01:41:30 | Sammy | morning all |
05:46:25 | Sammy | lunch... |
05:53:44 | amrutha | seletz: hello!! |
06:54:32 | amrutha | seletz:busy? |
09:23:23 | FM | l00 |
10:17:14 | skanda | erikm: I hope you got my mail |
10:42:08 | erikm | skanda: yes, I got your mail, but seletz already fixed the CVS version |
10:42:22 | erikm | skanda: (last thursday, IIRC) |
10:55:26 | Sammy | cool ...http://www.eink.com/iim/ |
11:12:21 | erikm | Sammy: *nod* |
11:12:35 | erikm | Sammy: btw, eink uses LART as well :) |
11:14:48 | Sammy | erikm: me **nod** too , and not only on LART :) |
11:33:50 | skanda | erik:as a test, suppose if we want to test the flash, how do we do it through software |
11:37:13 | seletz | hi ! |
11:37:43 | Sammy | yo seletz :) |
11:37:57 | seletz | hi sammy! |
11:38:27 | skanda | ok, lets say that I am manufacturing Assabets and need to determine if all the memory and the flash are working fine |
11:39:02 | skanda | I mean like a Power On Self Test, any suggestions how do I go about it |
11:40:46 | skanda | any one with any suggestions!!! |
11:40:56 | seletz | skanda: well, a POST routine would be nice. diag has quite some support for memory test, very good support IMHO (like memtest86). Lcd support is in too. |
11:41:53 | seletz | skanda: a POST routine would likely go into diag and call all available tests. That could be even done with erik's initcall feature. |
11:42:58 | seletz | skanda: for flash IMHI it would be best if you write a CRC32 routine, write a image to flash and calculate the CRC32 on your host (via the "sum" command) and then compare it to the assabed-generated crc. |
11:44:03 | amrutha | seletz:hai!! |
11:44:05 | seletz | skanda: then all you need is some support for the additional HW on assabet (audio, usb slave) if you really want to do a POST of all HW related things. |
11:44:15 | seletz | amrutha: hi ya! |
11:44:26 | skanda | actually I wanted to do a complete test on everything |
11:44:28 | amrutha | seletz:looks like you were busy!! |
11:45:02 | amrutha | seletz:finally got that assabet out of sleep :-) |
11:45:06 | seletz | skanda: those tests shoeld be hw-related, not assabet-related. so write the test functions for i.e. USB in general. |
11:45:28 | seletz | amrutha: yes, i were busy. got my ehernet chip toi work, now working on PCMCIA. |
11:45:32 | skanda | where exactly should I implement any tests for POST |
11:45:45 | seletz | amrutha: cool! What was the problem? |
11:46:08 | seletz | skanda: just add a command in diag for now. |
11:46:17 | amrutha | seletz: as usual some memory config..problem..now works fine... |
11:46:19 | skanda | ok |
11:46:36 | amrutha | seletz: but when i try to probe the sd_cke pin the process hangs. |
11:46:52 | seletz | amrutha: did you use blob? |
11:47:07 | amrutha | seletz:any suggestions...how concretely i can make sure that it has gone to sleep and woken up?? |
11:47:34 | amrutha | yes...the modified blob.. |
11:47:55 | seletz | amrutha: well, IIRC there is a power up reason register |
11:48:30 | seletz | amrutha: there are some bits that indicate the reset reason (HW reset, Sleep reset etc) |
11:48:46 | seletz | amrutha: hmmm, blob patches? |
11:48:58 | amrutha | seletz: yeah...that's all fine..i mean by probing the pins..i want to know.. |
11:48:59 | seletz | amrutha: which blob version? |
11:49:27 | amrutha | blob-1.0.8-pre2 |
11:49:32 | seletz | amrutha: you mean probing in HW from external HW? |
11:49:56 | amrutha | yes..u'r right. |
11:50:11 | seletz | amrutha: could you please try it with 2.0.5 and send patches? Please :) |
11:51:17 | seletz | amrutha: well, i'm not a hw guru, so i cant help you with that. Try asking erikm when he's back from lunch. |
11:51:42 | seletz | amrutha: (beeing a pure sw developer actually :) |
11:51:57 | amrutha | seletz: i shall try that out with 2.0.5 and get back to u.. :-) |
11:52:38 | amrutha | seletz: my next POA is to import this kernel onto my own board..but i do not have a Jtag interface...:-) |
11:52:48 | skanda | seletz: i think you are trying to say that I should implement my POST in my __initcall function, am i right |
11:53:54 | skanda | 8 |
11:54:10 | skanda | 6well |
11:54:17 | seletz | skanda: first do it in a simple command. Then try it out and make it working. _then_ we think about __initcall :) |
11:54:47 | skanda | how do I do that please |
11:55:07 | seletz | amrutha: hmm, no need for jtag, just upload it with blob over serial line and flash it. Whats the problem with that? |
11:55:49 | seletz | skanda: get 2.0.5 cvs blob version. Look into blob/src/diag. |
11:56:33 | skanda | yup, I am looking at it right now |
11:56:43 | skanda | is it in commands.c |
11:56:45 | amrutha | but that shud be permanent blob ..right?? i am just bothered if that blob does not work.. :-( |
11:56:46 | seletz | skanda: system3.c ist for ex. my personal testing field. |
11:57:23 | skanda | right, I got it |
11:57:28 | amrutha | seletz: how do i get the cvs version of blob 2.0.5?? i shall check my assabet with it.. |
11:57:35 | seletz | amrutha: why? Np if you see some otput and can issue commands. |
11:58:08 | seletz | amrutha: have linux running and internet connection? :) |
11:58:20 | amrutha | yes..pretty much.. |
11:58:30 | seletz | amrutha: fire up a bash and issue: |
11:59:08 | seletz | amrutha: on moment pls |
11:59:22 | amrutha | yeah..sure.. |
12:00:04 | seletz | amrutha: ok, fire netscape up and look at "http://sourceforge.net/projects/blob/" |
12:00:30 | seletz | amrutha: follow the cvs link. there is a description. |
12:00:42 | seletz | skanda: skimmed system3.c |
12:00:50 | skanda | yup, |
12:01:22 | skanda | but I am having trouble understanding certain things, like support for SA1111 |
12:01:41 | skanda | problem is I do not have SA1111 companion chip in my Assabet |
12:02:01 | seletz | skanda: nah, dont try to understand that yet. Yust wanted you to get some idea. Look at how commands are added. |
12:02:38 | skanda | ok, i think I got a fair idea about it |
12:03:04 | seletz | skanda: look for ex at the regs_show() function. |
12:03:52 | seletz | skanda: basically every command gets (argc, argv) as in a c_main(). Then you add a help string and the __commandlist() macro. See? |
12:04:04 | skanda | will do that, just hold on for a sec please |
12:04:24 | seletz | skanda: line 275 |
12:04:57 | skanda | am seeing that |
12:06:10 | seletz | skanda: ok. Now, "system3.c" is conditionally compiled into diag when blob is configured with "--with-board=system3". So, to add support in diag for assabet you'll have to add an "assabet.c". fine? |
12:07:08 | seletz | skanda: in assabet.c you can basically do everything you want. Add there a command "post" or similar. re-use the memtest routines from blob. |
12:07:18 | skanda | so you mean to say..when i configure the file with--board assabet...shud i add with--board=system3 too? |
12:07:24 | skanda | ok |
12:08:01 | seletz | skanda: no. either you have a system 3 board or an assabet board. not both :) |
12:08:14 | skanda | ok |
12:09:14 | Sammy | seletz: this let me remember something , I still don't send the puppy diff to you .... |
12:09:21 | seletz | skanda: later, if your post is working, we'll eventually add some board-independent facility to support POST calls. |
12:09:53 | skanda | okie...when i have assabet board..to add system3.c .....where do i add that?? |
12:10:05 | seletz | skanda: actually ther _is_ such a facility already. Its just not as of now used for POST things. |
12:10:23 | skanda | wheres that |
12:11:17 | seletz | skanda: ok, look at Makefile.am in the src/diag dfir. |
12:11:45 | seletz | skanda: blob uses autoconf/automake for automagically generating makefiles. |
12:11:53 | skanda | ok |
12:12:12 | seletz | skanda: dont try to understand all of it yet. Just get a basic idea. :) |
12:12:44 | skanda | so instead of system3.c, i put assabet.c in the Makefile and rename sy..3.c to assabet.c, |
12:12:53 | skanda | am i correct |
12:13:01 | seletz | skanda: NO. absolutely NO. |
12:13:10 | seletz | skanda: :) |
12:13:12 | skanda | well, guess i am lost again |
12:13:36 | seletz | skanda: one moment pls. figuring out a good expl. |
12:13:56 | skanda | thanxx...a million |
12:14:02 | amrutha | seletz: his doubt is "where shud he add assabet.c" if he is using another board? |
12:14:34 | seletz | :!ls |
12:14:39 | seletz | (wrong window :) |
12:15:51 | seletz | skanda: ok, look at configure.in in the base dir. |
12:16:10 | Sammy | skanda: you should see this http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/blob/blob/doc/porting.txt?rev=1.1&content-type=text/vnd.viewcvs-markup |
12:16:22 | seletz | skanda: this is the main config file for configure and automake/autoconf. From this definition makefiles are generated. |
12:16:26 | skanda | ok |
12:16:42 | skanda | right i understand |
12:17:00 | Sammy | that's the example for ipaq, and the same for assabet ... |
12:17:19 | seletz | skanda: yes, porting.txt is fine too. To get an idea how blob is organized for each platform. Doh, cold have thought of it sooner :) |
12:18:01 | seletz | skanda: but while you are at it, look at the case construct, at system3) |
12:18:32 | skanda | ok |
12:19:42 | seletz | skanda: there look at the line DIAG_PLATFORM_OBJ="system3.o" |
12:20:19 | seletz | skanda: that's how i tell the blob config system to actually compile system3.c into diag for my Board. |
12:21:03 | seletz | skanda: so, for assabet you have to add a line similar to that, just changing the name of the source you want to add. |
12:21:09 | skanda | but I am unable to get DIAG_PLATFORM_OBJ="system3.o |
12:21:18 | skanda | i am unable to find it anywhere |
12:21:27 | skanda | is this in the CVS version |
12:21:43 | seletz | skanda: then do a configure again ant _zonk_ your code will get compiled into diag. Fine, isn't it? |
12:22:15 | seletz | skanda: hmmm. |
12:22:22 | skanda | wait pls |
12:22:23 | seletz | hi erik! |
12:22:26 | erikm | hi seletz |
12:23:26 | seletz | skanda: found it? |
12:24:52 | erikm | seletz: I'm going on a one and a half week holiday this friday |
12:25:33 | erikm | seletz: I'm going to visit a friend in New York, so I have a 6 hour transatlantic flight. nice for writing documentation :) |
12:25:41 | seletz | erikm: cool! |
12:26:03 | seletz | erikm: well, flights to NY are quite cheap these days ... |
12:26:40 | erikm | seletz: yep, I booked mine less than two weeks ago for about 340 euro |
12:26:50 | seletz | erikm: something to do until then, to get blob into a nice form? Help needed? |
12:27:04 | seletz | erikm: Now, _that_ is cheap. |
12:27:43 | seletz | erikm: have a laptop that surrives 6 hrs? |
12:27:52 | erikm | seletz: no, it survives four hours |
12:28:28 | erikm | seletz: but that should be good enough. there's also lunch, and they even might have a nice movie |
12:28:51 | seletz | erikm: well, sounds fun :) |
12:29:39 | erikm | seletz: actually, it's cheaper than last summer, where I booked more than a month in advance for about 450 euro |
12:30:30 | skanda | got it at last |
12:30:38 | skanda | was searching the wrong file till now |
12:32:39 | seletz | erikm: Visiting NY often, do you? |
12:33:58 | seletz | man, my kernel oopsed just again. sigh. |
12:34:18 | seletz | staring at my code ... |
12:35:55 | erikm | seletz: well, I have a friend who lives over there. summer 2000 I was there for holiday, januari 2001 for the intel strongarm linux summit, this summer again for holiday |
12:36:21 | seletz | erikm: intel strongarm linux summit? Regular happening? |
12:37:12 | erikm | seletz: and now my boss complained that I had to get rid of the huge pile of days off. that friend jost got a new house which needs some work, so that's a nice goal to visit again |
12:37:42 | erikm | seletz: no, that was the first one. it was nice, I met a lot of people over there I only knew from mailing lists |
12:38:48 | seletz | erikm: gee, i just started in april with this arm/linux buisness. sigh. Hope they'll have another summit. |
12:40:09 | erikm | seletz: well, it was by invitation only. there were about 25 people, and they had a lot more people who'd like to attend |
12:40:38 | seletz | erikm: well, ok :) Got the message :) |
12:41:32 | seletz | erikm: actually :( |
12:42:19 | erikm | seletz: nico and me shared a hotel room. we changed it into a hackers lab cause we wanted to have linux running on the XScale so we could show it on linuxworld expo |
12:42:41 | erikm | seletz: yeah, intel had do disappoint quite some people |
12:43:01 | erikm | seletz: they really didn't expect so many people |
12:43:36 | seletz | erikm: ah, it was sponsored by intel. I see. |
12:44:12 | seletz | erikm: got it running? XScale not avaliable up to now, is it? |
12:45:45 | seletz | afk, bbl |
12:49:25 | erikm | seletz: we got it running |
12:49:50 | seletz | back |
12:50:05 | erikm | seletz: see http://linuxpr.com/releases/3363.html |
12:50:35 | seletz | erikm: cool. |
12:50:37 | erikm | seletz: that was only the initial version. nico had been working on it for quite some time, but couldn't get it working |
12:50:49 | seletz | erikm: hmm. Are you working for montavista? |
12:50:55 | erikm | seletz: no. but nico is |
12:51:21 | seletz | erikm: ah. i see. |
12:52:14 | erikm | seletz: we had a late night debugging session. we started at 22:30, at 2:00 we got a bash prompt, and at 4:00 we even had a nice demo |
12:52:24 | seletz | erikm: lots of buzzwords in there, tough. |
12:52:47 | erikm | seletz: it's a press release, what do you expect? |
12:53:14 | seletz | erikm: well, its always like that when getting someting hacked up for demos and such, isn't it? |
12:54:18 | seletz | erikm: you strarted not from scratch, did you? Took SA1110 as a base? |
12:55:17 | erikm | seletz: nico did it. he had to create a new arm architecture specific directory |
12:55:32 | erikm | seletz: here's the news with less buzzwords: http://slashdot.org/article.pl?sid=01/02/03/2328241&mode=thread |
12:57:58 | seletz | erikm: well, ok. But have it working in 4 Hrs is, erm, fast. Then hacking up a demo (what demo, btw?) |
12:58:27 | seletz | erikm: you guys must be thinking in C, don't you? |
12:58:33 | seletz | erikm: :)) |
12:58:40 | erikm | seletz: although the xscale is a member of the arm family, it's sufficiently different from the sa1110. especially because this particular core+support chip has PCI, so he had to setup PCI as well |
12:59:33 | erikm | seletz: the demo was just a hacked up ramdisk with enough "xscale" and "montavista" in it. it didn't do anything special besides running the lmbench benchmark |
13:00:35 | seletz | erikm: LOL! So demo == demo, as everywhere..... even in linux :) |
13:01:07 | Sammy | move body to home |
13:01:30 | erikm | seletz: demos are always the same. as long as it looks good, it is ok |
13:01:52 | seletz | erikm: actually one of _tis_ demos caused my boss tho think "well, linux runs on these boards. Ok, lets get some hw hacked together and we're into market in a minute." |
13:01:56 | Sammy | guys , see all of you later :) |
13:02:03 | seletz | later sammy... |
13:02:20 | erikm | bye sammy |
13:03:15 | erikm | seletz: actually porting linux to an already supported architecture is quite easy |
13:04:27 | seletz | well, ok. Yes. That's the reason even i managed to get my board running. sort of. |
13:12:49 | erikm | cool, somebody with waaaay to much free time: telnet 62.250.7.101 |
13:14:16 | seletz | erikm: hmm, no connection via telnet |
13:15:16 | erikm | seletz: it works for me. it's a star wars movie. in ascii. |
13:16:00 | seletz | erikm: _what_? in ASCII? oh my god, that needs some tome .... :) |
13:27:00 | erikm | walks of to do some soldering, bbl |
13:31:40 | erikm_afk | seletz: just try a couple of times until you get "Original Work" |
13:33:40 | seletz | _sloooowwwww_ |
16:00:07 | sammy_wms | how long is that movie ? |
16:02:36 | sammy_wms | 5 minute pass ...maybe that's all movie not just a part :) |
16:02:49 | sammy_wms | cooool |
16:18:32 | seletz | heading home |