11:37karolherbst: imirkin: mind taking a look at https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/3724 ? I'd like to merge the modifier stuff soonish
11:37emersion: \o/
15:00karolherbst: nice... started the android emulator and the kernel got stuck in an rcu handler...
15:00karolherbst: fun fact: wasn't using nouveau :/
15:02imirkin: you mean there are kernel bugs that aren't in nouveau?
15:04karolherbst: exactly
15:25ccr: karolherbst, Intel?
15:27ccr: if yes, could be this https://gitlab.freedesktop.org/drm/intel/-/issues/2905 (fix is coming, patch also mentioned in comments)
15:28karolherbst: ccr: could be, yes
15:28karolherbst: but only happened now with the android emulator
15:29ccr: it's pretty rare for me
15:29ccr: [14:13] <ickle> you need something that is generating new framebuffers virtually every frame, so a DRI3 client or compositor
15:29ccr: [14:13] <ickle> and then bad luck to race against a RCU worker
15:30ccr: so depends on what one is running and position of the moon
16:28karolherbst: wayland compositor?
16:28karolherbst: and I have tons of addfb2 ioctl calls :p
16:33ccr: it could be a winning combination
16:36karolherbst: yeah, I guess
16:38isapgswell: hi everyone
16:38isapgswell: what is nouveau firmware?
16:38isapgswell: nvidia firmware replacement?
16:40karolherbst: isapgswell: depends on the firmware
16:41isapgswell: karolherbst initial support
16:41isapgswell: karolherbst my gpu is new. gtx 1060
16:41karolherbst: then it's original firmware from nvidia
16:42isapgswell: karolherbst why not nvidia free their firmware?
16:42karolherbst: ask nvidia
16:42karolherbst: but I guess a big reason is drm
16:43isapgswell: karolherst
16:43karolherbst: and probably some other PR reasons like "security"
16:43isapgswell: karolberst yes
16:44isapgswell: karolherst security is controversy, we say spying us
16:45isapgswell: karolherbst maybe
16:46isapgswell: karolherbst is there a way to me to flash nouveau firmware or put /lib/firmware/nouveau*
16:46karolherbst: if you have the signing key then yes
16:47isapgswell: karolherbst or if i have a eeprom programmer
16:47karolherbst: nope
16:47karolherbst: it has to be signed
16:47karolherbst: and they are not on the GPU in the first place
16:47isapgswell: karolherbst oh my god
16:47karolherbst: firmware is not the same as the vbios ;)
16:48isapgswell: karolherst intel is worst than nvidia
16:48karolherbst: isapgswell: well, if this surprises you, maybe you want to throw away your own computer :p
16:48karolherbst: because nearly every piece of hw has this these days
16:48isapgswell: karolherbst vbios is like option rom?
16:49karolherbst: more or less
17:02juri_: karolherbst: generic positive words of encouragement and thanks here.
17:12RSpliet: isapgswell: the fact that NVIDIA required signed firmware doesn't change the fact that you can still disassemble them and see what they do. Feel free to run them through the envytools and take a peek if you're concerned.
17:13RSpliet: I however, am too busy to assist you with that
17:15RSpliet: Last time I looked at their firmware blobs, 5-or-so-years-ago, there was no signs of "spying"
17:15RSpliet: Intel ME has given firmwares a bad name
17:19ccr: mismanagement engine
17:19RSpliet: I wonder if that's been the driving force behind Intel's management in the past few years :-P
17:40isapgswell: juri_ do you support the libreboot?
17:40isapgswell: RSpliet yes intel i dont put my hand into the fire
19:41juri_: isapgswell: I try. :)
20:33karolherbst: RSpliet: well, ime has some fancy features
20:33karolherbst: it's just full of holes
20:34RSpliet: IME being the Intel (Mis)Management Engine?
20:35RSpliet: if so, that concurs with what I said
20:35RSpliet: (and thank god it is, we wouldn't be so nearly as good at breaking it to the point of being effectively disabled otherwise :-D)
23:18karolherbst: RSpliet: well, for remote power control it is quite useful :p