Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 26 Oct 2021 23:55:43 -0700
From:   Christoph Hellwig <hch@infradead.org>
To:     Mark-PK Tsai <mark-pk.tsai@mediatek.com>
Cc:     agk@redhat.com, snitzer@redhat.com, dm-devel@redhat.com,
        matthias.bgg@gmail.com, linux-kernel@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org,
        linux-mediatek@lists.infradead.org, tequila.yuang@mediatek.com,
        yj.chiang@mediatek.com
Subject: Re: [PATCH] dm: add module parameter bdev_wait
Message-ID: <YXj37wXtN0wQqwZx@infradead.org>
References: <20211022064748.30136-1-mark-pk.tsai@mediatek.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20211022064748.30136-1-mark-pk.tsai@mediatek.com>
Precedence: bulk

On Fri, Oct 22, 2021 at 02:47:47PM +0800, Mark-PK Tsai wrote:
> dm_early_create() fail if the target block device not found in
> the late init stage.
> The block device is created in mmc_rescan() which kernel do
> it async by queue it into system_freezable_wq.
> Add module param bdev_wait to support waiting the block device
> ready before dm_early_create() like prepare_namespace() does.

This adds a busy loop.  There is a reason why we usually set up
stacking drivers from userspace, in that we have a working event
system there.  So the answer is not to add further hacks to the
in-kernel DM setup hack, but to stop using it.