01:46:17 | Sammy | hello all |
01:46:45 | Sammy | BZFlag : you get my mail ? :) |
01:47:12 | Sammy | er maybe got |
01:49:08 | Sammy | oh... I "got" your mail :) |
02:00:19 | Sammy | BZFlag: this is so strange , first time order come from that I don't know the wire need to charge another $, so that short 6 dollars , and sorry about that |
02:02:00 | Sammy | but this time I'd be careful about the wire charge , so I add 6 dollars ,and think that should be right , but why this time need to charge 30 dollars ? |
02:05:54 | Sammy | I think is that charge money choose from how much I snet ? means if I sent more it's will charge more ? |
02:06:58 | Sammy | if that so , then that we will never know how much it's will charge on me .... |
02:07:45 | Sammy | and I will always sorry about that everytime ..... it should't be that :( |
02:13:38 | Sammy | ahhh, Sorry about the muttering here ...... |
02:26:13 | BZFlag | no problem. it's close enough. |
02:26:27 | BZFlag | preping the order tonight. ship tomorrow morning. thanx! |
05:30:43 | BZFlag | Sammy: you here? |
05:30:51 | BZFlag | did you ever send me an address? |
07:08:20 | harsha | need some help regrading the /proc file system |
07:08:28 | harsha | can someone respond |
10:08:11 | seletz | morning |
10:10:06 | Sammy | howdy seletz :) |
10:10:30 | seletz | Sammy: hi, how's it going? |
10:11:19 | Sammy | fine , now is PM 6:00 here comes the weekend :) |
10:11:24 | Sammy | and you ? |
10:11:53 | seletz | grrr, the whole day to go, and tomorrow. |
10:12:09 | seletz | But i'm fine, thanks. |
10:13:23 | seletz | has to port 60.000+ LOC to AIX this weekend |
10:13:57 | Sammy | ||o|| |
11:01:00 | Sammy | bye seletz & all ... |
11:01:10 | Sammy | goes home.. |
13:14:23 | erikm | hello * |
17:21:42 | akira|work | hi |
17:22:02 | | what's up, erikm |
17:22:02 | erikm | hello |
17:22:10 | erikm | stupid bots |
17:31:52 | prpplague | erikm: hows life today? break any code? i've already torn up two apps today!lol |
17:33:58 | erikm | prpplague: not really. I'm fixing my CVS |
17:34:29 | erikm | russ or russ_: ayt? |
17:34:34 | erikm | s/russ/Russ/ |
17:42:28 | akira|work | anyone seen signs of life from seletz today ? :) |
17:42:34 | erikm | not really |
17:42:45 | akira|work | oh 7 hours idle... |
17:47:40 | prpplague | seletz: negative |
17:55:59 | akira|work | l8er |
18:38:53 | Russ | yes? |
18:39:51 | erikm | Russ: I'm thinking about rewriting the cs89x0 driver |
18:40:03 | erikm | Russ: care to help? |
18:40:12 | Russ | sure, what are the goals? |
18:40:20 | Russ | eeprom support? continued ISA support? |
18:40:20 | erikm | 1) make it not suck |
18:40:28 | erikm | 2) remove all PCisms |
18:40:38 | erikm | 3) make it work on non-ISA systems as well |
18:41:03 | Russ | what about DMA mode and memory mode? |
18:41:28 | Russ | memory mode on most non-ISA system probably gives a 1% performance gain |
18:41:39 | erikm | well, let's not start with it, but it would be nice |
18:42:01 | Russ | the main problem I see is initialization |
18:42:08 | Russ | it should really be broken out a bit |
18:42:41 | erikm | I've been talking with jgarzik about it as well, he also got some nice ideas |
18:43:41 | Russ | can you get an ISA cs89x0 card for testing? |
18:43:49 | erikm | the most important is to probe for devices at module load time, not in net_device::init() |
18:44:26 | Russ | I currently use the ether= command line to make it detect two devices with the module compiled in |
18:44:49 | erikm | oh, we could still support that |
18:45:10 | erikm | I didn't try yet to get an ISA cs89x0 |
18:45:16 | erikm | I can ask andrew morton |
18:45:44 | Russ | I thikn that an init function should either attempt to autodetect the next card, or detect a specified card |
18:46:24 | Russ | I don't know how much of that should be in the module_init() function |
18:46:31 | erikm | well, jgarzik's idea was that init() should do what it was meant to do: initialise the device |
18:46:42 | | which one is in my rootfs? |
18:46:42 | Russ | which one? |
18:47:01 | erikm | well, the device you're trying to bring up |
18:47:05 | erikm | looks for the log |
18:47:41 | erikm | <jgarzik> erikm: I have some ideas about ISA and probing in 2.4 |
18:47:51 | erikm | <jgarzik> erikm: Namely, when you build the driver into the kernel, you have a list of ports to probe. You also have a list of ports to probe when built as a module. Take this list, and call probe1() for each port. Use alloc_etherdev at the start of probe1() to get a struct net_device, -instead of- the current method of initializating the net_device::init() member, and letting that do the probe. |
18:48:00 | erikm | <erikm> jgarzik: sounds sane to me. so you actually want to probe from the init_module() function and let the net_device::init() do what it was meant for: initialising the device |
18:48:07 | erikm | <jgarzik> erikm: no need for a net_device::init() at all in the driver |
18:48:16 | erikm | <erikm> jgarzik: hmm, actually no. you can do the same from net_device::open() |
18:48:26 | erikm | <jgarzik> erikm: init_module calls probe1() once for each port. probe1() should stay pretty much unchanged, except that it now calls register_netdev at the end |
18:48:36 | erikm | <jgarzik> erikm: open()? init()? init() is not necessary. open() is :) |
18:48:54 | erikm | <erikm> jgarzik: so probe1() returns a pointer to a struct net_device |
18:48:54 | Russ | I thought you could one have the kernel come back with one device when it called the netcard init function |
18:49:11 | Russ | I currently do "ether=0,0,0,eth0 ether=0,0,0,eth1" on the command line |
18:49:12 | erikm | <jgarzik> erikm: probe1() takes an I/O port (and irq, etc.) as argument, returns zero on success, negative value on error |
18:49:47 | erikm | we can still do that |
18:49:59 | Russ | my main issues is how non clean it is |
18:50:30 | erikm | well, if we set it up properly we can do good auto-probing |
18:50:36 | Russ | mainly, the probe1(), how it behave differently base on the port number value, has all the eeprom stuff mixed in, the media detection stuff |
18:50:48 | Russ | s/behave/behaves/ |
18:50:49 | Russ | brb |
18:57:33 | Russ | back |
18:57:50 | erikm | k |
18:58:21 | Russ | ok, so you are basically saying remove the call for the net subsystem to init the cs89x0, and make it init itself |
18:58:38 | erikm | yes, that's what jgarzik proposed |
18:59:08 | Russ | so they way I'm using ether= would be obsolete |
18:59:37 | Russ | but if someone wanted to use it to specify io port/irq, it would no longer work, correct? |
18:59:44 | erikm | yes |
19:02:06 | Russ | some people would consider not being able to specify that on the command line broken (I'm not among them) |
19:03:56 | erikm | well, jgarzik has plans to change all device drivers accordingly, so I guess that he'll fix the ether= stuff as well |
19:06:08 | Russ | ah |
19:06:15 | Russ | I'll be in at work in a bit, gotta go |
19:06:59 | erikm | ok, cya |