Message ID | B8D63297-0A88-42A3-B789-3849895CF201@frei.media (mailing list archive) |
---|---|
State | Accepted |
Commit | 47cc0ca91c9e4dde68d69f35e139360ef356267f |
Delegated to: | Kalle Valo |
Headers | show |
Matthias Frei <mf@frei.media> writes: > Set the a-mpdu reference number in ath10k to make it accessible in the receivers radiotap header. Implemented as in ath9k. > The reference number is needed for troubleshooting and research at the receivers site (e.g. to identify mpdu's that were aggregated in an a-mpdu) > > Signed-off-by: Matthias Frei <mf@frei.media> I did some changes in pending branch, please review them: https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/commit/?h=pending&id=f722728460a5c9e9200a7f1362fa605a714c1968 > --- a/drivers/net/wireless/ath/ath10k/core.h > +++ b/drivers/net/wireless/ath/ath10k/core.h > @@ -989,6 +989,9 @@ struct ath10k { > u32 reg_ack_cts_timeout_orig; > } fw_coverage; > > + /* AMPDU */ > + u32 ampdu_ref; The comment is not providing any extra value, I'll remove that. What about locking? How is this protected or doesn't it need anything? Also I renamed this to ampdu_reference just to be consistent with the mac80211 name. > --- a/drivers/net/wireless/ath/ath10k/htt_rx.c > +++ b/drivers/net/wireless/ath/ath10k/htt_rx.c > @@ -877,16 +877,24 @@ static void ath10k_htt_rx_h_ppdu(struct ath10k *ar, > status->nss = 0; > status->encoding = RX_ENC_LEGACY; > status->bw = RATE_INFO_BW_20; > - status->flag &= ~RX_FLAG_MACTIME_END; > - status->flag |= RX_FLAG_NO_SIGNAL_VAL; > + status->flag &= ~(RX_FLAG_MACTIME_END | RX_FLAG_AMPDU_IS_LAST); > + status->flag |= RX_FLAG_NO_SIGNAL_VAL | RX_FLAG_AMPDU_DETAILS | RX_FLAG_AMPDU_LAST_KNOWN; This added a new warning: drivers/net/wireless/ath/ath10k/htt_rx.c:894: line over 90 characters I fixed that by separating the setting of ampdu flags. > + /* set ampdu ref */ > + status->ampdu_reference = ar->ampdu_ref; The comment is not telling anything new so removed it. > - if (is_last_ppdu) > + if (is_last_ppdu) { > ath10k_htt_rx_h_mactime(ar, status, rxd); > + > + /* set ampdu last segment flag */ > + status->flag |= RX_FLAG_AMPDU_IS_LAST; > + ar->ampdu_ref++; > + } So this counter is per wiphy, not per vdev. Is that ok?
> So this counter is per wiphy, not per vdev. Is that ok? Well, I think this should be okay. The counter depends on the point of view. Counting per wiphy and looking on the physical transmission it counts every incoming ampdu that was received by the physical wireless card. On the other hand, looking „down the network stack“ from the application layer, you may want a consistent ampdu reference number. I think doing the counting per wiphy should be okay, because it is still a virtual counter on the receivers site. Doing the counting per wiphy still allows to identify aggregated mpdu’s on the application layer, as you just search for mpdus with the same reference number. Additionally, counting per whipy enables you to get a feeling on the application layer, how much incoming load the wiphy has to handle at the moment. Considering this counter is only used for debugging and research, this additional information can be useful or if not, it does not change anything. > What about locking? How is this protected or doesn't it need anything? I don’t think locking is needed here since the counter is per wiphy and it is only incremented on the last mpdu of a ampdu. Additionally, locking is also not used in ath9k. Thanks! Best, Matthias Frei > Am 06.07.2017 um 09:42 schrieb Kalle Valo <kvalo@qca.qualcomm.com>: > > Matthias Frei <mf@frei.media> writes: > >> Set the a-mpdu reference number in ath10k to make it accessible in the receivers radiotap header. Implemented as in ath9k. >> The reference number is needed for troubleshooting and research at the receivers site (e.g. to identify mpdu's that were aggregated in an a-mpdu) >> >> Signed-off-by: Matthias Frei <mf@frei.media> > > I did some changes in pending branch, please review them: > > https://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git/commit/?h=pending&id=f722728460a5c9e9200a7f1362fa605a714c1968 > >> --- a/drivers/net/wireless/ath/ath10k/core.h >> +++ b/drivers/net/wireless/ath/ath10k/core.h >> @@ -989,6 +989,9 @@ struct ath10k { >> u32 reg_ack_cts_timeout_orig; >> } fw_coverage; >> >> + /* AMPDU */ >> + u32 ampdu_ref; > > The comment is not providing any extra value, I'll remove that. > > What about locking? How is this protected or doesn't it need anything? > > Also I renamed this to ampdu_reference just to be consistent with the > mac80211 name. > >> --- a/drivers/net/wireless/ath/ath10k/htt_rx.c >> +++ b/drivers/net/wireless/ath/ath10k/htt_rx.c >> @@ -877,16 +877,24 @@ static void ath10k_htt_rx_h_ppdu(struct ath10k *ar, >> status->nss = 0; >> status->encoding = RX_ENC_LEGACY; >> status->bw = RATE_INFO_BW_20; >> - status->flag &= ~RX_FLAG_MACTIME_END; >> - status->flag |= RX_FLAG_NO_SIGNAL_VAL; >> + status->flag &= ~(RX_FLAG_MACTIME_END | RX_FLAG_AMPDU_IS_LAST); >> + status->flag |= RX_FLAG_NO_SIGNAL_VAL | RX_FLAG_AMPDU_DETAILS | RX_FLAG_AMPDU_LAST_KNOWN; > > This added a new warning: > > drivers/net/wireless/ath/ath10k/htt_rx.c:894: line over 90 characters > > I fixed that by separating the setting of ampdu flags. > >> + /* set ampdu ref */ >> + status->ampdu_reference = ar->ampdu_ref; > > The comment is not telling anything new so removed it. > >> - if (is_last_ppdu) >> + if (is_last_ppdu) { >> ath10k_htt_rx_h_mactime(ar, status, rxd); >> + >> + /* set ampdu last segment flag */ >> + status->flag |= RX_FLAG_AMPDU_IS_LAST; >> + ar->ampdu_ref++; >> + } > > So this counter is per wiphy, not per vdev. Is that ok? > > -- > Kalle Valo
Matthias Frei <mf@frei.media> wrote: > Set the a-mpdu reference number in ath10k to make it accessible in the > receivers radiotap header. Implemented as in ath9k. The reference number is > needed for troubleshooting and research at the receivers site (e.g. to identify > mpdu's that were aggregated in an a-mpdu) > > Signed-off-by: Matthias Frei <mf@frei.media> > [kvalo@qca.qualcomm.com: fix checkpatch warning, commit log cleanup] > Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com> Patch applied to ath-next branch of ath.git, thanks. 47cc0ca91c9e ath10k: set a-mpdu receiver reference number
diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h index bf09151..82adf55 100644 --- a/drivers/net/wireless/ath/ath10k/core.h +++ b/drivers/net/wireless/ath/ath10k/core.h @@ -989,6 +989,9 @@ struct ath10k { u32 reg_ack_cts_timeout_orig; } fw_coverage; + /* AMPDU */ + u32 ampdu_ref; + /* must be last */ u8 drv_priv[0] __aligned(sizeof(void *)); }; diff --git a/drivers/net/wireless/ath/ath10k/htt_rx.c b/drivers/net/wireless/ath/ath10k/htt_rx.c index 84b6067..a4de02b 100644 --- a/drivers/net/wireless/ath/ath10k/htt_rx.c +++ b/drivers/net/wireless/ath/ath10k/htt_rx.c @@ -877,16 +877,24 @@ static void ath10k_htt_rx_h_ppdu(struct ath10k *ar, status->nss = 0; status->encoding = RX_ENC_LEGACY; status->bw = RATE_INFO_BW_20; - status->flag &= ~RX_FLAG_MACTIME_END; - status->flag |= RX_FLAG_NO_SIGNAL_VAL; + status->flag &= ~(RX_FLAG_MACTIME_END | RX_FLAG_AMPDU_IS_LAST); + status->flag |= RX_FLAG_NO_SIGNAL_VAL | RX_FLAG_AMPDU_DETAILS | RX_FLAG_AMPDU_LAST_KNOWN; + + /* set ampdu ref */ + status->ampdu_reference = ar->ampdu_ref; ath10k_htt_rx_h_signal(ar, status, rxd); ath10k_htt_rx_h_channel(ar, status, rxd, vdev_id); ath10k_htt_rx_h_rates(ar, status, rxd); } - if (is_last_ppdu) + if (is_last_ppdu) { ath10k_htt_rx_h_mactime(ar, status, rxd); + + /* set ampdu last segment flag */ + status->flag |= RX_FLAG_AMPDU_IS_LAST; + ar->ampdu_ref++; + } } static const char * const tid_to_ac[] = {
Set the a-mpdu reference number in ath10k to make it accessible in the receivers radiotap header. Implemented as in ath9k. The reference number is needed for troubleshooting and research at the receivers site (e.g. to identify mpdu's that were aggregated in an a-mpdu) Signed-off-by: Matthias Frei <mf@frei.media> --- drivers/net/wireless/ath/ath10k/core.h | 3 +++ drivers/net/wireless/ath/ath10k/htt_rx.c | 14 +++++++++++--- 2 files changed, 14 insertions(+), 3 deletions(-)