Re: [Intel-wired-lan] [PATCH iwl-next v4] idpf: export RX hardware timestamping information to XDP

2026-01-13 Thread Alexander Lobakin
From: Paul Menzel 
Date: Sat, 10 Jan 2026 09:35:58 +0100

> Dear Mina,
> 
> 
> Thank you for your patch. Some minor comments, should you resend.

[...]

>> @@ -166,6 +173,19 @@ idpf_xdp_get_qw2(struct idpf_xdp_rx_desc *desc,
>>   #endif
>>   }
>>   +static inline void
>> +idpf_xdp_get_qw3(struct idpf_xdp_rx_desc *desc,
>> + const struct virtchnl2_rx_flex_desc_adv_nic_3 *rxd)
>> +{
>> +#ifdef __LIBETH_WORD_ACCESS
>> +    desc->qw3 = ((const typeof(desc))rxd)->qw3;
>> +#else
>> +    desc->qw3 = ((u64)le32_to_cpu(rxd->ts_high) << 32) |
>> +    ((u64)le16_to_cpu(rxd->fmd6) << 16) |
>> +    le16_to_cpu(rxd->l2tag1);
>> +#endif
> 
> It’s done elsewhere in the file, but I wonder why use the preprocessor
> and not plain C code, and let the linker(?) remove the unneeded branch?

Do you mean IS_ENABLED()?
In the libeth_xdp code, several paths won't build with IS_ENABLED(),
here in idpf it might work out, but I did it the same way as in
libeth_xdp for consistency.

> 
>> +}
>> +
>>   void idpf_xdp_set_features(const struct idpf_vport *vport);
>>     int idpf_xdp(struct net_device *dev, struct netdev_bpf *xdp);
> 
> 
> Kind regards,
> 
> Paul

Thanks,
Olek


Re: [Intel-wired-lan] [PATCH iwl-next v4] idpf: export RX hardware timestamping information to XDP

2026-01-11 Thread Mina Almasry via Intel-wired-lan
On Sat, Jan 10, 2026 at 12:36 AM Paul Menzel  wrote:
>
> Dear Mina,
>
>
> Thank you for your patch. Some minor comments, should you resend.
>

Thanks, looks like I have reviews and this is on its way, but should I
resend I will fix the minor comments.

> Am 23.12.25 um 20:46 schrieb Mina Almasry via Intel-wired-lan:
> > From: YiFei Zhu 
> >
> > The logic is similar to idpf_rx_hwtstamp, but the data is exported
> > as a BPF kfunc instead of appended to an skb to support grabbing
> > timestamps in xsk packets.
> >
> > A idpf_queue_has(PTP, rxq) condition is added to check the queue
> > supports PTP similar to idpf_rx_process_skb_fields.
> >
> > Tested using an xsk connection and checking xdp timestamps are
> > retreivable in received packets.
>
> retr*ie*vable
>
> It’d be great if you could share the commands.
>

I don't have easy repro to share in the commit message. The test
involves hacking up the xsk_rr Sami used for his busypolling patch to
enable xdp metadata and retrieve timestamps, or (what I did) actually
set up openonload with this patch and test onload can get the
timestamps. Let me see what I can do, but it's likely too much context
for someone unfamiliar to piece together.

-- 
Thanks,
Mina


Re: [Intel-wired-lan] [PATCH iwl-next v4] idpf: export RX hardware timestamping information to XDP

2026-01-10 Thread Paul Menzel

Dear Mina,


Thank you for your patch. Some minor comments, should you resend.

Am 23.12.25 um 20:46 schrieb Mina Almasry via Intel-wired-lan:

From: YiFei Zhu 

The logic is similar to idpf_rx_hwtstamp, but the data is exported
as a BPF kfunc instead of appended to an skb to support grabbing
timestamps in xsk packets.

A idpf_queue_has(PTP, rxq) condition is added to check the queue
supports PTP similar to idpf_rx_process_skb_fields.

Tested using an xsk connection and checking xdp timestamps are
retreivable in received packets.


retr*ie*vable

It’d be great if you could share the commands.


Cc: [email protected]
Signed-off-by: YiFei Zhu 
Signed-off-by: Mina Almasry 
Reviewed-by: Aleksandr Loktionov 

---

v4: 
https://lore.kernel.org/netdev/[email protected]/
- Fix indentation (lobakin)
- I kept the (u64) casts for all bit shifted bits in idpf_xdp_get_qw3
   and friends as I see all idpf_xdp_get_qw* functions do the cast in all
   bit-shifted variables.

v3: 
https://lore.kernel.org/netdev/[email protected]/
- Do the idpf_queue_has(PTP) check before we read qw1 (lobakin)
- Fix _qw1 not copying over ts_low on on !__LIBETH_WORD_ACCESS systems
   (AI)

v2: 
https://lore.kernel.org/netdev/[email protected]/
- Fixed alphabetical ordering
- Use the xdp desc type instead of virtchnl one (required some added
   helpers)

---
  drivers/net/ethernet/intel/idpf/xdp.c | 31 +++
  drivers/net/ethernet/intel/idpf/xdp.h | 20 +
  2 files changed, 51 insertions(+)

diff --git a/drivers/net/ethernet/intel/idpf/xdp.c 
b/drivers/net/ethernet/intel/idpf/xdp.c
index 958d16f87424..0916d201bf98 100644
--- a/drivers/net/ethernet/intel/idpf/xdp.c
+++ b/drivers/net/ethernet/intel/idpf/xdp.c
@@ -2,6 +2,7 @@
  /* Copyright (C) 2025 Intel Corporation */
  
  #include "idpf.h"

+#include "idpf_ptp.h"
  #include "idpf_virtchnl.h"
  #include "xdp.h"
  #include "xsk.h"
@@ -391,8 +392,38 @@ static int idpf_xdpmo_rx_hash(const struct xdp_md *ctx, 
u32 *hash,
pt);
  }
  
+static int idpf_xdpmo_rx_timestamp(const struct xdp_md *ctx, u64 *timestamp)

+{
+   const struct libeth_xdp_buff *xdp = (typeof(xdp))ctx;
+   struct idpf_xdp_rx_desc desc __uninitialized;
+   const struct idpf_rx_queue *rxq;
+   u64 cached_time, ts_ns;
+   u32 ts_high;
+
+   rxq = libeth_xdp_buff_to_rq(xdp, typeof(*rxq), xdp_rxq);
+
+   if (!idpf_queue_has(PTP, rxq))
+   return -ENODATA;
+
+   idpf_xdp_get_qw1(&desc, xdp->desc);
+
+   if (!(idpf_xdp_rx_ts_low(&desc) & VIRTCHNL2_RX_FLEX_TSTAMP_VALID))
+   return -ENODATA;
+
+   cached_time = READ_ONCE(rxq->cached_phc_time);
+
+   idpf_xdp_get_qw3(&desc, xdp->desc);
+
+   ts_high = idpf_xdp_rx_ts_high(&desc);
+   ts_ns = idpf_ptp_tstamp_extend_32b_to_64b(cached_time, ts_high);
+
+   *timestamp = ts_ns;
+   return 0;
+}
+
  static const struct xdp_metadata_ops idpf_xdpmo = {
.xmo_rx_hash= idpf_xdpmo_rx_hash,
+   .xmo_rx_timestamp   = idpf_xdpmo_rx_timestamp,


Append the unit?


  };
  
  void idpf_xdp_set_features(const struct idpf_vport *vport)

diff --git a/drivers/net/ethernet/intel/idpf/xdp.h 
b/drivers/net/ethernet/intel/idpf/xdp.h
index 479f5ef3c604..1748a0d73547 100644
--- a/drivers/net/ethernet/intel/idpf/xdp.h
+++ b/drivers/net/ethernet/intel/idpf/xdp.h
@@ -112,11 +112,13 @@ struct idpf_xdp_rx_desc {
aligned_u64 qw1;
  #define IDPF_XDP_RX_BUF   GENMASK_ULL(47, 32)
  #define IDPF_XDP_RX_EOP   BIT_ULL(1)
+#define IDPF_XDP_RX_TS_LOW GENMASK_ULL(31, 24)
  
  	aligned_u64		qw2;

  #define IDPF_XDP_RX_HASH  GENMASK_ULL(31, 0)
  
  	aligned_u64		qw3;

+#define IDPF_XDP_RX_TS_HIGHGENMASK_ULL(63, 32)
  } __aligned(4 * sizeof(u64));
  static_assert(sizeof(struct idpf_xdp_rx_desc) ==
  sizeof(struct virtchnl2_rx_flex_desc_adv_nic_3));
@@ -128,6 +130,8 @@ static_assert(sizeof(struct idpf_xdp_rx_desc) ==
  #define idpf_xdp_rx_buf(desc) FIELD_GET(IDPF_XDP_RX_BUF, (desc)->qw1)
  #define idpf_xdp_rx_eop(desc) !!((desc)->qw1 & IDPF_XDP_RX_EOP)
  #define idpf_xdp_rx_hash(desc)FIELD_GET(IDPF_XDP_RX_HASH, (desc)->qw2)
+#define idpf_xdp_rx_ts_low(desc)   FIELD_GET(IDPF_XDP_RX_TS_LOW, 
(desc)->qw1)
+#define idpf_xdp_rx_ts_high(desc)  FIELD_GET(IDPF_XDP_RX_TS_HIGH, 
(desc)->qw3)
  
  static inline void

  idpf_xdp_get_qw0(struct idpf_xdp_rx_desc *desc,
@@ -149,6 +153,9 @@ idpf_xdp_get_qw1(struct idpf_xdp_rx_desc *desc,
desc->qw1 = ((const typeof(desc))rxd)->qw1;
  #else
desc->qw1 = ((u64)le16_to_cpu(rxd->buf_id) << 32) |
+   ((u64)rxd->ts_low << 24) |
+   ((u64)rxd->fflags1 << 16) |
+   ((u64)rxd->status_err1 << 8) |
rxd->status_err0_qw1;
  #endif
  }
@@ -166,6 +173,19 @@ idpf_xdp_ge

Re: [Intel-wired-lan] [PATCH iwl-next v4] idpf: export RX hardware timestamping information to XDP

2026-01-08 Thread Simon Horman
On Tue, Dec 23, 2025 at 07:46:46PM +, Mina Almasry wrote:
> From: YiFei Zhu 
> 
> The logic is similar to idpf_rx_hwtstamp, but the data is exported
> as a BPF kfunc instead of appended to an skb to support grabbing
> timestamps in xsk packets.
> 
> A idpf_queue_has(PTP, rxq) condition is added to check the queue
> supports PTP similar to idpf_rx_process_skb_fields.
> 
> Tested using an xsk connection and checking xdp timestamps are
> retreivable in received packets.
> 
> Cc: [email protected]
> Signed-off-by: YiFei Zhu 
> Signed-off-by: Mina Almasry 
> Reviewed-by: Aleksandr Loktionov 

Reviewed-by: Simon Horman 



Re: [Intel-wired-lan] [PATCH iwl-next v4] idpf: export RX hardware timestamping information to XDP

2026-01-05 Thread Alexander Lobakin
From: Mina Almasry 
Date: Tue, 23 Dec 2025 19:46:46 +

> From: YiFei Zhu 
> 
> The logic is similar to idpf_rx_hwtstamp, but the data is exported
> as a BPF kfunc instead of appended to an skb to support grabbing
> timestamps in xsk packets.
> 
> A idpf_queue_has(PTP, rxq) condition is added to check the queue
> supports PTP similar to idpf_rx_process_skb_fields.
> 
> Tested using an xsk connection and checking xdp timestamps are
> retreivable in received packets.
> 
> Cc: [email protected]
> Signed-off-by: YiFei Zhu 
> Signed-off-by: Mina Almasry 
> Reviewed-by: Aleksandr Loktionov 

Reviewed-by: Alexander Lobakin 

Thanks,
Olek