#asahi · 2023-02-06 23:05
karpouzihmm actually the part that was actually stuck was 'stop-sigterm'#asahi · 2023-02-06 23:04
karpouzisystemd[1]: `using hardwawre watchdog 'Apple SoC Watchdog', version 0, device /dev/watchdog0; Failed to set timeout to 10min: Invalid argument#asahi · 2023-02-06 23:03
karpouzihmm, actually no#asahi · 2023-02-06 22:57
karpouzilooks like this is probably systemd-networkd and networkmanager fighting each other#asahi · 2023-02-06 22:55
karpouzianyone know how to get logs off a machine while it's shutting down?#asahi · 2023-02-06 22:54
karpouzirather, sysrq does something, but the TTY says "this sysrq is disabled"#asahi · 2023-02-06 22:53
karpouziit eventually shuts down or reboots as requested, but the hanging is odd. haven't seen it on other machines#asahi · 2023-02-06 22:53
karpouzii'm having an odd issue - an M1 mac mini hangs on the shutdown TTY for about 2-3 minutes with no error messages. sysrqs seem to do nothing?#asahi · 2023-02-03 23:54
karpouziwish I could just buy an M1 machine in an iPhone form factor. it's just that They don't want us to write apps on hardware that we own b/c they can't monetize them that way #asahi · 2023-02-03 23:51
karpouzithanks for the link sven, i wasn't able to find that kind of doc through a simple search#asahi · 2023-02-03 23:49
karpouzibut actually they're just using some nvmem in the chip to most of the heavy lifting so they can still outsource the flash/ram#asahi · 2023-02-03 23:49
karpouzibut i had been under the impression that was what Apple did also#asahi · 2023-02-03 23:48
karpouzii guess intel and micron and samsung probably already have patents for RAM/flash controllers with transaction signing#asahi · 2023-02-03 23:42
karpouzii had thought Apple went to the length of getting some special ones that would reject readout without signed transactions#asahi · 2023-02-03 23:42
karpouziyeah, but the flash chips are apparently just some ordinary flash chips right.#asahi · 2023-02-03 23:41
karpouziand try to read out hundreds of gigabytes or terabytes of data manually without the readout circuits' help#asahi · 2023-02-03 23:40
karpouziwhat i mean is that it's completely impractical for someone to take a flash chip whose readout circuitry won't accept or respond to any transactions that aren't signed#asahi · 2023-02-03 23:40
karpouzihmm#asahi · 2023-02-03 23:40
karpouzithat's actually fairly practical#asahi · 2023-02-03 23:40
karpouzigetting the internal keys isn't that bad if they're only 256 bits or whatever on some silicon#asahi · 2023-02-03 23:39
karpouziso perhaps only the M1 could even read the flash or something#asahi · 2023-02-03 23:39
karpouzitpw_rules: i would have expected some level of flash security e.g. additional anti-rollback fuse bits in the flash to have been used#asahi · 2023-02-03 23:37
karpouzii saw that comic when it came out ok#asahi · 2023-02-03 23:37
karpouziboring#asahi · 2023-02-03 23:36
karpouzithe scenario that still exists is this: i have some file on my machine, they clone the chips and hold onto the clones, i destroy my machine by breaking the PCB (snapping the flash, but the M1 CPU survives), they extract the keys from my M1 and decrypt my drive#asahi · 2023-02-03 23:35
karpouzitheir M1's keys don't match my M1's keys, so they won't be able to boot the cloned chips, right?#asahi · 2023-02-03 23:35
karpouziwhat do you mean 'boot a clone on their machine'? #asahi · 2023-02-03 23:34
karpouzithe point of the exercise is that they get copies of my data, in a way which I cannot detect, and they "can" decrypt it later on provided they somehow acquire my machine's internal keys#asahi · 2023-02-03 23:33
karpouzithat's good i suppose#asahi · 2023-02-03 23:32
karpouzioh, so as long as they don't have my M1 CPU, they can't make it boot up with my flash?#asahi · 2023-02-03 23:32
karpouzii was under the impression there's some NVMEM in the M1 which keeps track of the serial number of the flash or something#asahi · 2023-02-03 23:31
karpouzihmm, that's disturbing#asahi · 2023-02-03 23:31
karpouzifollow up: if they took #asahi · 2023-02-03 23:30
karpouziso i have an incredibly stupid question. if someone took my M1 machine, desoldered the flash chips, cloned the contents of each flash chip to identical parts, then soldered the clones back to my board, would I be able to tell the difference?#asahi · 2023-02-03 05:51
karpouzioh, interesting. thanks for the answer marcan!#asahi · 2023-02-03 05:43
karpouzithere's an opengl error when i fire it up: "no openGL context found". i take it the target with asahi-gpu is to drive development on vulkan -- not opengl right?#asahi · 2023-02-03 05:42
karpouzianyone tried out blender either?#asahi · 2023-02-03 05:14
karpouzinobody else seen this?#asahi · 2023-02-03 03:29
karpouzis/error/bug#asahi · 2023-02-03 03:28
karpouzifar from a priority but it would make sense to make an error report for this, right?#asahi · 2023-02-03 03:27
karpouzitg3 driver looks completely unchanged#asahi · 2023-02-03 03:22
karpouzipoints to bcm57762. but i'd expect that to not have changed vs upstream right#asahi · 2023-02-03 03:14
karpouzisame issue with netperf :/#asahi · 2023-02-03 03:02
karpouzii guess this is an iperf3 problem, not an asahi problem xD#asahi · 2023-02-03 03:02
karpouziiperf3 can send avg of 936 mbit/s *to* an M1 mini from a random e1000e laptop, but the m1 itself can't seem to send faster than 836 mbit/s#asahi · 2023-02-03 02:57
karpouzii've disabled EEE fwiw#asahi · 2023-02-03 02:57
karpouzianyone noticed that native ethernet throughput between two M1 minis isn't quite full speed?#asahi · 2023-01-31 03:30
karpouzieeyup#asahi · 2023-01-31 03:29
karpouzioh maybe it's because i've been leaving a thunderbolt to HDMI adapter plugged in >.<#asahi · 2023-01-31 03:28
karpouziseems to hang there for a minute or two on the m1 macmini.