panic: resource_list_add: resource entry is busy

[Hans Petter Selasky]

On 05/17/14 14:14, John Baldwin wrote:
> On 5/16/14, 12:21 PM, Hans Petter Selasky wrote:
>> Hi,
>>
>> I see the following panic:
>>
>> panic: resource_list_add: resource entry is busy
>>
>> When trying to kldload an older pccard driver. The call comes from the
>> driver_added bus method somewhere down the tree. Loading the module
>> before the kernel boots fixes the problem temporarily.
>>
>> Any bells ringing or patch suggestions?
>>
>> Seeing this on 9-stable, also believed that the same issue exists with
>> 10-current.
>
> Something is doing a duplicate resource_list_add() for the same RID.
> The trace will tell you where the second resource_list_add() is
> occurring.  Some tracing should let you see where the first one occurs.
>

Hi,

Here is the backtrace. Is this information sufficient to debug further:

KDB: stack backtrace:
#0 0xffffffff8089e5d6 at kdb_backtrace+0x66
#1 0xffffffff808960f7 at resource_list_add+0x57
#2 0xffffffff805ea939 at pccard_function_init+0x279
#3 0xffffffff805ebede at pccard_probe_and_attach_child+0x7e
#4 0xffffffff805ec6ce at pccard_driver_added+0x5e
#5 0xffffffff808951b5 at devclass_driver_added+0x75
#6 0xffffffff80895cf3 at devclass_add_driver+0x103
#7 0xffffffff8084f1e8 at module_register_init+0xa8
#8 0xffffffff80846cde at linker_load_module+0x85e
#9 0xffffffff808473f4 at kern_kldload+0xb4
#10 0xffffffff80847504 at sys_kldload+0x84
#11 0xffffffff80c4a48a at amd64_syscall+0x5ea
#12 0xffffffff80c34b27 at Xfast_syscall+0xf7

--HPS