From: Benjamin Lin <[email protected]>
During runtime, the GI value in the WTBL is not updated in real-time. To
obtain the latest results for the TX GI, switch to use an MCU command.
Signed-off-by: Benjamin Lin <[email protected]>
Signed-off-by: Shayne Chen <[email protected]>
---
.../wireless/mediatek/mt76/mt76_connac_mcu.h | 2 +-
.../net/wireless/mediatek/mt76/mt7996/mac.c | 48 ++-----------------
.../net/wireless/mediatek/mt76/mt7996/main.c | 1 +
.../net/wireless/mediatek/mt76/mt7996/mcu.c | 47 ++++++++++++++++++
.../net/wireless/mediatek/mt76/mt7996/mcu.h | 22 +++++++++
5 files changed, 74 insertions(+), 46 deletions(-)
diff --git a/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h b/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
index 65844de6dccd..0185804d8ce3 100644
--- a/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
+++ b/drivers/net/wireless/mediatek/mt76/mt76_connac_mcu.h
@@ -1328,7 +1328,7 @@ enum {
};
enum UNI_ALL_STA_INFO_TAG {
- UNI_ALL_STA_TX_RATE,
+ UNI_ALL_STA_TXRX_RATE,
UNI_ALL_STA_TX_STAT,
UNI_ALL_STA_TXRX_ADM_STAT,
UNI_ALL_STA_TXRX_AIR_TIME,
diff --git a/drivers/net/wireless/mediatek/mt76/mt7996/mac.c b/drivers/net/wireless/mediatek/mt76/mt7996/mac.c
index 56dfbeb51504..9db610e2645f 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7996/mac.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7996/mac.c
@@ -102,7 +102,6 @@ static void mt7996_mac_sta_poll(struct mt7996_dev *dev)
};
struct ieee80211_sta *sta;
struct mt7996_sta *msta;
- struct rate_info *rate;
u32 tx_time[IEEE80211_NUM_ACS], rx_time[IEEE80211_NUM_ACS];
LIST_HEAD(sta_poll_list);
int i;
@@ -118,7 +117,6 @@ static void mt7996_mac_sta_poll(struct mt7996_dev *dev)
u32 addr, val;
u16 idx;
s8 rssi[4];
- u8 bw;
spin_lock_bh(&dev->mt76.sta_poll_lock);
if (list_empty(&sta_poll_list)) {
@@ -174,49 +172,6 @@ static void mt7996_mac_sta_poll(struct mt7996_dev *dev)
ieee80211_sta_register_airtime(sta, tid, tx_cur, rx_cur);
}
- /* We don't support reading GI info from txs packets.
- * For accurate tx status reporting and AQL improvement,
- * we need to make sure that flags match so polling GI
- * from per-sta counters directly.
- */
- rate = &msta->wcid.rate;
-
- switch (rate->bw) {
- case RATE_INFO_BW_320:
- bw = IEEE80211_STA_RX_BW_320;
- break;
- case RATE_INFO_BW_160:
- bw = IEEE80211_STA_RX_BW_160;
- break;
- case RATE_INFO_BW_80:
- bw = IEEE80211_STA_RX_BW_80;
- break;
- case RATE_INFO_BW_40:
- bw = IEEE80211_STA_RX_BW_40;
- break;
- default:
- bw = IEEE80211_STA_RX_BW_20;
- break;
- }
-
- addr = mt7996_mac_wtbl_lmac_addr(dev, idx, 6);
- val = mt76_rr(dev, addr);
- if (rate->flags & RATE_INFO_FLAGS_EHT_MCS) {
- addr = mt7996_mac_wtbl_lmac_addr(dev, idx, 5);
- val = mt76_rr(dev, addr);
- rate->eht_gi = FIELD_GET(GENMASK(25, 24), val);
- } else if (rate->flags & RATE_INFO_FLAGS_HE_MCS) {
- u8 offs = 24 + 2 * bw;
-
- rate->he_gi = (val & (0x3 << offs)) >> offs;
- } else if (rate->flags &
- (RATE_INFO_FLAGS_VHT_MCS | RATE_INFO_FLAGS_MCS)) {
- if (val & BIT(12 + bw))
- rate->flags |= RATE_INFO_FLAGS_SHORT_GI;
- else
- rate->flags &= ~RATE_INFO_FLAGS_SHORT_GI;
- }
-
/* get signal strength of resp frames (CTS/BA/ACK) */
addr = mt7996_mac_wtbl_lmac_addr(dev, idx, 34);
val = mt76_rr(dev, addr);
@@ -1298,6 +1253,8 @@ mt7996_mac_add_txs_skb(struct mt7996_dev *dev, struct mt76_wcid *wcid,
goto out;
rate.flags = RATE_INFO_FLAGS_VHT_MCS;
+ if (wcid->rate.flags & RATE_INFO_FLAGS_SHORT_GI)
+ rate.flags |= RATE_INFO_FLAGS_SHORT_GI;
break;
case MT_PHY_TYPE_HE_SU:
case MT_PHY_TYPE_HE_EXT_SU:
@@ -2312,6 +2269,7 @@ void mt7996_mac_work(struct work_struct *work)
mt7996_mac_update_stats(phy);
+ mt7996_mcu_get_all_sta_info(phy, UNI_ALL_STA_TXRX_RATE);
if (mtk_wed_device_active(&phy->dev->mt76.mmio.wed)) {
mt7996_mcu_get_all_sta_info(phy, UNI_ALL_STA_TXRX_ADM_STAT);
mt7996_mcu_get_all_sta_info(phy, UNI_ALL_STA_TXRX_MSDU_COUNT);
diff --git a/drivers/net/wireless/mediatek/mt76/mt7996/main.c b/drivers/net/wireless/mediatek/mt76/mt7996/main.c
index 7336eaa7b9ae..d9ba57ae9fdc 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7996/main.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7996/main.c
@@ -998,6 +998,7 @@ static void mt7996_sta_statistics(struct ieee80211_hw *hw,
sinfo->txrate.he_gi = txrate->he_gi;
sinfo->txrate.he_dcm = txrate->he_dcm;
sinfo->txrate.he_ru_alloc = txrate->he_ru_alloc;
+ sinfo->txrate.eht_gi = txrate->eht_gi;
}
sinfo->txrate.flags = txrate->flags;
sinfo->filled |= BIT_ULL(NL80211_STA_INFO_TX_BITRATE);
diff --git a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
index 8141c24ade50..b8d0b52be1e7 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
@@ -449,6 +449,43 @@ mt7996_mcu_ie_countdown(struct mt7996_dev *dev, struct sk_buff *skb)
}
}
+static int
+mt7996_mcu_update_tx_gi(struct rate_info *rate, struct all_sta_trx_rate *mcu_rate)
+{
+ switch (mcu_rate->tx_mode) {
+ case MT_PHY_TYPE_CCK:
+ case MT_PHY_TYPE_OFDM:
+ break;
+ case MT_PHY_TYPE_HT:
+ case MT_PHY_TYPE_HT_GF:
+ case MT_PHY_TYPE_VHT:
+ if (mcu_rate->tx_gi)
+ rate->flags |= RATE_INFO_FLAGS_SHORT_GI;
+ else
+ rate->flags &= ~RATE_INFO_FLAGS_SHORT_GI;
+ break;
+ case MT_PHY_TYPE_HE_SU:
+ case MT_PHY_TYPE_HE_EXT_SU:
+ case MT_PHY_TYPE_HE_TB:
+ case MT_PHY_TYPE_HE_MU:
+ if (mcu_rate->tx_gi > NL80211_RATE_INFO_HE_GI_3_2)
+ return -EINVAL;
+ rate->he_gi = mcu_rate->tx_gi;
+ break;
+ case MT_PHY_TYPE_EHT_SU:
+ case MT_PHY_TYPE_EHT_TRIG:
+ case MT_PHY_TYPE_EHT_MU:
+ if (mcu_rate->tx_gi > NL80211_RATE_INFO_EHT_GI_3_2)
+ return -EINVAL;
+ rate->eht_gi = mcu_rate->tx_gi;
+ break;
+ default:
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
static void
mt7996_mcu_rx_all_sta_info_event(struct mt7996_dev *dev, struct sk_buff *skb)
{
@@ -465,6 +502,16 @@ mt7996_mcu_rx_all_sta_info_event(struct mt7996_dev *dev, struct sk_buff *skb)
struct mt76_wcid *wcid;
switch (le16_to_cpu(res->tag)) {
+ case UNI_ALL_STA_TXRX_RATE:
+ wlan_idx = le16_to_cpu(res->rate[i].wlan_idx);
+ wcid = rcu_dereference(dev->mt76.wcid[wlan_idx]);
+
+ if (!wcid)
+ break;
+
+ if (mt7996_mcu_update_tx_gi(&wcid->rate, &res->rate[i]))
+ dev_err(dev->mt76.dev, "Failed to update TX GI\n");
+ break;
case UNI_ALL_STA_TXRX_ADM_STAT:
wlan_idx = le16_to_cpu(res->adm_stat[i].wlan_idx);
wcid = rcu_dereference(dev->mt76.wcid[wlan_idx]);
diff --git a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.h b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.h
index 1562c8a6a821..328edc354165 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.h
+++ b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.h
@@ -175,6 +175,27 @@ struct mt7996_mcu_mib {
__le64 data;
} __packed;
+struct all_sta_trx_rate {
+ __le16 wlan_idx;
+ u8 __rsv1[2];
+ u8 tx_mode;
+ u8 flags;
+ u8 tx_stbc;
+ u8 tx_gi;
+ u8 tx_bw;
+ u8 tx_ldpc;
+ u8 tx_mcs;
+ u8 tx_nss;
+ u8 rx_rate;
+ u8 rx_mode;
+ u8 rx_nsts;
+ u8 rx_gi;
+ u8 rx_coding;
+ u8 rx_stbc;
+ u8 rx_bw;
+ u8 __rsv2;
+} __packed;
+
struct mt7996_mcu_all_sta_info_event {
u8 rsv[4];
__le16 tag;
@@ -185,6 +206,7 @@ struct mt7996_mcu_all_sta_info_event {
u8 rsv3[2];
union {
+ struct all_sta_trx_rate rate[0];
struct {
__le16 wlan_idx;
u8 rsv[2];
--
2.39.2
On 11/2/23 03:02, Shayne Chen wrote:
> From: Benjamin Lin <[email protected]>
>
> During runtime, the GI value in the WTBL is not updated in real-time. To
> obtain the latest results for the TX GI, switch to use an MCU command.
Hello,
I do not see this callback happening on my system. What firmware version
is needed for this to work?
And where to find it...
Thanks,
Ben
--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com
On Mon, 2023-12-04 at 13:57 -0800, Ben Greear wrote:
>
> External email : Please do not click links or open attachments until
> you have verified the sender or the content.
> On 11/2/23 03:02, Shayne Chen wrote:
> > From: Benjamin Lin <[email protected]>
> >
> > During runtime, the GI value in the WTBL is not updated in real-
> time. To
> > obtain the latest results for the TX GI, switch to use an MCU
> command.
>
> Hello,
Hi Ben,
>
> I do not see this callback happening on my system. What firmware
> version
> is needed for this to work?
>
> And where to find it...
>
Please get testing firmware files from the following link to see if it
works on your environment:
https://github.com/csyuanc/linux-firmware
Thanks,
Shayne
> Thanks,
> Ben
>
On 12/7/23 09:07, [email protected] wrote:
> On Mon, 2023-12-04 at 13:57 -0800, Ben Greear wrote:
>>
>> External email : Please do not click links or open attachments until
>> you have verified the sender or the content.
>> On 11/2/23 03:02, Shayne Chen wrote:
>>> From: Benjamin Lin <[email protected]>
>>>
>>> During runtime, the GI value in the WTBL is not updated in real-
>> time. To
>>> obtain the latest results for the TX GI, switch to use an MCU
>> command.
>>
>> Hello,
>
> Hi Ben,
>>
>> I do not see this callback happening on my system. What firmware
>> version
>> is needed for this to work?
>>
>> And where to find it...
>>
>
> Please get testing firmware files from the following link to see if it
> works on your environment:
> https://github.com/csyuanc/linux-firmware
I do see the callback happening with that latest firmware, thanks for the link to that.
tx_gi is always reported at zero though, which seems unlikely to be correct.
[ 625.875443] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 13 rxd->seq: 12 cmd: 130022
[ 625.988751] update-tx-gi, mode: 8 tx_gi: 0
[ 625.988755] update-tx-gi, mode: 0 tx_gi: 0
[ 626.091378] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 1 rxd->seq: 15 cmd: 130022
[ 626.204917] update-tx-gi, mode: 8 tx_gi: 0
[ 626.204920] update-tx-gi, mode: 0 tx_gi: 0
[ 626.307376] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 4 rxd->seq: 3 cmd: 130022
[ 626.421332] update-tx-gi, mode: 8 tx_gi: 0
[ 626.421335] update-tx-gi, mode: 0 tx_gi: 0
[ 626.523436] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 7 rxd->seq: 6 cmd: 130022
[ 626.636900] update-tx-gi, mode: 8 tx_gi: 0
[ 626.636904] update-tx-gi, mode: 0 tx_gi: 0
[ 626.739408] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 10 rxd->seq: 9 cmd: 130022
[ 626.852727] update-tx-gi, mode: 8 tx_gi: 0
[ 626.852731] update-tx-gi, mode: 0 tx_gi: 0
[ 626.955475] mt7996e 0000:0d:00.0: ERROR: MCU: Sequence mismatch in response, seq: 13 rxd->seq: 12 cmd: 130022
diff --git a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
index 407894c13d91..8aad38a21cd4 100644
--- a/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
+++ b/drivers/net/wireless/mediatek/mt76/mt7996/mcu.c
@@ -534,6 +534,9 @@ mt7996_mcu_ie_countdown(struct mt7996_dev *dev, struct sk_buff *skb)
static int
mt7996_mcu_update_tx_gi(struct rate_info *rate, struct all_sta_trx_rate *mcu_rate)
{
+ pr_info("update-tx-gi, mode: %d tx_gi: %d\n",
+ mcu_rate->tx_mode, mcu_rate->tx_gi);
+
switch (mcu_rate->tx_mode) {
case MT_PHY_TYPE_CCK:
case MT_PHY_TYPE_OFDM:
[greearb@ben-dt5 mediatek]$
And lots of seq mismatch warnings in this firmware...could that be because of the new guard-interval
callback perhaps? Do the messages sent from FW increment the seq number? The parse-response code appears
to assume that it's responses are the only thing that would increment the seq number...
Here's my code to debug the seq mismatch:
static int
mt7996_mcu_parse_response(struct mt76_dev *mdev, int cmd,
struct sk_buff *skb, int seq)
{
struct mt7996_mcu_rxd *rxd;
struct mt7996_mcu_uni_event *event;
int mcu_cmd = FIELD_GET(__MCU_CMD_FIELD_ID, cmd);
int ret = 0;
if (!skb) {
const char* first = "Secondary";
mdev->mcu_timeouts++;
if (!mdev->first_failed_mcu_cmd)
first = "Initial";
dev_err(mdev->dev, "MCU: %s Failure: Message %08x (cid %lx ext_cid: %lx seq %d) timeout (%d/%d). Last successful cmd: 0x%x\n",
first,
cmd, FIELD_GET(__MCU_CMD_FIELD_ID, cmd),
FIELD_GET(__MCU_CMD_FIELD_EXT_ID, cmd), seq,
mdev->mcu_timeouts, MAX_MCU_TIMEOUTS,
mdev->last_successful_mcu_cmd);
if (!mdev->first_failed_mcu_cmd)
mdev->first_failed_mcu_cmd = cmd;
return -ETIMEDOUT;
}
mdev->mcu_timeouts = 0;
mdev->last_successful_mcu_cmd = cmd;
if (mdev->first_failed_mcu_cmd) {
dev_err(mdev->dev, "MCU: First success after failure: Message %08x (cid %lx ext_cid: %lx seq %d)\n",
cmd, FIELD_GET(__MCU_CMD_FIELD_ID, cmd),
FIELD_GET(__MCU_CMD_FIELD_EXT_ID, cmd), seq);
mdev->first_failed_mcu_cmd = 0;
} else {
/* verbose debugging
dev_err(mdev->dev, "MCU: OK response to message %08x (cid %lx ext_cid: %lx seq %d)\n",
cmd, FIELD_GET(__MCU_CMD_FIELD_ID, cmd),
FIELD_GET(__MCU_CMD_FIELD_EXT_ID, cmd), seq);
*/
}
rxd = (struct mt7996_mcu_rxd *)skb->data;
if (seq != rxd->seq) {
dev_err(mdev->dev, "ERROR: MCU: Sequence mismatch in response, seq: %d rxd->seq: %d cmd: %0x\n",
seq, rxd->seq, cmd);
return -EAGAIN;
}
if (cmd == MCU_CMD(PATCH_SEM_CONTROL)) {
skb_pull(skb, sizeof(*rxd) - 4);
ret = *skb->data;
} else if ((rxd->option & MCU_UNI_CMD_EVENT) &&
rxd->eid == MCU_UNI_EVENT_RESULT) {
skb_pull(skb, sizeof(*rxd));
event = (struct mt7996_mcu_uni_event *)skb->data;
ret = le32_to_cpu(event->status);
/* skip invalid event */
if (mcu_cmd != event->cid)
ret = -EAGAIN;
} else {
skb_pull(skb, sizeof(struct mt7996_mcu_rxd));
}
return ret;
}
Thanks,
Ben
--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com