Merge: SUNRPC: Cleanup/fix initial rq_pages allocation

MR: https://gitlab.com/redhat/centos-stream/src/kernel/centos-stream-9/-/merge_requests/7068

JIRA: https://issues.redhat.com/browse/RHEL-91267
Conflicts:  This RHEL-9 version is quite a bit different from the upstream
    version because the original fix for this problem conflicted with a
    series to "Allocate payload arrays dynamically":
    https://lore.kernel.org/linux-nfs/[email protected]/
    Thus, the upstream fix referenced here keeps svc_init_buffer to allow
    the page array allocation.  We're not ready for this new work to land in
    RHEL-9 and if we do take it we'll need to take this fix again afterward.
    This version corresponds to the v1 fix posted here:
    https://lore.kernel.org/linux-nfs/458f45b2b7259c17555dd65aa7cdbbf1a459d5e6.1749131924.git.bcodding@redhat.com/
    The result is the same - we don't do the bulk array allocation in
    svc_init_buffer() because we lack the required retry logic to allocate
    the whole range.

commit 8b3ac9f
Author: Benjamin Coddington <[email protected]>
Date:   Mon Jun 9 13:21:56 2025 -0400

    SUNRPC: Cleanup/fix initial rq_pages allocation

    While investigating some reports of memory-constrained NUMA machines
    failing to mount v3 and v4.0 nfs mounts, we found that svc_init_buffer()
    was not attempting to retry allocations from the bulk page allocator.
    Typically, this results in a single page allocation being returned and
    the mount attempt fails with -ENOMEM.  A retry would have allowed the mount
    to succeed.

    Additionally, it seems that the bulk allocation in svc_init_buffer() is
    redundant because svc_alloc_arg() will perform the required allocation and
    does the correct thing to retry the allocations.

    The call to allocate memory in svc_alloc_arg() drops the preferred node
    argument, but I expect we'll still allocate on the preferred node because
    the allocation call happens within the svc thread context, which chooses
    the node with memory closest to the current thread's execution.

    This patch cleans out the bulk allocation in svc_init_buffer() to allow
    svc_alloc_arg() to handle the allocation/retry logic for rq_pages.

    Signed-off-by: Benjamin Coddington <[email protected]>
    Reviewed-by: Jeff Layton <[email protected]>
    Fixes: ed603bc ("sunrpc: Replace the rq_pages array with dynamically-allocated memory")
    Signed-off-by: Chuck Lever <[email protected]>

Signed-off-by: Benjamin Coddington <[email protected]>

Approved-by: Olga Kornievskaia <[email protected]>
Approved-by: Paulo Alcantara <[email protected]>
Approved-by: Scott Mayhew <[email protected]>
Approved-by: CKI KWF Bot <[email protected]>

Merged-by: Augusto Caringi <[email protected]>

Author: caringi

net/sunrpc/svc.c

@@ -630,27 +630,6 @@ svc_destroy(struct svc_serv **servp)
 }
 EXPORT_SYMBOL_GPL(svc_destroy);
 
-static bool
-svc_init_buffer(struct svc_rqst *rqstp, unsigned int size, int node)
-{
-	unsigned long pages, ret;
-
-	/* bc_xprt uses fore channel allocated buffers */
-	if (svc_is_backchannel(rqstp))
-		return true;
-
-	pages = size / PAGE_SIZE + 1; /* extra page as we hold both request and reply.
-				       * We assume one is at most one page
-				       */
-	WARN_ON_ONCE(pages > RPCSVC_MAXPAGES);
-	if (pages > RPCSVC_MAXPAGES)
-		pages = RPCSVC_MAXPAGES;
-
-	ret = alloc_pages_bulk_array_node(GFP_KERNEL, node, pages,
-					  rqstp->rq_pages);
-	return ret == pages;
-}
-
 /*
  * Release an RPC server buffer
  */
@@ -690,9 +669,6 @@ svc_rqst_alloc(struct svc_serv *serv, struct svc_pool *pool, int node)
 	if (!rqstp->rq_resp)
 		goto out_enomem;
 
-	if (!svc_init_buffer(rqstp, serv->sv_max_mesg, node))
-		goto out_enomem;
-
 	return rqstp;
 out_enomem:
 	svc_rqst_free(rqstp);