#asahi · 2022-01-10 11:54
svenknowing apple it may also be called both :D#asahi · 2022-01-10 11:53
svenrtkit it is then :D#asahi · 2022-01-10 11:52
svenfine with me :D#asahi · 2022-01-10 11:52
svenah#asahi · 2022-01-10 11:52
svenah, heh#asahi · 2022-01-10 11:52
svenhuh, where did that come from?#asahi · 2022-01-10 11:51
svenakf you mean?#asahi · 2022-01-10 11:51
sven_jannau_: it unfortunately is. I can write some more details later#asahi · 2022-01-10 11:39
svensince you’ve already spent some time with DCP.. wanna give proper locked dart support a shot? :>#asahi · 2022-01-10 11:38
svenyeah, I think we can just always send it#asahi · 2022-01-10 11:36
svenit is required for DCP though since iboot puts it into hibernation instead of deep sleep#asahi · 2022-01-10 11:36
svenwake isn’t required for all coprocs but I also think it doesn’t hurt #asahi · 2022-01-10 11:35
svenoslog or something like that#asahi · 2022-01-10 11:35
sveniirc another syslog #asahi · 2022-01-10 11:21
svenjannau: nice!#asahi · 2022-01-08 14:52
sven:)#asahi · 2022-01-08 13:31
svennamespace#asahi · 2022-01-06 15:30
svenor do that cute hack where we stuff a new version into this auxKC thing and let iBoot load it for us but i believe marcan said that doesn't work if you have a custom boot object anyway#asahi · 2022-01-06 15:29
svenj`ey: yeah, that would be possible#asahi · 2022-01-06 15:29
svenb) is more tricky because it's not clear if it's possible to update the firmware *while keeping the custom boot object*#asahi · 2022-01-06 15:28
svena) would also also possible without that roundtrip if we added chainloading support to m1n1 itself (or if this auxKC can be used together with fuos)#asahi · 2022-01-06 15:28
svenboth are definitely possible with a round-trip through 1TR#asahi · 2022-01-06 15:28
sventhere are two things: a) updating m1n1 (the custom boot object pr fuos or however it's called) b) updating the firmware on the stub partition#asahi · 2022-01-06 15:27
svenno, it's not.#asahi · 2022-01-06 15:26
sventhat's separate from the question you just asked#asahi · 2022-01-06 15:22
svenmight not even be possible if the normal update procedure also updates the "custom boot object". if that's the case we'll alway have to go through 1TR for that#asahi · 2022-01-06 15:21
sventhat's gonna be a bit more scary. it will require APFS write support and a pretty well understood SEP API#asahi · 2022-01-06 14:42
sven🤷🏻♂️#asahi · 2022-01-06 14:41
svenI’ll need to write a dumb nvme driver for m1n1 to run some experiments #asahi · 2022-01-06 14:40
svenwe’ll figure something out#asahi · 2022-01-06 14:39
svenalso happens in the T2 Macs already #asahi · 2022-01-06 14:39
svenrandom writes plus flush brings the performance down to ~100 kb/s or so#asahi · 2022-01-06 14:38
svenyeah, just don’t issue flush commands :p#asahi · 2022-01-06 14:34
sven64 at most, but those slots are shared between the admin and io queue#asahi · 2022-01-06 14:31
sventhe completion queue still exists and works as usual though#asahi · 2022-01-06 14:31
svenand linear means there’s no head/tail pointer but just an index into the array. It’s technically no longer a queue#asahi · 2022-01-06 14:29
svenit also only has a single shared interrupt so even if you could have more queues they’d only make sense as poll queues#asahi · 2022-01-06 14:29
svenon the apple chip there’s always just one IO queue even if you don’t enable linear sq#asahi · 2022-01-06 14:29
svenmore or less#asahi · 2022-01-06 14:26
svenlinear submission “queue” instead of the normal nvme submission queue#asahi · 2022-01-06 10:36
svenAlso makes us stop wasting time on bugs that are really just wrong config options #asahi · 2022-01-06 08:37
svenlooks useful, thanks!#asahi · 2022-01-05 14:10
svenjannau: yeah.. i'll also have to get back to that series that will allow PCIe with 4k pages#asahi · 2022-01-05 13:57
svenso i'm really not sure how that happened#asahi · 2022-01-05 13:57
svenCONFIG_USB_GADGET is in arm64 defconfig which should enable CONFIG_USB_DWC3_DUAL_ROLE#asahi · 2022-01-05 13:54
sveni'm kinda surprised that isn't enabled fwiw#asahi · 2022-01-05 13:49
svencould be#asahi · 2022-01-05 13:48
sventhen i'm pretty sure you're doing something wrong#asahi · 2022-01-05 13:47
svenas long as the i2c nodes, the usb-pd and the dwc3 nodes in the device tree are present and you use that commit usb hotplug should work#asahi · 2022-01-05 13:47
svenyeah