**** BEGIN LOGGING AT Thu Dec 22 02:59:58 2005 Dec 22 03:23:00 ka6sox: qemu emulates the FPA floating point unit Dec 22 03:23:09 ka6sox: which means you can use both hardfloat and softfloat binaries Dec 22 09:21:09 <[g2]> ka6sox ka6sox-office ping Dec 22 09:21:34 <[g2]> I'm wondering what's up with the board Dec 22 11:30:47 [g2]: hey man Dec 22 11:30:56 prpplague: yo Dec 22 11:31:26 beewoolie-afk: hey Dec 22 11:31:31 beewoolie-afk: whats cookin? Dec 22 11:31:51 prpplague: Adding MMU enabling code APEX at the moment. Dec 22 11:31:54 beewoolie-afk: btw, the lnode80 goes into full production monday Dec 22 11:32:10 It turns out to be necessary (desirable) for splash screen decode. Dec 22 11:32:16 prpplague: kudos Dec 22 11:32:26 beewoolie-afk: i've got a complete kit set aside to send you Dec 22 11:32:44 prpplague: sweet. Did we finish with the patch? Dec 22 11:32:53 beewoolie-afk: for the lnode80 yea Dec 22 11:32:58 beewoolie-afk: not for the s3c2410 Dec 22 11:33:03 k. Dec 22 11:33:05 beewoolie-afk: i got pulled off for a fire Dec 22 11:33:15 beewoolie-afk: major $$$ project Dec 22 11:33:21 Always a good ting. Dec 22 11:33:30 beewoolie-afk: 3500 very expensive handhelds Dec 22 11:33:47 s/very/VERY Dec 22 11:34:32 beewoolie-afk: how stable is the usb slave on the 79524/5 series? Dec 22 11:34:53 prpplague: I don't think that the slave has ever been coded. Dec 22 11:35:01 wow Dec 22 11:35:03 At least, I've not noticed it. Dec 22 11:35:09 No one has asked for t. Dec 22 11:35:20 interesting Dec 22 11:35:46 seems odd, but then i guess the 79524/5 isn't geared for handhelds Dec 22 11:36:10 Well, no one has asked for the client (slave) interface for any of the LH processors. Dec 22 11:36:23 interesting Dec 22 11:36:28 seems very odd Dec 22 11:36:41 It's possible someone has done it as I recall that there was some gadget work. Isn't that usually on the slave? Dec 22 11:36:41 that would be one of the first things we'd have looked for Dec 22 11:37:21 yea Dec 22 11:37:52 Then I think there has been something written for it. Dec 22 11:38:27 ahh Dec 22 11:38:33 There is a file i the tree...drivers/usb/gadget/lh7a40x_udc.c Dec 22 11:38:52 well, i was kinda thinking about doing a small board for own stuff Dec 22 11:39:14 but as far as work goes, sharp really shot themselves in the foot with us Dec 22 11:39:26 So you said. Dec 22 11:39:38 i still like the 79520 for small stuff Dec 22 11:39:53 thats one reason we still went forward with the lnode80 board Dec 22 11:40:02 Is it really better than the samsung parts for that? Dec 22 11:40:27 beewoolie-afk: no not really Dec 22 11:40:38 jsut we'd already put alot of work into the design Dec 22 11:40:44 I'm not advocating, here, I'm jus curious. Dec 22 11:41:05 If you already have a design, what is it that Sharp isn't doing for you? Dec 22 11:41:32 beewoolie-afk: well we have about 6 designs we were gonna do Dec 22 11:41:42 beewoolie-afk: only 1 survived with the lh79520 Dec 22 11:41:50 beewoolie-afk: the rest got moved to samsung Dec 22 11:42:06 because...? Dec 22 11:42:45 beewoolie-afk: mainly because we weren't getting any help with the initial board bringing Dec 22 11:42:55 beewoolie-afk: there were tons of things that were missing from the docs Dec 22 11:43:02 Hmm. Dec 22 11:43:05 beewoolie-afk: and we had to find them out ourselves Dec 22 11:43:08 LPD or Sharp? Dec 22 11:43:13 beewoolie-afk: sharp Dec 22 11:43:35 Interesting. I've not found the LH79520 docs to be lacking, but I haven't done any hardware design for it. Dec 22 11:43:52 LPD OTOH has lots of gaps in their docs. Dec 22 11:44:21 tons of things, like info on the lcd interface timmings, the sdram inits, problems with the jtag Dec 22 11:45:09 like all the docs are geared for x32 sdram configs, and no information at all for running with a single x16 chip Dec 22 11:45:17 Hmm. I never found those things to be a problem. Dec 22 11:46:06 Ah. Right. I was always working with 32 bit memory. However, I did configure the lh79525 with 16bit memory. The docs were fine, tho. All I needed was a schematic so that I could get the right signals setup. Dec 22 11:46:47 prpplague: BTW, I did mention that I'd spoken with a customer who had dropped the LH chips because of these kinds of problems. Dec 22 11:47:03 beewoolie-afk: and what did they say? Dec 22 11:48:18 There was no specific response. They acknowledged that a transition on the marketing team had left some gaps in support. Dec 22 11:48:41 They said that they've added a forum to help customers get their questions answered. Dec 22 11:48:42 beewoolie-afk: i've got like 10 posts on their forum for the lh79520, all of which they say they will fix in the docs, however almost 9 months later, the docs haven't been updated Dec 22 11:48:50 On the Sharp site? Dec 22 11:48:53 yea Dec 22 11:49:06 Hmm. I'll look at them and see what is up. Dec 22 11:50:29 just have a look at how many times the sdram init problems are posted there Dec 22 11:51:03 beewoolie-afk: kinda like this one - http://forums.sharpsma.com/viewtopic.php?t=168 Dec 22 11:53:17 beewoolie-afk: btw, if it seems i'm blasting you for sharp's mistakes, i don't mean it Dec 22 11:53:39 NP. I know these people, so I think the situation looks different to me. Dec 22 11:53:45 Let me see if I understand the issue. Dec 22 11:53:47 ...in this case. Dec 22 11:54:08 When you use 16 bit SDRAM, you need to double the size of a burst because you need to fill a cache line. Dec 22 11:54:54 that is correct Dec 22 11:55:18 The burst of 4 is 4 accesses at the bus width. 128 bits or 16 bytes. On a 16 bit bus, the burst has to be 8 so that you get 16 bytes. Dec 22 11:55:36 beewoolie-afk: thats my understanding Dec 22 11:55:56 I guess that I don't see this as a big omission on their part. Dec 22 11:56:23 After all, that is what a burst is about. Dec 22 11:56:55 beewoolie-afk: well, you see, when i was running into trouble during the init problem i post and called about 20 times about the fact there was no example about how to calculate the proper burst rate Dec 22 11:57:07 beewoolie-afk: only after i posted the answer did someone respond Dec 22 11:57:24 That is definitely not good. Dec 22 11:57:32 beewoolie-afk: if you look at the sa-1110 manual Dec 22 11:57:42 beewoolie-afk: every register has a equation listed Dec 22 11:57:46 Sure, I can see how this would be frustrating. Dec 22 11:57:54 beewoolie-afk: there is no question about to calculate the value Dec 22 11:58:09 They added this sort of thing to the 79524 manual. Dec 22 11:58:23 In fact, they have a table with lots of memory combinations. Dec 22 11:58:36 beewoolie-afk: every one of them 32-bit Dec 22 11:58:49 beewoolie-afk: i dare you to find a 16-bit example Dec 22 11:58:56 beewoolie-afk: http://forums.sharpsma.com/viewtopic.php?t=142 Dec 22 11:59:01 beewoolie-afk: there is my first post Dec 22 11:59:18 beewoolie-afk: i sent out schematic and info as posted, never got a response Dec 22 11:59:32 beewoolie-afk: called them, they said they had gotten the info and would look Dec 22 11:59:37 beewoolie-afk: never heard a word back Dec 22 11:59:40 Look at page 7-20 of the lh79524 manual. Dec 22 11:59:51 beewoolie-afk: called almost every day for 2 weeks Dec 22 12:00:09 prpplague: I think that they got the message...eventually. Dec 22 12:00:33 * prpplague grabs the 79524 manual even though it didn't exist at the time and we using the 79520 Dec 22 12:01:05 prpplague: I'm not saying that they were on it when you had your trouble. I'm just saying that they did get the message eventually. Dec 22 12:01:22 In the dialog that you had on the forum, you got a reply from Doug Jones. He was the fellow who hired me. Dec 22 12:02:00 He was consciencious and put a lot of effort into helping customers get what they needed. Dec 22 12:03:05 beewoolie-afk: granted that that was pretty much end of us trying to get help from sharp, as we had already made the decision not to use sharp for the other projects Dec 22 12:03:18 Understood. Dec 22 12:03:20 beewoolie-afk: they may have changed, i don't know Dec 22 12:03:38 Well, I wouldn't say they are 100%. Dec 22 12:04:05 beewoolie-afk: have a look on the forum about how many ppl are still having jtag issues Dec 22 12:04:06 I suspect that they're like a lot of places where the only advocates the customers have are the sales folks. Dec 22 12:04:13 will do... Dec 22 12:04:51 beewoolie-afk: it wouldn't take sharp much to post a nice linux wiggler app that would be generic enough to fit alot of the flash needs Dec 22 12:06:46 It would require someone writing one. Dec 22 12:06:49 beewoolie-afk: i mean, think of it like my boss, where his questions to me are like, "well if we are having this much trouble jtag flashing this device during development, what kind of problems will we have during production?" Dec 22 12:07:21 it's funny that those folks are having these problems because I know that the OCD and BDI software work. Dec 22 12:07:35 beewoolie-afk: duh, i already have one that i've sent to lij Dec 22 12:08:04 beewoolie-afk: we there are a ton of issues with the lh7952x jtag Dec 22 12:08:26 Yeah. I get that. It also isn't quite fair since it's like saying: "I'm driving this car with home-made gasoline and it's missing and sputtering. What other sorts of problems will I have?" Dec 22 12:08:47 I'll as Jun about it... Dec 22 12:08:52 * beewoolie-afk sending email Dec 22 12:09:21 beewoolie-afk: sorry that doesn't fly with me due to the fact that if the lh79520 jtag was IEEE compliant these wouldn't be issues Dec 22 12:09:53 all of the issues are due to that fact Dec 22 12:09:56 As I said, it works with the ARM hardware, the BDI hardware, and the Wiggler's software. Dec 22 12:10:11 so, I was never aware of an issue. Dec 22 12:11:07 beewoolie-afk: yea, thats because sharp worked with them to fix the issue(as i understand it from a friend who works a macgraigor) Dec 22 12:11:45 prpplague: They didn't do anything special for Abatron. Dec 22 12:12:24 are the sharp arms -S cores? or hard cores? Dec 22 12:12:36 vmaster: Don't know. Dec 22 12:12:47 vmaster: Where would one find that info? Dec 22 12:12:52 beewoolie-afk: i don't see how they couldn't have Dec 22 12:13:13 mhh, what core do they use? Dec 22 12:13:19 beewoolie-afk: http://document.sharpsma.com/Handler/getfile_handler2.asp?nodeid=Documents&fileid=88EDF5CF792946AF93D03CEC1DF1872E Dec 22 12:13:29 prpplague: I'm certain that Abatron hadn't heard of the LH79520 before I bought a BDI2000 to use with it. Dec 22 12:14:02 beewoolie-afk: hmm, and it worked out of the box? Dec 22 12:14:33 prpplague: That page isn't working for me. Dec 22 12:14:44 and yeah, it worked out of the box. Dec 22 12:14:54 beewoolie-afk: thats interesting Dec 22 12:15:18 I mean, I had to ask for help with the config because I was being a dumb-ass. But they made no FW changes for me. Dec 22 12:18:07 beewoolie-afk: hmm, it seems i can't directly link to that doc Dec 22 12:18:21 How did you get there? Dec 22 12:18:28 beewoolie-afk: http://tinyurl.com/dn6b8 Dec 22 12:18:31 try that Dec 22 12:18:34 should be a pdf Dec 22 12:19:00 It wants a login but it doesn't accept my sharp login. Dec 22 12:19:21 beewoolie-afk: its on the lh79520 page, Errata 07/11/05 Dec 22 12:20:16 beewoolie-afk: my lauterbach had to have a firmware update for it work with the lh79520 for which they reference the errata Dec 22 12:21:06 they only talk about boundary scanning Dec 22 12:21:32 beewoolie-afk: go there first: http://www.sharpsma.com/part.php?PartID=173 Dec 22 12:21:50 looks like a deep-link protection Dec 22 12:21:52 I got the erratum. Dec 22 12:21:57 this is the main kicker "the LH79520 outputs change state on the rising edge of TCK. The IEEE 1149.1 specification states that signals driventhe LH79520 outputs change state on the rising edge of TCK out of the component via the boundary scan logic must change state relative to the falling edge of TCK." Dec 22 12:22:41 oooh, bad bad Dec 22 12:22:43 thats the item that has broken almost every app i tried Dec 22 12:22:55 you can _not_ change TDO on rising_edge(TCK) Dec 22 12:23:03 hehehe, you shouldn't at least Dec 22 12:23:08 that's when the tap master samples tdo Dec 22 12:23:11 the other one is this "The test mode select pins nTSTA, TEST1, and TEST2 are included in the boundary scan chain. These pins, when set to 101 (respectively) before a rising edge of nRESETIN, are used to activate the JTAG interface. However, boundary scan vectors that are generated using the BSDL file float these inputs because they are part of the boundary scan chain. As a result, the JTAG interface enters an undefined state when the vectors a Dec 22 12:23:11 re exercised, causing the boundary scan to be non-functional." Dec 22 12:23:36 prpplague: yes, but that only refers to boundary-scanning, i.e. around all the device pins Dec 22 12:23:36 I read that. Dec 22 12:23:37 vmaster: tap master can sample TDO anywhere between falling_edge + delta and raising_edge - delta Dec 22 12:23:54 vmaster: correct which is how alot of flash apps work Dec 22 12:24:19 vmaster: AND my last check the BSDL file that sharp distrubtes DOESN'T have those pins set properly Dec 22 12:24:34 vmaster: actually, anywhere between falling_edge + delta and falling_edge - delta Dec 22 12:24:52 I'll query them about this. It looks like it's just a bug in the silicon. Dec 22 12:25:14 "Solution: Permanent erratum; this will not be fixed." Dec 22 12:25:24 prpplague: as there is a work-around. Dec 22 12:25:35 they can't claim JTAG compliance then Dec 22 12:25:37 prpplague: yeah, but that explains why abatron works Dec 22 12:26:00 beewoolie-afk: exactly my point, if you are gonna use a f'd up jtag, you should atleast provide some code that helps people use your f'd up jtag Dec 22 12:26:22 lennert: the erratum states that they aren't compliant. Dec 22 12:26:42 prpplague: they do indicate some products tha work. Dec 22 12:26:43 beewoolie-afk: but they don't advertise that fact Dec 22 12:26:49 prpplague: It sounds like you really just want to hate them. Dec 22 12:26:57 beewoolie-afk: hehe, no Dec 22 12:27:11 I'm hearing a lot of hate. Dec 22 12:27:36 beewoolie-afk: i got frustrated because i was collecting all this info and trying to help sharp make their items better, and all i got on the other end was a dead line Dec 22 12:27:37 After all, they're not claiming that there are weapons of mass desctruction. Dec 22 12:27:38 i hear a lot of frustration, rather Dec 22 12:28:06 beewoolie-afk: i'm still willing to provide them help Dec 22 12:29:02 prpplague: they really can't do a lot more than admitting that they blew it Dec 22 12:29:35 vmaster: i disagree, providing clear documentation and sample code would go a long way in my book Dec 22 12:30:02 vmaster: and i have a list of about 20 items like this Dec 22 12:30:31 prpplague: yeah, regarding the other things you've mentioned, but not the jtag part Dec 22 12:30:33 Sounds like prplague has some 'letting go' to do. Dec 22 12:30:40 hehe Dec 22 12:30:42 :-) Dec 22 12:30:50 yea i get passionate about stuff Dec 22 12:31:49 philips lpc are just the same - several debugger vendors contacted them about a problem with their devices, but their remarks didn't even make it into the erratas Dec 22 12:32:21 i contacted their support, and they didn't even know there was a problem Dec 22 12:32:36 I believe that a better solution than trying to get the vendors to comply is to produce a hardware/software jtag solution... Dec 22 12:32:43 Oh right. this is openjtag.... Dec 22 12:32:46 ;) Dec 22 12:32:52 hehe Dec 22 12:33:18 beewoolie-afk: its not like off the shelf code doesn't work, just got to make the nessary changes Dec 22 13:12:45 beewoolie-afk: didn't mean to sound so harsh Dec 22 13:13:02 beewoolie-afk: hehe, you still get an lnode80 if you want one Dec 22 13:13:26 prpplague: NP. No offense taken. Dec 22 13:13:30 And yeah, I'd like one. Dec 22 13:13:48 It means I can test on another target. Dec 22 13:15:11 beewoolie-afk: the entire main board is just a little bigger than a 802.11b CF card Dec 22 13:15:30 Nice. Dec 22 13:16:36 beewoolie-afk: has two leds, reset button, 2x7 jtag header, mmc/sd slot, onboard regulator, 4MB flash, 16MB ram, all with .1" headers Dec 22 13:17:10 What about that MMC code you were going to send my way? Dec 22 13:17:30 beewoolie-afk: thought i did already Dec 22 13:17:36 * prpplague double checks Dec 22 13:17:37 Let me check again... Dec 22 13:18:33 i show i sent it, right before i sent the first round of s3c2410 patches for apex Dec 22 13:18:44 i think that might have been when you had some email server issue Dec 22 13:18:49 i'll resend Dec 22 13:18:56 prpplague: That could be. Dec 22 13:19:04 Last I have is that you wanted to wait. Dec 22 13:20:43 ahh Dec 22 13:21:04 i'll resend it Dec 22 13:21:15 its still pretty ugly but you can get a feel for it Dec 22 13:21:27 I just want to see something that works. Dec 22 13:21:29 its a nasty hack right now Dec 22 13:21:43 I spend some time trying to get the ID phase working and never succeeded. Dec 22 13:28:22 <[g2]> hey beewoolie-afk ! Dec 22 13:28:35 [g2]: yo Dec 22 13:29:26 <[g2]> happy holidays beewoolie-afk Dec 22 13:29:37 Likewise. Dec 22 13:29:42 <[g2]> thx Dec 22 13:29:53 How's the fam? Dec 22 13:30:03 <[g2]> fine thx Dec 22 13:30:23 <[g2]> I hope you and yours are well too Dec 22 13:30:49 Whenever I say that I've been on chat, doll face asks if I've talked to you, [g2]. Dec 22 13:31:02 <[g2]> heh Dec 22 13:31:23 <[g2]> I hope doll face it well too Dec 22 13:31:27 Indeed. Dec 22 13:31:47 She's become a fan of the TV show Lost. Dec 22 13:32:22 <[g2]> I haven't seen that one Dec 22 13:32:38 beewoolie-afk: interesting show Dec 22 13:33:15 Reminds me of Twin Peaks. spooky at first and then evolves into something seriously wierd. Dec 22 13:33:59 <[g2]> I missed that one too Dec 22 13:34:16 took me some time to get into lost, but then i started missing it after the season ended Dec 22 13:34:22 <[g2]> beewoolie-afk before I forget there some new hw will be playing with Dec 22 13:34:46 [g2]: What Dec 22 13:34:51 [g2]: what's up? Dec 22 13:34:56 <[g2]> www.glomationinc.com Dec 22 13:35:01 <[g2]> 9312-sx Dec 22 13:35:09 <[g2]> I've got a hw sponsor Dec 22 13:35:18 <[g2]> we'll be getting 5 boards to begin with Dec 22 13:35:21 wheee Dec 22 13:35:29 [g2]: my head is spinning. Dec 22 13:35:43 <[g2]> glomation isn't the sponsor Dec 22 13:35:53 [g2]: now it's really spinning. Dec 22 13:35:59 <[g2]> heh Dec 22 13:36:49 <[g2]> lennert dyoung-away rwhitby me dwery are in the first pass Dec 22 13:37:23 <[g2]> If the boards look good I can probably get some more (I'm sure you'd qualify if you wanted one :) Dec 22 13:37:29 I do. Dec 22 13:37:42 <[g2]> well we have to see how the boards look Dec 22 13:37:49 I like the slugs, but the lack of RAM is causing some grief. Dec 22 13:38:05 <[g2]> The Loft's have 64MB :) Dec 22 13:38:07 I've been unable to get them stable with more than file sharing. Dec 22 13:38:16 <[g2]> I've got a 128 and 256MB sample too Dec 22 13:38:23 [g2]: I was hoping for one of those but for some reason... Dec 22 13:38:35 My slug crashes when I added NTP. Dec 22 13:38:56 <[g2]> hmmm odd, I've run NTP for a long time Dec 22 13:39:01 I know. Dec 22 13:39:13 <[g2]> You'll have to let me know your setup Dec 22 13:39:14 It doesn't crash right away, but it does so reliably. Dec 22 13:39:23 <[g2]> do you do DHCP ? Dec 22 13:39:28 I need to upgrade the kernel. Dec 22 13:39:28 <[g2]> with Debian ? Dec 22 13:39:35 Debian, but older kernel. Dec 22 13:39:37 No dhcp. Dec 22 13:39:47 I need modules that the package feed doesn't supply. Dec 22 13:39:56 I've been doing other things so I haven't had time to build a newer kernel. Dec 22 13:39:58 <[g2]> Ok... I was seeing an issue with the dhcpclient after like 12-14 hours Dec 22 13:40:13 Not good either. Dec 22 13:40:28 <[g2]> it's with the IAL driver Dec 22 13:40:33 <[g2]> IAL/CSR Dec 22 13:40:37 <[g2]> it's not that bad actually Dec 22 13:41:12 I expect more. Dec 22 13:41:18 <[g2]> I haven't tried with the latest versions of the drivers Dec 22 13:41:28 beewoolie-afk: is your ram buggered? Dec 22 13:41:31 <[g2]> and it only lock up the ethernet on that port Dec 22 13:41:39 beewoolie-afk: my slug's ram is buggered Dec 22 13:41:50 lennert: I don't think so. It runs fine for weeks without NTP. Dec 22 13:41:59 mine too Dec 22 13:42:01 I've been writing a lot of data and never had a glitch. Dec 22 13:42:02 <[g2]> lennert what's up with your RAM ? Dec 22 13:42:03 mine runs fine for weeks Dec 22 13:42:14 but if i untar a kernel tarball, i get bzip2 corruption errors Dec 22 13:42:14 I gues we need a ram test? Dec 22 13:42:17 yeah Dec 22 13:42:26 I don't see anything like that. Dec 22 13:42:33 I unpack lots of packages. Dec 22 13:42:41 okay, it's probably okay then Dec 22 13:42:47 I'll try a kernel tarball just for kicks. Dec 22 13:43:24 <[g2]> lennert are you still playing with the slug ? Dec 22 13:45:46 <[g2]> beewoolie-afk BTW lennert gave me some code to test the speed of the GPIO and it looks like we could bit bang JTAG pretty fast from a Loft Dec 22 13:46:05 Interesting. Dec 22 13:46:12 Faster than USB dongle trick? Dec 22 13:46:36 <[g2]> Well writing is at 11M/s Dec 22 13:46:48 Holymackerel! Dec 22 13:46:48 <[g2]> so on writes we could toggle around 4.5M Dec 22 13:47:02 <[g2]> M/s Dec 22 13:47:15 IIRC, that's about the same speed as the USB device. Dec 22 13:47:23 [g2]: i don't play with my slug as much as i should.. as it's kind of buggered Dec 22 13:47:25 Only advantage of USB would be overall cost. Dec 22 13:47:32 [g2]: it kind of sucks that my last slug also had buggered memory :( Dec 22 13:47:48 beewoolie-afk: the ft2232c has a tck maximum of 6mhz, but effectively you wont get more than 1-2mhz Dec 22 13:47:57 [g2]: i'd be happy to send it to someone who can do something else with it Dec 22 13:48:10 <[g2]> maybe dyoung-away Dec 22 13:48:28 vmaster: I thought that it could write the scan chain that fast, ~5MHz. Dec 22 13:48:41 <[g2]> vmaster also pointed out that the phillps can bit bang around 15M Dec 22 13:48:42 vmaster: are you talking about the scan-chain delay? Dec 22 13:49:06 beewoolie-afk: i'm talking about real-life debugging of arm7 and arm9 cores Dec 22 13:49:18 beewoolie-afk: at some point, you're waiting for the results Dec 22 13:49:49 vmaster: I was just asking about an equivalent comparison. Dec 22 13:50:30 beewoolie-afk: hmm? Dec 22 13:50:32 vmaster: if the scan chain clock is XMhz on one and YMHz on another, doesn't that give similar results? Or are you adding in a latency for the USB transfer? Dec 22 13:50:59 beewoolie-afk: yes, it's mostly latency Dec 22 13:51:27 Hmm. Dec 22 13:51:59 Seems like we still haven't found the right combination of inexpensive hardware and efficient (very fast) operation. Dec 22 13:52:17 beewoolie-afk: i've heard from olimex that they're working on atmel sam7 based jtag device Dec 22 13:52:25 [g2]: Also, doesn't the IEEE standard limit the clock speed to about 8MHz? Dec 22 13:52:48 beewoolie-afk: no, the standards places no limits at all Dec 22 13:53:02 beewoolie-afk: individual devices specify minimum high-low times for the tck Dec 22 13:53:06 <[g2]> beewoolie-afk dunno, I here the IXP42x's max out around 10M Dec 22 13:53:24 vmaster: ah, good. Dec 22 13:54:54 the at91rm9200 specifies a minimum tck period of 20ns Dec 22 13:55:01 that would be 50mhz jtagging Dec 22 13:55:11 <[g2]> that'd be fun :) Dec 22 13:55:25 vmaster: So it would be nice if we could get something capable of that sort of speed. Dec 22 13:55:37 <[g2]> lennert do you know voltage ouput levels for the S3 I/O pins ? Dec 22 13:55:51 Which is why, IIRC, we were looking into the sparton. Dec 22 13:56:13 [g2]: 3.3V ? Dec 22 13:56:20 3.3v Dec 22 13:56:24 max Dec 22 13:56:26 <[g2]> hey Dec 22 13:56:29 <[g2]> hey ka6sox-office Dec 22 13:56:34 heya Dec 22 13:56:48 [g2]: the pins are 5V tolerant though, i think Dec 22 13:56:58 sorta here...just trying to unbury myself from all this paperwork that has been piling up in my absense Dec 22 13:57:00 <[g2]> Ok.. so Digilent has these flying lead connectors for < $5 Dec 22 13:57:10 yep Dec 22 13:57:17 I use flying lead. Dec 22 13:57:22 <[g2]> They plug right into the board Dec 22 13:57:47 <[g2]> I think we could plug from the S3 I/O pins straight into the Loft with the S3 board Dec 22 13:58:01 <[g2]> and 6 or 12 leads would do the trick Dec 22 13:58:14 heh, 50mhz with flying leads? Dec 22 13:58:26 <[g2]> vmaster I was thinking 10M Dec 22 13:58:27 impedance... Dec 22 13:58:57 arm's realview tools use differential signaling down to the target connector Dec 22 13:59:03 iirc Dec 22 14:00:03 <[g2]> lennert what about impedance Dec 22 14:00:35 50mhz can be problematic if you don't use impedance matched stuff Dec 22 14:00:59 <[g2]> Oh sure, but what about 5-10Mhz Dec 22 14:01:18 the cable is critical at that speed Dec 22 14:01:27 i've had problems with cable-length at 6mhz Dec 22 14:01:50 <[g2]> vmaster how long a cable are we talking ? Dec 22 14:02:01 [g2]: 20cm caused problems, <10cm worked fine Dec 22 14:02:15 that was the standard arm layout, e.g. a gnd between every signal Dec 22 14:02:55 on the lpc mailing list, people report problems even at wiggler speed Dec 22 14:03:40 bdi comes with ~10cm cable Dec 22 14:03:51 <[g2]> 10cm is like 3.9 Inches right ? Dec 22 14:04:04 yeah Dec 22 14:06:59 <[g2]> one of the problems with the impedeance mismatch is the relfection of the signal back down the line right ? Dec 22 14:07:15 yeah Dec 22 14:09:59 anyway, this can be solved Dec 22 14:10:05 but first we need to achieve that speed Dec 22 14:10:10 <[g2]> lennert do you know offhand how much power the S3 draws ? Dec 22 14:10:26 kind of depends what you do with it :) Dec 22 14:11:08 there is a power analyser with in the xilinx webpack Dec 22 14:12:59 <[g2]> I'm wondering if put an S3 right on top (or .5/1 inch) away and powered it from the board Dec 22 14:13:20 lennert: BTW, no problems with untar of kernel. I wish that that had been the case. Dec 22 14:13:28 <[g2]> And then talked some other protocol off S3 Dec 22 14:13:43 beewoolie-afk: you wished your device had broken ram? ;) Dec 22 14:13:44 beewoolie-afk: i can't even complete "bunzip2 < linux-2.6.14.tar.bz2 > /dev/null" successfully once Dec 22 14:13:59 vmaster: yeah. That would have explained my problems. :-) Dec 22 14:14:31 <[g2]> beewoolie-afk you are probably no aware of jacques work lately Dec 22 14:14:38 <[g2]> s/no/not Dec 22 14:14:42 [g2]: hit me Dec 22 14:14:58 <[g2]> I think he was running debian arm in QEMU Dec 22 14:15:07 I read about thay. Dec 22 14:15:21 <[g2]> and his perl tests ran faster than a 266Mhz slug Dec 22 14:15:24 I saw a thread about building packages that way. Dec 22 14:15:36 Kinda sweet. Dec 22 14:16:12 <[g2]> I think that means the 95% of all program synchronuous problems can be resolved in the emulator if reproducable Dec 22 14:16:54 Hmm. Depends on what type of problem. I thought that most of our troubles have been with the IAL and the NPEs. Dec 22 14:17:23 <[g2]> I think there's only 1 outstanding issue there Dec 22 14:17:29 <[g2]> the one I mentioned earlier Dec 22 14:17:40 <[g2]> other than just general Open Source issues Dec 22 14:17:56 i think those are being worked on Dec 22 14:18:04 I haven't been aware of many (if any) ARM specific issues. In fact, that has been discussed on the Debian lists. Dec 22 14:18:04 <[g2]> The glomation board is _fully_ open from my understanding Dec 22 14:18:15 Most of the arch issues are really issues with *all* arches. Dec 22 14:18:26 Which CPU is glomation using? Dec 22 14:18:33 <[g2]> Cirrus 93xx Dec 22 14:18:51 I think I got a link about that... Dec 22 14:19:06 <[g2]> ARM 920 based with a hw FP instructions Dec 22 14:19:21 Yep. My stepdad asked about this. Dec 22 14:19:27 Nice. Dec 22 14:19:55 He wondered if it wouldn't be adequate for his needs. They have some reasonably priced boards. Dec 22 14:20:23 <[g2]> how the ethernet support in APEX going ? Dec 22 14:20:49 [g2]: I've got support for the SMC chip and the Sharp MAC. Dec 22 14:21:07 The IAL isn't happening because it's such a load of work *and* because of the licensing. Dec 22 14:21:26 <[g2]> I mean for the glomation board :) Dec 22 14:21:38 beewoolie-afk: there is a copy of the IAL available under the GPL and/or BSD, i've heard Dec 22 14:21:40 I was thinking of buying one of the glomation boards so that I could implement support for the cs8900. Dec 22 14:22:00 lennert: DOH! really? Does it work in the kernel? Dec 22 14:22:13 beewoolie-afk: see netdev@vger.kernel.org archives of today Dec 22 14:22:19 beewoolie-afk: some guy wants to clean it up and integrate it Dec 22 14:22:26 That would rock! Dec 22 14:22:37 beewoolie-afk: still a bunch of work though, as you said Dec 22 14:22:40 * beewoolie-afk is rummaging Dec 22 14:22:57 The licensing is the worst part. Once it's open, we can do the work once. Dec 22 14:23:35 <[g2]> lennert knows all the juicey stuff :) Dec 22 14:24:24 [g2]: i'm on all the interesting mailing lists :) Dec 22 14:24:34 <[g2]> it is kinda cool that the slug has 3 cores on it and the loft had 4 cores really Dec 22 14:24:52 <[g2]> lennert it's that you actuallly _read_ the ml's :) Dec 22 14:26:53 lennert: You know, I saw your post about putting it in the kernel and I didn't believe it. Dec 22 14:27:11 I thought that I was hallucinating about IAL code going into the kernel. Dec 22 14:27:19 beewoolie-afk: hehe Dec 22 14:27:33 beewoolie-afk: well, the IAL code is kind of.. how to say.. Dec 22 14:27:41 DON'T! Dec 22 14:27:41 beewoolie-afk: 'not up to the standards' Dec 22 14:27:46 EEK! Dec 22 14:27:48 oops, sorry, too late Dec 22 14:28:00 The secret has been revealed. Dec 22 14:28:14 well Dec 22 14:28:16 I'll take another look. Do I just download from the normal place at intel? Dec 22 14:28:21 i don't think it was a well-kept secret Dec 22 14:28:27 STOP! Dec 22 14:28:28 not sure, i've not looked into it for a while Dec 22 14:28:38 What version are you looking at? Dec 22 14:28:46 Or, are they looking at? Dec 22 14:28:52 not sure Dec 22 14:28:54 it should be in the thread Dec 22 14:29:35 He just says that he 'has a version'. Dec 22 14:29:45 one guy 'has a version' under BSD Dec 22 14:29:57 Perhaps I should wait until he posts apatch. Dec 22 14:30:03 the other guy says that there's a downloadable version on the intel site that is eCos-GPL licensed, and gives a link (iirc) Dec 22 14:31:08 <[g2]> I'm looking on lkml ? what's the date ? Dec 22 14:31:14 [g2]: netdev Dec 22 14:31:24 I've been looking at the netdev archive in MARC. Dec 22 14:32:05 <[g2]> that'll make a big difference :) Dec 22 14:32:10 <[g2]> thx guyx Dec 22 14:32:15 http://marc.theaimsgroup.com/?t=113516654600005&r=1&w=2 Dec 22 14:32:26 http://marc.theaimsgroup.com/?l=linux-netdev&m=113524802815054&w=2 says: Dec 22 14:32:31 FWIW, the parts of the IXP400 Access Library relevant to ethernet Dec 22 14:32:32 (except the microcode which could be made into a binary blob for loading Dec 22 14:32:32 by the firmware loader) are available (from Intel) under the eCos Dec 22 14:32:32 variant of the GPL. Dec 22 14:32:34 --- Dec 22 14:32:36 but he didn't give a link Dec 22 14:33:15 <[g2]> I had heard that Intel was going to do that Dec 22 14:33:53 <[g2]> but it would be great if that was really true Dec 22 14:38:58 okay, help me out.. fluke 112 or fluke 177? :) Dec 22 14:40:20 <[g2]> I'm not much help there Dec 22 14:40:54 lennert: 177 Dec 22 14:41:13 lennert: Actually, I bought the 179 Dec 22 14:41:24 One of the best tools I have. Dec 22 14:41:26 beewoolie-afk: why the 177? Dec 22 14:42:13 177 is E70 more.. for 0.09% accuracy instead of 0.7%, 50Mohm instead of 40, 100kHz instead of 50, and lifetime warranty Dec 22 14:43:11 The data sheet shows the accuracy being thesame. Dec 22 14:43:15 I got a temp probe. Dec 22 14:43:25 sorry, i mean, the difference between the 177 and the 112 Dec 22 14:43:35 http://www.fluke.com/Download/DigitalMultimeters/2155a.pdf Dec 22 14:43:38 the 179 has a temperature probe while the 177 hasn't Dec 22 14:43:44 accuracy of the 177 and 179 is the same Dec 22 14:43:49 Oh. I don'tknow the 112. Dec 22 14:43:52 let me see... Dec 22 14:43:52 the 112 has 0.7% accuracy, the 177 has 0.09% Dec 22 14:44:12 i'm trying to convince my wife as well as myself.. Dec 22 14:44:23 my wife is already convinced, i think, i just need to convince myself.. :) Dec 22 14:45:45 The 17x series has hold measurement features. I use them more than I thought I would. Dec 22 14:45:53 ah Dec 22 14:46:23 Also MIN MAX AVG. Dec 22 14:46:34 right Dec 22 14:46:36 that sounds useful Dec 22 14:47:05 What does the wife say? Dec 22 14:47:16 they all look good :) Dec 22 14:47:27 and i must say i agree Dec 22 14:47:27 BTW, backlight is really important, too. Both have it. Dec 22 14:48:20 If I were at a company and I was buying 100 units and I didn't think we needed HOLD,then I'd probably buy the 112. For me, I hang onto these things for a long time. Dec 22 14:48:41 $US80 is nothing compared to frustration of not having it when I need it. Dec 22 14:48:48 yeah Dec 22 14:49:03 they sell DMMs for as low as E5 here Dec 22 14:49:09