**** BEGIN LOGGING AT Sun Nov 06 02:59:59 2005 Nov 06 07:09:39 vmaster: there ? Nov 06 07:09:55 hey key2 Nov 06 07:10:04 ok Nov 06 07:10:07 just a lil explanation Nov 06 07:10:12 for the ARM Nov 06 07:10:32 i do a gcc myfile.c so I get a a.out then I use objcompy to make the a.out >> a.hex ? Nov 06 07:10:40 or should I first make an elf Nov 06 07:10:49 from the .s file ? Nov 06 07:11:18 it's easier to use -c to build intermediate object files, then link them together with ld Nov 06 07:11:39 ok Nov 06 07:11:47 so I have my toto.o Nov 06 07:11:50 what do I do with it ? Nov 06 07:11:57 either use a linker script, or give section start addresses as arguments to ld Nov 06 07:12:20 like: ld toto.o -o toto -Ttext=0x40000000 Nov 06 07:12:24 what's the start 0x0 ? Nov 06 07:12:39 0x4000000 is the beginning of the ram? Nov 06 07:13:03 on the lpc2xxx, internal sram starts at 0x40000000, yeah Nov 06 07:13:24 for more complex setups (text, data, bss...) use a linker script Nov 06 07:13:38 lp2106.ld / Nov 06 07:13:38 ? Nov 06 07:14:11 if that file is a GNU ld linker script, yeah Nov 06 07:15:09 ok Nov 06 07:15:58 then I go from the .o to .hex ? or .elf to .hex ? Nov 06 07:16:58 from the linker output to .hex Nov 06 07:17:14 ok Nov 06 07:17:19 then for uploading it to the module ? Nov 06 07:17:24 do I need like a max232 ? Nov 06 07:17:26 or jtag ? Nov 06 07:17:48 do you want to build the board yourself? or do you want to use a prebuilt board? Nov 06 07:18:01 I wanna use the very little board Nov 06 07:18:04 http://www.olimex.com/dev/lpc-h40.html Nov 06 07:18:37 but there are no max232 on it and I don't see either the jtag port, I have a max232 tho so I could connect it to it Nov 06 07:19:02 how do you want to interface that board do your host pc? Nov 06 07:19:14 I'll use a FTDI chip Nov 06 07:19:19 USB Nov 06 07:19:48 that's what the larger olimex boards use - connect a FT232 to the serial port, and you can use that for both programming and communication Nov 06 07:20:04 yeah but ft232 is slow Nov 06 07:20:13 I need parallal Nov 06 07:20:47 the ft2232 has two parallal port, I might even use both of them to be faster Nov 06 07:20:57 If I can read fast enough from the jtag, then I'm fine Nov 06 07:22:39 all I need to know right now is if the ARM is fast enough for reading/writing to the jtag Nov 06 07:23:42 or if I'll end up using a fpga Nov 06 07:28:38 mhh, the ft232 allows up to 3mbaud - i doubt you're going to achieve a lot more than that Nov 06 08:22:22 do I have to write my own driver for rs232 on the ARM or is in build in ? Nov 06 08:24:48 there should be examples for lpc uart drivers Nov 06 08:24:59 well Nov 06 08:25:08 would be the same probleme if I wanted to bitbang at 3mb Nov 06 08:25:17 I don't use the port of the lpc Nov 06 08:25:17 hmm? Nov 06 08:25:27 so why not just putting a ft2232 Nov 06 08:25:32 with both port Nov 06 08:26:05 the interesting part of it is that it has 1MB of ram Nov 06 08:26:28 so I could eventually load in the ram big packets Nov 06 08:26:37 and then send them back to the usb Nov 06 08:26:46 but I'm more or less familiar with the ft2232 Nov 06 08:26:48 so I'll see Nov 06 08:27:15 otherwise, I have one of those board: http://www.fpga4fun.com/board_dragon.html Nov 06 08:27:28 vmaster: do you think It would be fast with that ? Nov 06 08:29:30 i guess you could make it fast - but if you want to implement the whole ejtag functionality inside the fpga, it's probably going to become quite complex Nov 06 08:30:09 vmaster: do you know where I could buy just the LPC1206 ? Nov 06 08:30:14 just the chip Nov 06 08:31:23 hmm, sorry, don't know much about distributors Nov 06 08:31:39 do you have anotherwise an idea of the retail price ? Nov 06 08:31:59 If I wanted to buy not big quantities Nov 06 08:34:21 Price: Nov 06 08:34:22 250 - $10.8076 Nov 06 08:34:27 where ? Nov 06 08:34:29 url ? Nov 06 08:34:30 avnet Nov 06 08:34:46 not sure if you can order as a private person Nov 06 08:35:13 a german online shop has it for 20 euros Nov 06 08:35:37 which seems extremely overpriced... Nov 06 08:36:24 I don't see the one on avnet Nov 06 08:36:31 you can send the link ? Nov 06 08:37:12 go to advanced part search, microcontrollers, philips, family lpc2000 arm7 Nov 06 08:48:58 yeah Nov 06 08:49:07 the pb is that they won't sell less than 250 Nov 06 08:49:45 i checked some german stores that sell single quantities - they charge 20-30 euros Nov 06 08:51:54 yeah Nov 06 08:51:56 same in paris Nov 06 08:52:03 so $10 is not bad Nov 06 08:52:37 dunno if jtag support 3.3v Nov 06 08:52:42 otherwise I need to put a buffer Nov 06 08:53:35 key2: JTAG uses whatever I/O voltage your target has Nov 06 09:09:05 ok Nov 06 09:09:10 my target is in 5v Nov 06 09:09:19 so I gues I'll need to convert the 3v >> 5 Nov 06 09:15:09 the lpc pins are 5v tolerant, iirc Nov 06 09:35:51 ~seen beewoolie Nov 06 09:35:58 beewoolie was last seen on IRC in channel #openjtag, 38d 22h 30m 58s ago, saying: 'Right. I get that part. My block flashing code is way faster than the byte flashing.'. Nov 06 09:35:59 ~seen bewoollie Nov 06 09:36:00 i haven't seen 'bewoollie', prpplague Nov 06 09:37:52 ~seen beewoolie Nov 06 09:37:53 beewoolie was last seen on IRC in channel #openjtag, 38d 22h 32m 53s ago, saying: 'Right. I get that part. My block flashing code is way faster than the byte flashing.'. Nov 06 09:44:08 anyone here doing apex dev? Nov 06 09:44:26 [g2] would know. Nov 06 09:56:25 <[g2]> would know what ? Nov 06 09:57:15 <[g2]> prpplague what's up with APEX ? Nov 06 10:01:05 [g2]: APEX the fpga ? Nov 06 10:01:16 apex the bootloader, i guess Nov 06 10:06:52 [g2]: just wondering the status of it Nov 06 10:07:04 [g2]: considering using it for a new project Nov 06 10:08:55 [g2]: only problem i have with apex is the cli Nov 06 10:11:41 <[g2]> send a message to bewoolie Nov 06 10:12:11 [g2]: hehe one step ahead of ya Nov 06 10:12:30 [g2]: you use apex? Nov 06 10:12:52 <[g2]> many months ago I was using it for may fatslug Nov 06 10:13:09 <[g2]> I've got my own boards now Nov 06 10:13:29 [g2]: which bootloader are you using now? Nov 06 10:14:27 <[g2]> Redboot V2.01 Nov 06 10:14:41 ahh Nov 06 10:15:09 <[g2]> with CF I load ... -m disk hda1:/boot/zImage ... Nov 06 10:15:21 <[g2]> CF/microdrive Nov 06 10:15:31 [g2]: i've been writing my own for most of our projects, its just getting old Nov 06 10:15:52 <[g2]> nod, having a base to work from is nice Nov 06 10:16:10 <[g2]> it's really a time versus function issue Nov 06 10:16:19 <[g2]> dev time versus what one needs Nov 06 10:16:43 [g2]: yea, our main problem has been finding something with the specs we need Nov 06 10:17:02 <[g2]> nod. Nov 06 10:17:02 [g2]: apex appears to have reached a point where it has it Nov 06 10:17:08 <[g2]> cool Nov 06 10:45:19 [g2]: ugh, apex's command line is soooooooooo cryptic Nov 06 11:46:05 wonder where bewoollie has been Nov 06 13:32:42 * prpplague wonders if bewoollie is alive Nov 06 13:38:59 hehehe :) Nov 06 13:39:23 prpplague: yo. Nov 06 13:40:33 hey beewoolie-afk Nov 06 13:41:03 lennert: did that reboot fix ever get resolved? Nov 06 13:41:13 beewoolie-afk: which reboot fix? Nov 06 13:42:01 I was in a convo with RMK and someone else (name escapes me) about how the IXP42x wouldn't reliably reboot. It seems to be due to the fact that flash isn't in the array mode. Nov 06 13:42:33 ixp4xx has no reset output for when the watchdog triggers Nov 06 13:42:45 kind of hard to fix Nov 06 13:43:19 Except that it isn't because we're using the watchdog to reset the system from a controlled location. Nov 06 13:43:37 ? Nov 06 13:43:37 The fix is to get the MTD/NOR flash code to put the flash into array mode on shutdown. Nov 06 13:43:48 yes.. Nov 06 13:43:51 There is a callback or it. Nov 06 13:44:00 s/or/for/ Nov 06 13:44:08 i didn't follow that particular discussion Nov 06 13:44:12 are you on linux-arm-kernel? Nov 06 13:44:31 Yeah. Nov 06 13:56:22 beewoolie-afk: hey Nov 06 13:56:39 beewoolie-afk: whats cookin? Nov 06 13:57:03 beewoolie-afk: getting ready to send you an email with some patches for my lh79520 based device Nov 06 14:27:43 prpplague: Hey! Nov 06 14:27:46 OK. Nov 06 14:28:06 beewoolie-afk: i'd like to start doing some apex dev Nov 06 14:28:07 I'm doing a purge of old gear. I'm kinda far from a console while I'm working. Nov 06 14:28:12 KO. Nov 06 14:28:18 beewoolie-afk: np Nov 06 14:28:22 I've started on the MMC driver. Nov 06 14:28:23 FYI. Nov 06 14:28:30 beewoolie-afk: yea i noticed Nov 06 14:28:42 I'm not having problems getting the card ID register. Nov 06 14:28:43 beewoolie-afk: alreadys started hacking around with it Nov 06 14:29:00 beewoolie-afk: i'm gonna be adding samsung s3c2410 support next week Nov 06 14:29:14 Yeah. It's in the latest APEX, but only in the exploration phase. Nov 06 14:29:23 What is the core on the s3c2410? Nov 06 14:30:05 Arm7 or Arm9? Nov 06 14:30:15 920t Nov 06 14:30:28 yea Nov 06 14:31:19 beewoolie-afk: just about done testing the patch Nov 06 14:32:29 beewoolie-afk: there is only one item that isn't complete, we've not registered the board yet as we don't want ppl getting wind of the board as of yet Nov 06 14:58:17 Yo G Dawg. Nov 06 14:58:49 beewoolie-afk: ok, patch is otw Nov 06 14:59:02 beewoolie-afk: i'm soon to head home for the day, so just drop me an email Nov 06 14:59:37 prpplague: k Nov 06 14:59:42 dyoung-zzzz: Hey man! Nov 06 14:59:56 beewoolie-afk: What new and happening ? Nov 06 15:00:23 beewoolie-afk: i've added apex to buildroot, should be submitting the patch for that tomorrow Nov 06 15:00:37 dyoung-zzzz: I'm cleaning out some old gear. I think I only need one extra CD rom drive, etc. Nov 06 15:00:49 prpplague: Nice. Nov 06 15:02:06 beewoolie-near: I'm still keeping my eyes open for a decent logic analyser, but so far no luck... Nov 06 15:02:11 beewoolie-near: i'm gonna need to add some sort of usb support to apex Nov 06 15:02:55 gotta run later Nov 06 15:02:58 dyoung-zzzz: I really wish that the USB devices were fast enough. Nov 06 15:03:00 prpplague: later. Nov 06 15:03:08 beewoolie-near: apex sure has come a long way, looks realllllly good Nov 06 15:03:15 beewoolie-near: let me know about the patch Nov 06 15:03:22 prpplague: thx. Nov 06 15:03:34 beewoolie-near: Yeah, that one z-foo branded thing looks nice. Nov 06 15:03:38 prpplague: BTW, did you notice that I added thumb support? Nov 06 15:03:58 dyoung-zzzz: And the problem is...? Nov 06 15:04:02 I fond one on ebay for like 1/3 price new; was thinking about it for low-bandwidth stuff.... Nov 06 15:04:06 beewoolie-near: yea noticed that, plan to test it with the juicebox when i have time Nov 06 15:04:17 no problem; its just slow ! Nov 06 15:04:23 prpplague: I haven't solved the call-by-r3 problem. Nov 06 15:04:31 But it does build. Nov 06 15:04:38 cool Nov 06 15:04:42 beewoolie-near: later Nov 06 15:04:43 dyoung-zzzz: Yeah. Slowness is a problem. Nov 06 15:05:13 dyoung-zzzz: I could use one right now. In fact, I don't need speed as the signals are only 300KHz. Nov 06 15:07:57 http://cgi.ebay.com/New-895-USBee-ZX-PC-Logic-Analyzer-Signal-Generator-NR_W0QQitemZ7560682243QQcategoryZ1504QQssPageNameZWDVWQQrdZ1QQcmdZViewItem Nov 06 15:08:00 and Nov 06 15:08:03 http://www.usbee.com/zx.html Nov 06 15:09:25 Still seems to be too expensive, but I'm a super cheapskate. Nov 06 15:12:00 Yeah, it is pricy considering theres only a couple chips in it. Nov 06 15:23:35 dyoung-zzzz: What's been up with you? Nov 06 15:24:16 Nothing cool. Nov 06 15:24:24 Just the same mundaneness Nov 06 15:24:51 although I have bee partcipating in the debian armeb port. :-) Nov 06 15:24:56 Do you need a good tsunami to get you out of the rut? Nov 06 15:25:03 Nice. Nov 06 15:28:00 I'm finding it a bit disappointing that the SE Asian Tsunami sense array is being largely politicised; with no real hope of ever providing early warning. Nov 06 15:31:20 Odd. Wouldn't it be great if the data from the array were simply available on the internet? Then anyone could monitor it. Nov 06 15:32:20 Would be neat but it wouldn't be very useful Nov 06 15:33:14 Why is that? Nov 06 15:33:40 (Hi T.) Nov 06 15:33:53 If J. Random User sees that something funky is going on. Who does he/she tell? Nov 06 15:33:59 Hey. Long time no see Nov 06 15:34:02 That's not what I mean. Nov 06 15:34:06 :-) Nov 06 15:34:11 ah :) Nov 06 15:34:24 If the data is public, then the analysis can be done by any organization interested in it. Nov 06 15:34:37 *THAT* organization's reputation would be the basis of a warning. Nov 06 15:34:59 Public data means that it is auditable. Nov 06 15:35:04 hmm Nov 06 15:35:40 There were some rumblings earlier this year from weather 'vendors'. They were mad that the US gov publishes the weather data for free. Nov 06 15:35:58 They felt it was preventing them from ... making money on public information..?! Nov 06 15:36:02 They never explained. Nov 06 15:36:10 I'm glad that the data is public. Nov 06 15:37:45 It'll be difficult to do what you say just because of those reasons unfortunately. Whoever does it isn't going to want everybody keeping track of what they're doing. Nov 06 15:40:28 On a similar note, a breathalizer manufacturer in the states refused to give up the code for their machine. Trouble is, the judge agreed with the defendant that there ought to be an opportunity to question the accuer. The judge is likely to dismiss the case. Nov 06 15:41:33 Oh, and I didn't say that I thought the tsunami sensor array data would be made public. I just thought that it would be a good idea, for the people. Nov 06 15:41:47 Yeah. I know :) Nov 06 15:42:07 Hasn't the breathalyser source been turned over to a "expert" third party? Nov 06 15:42:25 At least that was in the short article I read the other day Nov 06 15:42:39 Nope. Last I heard was that the manufacturer refused that tack. Nov 06 15:42:42 let me see... Nov 06 15:43:33 http://www.freedom-to-tinker.com/?p=914 Nov 06 15:44:36 http://news.zdnet.com/2100-1040_22-5931553.html Nov 06 15:44:47 ah okay. I guess it was what the judge was proposing Nov 06 15:47:03 I seem to recall that the manufacturer balked at the court order, but I'm not seeing it any more. Nov 06 15:47:27 I heard that someone is making a browser that indexes everything that a user reads. That would help me find things I've read. Nov 06 15:48:51 Suggest it to Google :) Nov 06 15:59:46 ;-) Nov 06 16:01:33 Oh the data is all public. Nov 06 16:02:03 dyoung-zzzz: Is it live? Nov 06 16:02:10 Almost. Nov 06 16:02:28 How does one get it? Nov 06 16:03:07 What I meant by that earlier statement is that the governments in the area want to have these sensors to say to their neighbor they have it; but will not fork out for the infrastructure for the actual warning systems. Nov 06 16:04:25 The data is raw and uncalibrated; but you can get it by PID from the satellite download place, or slightly processed at http://uhslc.soest.hawaii.edu . Nov 06 16:08:24 Interesting. So it looks like it would be possible to code something from the raw data. Nov 06 16:08:40 I was only seeing plots and MPEGs. Nov 06 16:08:46 Yep, thats my day job. Nov 06 16:08:57 oh, in RSL Nov 06 16:09:14 What did the data look like last December just before thee wave? Nov 06 16:09:38 ah, okay its not that realtime. Nov 06 16:09:55 there are only a few minutes of transmission slots per hour. Nov 06 16:10:06 so its as realtime as it can be. Nov 06 16:10:17 The newer stuff wil be BGAN for truly realtime. Nov 06 16:14:57 Ah. OK. That's the part that is crucal. Nov 06 16:21:26 hi, on my slug, apex boots almost perfectly into OpenSlug-2.7-beta, except that the kernel does not detect ttyS1 (ttyS1 works under redboot/same kernel) Nov 06 16:21:41 ideas, anyone? Nov 06 16:22:52 cegg: The console is ttyAM0. Nov 06 16:23:13 cegg: let me verify that... Nov 06 16:24:49 cegg, are you talking about the second serial port? Nov 06 16:25:00 beewoolie-near: yes Nov 06 16:25:18 I've never look at that. It's possible that APEX needs to initialize something on that port. let me check. Nov 06 16:27:00 <[g2]> hey cegg Nov 06 16:27:06 <[g2]> I just posted on nslu2-linux Nov 06 16:27:12 Hey [g2]. Nov 06 16:27:13 <[g2]> the IRC channel that is Nov 06 16:27:19 <[g2]> HEY beewoolie-near Nov 06 16:27:28 [g2]: oh, i didn't check Nov 06 16:27:37 <[g2]> cegg NP Nov 06 16:27:41 <[g2]> just FYI Nov 06 16:28:03 the watchdog works, that was my mistake Nov 06 16:28:43 <[g2]> np Nov 06 16:29:11 cegg, same kernel in both cases, right? Nov 06 16:29:18 <[g2]> cegg are you familiar with the nslu2 kernel specific stuff Nov 06 16:29:57 [g2]: well, not Nov 06 16:30:21 <[g2]> you're building from sources right ? Nov 06 16:30:36 [g2]: thats right Nov 06 16:31:24 <[g2]> look at nslu2-setup.c or nslu2-misc in the openslug kernel under tmp Nov 06 16:31:45 <[g2]> so beewoolie-near how they he** are you dude! Nov 06 16:32:15 Been out of town a lot lately. Nov 06 16:32:22 <[g2]> good stuff ? Nov 06 16:32:26 All of it! Nov 06 16:32:30 <[g2]> excellent Nov 06 16:32:32 [g2]: That's fixed already. That (2.6.12.2/openslug 2.7beta) kernel works with redboot already Nov 06 16:32:42 Last week, doll-face and I went to Vancouver and victoria. Nov 06 16:32:52 <[g2]> you know irc works from different places :) Nov 06 16:33:04 Only if I have computer access. Nov 06 16:33:06 :-) Nov 06 16:33:12 <[g2]> hey ski season should be getting kicked off soon Nov 06 16:33:35 I think that some of the lifts just opened in Washington. Nov 06 16:33:43 <[g2]> when's doll-face gonna make an honest man outta ya ? :)) Nov 06 16:34:10 :-) Nov 06 16:34:22 Interesting tidbit. The slug has allocated interrupt 15 but not 13. Nov 06 16:34:34 15 is for the 'high-speed' uart and 13 is for the console. Nov 06 16:34:41 Are we using the console uart at all? Nov 06 16:34:47 [g2] cegg mentioned that it the 2nd serial port worked under redboot just fine. So that means the kernel setup is right; and I can verify that as it was my commmit. Nov 06 16:34:48 [g2]: But not under apex. Just omits the "ttyS1 at MMIO 0xc8001000 (irq = 13) is a XScale" Nov 06 16:34:58 <[g2]> I hear skiing up north is good as long as dyoung-zzzz isn't there (he warms the place up too much) Nov 06 16:35:04 beewoolie-near: yes. same kernel. Nov 06 16:35:27 I'm going to Banff next next week. :-) Nov 06 16:35:35 beewoolie-near: both loaded with xmodem and checksummed before exec Nov 06 16:36:20 <[g2]> so beewoolie-near I've got units :) Nov 06 16:36:26 The x1205 is filling my kernel message log with garbage, so I cannot see the startup messages. Nov 06 16:36:32 [g2]: DUUDE! Nov 06 16:36:51 dyoung-zzzz: Bring the woolies! Nov 06 16:37:31 cegg: what does /proc/iomem show when you boot from redboot? Nov 06 16:37:47 beewoolie-near: second.. Nov 06 16:38:13 <[g2]> so beewoolie-near I've been playing with ebtables a little and doing bridging Nov 06 16:38:53 ...and? Nov 06 16:39:09 <[g2]> well the bridging is rock'n Nov 06 16:39:13 I'm still not finding anything in the HW to enable the serial ports.... Nov 06 16:39:22 <[g2]> runs nearl line rate with 60 Nov 06 16:39:28 <[g2]> % of CPU left Nov 06 16:39:29 [g2]: You've got two interfaces and want a transparent firewall? Nov 06 16:39:37 <[g2]> I've been running that Nov 06 16:40:22 <[g2]> I've been playing with both transparent and non-transparent Nov 06 16:40:57 <[g2]> my scp x-fers were clocking 11.0+MBs when the CPU was only 40% rolling Nov 06 16:41:17 <[g2]> I thinks if I switched to polled it'd be a lot lower Nov 06 16:41:29 beewoolie-near: last lines:c8000000-c8000fff : serial8250.0 Nov 06 16:41:29 c8000000-c800001f : serial Nov 06 16:41:29 c8001000-c8001fff : serial8250.0 Nov 06 16:41:31 c8001000-c800101f : serial Nov 06 16:42:09 cegg: which kernel version are you using? Nov 06 16:42:25 <[g2]> cegg My guess is that Redbot (as it mostly configures an IXP425 dev board) configs both serial ports Nov 06 16:42:27 beewoolie-near: 2.6.12.2 from openslug2.7beta Nov 06 16:42:55 <[g2]> cegg I don't think the kernel specifically sets up the second, but doesn't disable it either Nov 06 16:42:58 OK. I'm using 2.6.11.2 Nov 06 16:43:37 <[g2]> So I'd guess that APEX just isn't setting up the second Nov 06 16:44:30 I switched to apex because redboot hangs sometimes, when sth. is connected to rxd1 Nov 06 16:44:38 [g2] I told it specifically to set it up. Nov 06 16:44:46 cegg, isn't it odd that it thinks it's an 8250? Nov 06 16:44:49 <[g2]> dyoung-zzzz the kernel ? Nov 06 16:44:58 [g2] yes Nov 06 16:45:30 maybe. this changed from 2.4-linksys to 2.6 Nov 06 16:45:32 cegg, you might need a pullup on the 2nd port Nov 06 16:45:34 <[g2]> dyoung-zzzz is the irq hooked and we see stuff in dmesg ? Nov 06 16:45:57 <[g2]> and it's just RX right Nov 06 16:46:11 dyoung-zzzz: i had success with a pulldown. Nov 06 16:46:20 <[g2]> and the getty is setup right in inittab Nov 06 16:46:35 [g2] from a 2.7 bbaased kernel: Nov 06 16:46:37 ttyS0 at MMIO 0xc8000000 (irq = 15) is a XScale Nov 06 16:46:38 ttyS1 at MMIO 0xc8001000 (irq = 13) is a XScale Nov 06 16:46:45 I'm sure I set it up as such. Nov 06 16:46:53 <[g2]> cegg what are you doing with both ports ? Nov 06 16:48:06 [g2]: i got help in the past here already. RxD1 is connected to a gps-receiver. Nov 06 16:48:35 <[g2]> nod. Nov 06 16:48:44 <[g2]> can you tell me about you app and setup ? Nov 06 16:48:45 [g2]: The first port is used for to connect a gprs-modem Nov 06 16:48:52 * [g2] is curious Nov 06 16:48:55 <[g2]> and sells hw Nov 06 16:49:14 <[g2]> so like a liitte telematic setup Nov 06 16:49:41 cegg: So, when you boot via APEX, is the MMIO entry for the second port missing? Nov 06 16:49:41 <[g2]> do you have cheap GPS and GPRS setups ? Nov 06 16:49:58 [g2]: yep. thats for a tracking system Nov 06 16:50:17 beewoolie-near: yes Nov 06 16:50:19 <[g2]> cegg for personal or business ? Nov 06 16:50:24 BTW, according to the data-sheet, 0xc8000000 is the High-Speed UART. and 0xc8001000 is the console. Nov 06 16:51:19 [g2]: the second Nov 06 16:51:33 * [g2] hugs cegg Nov 06 16:51:38 dyoung-zzzz: Seems like there has to be something in the UART kernel driver that is have problems finding the port. Nov 06 16:52:25 Could be; I never actually tested it. :-) Nov 06 16:52:37 but iirc, I made that commit for cegg. Nov 06 16:53:06 and it was reportedly working (with redboot?) at some point. Something may have changed since then... Nov 06 16:53:06 Well, if as cegg says, it works when RedBoot boots and not APEX, then I'd guess that the kernel driver is looking for the UART to be in a particular mode. Nov 06 16:53:24 <[g2]> cegg I've to a completed system ready to go Nov 06 16:53:48 * dyoung-zzzz hmms Nov 06 16:53:48 dyoung-zzzz: Why does the /proc/iomem show an 8250 driver? Nov 06 16:53:54 <[g2]> can have two full serial ports and runs POE and boots of CF Nov 06 16:54:09 <[g2]> and runs all the openslug apps Nov 06 16:54:17 <[g2]> well 99.5% Nov 06 16:55:00 beewoolie-near: not sure; all the ixp42x board setup are inherieting the 8250 bits; so I just copied. Nov 06 16:55:06 [g2]: what do you mean? (Im not native-english) Nov 06 16:55:32 [g2] you need to have a picture. Nov 06 16:55:38 dyoung-zzzz: which driver source file are we using on the ixp? Nov 06 16:55:50 <[g2]> cegg I've got custom IXP42x boards Nov 06 16:56:14 I copied the ixdp425 one; but checked against the coyote and a couple others to make sure. Nov 06 16:56:53 How would you like to past the file into the pastebin so I can see it? Nov 06 16:57:03 [g2]: that means, well, non-linksys? Nov 06 16:57:26 <[g2]> yup, full Redboot, a bunch of extras Nov 06 16:57:48 <[g2]> can go industrial temp if you need it Nov 06 16:57:49 [g2]: nice Nov 06 16:57:59 I havnt looked at it for several months; lemme dig it up Nov 06 16:58:09 <[g2]> POE tons of goodies Nov 06 16:59:25 dyoung-zzzz: I know I have some patches here some where. Nov 06 17:01:26 http://pastebin.ca/27819 Nov 06 17:03:33 OK. Here's the wierd part. If the driver cannot open the device, it should release the resources. Nov 06 17:05:31 ...except that the code is really lame and does not checking. Nov 06 17:05:37 s/not/no/ Nov 06 17:06:11 heh Like I mentioned; I just pilfered it from someplace else. :-) Nov 06 17:06:24 It's not your doing. I'm reading the core 8250 code. Nov 06 17:06:33 ok. :-) Nov 06 17:07:30 Are you seeing a dmesg like this: Nov 06 17:07:39 Serial: 8250/16550 driver ... ? Nov 06 17:07:40 <[g2]> ttyS1 works fine on the Loft Nov 06 17:07:48 [g2]: With APEX? Nov 06 17:07:53 <[g2]> with Redboot Nov 06 17:08:10 <[g2]> I haven't gotten around to APEX on the Loft yet Nov 06 17:08:13 ttyS1 works fine on a nslu2 too. Thats not the issue. Nov 06 17:08:28 <[g2]> the latest version of Redboot can load the kernel from ext2 on CF/Microdrive Nov 06 17:08:29 I wonder if the easiest way to fix this would be to initialize ttyS1 in APEX. It shouldn't be necessary, but it might fix the problem. Nov 06 17:08:53 [g2]: Well isn't that just special...:-) Nov 06 17:09:37 <[g2]> yeah on a new unit I just change the command string to load .... -m disk hda1:/boot/zImage and put the microdrive in and boot Nov 06 17:09:45 <[g2]> root=/dev/hda1 Nov 06 17:09:54 cegg: I can add code to APEX to initialize the second serial port. I don't have the same kernel you have, so I cannot verify that it is working. Nov 06 17:10:03 <[g2]> it's really _just_ too sane Nov 06 17:10:23 [g2]: I think that their -m syntax is lame. Nov 06 17:11:19 <[g2]> it may be but it's something that works Nov 06 17:11:26 beewoolie-near: ok. where to add that? Nov 06 17:12:10 I'm making some changes... Nov 06 17:13:00 beewoolie-near: I have used apex-1.3.8 so far Nov 06 17:13:13 NP. Nov 06 17:19:05 cegg: OK. I've made the patch. How do you want to receive it? I can make a patch file. I can send a binary. Nov 06 17:20:42 Actually, I think I see exactly what is the problem. Once of the uart registers has an enable bit in it. Nov 06 17:20:59 It might be specific to this chip and not a generic 8250 feature. Nov 06 17:21:28 What I recommend we do is try the init code as I've written it. If this work, we can remove all but the setting of that one bit to see if that makes it work. Nov 06 17:21:54 beewoolie-near: email is ok Nov 06 17:25:11 cegg. Did you get my dialog messages? Nov 06 17:25:37 beewoolie-near: no. take johnnywhoo at gmx de (Private messages from unregistered users are currently blocked due to spam problems) Nov 06 17:26:09 cegg. Right. It's easy to register. Nov 06 17:26:16 Do you want a patch? or binary? Nov 06 17:26:29 beewoolie-near: patch is ok Nov 06 17:26:51 OK. BTW, I think that this code should really go into the kernel startup code for our target. it doesn't belong in the boot loader. Nov 06 17:27:43 <[g2]> beewoolie-near unless you want to use ttyS1 as your APEX console Nov 06 17:27:58 g Nov 06 17:28:04 [g2]: That's already accounted for. Nov 06 17:28:24 Initializing the second uart isn't something that the boot loader should do. Nov 06 17:28:48 Moreover, the bootloader could, and perhaps should, be able to disable both uarts once the kernel is running. Nov 06 17:28:57 It will save power for systems that don't need uarts. Nov 06 17:29:18 beewoolie-near: For testing purposes, it'll do. That should be done soon Nov 06 17:29:20 cegg: It's on it's way. Nov 06 17:29:42 Right. I'm not going to push this for now. if it works, I'll release another version of APEX. Nov 06 17:30:13 The only line that should be necessary is the last one in the init code: Nov 06 17:30:15 UART_O_IER = UART_IER_UUE; Nov 06 17:31:09 In fact, that's all we ought to need in the kernel, too. Nov 06 17:31:40 cegg: Perhaps you'd be better off testing the kernel with a small change to write that bit. Nov 06 17:32:27 cegg: email bounced. Are you sure you gave me the right name/ Nov 06 17:33:07 scheisse, replace the .de with .net Nov 06 17:33:38 beewoolie-near: the rest is correct Nov 06 17:34:21 beewoolie-near: received Nov 06 17:34:41 It's hard for me to tell if the code is built correctly. Nov 06 17:35:20 As I said, I think you'd be better off writing the single bit to the uart register in the kernel instead of hacking the bootloader. Safer and easier to test. Nov 06 17:36:02 *(unsigned long*) (0xc8000004) = (1<<6); Nov 06 17:36:59 Er, I think that's 0xc8001004 Nov 06 17:37:37 beewoolie-near: ok. Compile & jtag will take some minutes... Nov 06 17:37:53 K. I may be afk. Nov 06 17:38:50 <[g2]> beewoolie-near give doll-face a hug when you're afk Nov 06 18:03:09 [g2]: :-) He's coming over with soup in a few hours. Nov 06 18:04:29 beewoolie-near: it works! great! At least the kernel-message says so, because, currently, the jffs2-partition is dead. Nov 06 18:05:20 Nice. I wonder if we remove everything except for the final line if the UART still works? Are you willing to test it? Nov 06 18:05:37 beewoolie-near: yes, sure Nov 06 18:05:41 thx Nov 06 18:05:56 I have working jtag... Nov 06 18:21:53 beewoolie-near: Ok. Seems like the last line does the job. The stripped one works too. Nov 06 18:22:22 Excellent. I'll include only that line in the next release of APEX. Nov 06 18:22:28 Thanks for the legwork. Nov 06 18:23:23 I'll recover mtdblock4 partition to get it fully checked. Nov 06 18:24:56 beewoolie-near: Well, I have to thank. Nov 06 19:18:00 [g2]: hello pal it's dollface Nov 06 19:20:36 <[g2]> beewoolie-near hey doll-face Nov 06 19:57:35 Later dudes. Nov 07 00:59:29 i just submitted an UART_IER_UUE patch to l-k some days ago Nov 07 01:02:09 the 8250 driver also supports xscale-style serial ports -- the only difference is the UUE bi Nov 07 01:02:12 t Nov 07 01:03:18 in their infinite wisdom, intel decided to invert the sense of the UUE bit from what would have been the right idea **** ENDING LOGGING AT Mon Nov 07 02:59:57 2005