SUNRPC: Revert 5f7fc5d69f6e92ec0b38774c387f5cf7812c5806
authorChuck Lever <chuck.lever@oracle.com>
Mon, 18 Dec 2023 22:05:40 +0000 (17:05 -0500)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 1 Jan 2024 12:42:26 +0000 (12:42 +0000)
[ Upstream commit bd018b98ba84ca0c80abac1ef23ce726a809e58c ]

Guillaume says:
> I believe commit 5f7fc5d69f6e ("SUNRPC: Resupply rq_pages from
> node-local memory") in Linux 6.5+ is incorrect. It passes
> unconditionally rq_pool->sp_id as the NUMA node.
>
> While the comment in the svc_pool declaration in sunrpc/svc.h says
> that sp_id is also the NUMA node id, it might not be the case if
> the svc is created using svc_create_pooled(). svc_created_pooled()
> can use the per-cpu pool mode therefore in this case sp_id would
> be the cpu id.

Fix this by reverting now. At a later point this minor optimization,
and the deceptive labeling of the sp_id field, can be revisited.

Reported-by: Guillaume Morin <guillaume@morinfr.org>
Closes: https://lore.kernel.org/linux-nfs/ZYC9rsno8qYggVt9@bender.morinfr.org/T/#u
Fixes: 5f7fc5d69f6e ("SUNRPC: Resupply rq_pages from node-local memory")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
net/sunrpc/svc_xprt.c

index 4cfe9640df4814cff2a68c28b565cc967966f88b..5cfe5c7408b74443deb5d7216b379d41dd5015e5 100644 (file)
@@ -666,9 +666,8 @@ static bool svc_alloc_arg(struct svc_rqst *rqstp)
        }
 
        for (filled = 0; filled < pages; filled = ret) {
-               ret = alloc_pages_bulk_array_node(GFP_KERNEL,
-                                                 rqstp->rq_pool->sp_id,
-                                                 pages, rqstp->rq_pages);
+               ret = alloc_pages_bulk_array(GFP_KERNEL, pages,
+                                            rqstp->rq_pages);
                if (ret > filled)
                        /* Made progress, don't sleep yet */
                        continue;