On Tue, May 09, 2017 at 08:48:21PM +0530, Abdul Haleem wrote:

A bisection for the above suspects resulted a bad commit;

c20cfc27a47307e811346f85959cf3cc07ae42f9 is the first bad commit
commit c20cfc27a47307e811346f85959cf3cc07ae42f9
Author: Christoph Hellwig [off-list ref]
Date:   Wed Apr 5 19:21:07 2017 +0200

    block: stop using blkdev_issue_write_same for zeroing

And this effectively switches us to use the write_zeroes for SCSI.

    
    We'll always use the WRITE ZEROES code for zeroing now.
    
    Signed-off-by: Christoph Hellwig [off-list ref]
    Reviewed-by: Martin K. Petersen [off-list ref]
    Reviewed-by: Hannes Reinecke [off-list ref]
    Signed-off-by: Jens Axboe [off-list ref]


@Christoph FYI, the machine configured with 64K page size

quoted

WARNING: CPU: 12 PID: 0 at block/blk-core.c:2651 .blk_update_request+0x4cc/0x4e0

Can you decode which warning this is?  Is it:

	WARN_ON_ONCE(req->rq_flags & RQF_SPECIAL_PAYLOAD);

?  In which case your setup did a partial completion of a WRITE SAME
command, which is perfectly legal according to SCSI, but a bit unusual.