[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZXF4geWqVrwMe8Xr@infradead.org>
Date: Wed, 6 Dec 2023 23:47:13 -0800
From: Christoph Hellwig <hch@...radead.org>
To: Sergei Shtepa <sergei.shtepa@...ux.dev>
Cc: axboe@...nel.dk, hch@...radead.org, corbet@....net,
snitzer@...nel.org, mingo@...hat.com, peterz@...radead.org,
juri.lelli@...hat.com, viro@...iv.linux.org.uk, brauner@...nel.org,
linux-block@...r.kernel.org, linux-doc@...r.kernel.org,
linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
Sergei Shtepa <sergei.shtepa@...am.com>
Subject: Re: [PATCH v6 10/11] blksnap: Kconfig and Makefile
> +config BLKSNAP_DIFF_BLKDEV
> + bool "Use an optimized algorithm to store difference on a block device"
> + depends on BLKSNAP
> + default y
> + help
> + The difference storage for a snapshot can be a regular file or a
> + block device. We can work with a block device through the interface
> + of a regular file. However, direct management of I/O units should
> + allow for higher performance.
Is there much of a point in making this option?
Btw, Linus hates defaul y, so we should not have one in there.
> +config BLKSNAP_CHUNK_DIFF_BIO_SYNC
> + bool "Use a synchronous I/O unit processing algorithm for the snapshot image"
> + depends on BLKSNAP
> + default n
n the default default, so no need to add it. But unless there is a
really good reason to have two different bio submission mechanisms
upstream I'd strongly suggest dropping one of them and this option.
Powered by blists - more mailing lists