Quoting Eugen Hristev (2024-10-30 01:28:14) > On 10/30/24 02:40, Stephen Boyd wrote: > > > > If the rpmh-rsc code didn't use writel() or readl() I'd believe that the > > data member is simply a u32 container. But those writel() and readl() > > functions are doing a byte swap, which seems to imply that the data > > member is a native CPU endian u32 that needs to be converted to > > little-endian. Sounds like BCM_TCS_CMD() should just pack things into a > > u32 and we can simply remove the cpu_to_l32() stuff in the macro? > > This review [1] from Evan Green on the original patch submission > requested the use of cpu_to_le32 > > So that's how it ended up there. > Thanks. I still don't see why this can't just be treated as a u32 and then we have writel() take care of it for us.
