On 06/07/2021 16:22, Jan Beulich wrote: > On 06.07.2021 17:13, Jan Beulich wrote: >> On 06.07.2021 16:11, Olaf Hering wrote: >>> Am Tue, 6 Jul 2021 14:58:04 +0200 >>> schrieb Olaf Hering <o...@aepfle.de>: >>> >>>> the reporting is broken since a while >>> A quick check on a Dell T320 with E5-2430L, which does not have "Page >>> Modification Logging", indicates that staging-4.5 appears to work, but >>> reporting in staging-4.6 is broken. >> Not surprising at all, considering PML support was added in 4.6. > Or perhaps still surprising, as the functions involved there don't > appear to be bypassing any core logic, so after some looking around > I can't say I see anything that's obviously wrong. > > Oh, and I notice only now the "not" in your original reply, which > renders my earlier statement completely pointless anyway. I'm sorry.
Ok, so my observations of dirty_count being reliable during development of migration v2 appears to be correct, and something regressed in the dev window where it got committed upstream... That is suspicious. ~Andrew