#
✕ nick:pipcet[m]
#asahi · 2021-04-29 13:58
pipcet[m]I think you're right for real RT tasks, but I'm just thinking about a performance hack to keep the video player from being interrupted by interrupts (which we will need until we've got the hardware scaler up and running). but, again, just trying to wrap my head around this IRQ-vs-FIQ thing
#asahi · 2021-04-29 13:50
pipcet[m]okay, thought there was a more serious issue. You're right that it's best done in userspace, though.
#asahi · 2021-04-29 13:48
pipcet[m]and what's wrong with running userspace code without enabling IRQs on 7 cores?
#asahi · 2021-04-29 13:29
pipcet[m]just thinking about possibilities here. it'd be silly to ruin a CPU's cache if another CPU is idle
#asahi · 2021-04-29 13:28
pipcet[m]on all but one core, at least :-)
#asahi · 2021-04-29 13:27
pipcet[m]so we could run userspace code with IRQs disabled, if we were using FIQ IPIs?
#asahi · 2021-04-29 13:27
pipcet[m]opticron: good point, we should print out a message for spurious interrupts
#asahi · 2021-04-29 13:22
pipcet[m]IRQs are enabled separately from FIQs, right?
#asahi · 2021-04-29 12:41
pipcet[m]so probably the first CPU to be considered "idle", by whatever standard?
#asahi · 2021-04-29 12:41
pipcet[m]I do wonder how the AIC selects which CPU to send the interrupt to. It appears to be mostly the first two ones to have their bits set, but once in a while it's the third.
#asahi · 2021-04-29 12:21
pipcet[m]turns out disabling your keyboard interrupt still isn't a good idea!
#asahi · 2021-04-29 12:01
pipcet[m](using it with the Corellium drivers, for the time being)
#asahi · 2021-04-29 11:59
pipcet[m]but really, those are very minor issues. The driver appears to work, though I did hack in a few things to get the other device drivers working and need to investigate more.
#asahi · 2021-04-29 11:58
pipcet[m]but with the x86 there's no real difference between the cores, is there :-)
#asahi · 2021-04-29 11:45
pipcet[m]I understand it's probably not worth it to make the change, but I do wonder why we have all the FIQ code in the AIC driver even though they appear to be totally independent. The one dependency is that we can use FIQ or IRQ IPIs, but not both, right?
#asahi · 2021-04-29 11:43
pipcet[m]but overall I'm glad I don't have to use that anymore, the existing device drivers work fine with the new aic driver, with a few modifications :-)
#asahi · 2021-04-29 11:42
pipcet[m]I think the not-to-be-mentioned old driver set several bits and there appeared to be some degree of IRQ balancing going on with it...
#asahi · 2021-04-29 11:02
pipcet[m]but just one at a time, right?
#asahi · 2021-04-29 06:55
pipcet[m]probably my own fault
#asahi · 2021-04-29 06:55
pipcet[m]marcan: thanks, I'll have a look to see why it's not working here :-)
#asahi · 2021-04-29 06:54
pipcet[m]marcan: the AIC driver currently sends all irq interrupts to CPU0, right?
#asahi · 2021-04-26 22:36
pipcet[m]looks like iBoot doesn't obey large alignments. My kernel isn't reliably aligned to 2 MiB, so it fails to boot, but it's just a few bits that are missing, so by sheer luck it will boot once in a while...
#asahi · 2021-04-26 17:54
pipcet[m]so use t8103 in the driver, have both in the DT?
#asahi · 2021-04-26 17:45
pipcet[m](not likely to matter if apple won't release specs, though)
#asahi · 2021-04-26 17:44
pipcet[m]kettenis: I think it might be a good idea to use t8103 (or a non-code name) if that's the only hardware we tested our driver on. We might be relying on unspec'd behavior that differs between those devices.
#asahi · 2021-04-26 17:41
pipcet[m]marcan: that sounds like a good strategy to me!
#asahi · 2021-04-26 05:35
pipcet[m]sven's better :-)
#asahi · 2021-04-26 05:35
pipcet[m]f.close()
#asahi · 2021-04-26 05:35
pipcet[m]f.write(readmem(Int64ul.parse(readmem(get_bootargs() + 12 * 8, 8)) - Int64ul.parse(readmem(get_bootargs() + 1 * 8, 8)) + Int64ul.parse(readmem(get_bootargs() + 2 * 8, 8)), Int32ul.parse(readmem(get_bootargs() + 13 * 8, 4))))
#asahi · 2021-04-26 05:35
pipcet[m]f = open("adt", "wb")
#asahi · 2021-04-26 05:26
pipcet[m]I don't think a reference ADT is a good idea. Legally, it's probably copyrighted. It also contains private information like WiFi keys.
#asahi · 2021-04-24 18:25
pipcet[m]so it's probably a different bug in the Corellium code and not SSPS that's the problem.
#asahi · 2021-04-24 18:24
pipcet[m]sven: i2c.py doesn't work on the USB port Linux used between booting and starting m1n1, but it does work on the other one, and it's still charging when booted into Linux (with the power port removed from the DT) afterwards
#asahi · 2021-04-24 17:46
pipcet[m]sven: your usb code doesn't work if the cable's disconnected! fix it at once ;-)
#asahi · 2021-04-24 16:33
pipcet[m]sven: do I need to run pmpgr_clocks_enable() before running i2c.py? Anything else?
#asahi · 2021-04-24 14:06
pipcet[m]sven; oh, doh, thanks :-)
#asahi · 2021-04-24 13:54
pipcet[m]I've created https://github.com/AsahiLinux/docs/wiki/FW:ADT, feel free to edit or delete as appropriate
#asahi · 2021-04-24 13:18
pipcet[m]true
#asahi · 2021-04-24 13:13
pipcet[m]the SMC gP pins have "names", if you count gP0b...
#asahi · 2021-04-24 13:12
pipcet[m]but since we don't, we can just use a pinctrl device and pin numbers?
#asahi · 2021-04-24 13:07
pipcet[m]we should definitely use pinctrl, maybe I misunderstand what pinmux is for
#asahi · 2021-04-24 12:58
pipcet[m]maybe it's more trouble to use pinmux than it's worth, then?
#asahi · 2021-04-24 12:48
pipcet[m]can we have two pinmux cells instead of splitting one 32-bit pinmux cell into two 16-bit values?
#asahi · 2021-04-24 12:22
pipcet[m]that's why I skip m1n1 entirely.
#asahi · 2021-04-24 12:21
pipcet[m]only for Image Linux, not for a macho
#asahi · 2021-04-24 12:21
pipcet[m]though now you say that I'm not sure that's actually true, m1n1 does the same thing unless you ask it to bring up the secondaries, doesn't it?
#asahi · 2021-04-24 12:20
pipcet[m]so I'm running on one (efficiency...) core until the final kernel's loaded, but it's hardly noticeable
#asahi · 2021-04-24 12:20
pipcet[m]my inital kernel disables SMP for that reason.
#asahi · 2021-04-24 12:19
pipcet[m]will do as soon as I can kexec m1n1
#asahi · 2021-04-24 12:18
pipcet[m]I suspect it's the SPSS that breaks charging though :/

plus ›