<sven>
maybe point him to alyssa's version which is already cleaned up a bit
<dhewg>
mac via cmdline is prolly not acceptable anyway
<dhewg>
via module arg I mean
<sven>
yeah, alyssa’s versions takes it from the device tree
<j_ey>
why is that v3, I see no other versions
<sven>
Probably the first time they submitted something to the ML. Takes a while to get used to it and eventually accept that the only sane way is git send-mail
<j_ey>
oh.. 3 patches in the series, v3, maybe some confusion there
egavinc has joined #asahi-dev
<marcan>
replied
<marcan>
also threw some shade at stan :p
<j_ey>
poor stan
<sven>
:/
<j_ey>
marcan: 286 firmware files... is that for all the different macs?
egavinc has quit [Quit: Leaving]
<j_ey>
so much less gunk in alyssas series for brcfmac
<marcan>
(plus the next commit in t6000/bringup-work for the DART fix)
<sven>
oh... ouch. so it just took the paddr straight from the PTE when walking the table
<sven>
er. wait. the other way around. it put the paddr straight to the PTE when installing a table
<sven>
and then when walking the table later everything exploded
<maz>
marcan: looking good! if you want to split the fast IPI from the t6k bring up, that could trickle into upstream quickly (5.17).
<marcan>
yeah, I'll try to break it up into the compatible/fastipi handling first, and then add t6k / groups
<marcan>
sven: yeah
<marcan>
it's clearly a bug
<marcan>
how did that ever work with 52-bit addrs?
<sven>
yeah, probably makes sense to send that separately from the t6000 dart stuff
<marcan>
yeah
<sven>
i think "iommu/io-pgtable:" is the iommu style commit prefix and make sure to cc robin murphy if get_maintainers.pl doesn't include him automatically
<marcan>
maz: I was wondering about the division in the mask/unmask functions; alternatively I could make hwirq sparse and use bitfields, but at that point I really need to move to irq_domain_create_tree
<marcan>
however, with ~3000 hwirqs at this point even without holes (but with 2 groups), maybe that makes sense already?
<marcan>
sven: thanks, and yeah, the commits are BS right now :-)
<marcan>
whole thing needs to be redone and reworded
<marcan>
(also comments are wrong etc)
<sven>
yeah, just making sure you don't need a v2 just because the commit prefix is slightly off :-)
<j_ey>
sven: now you're the eager one :P
<sven>
j_ey: true :D
<maz>
marcan: +1 for a tree. it would even make sense today, where we only use a handful of interrupts, and yet allocate a fixed-size domain.
<marcan>
tree it is then, and then I'll just split the hwirq numbers into 16/16 bits and avoid the division
<j_ey>
marcan: how many interrupts did AICv1 have?
<marcan>
up to 1024 I think? 700-odd in practice IIRC
<j_ey>
ok, so quite a big jump
<marcan>
well, AICv2 has up to 4096 per block
<marcan>
but the current implementation has 1806 implemented, times 2 blocks
<marcan>
only one of which is in use right now
<marcan>
I strongly suspect the second block is where the second die's IRQs will show up on the rumored 2-die systems
<marcan>
though currently macOS ignores it completely (doesn't even set the masks)
<marcan>
it does work though, I can sw_gen IRQs in the second block and unmask them and they get delivered
<marcan>
so I implemented support for it, maybe I'll get lucky and the 2-die systems will need no AIC patches ;)
<sven>
:>
aleasto has joined #asahi-dev
<kettenis>
one can hope...
thunfisch is now known as Guest4804
thunfisch has joined #asahi-dev
Guest4804 has quit [Read error: Permission denied]
<kettenis>
marcan: any progress on the pmgr bindings?
<marcan>
kettenis: was kinda hoping for more review comments, since other than the spurious |s in the DT bindings I think everything else is kosher?
<marcan>
Krzysztof didn't comment on the v2 of the serial runtime-PM patch yet
riker77 has quit [Ping timeout: 480 seconds]
<kettenis>
the discussion about the compatibles confused me a bit, but do I understand correctly that you don't plan to remove the apple,pmgr one from the parent node?
<kettenis>
if the number of child nodes gets large, I may want to use that in OpenBSD to avoid cluttering dmesg
<kettenis>
attach a single driver to the parent node instead of separate drivers to the child nodes
<kettenis>
(we print a line for every driver that attaches)
<marcan>
yeah, apple,pmgr can stay
riker77 has joined #asahi-dev
aleasto has quit [Quit: Konversation terminated!]
aleasto has joined #asahi-dev
ihalip has quit [Remote host closed the connection]
kettenis has quit [Read error: Connection reset by peer]
kettenis has joined #asahi-dev
kov has joined #asahi-dev
aleasto has quit [Remote host closed the connection]
aleasto has joined #asahi-dev
aleasto has quit [Remote host closed the connection]
aleasto has joined #asahi-dev
yuyichao_ has quit [Ping timeout: 480 seconds]
aleasto has quit [Remote host closed the connection]
aleasto has joined #asahi-dev
aleasto has quit []
yuyichao_ has joined #asahi-dev
aleasto has joined #asahi-dev
aleasto has quit []
aleasto has joined #asahi-dev
aleasto has quit [Remote host closed the connection]
aleasto has joined #asahi-dev
aleasto has quit [Remote host closed the connection]
aleasto has joined #asahi-dev
aleasto has quit [Remote host closed the connection]