#
✕ nick:kettenis
#asahi · 2021-02-27 17:57
kettenisthe way I read it is that dmb is enough to guarantee the ordering
#asahi · 2021-02-22 14:32
kettenishardware that icorrectly tags a BAR as prefetchable exists
#asahi · 2021-02-22 14:09
kettenis(write-combining tends to speed things up a lot though)
#asahi · 2021-02-22 14:08
kettenisso it doesn't really offer a speed benefit over write-combining
#asahi · 2021-02-22 14:07
kettenisbtw, the reason nobody bothers mapping framebuffers write-through is that most grapphics code these days only writes to the framebuffer
#asahi · 2021-02-22 14:04
kettenismy understanding is that you need a memory barrier to guarantee that store buffers have been flushed
#asahi · 2021-02-21 15:31
kettenisif m1n1 turns into some sort of runtime firmware component to implement psci it could also implement scmi to do clock management
#asahi · 2021-02-21 15:29
kettenisclock management I mean
#asahi · 2021-02-21 15:28
kettenisalthough I think that if you want to apply the tunables, you'll need to do clocl management as well
#asahi · 2021-02-21 15:27
kettenisnothing really *needs* to move into m1n1
#asahi · 2021-02-21 15:25
kettenisyes, u-boot does the tunable pokes
#asahi · 2021-02-21 12:44
kettenisI don't have any thunderbolt hardware so that isn't really a priority for me
#asahi · 2021-02-21 12:40
kettenischeating a bit by doing a 1:1 mapping of about 1G of RAM and telling the kernel to use that block for DMA
#asahi · 2021-02-21 12:39
kettenisSince I haven't found a way to but the DART in bypass mode I manage the DART from OpenBSD
#asahi · 2021-02-21 12:37
kettenisI also rely on U-Boot enabling the DART clocks
#asahi · 2021-02-21 12:37
kettenisI currently rely on U-Boot bringing up PCIe which involves some clock gating, pin muxing, gpio poking and a hanshake with the SMC over a mailbox
#asahi · 2021-02-15 18:47
kettenisgood luck finding an ss20 with a working idprom
#asahi · 2021-02-15 15:41
kettenisthat should work; most DTs don't specify a baudrate
#asahi · 2021-02-15 14:28
kettenisI can see why you want it for uploading binaries, but as a default 115200 would be better
#asahi · 2021-02-15 14:27
kettenismany usb-to-serial converters don't support it and at a connection at that speed is quite susceptible to line noise
#asahi · 2021-02-15 14:26
kettenismarcan: 1.5M is somewhat annoying as a default baudrate for a serial console
#asahi · 2021-02-15 12:40
kettenisis that intentional, or is that just a placeholder that gets filled in by m1n1 with the real baudrate?
#asahi · 2021-02-15 12:39
kettenisstdout-path = "serial0:1500000";
#asahi · 2021-02-14 15:58
kettenisgood enough for a serial console ;)
#asahi · 2021-02-14 15:56
kettenisand a uart drive could be implemented in a way such that it simply skips setting the baudrate if it can't determine the clock frequency of the clock feeding it
#asahi · 2021-02-14 15:54
kettenisif fimrware (e.g. m1n1) sets up clocks and power a kernel without proper clock/power support might still work
#asahi · 2021-02-14 15:39
kettenisnot necessarily
#asahi · 2021-02-14 12:44
kettenisthe idea is that it is a proprty of the bus isn't it? not of the devices themselves
#asahi · 2021-02-14 12:44
kettenismaybe a separate file in bindings/bus/ is the way to go?
#asahi · 2021-02-13 18:58
ketteniswhat makes you think so?
#asahi · 2021-02-13 18:12
kettenisbut without further work on m1n1 it won't do anything useful
#asahi · 2021-02-13 18:11
kettenisit should be possible to use the serial console and see output in the framebuffer console
#asahi · 2021-02-13 18:10
kettenisat that point the u-boot pcie driver can be replaced with something that is much simpler
#asahi · 2021-02-13 18:09
kettenismarcan is planning to fully initialize pcie in m1n1
#asahi · 2021-02-13 18:08
kettenisyes, the corellium preloader copies these over from the apple device tree
#asahi · 2021-02-12 10:57
kettenisnope
#asahi · 2021-02-12 10:51
kettenisbut I'll adjust things as those come along
#asahi · 2021-02-12 10:50
kettenisso I can't really upstream things very well until we have "official" bindings
#asahi · 2021-02-12 10:49
kettenisso one issue with my u-boot work is that up until now it uses the device tree bindings invented by corellium
#asahi · 2021-02-12 10:40
kettenisOpenBSD doesn't actually run yet since we don't have working interrupts yet
#asahi · 2021-02-12 10:39
kettenisand I have USB support as well so a USB keyboard works
#asahi · 2021-02-12 10:38
kettenisbut EFI GOP is there so it should work with Linux as well
#asahi · 2021-02-12 10:38
kettenisstill with the corellium preloader
#asahi · 2021-02-12 10:37
kettenisarnd: not tried booting a Linux kernel yet; only OpenBSD
#asahi · 2021-02-11 23:56
kettenisjust pushed changes to make the serial port work in u-boot
#asahi · 2021-02-10 17:27
kettenisthanks for looking at it
#asahi · 2021-02-10 17:25
kettenisalready sent that bit upstream, but I'll roll a v3
#asahi · 2021-02-10 17:22
kettenisand it works as expected
#asahi · 2021-02-10 17:15
kettenisthere is an isb conveniently located close by
#asahi · 2021-02-10 17:13
kettenisthat should work for vbar_el1 as well isn't it?

plus ›