[PATCH net v2] net: sparx5: Fix invalid timestamps

Aakash Menon aakash.r.menon at gmail.com
Mon Sep 16 22:18:29 PDT 2024


Bit 270-271 are occasionally unexpectedly set by the hardware. This issue
was observed with 10G SFPs causing huge time errors (> 30ms) in PTP. Only
30 bits are needed for the nanosecond part of the timestamp, clear 2 most
significant bits before extracting timestamp from the internal frame
header.

Fixes: 70dfe25cd866 ("net: sparx5: Update extraction/injection for timestamping")
Signed-off-by: Aakash Menon <aakash.menon at protempis.com>
---
v2:
- Wrap patch descriptions at 75 characters wide.
- Use GENMASK(5,0) instead of masking with 0x3F
- Update Fixes tag to be on the same line
- Link to v1 -https://lore.kernel.org/r/20240913193357.21899-1-aakash.menon@protempis.com
drivers/net/ethernet/microchip/sparx5/sparx5_packet.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/microchip/sparx5/sparx5_packet.c b/drivers/net/ethernet/microchip/sparx5/sparx5_packet.c
index f3f5fb420468..70427643f777 100644
--- a/drivers/net/ethernet/microchip/sparx5/sparx5_packet.c
+++ b/drivers/net/ethernet/microchip/sparx5/sparx5_packet.c
@@ -45,8 +45,12 @@ void sparx5_ifh_parse(u32 *ifh, struct frame_info *info)
 	fwd = (fwd >> 5);
 	info->src_port = FIELD_GET(GENMASK(7, 1), fwd);
 
+	/*
+	 * Bit 270-271 are occasionally unexpectedly set by the hardware,
+	 * clear bits before extracting timestamp
+	 */
 	info->timestamp =
-		((u64)xtr_hdr[2] << 24) |
+		((u64)(xtr_hdr[2] & GENMASK(5, 0)) << 24) |
 		((u64)xtr_hdr[3] << 16) |
 		((u64)xtr_hdr[4] <<  8) |
 		((u64)xtr_hdr[5] <<  0);
-- 
2.46.0




More information about the linux-arm-kernel mailing list