01:02benjaminl: is there a place where CI status is tracked? I'm generally not sure whether to retry merges that fail due to timeout vs waiting for a fix
13:53tzimmermann: hi. how do we handle driver functionality that requires a minimum firmware version? any suggestions?
14:32mlankhorst: Optional or mandatory feature?
15:25tzimmermann: mlankhorst, optional i guess. it is addition resolutions https://lore.kernel.org/dri-devel/20241225003215.1343996-1-jammy_huang@aspeedtech.com/
15:25tzimmermann: or rather display modes
15:26pinchartl: tzimmermann: check the firmware version in the driver, and enable the additional features conditionally ?
15:28tzimmermann: pinchartl, ok sure
16:53Lynne: zzoon[m]: I've merged the ffmpeg changes to remove the query
17:08alyssa: hakzsam: re vk meta rect list, I've wanted to have vk_meta pass a regular triangle topology in that case since the special cases complicates the vk->hw enum translation
17:09alyssa: and when i looked a few months ago, the dedicated topology didn't actually do anything useful for any of the then consumers
17:09alyssa: I guess this is load bearing for radv though now?
17:09alyssa: s/complicates/slows down/, I guess
19:30dcbaker: kusma: 'Revert "panfrost: Disable CRC by default"' is auto nominated for 24.3 since it's a revert. It doesn't look like it should be applied so I've dropped it, just verifying that's the right thing to do
19:53kusma: Please don't backport that one, no. But sigh, I thought we ended up not automatically backporting reverts, that's a bad idea, because we don't know automatically why something got reverted...
19:53kusma: Reverts should have fixes tags if they should get backported...
20:36kisak: it's technically not automatically applied. It's queued for human review by a package maintainer as demonstrated a little bit ago.
21:55dcbaker: kusma: I (and I think Eric) want to get there, sort of like we'd like to do away with `CC` and use the `Backport` tag instead... It's a slow process
22:41llyyr: a certain anime game hits the following assert on builds with asserts enabled: ../src/compiler/nir/nir_opt_varyings.c:2487: deduplicate_outputs: Assertion `list_index == 0' failed.
22:42llyyr: the game has anticheat that makes it not work on linux, and requires patching to bypass it (and the author of the patch would prefer if I don't mention it in public places like mesa issue tracker), how would one go about reporting this?
22:43llyyr: this is with radv fwiw
22:43llyyr: (line number is off because I was trying older versions to see if it's a regression but doesn't appear to be)
22:45bnieuwenhuizen: honestly in general see if you can capture it with fossilize and then have the foz of the failing pipeline as repro without mentioning the app is probably the way I'd go for any app you don't want to discuss in an issue (though technically the appname is recorded in the foz, so if you want even that gone you might need to patch fossilize)
22:47llyyr: that probably works, it just needs to be hidden enough for search engines to not pick it up
22:54cmarcelo: fossilize question: is the instruction at the end of https://gitlab.freedesktop.org/mesa/shader-db/-/blob/master/README.md still valid? my $APPID/fozpipelinesv6 directory tree looks like https://pastebin.com/NkG0KaZq (after playing a couple of minutes)... trying the steamapp_*foz in the subdirectory complains about missing resources and eventually fails.
22:55cmarcelo: the error looks like: https://pastebin.com/wGhavBpP
22:55cmarcelo: it's not clear what the "common cache" that fossilize is refering to should be
22:56cmarcelo: (to be clear, I'm talking about using fossilize-replay with the pre-cached game fossil)