01:21 <alyssa> piped through semi-fake vblanks, so it's less horribly broken now

01:22 <alyssa> weston is artefact free now, waow :)

01:24 <alyssa> ...closing weston causes a kernel oops, but, you know, technicalities..

02:36 <alyssa> [ 54.113366] apple-dcp 231c00000.dcp: RTKit: syslog message: FramebufferDCP.cpp:1824: virtual IOMFBStatus IOMobileFramebuffer::swap_submit_dcp(const IOMFBSwap *, IOMFB::Surface **, const uint32_t *, bool, double,

02:36 <alyssa> wonder why this is cutting off

02:41 <alyssa> (0, 0) + 125829120x70778880 --> (0, 0) + 1920x1080

02:41 <alyssa> ummm that would explain a bit..

02:42 <alyssa> oh, that's in 16.16 fixed point... ok

02:45 <alyssa> marcan: what are the surface IDs in the swap call?

02:45 <alyssa> are those meaningful to the DCP, just debug, just need to be internally consistent, ...?

02:47 <marcan> alyssa: I think they just need to be consistent

02:47 <marcan> there is no alloc call

02:48 <marcan> macos just throws some numbers at DCP as far as I can tell

02:48 <alyssa> Alright

02:48 <marcan> might even be ignored and not matter

02:48 <alyssa> so I can get away with just 0/1 for the two planes or something like that?

02:48 <marcan> feel free to try

02:48 <alyssa> 👍

02:48 <alyssa> trying to get rid of the magic numbers in the code and that's an easy one

02:48 <alyssa> the rt bandwidth structure otoh...

02:48 <marcan> I mean I *think* so but I'm not going to go and claim there isn't some deeper significance to using different IDs e.g. for different surface addresses :p

02:48 <marcan> I think macos basically uses them to identify an allocation

02:48 <alyssa> Ack

02:48 <marcan> but it could well not matter at all

02:49 <marcan> so feel free to go with static numbers initially

02:49 <marcan> if things break, that'd be a good thing to try changing

02:49 <alyssa> Yeah, I would totally believe Apple serializing things that are used by macOS but not the DCP just because it's in their IOSurface class..

02:49 <marcan> and yeah that RT stuff is something i need to look at

02:49 <marcan> yeah, there is definitely unused stuff in those calls

02:50 <marcan> alyssa: re syslog, the buffer size is fixed

02:50 <marcan> yes, it gets cut off, and yes, it sucks

02:51 <alyssa> Groan.

02:52 <alyssa> rt bandwidth is clearly a structure, not a single thing

02:52 <marcan> yes, of course

02:52 <marcan> and it probably has something to do with clocking

02:52 <alyssa> "lClock" eh?

02:52 <marcan> I wouldn't spend too much time on that right now though

02:52 <marcan> that's a whole can of worms we'll be better equipped to open once we understand system clocks better

02:52 <alyssa> Sure, okay

02:53 <marcan> plus I'd want to gather different samples from the different machines, see if it's always consistent or not

03:03 <alyssa> 👍