3.11.0-1 (latest) and the 1wire bus

[crashmaster]

Hello,



this version runs on most of my raspberries without problems. Two of them have DS18B20 temperature sensors using the 1wire bus (modules w1_therm, w1_gpio). Both work fine with the previous but latest kernel version (3.10.9-1).

EDIT:
---------------------------------------------------------------------
Although one boots and the other hangs during boot when loading the w1 modules, the temperature can't be read anymore.

My following description is wrong, generally w1 seems not to work correctly.
---------------------------------------------------------------------

Although both have the same configuration (i checked it twice and used default settings), curiously one of both hangs at bootup (while the other works fine). The only difference is: the working one is a B model using LAN and the not working a A model using WLAN but that shouldn't be a reason.
The reason for hanging is: "systemd-modules-load.service" never returns. So i masked it and system boots without loading the w1 modules. I loaded them manually but it didn't get the sensor device. Switching back to the previous kernel version all works fine again.

Just to tell my experience

Greetings, CM

[moonman]

Thanks for the reports crashmaster, but the best thing you can do is report these upstream. https://github.com/raspberrypi/linux/issues

[crashmaster]

Thanks for the info moonman, that's also the way to go for problems with future releases?

Until now i didn't know about (additional) patches from yourselve or others so i didn't use upstream for feedback to avoid confusion.

Greets, CM

[moonman]

The only patch I had included myself is the bfq i/o scheduler and the kernel config is different from the standard one upstream https://github.com/archlinuxarm/PKGBUIL ... ypi-latest but i don't think problems like this would be caused by these differences, though it is possible.

[trilader]

Hi there,

I have the same problem on my Pi. For me the Pi hangs after I do modprobe w1_therm.

Here is the relevant part of the journalctl output. I also posted this in the issue at github.

$this->bbcode_second_pass_code('', 'Sep 29 11:25:24 redpi kernel: INFO: task modprobe:610 blocked for more than 120 seconds.
Sep 29 11:25:24 redpi kernel: "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Sep 29 11:25:24 redpi kernel: modprobe D c068f5fc 0 610 609 0x00000000
Sep 29 11:25:24 redpi kernel: [<c068f5fc>] (__schedule+0x284/0x5c0) from [<c0690028>] (schedule_preempt_disabled+0x14/0x20)
Sep 29 11:25:24 redpi kernel: [<c0690028>] (schedule_preempt_disabled+0x14/0x20) from [<c068ec18>] (__mutex_lock_slowpath+0x9c/0x12c)
Sep 29 11:25:24 redpi kernel: [<c068ec18>] (__mutex_lock_slowpath+0x9c/0x12c) from [<bf0e2008>] (w1_reconnect_slaves+0x3c/0xf8 [wire])
Sep 29 11:25:24 redpi kernel: [<bf0e2008>] (w1_reconnect_slaves+0x3c/0xf8 [wire]) from [<bf0e2ac0>] (w1_register_family+0x9c/0xb8 [wire])
Sep 29 11:25:24 redpi kernel: [<bf0e2ac0>] (w1_register_family+0x9c/0xb8 [wire]) from [<bf0f401c>] (w1_therm_init+0x1c/0x3c [w1_therm])
Sep 29 11:25:24 redpi kernel: [<bf0f401c>] (w1_therm_init+0x1c/0x3c [w1_therm]) from [<c00085d4>] (do_one_initcall+0x108/0x170)
Sep 29 11:25:24 redpi kernel: [<c00085d4>] (do_one_initcall+0x108/0x170) from [<c006de78>] (load_module+0x16e0/0x1e8c)
Sep 29 11:25:24 redpi kernel: [<c006de78>] (load_module+0x16e0/0x1e8c) from [<c006e6cc>] (SyS_init_module+0xa8/0xec)
Sep 29 11:25:24 redpi kernel: [<c006e6cc>] (SyS_init_module+0xa8/0xec) from [<c000e1a0>] (ret_fast_syscall+0x0/0x30)')

[crashmaster]

I have a workaround now: Load "w1-therm" before "w1-gpio".

I tested it on 2 RPis and changed the order in my conf file in /etc/modules-load.d/ - it works.