12:08 jfalempe: javierm: can you please review https://patchwork.freedesktop.org/patch/649308/?series=147977 ?
13:25 mripard: mlankhorst: it looks like we haven't sent a drm-misc-next PR for a while?
13:43 mlankhorst: mripard: Yeah, will do so in a bit!
13:48 alyssa: sima: bbrezillon: what I'm trying to remember is why is the dma-fence memalloc a *cross driver* contract, given the docs *also* say:
13:49 alyssa: > Note that only GPU drivers have a reasonable excuse for both requiring mmu_interval_notifier and shrinker callbacks at the same time as having to track asynchronous compute work using dma_fence. No driver outside of drivers/gpu should ever call dma_fence_wait() in such contexts.
13:49 alyssa: does that imply the deadlocks can only happen if
13:50 alyssa: 1. the GPU driver has a shrinker callback that actually tries to block on fences (instead of just hypotehtically being allowed to), or
13:50 alyssa: 2. multiple GPU drivers are in-use simultaneously
13:50 alyssa: that is - the whole cross-driver contract here exists specifically to ensure hybrid GPU setups are correct
13:50 sima: 3. kernel maintainers would like to keep a tiny amount of sanity and not have to remember per-driver rules :-)
13:50 sima: but yeah, technically it's just for hybrid gpus
13:50 alyssa: yeah ok
13:51 sima: which is why out of tree android drivers have this all really easy
13:51 alyssa: which would explain why - despite all the mobile parts being "wrong" - this has never actually blown up in the wild
13:51 sima: just bend things until it works for you and done
13:51 sima: alyssa, it's also really hard to hit in practice
13:51 alyssa: well yes
13:52 sima: even on big desktop gpus when you run the precise nasty workloads
13:52 alyssa: but like. as long as the mobile gpu shrinkers don't wait on their own fences... there's no problem, these crappy android SoCs don't have working pcie controllers anyway :p
13:55 sima: alyssa, people in various discussions did suggest we try to tag fences so you can have different ones with different rules
13:55 sima: and know when not to wait on some and bail
13:55 alyssa: this sounds like a different kind of hell.
13:56 sima: one hang-up is that the mmu notifier path can't fail (unless that process has suffered death-by-oom already)
13:56 sima: alyssa, yeah consensus I think still is that it's not worth the pain
13:56 alyssa: what is the mmu notifier path exactly?
13:57 sima: alyssa, pte clearing when you userptr that range and must stay in sync
13:58 sima: the easy option out is userptr with pin_user_pages(FOLL_LONGTERM)
13:58 alyssa: i see
13:58 sima: except for the long term pin part
13:58 jannau: even as "mobile" SoCs as this https://www.apple.com/v/mac-pro/q/images/overview/expansion__b121gij0y42u_large.jpg don't have working pcie controllers (by x86 standards)
13:58 alyssa: 🤡
14:02 sima: my tongue-in-cheek reply on this entire mess is that memory management is trivial, right until you add reclaim
14:02 sima: then it becomes impossible
14:05 alyssa: real
16:00 sima: mripard, https://lore.kernel.org/dri-devel/4bac662a-228e-4739-b627-5d81df3d4842@amd.com/
17:20 jannau: alyssa: do you plan to push my drm/adp fixes? if so please push https://lore.kernel.org/dri-devel/20250304-drm_debug_dp_mst_topo_kconfig-v1-1-e16fd152f258@jannau.net/ as well
17:33 alyssa: jannau: I can try
17:34 alyssa: once again staring at https://drm.pages.freedesktop.org/maintainer-tools/committer/committer-drm-misc.html#where-do-i-apply-my-patch
17:34 alyssa: i guess these are bug fixes
17:34 alyssa: that affect the current rc
17:34 alyssa: so.. drm-misc-fixes?
17:34 alyssa: for the whole adp series and also that thing?
17:35 alyssa: and then the vsprintf stuff i guess goes into -misc-next?
17:35 alyssa: i regret taking that one on
17:36 alyssa: tangentially
17:36 alyssa: jannau: you should possibly have drm commit rights yourself
17:39 jannau: alyssa: agreed, when I start sending out patches for the apple KMS driver I should have commit rights
17:39 alyssa: ack
17:41 jannau: yes, all fixes for v6.15. adp was merged for 6.15. the Kconfig patch is not that pressing but I expect it will be picked for once it in Linus' tree
17:42 alyssa: i'd argue the kconfig one is still bug fix enough to go with the rest
17:43 alyssa: this is a post-hoc argument because i already picked it and started the smoke buildtest
17:46 jannau: yes, it's a fix
17:46 alyssa: cool
17:46 alyssa: will push once the build test finishes
17:48 jannau: thanks
18:28 alyssa: jannau: in theory it's pushed
18:28 alyssa: i dont see it on cgit tho
18:29 alyssa: oh there it is on gitlab tho
18:30 alyssa: https://gitlab.freedesktop.org/drm/misc/kernel/-/tree/drm-misc-fixes?ref_type=heads
18:31 jannau: isn't cgit deprecated? I think recent pull request were using gitlab
18:33 alyssa: noted
19:04 anholt: daniels: since I'm really not maintaining marge, could we get its owner email switched off of me to some sort of /dev/null?
19:15 daniels: anholt: oh sure, don’t
19:15 daniels: *done
19:15 anholt: it does generate a good bit of mail
19:36 daniels: not anymore it doesn’t
20:34 daniels: pour one out for kernel+rootfs https://gitlab.freedesktop.org/mesa/mesa/-/pipelines/1415257