[PATCH iwl-next] ice: init desired_dcbx_cfg in default DCB config

[PATCH iwl-next] ice: init desired_dcbx_cfg in default DCB config

[Aleksandr Loktionov]

From: Lukasz Czapnik <lukasz.czapnik@intel.com>

When DCBX is disabled in firmware the driver falls back to software LLDP
mode and applies a default DCB configuration via ice_dcb_sw_dflt_cfg().
This function properly initializes local_dcbx_cfg with valid parameters
including etscfg.maxtcs from hardware capabilities. However,
desired_dcbx_cfg was never initialized in this path.

All DCB netlink functions (ice_dcbnl_setpfc, ice_dcbnl_setets, etc.)
use desired_dcbx_cfg as the base configuration for user-requested changes.
When desired_dcbx_cfg remains uninitialized with etscfg.maxtcs=0, the
firmware rejects the configuration for 4+ port NICs, causing DCB
configuration commands to fail.

It is not a problem for 1 or 2 port NICs where we support 8 TCs - in
that case FW accepts maxtc=0, treating it as 8.

Fix it by copying local_dcbx_cfg (which was freshly initialized) into
desired_dcbx_cfg after the default config is applied.

Fixes: b94b013eb626 ("ice: Implement DCBNL support")
Signed-off-by: Lukasz Czapnik <lukasz.czapnik@intel.com>
Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_dcb_lib.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/net/ethernet/intel/ice/ice_dcb_lib.c b/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
index bd77f1c..060744c 100644
--- a/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
+++ b/drivers/net/ethernet/intel/ice/ice_dcb_lib.c
@@ -669,6 +669,9 @@ int ice_dcb_sw_dflt_cfg(struct ice_pf *pf, bool ets_willing, bool locked)
 	if (ret)
 		return ret;
 
+	/* init desired_dcbx_cfg from local_dcbx_cfg */
+	pi->qos_cfg.desired_dcbx_cfg = pi->qos_cfg.local_dcbx_cfg;
+
 	return ice_query_port_ets(pi, &buf, sizeof(buf), NULL);
 }
 
-- 
2.52.0

Re: [PATCH iwl-next] ice: init desired_dcbx_cfg in default DCB config

[Simon Horman]

On Fri, Mar 20, 2026 at 06:05:41AM +0100, Aleksandr Loktionov wrote:
> From: Lukasz Czapnik <lukasz.czapnik@intel.com>
> 
> When DCBX is disabled in firmware the driver falls back to software LLDP
> mode and applies a default DCB configuration via ice_dcb_sw_dflt_cfg().
> This function properly initializes local_dcbx_cfg with valid parameters
> including etscfg.maxtcs from hardware capabilities. However,
> desired_dcbx_cfg was never initialized in this path.
> 
> All DCB netlink functions (ice_dcbnl_setpfc, ice_dcbnl_setets, etc.)
> use desired_dcbx_cfg as the base configuration for user-requested changes.
> When desired_dcbx_cfg remains uninitialized with etscfg.maxtcs=0, the
> firmware rejects the configuration for 4+ port NICs, causing DCB
> configuration commands to fail.
> 
> It is not a problem for 1 or 2 port NICs where we support 8 TCs - in
> that case FW accepts maxtc=0, treating it as 8.
> 
> Fix it by copying local_dcbx_cfg (which was freshly initialized) into
> desired_dcbx_cfg after the default config is applied.
> 
> Fixes: b94b013eb626 ("ice: Implement DCBNL support")
> Signed-off-by: Lukasz Czapnik <lukasz.czapnik@intel.com>
> Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>

Reviewed-by: Simon Horman <horms@kernel.org>