#asahi-dev · 2021-11-22 10:00
kettenisbut I don't blame you for missing that ;)#asahi-dev · 2021-11-22 09:59
kettenismarcan: yes I tried to point out that t8103 also changed#asahi-dev · 2021-11-22 07:35
kettenismarcan: yes, I tried pointing that out to you during one of the streams ;)#asahi-dev · 2021-11-21 22:24
kettenisdoes your DT have the necessary power-domain properties on the dart nodes?#asahi-dev · 2021-11-21 17:46
kettenisotherwise it looks good#asahi-dev · 2021-11-21 17:36
kettenisnot really a problem since you're probably going to end up applying those yourself#asahi-dev · 2021-11-21 17:35
kettenismarcan: since you rebased the PMGR stuff on top of a tree that has more stuff than what has been submitted upstream so far, the bits that actually add the nodes to the device tree won't apply#asahi-dev · 2021-11-21 16:03
kettenismarcan: some of the link speed related setup is actually done by the tunables applied by m1n1#asahi-dev · 2021-11-21 15:13
ketteniss/throughly/thoroughly/#asahi-dev · 2021-11-21 15:12
kettenisthe existing dwc code has the 1 contoller 1 port assumption throughly built in#asahi-dev · 2021-11-20 12:20
kettenispushing things until they break is always fun!#asahi-dev · 2021-11-20 12:09
kettenisdoubt that's a good idea#asahi-dev · 2021-11-20 12:04
kettenis(trying to convert u-boot to the new mailbox binding)#asahi-dev · 2021-11-20 12:03
kettenisand SMC doesn't need it#asahi-dev · 2021-11-20 12:02
kettenisso that's why you added that 2nd range to the nvme node#asahi-dev · 2021-11-20 11:59
kettenisah, ok#asahi-dev · 2021-11-20 11:57
kettenishow is that supposed to work?#asahi-dev · 2021-11-20 11:55
kettenisthe register to start the coprocessor isn't exposed in your binding#asahi-dev · 2021-11-20 11:55
ketteniseh, sven, regarding the mailbox#asahi-dev · 2021-11-20 11:35
ketteniswe probably want the NVMMU anyway#asahi-dev · 2021-11-20 11:34
kettenisfair enough#asahi-dev · 2021-11-20 11:32
ketteniswell, this might change their minds#asahi-dev · 2021-11-20 11:29
kettenisbecause macos doesn't use it#asahi-dev · 2021-11-20 11:29
kettenisthe risk of using this other mode is that Apple introduces bugs into it or removes it altogether in the future#asahi-dev · 2021-11-20 11:26
kettenisbut even that might not be the case#asahi-dev · 2021-11-20 11:26
kettenisjust different firmware#asahi-dev · 2021-11-20 11:26
kettenismy assumption has always been that the underlying hardware was very similar#asahi-dev · 2021-11-20 11:16
kettenisI bet we'd need the same quirks for the M1 in this mode#asahi-dev · 2021-11-20 11:15
kettenisbut on those NVMe was provided by Apples own implementation in the T2 chip#asahi-dev · 2021-11-20 11:14
kettenismakes some sense since Intel Macs uses "normal" NVMe#asahi-dev · 2021-11-20 11:13
kettenisinteresting#asahi-dev · 2021-11-15 13:53
kettenisgreat; that'll help OpenBSD when I get around to supporting the new SoCs#asahi-dev · 2021-11-15 13:35
kettenisI may be misreading the code though#asahi-dev · 2021-11-15 13:35
kettenisand as far as I can follow the code path Linux doesn't check on arm64#asahi-dev · 2021-11-15 13:35
kettenisso I remember removing the check of the "status" property on OpenBSD for that#asahi-dev · 2021-11-15 13:34
kettenisyeah, there are (broken) DTs out there that have all but the primary CPU node disabled#asahi-dev · 2021-11-15 13:33
kettenismarcan: I believe you need to remove the cpu nodes from the device tree in m1n1 instead of disabling them#asahi-dev · 2021-11-15 11:55
kettenisthese are 1:1 mappings, but they can be relocated by calling SetVirtualAddressMap#asahi-dev · 2021-11-15 11:55
kettenismarcan: the UEFI memory map basically indicates memory ranges that need to be preserved for runtime services#asahi-dev · 2021-11-14 12:02
kettenisok, my u-boot watchdog barks while sven's linux one bites ;)#asahi-dev · 2021-11-13 19:58
kettenisthere are some benefits in specifying the smallest possible size and in making it a power of 2#asahi-dev · 2021-11-13 13:35
kettenisso maybre the magic nvram writes are necessary after all#asahi-dev · 2021-11-13 13:33
kettenisas in, it doesn't actually power the machine off?#asahi-dev · 2021-11-13 12:43
kettenisbut it may be needed to prevent the machine booting into recovery mode or something when powered up again#asahi-dev · 2021-11-13 12:43
kettenisit would surprise me that writing to nvram is really necessary#asahi-dev · 2021-11-13 12:42
kettenisthe corellium code does indeed touch nvram#asahi-dev · 2021-11-13 12:24
kettenismaz's commit messages are always a joy to read ;)#asahi-dev · 2021-11-12 23:07
kettenismeanwhile I can confirm that the touchpad does indeed produce the same kind of packets as those handled by the bcm5974.c linux driver when in raw mode#asahi-dev · 2021-11-12 23:00
kettenisoh wait...#asahi-dev · 2021-11-12 23:00
kettenismaybe you should implement a battery status reader to see how fast you can drain your battery