[PATCH net 6/6] ethtool: rss: avoid device context leak on reply-build failure

[Jakub Kicinski <kuba@kernel.org>]

We wait with filling the reply for new RSS context creation
until after the driver ->create_rxfh_context call. The driver
needs to fill some of the defaults in the context. The failure
of rss_fill_reply() is somewhat theoretical, but doesn't take
much effort to handle it properly. Call ->remove_rxfh_context().

If the driver's remove callback fails (some implementations like sfc
can return real command errors from firmware RPCs) - skip the xa_erase
and kfree, leaving the context in the xarray. This matches how
ethnl_rss_delete_doit() behaves.

Fixes: a166ab7816c5 ("ethtool: rss: support creating contexts via Netlink")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
---
CC: andrew@lunn.ch
CC: gal@nvidia.com
---
 net/ethtool/rss.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/net/ethtool/rss.c b/net/ethtool/rss.c
index f5cf214f8f85..53792f53f922 100644
--- a/net/ethtool/rss.c
+++ b/net/ethtool/rss.c
@@ -1106,7 +1106,7 @@ int ethnl_rss_create_doit(struct sk_buff *skb, struct genl_info *info)
 	ntf_fail |= rss_fill_reply(rsp, &req.base, &data.base);
 	if (WARN_ON(!hdr || ntf_fail)) {
 		ret = -EMSGSIZE;
-		goto exit_unlock;
+		goto err_remove_ctx;
 	}
 
 	genlmsg_end(rsp, hdr);
@@ -1134,6 +1134,10 @@ int ethnl_rss_create_doit(struct sk_buff *skb, struct genl_info *info)
 	nlmsg_free(rsp);
 	return ret;
 
+err_remove_ctx:
+	if (ops->remove_rxfh_context(dev, ctx, req.rss_context, NULL))
+		/* leave the context on failure, like ethnl_rss_delete_doit() */
+		goto exit_unlock;
 err_ctx_id_free:
 	xa_erase(&dev->ethtool->rss_ctx, req.rss_context);
 err_unlock_free_ctx:
-- 
2.54.0