Prev: [PATCH 3/3] vmscan: Put kswapd to sleep on its own waitqueue, not congestion
Next: [PATCH 1/4] x86: remove rdc321x_defs.h
From: Paul Mundt on 8 Mar 2010 06:50
On Sun, Mar 07, 2010 at 10:53:25PM +0100, Samuel Ortiz wrote:
> H Hartley Sweeten (3):
> mfd: Use resource_size() in sm501
> mfd: Use resource_size() for t7l66xb
> mfd: Use resouce_size for tc6393xb
>
These need to be looked at carefully, at least the sm501 change is
totally bogus, breaks everything, and stems from a total lack of
understanding of how the drivers interact.

The I/O resource that's remapped and handled by the MFD takes the full
size, while the region request is purposely trimmed down so only the
registers needed by the MFD driver are reserved by the MFD driver. Other
drivers (like sm501fb) will have their own region requests within the
same resource at different offsets, all of which are broken after this
change.

As such, 311e54c07e784fe63d421f724dcc597d1f82ccad at least needs to be
reverted. I haven't looked at the other two, but if it's a similar
mechanical change then those should be dropped, too.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: H Hartley Sweeten on 8 Mar 2010 10:20
On Monday, March 08, 2010 4:49 AM, Paul Mundt wrote:
> On Sun, Mar 07, 2010 at 10:53:25PM +0100, Samuel Ortiz wrote:
>> H Hartley Sweeten (3):
>> mfd: Use resource_size() in sm501
>> mfd: Use resource_size() for t7l66xb
>> mfd: Use resouce_size for tc6393xb
>>
> These need to be looked at carefully, at least the sm501 change is
> totally bogus, breaks everything, and stems from a total lack of
> understanding of how the drivers interact.
>
> The I/O resource that's remapped and handled by the MFD takes the full
> size, while the region request is purposely trimmed down so only the
> registers needed by the MFD driver are reserved by the MFD driver. Other
> drivers (like sm501fb) will have their own region requests within the
> same resource at different offsets, all of which are broken after this
> change.
>
> As such, 311e54c07e784fe63d421f724dcc597d1f82ccad at least needs to be
> reverted. I haven't looked at the other two, but if it's a similar
> mechanical change then those should be dropped, too.

Hello Paul,

I took a look at the sm501 patch and see your concern.

The ioremap change to use resource_size looks ok. But the two request_mem_region
changes are an issue due to the point you mentioned above.

The patches to t7166xb and tc6393xb both appear to be ok. They do not have the
same issue.

Regards,
Hartley
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: [PATCH 3/3] vmscan: Put kswapd to sleep on its own waitqueue, not congestion
Next: [PATCH 1/4] x86: remove rdc321x_defs.h