#
✕ nick:maz
#asahi · 2022-03-07 14:25
mazthere is probably going to be more, such as entering the kernel with the MMU on, which speeds things up.
#asahi · 2022-03-07 11:53
mazI think that's the equivalent of the UUID for a partition. only much smaller.
#asahi · 2022-02-23 18:00
mazsven: Robin is on Libera Chat if you need to get hold of him (I saw him in #linux-rockchip).
#asahi · 2022-02-23 17:53
mazonce*
#asahi · 2022-02-23 17:52
mazmarcan: I'll probably close the door on the irqchip stuff one -rc7 is out, so if you aiming for 5.18, the time is now...
#asahi · 2022-02-23 17:48
mazyup. I only use mainline on my mini, no extra patch whatsoever.
#asahi · 2022-02-14 18:01
mazj`ey: if that happens, I may have to relegate the D05 to be a pure test box. until then...
#asahi · 2022-02-14 17:41
mazstop it. I've been trying not to add an extra piece of HW to the already huge zoo...
#asahi · 2022-02-14 17:40
maz5.5h is about what I get on my X1 with all the corp crap running...
#asahi · 2022-02-10 16:49
mazkettenis: hey, I now take proper page faults there. totally fun! ;-)
#asahi · 2022-02-10 16:47
maz(crazy idea, I know)
#asahi · 2022-02-10 16:47
mazkettenis: I've started adding exception handling to u-boot for $WORK reasons. could use that to actually handle keyboard interrupts!
#asahi · 2022-02-10 16:10
maz:D
#asahi · 2022-02-10 16:09
mazyup. yet another Linux86 effect... ;-)
#asahi · 2022-02-10 16:06
mazaxboe: odd. Only x86 selects HAVE_KVM_DIRTY_RING.
#asahi · 2022-02-10 16:01
mazcheers.
#asahi · 2022-02-10 16:00
mazI am. haven't seen it yet, but that code has been very recently reworked.
#asahi · 2022-02-10 15:59
mazfair enough. please shout on the list (and cc me) if you manage to catch a trace involving KVM.
#asahi · 2022-02-10 15:58
mazmitigation is enabled, so this is something else.
#asahi · 2022-02-10 15:57
mazcheers.
#asahi · 2022-02-10 15:56
mazcurrent on. I assume you can't access the one that's hung...
#asahi · 2022-02-10 15:56
mazjust to check that we actively mitigate it.
#asahi · 2022-02-10 15:55
mazaxboe: The original M1 has a really bad bug which we mitigate by a certain amount of trapping. Could you dump me a dmesg of the booted host somewhere?
#asahi · 2022-02-10 15:53
mazaxboe: huh. using KVM?
#asahi · 2022-02-10 15:51
maz(and by QEMU I mean in conjunction with KVM)
#asahi · 2022-02-10 15:50
mazaxboe: what's your qemu issue? I recently fixed a bunch of issues with it and KVM, but that should only affect the original M1. Pro/Max are somewhat saner.
#asahi · 2022-02-09 12:15
mazwhat is going to be interesting is to make that DT evolve without introducing incompatibilities (like it just happened with PMGR).
#asahi · 2022-02-09 12:07
maz+1
#asahi · 2022-02-09 12:07
mazDT should come from the firmware, m1n1 is the firmware, job done.
#asahi · 2022-02-09 12:06
mazI would say the opposite. using DT from the kernel is a license to hack it in incompatible way
#asahi · 2022-02-09 09:26
mazif you want PCI, yes. hopefully we'll be able to relax this at some point.
#asahi · 2022-02-08 19:02
maz20220208185604.1097957-1-maz@kernel.org is what you are looking for, until the last patch trickles in...
#asahi · 2022-02-08 19:02
mazaxboe: ah, that's because I copy-pasted the last patch instead of the first (had a 1 in it, could have been good).
#asahi · 2022-02-08 19:00
mazaxboe: couple of minutes ago.
#asahi · 2022-02-08 18:58
mazaxboe: 20220208185604.1097957-11-maz@kernel.org <- PMU patches. plumbing this on pro/max is left as an exercise for the reader.
#asahi · 2022-02-08 18:46
mazTIGON3, actually
#asahi · 2022-02-08 18:46
mazTG3
#asahi · 2022-02-08 18:28
mazsven: yeah, 4k support would be great. I'm tired of having a special M1 build! ;-)
#asahi · 2022-02-08 18:27
mazanyway, I need to repost the PMU support so that you can all have fun counting cycles!
#asahi · 2022-02-08 18:26
mazI guess that Pro/Max is a wee bit faster than my standard M1.
#asahi · 2022-02-08 18:20
mazrunning fio in a VM is pretty interesting: IOPS=6925K with 16k pages, IOPS=4235K with 4k pages.
#asahi · 2022-02-08 18:10
mazIOPS=3626K on the small cores.
#asahi · 2022-02-08 18:09
mazIOPS=6019K, if that matters.
#asahi · 2022-02-08 18:09
mazyou're probably running with the P cores at their reset freq.
#asahi · 2022-02-08 18:08
mazon P cores.
#asahi · 2022-02-08 18:08
maz1.7 sec here for the getpid test
#asahi · 2022-02-08 08:23
mazthanks! :)
#asahi · 2022-02-08 08:21
mazI'm happy to drag the rtkit dependency in my tree, but not *everything* that's on that branch! that's pretty much everything that's currently in flight ;-)
#asahi · 2022-02-08 08:18
mazmarcan: is there a minimal SMC branch I can pick in order to start looking at this PCIe hotplug issue? the current smc/work has all sort of stuff piled in...
#asahi · 2022-01-22 11:01
mazas for the 32bit support, that's also why M1 is rather good. none of the old crap is supported (even when it was mandatory).

plus ›