[PATCH net v2] ibmvnic: Harden device login requests
From: Thomas Falcon <hidden>
Date: 2020-06-15 15:33:05
Also in:
netdev
Subsystem:
ibm power sriov virtual nic device driver, linux for powerpc (32-bit and 64-bit), networking drivers, the rest · Maintainers:
Haren Myneni, Rick Lindsley, Madhavan Srinivasan, Andrew Lunn, "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds
The VNIC driver's "login" command sequence is the final step in the driver's initialization process with device firmware, confirming the available device queue resources to be utilized by the driver. Under high system load, firmware may not respond to the request in a timely manner or may abort the request. In such cases, the driver should reattempt the login command sequence. In case of a device error, the number of retries is bounded. Signed-off-by: Thomas Falcon <redacted> --- v2: declare variables in Reverse Christmas tree format --- drivers/net/ethernet/ibm/ibmvnic.c | 21 +++++++++++++++++---- 1 file changed, 17 insertions(+), 4 deletions(-)
diff --git a/drivers/net/ethernet/ibm/ibmvnic.c b/drivers/net/ethernet/ibm/ibmvnic.c
index 1b4d04e..2baf7b3 100644
--- a/drivers/net/ethernet/ibm/ibmvnic.c
+++ b/drivers/net/ethernet/ibm/ibmvnic.c@@ -842,12 +842,13 @@ static int ibmvnic_login(struct net_device *netdev) struct ibmvnic_adapter *adapter = netdev_priv(netdev); unsigned long timeout = msecs_to_jiffies(30000); int retry_count = 0; + int retries = 10; bool retry; int rc; do { retry = false; - if (retry_count > IBMVNIC_MAX_QUEUES) { + if (retry_count > retries) { netdev_warn(netdev, "Login attempts exceeded\n"); return -1; }
@@ -862,11 +863,23 @@ static int ibmvnic_login(struct net_device *netdev) if (!wait_for_completion_timeout(&adapter->init_done, timeout)) { - netdev_warn(netdev, "Login timed out\n"); - return -1; + netdev_warn(netdev, "Login timed out, retrying...\n"); + retry = true; + adapter->init_done_rc = 0; + retry_count++; + continue; } - if (adapter->init_done_rc == PARTIALSUCCESS) { + if (adapter->init_done_rc == ABORTED) { + netdev_warn(netdev, "Login aborted, retrying...\n"); + retry = true; + adapter->init_done_rc = 0; + retry_count++; + /* FW or device may be busy, so + * wait a bit before retrying login + */ + msleep(500); + } else if (adapter->init_done_rc == PARTIALSUCCESS) { retry_count++; release_sub_crqs(adapter, 1);
--
1.8.3.1