Re: [PATCH v3] loop: drop caches if offset or block_size are changed

[Bart Van Assche]

On Tue, 2018-12-18 at 14:41 -0800, Jaegeuk Kim wrote:
+AD4 +AFs ... +AF0

Please post new versions of a patch as a new e-mail thread instead of
as a reply to a previous e-mail.

+AD4 +AFs ... +AF0
+AD4
+AD4  	if (lo-+AD4-lo+AF8-offset +ACEAPQ info-+AD4-lo+AF8-offset +AHwAfA
+AD4  	    lo-+AD4-lo+AF8-sizelimit +ACEAPQ info-+AD4-lo+AF8-sizelimit) +AHs
+AD4 +-		/+ACo kill+AF8-bdev should have truncated all the pages +ACo-/
+AD4 +-		if (lo-+AD4-lo+AF8-device-+AD4-bd+AF8-inode-+AD4-i+AF8-mapping-+AD4-nrpages) +AHs
+AD4 +-			err +AD0 -EAGAIN+ADs
+AD4 +-			goto exit+ADs
+AD4 +-		+AH0

Please add a pr+AF8-info() or pr+AF8-warn() statement here such that it becomes
easy for the user to figure out why EAGAIN has been returned.

+AD4  	blk+AF8-mq+AF8-freeze+AF8-queue(lo-+AD4-lo+AF8-queue)+ADs
+AD4  
+AD4 +-	/+ACo kill+AF8-bdev should have truncated all the pages +ACo-/
+AD4 +-	if (lo-+AD4-lo+AF8-queue-+AD4-limits.logical+AF8-block+AF8-size +ACEAPQ arg +ACYAJg
+AD4 +-			lo-+AD4-lo+AF8-device-+AD4-bd+AF8-inode-+AD4-i+AF8-mapping-+AD4-nrpages) +AHs
+AD4 +-		err +AD0 -EAGAIN+ADs
+AD4 +-		goto out+ADs
+AD4 +-	+AH0

Same comment here. Additionally, please consider renaming the +ACI-out+ACI label
into +ACI-unfreeze+ACI or so. I think that will make the use of label names more
consistent with the rest of the block layer. Once these two comments are
addressed, feel free to add:

Reviewed-by: Bart Van Assche +ADw-bvanassche+AEA-acm.org+AD4