#asahi · 2021-02-09 10:47
marcanjust like nintendo, NXP, and almost every other silicon vendor#asahi · 2021-02-09 10:47
marcannvidia is on the That Key Recovery Exploit hall of fame#asahi · 2021-02-09 10:47
marcanand then they have bugs#asahi · 2021-02-09 10:47
marcannever_released: for *some* security stuff#asahi · 2021-02-09 10:47
marcannobody ships systems with *all of that* abstracted out behind EL3#asahi · 2021-02-09 10:46
marcanthere are always a million knobs EL2/1 can turn to cause madness#asahi · 2021-02-09 10:46
marcanbkero: have you *seen* the TRM for a modern SoC?#asahi · 2021-02-09 10:46
marcaneveryone else doesn't have the slightest clue#asahi · 2021-02-09 10:46
marcanApple, and Microsoft#asahi · 2021-02-09 10:46
marcanas far as I'm concerned there are two companies who know how to do SoC silicon security#asahi · 2021-02-09 10:45
marcanmodern SoCs are hard to get right enough as it is, trying to share a CPU with secure code is double madness#asahi · 2021-02-09 10:45
marcanand you're back to square 1#asahi · 2021-02-09 10:45
marcanand then that will have bugs#asahi · 2021-02-09 10:45
marcanincluding all power/memory/bus/etc control#asahi · 2021-02-09 10:45
marcanthere is no way to securely do that unless you put half of your kernel in EL3#asahi · 2021-02-09 10:44
marcanjust ask intel about SGX#asahi · 2021-02-09 10:44
marcanbut I'm talking about trying to treat TrustZone as a secure enclave#asahi · 2021-02-09 10:44
marcanyes, that was cute#asahi · 2021-02-09 10:43
marcanthat's why Apple doesn't use it#asahi · 2021-02-09 10:43
marcanthere is no way to implement TrustZone securely in a sane system#asahi · 2021-02-09 10:43
marcanwhich is a fundamentally broken model anyway#asahi · 2021-02-09 10:43
marcanwell again that's just part of the ridiculousness of TrustZone#asahi · 2021-02-09 10:42
marcanyou can't give vendors the power to unilaterally schedule their code or they will abuse it, every time#asahi · 2021-02-09 10:42
marcanthe problem is when it's part of the vendorblob#asahi · 2021-02-09 10:42
marcanif it's open source it's fine#asahi · 2021-02-09 10:41
marcanit's insane#asahi · 2021-02-09 10:41
marcanI heard about some Intel system that had >1k SMIs per second to work around bugs in their xHCI controller#asahi · 2021-02-09 10:41
marcanthe problem is when vendors throw other junk in there, like they do with SMM#asahi · 2021-02-09 10:41
marcanthough I don't really care about EL3 as long as it literally only executes code when called, for PSCI and such#asahi · 2021-02-09 10:40
marcanyeah, that's ARM's fault for pushing TrustZone#asahi · 2021-02-09 10:40
marcanit was always a terrible idea for security, and it hurts other things#asahi · 2021-02-09 10:40
marcangood riddance too#asahi · 2021-02-09 10:38
marcanI actually have a chance at getting decent realtime latencies on these things#asahi · 2021-02-09 10:37
marcanand one thing I'm very glad does not exist on M1#asahi · 2021-02-09 10:37
marcanSMM is another problem with ACPI/x86#asahi · 2021-02-09 10:34
marcanpretty sure that's ACPI's fault#asahi · 2021-02-09 10:34
marcane.g. I have a laptop where the SSD of all things stays warm in sleep mode#asahi · 2021-02-09 10:34
marcanbut even that stuff is often more broken than not#asahi · 2021-02-09 10:34
marcanI was (half) joking#asahi · 2021-02-09 10:33
marcanACPI rarely "implements" anything; more commonly it's just a random pile of hacks and throw-at-the-wall copypasta that happen to work on windows because they were tested there#asahi · 2021-02-09 10:29
marcanACPI is code#asahi · 2021-02-09 10:29
marcandiddledan: dt is a lot simpler and more general than ACPI. it's just telling you what hardware you have#asahi · 2021-02-09 10:23
marcanit does#asahi · 2021-02-09 10:23
marcansomewhere between 5.6 and 5.9 it broke#asahi · 2021-02-09 10:23
marcanI should bisect that at some point#asahi · 2021-02-09 10:23
marcanalso Linux broke the ACPI backlight method on the iMac I'm typing this on at some point... and I wouldn't be surprised if it's the firmware's fault#asahi · 2021-02-09 10:22
marcannobody implements it properly#asahi · 2021-02-09 10:22
marcanACPI is a joke; one of my first Linux patches was a workaround for broken ACPI tables in my laptop at the time#asahi · 2021-02-09 10:20
marcanOpen Firmware existed since, what, 1994?#asahi · 2021-02-09 10:18
marcan*late