Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
ice: cleanups and preparation for live migration

Jake Keller says:

Various cleanups and preparation to the ice driver code for supporting
SR-IOV live migration.

The logic for unpacking Rx queue context data is added. This is the inverse
of the existing packing logic. Thanks to <linux/packing.h> this is trivial
to add.

Code to enable both reading and writing the Tx queue context for a queue
over a shared hardware register interface is added. Thanks to ice_adapter,
this is locked across all PFs that need to use it, preventing concurrency
issues with multiple PFs.

The RSS hash configuration requested by a VF is cached within the VF
structure. This will be used to track and restore the same configuration
during migration load.

ice_sriov_set_msix_vec_count() is updated to use pci_iov_vf_id() instead of
open-coding a worse equivalent, and checks to avoid rebuilding MSI-X if the
current request is for the existing amount of vectors.

A new ice_get_vf_by_dev() helper function is added to simplify accessing a
VF from its PCI device structure. This will be used more heavily within the
live migration code itself.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
  ice: introduce ice_get_vf_by_dev() wrapper
  ice: avoid rebuilding if MSI-X vector count is unchanged
  ice: use pci_iov_vf_id() to get VF ID
  ice: expose VF functions used by live migration
  ice: move ice_vsi_update_l2tsel to ice_lib.c
  ice: save RSS hash configuration for migration
  ice: add functions to get and set Tx queue context
  ice: add support for reading and unpacking Rx queue context
====================

Link: https://patch.msgid.link/20250710214518.1824208-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

1 files changed, 35 insertions, 0 deletions

diff --git a/drivers/net/ethernet/intel/ice/ice_lib.c b/drivers/net/ethernet/intel/ice/ice_lib.c
index 2f1782e9357f..1be1e429a7c8 100644
--- a/drivers/net/ethernet/intel/ice/ice_lib.c
+++ b/drivers/net/ethernet/intel/ice/ice_lib.c
@@ -4020,3 +4020,38 @@ ice_vsi_update_local_lb(struct ice_vsi *vsi, bool set)
 	vsi->info = ctx.info;
 	return 0;
 }
+
+/**
+ * ice_vsi_update_l2tsel - update l2tsel field for all Rx rings on this VSI
+ * @vsi: VSI used to update l2tsel on
+ * @l2tsel: l2tsel setting requested
+ *
+ * Use the l2tsel setting to update all of the Rx queue context bits for l2tsel.
+ * This will modify which descriptor field the first offloaded VLAN will be
+ * stripped into.
+ */
+void ice_vsi_update_l2tsel(struct ice_vsi *vsi, enum ice_l2tsel l2tsel)
+{
+	struct ice_hw *hw = &vsi->back->hw;
+	u32 l2tsel_bit;
+	int i;
+
+	if (l2tsel == ICE_L2TSEL_EXTRACT_FIRST_TAG_L2TAG2_2ND)
+		l2tsel_bit = 0;
+	else
+		l2tsel_bit = BIT(ICE_L2TSEL_BIT_OFFSET);
+
+	for (i = 0; i < vsi->alloc_rxq; i++) {
+		u16 pfq = vsi->rxq_map[i];
+		u32 qrx_context_offset;
+		u32 regval;
+
+		qrx_context_offset =
+			QRX_CONTEXT(ICE_L2TSEL_QRX_CONTEXT_REG_IDX, pfq);
+
+		regval = rd32(hw, qrx_context_offset);
+		regval &= ~BIT(ICE_L2TSEL_BIT_OFFSET);
+		regval |= l2tsel_bit;
+		wr32(hw, qrx_context_offset, regval);
+	}
+}