| Message ID | 20240108124717.1845481-4-karol.kolacinski@intel.com (mailing list archive) |
|---|---|
| State | Superseded |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | ice: fix timestamping in reset process | expand |
> -----Original Message----- > From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf Of Karol Kolacinski > Sent: Monday, January 8, 2024 6:17 PM > To: intel-wired-lan@lists.osuosl.org > Cc: Keller, Jacob E <jacob.e.keller@intel.com>; netdev@vger.kernel.org; Kolacinski, Karol <karol.kolacinski@intel.com>; Nguyen, Anthony L <anthony.l.nguyen@intel.com>; Brandeburg, Jesse <jesse.brandeburg@intel.com> > Subject: [Intel-wired-lan] [PATCH v5 iwl-next 3/6] ice: rename verify_cached to has_ready_bitmap > > From: Jacob Keller <jacob.e.keller@intel.com> > > The tx->verify_cached flag is used to inform the Tx timestamp tracking > code whether it needs to verify the cached Tx timestamp value against > a previous captured value. This is necessary on E810 hardware which does > not have a Tx timestamp ready bitmap. > > In addition, we currently rely on the fact that the > ice_get_phy_tx_tstamp_ready() function returns all 1s for E810 hardware. > Instead of introducing a brand new flag, rename and verify_cached to > has_ready_bitmap, inverting the relevant checks. > > Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> > Signed-off-by: Karol Kolacinski <karol.kolacinski@intel.com> > Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> > --- > drivers/net/ethernet/intel/ice/ice_ptp.c | 31 ++++++++++++------------ > drivers/net/ethernet/intel/ice/ice_ptp.h | 8 +++--- > 2 files changed, 20 insertions(+), 19 deletions(-) > Tested-by: Pucha Himasekhar Reddy <himasekharx.reddy.pucha@intel.com> (A Contingent worker at Intel)
On Mon, Jan 08, 2024 at 01:47:14PM +0100, Karol Kolacinski wrote: > From: Jacob Keller <jacob.e.keller@intel.com> > > The tx->verify_cached flag is used to inform the Tx timestamp tracking > code whether it needs to verify the cached Tx timestamp value against > a previous captured value. This is necessary on E810 hardware which does > not have a Tx timestamp ready bitmap. > > In addition, we currently rely on the fact that the > ice_get_phy_tx_tstamp_ready() function returns all 1s for E810 hardware. > Instead of introducing a brand new flag, rename and verify_cached to > has_ready_bitmap, inverting the relevant checks. From the above I understand what this patch does. But not why this change is desirable. I think it would be useful to state that. Also, perhaps it just me, but it seems that renaming verify_cached and weeding out assumptions about the return value of ice_get_phy_tx_tstamp_ready() are separate, albeit related changes: I might have gone for two patches instead of one. > Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> > Signed-off-by: Karol Kolacinski <karol.kolacinski@intel.com> > Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> ...
> -----Original Message----- > From: Simon Horman <horms@kernel.org> > Sent: Monday, January 15, 2024 2:37 AM > To: Kolacinski, Karol <karol.kolacinski@intel.com> > Cc: intel-wired-lan@lists.osuosl.org; netdev@vger.kernel.org; Nguyen, Anthony L > <anthony.l.nguyen@intel.com>; Brandeburg, Jesse > <jesse.brandeburg@intel.com>; Keller, Jacob E <jacob.e.keller@intel.com> > Subject: Re: [PATCH v5 iwl-next 3/6] ice: rename verify_cached to > has_ready_bitmap > > On Mon, Jan 08, 2024 at 01:47:14PM +0100, Karol Kolacinski wrote: > > From: Jacob Keller <jacob.e.keller@intel.com> > > > > The tx->verify_cached flag is used to inform the Tx timestamp tracking > > code whether it needs to verify the cached Tx timestamp value against > > a previous captured value. This is necessary on E810 hardware which does > > not have a Tx timestamp ready bitmap. > > > > In addition, we currently rely on the fact that the > > ice_get_phy_tx_tstamp_ready() function returns all 1s for E810 hardware. > > Instead of introducing a brand new flag, rename and verify_cached to > > has_ready_bitmap, inverting the relevant checks. > > From the above I understand what this patch does. > But not why this change is desirable. > I think it would be useful to state that. > The main motivation is that it is easier to understand the resulting behavior than to rely on a hidden assumption of how ice_get_phy_tx_stamp_ready() works. > Also, perhaps it just me, but it seems that > renaming verify_cached and weeding out assumptions > about the return value of ice_get_phy_tx_tstamp_ready() > are separate, albeit related changes: > I might have gone for two patches instead of one. > It could possibly be split up. The motivation was to stop relying on the fact that ice_get_phy_tx_tstamp_ready() returns all 1s, since we already know whether we can safely check it or not. But then using "verify_cached" to do so reads weird with the code because its name no longer clearly indicates its purpose. Instead, we rename it to better reflect its intention. Thanks, Jake > > Signed-off-by: Jacob Keller <jacob.e.keller@intel.com> > > Signed-off-by: Karol Kolacinski <karol.kolacinski@intel.com> > > Reviewed-by: Jacob Keller <jacob.e.keller@intel.com> > > ...
diff --git a/drivers/net/ethernet/intel/ice/ice_ptp.c b/drivers/net/ethernet/intel/ice/ice_ptp.c index c309d3fd5a4e..4b1b2c577df7 100644 --- a/drivers/net/ethernet/intel/ice/ice_ptp.c +++ b/drivers/net/ethernet/intel/ice/ice_ptp.c @@ -601,17 +601,13 @@ void ice_ptp_complete_tx_single_tstamp(struct ice_ptp_tx *tx) /* Read the low 32 bit value */ raw_tstamp |= (u64)rd32(&pf->hw, PF_SB_ATQBAH); - /* For PHYs which don't implement a proper timestamp ready bitmap, - * verify that the timestamp value is different from the last cached - * timestamp. If it is not, skip this for now assuming it hasn't yet - * been captured by hardware. + /* Devices using this interface always verify the timestamp differs + * relative to the last cached timestamp value. */ - if (!drop_ts && tx->verify_cached && - raw_tstamp == tx->tstamps[idx].cached_tstamp) + if (raw_tstamp == tx->tstamps[idx].cached_tstamp) return; - if (tx->verify_cached && raw_tstamp) - tx->tstamps[idx].cached_tstamp = raw_tstamp; + tx->tstamps[idx].cached_tstamp = raw_tstamp; clear_bit(idx, tx->in_use); skb = tx->tstamps[idx].skb; tx->tstamps[idx].skb = NULL; @@ -701,9 +697,11 @@ static void ice_ptp_process_tx_tstamp(struct ice_ptp_tx *tx) hw = &pf->hw; /* Read the Tx ready status first */ - err = ice_get_phy_tx_tstamp_ready(hw, tx->block, &tstamp_ready); - if (err) - return; + if (tx->has_ready_bitmap) { + err = ice_get_phy_tx_tstamp_ready(hw, tx->block, &tstamp_ready); + if (err) + return; + } /* Drop packets if the link went down */ link_up = ptp_port->link_up; @@ -731,7 +729,8 @@ static void ice_ptp_process_tx_tstamp(struct ice_ptp_tx *tx) * If we do not, the hardware logic for generating a new * interrupt can get stuck on some devices. */ - if (!(tstamp_ready & BIT_ULL(phy_idx))) { + if (tx->has_ready_bitmap && + !(tstamp_ready & BIT_ULL(phy_idx))) { if (drop_ts) goto skip_ts_read; @@ -751,7 +750,7 @@ static void ice_ptp_process_tx_tstamp(struct ice_ptp_tx *tx) * from the last cached timestamp. If it is not, skip this for * now assuming it hasn't yet been captured by hardware. */ - if (!drop_ts && tx->verify_cached && + if (!drop_ts && !tx->has_ready_bitmap && raw_tstamp == tx->tstamps[idx].cached_tstamp) continue; @@ -761,7 +760,7 @@ static void ice_ptp_process_tx_tstamp(struct ice_ptp_tx *tx) skip_ts_read: spin_lock_irqsave(&tx->lock, flags); - if (tx->verify_cached && raw_tstamp) + if (!tx->has_ready_bitmap && raw_tstamp) tx->tstamps[idx].cached_tstamp = raw_tstamp; clear_bit(idx, tx->in_use); skb = tx->tstamps[idx].skb; @@ -1014,7 +1013,7 @@ ice_ptp_init_tx_e82x(struct ice_pf *pf, struct ice_ptp_tx *tx, u8 port) tx->block = port / ICE_PORTS_PER_QUAD; tx->offset = (port % ICE_PORTS_PER_QUAD) * INDEX_PER_PORT_E82X; tx->len = INDEX_PER_PORT_E82X; - tx->verify_cached = 0; + tx->has_ready_bitmap = 1; return ice_ptp_alloc_tx_tracker(tx); } @@ -1037,7 +1036,7 @@ ice_ptp_init_tx_e810(struct ice_pf *pf, struct ice_ptp_tx *tx) * verify new timestamps against cached copy of the last read * timestamp. */ - tx->verify_cached = 1; + tx->has_ready_bitmap = 0; return ice_ptp_alloc_tx_tracker(tx); } diff --git a/drivers/net/ethernet/intel/ice/ice_ptp.h b/drivers/net/ethernet/intel/ice/ice_ptp.h index afe454abe997..aa7a5588d11d 100644 --- a/drivers/net/ethernet/intel/ice/ice_ptp.h +++ b/drivers/net/ethernet/intel/ice/ice_ptp.h @@ -100,7 +100,7 @@ struct ice_perout_channel { * the last timestamp we read for a given index. If the current timestamp * value is the same as the cached value, we assume a new timestamp hasn't * been captured. This avoids reporting stale timestamps to the stack. This is - * only done if the verify_cached flag is set in ice_ptp_tx structure. + * only done if the has_ready_bitmap flag is not set in ice_ptp_tx structure. */ struct ice_tx_tstamp { struct sk_buff *skb; @@ -130,7 +130,9 @@ enum ice_tx_tstamp_work { * @init: if true, the tracker is initialized; * @calibrating: if true, the PHY is calibrating the Tx offset. During this * window, timestamps are temporarily disabled. - * @verify_cached: if true, verify new timestamp differs from last read value + * @has_ready_bitmap: if true, the hardware has a valid Tx timestamp ready + * bitmap register. If false, fall back to verifying new + * timestamp values against previously cached copy. * @last_ll_ts_idx_read: index of the last LL TS read by the FW */ struct ice_ptp_tx { @@ -143,7 +145,7 @@ struct ice_ptp_tx { u8 len; u8 init : 1; u8 calibrating : 1; - u8 verify_cached : 1; + u8 has_ready_bitmap : 1; s8 last_ll_ts_idx_read; };