| Message ID | 1463489221-24989-1-git-send-email-michal.kazior@tieto.com (mailing list archive) |
|---|---|
| State | Not Applicable |
| Delegated to: | Kalle Valo |
| Headers | show |
Michal Kazior <michal.kazior@tieto.com> writes: > Some setups suffer performance regressions with > current wake_tx_queue implementation. > > Signed-off-by: Michal Kazior <michal.kazior@tieto.com> > --- > Hi Roman, > > Can you give this patch a try and see if it helps > with your performance problems, please? Did anyone seeing the performance regression manage to test this? It would be good to get feedback, good or bad, before I push this to 4.7. Full patch here: https://patchwork.kernel.org/patch/9111981/
On 2 June 2016 at 16:14, Valo, Kalle <kvalo@qca.qualcomm.com> wrote: > Michal Kazior <michal.kazior@tieto.com> writes: > >> Some setups suffer performance regressions with >> current wake_tx_queue implementation. >> >> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> >> --- >> Hi Roman, >> >> Can you give this patch a try and see if it helps >> with your performance problems, please? > > Did anyone seeing the performance regression manage to test this? It > would be good to get feedback, good or bad, before I push this to 4.7. > > Full patch here: > > https://patchwork.kernel.org/patch/9111981/ Sorry, didn't try yet, bad timing for me :( Will try to test by the end of the week Regards, Roman -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Roman Yeryomin <leroi.lists@gmail.com> writes: > On 2 June 2016 at 16:14, Valo, Kalle <kvalo@qca.qualcomm.com> wrote: >> Michal Kazior <michal.kazior@tieto.com> writes: >> >>> Some setups suffer performance regressions with >>> current wake_tx_queue implementation. >>> >>> Signed-off-by: Michal Kazior <michal.kazior@tieto.com> >>> --- >>> Hi Roman, >>> >>> Can you give this patch a try and see if it helps >>> with your performance problems, please? >> >> Did anyone seeing the performance regression manage to test this? It >> would be good to get feedback, good or bad, before I push this to 4.7. >> >> Full patch here: >> >> https://patchwork.kernel.org/patch/9111981/ > > Sorry, didn't try yet, bad timing for me :( > Will try to test by the end of the week Roman, did you have a chance to test this? Michal, could you submit this patch as a formal patch, please? I think we should take it even if Roman hasn't tested it.
diff --git a/drivers/net/wireless/ath/ath10k/core.h b/drivers/net/wireless/ath/ath10k/core.h index 1379054000f9..44e7c2cd2e9b 100644 --- a/drivers/net/wireless/ath/ath10k/core.h +++ b/drivers/net/wireless/ath/ath10k/core.h @@ -657,6 +657,7 @@ struct ath10k_fw_components { struct ath10k { struct ath_common ath_common; struct ieee80211_hw *hw; + struct ieee80211_ops *ops; struct device *dev; u8 mac_addr[ETH_ALEN]; diff --git a/drivers/net/wireless/ath/ath10k/mac.c b/drivers/net/wireless/ath/ath10k/mac.c index 0e24f9ee8bff..5fa3888afa41 100644 --- a/drivers/net/wireless/ath/ath10k/mac.c +++ b/drivers/net/wireless/ath/ath10k/mac.c @@ -7396,21 +7396,32 @@ static const struct ieee80211_channel ath10k_5ghz_channels[] = { struct ath10k *ath10k_mac_create(size_t priv_size) { struct ieee80211_hw *hw; + struct ieee80211_ops *ops; struct ath10k *ar; - hw = ieee80211_alloc_hw(sizeof(struct ath10k) + priv_size, &ath10k_ops); - if (!hw) + ops = kmemdup(&ath10k_ops, sizeof(ath10k_ops), GFP_KERNEL); + if (!ops) return NULL; + hw = ieee80211_alloc_hw(sizeof(struct ath10k) + priv_size, ops); + if (!hw) { + kfree(ops); + return NULL; + } + ar = hw->priv; ar->hw = hw; + ar->ops = ops; return ar; } void ath10k_mac_destroy(struct ath10k *ar) { + struct ieee80211_ops *ops = ar->ops; + ieee80211_free_hw(ar->hw); + kfree(ops); } static const struct ieee80211_iface_limit ath10k_if_limits[] = { @@ -7838,6 +7849,15 @@ int ath10k_mac_register(struct ath10k *ar) ath10k_warn(ar, "failed to initialise DFS pattern detector\n"); } + /* Current wake_tx_queue implementation imposes a significant + * performance penalty in some setups. The tx scheduling code needs + * more work anyway so disable the wake_tx_queue unless firmware + * supports the pull-push mechanism. + */ + if (!test_bit(ATH10K_FW_FEATURE_PEER_FLOW_CONTROL, + ar->running_fw->fw_file.fw_features)) + ar->ops->wake_tx_queue = NULL; + ret = ath_regd_init(&ar->ath_common.regulatory, ar->hw->wiphy, ath10k_reg_notifier); if (ret) {
Some setups suffer performance regressions with current wake_tx_queue implementation. Signed-off-by: Michal Kazior <michal.kazior@tieto.com> --- Hi Roman, Can you give this patch a try and see if it helps with your performance problems, please? drivers/net/wireless/ath/ath10k/core.h | 1 + drivers/net/wireless/ath/ath10k/mac.c | 24 ++++++++++++++++++++++-- 2 files changed, 23 insertions(+), 2 deletions(-)