Message ID | 20200116210859.7376-1-f.fainelli@gmail.com |
---|---|
State | Accepted |
Delegated to: | David Miller |
Headers | show |
Series | [net] net: systemport: Fixed queue mapping in internal ring map | expand |
From: Florian Fainelli <f.fainelli@gmail.com> Date: Thu, 16 Jan 2020 13:08:58 -0800 > We would not be transmitting using the correct SYSTEMPORT transmit queue > during ndo_select_queue() which looks up the internal TX ring map > because while establishing the mapping we would be off by 4, so for > instance, when we populate switch port mappings we would be doing: > > switch port 0, queue 0 -> ring index #0 > switch port 0, queue 1 -> ring index #1 > ... > switch port 0, queue 3 -> ring index #3 > switch port 1, queue 0 -> ring index #8 (4 + 4 * 1) > ... > > instead of using ring index #4. This would cause our ndo_select_queue() > to use the fallback queue mechanism which would pick up an incorrect > ring for that switch port. Fix this by using the correct switch queue > number instead of SYSTEMPORT queue number. > > Fixes: 3ed67ca243b3 ("net: systemport: Simplify queue mapping logic") > Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> Applied, but I had to fix the SHA1-ID of the Fixes tag to be: Fixes: 25c440704661 ("net: systemport: Simplify queue mapping logic") Thanks.
On 1/17/2020 4:32 AM, David Miller wrote: > From: Florian Fainelli <f.fainelli@gmail.com> > Date: Thu, 16 Jan 2020 13:08:58 -0800 > >> We would not be transmitting using the correct SYSTEMPORT transmit queue >> during ndo_select_queue() which looks up the internal TX ring map >> because while establishing the mapping we would be off by 4, so for >> instance, when we populate switch port mappings we would be doing: >> >> switch port 0, queue 0 -> ring index #0 >> switch port 0, queue 1 -> ring index #1 >> ... >> switch port 0, queue 3 -> ring index #3 >> switch port 1, queue 0 -> ring index #8 (4 + 4 * 1) >> ... >> >> instead of using ring index #4. This would cause our ndo_select_queue() >> to use the fallback queue mechanism which would pick up an incorrect >> ring for that switch port. Fix this by using the correct switch queue >> number instead of SYSTEMPORT queue number. >> >> Fixes: 3ed67ca243b3 ("net: systemport: Simplify queue mapping logic") >> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> > > Applied, but I had to fix the SHA1-ID of the Fixes tag to be: > > Fixes: 25c440704661 ("net: systemport: Simplify queue mapping logic") Just like the other patch what happened is that the check_fixes script from Stephen was able to resolve this incorrect SHA1 because I have both the Broadcom STB downstream remote configured as well as upstream and the script was not yet updated to make sure this was an object that can be reached in the upstream remotes (yours, Linus...). This should be corrected now, sorry about that.
diff --git a/drivers/net/ethernet/broadcom/bcmsysport.c b/drivers/net/ethernet/broadcom/bcmsysport.c index 825af709708e..d6b1a153f9df 100644 --- a/drivers/net/ethernet/broadcom/bcmsysport.c +++ b/drivers/net/ethernet/broadcom/bcmsysport.c @@ -2323,7 +2323,7 @@ static int bcm_sysport_map_queues(struct notifier_block *nb, ring->switch_queue = qp; ring->switch_port = port; ring->inspect = true; - priv->ring_map[q + port * num_tx_queues] = ring; + priv->ring_map[qp + port * num_tx_queues] = ring; qp++; } @@ -2338,7 +2338,7 @@ static int bcm_sysport_unmap_queues(struct notifier_block *nb, struct net_device *slave_dev; unsigned int num_tx_queues; struct net_device *dev; - unsigned int q, port; + unsigned int q, qp, port; priv = container_of(nb, struct bcm_sysport_priv, dsa_notifier); if (priv->netdev != info->master) @@ -2364,7 +2364,8 @@ static int bcm_sysport_unmap_queues(struct notifier_block *nb, continue; ring->inspect = false; - priv->ring_map[q + port * num_tx_queues] = NULL; + qp = ring->switch_queue; + priv->ring_map[qp + port * num_tx_queues] = NULL; } return 0;
We would not be transmitting using the correct SYSTEMPORT transmit queue during ndo_select_queue() which looks up the internal TX ring map because while establishing the mapping we would be off by 4, so for instance, when we populate switch port mappings we would be doing: switch port 0, queue 0 -> ring index #0 switch port 0, queue 1 -> ring index #1 ... switch port 0, queue 3 -> ring index #3 switch port 1, queue 0 -> ring index #8 (4 + 4 * 1) ... instead of using ring index #4. This would cause our ndo_select_queue() to use the fallback queue mechanism which would pick up an incorrect ring for that switch port. Fix this by using the correct switch queue number instead of SYSTEMPORT queue number. Fixes: 3ed67ca243b3 ("net: systemport: Simplify queue mapping logic") Signed-off-by: Florian Fainelli <f.fainelli@gmail.com> --- drivers/net/ethernet/broadcom/bcmsysport.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-)