On Wed, Dec 28, 2022 at 04:51:11PM +0100, Christoph Hellwig wrote:

On Wed, Dec 28, 2022 at 10:04:16AM +0800, kernel test robot wrote:

quoted

quoted

quoted

block/blk-sysfs.c:254:20: warning: comparison of distinct pointer types ('typeof (max_hw_sectors_kb) *' (aka 'unsigned long *') and 'typeof (2560U >> 1) *' (aka 'unsigned int *')) [-Wcompare-distinct-pointer-types]

Seems like the local max_sectors_kb variable also needs to be turned
into an unsigned int from the current unsigned long.

Yeah, and gcc didn't complain about it, so looks like I need to add
clang to my dev setup.
 

And btw, while I'm getting philosophical during my early morning
coffee here, I wonder if we need to actually get rid of or at least
bump BLK_DEF_MAX_SECTORS.  It's pretty ridiculous for modern devices,
and with CDL support that defintion would also include hard drives
eventually.

It looks like there's a bit of history around that value. I was hoping
we could just get rid of it entirely and let drivers choose their own
defaults.