| |
| |
| |
|
Page: 1 2 3 4 5
Comments:
<0> redwood city
<1> ah
<0> bay area
<1> i worked there before
<1> at a company called Determina
<1> working on some really cool security software
<1> just down the street from Oracle
<1> well, i worked here in idaho. but i'd fly there lots to meet up with the other engineers
<0> niiice
<0> how is the security software industry?
<0> I have a friend that's working at zone alarm
<0> *zone labs
<0> whatever the hell they call it.
<1> well, if you like it, its fun
<1> haha
<1> Determina is doing really well
<1> but i hate doing remote work
<1> so i quit and did my own thing for a while
<1> and then 2 weeks ago McAfee came to me with a job offer
<1> so i accepted and i'm just waiting for them to finish the inner company hiring process
<0> awesome!
<0> I'm hoping I'll get to work at apple next year
<0> or maybe go back to lawrence livermore
<0> one of the people I worked for said he'd hook me up
<1> thats cool
<1> i have a friend over at apple
<2> hey
<3> http://www.rollmop.org/ferrets/images/easter_ferret.jpg
<4> !
<5>
<6> hey babies
<6> and dudes
<7> oh lord. it still works
<6> what? the brain? the floppy? the mojo?
<6> Evil-Bill, done any RTL8139D coding?
<8> howdy
<6> hey, dude
<6> eieio, i'm scratching my head of over the rtl... though I'd spent an hour or so today as well
<9> gr00ber which rtl?
<6> all I wanna achieve now is receiving a packet (from a broadcast)
<8> ah, what's up wit it?
<6> RTL8139D
<6> Dr_Evil, you wrote a driver for that one?
<9> oh that one, i once wrote a driver for rtl8169, the gigabit chipset, for BeOS
<6> now, what I've done so far is set up the irq mechanism and pci stuff (I get media change events when I take out the cable :)
<9> it deosnt share much with the 39, sorry
<6> other wise, i've written 0xff into the multicast registers and set the receive config register to allow all packets
<6> I also unmasked all irq's
<6> The spec, alas, doesn't spell out the steps to init the card
<9> i remember one strange "feature", you had to enable receiver and transmitter *before* configuring them
<6> oh, I do it afterwards
<6> lemme fix that, thx dude
<9> but that was the 69 chipset, perhaps it doesnt apply
<6> The spec doesn't mention this feature, I think ;:|
<6> i'll try
<6> I also wonder about Bus Master Enable - it wasn't set in the PCI cmd reg
<6> and the spec says it should be... so I'm setting it manually
<6> gnorf
<6> Then there's the 32-bit "Rx Buffer Start Address"
<6> this is any old physically contiguous memory address, no?
<6> HOLY ****
<9> you need to manualy set pci command register
<6> Dr_Evil, it was indeed the rx/rc enable-before-config - I get packages now :)
<6> what should I do without Dr Evil ?
<8> nice
<6> maybe ask Mini Me
<6> Dr_Evil, where did you get that info? Or the hard way?
<6> Thing is, the Realtek spec is REALLY ****ty
<9> the hard way
<6> I thought it was good yesterday, but now as I gr0k more, I find it... inadequat
<6> Dr_Evil, i c
<9> its not ducumented
<6> hm
<8> oh look, the initialization sequense is in to programming guide
<6> what programming guide?
<6> i have a guide for the 8029, but not this one
<8> umm, i don't know where it came from so ill just post it for you
<6> The only thing i've found is the "RTL8139D Data Sheet"
<6> eieio, excellent
<6> ping -b 10.0.0.255 is my friend
<6> I wonder why I don't have permissions to run arping
<9> eieio, gr00ber can I get those documents, too?
<6> the data sheet is at realtek.com.tw
<9> programming guide, too?
<8> let me know when you've got it
<8> i'm going to take it down after you two get it
<9> thanks, need to leave
<8> did you get it gr00ber?
<6> eieio, sorry, was away
<6> downloading now
<8> k, np
<6> hm, that's for A/B
<6> I have a D card
<6> But I'm sure a lot of it will be useful, so thanks
<8> yeah, well isn't most of the D stuff related to the more modern DMA descriptor style?
<6> I have no idea yet
<2> Modern? :-/
<6> about the differences
<2> brrr...
<6> i'll figure it out
<8> "more modern"
<6> marco_g, don't tell about modern. I have ne2000 clone to be hacked soon enuff :)
<2> gr00ber: ISA, I hope? ;-)
<6> gee, I have both PCI and ISA NE2000 clones
<6> the 8029 version i have is pci
<6> but 8139D first : )
<8> so now i need to implement a buffer cache
<6> ya
<6> hm, seems it starts up in half-***ed half-duplex mode.
<6> anyway, time to parse those pesky ethernet packets
<8> i defined an interface a wile ago, but i haven't needed to implement it since Mac OS was doing p*** through IO and has its own buffer cache
<8> no point in double caching the pages
<8> but now i'm using native FS support for more things and it's slooooow
<6> ok, so when pinging the broadcast address 10.0.0.255 I get packet received irq's. I guess the buffer now contains a [ethernet [ip [icmp] ] structure?
<6> or is there even more lower-level headers to be stripped of that the nic doesn't do?
<6> like preambles and ****
<6> boy, am I a nic noob
<8> umm, i don't remember if the nic leaves the eth header on or not
<2> AFAIK it leaves the ethernet header.
<2> You need it to see who sent the frame.
<8> i suspect it does, though i know you can change that in more modern chipsets
<2> :-)
<6> hm
<6> ok, so after a while with pinging here I get the (not-unexpected) Rx Buffer overflow
<6> since I've exhausted the buffer
<6> does it start on 0 again automagically after ack'ing this?
<6> or am I supposed to copy each packet to a separate buffer ASAP and then reset the rx ptr (if there is such a thing)?
<6> man, I need a NIC 101
<8> i think you can either copy it, or ***ign a new rx buffer
<8> copying is probably the better choice
<6> especially on concurrent requests... :)
<6> ok, so the spec has a section "Receive Status Register in Rx Packet Header"
<6> there is no such register in the register index
<6> is this a field in the received header?
<6>
<6> must be something like that since it is the only register mentioned without a offset
<8> yawn
<8> i've been gettin a lot done at work today. feels good
<10> http://xkcd.com/c149.html
<10> i just woke up!
<10> but I was up till 3 or 4am so I dont feel so bad
<11> hmm, what's time over there?
<10> noon
<11> mkey, good time to get up IMO :)
<8> heh
<8> also a good one: http://xkcd.com/c10.html
<6> Evil-Bill, how 'bout that Rx Packet Header status register?
Return to
#osdev
or
Go to some related
logs:
undelete files knoppix ext3
15-av_scanners ubuntu
#python
postfix Servname not supported for ai_socktype
#linux
strife.swf
emerge kde-header
#php
openwrt sonicwall
#centos
|
|
