On Tue, Oct 27, 2026 at 05:41:21AM -0700, Shyam Saini wrote: > Hi, > > Any comments/feedback on this patch? > > Would it be better to discard all PMEM regions if a size mismatch is > detected for any PMEM partition, rather than validating the data > offset as proposed in this patch?
Hi Shyam, Before considering changes to the validation or discarding PMEM regions, we need a better understanding of the problem. For v2, please expand the commit message to describe the failure and how to reproduce it. Include the platform, how PMEM is exposed, the namespace configuration, and what changes across reboots. The explanation about DRAM not being zeroed is also confusing. PMEM metadata is expected to persist across reboots, so why is the existing superblock considered stale? What makes it invalid? Also, please address the serious correctness concerns raised by Sashiko. With that info, we can begin to evaluate the proposed fix. Thanks, Alison > > Thanks, > Shyam
