Re: WARNING and PANIC in irq_matrix_free

[Song Liu]

Adding Broadcom developers to the thread.

Hi Michael, Vasundhara, and Andy,

Could you please help look into the case I found? ethtool crashes the system
for both net/master and net-next/master.

Thanks,
Song

On Fri, May 25, 2018 at 1:10 PM, Song Liu <liu.song.a23@xxxxxxxxx> wrote:
> Hi,
>
> We are seeing something probably related.
>
> We run ethtool on a system with Broadcom NIC to increase number of
> combined queues.
>
>
>     [root@ ~]# ethtool -l eth0
>     Channel parameters for eth0:
>     Pre-set maximums:
>     RX:             9
>     TX:             8
>     Other:          0
>     Combined:       17
>     Current hardware settings:
>     RX:             0
>     TX:             0
>     Other:          0
>     Combined:       8
>
>     [root@ ~]# ethtool -L eth0 combined 16
>
> The last command PANIC the kernel easily (5 out of 5 in my tests).
>
> I haven't got luck to catch much console output, the only line I got is:
>
>     [ 504.727865] WARNING: CPU: 10 PID: 0 at kernel/irq/matrix.c:371
> irq_matrix_free+0x32/0xd0
>
> The NIC we have is
>
>     Broadcom Limited BCM57302 NetXtreme-C 10Gb/25Gb Ethernet Controller
>
> Thanks,
> Song
>
> On Wed, May 23, 2018 at 1:49 AM, Thomas Gleixner <tglx@xxxxxxxxxxxxx> wrote:
>> On Wed, 23 May 2018, Tariq Toukan wrote:
>>> On 19/05/2018 2:20 PM, Thomas Gleixner wrote:
>>> > On Fri, 18 May 2018, Dmitry Safonov wrote:
>>> > > I'm not entirely sure that it's the same fault, but at least backtrace
>>> > > looks resembling.
>>> >
>>> > Yes, it's similar, but not the same issue. I'll stare are the code ...
>>> >
>>> > Thanks,
>>> >
>>> >     tglx
>>> >
>>>
>>> We still see the issue in our daily regression runs.
>>> I have your patch merged into my internal branch, it prints the following:
>>>
>>> [ 4898.226258] Trying to clear prev_vector: 0
>>> [ 4898.226439] Trying to clear prev_vector: 0
>>>
>>> i.e. vector(0) is lower than FIRST_EXTERNAL_VECTOR.
>>
>> Could you please enable the vector and irq matrix trace points and capture
>> the trace when this happens?
>>
>> Thanks,
>>
>>         tglx
>>