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: <25cfb176-ff2d-36e5-d12e-a2413ad2d5c2@linux.alibaba.com>
Date:   Thu, 24 Nov 2022 11:19:50 +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 v4 1/2] fscache,cachefiles: add prepare_ondemand_read()
 callback
Content-Language: en-US
To:     David Howells <dhowells@redhat.com>
Cc:     xiang@kernel.org, chao@kernel.org, jlayton@kernel.org,
        linux-erofs@lists.ozlabs.org, linux-cachefs@redhat.com,
        linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org
References: <20221117053017.21074-2-jefflexu@linux.alibaba.com>
 <20221117053017.21074-1-jefflexu@linux.alibaba.com>
 <1609247.1669221883@warthog.procyon.org.uk>
From:   Jingbo Xu <jefflexu@linux.alibaba.com>
In-Reply-To: <1609247.1669221883@warthog.procyon.org.uk>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

Hi David,

Really thanks for the comment.

On 11/24/22 12:44 AM, David Howells wrote:
> Jingbo Xu <jefflexu@linux.alibaba.com> wrote:
> 
>> -/*
>> - * Prepare a read operation, shortening it to a cached/uncached
>> - * boundary as appropriate.
>> - */
>> -static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *subreq,
>> -						      loff_t i_size)
>> +static inline enum netfs_io_source
>> +cachefiles_do_prepare_read(struct netfs_cache_resources *cres,
>> +			   loff_t start, size_t *_len, loff_t i_size,
>> +			   unsigned long *_flags)
> 
> That's not exactly what I meant, but I guess it would work as the compiler
> would probably inline it into both callers.

Yeah, I just have no better way if we don't want another function
calling introduced by this patch.  If we keep cachefiles_prepare_read()
untouched, the on-demand users need to construct a temporary subrequest
on the stack, and Jeff pointed out that this way is somewhat fragile and
not robust enough.

Anyway I would keep moving forward in the direction of current patch,
and add back the netfs_inode number to the tracepoint.  I would send v5
soon.

Thanks again for the reply :-)

> 
>> -		      __entry->netfs_inode, __entry->cache_inode)
>> +		      __entry->cache_inode)
> 
> Can you not lose the netfs_inode number from the tracepoint, please?  Feel
> free to display 0 there for your purposes.
> 

Sure.


-- 
Thanks,
Jingbo