[PATCH rtw-next v2] wifi: rtw89: usb: shorten the RX aggregation hold on RTL8922AU over USB 3

[PATCH rtw-next v2] wifi: rtw89: usb: shorten the RX aggregation hold on RTL8922AU over USB 3

[Devin Wittmayer]

The RTL8922AU holds received frames for up to a millisecond before
handing them to the host, a value carried over from the RTL8852CU.
On USB 3 that adds up to a millisecond of round trip, and under load
at 160 MHz the access point drops to lower rates and download suffers.

Shorten the hold on USB 3 only. On USB 2 the bus is the limit, and a
slow host needs the long hold to keep up.

RTL8922AU on USB 3 at 160 MHz, power save off for the round trip:

                         x86, 6 GHz    x86, 5 GHz    Pi 5, 6 GHz
                        before after  before after  before after
  TCP down, 1 flow        613  1002     579   948     587  1010  Mbps
  TCP down, 8 flows       656  1187     629  1080     605  1122  Mbps
  round trip             2.33  1.29    2.25  1.18    1.42  0.84  ms
  busiest core, UDP 200M  8.4  12.8     8.3  12.9     2.1   3.3  %

RTL8922AU on USB 2, Raspberry Pi 1 B+ at 5 GHz, 80 MHz:

                        long hold  short hold
  TCP up, 1 flow            102.3        87.7  Mbps
  CPU, UDP 20M               50.5        77.4  %
  loss, UDP 50M               3.8        85.7  %

Fixes: 32cee8a9bb66 ("wifi: rtw89: usb: Enable RX aggregation for RTL8922AU")
Reported-by: Nick Morrow <morrownr@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Devin Wittmayer <lucid_duck@justthetip.ca>
---
v2:
 - short hold on USB 3 only
 - Pi 5 column now includes a run on v2
 - dropped Ping-Ke's ack, since the patch changed
v1: https://lore.kernel.org/linux-wireless/20261006190618.258037-1-lucid_duck@justthetip.ca/

Ping-Ke, thanks for the low-end CPU point. A Pi 1 on USB 2 showed it,
second table above.

The USB 2 numbers you asked about, 6 GHz, 160 MHz:

                         x86            Pi 5
                         long  short    long  short
  TCP down, 1 flow        328    327     297    296  Mbps
  round trip             2.27   1.21    1.47   0.65  ms
  busiest core, UDP 200M  7.2   10.8     1.9    2.5  %

Small-packet mode is at the link below. It sets both the hold and the
size to 1, and nothing in that driver selects it.

https://github.com/morrownr/rtl8852cu-20251113/blob/fcb70aac55585a8e63c72da4e6d24c39d7494307/phl/hal_g6/hal_rx.c#L805-L808

 drivers/net/wireless/realtek/rtw89/usb.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/net/wireless/realtek/rtw89/usb.c b/drivers/net/wireless/realtek/rtw89/usb.c
index dbad42c17761..8dc2c93e502e 100644
--- a/drivers/net/wireless/realtek/rtw89/usb.c
+++ b/drivers/net/wireless/realtek/rtw89/usb.c
@@ -819,10 +819,15 @@ static void rtw89_usb_rx_agg_cfg_v2(struct rtw89_dev *rtwdev)
 
 static void rtw89_usb_rx_agg_cfg_v3(struct rtw89_dev *rtwdev)
 {
-	const u32 rxagg_0 = FIELD_PREP_CONST(B_BE_RXAGG_0_EN, 1) |
-			    FIELD_PREP_CONST(B_BE_RXAGG_0_NUM_TH, 255) |
-			    FIELD_PREP_CONST(B_BE_RXAGG_0_TIME_32US_TH, 32) |
-			    FIELD_PREP_CONST(B_BE_RXAGG_0_BUF_SZ_1K, 20);
+	struct rtw89_usb *rtwusb = rtw89_usb_priv(rtwdev);
+	u32 rxagg_0 = FIELD_PREP_CONST(B_BE_RXAGG_0_EN, 1) |
+		      FIELD_PREP_CONST(B_BE_RXAGG_0_NUM_TH, 255) |
+		      FIELD_PREP_CONST(B_BE_RXAGG_0_BUF_SZ_1K, 20);
+
+	if (rtwusb->udev->speed == USB_SPEED_SUPER)
+		rxagg_0 |= FIELD_PREP_CONST(B_BE_RXAGG_0_TIME_32US_TH, 1);
+	else
+		rxagg_0 |= FIELD_PREP_CONST(B_BE_RXAGG_0_TIME_32US_TH, 32);
 
 	rtw89_write32(rtwdev, R_BE_RXAGG_0_V1, rxagg_0);
 	rtw89_write32(rtwdev, R_BE_RXAGG_1_V1, 0x1F);
-- 
2.55.0

RE: [PATCH rtw-next v2] wifi: rtw89: usb: shorten the RX aggregation hold on RTL8922AU over USB 3

[Ping-Ke Shih]

Devin Wittmayer <lucid_duck@justthetip.ca> wrote:
> The RTL8922AU holds received frames for up to a millisecond before
> handing them to the host, a value carried over from the RTL8852CU.
> On USB 3 that adds up to a millisecond of round trip, and under load
> at 160 MHz the access point drops to lower rates and download suffers.
> 
> Shorten the hold on USB 3 only. On USB 2 the bus is the limit, and a
> slow host needs the long hold to keep up.
> 
> RTL8922AU on USB 3 at 160 MHz, power save off for the round trip:
> 
>                          x86, 6 GHz    x86, 5 GHz    Pi 5, 6 GHz
>                         before after  before after  before after
>   TCP down, 1 flow        613  1002     579   948     587  1010  Mbps
>   TCP down, 8 flows       656  1187     629  1080     605  1122  Mbps
>   round trip             2.33  1.29    2.25  1.18    1.42  0.84  ms
>   busiest core, UDP 200M  8.4  12.8     8.3  12.9     2.1   3.3  %
> 
> RTL8922AU on USB 2, Raspberry Pi 1 B+ at 5 GHz, 80 MHz:
> 
>                         long hold  short hold
>   TCP up, 1 flow            102.3        87.7  Mbps
>   CPU, UDP 20M               50.5        77.4  %
>   loss, UDP 50M               3.8        85.7  %
> 
> Fixes: 32cee8a9bb66 ("wifi: rtw89: usb: Enable RX aggregation for RTL8922AU")
> Reported-by: Nick Morrow <morrownr@gmail.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Devin Wittmayer <lucid_duck@justthetip.ca>

Acked-by: Ping-Ke Shih <pkshih@realtek.com>

> 
> Small-packet mode is at the link below. It sets both the hold and the
> size to 1, and nothing in that driver selects it.
> 
> https://github.com/morrownr/rtl8852cu-20251113/blob/fcb70aac55585a8e63c72da4e6d24c39d7494307/phl/h
> al_g6/hal_rx.c#L805-L808

Thanks for the info. I quickly check the internal commit log that people want
dynamically select the modes depends on RX conditions. I think the actually
selection might be only existing in special branches. 

Let's take this patch in advance. If you or other people can fine tune 
the threshold by certain conditions dynamically, we can apply the patch
in the future.

Ping-Ke

Re: [PATCH rtw-next v2] wifi: rtw89: usb: shorten the RX aggregation hold on RTL8922AU over USB 3

[Ping-Ke Shih]

Devin Wittmayer <lucid_duck@justthetip.ca> wrote:

> The RTL8922AU holds received frames for up to a millisecond before
> handing them to the host, a value carried over from the RTL8852CU.
> On USB 3 that adds up to a millisecond of round trip, and under load
> at 160 MHz the access point drops to lower rates and download suffers.
> 
> Shorten the hold on USB 3 only. On USB 2 the bus is the limit, and a
> slow host needs the long hold to keep up.
> 
> RTL8922AU on USB 3 at 160 MHz, power save off for the round trip:
> 
>                          x86, 6 GHz    x86, 5 GHz    Pi 5, 6 GHz
>                         before after  before after  before after
>   TCP down, 1 flow        613  1002     579   948     587  1010  Mbps
>   TCP down, 8 flows       656  1187     629  1080     605  1122  Mbps
>   round trip             2.33  1.29    2.25  1.18    1.42  0.84  ms
>   busiest core, UDP 200M  8.4  12.8     8.3  12.9     2.1   3.3  %
> 
> RTL8922AU on USB 2, Raspberry Pi 1 B+ at 5 GHz, 80 MHz:
> 
>                         long hold  short hold
>   TCP up, 1 flow            102.3        87.7  Mbps
>   CPU, UDP 20M               50.5        77.4  %
>   loss, UDP 50M               3.8        85.7  %
> 
> Fixes: 32cee8a9bb66 ("wifi: rtw89: usb: Enable RX aggregation for RTL8922AU")
> Reported-by: Nick Morrow <morrownr@gmail.com>
> Cc: stable@vger.kernel.org
> Signed-off-by: Devin Wittmayer <lucid_duck@justthetip.ca>
> Acked-by: Ping-Ke Shih <pkshih@realtek.com>

1 patch(es) applied to rtw-next branch of rtw.git, thanks.

83de3a16c7b3 wifi: rtw89: usb: shorten the RX aggregation hold on RTL8922AU over USB 3

---
https://github.com/pkshih/rtw.git