The ERR007885 will lead to a TDAR race condition for mutliQ when the
driver sets TDAR and the UDMA clears TDAR simultaneously or in a small
window (2-4 cycles). And it will cause the udma_tx and udma_tx_arbiter
state machines to hang. Therefore, the commit 53bb20d1faba ("net: fec:
add variable reg_desc_active to speed things up") and the commit
a179aad12bad ("net: fec: ERR007885 Workaround for conventional TX") have
added the workaround to fix the potential issue for the conventional TX
path. Similarly, the XDP TX path should also have the potential hang
issue, so add the workaround for XDP TX path.

Fixes: 6d6b39f180b8 ("net: fec: add initial XDP support")
Signed-off-by: Wei Fang <wei.fang@nxp.com>
---
 drivers/net/ethernet/freescale/fec_main.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/freescale/fec_main.c b/drivers/net/ethernet/freescale/fec_main.c
index 3222359ac15b..e2b75d1970ae 100644
--- a/drivers/net/ethernet/freescale/fec_main.c
+++ b/drivers/net/ethernet/freescale/fec_main.c
@@ -3948,7 +3948,12 @@ static int fec_enet_txq_xmit_frame(struct fec_enet_private *fep,
 	txq->bd.cur = bdp;
 
 	/* Trigger transmission start */
-	writel(0, txq->bd.reg_desc_active);
+	if (!(fep->quirks & FEC_QUIRK_ERR007885) ||
+	    !readl(txq->bd.reg_desc_active) ||
+	    !readl(txq->bd.reg_desc_active) ||
+	    !readl(txq->bd.reg_desc_active) ||
+	    !readl(txq->bd.reg_desc_active))
+		writel(0, txq->bd.reg_desc_active);
 
 	return 0;
 }
-- 
2.34.1

On 11/28/25 3:59 AM, Wei Fang wrote:
> The ERR007885 will lead to a TDAR race condition for mutliQ when the
> driver sets TDAR and the UDMA clears TDAR simultaneously or in a small
> window (2-4 cycles). And it will cause the udma_tx and udma_tx_arbiter
> state machines to hang. Therefore, the commit 53bb20d1faba ("net: fec:
> add variable reg_desc_active to speed things up") and the commit
> a179aad12bad ("net: fec: ERR007885 Workaround for conventional TX") have
> added the workaround to fix the potential issue for the conventional TX
> path. Similarly, the XDP TX path should also have the potential hang
> issue, so add the workaround for XDP TX path.
> 
> Fixes: 6d6b39f180b8 ("net: fec: add initial XDP support")
> Signed-off-by: Wei Fang <wei.fang@nxp.com>
> ---
>  drivers/net/ethernet/freescale/fec_main.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/ethernet/freescale/fec_main.c b/drivers/net/ethernet/freescale/fec_main.c
> index 3222359ac15b..e2b75d1970ae 100644
> --- a/drivers/net/ethernet/freescale/fec_main.c
> +++ b/drivers/net/ethernet/freescale/fec_main.c
> @@ -3948,7 +3948,12 @@ static int fec_enet_txq_xmit_frame(struct fec_enet_private *fep,
>  	txq->bd.cur = bdp;
>  
>  	/* Trigger transmission start */
> -	writel(0, txq->bd.reg_desc_active);
> +	if (!(fep->quirks & FEC_QUIRK_ERR007885) ||
> +	    !readl(txq->bd.reg_desc_active) ||
> +	    !readl(txq->bd.reg_desc_active) ||
> +	    !readl(txq->bd.reg_desc_active) ||
> +	    !readl(txq->bd.reg_desc_active))
> +		writel(0, txq->bd.reg_desc_active);
>  
>  	return 0;
>  }

LGTM!

Side note for a net-next follow-up: please consider moving this logic in
a reusable helper, since you already have a few potentials call sites.

/P