[linux-yocto] Fwd: [PATCH] serial: pch_uart: no need to disable irq for -rt kernel

[Bruce Ashfield]

On 12-11-21 8:46 PM, Liang Li wrote:
> On 2012-11-22 00:06, Darren Hart<dvhart at linux.intel.com>  wrote:
>>
>>
>> On 11/21/2012 07:53 AM, Bruce Ashfield wrote:
>>> On 12-11-21 10:46 AM, Bruce Ashfield wrote:
>>>> Looping this via the linux-yocto mailing list as well, since I plan to
>>>> merge this to the 3.4 yocto kernel (with a slightly tweaked header),
>>>> which shares the same base and will have the same problem.
>>>>
>>>> Darren: does this look fine to you ?
>>>
>>> Cancel this. I *knew* the patch looked familiar .. I already have
>>> this fix .. from Darren :)
>>
>> Right, that. ;-)
>>
>
> I did not see the patch on the linux-yocto mailing list .. so I guess
> you might just send it to Bruce directly.

I already had it .. just only on the fri2 branch, where other BSPs
couldn't take advantage of it .. which is why I originally missed it
when I saw yours (I checked the base -rt branch).

We are all good for linux-yocto and derived kernels now.

Bruce

>
> Thanks.
>
> 	Liang Li
>
>> --
>> Darren
>>
>>>
>>> Cheers,
>>>
>>> Bruce
>>>
>>>>
>>>> Cheers,
>>>>
>>>> Bruce
>>>>
>>>>
>>>> -------- Original Message --------
>>>> Subject: [PATCH] serial: pch_uart: no need to disable irq for -rt kernel
>>>> Date: Wed, 21 Nov 2012 16:23:30 +0800
>>>> From: Liang Li<liang.li at windriver.com>
>>>>
>>>> Fix the following call trace when ttyPCH* is used as serial console:
>>>>
>>>> BUG: sleeping function called from invalid context at
>>>> ../linux/kernel/rtmutex.c:646
>>>>
>>>> in_atomic(): 0, irqs_disabled(): 1, pid: 12, name: kworker/0:1
>>>> Pid: 12, comm: kworker/0:1 Tainted: G W
>>>> 3.4.16-rt17-preempt-rt #1
>>>> Call Trace:
>>>> [<c1057236>] __might_sleep+0xd6/0x100
>>>> [<c163892c>] rt_spin_lock+0x1c/0x40
>>>> [<c1351c18>] pch_console_write+0x58/0x1c0
>>>> [<c102a4cb>] __call_console_drivers+0x7b/0xa0
>>>> [<c102a653>] _call_console_drivers+0x73/0x120
>>>> [<c163c51d>] ? sub_preempt_count+0x7d/0xb0
>>>> [<c102a9c1>] console_unlock+0xf1/0x250
>>>> [<c1346179>] console_callback+0x79/0x120
>>>> [<c1045177>] process_one_work+0x107/0x410
>>>> [<c1346100>] ? poke_blanked_console+0xc0/0xc0
>>>> [<c1346100>] ? poke_blanked_console+0xc0/0xc0
>>>> [<c10459c5>] worker_thread+0x135/0x2e0
>>>> [<c1045890>] ? manage_workers.isra.26+0x200/0x200
>>>> [<c1045890>] ? manage_workers.isra.26+0x200/0x200
>>>> [<c104a5e3>] kthread+0x73/0x80
>>>> [<c104a570>] ? __init_kthread_worker+0x40/0x40
>>>> [<c163f9fa>] kernel_thread_helper+0x6/0x10
>>>> BUG: sleeping function called from invalid context at ...
>>>>
>>>> Signed-off-by: Liang Li<liang.li at windriver.com>
>>>> ---
>>>>    drivers/tty/serial/pch_uart.c | 4 ++--
>>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/tty/serial/pch_uart.c b/drivers/tty/serial/pch_uart.c
>>>> index 7d47514..604fcc0 100644
>>>> --- a/drivers/tty/serial/pch_uart.c
>>>> +++ b/drivers/tty/serial/pch_uart.c
>>>> @@ -1551,7 +1551,7 @@ pch_console_write(struct console *co, const char
>>>> *s, unsigned int count)
>>>>
>>>>        touch_nmi_watchdog();
>>>>
>>>> -    local_irq_save(flags);
>>>> +    local_irq_save_nort(flags);
>>>>        if (priv->port.sysrq) {
>>>>            spin_lock(&priv->lock);
>>>>            /* serial8250_handle_port() already took the port lock */
>>>> @@ -1584,7 +1584,7 @@ pch_console_write(struct console *co, const char
>>>> *s, unsigned int count)
>>>>            spin_unlock(&priv->port.lock);
>>>>        if (priv_locked)
>>>>            spin_unlock(&priv->lock);
>>>> -    local_irq_restore(flags);
>>>> +    local_irq_restore_nort(flags);
>>>>    }
>>>>
>>>>    static int __init pch_console_setup(struct console *co, char *options)
>>>
>>
>> --
>> Darren Hart
>> Intel Open Source Technology Center
>> Yocto Project - Technical Lead - Linux Kernel