01:06 robclark: HdkR: hmm, that trace replays fine on x13s (a690).. not sure what the user had.. would be interesting to know if the apitrace can reproduce the issue or if app is doing something different based on renderer string
01:19 HdkR: robclark: Looks like they were running on some FD660 device, likely on a phone with kgsl.
01:19 HdkR: So getting them to run apitrace on their phone would be funny.
01:21 HdkR: I might be able to dig my X13s out of my shelf over the next couple of days. Might take a bit to get it up to speed.
01:22 HdkR: At least then I could run the demo directly
01:22 HdkR: directly on an adreno*
01:22 robclark: x13s is what I have with me, so no need
01:24 robclark: a660 and a690 should be pretty similar.. there is some missing tess related programming on a7xx+ (both tu and fd), but afaict it doesn't seem to be load bearing
01:31 robclark: also.. how were they using gallium driver w/ kgsl? I guess either app is doing something different to expose a bug, depending on renderer string, or it is bug caused by some whatever collection of downstream patches the user had
01:31 robclark:suspects lolz
01:32 HdkR: Good question
01:35 HdkR: Main thing if it can't be reproduced is I want the evidence so they can stop bagging on Freedreno so hard.
01:44 robclark: Basically ball is in their court to provide a way to repro the issue, since we can't ;-)
01:45 HdkR: Well, I'll want to get it running directly on the laptop and say it works fine here :P
01:45 robclark: ofc non-zero possibility that they have some downstream patches that break things, etc.. I've no idea what they are running
01:45 robclark: wfm