[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <1029982.1647872043@warthog.procyon.org.uk>
Date: Mon, 21 Mar 2022 14:14:03 +0000
From: David Howells <dhowells@...hat.com>
To: Matthew Wilcox <willy@...radead.org>
Cc: dhowells@...hat.com, Jeffle Xu <jefflexu@...ux.alibaba.com>,
linux-cachefs@...hat.com, xiang@...nel.org, chao@...nel.org,
linux-erofs@...ts.ozlabs.org, torvalds@...ux-foundation.org,
gregkh@...uxfoundation.org, linux-fsdevel@...r.kernel.org,
joseph.qi@...ux.alibaba.com, bo.liu@...ux.alibaba.com,
tao.peng@...ux.alibaba.com, gerry@...ux.alibaba.com,
eguan@...ux.alibaba.com, linux-kernel@...r.kernel.org,
luodaowen.backend@...edance.com
Subject: Re: [PATCH v5 03/22] cachefiles: introduce on-demand read mode
Matthew Wilcox <willy@...radead.org> wrote:
> Why do you have a separate rwlock when the xarray already has its own
> spinlock? This is usually a really bad idea.
Jeffle wants to hold a lock across the CACHEFILES_DEAD check and the xarray
access.
However, he tells xarray to do a GFP_KERNEL alloc whilst holding the rwlock:-/
David
Powered by blists - more mailing lists