Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <6064150a-7517-c0e1-72bb-e1a8adcfae74@linux.alibaba.com>
Date:   Fri, 4 Nov 2022 20:20:09 +0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0)
 Gecko/20100101 Thunderbird/102.4.0
Subject: Re: [PATCH 2/2] erofs: switch to prepare_ondemand_read() in fscache
 mode
Content-Language: en-US
To:     Jeff Layton <jlayton@kernel.org>, dhowells@redhat.com,
        xiang@kernel.org, chao@kernel.org, linux-cachefs@redhat.com,
        linux-erofs@lists.ozlabs.org
Cc:     linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org
References: <20221104072637.72375-1-jefflexu@linux.alibaba.com>
 <20221104072637.72375-3-jefflexu@linux.alibaba.com>
 <2e2eceeb11972462bb9161a73c00a9c77f8af8d2.camel@kernel.org>
From:   JeffleXu <jefflexu@linux.alibaba.com>
In-Reply-To: <2e2eceeb11972462bb9161a73c00a9c77f8af8d2.camel@kernel.org>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk


On 11/4/22 7:46 PM, Jeff Layton wrote:
> On Fri, 2022-11-04 at 15:26 +0800, Jingbo Xu wrote:
>> Switch to prepare_ondemand_read() interface and a self-contained request
>> completion to get rid of netfs_io_[request|subrequest].
>>
>> The whole request will still be split into slices (subrequest) according
>> to the cache state of the backing file.  As long as one of the
>> subrequests fails, the whole request will be marked as failed. Besides
>> it will not retry for short read.  Similarly the whole request will fail
>> if that really happens. 
>>
> 
> That's sort of nasty. The kernel can generally give you a short read for
> all sorts of reasons, some of which may have nothing to do with the
> underlying file or filesystem.
> 
> Passing an error back to an application on a short read is probably not
> what you want to do here. The usual thing to do is just to return what
> you can, and let the application redrive the request if it wants.
> 

Yeah, thanks for your comment. We can fix this either in current
patchset or a separate series. As we just discussed on IRC, we will fix
in the following series.


-- 
Thanks,
Jingbo