#asahi-dev · 2023-12-26 18:58
jannegi.e. we can't have it as reg since it must not be mapped to the second's dies MMIO range via ranges#asahi-dev · 2023-12-26 18:57
jannegthe disp-4 register got moved to "apple,bw-scratch" because it is apple's wisdom always pmgr on die0#asahi-dev · 2023-12-26 18:55
jannegkettenis: or most probably incompatible driver and dtb#asahi-dev · 2023-12-26 18:47
jannegkettenis: I can't remember having seen that ever. which device and OS firmware?#asahi-dev · 2023-12-26 12:43
jannegt8112 make the inconsistency even more obvious, see vm-base for dart-disp0 (0) and dart-dcp (0x800000000)#asahi-dev · 2023-12-26 12:41
jannegthat is already present in the adt. on t8103 it's based on asc-dram-mask in the ADT which only applies to dcp#asahi-dev · 2023-12-26 11:25
jannegkettenis: what do you mean? iova's out of the annotated dma-range? dma-range came later and nobody updated m1n1#asahi-dev · 2023-12-22 07:12
jannegmariogrip: yes, see https://github.com/AsahiLinux/macvdmtool#note-about-macos-12 and adapt#asahi-dev · 2023-12-20 21:33
jannegthe interface is stable-ish but there is enough churn from release to release to be annoying#asahi-dev · 2023-12-20 21:30
jannegI want to rewrite it in Rust but I'm not sure if I ever find time for it#asahi-dev · 2023-12-20 21:28
jannegLyude: it's mostly annoying to do in C. It uses #defines with multiple inclusion of template fiels#asahi-dev · 2023-12-20 21:25
jannegcyrinux: no, just the kernel#asahi-dev · 2023-12-20 20:46
jannegthere is a problem with the local version of the edge kernel#asahi-dev · 2023-12-20 20:45
jannegcyrinux: use the non-edge kernel. they are now fnuctionally the same#asahi-dev · 2023-12-20 15:25
jannegah, it's already in soc/for-next#asahi-dev · 2023-12-20 15:10
jannegmarcan: asahi-soc-mailbox-6.8 is missing from asahi-soc/for-next#asahi-dev · 2023-12-20 13:30
jannegkettenis: I don't think so. my motivation was to untangle modeset and dcp_flush()#asahi-dev · 2023-12-20 13:26
jannegrefuses to start in what sense? rtkit_wakeup fails? dcp_start()?#asahi-dev · 2023-12-20 13:25
jannegdrm_fb_helper_blank should result .atomic_enable / .atomic_disable calls#asahi-dev · 2023-12-20 13:16
jannegthat already has the delay to to see if a swap has happened#asahi-dev · 2023-12-20 13:15
jannegwe certyainly can't issue a flush directly from iomfbep_cb_enable_backlight_message_ap_gated(). the easiest option would be to call dcp_set_brightness() from a a workqueue#asahi-dev · 2023-12-20 13:14
jannegkettenis: sure, the modesetting/tearfree comment was also about improving the experience on xorg in general#asahi-dev · 2023-12-20 11:55
jannegkettenis: for xorg/modesetting you could try if the tearfree option helps, that should do swaps but iirc I've never tested it#asahi-dev · 2023-12-20 11:52
jannegleio: no#asahi-dev · 2023-12-20 11:49
jannegthis is one reason why we don't want to support Xorg on linux#asahi-dev · 2023-12-20 11:48
jannegkettenis: known issue. lina and I tried to workaround that but I'm not sure if that ever was relieable. not related to the gpu driver but just how the application uses the kms device#asahi-dev · 2023-12-20 09:56
jannegsven: could the 14 vs. 0x14 error in PIPEHANDLER_OVERRIDE_VALUES explain the broken suspend or other issue we've seen with usb3?#asahi-dev · 2023-12-19 17:55
jannegmissing: EAS (energy aware scheduler) support by chadmed#asahi-dev · 2023-12-19 17:53
jannegeiln: trackpad fix is by marcan#asahi-dev · 2023-12-19 17:28
jannegeiln: upcoming projects: make hdmi out on m2 desktops and 14/16-inch macbook pros robust, hdmi audio#asahi-dev · 2023-12-19 17:21
jannegI did some pointless changes to make diffing of macos and linux traces easier#asahi-dev · 2023-12-19 17:19
jannegwe need to find a way to bring thta together with the type-c mux/switch#asahi-dev · 2023-12-19 17:18
jannegsven: the dp-phy callbacks can now configure things and are called from dptx#asahi-dev · 2023-12-19 17:14
jannegI probably regressed dp-altmode slightly with hdmi out via atc3-phy#asahi-dev · 2023-12-19 17:06
jannegeiln: correct limit sensor / size for the trackpad on devices != j314#asahi-dev · 2023-12-19 08:12
jannegkettenis: \o/ why is smc needed? due to the current dts? early september dcp shouldn't have dptx / dp2hdmi iirc#asahi-dev · 2023-12-17 14:24
jannegofficially supported for asahilinux#asahi-dev · 2023-12-17 14:09
jannegleio: no idea. officially up to 4k@60 is supported which doesn't need DSC#asahi-dev · 2023-12-17 12:04
janneg6k@60 requires DSC even for 8-bit#asahi-dev · 2023-12-17 12:03
jannegcolor mode "selection" is not implemented. you can list timing and color modes with `trace-cmd stream -e dcp:iomfb_color_mode -e dcp:iomfb_timing_mode`#asahi-dev · 2023-12-17 11:55
jannegleio: none of the kernel source patches are necessay anymore (never were)#asahi-dev · 2023-12-17 11:34
janneg-a#asahi-dev · 2023-12-17 11:34
jannegfeel free to teast if backing out the 4k limit works. DSC seems to work transparently#asahi-dev · 2023-12-17 11:32
jannegno need for dcpext5. making dcp work as built-in is not a priority but I'll keep thinking (and testing) why that breaks#asahi-dev · 2023-12-17 11:04
jannegconfirmed. dcp works with changing just DRM_APPLE to m (and removing the asmedia firmware and enabling initramfs compression)#asahi-dev · 2023-12-17 10:54
jannegleio: maybe fixed by building drm_Apple as module#asahi-dev · 2023-12-17 10:14
jannegindeed, sorry. https://paste.centos.org/view/4c28bcf7#asahi-dev · 2023-12-17 10:08
jannegleio: https://dpaste.com/7EJU4CST3.txt works for me, trying to find the difference. it is not CONFIG_INIT_STACK_* which was my first suspect#asahi-dev · 2023-12-17 09:37
jannegthe interrupt error after modeset#asahi-dev · 2023-12-17 09:35
jannegleio: issue reproduced with your config