-----Original Message-----
From: Keller, Jacob E <jacob.e.keller@intel.com>
Sent: Thursday, March 26, 2026 12:42 AM
To: Korba, Przemyslaw <redacted>; intel-wired-lan@lists.osuosl.org
Cc: netdev@vger.kernel.org; Nguyen, Anthony L <anthony.l.nguyen@intel.com>; Kitszel, Przemyslaw <przemyslaw.kitszel@intel.com>
Subject: Re: [Intel-wired-lan] [PATCH iwl-net v2] ice: fix posted write support for sideband queue operations

On 3/19/2026 4:13 AM, Przemyslaw Korba wrote:

quoted

On E830, PTP time adjustment commands sent via
SBQ don't generate completion responses, causing the driver to
timeout waiting and return -EIO, when trying:

 Hm. Is this really for E830? I didn't realize that device actually
sends SBQ commands for PTP?

According to ice_ptp_adj_clock for E830, it says "E830 sync PHYs
automatically after setting GLTSYN_SHADJ".

Did you mean E825-C here? Can you confirm the device type affected? Or
am I missing something?

quoted

phc_ctl eth8 get adj 2 get
dmesg: ice 0000:1a:00.0: PTP failed to adjust time, err -5

Add support for posted mode not to wait for completion response.

Fixes: 8f5ee3c477a8 ("ice: add support for sideband messages")
Signed-off-by: Przemyslaw Korba <redacted>
---
v2:
- change "postpone" to "posted"
- init struct with {} instead of {0}
v1:
https://lore.kernel.org/intel-wired-lan/20260310110700.345904-1-przemyslaw.korba@intel.com/ (local)

 drivers/net/ethernet/intel/ice/ice_common.c   | 7 ++++++-
 drivers/net/ethernet/intel/ice/ice_controlq.c | 4 ++++
 drivers/net/ethernet/intel/ice/ice_controlq.h | 1 +
 3 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_common.c b/drivers/net/ethernet/intel/ice/ice_common.c
index 8866902efb91..c89c6ca1281b 100644
--- a/drivers/net/ethernet/intel/ice/ice_common.c
+++ b/drivers/net/ethernet/intel/ice/ice_common.c

@@ -1765,6 +1765,7 @@ int ice_sbq_rw_reg(struct ice_hw *hw, struct ice_sbq_msg_input *in, u16 flags)
 {

This affects the ice_sbq_rw_reg function which is used by several
variants including the E822 devices, E825-C with ETH56G, and even E810
devices.

Do all these devices not provide completion? Or do we simply not care
about waiting?

I don't see a single call to ice_sbq_rw_reg for E830, so I suspect this
is correct but for a different device, and the commit message is just a
typo?

Hi Jake,
Thanks for review! Grzegorz's patch he mentioned indeed makes it so E830 uses this functionality.
You are right there are other devices we need to consider. I've found more complete Karol's patch that does exactly that, but got stuck on reviews:
https://lore.kernel.org/intel-wired-lan/20250520110823.1937981-7-karol.kolacinski@intel.com/ (local)
I addressed past reviews, and will try to merge this patch, I sent updated version to our internal mailing list for review : )

quoted

 	struct ice_sbq_cmd_desc desc = {0};
 	struct ice_sbq_msg_req msg = {0};
+	struct ice_sq_cd cd = {};
 	u16 msg_len;
 	int status;

@@ -1785,10 +1786,14 @@ int ice_sbq_rw_reg(struct ice_hw *hw, struct ice_sbq_msg_input *in, u16 flags)
 		 */
 		msg_len -= sizeof(msg.data);

+	if (in->opcode == ice_sbq_msg_wr)
+		cd.posted = 1;
+
 	desc.flags = cpu_to_le16(flags);
 	desc.opcode = cpu_to_le16(ice_sbq_opc_neigh_dev_req);
 	desc.param0.cmd_len = cpu_to_le16(msg_len);
-	status = ice_sbq_send_cmd(hw, &desc, &msg, msg_len, NULL);
+	status = ice_sbq_send_cmd(hw, &desc, &msg, msg_len, &cd);
+
 	if (!status && !in->opcode)
 		in->data = le32_to_cpu
 			(((struct ice_sbq_msg_cmpl *)&msg)->data);

diff --git a/drivers/net/ethernet/intel/ice/ice_controlq.c b/drivers/net/ethernet/intel/ice/ice_controlq.c
index dcb837cadd18..a6008dc77fa4 100644
--- a/drivers/net/ethernet/intel/ice/ice_controlq.c
+++ b/drivers/net/ethernet/intel/ice/ice_controlq.c

@@ -1086,6 +1086,10 @@ ice_sq_send_cmd(struct ice_hw *hw, struct ice_ctl_q_info *cq,
 	wr32(hw, cq->sq.tail, cq->sq.next_to_use);
 	ice_flush(hw);

+	/* If the message is posted, don't wait for completion. */
+	if (cd && cd->posted)
+		goto sq_send_command_error;
+
 	/* Wait for the command to complete. If it finishes within the
 	 * timeout, copy the descriptor back to temp.
 	 */

diff --git a/drivers/net/ethernet/intel/ice/ice_controlq.h b/drivers/net/ethernet/intel/ice/ice_controlq.h
index 788040dd662e..c50d6fcbacba 100644
--- a/drivers/net/ethernet/intel/ice/ice_controlq.h
+++ b/drivers/net/ethernet/intel/ice/ice_controlq.h

@@ -77,6 +77,7 @@ struct ice_ctl_q_ring {
 /* sq transaction details */
 struct ice_sq_cd {
 	struct libie_aq_desc *wb_desc;
+	u8 posted : 1;
 };

 /* rq event information */

base-commit: acd2abc52dea91c3bc3d1b6dd8a92b9631d48bbf