Received-SPF: pass (google.com: domain of linux-nfs+bounces-3214-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
From: Benjamin Coddington <bcodding@redhat.com>
To: Olga Kornievskaia <olga.kornievskaia@gmail.com>
Cc: trond.myklebust@hammerspace.com, anna.schumaker@netapp.com,
 linux-nfs@vger.kernel.org
Subject: Re: [PATCH 1/1] pNFS/filelayout: check layout segment range
Date: Wed, 08 May 2024 14:23:53 -0400
Message-ID: <33B5E0C2-9B94-4C54-BD6F-475E4AC0EB75@redhat.com>
In-Reply-To: <CAN-5tyFWrqihMv8UwvAmrcxuFdM6S5qMvnZgWkJqWXm6iFJeVQ@mail.gmail.com>
References: <20240507195933.45683-1-olga.kornievskaia@gmail.com>
 <9D9DB9E5-E772-462B-BD38-7F703459A0FC@redhat.com>
 <CAN-5tyFWrqihMv8UwvAmrcxuFdM6S5qMvnZgWkJqWXm6iFJeVQ@mail.gmail.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

On 8 May 2024, at 13:52, Olga Kornievskaia wrote:

> On Wed, May 8, 2024 at 10:50 AM Benjamin Coddington <bcodding@redhat.com> wrote:
>>
>> On 7 May 2024, at 15:59, Olga Kornievskaia wrote:
>>
>>> From: Olga Kornievskaia <kolga@netapp.com>
>>>
>>> Before doing the IO, check that we have the layout covering the range of
>>> IO.
>>>
>>> Signed-off-by: Olga Kornievskaia <kolga@netapp.com>
>>> ---
>>>  fs/nfs/filelayout/filelayout.c | 2 ++
>>>  1 file changed, 2 insertions(+)
>>>
>>> diff --git a/fs/nfs/filelayout/filelayout.c b/fs/nfs/filelayout/filelayout.c
>>> index 85d2dc9bc212..bf3ba2e98f33 100644
>>> --- a/fs/nfs/filelayout/filelayout.c
>>> +++ b/fs/nfs/filelayout/filelayout.c
>>> @@ -868,6 +868,7 @@ filelayout_pg_init_read(struct nfs_pageio_descriptor *pgio,
>>>                       struct nfs_page *req)
>>>  {
>>>       pnfs_generic_pg_check_layout(pgio);
>>> +     pnfs_generic_pg_check_range(pgio, req);
>>>       if (!pgio->pg_lseg) {
>>>               pgio->pg_lseg = fl_pnfs_update_layout(pgio->pg_inode,
>>>                                                     nfs_req_openctx(req),
>>> @@ -892,6 +893,7 @@ filelayout_pg_init_write(struct nfs_pageio_descriptor *pgio,
>>>                        struct nfs_page *req)
>>>  {
>>>       pnfs_generic_pg_check_layout(pgio);
>>> +     pnfs_generic_pg_check_range(pgio, req);
>>>       if (!pgio->pg_lseg) {
>>>               pgio->pg_lseg = fl_pnfs_update_layout(pgio->pg_inode,
>>>                                                     nfs_req_openctx(req),
>>> --
>>> 2.39.1
>>
>> Looks right, less duplication to just call pnfs_generic_pg_check_range()
>> from pnfs_generic_pg_check_layout() now.
>
> filelayout.c is not the only caller of pnfs_generic_pg_check_layout().
> flexfilelayout has a wrapper around those 2 functions and calls that.
> however, the argument about duplicated code frustrates me because
> currently the code has 4lines. but if we were to re-write the same
> with a function, it would be more lines used in total (flexfiles has 8
> lines for it).

sorry - not trying to frustrate.  Since everyone is now calling both
pnfs_generic_pg_check_layout /and/ pnfs_generic_pg_check_range in the same
place, I thought you could just put
!pnfs_lseg_request_intersecting(pgio->pg_lseg, req) in the first test within
pnfs_generic_pg_check_layout(pgio, req), making that function do the work of
both.  Then you can delete pnfs_generic_pg_check_range entirely for
everyone.  The name still makes sense.

Ben