**** BEGIN LOGGING AT Fri Dec 23 02:59:56 2005 Dec 23 05:12:10 ok, so, toggling the GPIOs on the pd10000 does actually work Dec 23 05:12:43 i.e., the levels change Dec 23 05:25:31 pd10000? Dec 23 05:35:37 via epia pd10000 mini-itx mainboard :) Dec 23 05:36:42 it has 8 exposed gpios Dec 23 05:37:59 is it hard to make a parallel cable 3 to parallel cable 4 adapter (so that i can program a device with a PC4 socket with a PC3 cable)? Dec 23 05:38:03 it doesn't seem hard.. Dec 23 07:30:33 the xilinx system ace docs keep saying that it's possible to configure the CFG JTAG port from the TST JTAG port Dec 23 07:30:44 but i can't find any special instruction codes in the .bsd file Dec 23 07:30:51 and there's no other docs about it Dec 23 07:31:24 i really hope there's another way of doing it than using SAMPLE/EXTEST.. but i guess not :-/ Dec 23 07:35:29 ah Dec 23 07:35:34 i think ds080 has a typo on page 33 Dec 23 07:35:50 CFGTDO in figure 17 should be TSTTDO, and TSTTDO should be CFGTDO Dec 23 07:35:58 if you look at it that way, it's indeed not possible directly Dec 23 07:46:44 or, well.. Dec 23 07:46:50 the usb2 interface cpld is on the 'test jtag chain' side Dec 23 07:46:57 and impact does recognise it Dec 23 07:47:04 so maybe there's some sekrit opcodes in the system ace Dec 23 08:06:20 i wonder if they disconnect the PC4 when the XC2C256 is active Dec 23 08:10:21 what board are you talking about? the digilent spartan3? Dec 23 08:11:03 the digilent xupv2p Dec 23 08:11:14 i.e. virtex2 pro Dec 23 10:06:59 morning Dec 23 10:07:13 ep1220: ping? Dec 23 10:09:03 okay, the platform usb cable can do it Dec 23 10:09:10 it's just not documented (as usual, with xilinx) Dec 23 10:13:30 ah Dec 23 10:13:41 this is the V2Pro? Dec 23 10:13:44 ka6sox: you had your v2pro course yet? Dec 23 10:13:46 yeah Dec 23 10:13:49 the digilent v2p board Dec 23 10:14:06 that course starts jan 6 Dec 23 10:14:11 ah Dec 23 10:14:22 my board is skedded to get here by the 2ond Dec 23 10:14:37 while you're at it, ask them why the bitstream format isn't open ;) Dec 23 10:14:43 ka6sox: which board are you getting? Dec 23 10:14:51 same as you. Dec 23 10:14:54 ah Dec 23 10:14:57 the xupv2p one Dec 23 10:15:00 yep Dec 23 10:15:15 i kind of have it working Dec 23 10:15:38 but it's just so cumbersome to have to load your config to a compatflash card, putt it in the board, put it back, etc. Dec 23 10:17:26 yep. Dec 23 10:17:35 gotta go fix the Water Heater... Dec 23 10:17:52 maybe it has jtag ;) Dec 23 10:20:05 heh, i wouldn't be surpised if it had Dec 23 10:29:23 actually, it's probably easier than i thought Dec 23 11:55:50 so Dec 23 11:55:55 it's not actually clear from the system ace docs Dec 23 11:56:08 but the idea is that you splice the system ace into the middle of your boundary scan chain somewhere Dec 23 11:56:21 when it's not configuring your devices, it'll just pass everything through Dec 23 11:56:36 but if it is configuring your devices, it'll shut off the host side from the fpga side Dec 23 11:57:24 some of the diagrams are faulty too Dec 23 12:30:22 if you live in the U.S., you should check the avnet website Dec 23 12:30:52 Antti Lukas writes in comp.arch.fpga that avnet offers spartan3e boards free of charge Dec 23 12:31:52 yeah, i know Dec 23 12:31:59 i've been trying to get registered with avnet Dec 23 12:32:07 just click on the 'register' link -- you get a 404 page Dec 23 12:32:11 emailed them about it a few days Dec 23 12:32:16 ago, heard nothing yet Dec 23 12:32:44 okay, they fixed it now Dec 23 12:35:40 registration confirmation is still broken Dec 23 13:54:49 i'm about halfway hacking something up that'll sample a couple of pins and send an ethernet packet every second with the values of those pins Dec 23 13:55:20 next step is to also make the virtex receive ethernet frame Dec 23 13:55:21 s Dec 23 16:03:58 [g2]: hey man. Are you going to be using stock part sold by, erm, that other company that uses Cirrus chips? Dec 23 16:10:05 linux doesn't support the cirrus ep9301, does it? Dec 23 16:18:23 If it doesn't,it'll be easy to add. Dec 23 16:18:47 hi, lennert Dec 23 16:18:53 Did you get a DMM? Dec 23 16:19:35 beewoolie-afk: why so? Dec 23 16:19:45 beewoolie-afk: nope, not yet -- can't find it anywhere in the neigbourhood Dec 23 16:19:54 Why will it be easy? Dec 23 16:20:01 yeah, why will it be easy? Dec 23 16:20:19 It's an arm9 core. The only tricky part about these processors is the interrupt handlers. Dec 23 16:20:27 okay.. Dec 23 16:20:30 I've done four already. Dec 23 16:20:47 One more won't be tough. Dec 23 16:21:03 are they different enough to have separate mach-blah dirs? Dec 23 16:21:07 We have a cs8900 driver already. Dec 23 16:21:12 Don't know. Dec 23 16:21:19 it depends on the interrupt handling code. Dec 23 16:21:22 okay, so, where did _you_ put the files? :) Dec 23 16:21:26 or, rather, the core. Dec 23 16:21:33 cs8900 is the ugliest driver i've ever seen.. Dec 23 16:21:33 I support the sharp LH processors. Dec 23 16:21:47 I have two directories, one for the ARM9's and another for the ARM7's. Dec 23 16:21:53 lol Dec 23 16:21:57 I haven't looked at it. Dec 23 16:22:09 cs8900 is like the local town whore Dec 23 16:22:23 everybody keeps adding their hacks to make it work on their boards Dec 23 16:22:25 I have a board here with that chip, but it doesn't recognize my jtag for some reason. Dec 23 16:22:27 so it's ifdef spaghetti Dec 23 16:22:36 Sounds like it needs some love. Dec 23 16:22:41 yeah Dec 23 16:22:45 some rewriting Dec 23 16:22:50 That too. Dec 23 16:23:14 there's an ixdp board with a cs8900 Dec 23 16:23:25 normally the registers have 16bit spacing Dec 23 16:23:31 but on the ixdp2401, they have 32bit spacing Dec 23 16:23:41 there's some real ugly hacks for that Dec 23 16:23:42 prplague has a cs8900 using board, too. Dec 23 16:23:50 i don't have any cs8900 board myself.. Dec 23 16:23:57 It should be done as it is in the smc91x driver. Dec 23 16:23:58 i just work on deepak's ixdps remotely Dec 23 16:24:06 All of the wierd stuff is in some macros. Dec 23 16:24:11 The code is relatively simple. Dec 23 16:24:14 yeah Dec 23 16:24:20 well Dec 23 16:24:21 :) Dec 23 16:24:30 did you have breakfast yet? Dec 23 16:24:39 Uh, huh. And lunch. Dec 23 16:24:47 okay Dec 23 16:24:51 I am not talking about the cs8900 at the moment. Dec 23 16:24:53 check include/asm-arm/arch-ixp2000/io.h Dec 23 16:25:03 grep for #ifdef CONFIG_ARCH_IXDP2X01 Dec 23 16:25:04 The smc91x driver is, more or less, a straightforward driver. Dec 23 16:25:09 (and try to keep your food in..) Dec 23 16:25:12 * beewoolie-afk starts grepping. Dec 23 16:25:40 note that this code gets run on _every_ I/O access on these platforms Dec 23 16:25:48 Oh good god NO!!!!! Dec 23 16:25:53 mu ha ha ha ha Dec 23 16:26:02 * lennert whistles Dec 23 16:26:07 BAD BAD is should burn in a pit of lead. Dec 23 16:26:13 yup Dec 23 16:26:14 ACK MY EYES Dec 23 16:26:18 lol Dec 23 16:26:20 hahaha Dec 23 16:26:23 OK. That is inexcusable Dec 23 16:26:26 yeah Dec 23 16:26:31 it was either that or hack cs8900 even further Dec 23 16:26:52 If I have a board, I'll take a look at writing some new code for it. Dec 23 16:26:53 http://ixp2xxx.sourceforge.net/cs89x0/ tries to clean it up Dec 23 16:27:02 and it moves the ixdp-specific stuff into cs89x0 itself Dec 23 16:27:08 It should be a generic driver, like the smc91x. Dec 23 16:27:13 but it didn't get accepted upstream Dec 23 16:27:34 Sounds like a proprietary issue. IOW, someone wants to *own* it. Dec 23 16:27:44 There have been several rewrites of bad drivers. Dec 23 16:27:54 it doesn't really have a maintainer, too Dec 23 16:28:06 i would maintain it if i had hardware Dec 23 16:28:10 but i don't Dec 23 16:28:15 and i need to bug deepak to check stuff for me Dec 23 16:28:35 As is usually the case, especially for embedded stuff. Dec 23 16:28:42 jeff garzik just applies reasonable patcges Dec 23 16:28:54 mine probably looked too big Dec 23 16:29:01 and noone acked it Dec 23 16:29:06 yeah Dec 23 16:29:16 Sometimes it takes some persistence. Dec 23 16:29:26 true Dec 23 16:29:27 That is, sometimes it takes more than is reasonable. Dec 23 16:29:36 But, that's how these things seem to work. Dec 23 16:29:45 but if i'm going to put effort into it anyway.. i might as well rewrite it as a platform device Dec 23 16:29:54 and be its maintainer Dec 23 16:30:02 Right. Dec 23 16:30:05 (or someone else) Dec 23 16:30:23 I have the data sheet somewhere... Dec 23 16:30:34 AOL Dec 23 16:31:25 Yep, it's right here. Dec 23 16:31:34 America OnLine? Dec 23 16:31:43 AOL = "Me too!" Dec 23 16:32:11 i.e. i have the datasheet somewhere too Dec 23 16:32:41 Yeah. I'll give it a read. usually, I start with a driver for APEX and work from that. Dec 23 16:36:24 I remember this chip. I had a problem with one where the eeprom wasn' t properly programmed. Dec 23 16:37:23 eep! rom Dec 24 02:37:48 lennert: avnet registration works if you replace the https:// with http:// **** ENDING LOGGING AT Sat Dec 24 02:59:57 2005