#
✕ nick:marcan
#asahi-dev · 2021-01-31 05:06
marcanPPC is way saner, at least the docs
#asahi-dev · 2021-01-31 05:06
marcanI used to like ARM, but I think I don't any more
#asahi-dev · 2021-01-31 05:05
marcanand yeah, half the documentation is incomprehensible
#asahi-dev · 2021-01-31 05:05
marcan0x1201111100001011
#asahi-dev · 2021-01-31 05:05
marcan>>> mrs(ID_AA64MMFR2_EL1)
#asahi-dev · 2021-01-31 05:05
marcan0x11212100
#asahi-dev · 2021-01-31 05:05
marcan>>> mrs(ID_AA64MMFR1_EL1)
#asahi-dev · 2021-01-31 05:05
marcan0x12120f100001
#asahi-dev · 2021-01-31 05:05
marcan>>> mrs(ID_AA64MMFR0_EL1)
#asahi-dev · 2021-01-31 05:03
marcanE2H?
#asahi-dev · 2021-01-31 05:03
marcanwhich ones?
#asahi-dev · 2021-01-31 05:00
marcanlike MCE on x86
#asahi-dev · 2021-01-31 04:59
marcanit's an asynchronous exception
#asahi-dev · 2021-01-31 04:59
marcanSystem Error
#asahi-dev · 2021-01-31 04:59
marcanand it might break other things
#asahi-dev · 2021-01-31 04:59
marcanbut that still means you're doing something wrong
#asahi-dev · 2021-01-31 04:59
marcanand those can be disabled in the control register
#asahi-dev · 2021-01-31 04:58
marcanbut a bad write, or speculation, or any of that stuff, is just SError
#asahi-dev · 2021-01-31 04:58
marcanactually a bad read will report a sync error
#asahi-dev · 2021-01-31 04:58
marcanwell a bad read or write from unmapped address ranges just does nothing, other than report the error
#asahi-dev · 2021-01-31 04:57
marcananyway, the L2 error reporting can be disabled, which is what corellium do - presumably because they have bad address faults lurking and they didn't want to bother fixing them
#asahi-dev · 2021-01-31 04:57
marcanlike the PC wasn't advancing at all
#asahi-dev · 2021-01-31 04:56
marcanso... what else would it be?
#asahi-dev · 2021-01-31 04:56
marcanmodwizcode: it was happening on *return* from a page fault
#asahi-dev · 2021-01-31 04:56
marcanand then I found this https://mrcn.st/t/Screenshot_20210131_135610.png
#asahi-dev · 2021-01-31 04:55
marcanand then I thought something must be broken because it's nGnRnE mapped and yet something was still hitting it speculatively... in what looked like an instruction fetch path even...
#asahi-dev · 2021-01-31 04:55
marcanso I tried just unmapping zero and it worked
#asahi-dev · 2021-01-31 04:55
marcanso... it sounded like something was hitting zero
#asahi-dev · 2021-01-31 04:54
marcanwell the L2 subsystem address was basically zero (ignoring the high bits which seem to have a special meaning)
#asahi-dev · 2021-01-31 04:54
marcansure
#asahi-dev · 2021-01-31 04:53
marcanthough the problem happened with caches off too, just with MMU on
#asahi-dev · 2021-01-31 04:53
marcancould be
#asahi-dev · 2021-01-31 04:35
marcanthis thing really does punch you in the face if you miss any of these details, moreso than any other CPU I've ever used
#asahi-dev · 2021-01-31 04:34
marcanspeculation/prediction strikes again
#asahi-dev · 2021-01-31 04:34
marcananyway
#asahi-dev · 2021-01-31 04:34
marcanbut then again I only saw this behavior, so far, on return from a page fault, which can only happen with the MMU on, so...
#asahi-dev · 2021-01-31 04:34
marcanthat is very weird, though, because you'd *think* that means it would also break with the MMU off, where there is no such thing as non-executable...
#asahi-dev · 2021-01-31 04:33
marcanthe M1 decided to speculate an instruction fetch from address 0.
#asahi-dev · 2021-01-31 04:33
marcanfigured out the sError thing. turns out mapping device memory as nGnRnE isn't enough to guard it, you need to also mark it non-executable...
#asahi-dev · 2021-01-30 17:42
marcanuhh if I'm fast enough I might have *prototypes* to send out to some people within a month :p
#asahi-dev · 2021-01-30 17:41
marcanmaybe starting within a week?
#asahi-dev · 2021-01-30 17:40
marcanI'm going to be designing a proper m1 serial cable soonish, will stream most of it
#asahi-dev · 2021-01-30 17:22
marcan*0806
#asahi-dev · 2021-01-30 17:22
marcan*08056
#asahi-dev · 2021-01-30 17:22
marcanI have resistor books with all of E24? I think it was? in 0803 and 0604
#asahi-dev · 2021-01-30 17:20
marcanhah :)
#asahi-dev · 2021-01-30 17:20
marcan1/8W 10% would be fine too, by the way ;)
#asahi-dev · 2021-01-30 17:20
marcanwhat, you mean you don't have the whole E6 range of resistors handy already? :-)
#asahi-dev · 2021-01-30 16:56
marcanI picked the values to be in the range of any sane driver from the standard E6 list
#asahi-dev · 2021-01-30 15:07
marcanwonder if I'll get a pretty picture in the morning

plus ›