How to suppress rtk_btcoex messages in journal

MilanKnizek · 2021-03-25 20:24:58

Hi,

I have got a new Bluetooth 5 usb dongle 0bda:8771 Realtek Semiconductor Corp. Bluetooth Radio and got it working with the help of AUR package rtl8761usb

$ lsmod | grep rtk_
rtk_btusb              65536  0
bluetooth             749568  55 bnep,rtk_btusb,rfcomm

btusb module is blacklisted.

Generally, it is working, however during the playback the journal is full of

Mar 25 20:56:53 athlon kernel: rtk_btcoex: count_a2dp_packet_timeout: a2dp_packet_count 127
Mar 25 20:56:54 athlon kernel: rtk_btcoex: count_a2dp_packet_timeout: a2dp_packet_count 126
Mar 25 20:56:55 athlon kernel: rtk_btcoex: count_a2dp_packet_timeout: a2dp_packet_count 127
Mar 25 20:56:56 athlon kernel: rtk_btcoex: count_a2dp_packet_timeout: a2dp_packet_count 65

That represents cca 90 % of all messages. The issue occurs on both USB 2.0 and 3.0 ports (without hub).

I can see that the messages are produced by usb/bluetooth_usb_driver/rtk_coex.c

#if LINUX_VERSION_CODE > KERNEL_VERSION(4, 14, 0)
static void count_a2dp_packet_timeout(struct timer_list *unused)
#else
static void count_a2dp_packet_timeout(unsigned long data)
#endif
{
    if (btrtl_coex.a2dp_packet_count)
        RTKBT_DBG("%s: a2dp_packet_count %d", __func__,
              btrtl_coex.a2dp_packet_count);
    if (btrtl_coex.a2dp_packet_count == 0) {
        if (is_profile_busy(profile_a2dp)) {
            RTKBT_DBG("%s: a2dp busy->idle!", __func__);
            update_profile_state(profile_a2dp, FALSE);
            if (btrtl_coex.profile_bitmap & BIT(profile_sink))
                update_profile_state(profile_sink, FALSE);
        }
    }

Although I believe it would be great to solve the timeout problems, yet, is there a way how to suppress the RTKBT_DBG messages? Either realtime, via module config or even by recompiling the module?

loqs · 2021-03-25 20:43:17

Try applying this diff to the package:

diff --git a/20201202_LINUX_BT_DRIVER/usb/bluetooth_usb_driver/rtk_misc.h b/20201202_LINUX_BT_DRIVER/usb/bluetooth_usb_driver/rtk_misc.h
index dbc5690..9fb1163 100644
--- a/20201202_LINUX_BT_DRIVER/usb/bluetooth_usb_driver/rtk_misc.h
+++ b/20201202_LINUX_BT_DRIVER/usb/bluetooth_usb_driver/rtk_misc.h
@@ -37,11 +37,13 @@
 /* #define RTKBT_TV_POWERON_WHITELIST */
 
 #if 1
-#define RTKBT_DBG(fmt, arg...) printk(KERN_INFO "rtk_btusb: " fmt "\n" , ## arg)
+#define RTKBT_DBG(fmt, arg...) printk(KERN_DEBUG "rtk_btusb: " fmt "\n" , ## arg)
 #define RTKBT_INFO(fmt, arg...) printk(KERN_INFO "rtk_btusb: " fmt "\n" , ## arg)
 #define RTKBT_WARN(fmt, arg...) printk(KERN_WARNING "rtk_btusb: " fmt "\n", ## arg)
 #else
 #define RTKBT_DBG(fmt, arg...)
+#define RTKBT_INFO(fmt, arg...)
+#define RTKBT_WARN(fmt, arg...)
 #endif
 
 #if 1

Edit:
Updated patch to use log level KERN_DEBUG instead of disabling the message. By default debug messages are disabled anyway.

Last edited by loqs (2021-03-25 21:15:55)

MilanKnizek · 2021-03-25 22:05:48

Thanks for the quick comment and effort. From some reason, the RTKBT_DBG macro still produced the full error message, despite the log level being changed to KERN_DEBUG from the original KERN_INFO - as per you patch.

My kernel log level is:

# cat /proc/sys/kernel/printk
1	4	1	4

So, the DEBUG messages should not show. They did.

Eventually, I ended up commenting the part of the code directly:

--- rtk_coex.c  2020-10-20 13:52:08.000000000 +0200
+++ rtk_coex.c_1    2021-03-25 22:26:30.158850666 +0100
@@ -1046,9 +1046,9 @@
 static void count_a2dp_packet_timeout(unsigned long data)
 #endif
 {
-   if (btrtl_coex.a2dp_packet_count)
-       RTKBT_DBG("%s: a2dp_packet_count %d", __func__,
-             btrtl_coex.a2dp_packet_count);
+   //if (btrtl_coex.a2dp_packet_count)
+       //RTKBT_DBG("%s: a2dp_packet_count %d", __func__,
+       //    btrtl_coex.a2dp_packet_count);
    if (btrtl_coex.a2dp_packet_count == 0) {
        if (is_profile_busy(profile_a2dp)) {
            RTKBT_DBG("%s: a2dp busy->idle!", __func__);

My journal is now happier :-)

loqs · 2021-03-25 22:19:09

Printk filtering only effects what is sent to the console but it is surprising debug messages are being emitted by the kernel.
Glad you fixed the issue anyway.
Edit:
Should have used pr_debug / dev_debug which would only be enabled by dynamic debugging instead of just changing the log level. Oh well fixed now.

Last edited by loqs (2021-03-25 22:38:16)

Arch Linux

#1 2021-03-25 20:24:58

How to suppress rtk_btcoex messages in journal

#2 2021-03-25 20:43:17

Re: How to suppress rtk_btcoex messages in journal

#3 2021-03-25 22:05:48

Re: How to suppress rtk_btcoex messages in journal

#4 2021-03-25 22:19:09

Re: How to suppress rtk_btcoex messages in journal

Board footer