04:54 alyssa: zmike: maybe break out the two agents.md changes to separate MRs
04:55 alyssa: given that both are controversial and they don't have any real relationship so it's possible one lands and the other doesn't (not sure which would which)
06:55 mlankhorst: airlied: would it still be useful to send a -fixes? Didn't see your mail yesterday
11:54 zmike: alyssa: more likely that I will just drop the first one anyway since our policy is actually "do not name the model"
20:29 uis: Are amdgpu/amdkfd people here? Is it normal for ioctl with KFD_IOC_ALLOC_MEM_FLAGS_MMIO_REMAP to fail with ENOMEM on polaris GPUs?
20:29 uis: It seems to be the culprit of failure of rocminfo.