On Wed, Feb 26, 2025 at 09:38:36AM -0800, Bart Van Assche wrote: > On 2/24/25 6:21 PM, Ming Lei wrote: > > diff --git a/block/blk.h b/block/blk.h > > index 90fa5f28ccab..9cf9a0099416 100644 > > --- a/block/blk.h > > +++ b/block/blk.h > > @@ -14,6 +14,7 @@ > > struct elevator_type; > > #define BLK_DEV_MAX_SECTORS (LLONG_MAX >> 9) > > +#define BLK_MIN_SEGMENT_SIZE 4096 > > /* Max future timer expiry for timeouts */ > > #define BLK_MAX_TIMEOUT (5 * HZ) > > Hi Ming, > > Would you agree with reducing BLK_MIN_SEGMENT_SIZE further, e.g. to 2048 > or 1024? Although I'm not aware of any storage devices that need this We never take < 4096 as min segment size, I'd rather not relax it for avoiding any potential regression. But if any kind of real device requires it, we may re-consider to relax it more. > change, this change would make it possible to test the new code paths > introduced by this patch on systems with a 4 KiB page size. I wrote > blktests tests for the new code paths before I posted my patch series > "Support limits below the page size" > (https://lore.kernel.org/linux-block/20230612203314.17820-1-bvanassche@xxxxxxx/). > The last two patches of that patch series are still needed to run these > blktests tests. Indeed, I guess you will re-send the two, and I am glad to review after it is posed out. thanks, Ming
