OK, the subject is technically true, but to most people "fix compile warning" means yet another pointless patch. This patch isn't pointless, though, is it? The driver is actually broken on big endian systems without it, so the subject should probably say this because it's important for distros triaging whether to backport this fix if they backport the nvme components. On Fri, 2017-06-30 at 16:56 -0700, Himanshu Madhani wrote: > drivers/scsi/qla2xxx/qla_nvme.c: In function 'qla2x00_start_nvme_mq': > include/uapi/linux/byteorder/big_endian.h:32:26: warning: large > integer implicitly truncated to unsigned type [-Woverflow] > #define __cpu_to_le32(x) ((__force __le32)__swab32((x))) This text doesn't explain why you've fixed the bug this way. It has to do with the 32 byte reads the card does for the mailbox, so you should explain what the actual problem the compile warning picked up is and why you fixed it by doing an entire 32 bit write. Preferably ASAP because the merge window will open on Monday. James
