Re: [RFC] Fix BUG splat during nubus_driver_register()

[Michael Schmitz]

Hi Finn,

Am 03.05.18 um 12:52 schrieb Finn Thain:
>> No need for a Mac test in nubus_proc_init() as it appears to work just 
>> fine.
>>
> Yes. I suggested a MACH_IS_MAC() test because zorro_proc_init() has a 
> similar test. The postcore_initcall() approach retains the MACH_IS_MAC() 
> test, so as to bypass nubus_proc_init(). Either way is fine with me.

OK, that'll eliminate /proc/nubus, and legacy software won't know about
/sys/nubus anyway.
>> Are we confusing any legacy user space tools that might check for 
>> presence of /proc/nubus with this patch?
> I doubt it.

Me too. But I've been wrong before.

>> Do you want an alternative patch tested (MACH_IS_MAC() test in 
>> nubus_driver_register() instead)?
>>
> The second patch might be worth testing, but I would wait to hear from 
> Geert before I did so.
>

Fair enough - your patch fixes the issue, that's what matters.

FWIW:

Tested-by: Michael Schmitz 

--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [RFC] Fix BUG splat during nubus_driver_register()

[Finn Thain]

On Thu, 3 May 2018, Michael Schmitz wrote:

> Hi Finn,
> 
> with your patch, 4.17rc2 boots again to user space.
> 

OK.

> # ls -lrt /sys/bus/nubus/devices/
> total 0
> # cat /proc/bus/nubus/devices
> [no output]
> 
> No need for a Mac test in nubus_proc_init() as it appears to work just 
> fine.
> 

Yes. I suggested a MACH_IS_MAC() test because zorro_proc_init() has a 
similar test. The postcore_initcall() approach retains the MACH_IS_MAC() 
test, so as to bypass nubus_proc_init(). Either way is fine with me.

> Are we confusing any legacy user space tools that might check for 
> presence of /proc/nubus with this patch?

I doubt it.

> Do you want an alternative patch tested (MACH_IS_MAC() test in 
> nubus_driver_register() instead)?
> 

The second patch might be worth testing, but I would wait to hear from 
Geert before I did so.

-- 

> Cheers,
> 
>   Michael
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [RFC] Fix BUG splat during nubus_driver_register()

[Michael Schmitz]

Hi Finn,

with your patch, 4.17rc2 boots again to user space.

# ls -lrt /sys/bus/nubus/devices/
total 0
# cat /proc/bus/nubus/devices
[no output]

No need for a Mac test in nubus_proc_init() as it appears to work just fine.

Are we confusing any legacy user space tools that might check for
presence of /proc/nubus with this patch? Do you want an alternative
patch tested (MACH_IS_MAC() test in
nubus_driver_register() instead)?

Cheers,

  Michael

On Wed, May 2, 2018 at 5:42 PM, Michael Schmitz  wrote:
> Hi Finn,
>
> I'll try that one - will require a new kernel though, and I can't
> currently reach elgar by ssh...
>
> Quite confident this is the right way to fix the issue.
>
> Regarding zorro bus drivers - bus_register is called unconditionally
> from a core initcall for the Zorro bus code. I suppose it's always
> initialized even on ARAnyM. But Geert should know - AFAIR he tests all
> his kernels using ARAnyM.
>
> Cheers,
>
> Michael
>
>
> Am 02.05.2018 um 16:24 schrieb Finn Thain:
>> Please consider the patch below. The approach here is to always register
>> the bus type, regardless of platform.
>>
>> An alternative approach would be to add another MACH_IS_MAC() test to
>> nubus_driver_register().
>>
>> Arguably this patch should add a MACH_IS_MAC() test in nubus_proc_init().
>>
>> BTW, does zorro_register_driver() cause the same failure? What happens if
>> you do 'modprobe zorro7xx' on Aranym?
>>
>> diff --git a/drivers/nubus/nubus.c b/drivers/nubus/nubus.c
>> index 4621ff98138c..5152dbdb0707 100644
>> --- a/drivers/nubus/nubus.c
>> +++ b/drivers/nubus/nubus.c
>> @@ -861,6 +861,9 @@ static void __init nubus_scan_bus(void)
>>  {
>>   int slot;
>>
>> + if (!MACH_IS_MAC)
>> + return 0;
>> +
>>   pr_info("NuBus: Scanning NuBus slots.\n");
>>   for (slot = 9; slot < 15; slot++) {
>>   nubus_probe_slot(slot);
>> @@ -871,9 +874,6 @@ static int __init nubus_init(void)
>>  {
>>   int err;
>>
>> - if (!MACH_IS_MAC)
>> - return 0;
>> -
>>   nubus_proc_init();
>>   err = nubus_bus_register();
>>   if (err)
>>
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [RFC] Fix BUG splat during nubus_driver_register()

[Michael Schmitz]

Hi Finn,

I'll try that one - will require a new kernel though, and I can't
currently reach elgar by ssh...

Quite confident this is the right way to fix the issue.

Regarding zorro bus drivers - bus_register is called unconditionally
from a core initcall for the Zorro bus code. I suppose it's always
initialized even on ARAnyM. But Geert should know - AFAIR he tests all
his kernels using ARAnyM.

Cheers,

Michael


Am 02.05.2018 um 16:24 schrieb Finn Thain:
> Please consider the patch below. The approach here is to always register 
> the bus type, regardless of platform.
> 
> An alternative approach would be to add another MACH_IS_MAC() test to 
> nubus_driver_register().
> 
> Arguably this patch should add a MACH_IS_MAC() test in nubus_proc_init().
> 
> BTW, does zorro_register_driver() cause the same failure? What happens if 
> you do 'modprobe zorro7xx' on Aranym?
> 
> diff --git a/drivers/nubus/nubus.c b/drivers/nubus/nubus.c
> index 4621ff98138c..5152dbdb0707 100644
> --- a/drivers/nubus/nubus.c
> +++ b/drivers/nubus/nubus.c
> @@ -861,6 +861,9 @@ static void __init nubus_scan_bus(void)
>  {
>   int slot;
>  
> + if (!MACH_IS_MAC)
> + return 0;
> +
>   pr_info("NuBus: Scanning NuBus slots.\n");
>   for (slot = 9; slot < 15; slot++) {
>   nubus_probe_slot(slot);
> @@ -871,9 +874,6 @@ static int __init nubus_init(void)
>  {
>   int err;
>  
> - if (!MACH_IS_MAC)
> - return 0;
> -
>   nubus_proc_init();
>   err = nubus_bus_register();
>   if (err)
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-m68k" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html