On Mon, Aug 3, 2020 at 5:11 PM Stafford Horne <shorne@xxxxxxxxx> wrote: > > On big-endian architectures like OpenRISC, sparse outputs below warnings on > asm-generic/io.h. This is due to io statements like: > > __raw_writel(cpu_to_le32(value), PCI_IOBASE + addr); > > The __raw_writel() function expects native endianness, however > cpu_to_le32() returns __le32. On little-endian machines these match up > and there is no issue. However, on big-endian we get warnings, for IO > that is defined as little-endian the mismatch is expected. > > The fix I propose is to __force to native endian. > > Warnings: > > ./include/asm-generic/io.h:166:15: warning: cast to restricted __le16 > ./include/asm-generic/io.h:166:15: warning: cast to restricted __le16 > ./include/asm-generic/io.h:166:15: warning: cast to restricted __le16 > ./include/asm-generic/io.h:166:15: warning: cast to restricted __le16 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:179:15: warning: cast to restricted __le32 > ./include/asm-generic/io.h:215:22: warning: incorrect type in argument 1 (different base types) > ./include/asm-generic/io.h:215:22: expected unsigned short [usertype] value > ./include/asm-generic/io.h:215:22: got restricted __le16 [usertype] > ./include/asm-generic/io.h:225:22: warning: incorrect type in argument 1 (different base types) > ./include/asm-generic/io.h:225:22: expected unsigned int [usertype] value > ./include/asm-generic/io.h:225:22: got restricted __le32 [usertype] > > Signed-off-by: Stafford Horne <shorne@xxxxxxxxx> Looks good to me. Acked-by: Arnd Bergmann <arnd@xxxxxxxx> Can you just merge that through your openrisc tree? I don't have any other asm-generic changes for this merge window.
