Hello Adrian,
I've prepared the attached patch, can you check if the situation improved with
it?
Best wishes
David
On 11/5/19 5:20 PM, David Bauer wrote:
> Hello Adrian,
>
> On 11/5/19 5:14 PM, Adrian Schmutzler wrote:
>> Hi David,
>>
>> thanks for your response.
>>
>> To me it looks like qca953x already uses 25 MHz clock, or am I looking at
>> the wrong value:
>
> Yes, however ath9k does not use this value but tries to determine the
> reference clock
> based on the bootstrap bit (see first link in my previous E-Mail), so the
> value
> from the device tree is never used for ath9k.
>
> Best wishes
> David
>
>>
>> https://github.com/openwrt/openwrt/blob/master/target/linux/ath79/dts/qca953x.dtsi#L27
>>
>> Best
>>
>> Adrian
>>
>>
>> On 5 November 2019 16:46:59 CET, David Bauer <m...@david-bauer.net> wrote:
>>
>> Hello Adrian,
>>
>> during the CPE210v2 bringup it was discovered that the CPE210 has the
>> wrong bootstrap option set
>> for it's 25 MHz reference clock. Because of this, the device was
>> originally not even booting with ar71xx.
>>
>> On ath79, the reference clock is not detected based on the bootstrap
>> option, but set by the DTS.
>> The twist however is the ath9k driver, whose OF patch still reads this
>> register. [0]
>>
>> On ar71xx, the platform data was always set to true for the QCA9533 [1]
>>
>> So you can try to force the settings for 25MHz reference clock for all
>> QCA953x regardless of the bootstrap
>> settings to keep the behavior in line with ar71xx.
>>
>> I have no device to verify this, however it's a good candidate for the
>> root cause. ;)
>>
>> [0]
>> https://github.com/openwrt/openwrt/blob/master/package/kernel/mac80211/patches/ath/552-ahb_of.patch#L237
>> [1]
>> https://github.com/openwrt/openwrt/blob/master/target/linux/ar71xx/patches-4.14/620-MIPS-ath79-add-support-for-QCA953x-SoC.patch#L260
>>
>> Best wishes
>> David
>>
>> On 11/5/19 3:05 PM, Adrian Schmutzler wrote:
>>
>> Hi,
>>
>> for quite some time already we are struggling with broken WiFi on
>> some TP-Link CPE devices having QCA9533 rev. 2 (QCA9533-BL3A SOC) in common.
>>
>> I'd be happy on some help here, since I've exhausted my debugging
>> capabilities.
>>
>>
>> 1. Symptoms: WiFi looks up on the device, some TX traffic is shown
>> in ifconfig, RX is zero. The AP cannot be found by clients. "iw dev wlan0
>> scan" returns nothing.
>>
>> 2. Affected devices: TP-Link CPE210 v2/v3, CPE220 v3 (all QCA9533
>> rev. 2?); no other QCA9533 devices known to be affected (specific to CPE or
>> to QCA9533 rev. 2?)
>>
>> 3. For a certain model, there are devices which are working
>> correctly and others which don't. There is no known indicator to find out
>> whether a device works or not. The state of a device does not change as far
>> as we know (always working or never working).
>>
>> 4. So far, only 2.4 GHZ-only devices were affected
>>
>> 5. There is no diagnostic output that indicates a WiFi problem.
>> dmesg/logread look normal, there is no difference when compared between
>> working and not-working devices (despite RX=0/scan as stated above)
>>
>> 6. The problem seems to be present from the beginning (device
>> support patches), it just has been overlooked since it's not occurring on
>> every device.
>>
>> 7. The ar71xx firmware for all devices works flawlessly, so it is an
>> ath79-specific problem.
>>
>>
>> Other findings that might be connected or not:
>>
>> a. On ath79 phy0 uses irq=11/irq=12 and on ar71xx irq=47. eth0 uses
>> irq=4 on both targets.
>>
>> b. The following gpio is only found on ar71xx: gpio-495 (
>> |ath9k-phy0 ) in lo
>>
>>
>> I currently own a CPE210v2 with the bug and can test suggestions (if
>> I'm capable of implementing them).
>> There is a device support PR for the CPE220 v3 suffering from the
>> same problem: https://github.com/openwrt/openwrt/pull/2514
>>
>> Despite, further reading may be found in forum discussion and bug
>> report:
>> https://bugs.openwrt.org/index.php?do=details&task_id=2333
>> https://bugs.openwrt.org/index.php?do=details&task_id=2478
>>
>> https://forum.openwrt.org/t/ath79-tp-link-cpe210-v2-0-wifi-not-working/40666
>>
>> Initial support for CPE210 v2/v3 was done by me and bluelineXY, both
>> already involved in the discussion. ;-)
>>
>> Thanks for any hints!
>>
>> Adrian
>>
>> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>> openwrt-devel mailing list
>> openwrt-devel@lists.openwrt.org
>> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>>
> _______________________________________________
> openwrt-devel mailing list
> openwrt-devel@lists.openwrt.org
> https://lists.openwrt.org/mailman/listinfo/openwrt-devel
>
>From b4451e9975faa66919a056080c897d327eec1e79 Mon Sep 17 00:00:00 2001
From: David Bauer <m...@david-bauer.net>
Date: Tue, 5 Nov 2019 22:28:39 +0100
Subject: [PATCH] mac80211 ath9k: force QCA953x clock to 25MHz
The QCA953x only supports 25 MHz refclk, however some OEMs set an
invalid bootstrap value for the REF_CLK option, which would break the
clock detection in ath9k.
Force the QCA953x refclk to 25MHz in ath9k, as this is (according to the
datasheet) the only valid frequency.
Signed-off-by: David Bauer <m...@david-bauer.net>
---
.../mac80211/patches/ath/552-ahb_of.patch | 19 +++++++++++++------
1 file changed, 13 insertions(+), 6 deletions(-)
diff --git a/package/kernel/mac80211/patches/ath/552-ahb_of.patch b/package/kernel/mac80211/patches/ath/552-ahb_of.patch
index 1170fc64bd..bbb29fc5b1 100644
--- a/package/kernel/mac80211/patches/ath/552-ahb_of.patch
+++ b/package/kernel/mac80211/patches/ath/552-ahb_of.patch
@@ -16,7 +16,7 @@
static const struct platform_device_id ath9k_platform_id_table[] = {
{
-@@ -69,6 +77,235 @@ static const struct ath_bus_ops ath_ahb_
+@@ -69,6 +77,242 @@ static const struct ath_bus_ops ath_ahb_
.eeprom_read = ath_ahb_eeprom_read,
};
@@ -232,7 +232,14 @@
+ pdata->external_reset = data->wmac_reset;
+ }
+
-+ if (data->bootstrap_reg && data->bootstrap_ref) {
++ if (data->dev_id == AR9300_DEVID_AR953X) {
++ /*
++ * QCA953x only supports 25MHz refclk.
++ * Some vencors have an invalid bootstrap option
++ * set, which would break the WMAC here.
++ */
++ pdata->is_clk_25mhz = true;
++ } else if (data->bootstrap_reg && data->bootstrap_ref) {
+ u32 t = ath79_reset_rr(data->bootstrap_reg);
+ if (t & data->bootstrap_ref)
+ pdata->is_clk_25mhz = false;
@@ -252,7 +259,7 @@
static int ath_ahb_probe(struct platform_device *pdev)
{
void __iomem *mem;
-@@ -80,6 +317,17 @@ static int ath_ahb_probe(struct platform
+@@ -80,6 +324,17 @@ static int ath_ahb_probe(struct platform
int ret = 0;
struct ath_hw *ah;
char hw_name[64];
@@ -270,7 +277,7 @@
if (!dev_get_platdata(&pdev->dev)) {
dev_err(&pdev->dev, "no platform data specified\n");
-@@ -122,13 +370,16 @@ static int ath_ahb_probe(struct platform
+@@ -122,13 +377,16 @@ static int ath_ahb_probe(struct platform
sc->mem = mem;
sc->irq = irq;
@@ -288,7 +295,7 @@
if (ret) {
dev_err(&pdev->dev, "failed to initialize device\n");
goto err_irq;
-@@ -159,6 +410,9 @@ static int ath_ahb_remove(struct platfor
+@@ -159,6 +417,9 @@ static int ath_ahb_remove(struct platfor
free_irq(sc->irq, sc);
ieee80211_free_hw(sc->hw);
}
@@ -298,7 +305,7 @@
return 0;
}
-@@ -168,6 +422,9 @@ static struct platform_driver ath_ahb_dr
+@@ -168,6 +429,9 @@ static struct platform_driver ath_ahb_dr
.remove = ath_ahb_remove,
.driver = {
.name = "ath9k",
--
2.23.0
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
