On Wednesday 26 October 2022 22:22:28 SCHNEIDER Johannes wrote: > a colleague pointed out the use of barriers in the quoted kernel code snippet > -> are we perhaps running into code-reordering issues; that checking the > status flags and writing into the queue gets out of sync?
I was always in impression that readl() and writel() calls in U-Boot cannot be reordered and so no explicit barrier is needed. Could somebody check & confirm this?