#asahi-dev · 2025-04-25 16:47
nickchanchaos_princess: very wip rtc driver if you want to take a look on how pmic rtc worked on a7-a11#asahi-dev · 2025-04-24 12:27
nickchanfl0_id: there is also one person that worked on postmarketOS that is intersted#asahi-dev · 2025-04-24 12:05
nickchanfl0_id: there are three github issues where people actually tried to compile things#asahi-dev · 2025-04-24 12:02
nickchanugh the details is probably not very suitable in this channel I will talk in #asahi-re#asahi-dev · 2025-04-24 12:02
nickchancurrently the hacked up tracing bootchain looks like this: iboot loads xnu -> stage does seprom firmware load race and apply xnu security patches, passes xnu entry point to m1n1 -> m1n1 performed tracing patches, restored chcicken bits and branch to xnu as next stage#asahi-dev · 2025-04-24 11:58
nickchancurrently relying on previous stages to load xnu though #asahi-dev · 2025-04-24 11:56
nickchanBut it's all for less binary reverse enginnering later on#asahi-dev · 2025-04-24 11:55
nickchanneed to say hacking xnu like that obviously involves binary reverse enginnering and i will not pretend otherwise#asahi-dev · 2025-04-24 11:55
nickchanfor pagetables the idea current is to make some TLB maintance op UNDEFINED#asahi-dev · 2025-04-24 11:54
nickchanand then the sysreg instructions are hooked by making them UNDEFINED#asahi-dev · 2025-04-24 11:53
nickchanhe is currenting working on hacking xnu for tracing and it's... kind of working? vbar and rvbar still points to m1n1 when xnu fully booted but the elephant in the room called pagetables are not solved yet#asahi-dev · 2025-04-24 11:52
nickchanso for the people that actually cared performance counting on those devices it would be me and the other person that's also doing some dev work (not in the way of kernel commits though)#asahi-dev · 2025-04-24 11:50
nickchanand then the rest a7-a9 is just convenience since it is minimal effort to support those too#asahi-dev · 2025-04-24 11:50
nickchanalso for spoofing ecore frequencies on a10/a10x/t2 appropriately#asahi-dev · 2025-04-24 11:49
nickchanlike i make it work so i have a better idea of whether my capacity-dmips-mhz on a11 is apprioriate but really coremark itself is already enough#asahi-dev · 2025-04-24 11:47
nickchanfl0_id: there are definitely some people writing github issue but i don't think they cared about performance counting#asahi-dev · 2025-04-24 11:34
nickchana case of no one caring#asahi-dev · 2025-04-24 11:32
nickchanproabbly would resend the a7-a11 perf support after -rc4#asahi-dev · 2025-04-16 13:26
nickchanchaos_princess: there are always on support in upstream pmgr, it has to be there otherwise the kernel will explode hard on things like the ps_dcs1 (lpddr interfaces)#asahi-dev · 2025-04-15 17:52
nickchan#KEY reads: 603 so this is every single key#asahi-dev · 2025-04-15 17:46
nickchanchaos_princess: https://gist.github.com/asdfugil/450c8da822aafcba98c67415483a5e50#asahi-dev · 2025-04-15 17:40
nickchanand i have a mass key dump#asahi-dev · 2025-04-15 17:40
nickchanchaos_princess: Ok i can say with confidence that CLKM does not exist#asahi-dev · 2025-04-15 17:03
nickchanchaos_princess: smc is in suspend when iboot boots kernel also the actual interface is also different#asahi-dev · 2025-04-15 17:02
nickchanthe iop interface is not compatible#asahi-dev · 2025-04-15 17:00
nickchanso m1 doesn't have year 2038#asahi-dev · 2025-04-15 17:00
nickchanbytes i mean#asahi-dev · 2025-04-15 17:00
nickchanokay rtc_offset is the same thing as a11 except there are two extra bits#asahi-dev · 2025-04-15 16:41
nickchanbut it looked really similar i need to check#asahi-dev · 2025-04-15 16:40
nickchanthe clock on a11 is still susceptible to year 2038 at least#asahi-dev · 2025-04-15 16:38
nickchanfor t2 and a11#asahi-dev · 2025-04-15 16:38
nickchanall things considered it probably does exist there#asahi-dev · 2025-04-15 16:37
nickchanmore specifically, overwrite the rest of the time stored in the pmic nvmem#asahi-dev · 2025-04-15 16:36
nickchanit's doing that because the 194 days counter cannot be reset and time becoming wrong just because iOS is not running when that counter overflow is not great#asahi-dev · 2025-04-15 16:35
nickchancan even reproduce by just setting the time in an iOS ramdisk#asahi-dev · 2025-04-15 16:35
nickchanand the reason i have not implemented that rtc just yet that iOS *will* overwrite the time every boot#asahi-dev · 2025-04-15 16:34
nickchanyou should keep it#asahi-dev · 2025-04-15 16:32
nickchan194 days of time#asahi-dev · 2025-04-15 16:32
nickchana rtc that can count 194 of time and then a 32-bit value in the pmic nvmem to store the rest of the time#asahi-dev · 2025-04-15 16:30
nickchanchaos_princess: I have verified the existence of an rtc in a11's spmi dialog da2422 pmic and is able to read time from it#asahi-dev · 2025-04-15 16:27
nickchanchaos_princess: I could reuse it for other purposes I suppose#asahi-dev · 2025-04-15 16:05
nickchansven: what weird bug?#asahi-dev · 2025-04-15 15:46
nickchannevermind#asahi-dev · 2025-04-15 09:15
nickchanm1n1 has a 4k paging code path too though it's not used on m4#asahi-dev · 2025-04-15 09:15
nickchansome people say it may be broken#asahi-dev · 2025-04-15 09:15
nickchanhave anyone tried 4k paging in el2 on m4?#asahi-dev · 2025-04-13 11:03
nickchancan probably try mapping physical memory to el0 from bridgeOS on t2 too to find out but i think it's better to hack on xnu vbar first#asahi-dev · 2025-04-13 11:02
nickchant2 is unknown because accessing both the i2c and related pmgr pwrstate regs in whatever state iboot puts the kernel in results in serror#asahi-dev · 2025-04-13 11:00
nickchani2c hwrev (0x28) is 6 since a12 and 3 on a11... not like the a12 part has any practical sigificance anyways#asahi-dev · 2025-04-13 10:46
nickchanthere's also https://lore.kernel.org/asahi/20250407-apple-cpmu-v6-0-ae8c2f225c1f@gmail.com/T which comes with some dts patches but nothing from the arm64 side so it continues to sit there