[PATCH intel-net v2] ice: fix reservation of resources for RDMA when disabled

[PATCH intel-net v2] ice: fix reservation of resources for RDMA when disabled

[Jesse Brandeburg]

From: Jesse Brandeburg <jbrandeburg@cloudflare.com>

If the CONFIG_INFINIBAND_IRDMA symbol is not enabled as a module or a
built-in, then don't let the driver reserve resources for RDMA. The result
of this change is a large savings in resources for older kernels, and a
cleaner driver configuration for the IRDMA=n case for old and new kernels.

Implement this by avoiding enabling the RDMA capability when scanning
hardware capabilities.

Note: Loading the out-of-tree irdma driver in connection to the in-kernel
ice driver, is not supported, and should not be attempted, especially when
disabling IRDMA in the kernel config.

Fixes: d25a0fc41c1f ("ice: Initialize RDMA support")
Signed-off-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Acked-by: Dave Ertman <david.m.ertman@intel.com>
---
v2: resend with acks, add note about oot irdma (Leon), reword commit
message
v1: https://lore.kernel.org/netdev/20241114000105.703740-1-jbrandeb@kernel.org/
---
 drivers/net/ethernet/intel/ice/ice_common.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index 7a2a2e8da8fa..1e801300310e 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c
@@ -2271,7 +2271,8 @@ ice_parse_common_caps(struct ice_hw *hw, struct ice_hw_common_caps *caps,
 			  caps->nvm_unified_update);
 		break;
 	case ICE_AQC_CAPS_RDMA:
-		caps->rdma = (number == 1);
+		if (IS_ENABLED(CONFIG_INFINIBAND_IRDMA))
+			caps->rdma = (number == 1);
 		ice_debug(hw, ICE_DBG_INIT, "%s: rdma = %d\n", prefix, caps->rdma);
 		break;
 	case ICE_AQC_CAPS_MAX_MTU:
-- 
2.43.0

RE: [Intel-wired-lan] [PATCH intel-net v2] ice: fix reservation of resources for RDMA when disabled

[Loktionov, Aleksandr]

> -----Original Message-----
> From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf Of
> Jesse Brandeburg
> Sent: Thursday, March 6, 2025 6:57 PM
> To: intel-wired-lan@lists.osuosl.org
> Cc: Brandeburg, Jesse <jbrandeburg@cloudflare.com>;
> netdev@vger.kernel.org; kernel-team@cloudflare.com; jbrandeb@kernel.org;
> leon@kernel.org; Kitszel, Przemyslaw <przemyslaw.kitszel@intel.com>;
> Ertman, David M <david.m.ertman@intel.com>
> Subject: [Intel-wired-lan] [PATCH intel-net v2] ice: fix reservation of resources
> for RDMA when disabled
> 
> From: Jesse Brandeburg <jbrandeburg@cloudflare.com>
> 
> If the CONFIG_INFINIBAND_IRDMA symbol is not enabled as a module or a
> built-in, then don't let the driver reserve resources for RDMA. The result of this
> change is a large savings in resources for older kernels, and a cleaner driver
> configuration for the IRDMA=n case for old and new kernels.
> 
> Implement this by avoiding enabling the RDMA capability when scanning
> hardware capabilities.
> 
> Note: Loading the out-of-tree irdma driver in connection to the in-kernel ice
> driver, is not supported, and should not be attempted, especially when
> disabling IRDMA in the kernel config.
> 
> Fixes: d25a0fc41c1f ("ice: Initialize RDMA support")
> Signed-off-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
> Acked-by: Dave Ertman <david.m.ertman@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>

> ---
> v2: resend with acks, add note about oot irdma (Leon), reword commit
> message
> v1: https://lore.kernel.org/netdev/20241114000105.703740-1-
> jbrandeb@kernel.org/
> ---
>  drivers/net/ethernet/intel/ice/ice_common.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/intel/ice/ice_common.c
> b/drivers/net/ethernet/intel/ice/ice_common.c
> index 7a2a2e8da8fa..1e801300310e 100644
> --- a/drivers/net/ethernet/intel/ice/ice_common.c
> +++ b/drivers/net/ethernet/intel/ice/ice_common.c
> @@ -2271,7 +2271,8 @@ ice_parse_common_caps(struct ice_hw *hw, struct
> ice_hw_common_caps *caps,
>  			  caps->nvm_unified_update);
>  		break;
>  	case ICE_AQC_CAPS_RDMA:
> -		caps->rdma = (number == 1);
> +		if (IS_ENABLED(CONFIG_INFINIBAND_IRDMA))
> +			caps->rdma = (number == 1);
>  		ice_debug(hw, ICE_DBG_INIT, "%s: rdma = %d\n", prefix, caps-
> >rdma);
>  		break;
>  	case ICE_AQC_CAPS_MAX_MTU:
> --
> 2.43.0

RE: [Intel-wired-lan] [PATCH intel-net v2] ice: fix reservation of resources for RDMA when disabled

[Rinitha, SX]

> -----Original Message-----
> From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf Of Jesse Brandeburg
> Sent: 06 March 2025 23:27
> To: intel-wired-lan@lists.osuosl.org
> Cc: Brandeburg, Jesse <jbrandeburg@cloudflare.com>; netdev@vger.kernel.org; kernel-team@cloudflare.com; jbrandeb@kernel.org; leon@kernel.org; Kitszel, Przemyslaw <przemyslaw.kitszel@intel.com>; Ertman, David M <david.m.ertman@intel.com>
> Subject: [Intel-wired-lan] [PATCH intel-net v2] ice: fix reservation of resources for RDMA when disabled
>
> From: Jesse Brandeburg <jbrandeburg@cloudflare.com>
>
> If the CONFIG_INFINIBAND_IRDMA symbol is not enabled as a module or a built-in, then don't let the driver reserve resources for RDMA. The result of this change is a large savings in resources for older kernels, and a cleaner driver configuration for the IRDMA=n case for old and new kernels.
>
> Implement this by avoiding enabling the RDMA capability when scanning hardware capabilities.
>
> Note: Loading the out-of-tree irdma driver in connection to the in-kernel ice driver, is not supported, and should not be attempted, especially when disabling IRDMA in the kernel config.
>
> Fixes: d25a0fc41c1f ("ice: Initialize RDMA support")
> Signed-off-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
> Acked-by: Dave Ertman <david.m.ertman@intel.com>
> ---
> v2: resend with acks, add note about oot irdma (Leon), reword commit message
> v1: https://lore.kernel.org/netdev/20241114000105.703740-1-jbrandeb@kernel.org/
> ---
> drivers/net/ethernet/intel/ice/ice_common.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>

Tested-by: Rinitha S <sx.rinitha@intel.com> (A Contingent worker at Intel)