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: <c529ee21-699d-dfc8-5f7d-2597fa00796d@linux.alibaba.com>
Date:   Thu, 17 Nov 2022 12:24:40 +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 v3 1/2] fscache,cachefiles: add prepare_ondemand_read()
 callback
Content-Language: en-US
To:     David Howells <dhowells@redhat.com>,
        Jeff Layton <jlayton@kernel.org>
Cc:     xiang@kernel.org, chao@kernel.org, linux-erofs@lists.ozlabs.org,
        linux-cachefs@redhat.com, linux-kernel@vger.kernel.org,
        linux-fsdevel@vger.kernel.org
References: <2b595b62f6ecd28298a860fcdc5b4941dcafd9eb.camel@kernel.org>
 <20221116104502.107431-1-jefflexu@linux.alibaba.com>
 <20221116104502.107431-2-jefflexu@linux.alibaba.com>
 <2968419.1668606101@warthog.procyon.org.uk>
From:   Jingbo Xu <jefflexu@linux.alibaba.com>
In-Reply-To: <2968419.1668606101@warthog.procyon.org.uk>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 11/16/22 9:41 PM, David Howells wrote:
> Jeff Layton <jlayton@kernel.org> wrote:
> 
>>> +static enum netfs_io_source cachefiles_do_prepare_read(struct netfs_cache_resources *cres,
>>> +					loff_t *_start, size_t *_len,
>>> +					unsigned long *_flags, loff_t i_size)
>>
>> _start is never changed, so it should be passed by value instead of by
>> pointer.
> 
> Hmmm.  The intention was that the start pointer should be able to be moved
> backwards by the cache - but that's not necessary in ->prepare_read() and
> ->expand_readahead() is provided for that now.  So yes, the start pointer
> shouldn't get changed at this point.

Okay.


> 
>> I'd also reverse the position of the arguments for _flags and i_size.
>> Otherwise, the CPU/compiler have to shuffle things around more in
>> cachefiles_prepare_ondemand_read before they call this.
> 
> Better to pass the flags in and then ignore them.  That way it can tail call,
> or just call cachefiles_do_prepare_read() directly from erofs.  If you're
> going to have a wrapper, then you might be just as well create a
> netfs_io_subrequest struct on the stack.

I would prefer letting cachefiles_prepare_ondemand_read() pass flags in
and then tail call cachefiles_do_prepare_read() directly.

Many thanks for the suggestion.


-- 
Thanks,
Jingbo