linux-kernel - Re: [PATCH 1/2] fscache,cachefiles: add prepare_ondemand

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Fri, 4 Nov 2022 20:22:48 +0800
From:   JeffleXu <jefflexu@...ux.alibaba.com>
To:     Jeff Layton <jlayton@...nel.org>, dhowells@...hat.com,
        xiang@...nel.org, chao@...nel.org, linux-cachefs@...hat.com,
        linux-erofs@...ts.ozlabs.org
Cc:     linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org
Subject: Re: [PATCH 1/2] fscache,cachefiles: add prepare_ondemand_read()
 callback



On 11/4/22 7:18 PM, Jeff Layton wrote:
> On Fri, 2022-11-04 at 15:26 +0800, Jingbo Xu wrote:
>> Add prepare_ondemand_read() callback dedicated for the on-demand read
>> scenario, so that callers from this scenario can be decoupled from
>> netfs_io_subrequest.
>>
>> To reuse the hole detecting logic as mush as possible, both the
>> implementation of prepare_read() and prepare_ondemand_read() inside
>> Cachefiles call a common routine.
>>
>> In the near future, prepare_read() will get enhanced and more
>> information will be needed and then returned to callers. Thus
>> netfs_io_subrequest is a reasonable candidate for holding places for all
>> these information needed in the internal implementation.
>>
>> Signed-off-by: Jingbo Xu <jefflexu@...ux.alibaba.com>
>> ---
>>  fs/cachefiles/io.c                | 42 +++++++++++++++++++++++++------
>>  include/linux/netfs.h             |  7 ++++++
>>  include/trace/events/cachefiles.h |  4 +--
>>  3 files changed, 43 insertions(+), 10 deletions(-)
>>
>> diff --git a/fs/cachefiles/io.c b/fs/cachefiles/io.c
>> index 000a28f46e59..6427259fcba9 100644
>> --- a/fs/cachefiles/io.c
>> +++ b/fs/cachefiles/io.c
>> @@ -385,16 +385,11 @@ static int cachefiles_write(struct netfs_cache_resources *cres,
>>  				  term_func, term_func_priv);
>>  }
>>  
>> -/*
>> - * 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 enum netfs_io_source cachefiles_do_prepare_read(struct netfs_io_subrequest *subreq,
>> +						       struct netfs_cache_resources *cres,
>> +						       loff_t i_size)
>>  {
>>  	enum cachefiles_prepare_read_trace why;
>> -	struct netfs_io_request *rreq = subreq->rreq;
>> -	struct netfs_cache_resources *cres = &rreq->cache_resources;
>>  	struct cachefiles_object *object;
>>  	struct cachefiles_cache *cache;
>>  	struct fscache_cookie *cookie = fscache_cres_cookie(cres);
>> @@ -501,6 +496,36 @@ static enum netfs_io_source cachefiles_prepare_read(struct netfs_io_subrequest *
>>  	return ret;
>>  }
>>  
>> +/*
>> + * 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)
>> +{
>> +	return cachefiles_do_prepare_read(subreq,
>> +			&subreq->rreq->cache_resources, i_size);
>> +}
>> +
>> +/*
>> + * Prepare an on-demand read operation, shortening it to a cached/uncached
>> + * boundary as appropriate.
>> + */
>> +static enum netfs_io_source cachefiles_prepare_ondemand_read(struct netfs_cache_resources *cres,
>> +		loff_t start, size_t *_len, loff_t i_size)
>> +{
>> +	enum netfs_io_source source;
>> +	struct netfs_io_subrequest subreq = {
>> +		.start	= start,
>> +		.len	= *_len,
>> +		.flags	= 1 << NETFS_SREQ_ONDEMAND,
>> +	};
>> +
> 
> Faking up a struct like this is sort of fragile. What if we change
> cachefiles_do_prepare_read to consult other fields in this structure
> later?

Indeed it's not robust somehow.


> 
> It might be best to have cachefiles_do_prepare_read take individual
> start, len, and flags values instead of doing this.
> 

I will give it a try if nobody objects this.


Thank you for your suggestions :)


-- 
Thanks,
Jingbo