague: no, but the last time I brought a lart was before september 11 |
22:38:09 | erikm | prpplague: in december I didn't bring it on purpose |
22:38:39 | BZFlag | I'm planning to bring the laptop and a couple zauri all with 802.11 cards. |
22:38:49 | BZFlag | not planning on packing any tuxscreens |
22:38:51 | prpplague | BZFlag: cool |
22:39:03 | erikm | prpplague: otoh: they never really cared about a lart before that |
22:39:45 | prpplague | erikm: i went to barbados right after 9/11 and had a scsi raid cabinet, it took me 7 hours to get through security in dallas |
22:39:46 | erikm | prpplague: and in december they were only interested in my laptop, which I had to put separately through the scanner |
22:41:24 | prpplague | erikm: one of our employees got stopped coming back from canada because they thought that his laptop was fake because it didn't have windows running on it |
22:41:57 | erikm | prpplague: well, gnome sort of looks like windows |
22:42:12 | prpplague | erikm: true, but he was running init 3 |
22:43:33 | BZFlag | time to take that xp snapshot and replace your background image I suppose. |
22:44:06 | erikm | BZFlag: lol :) |
22:44:25 | erikm | well, my laptop still dual boots. have to remember to let it run windows when they ask |
22:44:51 | BZFlag | heh, that would mean a reboot. I always suspend/resume mine. ;-) |
22:45:19 | erikm | BZFlag: doesn't work with mine. ACPI prevents that |
22:46:11 | BZFlag | ugh. |
22:47:50 | Russ|werk | erikm: just get the xp themes |
22:48:16 | prpplague | BZFlag: not going to bring one of your new pda's with the fold out keyboard? |
22:48:19 | Russ|werk | and that palm tree background |
22:48:56 | erikm | Russ|werk: I think telling lilo to boot windows as a default will do the job. it will boot unattended into windows |
22:49:51 | Russ|werk | http://russ.dhs.org:8000/files/xpscreenshot.png |
22:50:07 | BZFlag | prpplague: if I have a non-developer board by then I'll bring it. I doubt I will though. |
22:50:20 | Russ|werk | I was bored |
22:50:34 | BZFlag | lilo? ugh. grub rox ;-) |
22:50:41 | Russ|werk | lilo is simple |
22:51:25 | prpplague | BZFlag: cool, i think seletz might have one of our "hack kits" available at OLS too |
22:51:57 | erikm | BZFlag: lilo does the job for me. has done it for seven years already or so |
22:52:10 | BZFlag | Russ|werk: nice "theme". Just hide the bar and it will easily fool most anyone. |
22:52:31 | BZFlag | erikm: I've been a grub convert for close to 4 years now. |
22:52:46 | erikm | Russ|werk: nice theme |
22:53:03 | Russ|werk | BZFlag: I was thinking about it, but then I wrote a cramfs loader for it |
22:53:03 | Russ|werk | ew |
22:53:14 | BZFlag | being able to boot a floppy and pick a kernel/initrd off anyones lilo-hosed machine is just a beautiful thing. ;-) |
22:53:33 | prpplague | lol |
22:53:47 | erikm | BZFlag: that's where the linuxcare bootable business card comes in :) |
22:53:56 | BZFlag | Russ|werk: didn't say the code was beautiful, just the capabilities. |
22:53:57 | rfs | wishes for a bootdisk with every possible disk repair tool included... |
22:54:32 | BZFlag | erikm: will it pick a kernel/initrd or does it just allow you to pivot_root etc. ? |
22:54:44 | BZFlag | rfs: start a project. ;-) |
22:55:14 | rfs | yarbt-0.1 here we come |
22:55:59 | BZFlag | or yabf-0.1 ? |
22:56:02 | erikm | BZFlag: it boots a standard kernel that tries to mount all filesystens it discovers |
22:56:37 | erikm | BZFlag: you can remount a filesystem rw, chroot into it and fix the problem |
23:00:16 | prpplague | heads home |
23:00:39 | prpplague | _rmk_: hope to see ya at OLS, and let me buy ya a beer as a thanks |
23:04:47 | erikm | goes zzz |
23:05:36 | Russ|werk | brb |
23:09:18 | erikm | see you guys at OLS |
23:35:27 | _rmk_ | vanishes |
23:36:05 | _rmk_ | see ppl at ols or maybe earlier |