|
what is the mechanism by which terminating launchd can render your system unusable without a reinstall?
|
#
?
Jan 15, 2022 15:54
|
|
|
|
| # ? Apr 26, 2024 08:30 |
|
|
infernal machines posted:what is the mechanism by which terminating launchd can render your system unusable without a reinstall? should have waited until it was safe to turn off your computer 
|
|
#
?
Jan 15, 2022 15:59
|
|
|
read the explanation in this of why people want to collect blu ray disks to instantly crumble into dust
|
|
#
?
Jan 15, 2022 16:01
|
|
|
they didnt even talk about the best part of bluray - the extremely high bitrate
|
|
#
?
Jan 15, 2022 16:07
|
|
|
Jenny Agutter posted:read the explanation in this of why people want to collect blu ray disks to instantly crumble into dust
|
|
#
?
Jan 15, 2022 16:11
|
|
|
Jabor posted:they're talking about this bit: The BSD process model is one of the things macOS inherited from BSD going all the way back to when NeXT was making NeXTSTEP (NeXTSTEPs kernel combined code from Mach from CMU and BSD from UCB); the modern implementation in macOS is from FreeBSD 7 (ie. that's the last time Apple did a major source pull from FreeBSD, and the process model hasn't changed since). In the BSD process model, the init process (whether it's BSD init found in the BSDs or launchd in macOS, has special handling for either going to single-user mode or shutting down the system. With all of that said, even if you could kill the init process using SIGKILL, which you can't, sending SIGKILL to a process isn't supposed to alter the binary executable image that's stored on the disk - and if it does, that's a huge bug. BlankSystemDaemon fucked around with this message at 18:07 on Jan 15, 2022 |
|
#
?
Jan 15, 2022 17:00
|
|
|
Jabor posted:they're talking about this bit: it is also completely incorrect. if you somehow manage to kill launchd and you need to reinstall the os afterwards, then you needed to reinstall the os before that and just didn't notice until you restarted
|
|
#
?
Jan 15, 2022 18:01
|
|
|
Carthag Tuek posted:turns out macos 10:13 can have a full screen ad that you cant stop w keyboard or mouse
|
|
#
?
Jan 15, 2022 18:03
|
|
|
things are getting weird in the secfuck thread
|
|
#
?
Jan 15, 2022 18:04
|
|
|
Carbon dioxide posted:... force quitting a task should not brick your system, come on mac you can do better than that. infernal machines posted:what is the mechanism by which terminating launchd can render your system unusable without a reinstall? Apparently lock files can be left behind on the drive that can halt further progress, even after the system is restarted. Note that I'm basing this on several peoples' described experiences on the Internet. This may no longer be the case.
|
|
#
?
Jan 15, 2022 23:15
|
|
|
Quackles posted:Apparently lock files can be left behind on the drive that can halt further progress, even after the system is restarted.
|
|
#
?
Jan 16, 2022 02:55
|
|
|
lock files aren’t always using flock
|
|
#
?
Jan 16, 2022 02:56
|
|
|
Subjunctive posted:lock files aren’t always using flock flocka locka flame
|
|
#
?
Jan 16, 2022 03:06
|
|
|
Subjunctive posted:lock files aren’t always using flock it's just metadata stored in the kernel and you can't hold a lock if your kernel crashed leading up to this fantasy land scenario we're talking about i also don't understand how lock files could possibly block kernel device initialization, fsck, mountroot or init/its functional equivalent
|
|
#
?
Jan 16, 2022 04:03
|
|
|
Here is the extremely unsubstantiated evidence. https://www.reddit.com/r/mac/comments/536bg2/help_i_stupidly_deleted_my_launchd_from_my_mac/ https://forums.macrumors.com/threads/accidentally-quit-launchd-process-now-chrome-wont-work.871126/
|
|
#
?
Jan 16, 2022 04:08
|
|
|
BlankSystemDaemon posted:lock files exist because bad developers don't use flock(2) or fcntl(2) properly (ie. via the aforementioned library functions or if all else fails, lockf(1)) You don't understand how software could have been written to attempt to acquire a lock before doing something critical to starting up the user environment?
|
|
#
?
Jan 16, 2022 04:09
|
|
|
Quackles posted:Here is the extremely unsubstantiated evidence. did you actually read either of those?
|
|
#
?
Jan 16, 2022 04:10
|
|
|
Jabor posted:You don't understand how software could have been written to attempt to acquire a lock before doing something critical to starting up the user environment? locks should be on a ram disk 
|
|
#
?
Jan 16, 2022 04:11
|
|
|
infernal machines posted:did you actually read either of those? Yes. They're kind of patchy, even taken together. Guy #1 seems to have broken his computer during or after a force quit of launchd, while Guy #2's woes may have been caused by one or more filtering softwares he had installed. That's why I called them "extremely unsubstantiated".
|
|
#
?
Jan 16, 2022 04:11
|
|
|
i mean, the second person outright refutes the claim themselvesquote:so i think launchd had nothing to do with it (other than forcing me to restart). it was all glimmerblocker. in trying to uninstall glimmerblocker completely, i reinstalled it and uninstalled it...which messed up the proxy-whatever (in these grounds i don't know what the heck i'm talking about). so i followed the advice of one post on another site and went to safari>>preferences>>advanced>>change settings... and unchecked "web proxy" both those anecdotes involved killing launchd but only one even resulted in an unbootable system, and that's because they also deleted it
|
|
#
?
Jan 16, 2022 04:15
|
|
|
very careless of apple to design an operating system that doesn't work without an init process
|
|
#
?
Jan 16, 2022 04:26
|
|
|
Jabor posted:You don't understand how software could have been written to attempt to acquire a lock before doing something critical to starting up the user environment? carry on then posted:very careless of apple to design an operating system that doesn't work without an init process
|
|
#
?
Jan 16, 2022 04:26
|
|
|
carry on then posted:very careless of apple to design an operating system that doesn't work without an init process they even kinda fixed that. these days the os is installed to a cryptographically signed read-only volume that can only be modified from recovery mode, so the only way to not have an init process is to not have a copy of the os at all.
|
|
#
?
Jan 16, 2022 05:11
|
|
|
Plorkyeran posted:they even kinda fixed that. these days the os is installed to a cryptographically signed read-only volume that can only be modified from recovery mode, so the only way to not have an init process is to not have a copy of the os at all.
|
|
#
?
Jan 16, 2022 05:14
|
|
|
BlankSystemDaemon posted:this is good for security apple walled garden enemies of FREEDOM
|
|
#
?
Jan 16, 2022 05:18
|
|
|
Plorkyeran posted:they even kinda fixed that. these days the os is installed to a cryptographically signed read-only volume that can only be modified from recovery mode, so the only way to not have an init process is to not have a copy of the os at all. hmm. how much space is available on the TPM chip to the OS? I think I have an idea... e: poo poo, nvm "The TPM SHALL support allocation of at least 68 indices, with a total minimum data size of 3834 Bytes (decimal)."
|
|
#
?
Jan 16, 2022 05:22
|
|
|
carry on then posted:apple walled garden enemies of FREEDOM I mean, I modified mine so I could have a neat tiling manager, so its not *completely* locked down
|
|
#
?
Jan 16, 2022 05:31
|
|
|
BlankSystemDaemon posted:lock files exist because bad developers don't use flock(2) or fcntl(2) properly file lock != lock file lock files are usually just open(O_EXCL | O_CREAT)
|
|
#
?
Jan 16, 2022 05:38
|
|
|
god drat it beaten ^^^
|
|
#
?
Jan 16, 2022 05:51
|
|
|
Subjunctive posted:file lock != lock file
|
|
#
?
Jan 16, 2022 05:54
|
|
|
BlankSystemDaemon posted:you didn't say file lock until just then, though indeed, you are the one who first mentioned file locks as though they were somehow related
|
|
#
?
Jan 16, 2022 06:19
|
|
|
flock(O_SEAGULLS)
|
|
#
?
Jan 16, 2022 06:31
|
|
|
BlankSystemDaemon posted:this is good for security mostly. the big downside is that it now takes multiple hours to install a security update for safari instead of five minutes and that's gonna reduce how many people are fully up-to-date.
|
|
#
?
Jan 16, 2022 07:23
|
|
|
Isn't Safari on the main volume?
|
|
#
?
Jan 16, 2022 09:57
|
|
|
You people are bricking the thread with this endless discussion.
|
|
#
?
Jan 16, 2022 10:13
|
|
|
distributed denial of posting
|
|
#
?
Jan 16, 2022 10:16
|
|
|
Captain Foo posted:flock(O_SEAGULLS) you'd think that in 2022 they could afford a few more characters in their function names and constants
|
|
#
?
Jan 16, 2022 10:18
|
|
|
byte wise, mutex foolish
|
|
#
?
Jan 16, 2022 10:34
|
|
|
Captain Foo posted:flock(O_SEAGULLS) run_away(DISTANCE_FAR)
|
|
#
?
Jan 16, 2022 14:35
|
|
|
|
| # ? Apr 26, 2024 08:30 |
|
|
Captain Foo posted:flock(O_SEAGULLS) 
|
|
#
?
Jan 16, 2022 15:19
|
|