Re: [PATCH v1 3/3] NFSD: Implement NFSD_IO_DIRECT for NFS READ

[Chuck Lever <chuck.lever@oracle.com>]

On 9/9/25 7:39 PM, Chuck Lever wrote:
>> diff --git a/fs/nfsd/vfs.c b/fs/nfsd/vfs.c
>> index 441267d877f9..9665454743eb 100644
>> --- a/fs/nfsd/vfs.c
>> +++ b/fs/nfsd/vfs.c
>> @@ -1074,6 +1074,79 @@ __be32 nfsd_splice_read(struct svc_rqst *rqstp, struct svc_fh *fhp,
>>  	return nfsd_finish_read(rqstp, fhp, file, offset, count, eof, host_err);
>>  }
>>  
>> +/*
>> + * The byte range of the client's READ request is expanded on both
>> + * ends until it meets the underlying file system's direct I/O
>> + * alignment requirements. After the internal read is complete, the
>> + * byte range of the NFS READ payload is reduced to the byte range
>> + * that was originally requested.
>> + *
>> + * Note that a direct read can be done only when the xdr_buf
>> + * containing the NFS READ reply does not already have contents in
>> + * its .pages array. This is due to potentially restrictive
>> + * alignment requirements on the read buffer. When .page_len and
>> + * @base are zero, the .pages array is guaranteed to be page-
>> + * aligned.
> Where do we test that this condition is met?
> 
> I can see that nfsd_direct_read() is only called if "base" is zero, but
> that means the current contents of the .pages array are page-aligned,
> not that there are now.

The above comment might be stale; I'm not sure the .page_len test is
necessary. As long as the payload starts on a page boundary, it should
meet most any plausible buffer alignment restriction.

However, if there are additional restrictions needed, checks can be
added in nfsd_iter_read() just before nfsd_direct_read() is called.


-- 
Chuck Lever