Re: [PATCH] wifi: rtl8xxxu: fixing IQK failures for rtl8192eu

From: Jun ASAKA
Date: Thu Dec 01 2022 - 09:27:10 EST


On 01/12/2022 9:09 pm, Bitterblue Smith wrote:

On 01/12/2022 04:18, Ping-Ke Shih wrote:
-----Original Message-----
From: Jun ASAKA <JunASAKA@xxxxxxxxxxxxx>
Sent: Thursday, December 1, 2022 9:39 AM
To: Ping-Ke Shih <pkshih@xxxxxxxxxxx>; Jes.Sorensen@xxxxxxxxx
Cc: kvalo@xxxxxxxxxx; davem@xxxxxxxxxxxxx; edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx;
linux-wireless@xxxxxxxxxxxxxxx; netdev@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
Subject: Re: [PATCH] wifi: rtl8xxxu: fixing IQK failures for rtl8192eu

On 01/12/2022 8:54 am, Ping-Ke Shih wrote:

-----Original Message-----
From: JunASAKA <JunASAKA@xxxxxxxxxxxxx>
Sent: Wednesday, November 30, 2022 10:09 PM
To: Jes.Sorensen@xxxxxxxxx
Cc: kvalo@xxxxxxxxxx; davem@xxxxxxxxxxxxx; edumazet@xxxxxxxxxx; kuba@xxxxxxxxxx; pabeni@xxxxxxxxxx;
linux-wireless@xxxxxxxxxxxxxxx; netdev@xxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx; JunASAKA
<JunASAKA@xxxxxxxxxxxxx>
Subject: [PATCH] wifi: rtl8xxxu: fixing IQK failures for rtl8192eu

Fixing "Path A RX IQK failed" and "Path B RX IQK failed"
issues for rtl8192eu chips by replacing the arguments with
the ones in the updated official driver.
I think it would be better if you can point out which version you use, and
people will not modify them back to old version suddenly.

Signed-off-by: JunASAKA <JunASAKA@xxxxxxxxxxxxx>
---
.../realtek/rtl8xxxu/rtl8xxxu_8192e.c | 76 +++++++++++++------
1 file changed, 54 insertions(+), 22 deletions(-)

diff --git a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
index b06508d0cd..82346500f2 100644
--- a/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
+++ b/drivers/net/wireless/realtek/rtl8xxxu/rtl8xxxu_8192e.c
[...]

@@ -891,22 +907,28 @@ static int rtl8192eu_iqk_path_b(struct rtl8xxxu_priv *priv)

rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x00000000);
rtl8xxxu_write_rfreg(priv, RF_B, RF6052_REG_UNKNOWN_DF, 0x00180);
- rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x80800000);

- rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x00000000);
+ rtl8xxxu_write_rfreg(priv, RF_B, RF6052_REG_WE_LUT, 0x800a0);
+ rtl8xxxu_write_rfreg(priv, RF_B, RF6052_REG_RCK_OS, 0x20000);
+ rtl8xxxu_write_rfreg(priv, RF_B, RF6052_REG_TXPA_G1, 0x0000f);
+ rtl8xxxu_write_rfreg(priv, RF_B, RF6052_REG_TXPA_G2, 0x07f77);
+
rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x80800000);

+ // rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x00000000);
+ // rtl8xxxu_write32(priv, REG_FPGA0_IQK, 0x80800000);
+
I think this is a test code of vendor driver. No need them here.


/* Path B IQK setting */
rtl8xxxu_write32(priv, REG_TX_IQK_TONE_A, 0x38008c1c);
rtl8xxxu_write32(priv, REG_RX_IQK_TONE_A, 0x38008c1c);
rtl8xxxu_write32(priv, REG_TX_IQK_TONE_B, 0x18008c1c);
rtl8xxxu_write32(priv, REG_RX_IQK_TONE_B, 0x38008c1c);

- rtl8xxxu_write32(priv, REG_TX_IQK_PI_B, 0x821403e2);
+ rtl8xxxu_write32(priv, REG_TX_IQK_PI_B, 0x82140303);
rtl8xxxu_write32(priv, REG_RX_IQK_PI_B, 0x68160000);

/* LO calibration setting */
- rtl8xxxu_write32(priv, REG_IQK_AGC_RSP, 0x00492911);
+ rtl8xxxu_write32(priv, REG_IQK_AGC_RSP, 0x00462911);

/* One shot, path A LOK & IQK */
rtl8xxxu_write32(priv, REG_IQK_AGC_PTS, 0xfa000000);
[...]

I have compared your patch with internal code, and they are the same.
But, I don't have a test.

Ping-Ke
I changed those arguments into the ones here:
https://github.com/Mange/rtl8192eu-linux-driver which works fine with my
rtl8192eu wifi dongle. But forgive my ignorant that I don't have enough
experience on wifi drivers, I just compared those two drivers and
figured that those codes fixing my IQK failures.
I do similar things as well. :-)

The github repository mentioned
"This branch is based on Realtek's driver versioned 4.4.1. master is based on 4.3.1.1 originally."
So, we can add something to commit message:
1. https://github.com/Mange/rtl8192eu-linux-driver
2. vendor driver version: 4.3.1.1

--
Ping-Ke

That repo is confusing, unfortunately. Indeed, the "master" branch seems to
contain v4.3.1.1_11320.20140505. But the last commit is from 2017.

The "realtek-4.4.x" branch is the one being actively maintained, and at some
point it was updated to v5.6.4_35685.20191108_COEX20171113-0047. README.md
was forgotten.

Well, it is and I am using the branch "realtek-4.4.x" for this patch that my problems has fixed correctly. But I has issued a v2 patch added "vendor driver version: 4.3.1.1" which may be incorrect as you mentioned. Is that means I need issue a v3 patch to change the commit messages to "version 5.6.4"? I can do it right now.


Jun ASAKA.