#
#asahi-re · 2021-01-07 15:37
Fanfweit's not handled by the T2, it's the USB mux controller (CD3217 from Texas Instruments) also known as ACE in Apple's own schematics
#asahi-re · 2021-01-07 15:35
davidrysk[m]it's not handled by a T2, but it's basically done the same way since the M1 is more or less a much expanded T2
#asahi-re · 2021-01-07 15:21
puhitakunever knew it's handled by T2! thanks a lot for the link
#asahi-re · 2021-01-07 15:20
davidrysk[m]there was also a tweet by marcan but I forget where it was
#asahi-re · 2021-01-07 15:19
davidrysk[m]puhitaku: they use vendor USB-PD commands to expose UART, etc over USB. https://t8012.dev/ is a good place to start
#asahi-re · 2021-01-07 15:07
puhitakumy concern is that special USB communication should be achieved on higher layers relatively. generally speaking USB communication is not available in the early stages of booting. I'm not sure neither how or if Apple provides it magically nor the entire boot sequence so I'm gonna learn them
#asahi-re · 2021-01-07 14:48
puhitakuyeah basically i hope it works well in any aspect of analysis
#asahi-re · 2021-01-07 14:35
davidrysk[m]puhitaku: you probably won't have to, if anything that special USB cable would be of use
#asahi-re · 2021-01-07 14:08
puhitakui've just purchased an M1 MBA just for experiments and i'd be glad to do disassembling and soldering if it's worth to try
#asahi-re · 2021-01-07 01:05
YuzuARM's dropping aarch32 support in the core after Matterhorn
#asahi-re · 2021-01-07 01:04
YuzuA32 is 32b only, A34 is 64b only, A35 is both. closely related cores other than that
#asahi-re · 2021-01-07 01:03
opticronI didn't know about the A34, that's neat
#asahi-re · 2021-01-07 01:02
opticronbut, I figure they will start releasing high end A64-only at some point soon
#asahi-re · 2021-01-07 01:02
opticronthe A77 is a mainstream/generic CPU directly from ARM, so it still makes sense
#asahi-re · 2021-01-07 01:01
opticronheh
#asahi-re · 2021-01-07 01:01
artemistoops
#asahi-re · 2021-01-07 01:01
artemist:q
#asahi-re · 2021-01-07 01:00
artemisttbh i'm surprised that they're still putting A32 and T32 in the Cortex-A77
#asahi-re · 2021-01-07 00:59
Yuzuand the A34, its 64b-only cousin
#asahi-re · 2021-01-07 00:59
artemistNot that anyone seriously used it
#asahi-re · 2021-01-07 00:58
artemistThe best part is that there's an ARMv8-A core that only runs A32 and T32: the Cortex-A32
#asahi-re · 2021-01-06 23:14
Yuzui believe that's the case of ThunderX as well, despite being a totally unrelated core otherwise
#asahi-re · 2021-01-06 23:12
opticronyeah, I remember the thunderx2 having that interesting attribute
#asahi-re · 2021-01-06 23:10
jn__Tegra 124 vs 132
#asahi-re · 2021-01-06 23:10
roxfanpossibly https://en.wikichip.org/wiki/cavium/thunderx2
#asahi-re · 2021-01-06 23:09
YuzuTegra K1-32 was A15, K1-64 was Denver
#asahi-re · 2021-01-06 23:09
opticronsorry, I was mixing denver k1 with tegra k1
#asahi-re · 2021-01-06 23:09
YuzuA15 K1 was aarch32
#asahi-re · 2021-01-06 23:09
YuzuDenver K1 was aarch64
#asahi-re · 2021-01-06 23:08
Yuzuno they weren't
#asahi-re · 2021-01-06 23:08
opticronyeah, but those were A32-only not armv8
#asahi-re · 2021-01-06 23:07
YuzuDenver K1 had no A57's
#asahi-re · 2021-01-06 23:07
opticronDenver also had A32 because it was directly paired with A57s
#asahi-re · 2021-01-06 23:06
roxfanthey used some "server arm" thingy I think
#asahi-re · 2021-01-06 23:06
roxfani've heard corellium had trouble running 32-bit apps because of this issue
#asahi-re · 2021-01-06 23:06
opticronQC's first was SD808 which was bog standard A53/A57 which definitely has A32
#asahi-re · 2021-01-06 23:05
opticronyeah
#asahi-re · 2021-01-06 23:05
j`eynot having A32 is recent i think
#asahi-re · 2021-01-06 23:04
opticrongoes to figure it out
#asahi-re · 2021-01-06 23:04
opticronI could have sworn the first NV and QC armv8 chips had A32
#asahi-re · 2021-01-06 23:02
roxfanused to own a 5s
#asahi-re · 2021-01-06 23:02
roxfanyep
#asahi-re · 2021-01-06 23:01
davidrysk[m]roxfan: the first Apple ARMv8 chips had A32, the newer ones don't
#asahi-re · 2021-01-06 22:27
diz3yah, wow
#asahi-re · 2021-01-06 22:26
j`ey(Patches here for supporting it https://lore.kernel.org/linux-arm-kernel/20201208132835.6151-1-will@kernel.org/)
#asahi-re · 2021-01-06 22:26
j`eyyou can have SoC where's only some CPU's have AArch32 at EL0
#asahi-re · 2021-01-06 22:23
diz3ybut not sure
#asahi-re · 2021-01-06 22:22
diz3yhm, an NV SoC would have n * Denver + n * A57 on the same SoC, I suspect they should have same architectural features
#asahi-re · 2021-01-06 22:13
roxfaniirc first nvidia or Qualcomm ARMv8 chips didn't have A32
#asahi-re · 2021-01-06 22:09
opticronthe AArch32 mode is also optional

plus ›