**** BEGIN LOGGING AT Wed Feb 23 23:59:56 2005 Feb 24 07:57:48 qbot: seen beewoolie Feb 24 07:57:49 beewoolie <~beewoolie@florence.buici.com> was last seen on IRC in channel #openjtag, 1d 8h 23m 13s ago, saying: 'night.'. Feb 24 10:59:23 <[g2]> boole2, new nick ? Feb 24 11:35:26 beewoolie-away: hey Feb 24 11:35:30 beewoolie-away: whats cookin? Feb 24 11:37:46 Just registering my mick. Feb 24 11:37:54 s/mick/nick/ Feb 24 11:38:16 I was also hoping that dyoung or [g2] would be willing to do a test. Feb 24 11:38:28 prpplague: what are you up to? Feb 24 11:39:56 <[g2]> beewoolie-away, how can I help ? What can I test ? Feb 24 11:40:31 [g2]: I have an hypothesis. I think that the only think we need to do to get the ixp0 driver running is a mac address. Feb 24 11:40:57 [g2]: If you'd be willing to boot with APEX and do ifconfig -a, we could get some info. Feb 24 11:41:11 If ifconfig can see the driver, all we may need to do is set the linkaddress. Feb 24 11:41:38 And if that that works, all I need to do is extract the mac address from flash and stuff it into a register before starting the kernel. Feb 24 11:41:42 <[g2]> OpenSlug automatically sets the hw addr :) Feb 24 11:41:53 Did you try it with APEX? Feb 24 11:42:18 <[g2]> I can pull the latest APEX and boot to it Feb 24 11:42:32 That would be keen-o-man. Feb 24 11:42:37 <[g2]> will do Feb 24 11:43:00 <[g2]> hey before I forget again http://www.usbee.com/ Feb 24 11:43:13 <[g2]> that was the usb analyzer stuff Feb 24 11:43:16 for the test to be interesting, APEX has to be written to the first flash block. Feb 24 11:43:55 Sounds like a zigbee product. Feb 24 11:43:57 <[g2]> beewoolie-away, I should be getting the FatSlug soon (today maybe) Feb 24 11:44:31 <[g2]> I don't have it yet, so running from memory won't help much ? Feb 24 11:46:20 beewoolie-away: legacy crap today Feb 24 11:46:49 beewoolie-away: gotta get it knocked out so i can start work on some drivers for apex tuesday Feb 24 11:46:54 [g2]: It has to boot to APEX, or redboot will perform HW setup and we won't be able to tell if OpenSlug needs the bootloader to init the NPE. Feb 24 11:47:23 [g2]: when you get the fat slug, we can test the memory init code for two banks. Feb 24 11:47:51 [g2]: That doesn't mean that the kernel will use the memory, but we can see if APEX does the init and can see all of the SDRAM. Feb 24 11:48:13 prpplague: :-) Feb 24 11:50:06 beewoolie-away: usb. cs8900 8bit polled, and spi-mmc Feb 24 11:51:12 [g2]: The bandwidth on that USBEE device isn't very high. It's about 10-12MHz. Feb 24 11:51:37 [g2]: Kinda expensive for such a low frequency device. Feb 24 11:54:41 <[g2]> beewoolie-away, are you taking the niquest theory into account I thought it was 24Mhz Feb 24 11:56:39 [g2]: Yeah. If it can take 24 million samples per second, then it can only resolve signals at less than 1/2 that frequency. With a margin for error, that puts the device at 10MHz or less. Moreover, it isn't clear to me if it shares that sampling BW among the channels which would significantly reduce the effective bandwidth. Feb 24 11:58:28 [g2]: This device is like many on the market. I've seen better ones that have sample at 100Ms/s for about the same money. IMHO, the price it too high for the features. Feb 24 11:59:15 anyone of you used jtag to read the register values and/or step through a program? Feb 24 12:06:50 <[g2]> beewoolie-away, are you initializing the ixp hw with APEX now ? Feb 24 12:07:17 [g2]: I am doing nothing to the NPE. AFAICT, there isn't anything I should have to do. Feb 24 12:07:26 That's the point of the test. Feb 24 12:07:40 The kernel driver already has the plumbing to upload the microcode. Feb 24 12:08:07 <[g2]> I thought I the npe hw was enabled in redboot Feb 24 12:08:11 AFAICT, there are no settings in the Ethernet register file that need to be tweaked. Feb 24 12:08:55 <[g2]> I guess my real question to you is why would 1.1.1 be any different thatn 1.0.27 ? Feb 24 12:09:02 [g2]: (I feel like we're going around a bit with this.) Yes, Redboot enables the NPE, but that's because it can use it. If APEX isn't going to use ethernet, and if the kernel can do the init, why add the code to APEX> Feb 24 12:09:19 [g2]: That's why I asked if you'd tested APEX with openslug. Feb 24 12:09:32 did you use apex 1.0.27 with openslug? Feb 24 12:09:38 Did you try ifconfig -a? Feb 24 12:09:40 <[g2]> from RAM Feb 24 12:10:00 From RAM, there isn't anything to test. Feb 24 12:10:05 <[g2]> which is the whole point that Redboot has run and done it's thing Feb 24 12:10:07 <[g2]> right Feb 24 12:10:13 Once redboot runs... Feb 24 12:10:19 <[g2]> exactly Feb 24 12:10:30 Did you ever boot APEX from the first flash block? Feb 24 12:10:49 <[g2]> not yet, hopefully the FatSlug arrives today Feb 24 12:10:57 OK. I thought that you had. Feb 24 12:11:01 <[g2]> I thought it might be here yesterday :( Feb 24 12:11:21 UPS Brown, across the country? Feb 24 12:11:41 <[g2]> I think he sent it 3-day last week Feb 24 12:21:17 Monday was a hiliday. Feb 24 12:21:41 <[g2]> ah... so maybe today ? Feb 24 12:39:27 oh @$@$%, assembly ppl fudged up one of the daughterboards Feb 24 12:39:39 cf socket is screwed Feb 24 12:43:45 whoops. Feb 24 12:47:45 * prpplague isn't very focused today Feb 24 12:48:59 beewoolie: did i understand correctly, that you are not adding ethernet support to apex? Feb 24 12:49:18 Not at the moment. Feb 24 12:49:39 Other features have higher priority. Feb 24 12:50:11 prpplague: Think about all that has to be done: Feb 24 12:50:40 prpplague: ethernet driver, specific to the HW; RARP/BOOTP/DHCP autoconf; TFTP Feb 24 12:51:07 Kind-of-a-lot. Even though there is existing code, it still takes time to rework it to fit into the loader. Feb 24 12:51:19 the tftp code from rmk can be used Feb 24 12:51:30 That isn't the point. Feb 24 12:51:31 as well as the much of the other stuff Feb 24 12:51:41 beewoolie: ok whats the point? Feb 24 12:51:50 Just because there is code, doesn't mean it can be used as-is. Existing code is not a 'library'. Feb 24 12:52:02 beewoolie: agreed Feb 24 12:52:09 There is existing ext2 code which I've been using to write the ext2 driver. Feb 24 12:52:24 The task isn't really that hard, but they do things very differently from me. Feb 24 12:52:56 IHMO, the Apex interfaces are simpler to program to. Feb 24 12:53:30 Besides, all of the protocols are documented. The hard thing about the NPE stuff is that the library is really the only way to interface to the chip. Feb 24 12:53:40 There is no documentation AFAICT. Feb 24 12:54:49 beewoolie: NPE? i'm not familiar with that Feb 24 12:55:29 NPE stands for Network Processing Engine. It's what handles networking on the IXP42x. Feb 24 12:55:42 ahh Feb 24 12:55:46 micrcode programmed cores. Feb 24 12:56:03 gotcha Feb 24 12:56:53 well 90% of the code in blob works for the ethernet i need to do, so i expect it shouldn't be difficult to add to apex Feb 24 13:02:44 prpplague: For example, APEX doesn't support malloc. For the time being, I don't plan to add it. Feb 24 13:03:37 prpplague: A polling driver should work, but you also need to handle all of the protocols. Doing so in a way that allows different drivers to run the network interface is an intergration task. That's where the work is. Feb 24 13:04:10 beewoolie: yea, blob doesn't use malloc either iirc Feb 24 13:05:31 prpplague: Don't recall. I think that redboot does and one of the other loaders I've used does, too. These things even implement ioctl. Kinda silly considering that ioctl is a loathesome hack. Why emulate it? Even Sharp emulates it in their sample code. Feb 24 13:06:17 beewoolie: yea i noticed that in the sharp code Feb 24 13:06:37 beewoolie: might have a quick look at the tftp/bootp code in blob Feb 24 13:06:47 prpplague: ejnoy. Feb 24 13:06:52 :-p Feb 24 13:06:55 beewoolie: no i mean you should Feb 24 13:07:04 beewoolie: i've already started moving it over to apex Feb 24 13:07:07 prpplague: As I said, it isn't very high on the list. Feb 24 13:07:13 hehe Feb 24 13:07:29 beewoolie: what is on the high priority list for apex? Feb 24 13:07:38 prpplague: The problem isn't that I don't think it can be done. The problem isn't that there is no sample code. The problem isn't that I don't want to do it. Feb 24 13:07:52 The problem is that I am limited in the time I can spend on it. Feb 24 13:08:20 beewoolie: you are saying no sample code for the npe stuff? Feb 24 13:08:25 right now, the ext2 driver is more important. After that, I need to add a bonafide configurator. Feb 24 13:08:33 beewoolie: cool Feb 24 13:08:53 beewoolie: the fat is better for us since 75% of our customers use blows on the desktop Feb 24 13:08:55 prpplague: No, the NPE bottleneck, and it is only for init, is that the library I tried, 1.5, doesn't build. Feb 24 13:09:15 beewoolie: ahh Feb 24 13:09:18 Right. That's why I implemented it, too. It is more useful to my client. Feb 24 13:09:38 beewoolie: well we will be using the cs8900, so implementation is very straight forward Feb 24 13:09:40 The ext2 driver is another step forward because it allows simpler integration. Feb 24 13:10:35 I've seen the cs8900 code before. Should be cake. Ethernet is a particularly tricky feature because the protocols have so many failure modes. Feb 24 13:10:43 i haven't looked, does the 1.1.0 have an uptodate todolist? Feb 24 13:10:53 yep Feb 24 13:10:55 I just keep a long of ideas. Feb 24 13:10:59 s/long/log/ Feb 24 13:11:14 No official TODO. The most important feature, actually, is the configurator. Feb 24 13:11:33 The biggest problem I am facing now, and will continue to face, is that fact that there are so many combinations to test. Feb 24 13:12:05 If I can make it easy to build lots of different loaders with many combinations of features, then I can at least be sure that it builds. Feb 24 13:12:14 You've seen how tricky this can be. Feb 24 13:12:52 I already have a target to build all of the _config's. It needs to be more sophisticated and be able to build lots of variations. Feb 24 13:23:38 beewoolie: yea Feb 24 13:24:00 beewoolie: i need to figure out how i'm gonna keep my patches in sync with your stuff Feb 24 13:24:15 beewoolie: you gonna make a cvs rep available? Feb 24 13:24:18 quilt is your friend. Feb 24 13:24:33 beewoolie: tried it last week Feb 24 13:24:38 beewoolie: doesn't work for me Feb 24 13:24:45 prpplague: I don't use CVS any more. SVN. I don't know if I'm gonna hav a repo. Feb 24 13:25:01 prpplague: Sounds like operator error. :-p Did you find the quilt paper? Feb 24 13:25:22 It's really quite easy to use. Feb 24 13:26:14 i didn't mean i couldn't make it work, it just doesn't "work" for me, hehe Feb 24 13:26:36 what do you expect that it doesn't do? Feb 24 13:27:30 i really didn't expect anything Feb 24 13:27:50 quilt new name Feb 24 13:27:57 quilt add file Feb 24 13:28:00 vi file Feb 24 13:28:03 quilt refresh Feb 24 13:28:12 quilt pop Feb 24 13:28:19 You now have a patch. Feb 24 13:28:24 nod Feb 24 13:28:45 nod Feb 24 13:28:56 dyoung-zzzz: the file got corrupted when it crashed. Feb 24 13:29:08 so I'm recreating the lower left again Feb 24 13:29:23 i guess i'm just a stick in the mud, diff -urN Feb 24 13:29:34 its not still ni your "sent mail" folder? Feb 24 13:30:21 prpplague: That is only so good. If you are going to make several different kinds of changes to a project, it doesn't work. besides, you'll need two copies of the source. quilt is simpler. Feb 24 13:30:27 <[g2]> dyoung-zzzz, Are you quilting on my now ? Feb 24 13:30:38 no..because I use webmail. Feb 24 13:30:53 quilting bEE Feb 24 13:30:54 and the attachment isn't there. Feb 24 13:31:04 let me check something on that. Feb 24 13:31:14 It *may* be . Feb 24 13:33:07 beewoolie: different worlds, our code is usally very static, which is why OE doesn't work well for us either Feb 24 13:35:02 prpplague: APEX is not going to be so static. I will be releasing patches for inter-release differences. Not happening right now. It'll be up to you to figure out how to keep in step. AFAICT, a repo isn't any better than patches. In fact, it's worse because it is harder to back-out changes. Feb 24 13:36:25 beewoolie: well, very 1.1.0 has pretty much all i need, so, unless i run into problems i don't expect to upgrade from there Feb 24 13:37:13 As long as you publish your changes, there's no problem. Feb 24 13:44:32 beewoolie: yea, we make a cd available with all they gpl code Feb 24 13:48:11 beewoolie: hehe, we comply with all licenses Feb 24 13:48:33 beewoolie: we actually have QA person who is in charge OSS license compliancy Feb 24 13:51:19 beewoolie: because it doesn't sound like you'll be adding anything new that would be benificial to us Feb 24 14:16:22 beewoolie: has sharp made any requests for additions to apex? Feb 24 14:17:29 prpplague: Nothing so far. Feb 24 14:18:54 beewoolie: so whats on the list after ext2 and configuration ? Feb 24 14:22:03 <[g2]> dyoung, can you do a flash test of APEX 1.1.1 with OpenSlug for beewoolie ? Feb 24 14:22:47 <[g2]> beewoolie, wants do know if the ixp driver actually loads Feb 24 14:23:03 <[g2]> and works Feb 24 14:25:21 <[g2]> beewoolie, we should also post the question on the ixp425 forums Feb 24 14:25:36 was there a salient change in that part of the code over 1.0.26 ? Feb 24 14:26:07 oops Feb 24 14:26:25 <[g2]> I don't think so regarding the ixp Feb 24 14:26:33 <[g2]> but beewoolie would be the best to answer it Feb 24 14:26:50 in that case we did the test. Feb 24 14:27:01 per your suggestion 1.5 weeks ago. Feb 24 14:27:20 I can flash it though. Feb 24 14:27:33 just not right now. Feb 24 14:35:18 okay which kernel? Feb 24 14:40:07 dyoung: I hypothesize that there isn't anything that the bootloader has to do to get the network interface working. Feb 24 14:40:50 dyoung: I suspect that the interface may have been lacking a MAC address, so it wasn't bringing up the link. Feb 24 14:41:18 The easy test is to do ifconfig -a once the system is running. If ifconfig can see the interface, then we probably only need to set the MAC address. Feb 24 14:41:53 prpplague: I don't plan that far ahead. Feb 24 14:43:20 this looks interesting - http://freshmeat.net/projects/fjtag/ Feb 24 14:43:32 <[g2]-away> I'll be back in a couple hours Feb 24 14:45:26 I went down that road before... Feb 24 14:45:43 dyoung: with fjtag? Feb 24 14:45:57 no, with ifconfig hw 1:2:3:4:6:7 Feb 24 14:46:01 ahh Feb 24 14:46:32 Okay just to be sure I 'll try it with unslung. Feb 24 14:46:43 dyoung: Unslung will definitely not work. Feb 24 14:46:45 and I wont touch the 2nd block so it will kn ow the mac. Feb 24 14:46:55 oh Feb 24 14:46:56 I just don't remember if we looked at this carefully. Feb 24 14:47:09 what do you mean no, with ifconfig hw 1:2:3:4:6:7 Feb 24 14:47:10 We did. Feb 24 14:47:22 Did ifconfig -a show the interface? Feb 24 14:47:26 It did. Feb 24 14:47:30 but th elight still wanst on. Feb 24 14:47:44 I tried ifcofnig eth0 1.2.3.4 up Feb 24 14:47:50 with the default mac Feb 24 14:47:54 which had no effect. Feb 24 14:48:09 ixp425_eth sets a default mac addr. Feb 24 14:48:58 OK. I just wanted to make sure. Feb 24 14:49:19 There is code in the driver to load the micrcode, so it seemed reasonable that it could do all of the init. Feb 24 14:49:42 Blame it on Intel. Feb 24 14:50:25 I blame Canada. Feb 24 14:50:30 wasn't that a song? Feb 24 14:50:38 ;-) Feb 24 14:51:04 Blame it on the Rain. Feb 24 14:52:37 I can re-try the experiment later this eve if you like; but I'm 99.999% sure we went this way before. Feb 24 14:53:46 dyoung: I just got a tip from someone else about getting the 1.5 access library to build. Feb 24 14:53:56 dyoung: I'll give it another shot next week. Feb 24 14:54:09 ok Feb 24 14:54:21 I got distracted with other stuff and havnt looked at 1.5csr since 2 weeks ago... Feb 24 14:54:24 heh Feb 24 14:57:19 If you are interested, the tip is that the some of the constants that the library uses come from the kernel. In 2.6, these changed from IXP425_xxxxx to IXP4XX_xxx. I haven't verified this yet. Feb 24 17:05:21 <[g2]-away> how interested are you guys in 425 hw ? Feb 24 17:17:07 dyoung...beewoolie..i did the ram upgrad on my slug..512Mbx2 -> 128MB slug...I also got the serial port working Feb 24 17:17:21 You're a superstar. Feb 24 17:17:34 You just couldn't wait, eh. Feb 24 17:17:38 :) Feb 24 17:17:58 beewoolie..i figured i'd get it done then help w/ apex testing... Feb 24 17:18:00 I suppose you might want to verify that the RAM works. Feb 24 17:18:13 well it boots. so at least 32MB works... Feb 24 17:18:14 Have you downloaded the source? Feb 24 17:18:23 not yet...gotta wait till i get home tonight Feb 24 17:18:35 i did the upgrade during lunch... Feb 24 17:18:38 at work. Feb 24 17:18:40 samms71-away: Hmm. I'm not sure what that means. Actually, it is kinda wierd that it works at 32MiB. Feb 24 17:19:05 Did you remove the old RAM and put on two new chips? Feb 24 17:19:08 well, it boots up and free says it has 32MB.. Feb 24 17:19:21 yeah, desoldered the old chips and put in two micron 512Mb chips... Feb 24 17:19:50 it seems happy..you think it shouldn't work? Feb 24 17:20:07 Hmm. I don't have time to look at this in detailright now. I suppose it might not matter ... thinking... Feb 24 17:20:24 It depends on how the chips implement the extra space. As rows or columns. Feb 24 17:21:14 It should be easy to test with APEX. When you get home, download the 1.1.1 source. You'll have to change a constant in the initialization file since I've not made the change for 512Mib parts. Feb 24 17:21:33 it's configured at extra columns. since the extra lines aren't driven it should look like the 1K pages that the old parts use...the only thing i was unsure about was the refresh. i thought the old parts used A10 as a refresh...not sure how the code configures the memory controller.. Feb 24 17:21:49 i can just follow the instructions at your wiki.....? Feb 24 17:22:17 Yeah. flash it to the sysconfig area and then you can see if the SDRAM init's properly. Feb 24 17:22:29 Hey dyoung-web Feb 24 17:22:38 hey dyoung.. Feb 24 17:22:58 hiyo.. Feb 24 17:23:02 I was under the impression that the only differences in the chips has to do with the way that address lines are multiplexed. Feb 24 17:23:22 The refresh should still work. Feb 24 17:23:31 Anyway, it seems to. Feb 24 17:23:53 i don't have the datasheets here but i thought i remembered something about A10 being used....i'll check later..it still seems to work...unslung booted and i could play around through the serial interface Feb 24 17:24:07 Sounds good. Feb 24 17:25:58 samms71-away: I hate to be the bearer of bad news. Feb 24 17:26:06 ? Feb 24 17:26:13 Wait, I spoke too soon. Feb 24 17:26:29 all you have to do is change the SDR_CONFIG constant. Feb 24 17:27:26 ...to SDR_CONFIG_2x32Mx16. Feb 24 17:28:15 Right now, it checks the number of banks. It doesn't check the size. Activate just one bank and then change the init to use that constant. Feb 24 17:28:31 If I get to it, I'll make some changes to the wiki. Feb 24 17:28:53 i grabbed 1.1.1..what's the path to this constant? Feb 24 17:29:22 found it... Feb 24 17:35:14 ok gotta go...thx beewoolie...i'll try it.. Feb 24 17:42:13 <[g2]> damn missed em Feb 24 17:42:26 samms71? Feb 24 17:45:18 <[g2]> yeah Feb 24 17:45:30 <[g2]> he did the apache/php ports Feb 24 17:46:23 <[g2]> also, he'll have have to use the dmabounce stuff because the PCI only supports the 64MB window Feb 24 17:46:40 <[g2]> those will be virgin tracks on the slug Feb 24 17:47:26 I thought that the dmabounce stuff was kinda automatic in the kernel. Feb 24 17:49:03 <[g2]> I think it is enabled automatically, but I don't know if it has ever been tested or used Feb 24 17:49:14 <[g2]> maybe on the ixp425 platform Feb 24 17:49:33 I don't think we're breaking *that* much new ground. Feb 24 17:52:01 <[g2]> beewoolie, this is 2.6 remember Feb 24 17:52:18 * beewoolie remembering Feb 24 17:53:05 <[g2]> so there it was ixp425 specific to start Feb 24 17:53:18 it took some *cough* *hacK* ing to make 2.6 go on the ixp target... Feb 24 17:53:43 and lots of patching to make CSR 1.4 compile. Feb 24 17:54:10 <[g2]> 2.6.9 got better, but a lot of this stuff is new Feb 24 17:55:49 beewoolie: the ka6sox 16M16 64M slug boots fine thinking it has 32M. Feb 24 17:57:06 dyoung-web: I'm going through it right now. I've gotten past the constant problem and am looking at some other things. I don't know how much there is, though. Feb 24 18:40:23 dyoung-web: So, I have an interesting hitch in the library. They're assuming that lots of things are statically mapped from physical to virtual addresses. This is true for all but one, the Q manager. sound familiar? Feb 24 18:54:53 <[g2]> kas11, howdy and ping! Feb 24 18:57:25 hiya [g2] ;) Feb 24 18:57:44 <[g2]> hi! Feb 24 18:58:07 [g2] wazzup? Feb 24 18:58:19 <[g2]> Hey I'm starting to look through the code, but you wouldn't remember the major/minor number for /dev/leds would you ? :) Feb 24 18:58:49 <[g2]> 60 0 ? Feb 24 18:59:17 <[g2]> #define N2RB_MAJOR 60 Feb 24 18:59:17 <[g2]> #define N2PB_MAJOR 61 Feb 24 18:59:17 <[g2]> #define N2BZ_MAJOR 62 Feb 24 18:59:17 <[g2]> #define N2LM_MAJOR 126 Feb 24 18:59:46 <[g2]> I'm trying to test the code here http://www.nslu2-linux.org/wiki/OpenSlug/UseLedAndBuzzerOnOpenslug Feb 24 19:00:29 <[g2]> if ((leds = open("/dev/leds", O_RDWR)) < 0) { fails because I don't have the device created Feb 24 19:00:56 126 Feb 24 19:02:02 <[g2]> 126 0 Feb 24 19:02:11 <[g2]> and it's a character device ? Feb 24 19:02:20 far as I know, it was working for VoodooZ.... i don't have anything to test on anymore Feb 24 19:02:33 <[g2]> that's ok it's *ready* to run Feb 24 19:03:42 if (register_chrdev(N2_LM_MAJOR, "n2_ldm", &n2_lm_fops) < NOERR) { Feb 24 19:03:42 printk(KERN_ERR "Led Manager Major %d not available\n", N2_LM_MAJOR); Feb 24 19:03:42 return -EBUSY; Feb 24 19:04:33 <[g2]> green lights! Feb 24 19:04:34 but we were using ioctls... Feb 24 19:04:53 <[g2]> it's a GO Feb 24 19:05:02 <[g2]> thx Feb 24 19:05:07 <[g2]> THX! Feb 24 19:05:14 kewl :) Feb 24 19:05:54 beewoolie: you know about the 2.6 patchs for csr 1.4 right? Feb 24 19:06:06 <[g2]> now I'll just change the default in the kernel to change the Ready/Status RED during boot then at then end of sysvinit GREEN Feb 24 19:06:16 I didn't find any. Feb 24 19:06:22 beewoolie: http://ixp4xx-osdg.sourceforge.net/linux-2.6.html Feb 24 19:06:33 that page didn't work last time I tried... Feb 24 19:06:41 I'm on it right now... Feb 24 19:06:50 Hmm. Must be a different URL. Feb 24 19:06:52 i have the patches too if thats more copnvenient. Feb 24 19:06:57 <[g2]> 404 Feb 24 19:07:09 <[g2]> It's up now Feb 24 19:07:14 http://ixp4xx-osdg.sourceforge.net/linux-2.6.html Feb 24 19:07:22 yeah I'm on it right now. Feb 24 19:07:35 Rob sent me a like earlier that was wrong. Wierd. Feb 24 19:07:55 Anyway, I have to stop for a while. Dollface is here and I'm heading for NYC tomorrow. Feb 24 19:08:12 I'll revisit it when I get back next week. Feb 24 19:08:17 Woo woo! Feb 24 19:08:21 <[g2]> beewoolie, HAVE a GREAT TIME! Feb 24 19:08:26 Have fun in the big apple. Feb 24 19:08:34 I put up a new version of APEX that should handle *all* of these fat slug configs. Feb 24 19:08:39 WOO! Feb 24 19:08:42 1.1.2 ? Feb 24 19:08:48 It recognizes the bank size and tries to to the right thing. Feb 24 19:08:51 Let samms know. Feb 24 19:09:07 Thanks beewoolie ! Feb 24 19:09:15 np. Feb 24 19:09:22 Youre the man. Feb 24 19:09:28 ~worship beewoolie Feb 24 19:09:28 bows down before beewoolie saying "We're not worthy! We're not worthy!" Feb 24 19:09:32 I've added some instructions to the page, too Feb 24 19:09:56 http://wiki.buici.com/twiki/bin/view/Main/ApexBootloader Feb 24 19:10:23 I'll mention it to samms71 when he comes around again. Feb 24 19:11:19 thx Feb 24 19:12:01 later d00dz Feb 24 19:12:09 have fun travel! Feb 24 19:27:13 I sure wish there was a less painful way of building the OE kernel...there is a lot of unfinished business in that driver Feb 24 19:30:07 it is a bit heavyweight.. but its not too bad if you know what youre going to be working with the kernel only. Feb 24 19:31:40 <[g2]> kas11, once you build the first time its easy to modify Feb 24 19:32:58 <[g2]> kas11 http://www.nslu2-linux.org/wiki/HowTo/TweakKernelOrPackagesInOE Feb 24 19:34:11 <[g2]> that builds the zimage, ipks and will rebuild the flashimage Feb 24 19:34:27 <[g2]> a bb openslug-kernel would just do the zImage and ipkgs Feb 24 19:34:54 and if you really want to get fancy you can use quilt to make you a patch on the spot. Feb 24 19:35:46 <[g2]> dyoung-web my quilting buddy :) Feb 24 19:36:00 <[g2]> I knew you were *old* but really :) Feb 24 19:36:13 Aw gee. Feb 24 19:36:37 I was trying to read 2 wiki pages at once...mix and match didn't play well...and since I have no domain name bk (iirc) got part way thru and blew me off Feb 24 19:37:21 <[g2]> BK_HOST= Feb 24 19:37:29 <[g2]> BK_USER Feb 24 19:37:37 <[g2]> you can set that to anything Feb 24 19:37:42 yeah, mine is set like this Feb 24 19:37:44 export BK_USER=dyoung Feb 24 19:37:44 export BK_HOST=builder.$none$ Feb 24 19:37:55 so it doenst grumble about FQDN's. Feb 24 19:38:31 builder being the host's name? Feb 24 19:38:42 yes ma'am! Feb 24 19:38:51 My Hostnames are so unorginal. Feb 24 19:38:51 gotcha :) Feb 24 19:39:05 the mini is mini. Feb 24 19:39:15 the VCR is PCDVR Feb 24 19:40:55 I started naming mine after planets sorta in order of size...17 computers later hosts is a very sorry looking mess Feb 24 19:41:33 I suppose when I get a imac, that will probbly be imac. Feb 24 19:41:48 there was my webserver webbie, but I lobotomized it. Feb 24 19:41:56 then sluggie. Feb 24 19:42:04 and toonie. Feb 24 19:42:09 I need a theme. Feb 24 19:43:00 and I have a real problem with the wrt54gsx and the gumstix...cuz they both want the same name... Feb 24 19:58:01 <[g2]> gsX ? Feb 24 19:58:29 Extreme? Feb 24 19:58:34 hmm Feb 24 19:58:41 <[g2]> I was just starting to setup dnsmasq on OpenSlug Feb 24 20:00:26 well, I settled on gsx for the gumstix and am using the gs for now...the gsx is on the shelf which makes the problem go away for now.. Feb 24 20:01:23 <[g2]> what's the X Feb 24 20:01:30 ok...where the hell is module bb...yes, I set pythonpath Feb 24 20:01:58 sounds like you need an update of bitbake Feb 24 20:02:12 blahblahblah something something missing bb ? Feb 24 20:02:55 There is a WRT54GX? Feb 24 20:03:01 Which looks really ugly IMO... Feb 24 20:04:05 "Global name 'bb' is not defined" ? Feb 24 20:04:18 that one is a bitbake update, so cd bitbake; svn up should do it Feb 24 20:04:30 these 2 wike pages really need to be fixed...stepbystep doesn't even mention needing to install bitbake...and the other one talks about path/to/openembedded before the OE instructions Feb 24 20:05:06 http://www.linksys.com/products/product.asp?grid=33&scid=35&prid=670 Feb 24 20:05:17 <[g2]> thx Feb 24 20:05:35 <[g2]> that's the mimo one right Feb 24 20:05:40 Yeah Feb 24 20:05:52 Looks like it Feb 24 20:05:54 I think rwhitby runs bitbake in place without installing it. Feb 24 20:06:13 Does SRX use more than one channel? Feb 24 20:06:23 <[g2]> "Multiple In, Multiple Out" (MIMO) Feb 24 20:06:40 Hmmm... Fun if you've got anybody else nearby with one then Feb 24 20:06:41 <[g2]> By overlaying the signals of two Wireless-G compatible radios, the "Multiple In, Multiple Out" (MIMO) technology effectively doubles the data rate Feb 24 20:06:59 <[g2]> Tiersten, reading challenged Feb 24 20:07:08 ? Feb 24 20:07:21 <[g2]> he didn't want to follow dyoung's link :) Feb 24 20:07:38 * dyoung-web pout Feb 24 20:07:50 <[g2]> Tiersten, I'm just taking-the-piss outta ya mate Feb 24 20:07:53 I was at that link before dyoung posted it. That's how I knew it looked really ugly and there was a WRT54GX :) Feb 24 20:07:58 hahaha Feb 24 20:08:11 I only went there after tiersten said it was ugly. Feb 24 20:08:16 I had to see for myself. Feb 24 20:08:23 and damn its UGLY. Feb 24 20:09:20 Only 3 external antennas as well Feb 24 20:09:21 well, there is the gs in the case...and there is the other one that has been hacked...which sorta got named gsx before the gumstix came along. Feb 24 20:13:20 <[g2]> Tiersten, Umm is that 2 RX (for diversity) and 1 TX Feb 24 20:14:25 Not sure. Might have one of those tiny strip antennas inside I guess Feb 24 20:15:24 <[g2]> We'll I think you get double the thoughput by using an uplink and downlink but I could be all wet Feb 24 20:17:22 <[g2]> hmmm, I'm all wet Feb 24 20:17:38 <[g2]> The diagram on this page looks cool though http://www.extremetech.com/article2/0,1558,1759082,00.asp Feb 24 20:19:02 Same channel then Feb 24 20:19:12 "Unlike channel-bonding techniques, MIMO can do its multiplexing within the bandwidth of a single 802.11 channel." Feb 24 20:19:36 Which is nice unlike some of the other nonstandard speed increase things we've had before Feb 24 20:19:49 this is making me feel dumber than usual...ERROR: Unable to open conf/bitbake.conf...perms are 664...775 don't help Feb 24 20:23:25 dunno... Feb 24 20:24:21 It in BBPATH? Feb 24 20:25:28 It's got this in the manual for BB "Currently, BitBake has hardcoded knowledge of a single configuration file. It expects to find 'conf/bitbake.conf' somewhere in the user specified BBPATH" Feb 24 20:26:15 BBPATH points to build...bitbake is in the same dir as build and bitbake has a dir named conf with bitbake.conf in it... Feb 24 20:26:46 You've reached the extent of my BitBake troubleshooting knowledge :) Feb 24 20:30:21 this wiki page is in serious need of editing Feb 24 20:31:01 Yeah. The pages on how to set up OE/BB in the wiki are really out of date and wrong in a few places. Feb 24 20:31:32 <[g2]> kas11, in the next couple days/ week I'll update the OE/BB for OpenSlug setup Feb 24 20:32:00 [g2] it will be much appreciated :) Feb 24 20:32:18 <[g2]> and you can be my alpha tester :) Feb 24 20:33:51 works for me Feb 24 20:34:29 <[g2]> Ok ping me in a day or two if you haven't heard anything Feb 24 20:34:46 [g2] which page were you going to repair? Feb 24 20:35:57 <[g2]> http://www.nslu2-linux.org/wiki/OpenEmbedded/StepByStepSetup Feb 24 20:36:00 <[g2]> is that the one ? Feb 24 20:36:20 http://www.nslu2-linux.org/wiki/OpenEmbedded/GettingStarted Feb 24 20:36:22 also Feb 24 20:36:28 This will create a bitbake/ directory in the current directory. Remain in this current directory (not the bitbake subdirectory) for the following steps (all the other directories go alongside the bitbake/ directory). Feb 24 20:36:34 they have conflicing information. Feb 24 20:36:51 Installing BitBake Software Feb 24 20:36:51 After downloading BitBake you will need to install the BitBake software. To install, cd into the BitBake directory you just created and type ./setup.py install --prefix=/usr/local --install-data=/usr/local/share. Feb 24 20:37:43 line sez stay here...very next instruction sez cd Feb 24 20:37:46 <[g2]> For Beta users, we should have the recipe for them to run and build the firmware Feb 24 20:38:23 agree. Feb 24 20:38:32 <[g2]> things have changes some since the original write-ups Feb 24 20:38:34 thats why I was wondreing which page you were gonna work on. Feb 24 20:38:44 <[g2]> I'll look at both of them Feb 24 20:38:49 2 eyes better than one. Feb 24 20:38:51 ok Feb 24 20:39:10 <[g2]> I think there may be a 3rd on the OE wiki Feb 24 20:39:10 I'll spend more time testing mt-daapd and less time hacking up the wiki. ;-) Feb 24 20:39:21 <[g2]> which is kergoth's domain Feb 24 20:40:36 <[g2]> the OpenSlug symlinks will need to be updated and maybe even removed Feb 24 20:40:51 openslug symlinks? Feb 24 20:41:02 <[g2]> nslu2-symlinks Feb 24 20:41:03 I've been using nslu2-package-symlinks. Feb 24 20:41:10 <[g2]> that one Feb 24 20:41:16 why removed? Feb 24 20:41:25 or do you mean for old stuff? Feb 24 20:41:31 <[g2]> it defeat the whole purpose of the just another OE release Feb 24 20:41:43 oh for the wiki. Feb 24 20:42:03 <[g2]> I have used the symlinks for OpenSlug for months Feb 24 20:42:04 thats fine because the symlink package is described elsewhere. Feb 24 20:42:28 <[g2]> and I just say bb frotz and it does Feb 24 20:42:28 http://www.nslu2-linux.org/wiki/OpenEmbedded/NSLU2PackageSymlinks Feb 24 20:42:45 yah Feb 24 20:42:58 <[g2]> bb frotz dnsmasq openssl php Feb 24 20:43:07 I've been using the symlink package from the day after you sent it to me. Feb 24 20:43:07 <[g2]> they all build Feb 24 20:43:30 <[g2]> it's a nice little work-around and fine for unslung Feb 24 20:43:45 <[g2]> the user can create his own for OpenSlug Feb 24 20:44:22 so what will go into the feed I wonder. Feb 24 20:44:38 <[g2]> the feed ? Feb 24 20:44:44 <[g2]> All packages that work and are useful Feb 24 20:44:56 nethack count? Feb 24 20:45:20 j/k Feb 24 20:45:38 <[g2]> ccsmart volunteered to help with the OpenSlug package status update Feb 24 20:45:47 <[g2]> I think there are others what would do the same Feb 24 20:46:20 ok Feb 24 20:46:25 <[g2]> We can either setup a build box to build all the packages every day and update the status, or have some rolling status with feedback from users Feb 24 20:46:59 nudi might be fast enough to handle it. Feb 24 20:47:22 I have bk setup on it now; and it pulls every hour. Feb 24 20:47:38 <[g2]> I mean every couple days or once a week would even be userful Feb 24 20:47:52 <[g2]> hmmm that's a funny typo .... userful Feb 24 20:48:02 I'm sure we can use nudi for that. Feb 24 20:48:07 come up with a schedule of some kind. Feb 24 20:49:27 do y'all really want to build as a user named slug...many examples I'm seeing say OEROOT = /home/slug...$(HOME)/slug, no? Feb 24 20:51:07 I dunno? Feb 24 20:51:31 Mine is all in $HOME/bbroot Feb 24 20:51:37 ...maybe that is where I'm going wrong...but I sure want things in my home directory Feb 24 20:52:10 yeah, that sounds better... Feb 24 20:52:15 $HOME/bbroot/{openembedded,bitbake,nslu2-package-symlinks} etc Feb 24 20:53:27 yeppers...that seems like what it outta say on the wiki...I did it like that so I really was *not* following instructions...not much new there tho... Feb 24 20:57:12 back lkater Feb 24 20:57:14 brr Feb 24 21:08:02 <[g2]> dyoung, was going to mention at bewoolie has a new APEX with additional memory configs Feb 24 21:08:21 1.1.1? Feb 24 21:08:38 <[g2]> I think it'll be 1.1.2 Feb 24 21:08:56 <[g2]> beewoole's on the way to NYC for several days Feb 24 21:09:04 ah..nice...i was going to try 1.1.1 tonight..dyoung tell you i upgraded my slug to 128MB... Feb 24 21:09:18 <[g2]> no I read it here Feb 24 21:09:31 k...forgot about the logs... ;) Feb 24 21:09:32 <[g2]> I just missed you when you left by a couple minutes Feb 24 21:09:50 <[g2]> I was in the channel just not at the terminal :) Feb 24 21:10:01 yeah, i'm usually on after the baby goes to bed...no play time before ~9pm. Feb 24 21:10:24 i haven't looked to see how you build apex...is there an image already built i can test? Feb 24 21:10:26 <[g2]> you must be gmt-8 Feb 24 21:10:30 yeah.. Feb 24 21:10:39 sorry..internet=world... Feb 24 21:10:40 <[g2]> GMT-5 Feb 24 21:10:49 whoa....night owl Feb 24 21:11:02 <[g2]> bedtime soon :) Feb 24 21:11:12 i'm in vancouver, WA....across the river from portland Feb 24 21:11:16 <[g2]> the README's in APEX are good Feb 24 21:11:29 k...apex... Feb 24 21:11:45 i wanted to also see if i could mod redboot..do you think it's worth a try? Feb 24 21:11:46 <[g2]> bummer, I'm near Raleigh, NC Feb 24 21:12:05 hehe, i spent a few months a jacksonville (camp lejuene sp?) Feb 24 21:12:12 <[g2]> you could help me with my solder challenged state :) Feb 24 21:12:19 ? Feb 24 21:12:39 * [g2] needs practice (lots) with the soldering iron Feb 24 21:13:09 hehehe....yeah i get alot of practice in my work (analog ee)..gotta fix all of my mistakes. :) Feb 24 21:13:17 what help do you need? Feb 24 21:13:43 <[g2]> practice, practice, practice Feb 24 21:14:07 1.1.2 is on the apex ftp...btw Feb 24 21:14:11 <[g2]> ka6sox-away, is sending my a FatSlug that's serial and JTAG enabled Feb 24 21:14:37 i just did both to my slug this afternoon...nice having the jtag/serial interface... Feb 24 21:14:39 <[g2]> you did apache and php right ? Feb 24 21:14:53 yeah...not alot of code work more ./configure work Feb 24 21:15:11 i just pretend i can do code stuff... ;) Feb 24 21:15:13 <[g2]> I'd like to get php up to 5.0.3 in OE Feb 24 21:15:37 <[g2]> and also integrated with thttpd that's in OE and get apached in OE working Feb 24 21:15:42 i had problems w/ php..couldn't get libxml built...ldconfig was missing and i don't know how to build w/o it.. Feb 24 21:15:49 php 5.0.3 that is.. Feb 24 21:16:12 <[g2]> I think another guy got that working with thttpd Feb 24 21:16:28 <[g2]> there was an e-mail on the list a couple days ago Feb 24 21:16:44 i saw that earlier...is libxml built for OE? Feb 24 21:16:54 <[g2]> dunno Feb 24 21:17:44 hmmm...can you build OE packages as a crosscompile? i assume you can since bitkeeper does most of the work... Feb 24 21:18:00 for config...etc... Feb 24 21:18:19 <[g2]> are you planning on playing with OpenSlug at all ? Feb 24 21:18:27 sorry..i really am dense about software stuff.. i used to do a fair bit of firmware but not a lot of package stuff... Feb 24 21:18:52 <[g2]> hey NP, we're a diverse set of ppl Feb 24 21:18:59 <[g2]> we all can't know it all Feb 24 21:19:14 i think i'm going to buy another slug and put openslug on it...my 1st slug i use for gallery serving baby pictures to the family so i can't take it down for openslug...wife would kill.. ;) Feb 24 21:19:15 <[g2]> then there'd be a bunch of duplication Feb 24 21:19:48