#asahi · 2022-01-22 05:26
mjg59DanB91: SPI#asahi · 2022-01-20 11:09
mjg59Taking an ecosystem that thrives on low-level differentiation and trying to re-model it on an ecosystem that has spent decades trying to be 100% compatible with a computer released in 1981 turns out to be hard#asahi · 2022-01-20 11:05
mjg59I'm still confused how Toshiba ended up involved in 1.0#asahi · 2022-01-20 11:01
mjg59ACPI suffers hugely from being a spec intended to solve a specific problem ("How do we provide some sort of abstraction layer for x86 that isn't int calls") that got kind of out of control#asahi · 2022-01-07 05:29
mjg59It is literally unable to prevent you booting the OS of your choice#asahi · 2022-01-07 05:29
mjg59The firmware running on Pluton in the AMD chips just exposes TPM functionality#asahi · 2021-10-18 20:09
mjg59Thanks!#asahi · 2021-10-18 20:07
mjg59marcan: I'm sure you mentioned this at some point, but I can't find anything online - does M1 implement Trustzone?#asahi · 2021-10-14 20:05
mjg59landscape15[m]: The Fedora GUI installer is written in Python. There's no need to use C for this kind of thing.#asahi · 2021-10-07 07:58
mjg59You need to punt decisions out to something that has more contextual understanding, and to toolkits that can make use of that extra context#asahi · 2021-10-07 07:58
mjg5996 is used as an override for the core X value because you can't assume the server made a good decision when confronted with a 1024x768 laptop and a 4K projector#asahi · 2021-10-07 07:57
mjg59tbf the 96 thing is also because the core X protocol has no concept of multiple displays with multiple DPIs#asahi · 2021-10-07 07:18
mjg59Oh, actually, https://lists.fedoraproject.org/pipermail/devel/2011-October/157760.html is more what I was thinking of and was literally over a decade ago#asahi · 2021-10-07 07:11
mjg59https://mjg59.dreamwidth.org/14802.html wow how did I write that in 2012#asahi · 2021-10-07 07:10
mjg59Not any more#asahi · 2021-06-18 17:44
mjg59It's hitting basically everything on the internet that's running an IRC daemon#asahi · 2021-05-27 22:12
mjg59Other than the extent that there's GPU support required, I think it's probably out of scope#asahi · 2021-05-27 22:11
mjg59rkjnsn[m]: The state of HDR in Linux overall isn't great at the moment#asahi · 2021-05-04 18:53
mjg59kexec_file() does the glue code stuff in-kernel instead, which means it can do signature validation (purgatory is hotpatched from userland with kernel data, so can't be signed)#asahi · 2021-05-04 18:52
mjg59The glue code is responsible for setting up state before entering the new kernel#asahi · 2021-05-04 18:52
mjg59kexec() basically just loads a kernel into RAM, loads some glue code (purgatory), tells the glue code where the kernel was loaded to and then jumps to the glue code#asahi · 2021-05-04 18:51
mjg59https://twitter.com/hugelgupf/status/1158466881373630464 is Windows being kexec()ed from Linux#asahi · 2021-05-04 18:50
mjg59kexec_file() assumes Linux#asahi · 2021-05-04 18:49
mjg59kexec() could absolutely load FreeBSD - you'd need a different purgatory#asahi · 2021-04-05 02:40
mjg59Communities vary on that a lot, but Cc:ing the poster is normal on kernel-related lists#asahi · 2021-02-17 06:51
mjg59Pushing RAM out to non-volatile storage is great if you're expecting to be asleep for an extended period of time, but in general this is implemented by suspending to RAM (and keeping the RAM in self refresh) and then setting a wakeup timer for some point in the future, and if that goes off then suspending to storage instead#asahi · 2021-02-17 06:49
mjg59(Actually more than that, because if you've powered down RAM you're going to need to go through firmware and start a bootloader that's able to start restoring that image and there's probably going to be some relocation involved as well)#asahi · 2021-02-17 06:48
mjg59Glanzmann: If you only get something like ~3GB/sec, it'd take ~5 seconds to restore RAM from NVMe#asahi · 2021-02-17 05:35
mjg59Glanzmann: Memory in self refresh takes very little power#asahi · 2021-01-08 13:15
mjg59If you're able to get to u-boot you can present a uefi layer#asahi · 2021-01-08 13:15
mjg59Yes it does#asahi · 2021-01-08 13:15
mjg59Shiz: ?#asahi · 2021-01-08 13:03
mjg59But there's no inherent UEFI#asahi · 2021-01-08 13:02
mjg59Lightsword: In theory u-boot could be used to provide a UEFI environment#asahi · 2021-01-08 12:32
mjg59raster: And they haven't gone after the jailbreak community in any meaningful way, even when they're working on dev releases#asahi · 2021-01-08 12:30
mjg59raster: I'm not aware of any of Apple's security models that rely on obfuscation#asahi · 2021-01-08 11:55
mjg59Once you're targeting a sufficiently low level it's easy enough to ensure that you can transform binaries#asahi · 2021-01-08 11:54
mjg59Yeah#asahi · 2021-01-08 11:54
mjg59(We use objcopy in the shim build process to transform an ELF object into a PE one)#asahi · 2021-01-08 11:53
mjg59Or you can use objcopy to transform them, but that's not going to work well in most cases#asahi · 2021-01-08 11:52
mjg59You build a toolchain that targets mach-O#asahi · 2021-01-08 05:42
mjg59https://github.com/ARM-software/u-boot/blob/master/doc/README.uefi#asahi · 2021-01-08 05:41
mjg59But once you're in u-boot you can do UEFI - there's support for providing enough UEFI boot services in u-boot to launch Linux#asahi · 2021-01-08 05:40
mjg59davidrysk[m]: I think that's (2), although you could theoretically have a first stage bootloader that just chained into u-boot#asahi · 2021-01-08 05:40
mjg59And you'd need to boot into recovery every time you update the kernel#asahi · 2021-01-08 05:39
mjg59(3) makes it easier to work with existing distros, but puts a lot more complexity in#asahi · 2021-01-08 05:38
mjg59Alex[m]17: Choices are basically: (1) kernel that looks like an iBoot payload (bad), (2) first stage bootloader that boots Linux directly, (3) first stage bootloader that presents a UEFI environment#asahi · 2021-01-08 05:31
mjg59So an equivalent model works well here#asahi · 2021-01-08 05:31
mjg59In UEFI land we have Shim as a simple first stage bootloader that changes rarely, and which then boots the actual bootloader#asahi · 2021-01-08 05:29
mjg59And yeah, it makes much more sense to have a simple first stage bootloader