Search logs:

channel logs for 2004 - 2010 are archived at ·· can't be searched

#osdev2 = #osdev @ Libera from 23may2021 to present

#osdev @ OPN/FreeNode from 3apr2001 to 23may2021

all other channels are on OPN/FreeNode from 2004 to present

Tuesday, 11 January 2022

04:55:00 <eryjus> pc boards how to
04:56:00 <eryjus> ha!
04:57:00 <kazinsal> "wait a second, this isn't my web browser"
04:58:00 <eryjus> obviously i look at my hands while I type
04:58:00 <klange> Sir, this is a Wendy's.
05:02:00 <kazinsal> can I get aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa uhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
05:02:00 <kazinsal> baconator
05:14:00 <brynet> and then?
05:36:00 <kazinsal> I'll have two number nines, a number nine large, a number six with extra dip, a number seven, two number 45s, one with cheese, and a large soda
05:42:00 <Mutabah> Would you like fries with that?
05:50:00 <geist> PCI question: i know that you're supposed to try to set all 1s in a BAR to see what the size is
05:50:00 <geist> but... question is it always safe to do that with a BAR that has been already configured, provided you know nothing is accessing it
05:50:00 <geist> ie, saving old value, writing 0xffff..., seeing the size, then putting the old value back?
05:51:00 <geist> given there's no other option as far as i can tell, i guess the answer is yes
05:52:00 <geist> my main concern would be if it somehow confused the device itself, but probably wont, as long as there are no outstanding mmio/io transactions against any of the BARs
05:52:00 <geist> since presumably the hardware being the bar config has to do with device's address decoding logic
05:53:00 <geist> so if it's temporarily set to 0xfffff.... then that's no harm if no transactions go against it
05:53:00 <geist> s/being/behind
07:28:00 <Mutabah> Pretty sure that's the intended usage, so should be safe
07:28:00 <Mutabah> as long as nothing else is accessing the device
07:37:00 <geist> yah i guess if you're going to get that information its probably safe to do it during initial probing since you know no drivers are accessing it at that point
07:37:00 <geist> vs on demand with some sort of 'get_bar_info()' api
07:39:00 <sham1> The BAR information should just be cached
07:39:00 <sham1> To the best of one's abilities
07:56:00 <geist> yah
08:17:00 <Belxjander> geist: on the OS I am using... everything is probed at the system startup and things of that nature are then cached as part of the hardware description which is then directly referenced on-demand whenever applications actually need that information
08:18:00 <geist> yah
08:18:00 <geist> i hadn't been doing that re: the bars but i probably should
09:01:00 <geist> hmm, wonder if it's possible to get qemu to put a 64bit pci device > 4GB
09:01:00 <geist> basically to test my 64bit BAR reading logic
09:07:00 <sham1> You might be able to hack it. If nothing else than by changing the source
11:41:00 <geist> looks like folks have asked before and one of the problems is the seabios doesn't understand it
11:43:00 <sham1> F
11:45:00 <zid> 64bit bars, or querying the bar size?
11:45:00 <zid> presumably the former
11:46:00 <zid> could you not reprogram the bars at boot, then run your 'normal' detection code?
12:16:00 <geist> Well device would need to have support
12:17:00 <geist> And afflict none do
12:17:00 <geist> Afaict
12:18:00 <geist> Since it chews up two bars some devices have to rearrange tHeir bar layout
12:19:00 <geist> And some never will, like say ahci which uses bar5 for the main mmio aperture
12:19:00 <geist> And is thus perpetually 32bit only
18:21:00 <zid> intel wanted me to make an account to grab a pdf.. mouser just had it via google, whatever
18:23:00 <gog> which pdf?
18:23:00 <zid> 82579
18:23:00 <gog> what
18:23:00 <gog> lmao
18:25:00 <zid> the software side pdf doesn't seem to exist though
18:25:00 <zid> just the fun product specs one
18:25:00 <gog> :/
18:26:00 <gog> society has moved beyond the need for intel
18:26:00 <ThinkT510> yeah, guessing is much better
18:27:00 <zid> ah I think because it's actually part of the PCH
18:27:00 <zid> because it's the management engine port
18:28:00 <zid> It's *probably* just an e1000 and you can just treat it like one, but who knows
18:30:00 <zid> e1000e*
18:33:00 <zid> hah I found it
18:33:00 <zid> hosted on intel's.. sourceforge page
18:33:00 <zid> saying the repo is out of date and to 'use'
18:37:00 <zid> if anyone has an opensdm_8257x-18.pdf newer than 18 lmk though
18:37:00 <zid> maybe geist bothered to sign up to intel openpartner or whatever it is at some point
18:38:00 <geist> probably
18:38:00 <gog> oh i have 325642
18:38:00 <gog> sorry
22:30:00 <Clockface> damn unreliable power
22:33:00 <Clockface> i really should make a script to set things up again automatically for when my computer gets forced off
22:42:00 <blockhead> Clockface: I got a laptop more for that reason than for being able to move it around easily