02:18 airlied[d]: all fun and games until you can't idle your channels
06:24 airlied[d]: 595 is proving a bit stubborn, booter mbox is 0, but gsp never goes
07:17 airlied[d]: notthatclippy[d]: https://people.freedesktop.org/~airlied/scratch/loginit.gz any chance you could decode that with 595.58.03 not that I'm hopeful, I'm not even getting to msg queue init on the gsp side
07:21 notthatclippy[d]: It's 99+% nulls, I don't think there's anything there
07:24 airlied[d]: ouch, guess I just have to keep hitting it with a stick
07:25 notthatclippy[d]: Did you do the new "sequencer" changes?
07:25 notthatclippy[d]: Though I guess that might come after the msgq init, so you might not be getting there at all
07:26 notthatclippy[d]: John has Nova working on 595 (and later), so maybe that gives you a smaller haystack to look through
07:27 notthatclippy[d]: notthatclippy[d]: yeah, I'm not getting anything out of this.
07:31 notthatclippy[d]: notthatclippy[d]: can't find this branch on github; maybe not pushed yet. You can just ask him directly for it
07:40 airlied[d]: yup have to get to the new msgs for sequencer, but not even seeing a msgq yet
07:44 airlied[d]: I'm very happy to see the sequencer gone 🙂
07:49 notthatclippy[d]: Heh, we _almost_ leaned into it all the way instead. Like.. parse gsp.bin, extract `.sequencer`, interpret it, and there's your GPU booted. Sounds nice, just don't think about it too much :)
07:53 airlied[d]: always like a bytecode to boot 😛
07:53 airlied[d]: okay looks like some libos changes might also have happened for 595, will see if it's those
07:53 notthatclippy[d]: What's the last version that you have booting successfully? Or at least getting to the msgq init?
07:54 airlied[d]: 580.142
07:55 notthatclippy[d]: ACK. And are you testing this on the spark, or are you first trying to get a dGPU up?
07:57 notthatclippy[d]: Sorry for suggesting the obvious, but, did you update the msgq init args, from `gsp_init_args.h`?
07:57 airlied[d]: trying on ada first 🙂
07:57 airlied[d]: yup found the new entries for GSP_ARGUMENTS_CACHED
08:00 airlied[d]: bleh the libos change to just use one addr instead of a pte array, is compatible, so not that
08:07 airlied[d]: I feel the gsp is waiting for me to initialise something before it continues to boot, but I've no idea
08:19 airlied[d]: ah well enough for today, found a 4096 that hsould have been a 12 but didn't help
15:13 mangodev[d]: notthatclippy[d]: didn't even realize nova was functional, nonetheless could run a whole driver
15:13 mangodev[d]: i guess there has been enough done on the rust drivers for linux part?
15:14 notthatclippy[d]: mangodev[d]: Working in this case means "can boot gsp and gsp tells you the gpu name and that's it"
15:15 mangodev[d]: fair point
15:15 mangodev[d]: was wondering if i missed something big about nova
15:15 mangodev[d]: i mean
15:15 mangodev[d]: it can boot, which is already an achievement
22:13 AlannaBaby[m]: Get in touch with this platform for greatness you’ll definitely thank me later
ℹ️❤️
👇🏻👇🏻👇🏻
22:13 AlannaBaby[m]: https://t.me/+NT0-GuEzsd9lNzVk
22:13 AlannaBaby[m]: Get in touch with this platform for greatness you’ll definitely thank me later
ℹ️❤️
👇🏻👇🏻👇🏻
22:13 AlannaBaby[m]: https://t.me/+NT0-GuEzsd9lNzVk
22:15 AlannaBaby[m]: Get in touch with this platform for greatness you’ll definitely thank me later
ℹ️❤️
👇🏻👇🏻👇🏻
22:15 AlannaBaby[m]: https://t.me/+NT0-GuEzsd9lNzVk
22:23 soreau:thanks airlied now, instead
23:59 mangodev[d]: in IRC? 💀