00:04 Lyude: karolherbst: does this mean anything to you:
00:04 Lyude: KBL079:
00:04 Lyude: A PCIe link should retrain when Retrain Link (bit 5) in the Link Control register (Bus
00:04 Lyude: 0; Device 1; Functions 0,1,2; Offset 0xB0) is set. Due to this erratum, if the link is in
00:04 Lyude: the L1 state, it may ignore the retrain request
00:04 karolherbst: uffff
00:04 Lyude: ...sorry, my phone made me think that would come out to only two lines of paste
00:04 karolherbst: the fuck
00:04 karolherbst: where did you get that from?
00:05 karolherbst: but I am sure if we force L0 on the GPU, it still fails
00:05 Lyude: karolherbst: literally a friend I just offhandedly mentioned this to, it's from "specification updates"
00:07 karolherbst: ohhhhhh
00:07 karolherbst: the bus controller refueses to retrain?
00:07 karolherbst: uhmmm
00:07 karolherbst: Lyude: I think I will check on that
00:08 karolherbst: that might actually sound like the issue we hit
00:08 karolherbst: nice find
00:13 Lyude: karolherbst: thank my friend! they found it, depressingly quickly :(
00:13 Lyude: granted, if even Bjorn didn't notice that one I don't feel too bad
00:15 karolherbst: :D
00:15 karolherbst: true
00:18 Lyude: https://cdrdv2.intel.com/v1/dl/getcontent/334663
00:18 Lyude: karolherbst: the link they just gave me
00:42 karolherbst: uff.. my system just peaked out at 27GB RAM usage :O
00:47 HdkR: Nice
00:50 karolherbst:running without swap
00:52 karolherbst: project ideas for people you want to "keep occupied": fix compiler warnings in chromium
00:52 Helios747: how many chrome tabs is that
00:52 karolherbst: that's not running chromium, it's compiling chromium :p
00:53 karolherbst: if it makes it better, the build dir is already at 41GB
00:58 imirkin: Lyude: ok. my point stands about it needing to happen on start.
00:58 imirkin: so you don't end up with stale state
14:22 uis: Hi
14:23 uis: Have somebody info about NVE4 init?
15:13 RSpliet: uis: there's lots of info to be found in envytools and the drivers. Do you have specific questions or problems?
19:47 uis: RSpliet: replace VBIOS
19:52 uis: And add init code to libgfxinit
19:52 uis: GPU init in coreboot
21:42 imirkin_: uis: have a read through this post for why i believe such an effort would be impractical. it's not 100% related to your question, but i do cover vbios things there. https://www.phoronix.com/forums/forum/linux-graphics-x-org-drivers/open-source-nvidia-linux-nouveau/998310-nouveau-persevered-in-2017-for-open-source-nvidia-but-2018-could-be-much-better?p=998427#post998427
21:44 imirkin_: uis: if otoh you're looking to use the vbios, just not its actual x86 code, then that should be just fine. but then you're stuck implementing a basic driver that gets you a fb displayed, i'm guessing.
23:16 imirkin: karolherbst: are you aware of any xf86-video-nouveau bugs?
23:16 imirkin: i guess RH wouldn't get too many, since they basically disabled it. debian too =/
23:17 karolherbst: not really... I don't use any ddx with my nvidia gpu
23:17 karolherbst: laptop.. and all ports wired to intel
23:17 karolherbst: so...
23:17 imirkin: right
23:17 imirkin: i just mean like ... have i forgotten about stuff?
23:17 karolherbst: no idea.. I could check on my desktop machine in the office if you have something I should test
23:17 imirkin: want to do a release with my latest batch of changes at some point
23:17 imirkin: nah
23:44 imirkin: Lyude: btw, it's limit = ctx->mem.addr + len - 1 (you're missing the -1)