#asahi · 2021-02-10 17:12
kettenismaz right, and I suspect that works fine#asahi · 2021-02-10 17:11
kettenisj'ey I don't know, but if you don't compile with -mgeneral-regs-only varargs functions such as snprintf save the fpu registers on the stack#asahi · 2021-02-10 17:10
kettenisis that not how other VHE supporting CPUs behave with W2H on?#asahi · 2021-02-10 17:09
ketteniswith E2H on, cptr_el2 has the same layout as cpacr_el1 (because they are aliased I suppose)#asahi · 2021-02-10 17:07
kettenisright but the layout of cptr_el2 changes if E2H is enabled#asahi · 2021-02-10 17:06
kettenistypically that is snprintf#asahi · 2021-02-10 17:06
kettenisand u-boot dies when the first varargs function gets hit#asahi · 2021-02-10 17:05
ketteniswithout that change it does *not* enable the fpu#asahi · 2021-02-10 17:05
kettenisno, I can't#asahi · 2021-02-10 17:01
kettenisso every byte counts ;)#asahi · 2021-02-10 17:01
ketteniswell, this is common code and on some boards it runs in SRAM#asahi · 2021-02-10 17:00
kettenisah, so that could save me three instructions#asahi · 2021-02-10 16:57
kettenisthe bit where it enables the fpu?#asahi · 2021-02-10 16:44
kettenisonly tested with the corellium pre-loader for now#asahi · 2021-02-10 16:43
kettenisthat can now load an UEFI bootloader from a usb disk connected to the type-A ports n the mini#asahi · 2021-02-10 16:42
kettenishttps://github.com/kettenis/u-boot/tree/apple-m1#asahi · 2021-02-10 14:43
kettenisthat would fit the OpenBSD bus/driver model nicely and U-Boot wouldn't worry about it#asahi · 2021-02-10 14:22
ketteniscorrect (as far as we know at this point)#asahi · 2021-02-10 14:19
kettenispart of the reason we (OpenBSD) don't need this flag is because we use nGnRnE by default#asahi · 2021-02-10 11:06
kettenisbus_space(9) is a fairly clever design, but the API makes your drivers a bit wordy#asahi · 2021-02-10 11:03
kettenisbut TBH I don't think you'd want to overhaul this part of the linux kernel design#asahi · 2021-02-10 11:03
ketteniswith some cleverness you could avoid the function pointer overhead for the actual access#asahi · 2021-02-10 11:01
kettenisalthough we don't actually use it for i2c#asahi · 2021-02-10 11:00
kettenisindeed#asahi · 2021-02-10 10:55
kettenishaving #addr-cells be 3 outside of pci space would be somewhat painful#asahi · 2021-02-10 10:54
kettenisso I think I can get away with that without any hints in the device tree#asahi · 2021-02-10 10:53
kettenisfor OpenBSD I was planning to use the bus_space(9) abstraction to make sure PCIe mmio address space gets mapped nGnRE#asahi · 2021-02-10 00:01
kettenisneed to clean up a few things before pushing the changes to my u-boot github repository#asahi · 2021-02-10 00:00
kettenisusb works and I can type commands using a usb keyboard and load and run the OpenBSD/arm64 UEFI bootloader#asahi · 2021-02-09 23:59
kettenisÃmade some progress on u-boot#asahi · 2021-02-09 10:41
kettenisdunno, having an open source firmware in EL3 that abstracts some of the platform specifics makes sense to me#asahi · 2021-02-09 10:36
kettenistypical PC ACPI code just triggers an SMI and lets the firmware handle the request in SMM mode#asahi · 2021-02-09 10:24
ketteniswe have our own aml interpreter in OpenBSD and I'm seeing issues all the time where AML makes assumptions that aren't justified by the language in the standard#asahi · 2021-02-09 10:23
kettenisthe acpi standard is hopelessly ambiguous#asahi · 2021-02-09 10:17
kettenisand sun before that#asahi · 2021-02-09 10:16
kettenisthe hard work was done for powerpc#asahi · 2021-02-09 01:11
kettenisanyway, zzz time here#asahi · 2021-02-09 01:11
kettenisnot yet; planning on giving that a go once the serial adapter arrives#asahi · 2021-02-09 01:08
kettenisbtw, if you have m1n1 bring up pcie, I can ditch pretty much all of the corellium code I'm now using ;)#asahi · 2021-02-09 01:03
kettenisme and a soldering iron don't get along very well#asahi · 2021-02-09 01:02
kettenisheh, that'd be great#asahi · 2021-02-09 00:59
kettenisI use the curl trick#asahi · 2021-02-09 00:59
kettenisBluerise sent me one but it got stuck in a snowstorm#asahi · 2021-02-09 00:59
kettenisnot yet#asahi · 2021-02-09 00:57
kettenisbut there is still something wrong as commands from the CPU to the controller time out#asahi · 2021-02-09 00:56
kettenismanaged to get the xhci(4) on the pcie bus to do DMA in u-boot#asahi · 2021-02-09 00:55
kettenisah, yes, I probably should get some zzz#asahi · 2021-02-09 00:24
kettenisinteresting, so 1G of mmio space for the TB4 PCIe controllers and 1.5G (512M + 1G) for the "normal" PCIe controller#asahi · 2021-02-07 21:43
kettenisgiven that arm and aarch64 are different archotectures, this doesn't really matter#asahi · 2021-02-07 19:31
ketteniswell, IMNSHO Red Hat choosing not to support device trees in RHEL is their problem