[PATCH] serial: of-serial: zero-out struct of_serial_info

[PATCH] serial: of-serial: zero-out struct of_serial_info

[Florian Fainelli]

of_platform_serial_probe() invokes of_platform_serial_setup() with a
kmalloc'd struct of_serial_info cookie, which is not much of a problem
for most of the struct of_serial_info fields we access but info->clk.

On platforms which do not provide a clock phandle for their UART nodes
but do provide a 'clock-frequency' property, we basically leave the
info->clk variable uninitialized, and this craps out during
suspend/resume with oopses, crashes or warnings in the clock code.

Fix this by using kzalloc() which also zeroes-out the structure and
provides an initialized info->clk member set to NULL.

Fixes: 2dea53bf57783 ("serial: of-serial: add PM suspend/resume support")
Fixes: e34b9c94b6e8d ("[POWERPC] of_serial: add port type checking")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
---
Greg, Jiri,

This fixes a regression introduced in 3.18-rc1, thanks!

 drivers/tty/serial/of_serial.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/tty/serial/of_serial.c b/drivers/tty/serial/of_serial.c
index 8bc2563335ae..56982da4a9e9 100644
--- a/drivers/tty/serial/of_serial.c
+++ b/drivers/tty/serial/of_serial.c
@@ -158,7 +158,7 @@ static int of_platform_serial_probe(struct platform_device *ofdev)
 	if (of_find_property(ofdev->dev.of_node, "used-by-rtas", NULL))
 		return -EBUSY;
 
-	info = kmalloc(sizeof(*info), GFP_KERNEL);
+	info = kzalloc(sizeof(*info), GFP_KERNEL);
 	if (info == NULL)
 		return -ENOMEM;
 
-- 
1.9.1

Re: [PATCH] serial: of-serial: zero-out struct of_serial_info

[Arnd Bergmann]

On Thursday 30 October 2014 14:18:33 Florian Fainelli wrote:
> of_platform_serial_probe() invokes of_platform_serial_setup() with a
> kmalloc'd struct of_serial_info cookie, which is not much of a problem
> for most of the struct of_serial_info fields we access but info->clk.
> 
> On platforms which do not provide a clock phandle for their UART nodes
> but do provide a 'clock-frequency' property, we basically leave the
> info->clk variable uninitialized, and this craps out during
> suspend/resume with oopses, crashes or warnings in the clock code.
> 
> Fix this by using kzalloc() which also zeroes-out the structure and
> provides an initialized info->clk member set to NULL.
> 
> Fixes: 2dea53bf57783 ("serial: of-serial: add PM suspend/resume support")
> Fixes: e34b9c94b6e8d ("[POWERPC] of_serial: add port type checking")
> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> 

I thought I'd acked this patch before, maybe someone else sent the same
one. This is definitely required, so

Acked-by: Arnd Bergmann <arnd@arndb.de>

Re: [PATCH] serial: of-serial: zero-out struct of_serial_info

[Florian Fainelli]

On 10/30/2014 02:55 PM, Arnd Bergmann wrote:
> On Thursday 30 October 2014 14:18:33 Florian Fainelli wrote:
>> of_platform_serial_probe() invokes of_platform_serial_setup() with a
>> kmalloc'd struct of_serial_info cookie, which is not much of a problem
>> for most of the struct of_serial_info fields we access but info->clk.
>>
>> On platforms which do not provide a clock phandle for their UART nodes
>> but do provide a 'clock-frequency' property, we basically leave the
>> info->clk variable uninitialized, and this craps out during
>> suspend/resume with oopses, crashes or warnings in the clock code.
>>
>> Fix this by using kzalloc() which also zeroes-out the structure and
>> provides an initialized info->clk member set to NULL.
>>
>> Fixes: 2dea53bf57783 ("serial: of-serial: add PM suspend/resume support")
>> Fixes: e34b9c94b6e8d ("[POWERPC] of_serial: add port type checking")
>> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
>>
> 
> I thought I'd acked this patch before, maybe someone else sent the same
> one. This is definitely required, so
> 
> Acked-by: Arnd Bergmann <arnd@arndb.de>
> 

Thanks, so who's taking that patch? Let me know if I need to resubmit
targeting a specific tree.
--
Florian

Re: [PATCH] serial: of-serial: zero-out struct of_serial_info

[Greg KH]

On Tue, Nov 04, 2014 at 10:23:49AM -0800, Florian Fainelli wrote:
> On 10/30/2014 02:55 PM, Arnd Bergmann wrote:
> > On Thursday 30 October 2014 14:18:33 Florian Fainelli wrote:
> >> of_platform_serial_probe() invokes of_platform_serial_setup() with a
> >> kmalloc'd struct of_serial_info cookie, which is not much of a problem
> >> for most of the struct of_serial_info fields we access but info->clk.
> >>
> >> On platforms which do not provide a clock phandle for their UART nodes
> >> but do provide a 'clock-frequency' property, we basically leave the
> >> info->clk variable uninitialized, and this craps out during
> >> suspend/resume with oopses, crashes or warnings in the clock code.
> >>
> >> Fix this by using kzalloc() which also zeroes-out the structure and
> >> provides an initialized info->clk member set to NULL.
> >>
> >> Fixes: 2dea53bf57783 ("serial: of-serial: add PM suspend/resume support")
> >> Fixes: e34b9c94b6e8d ("[POWERPC] of_serial: add port type checking")
> >> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
> >>
> > 
> > I thought I'd acked this patch before, maybe someone else sent the same
> > one. This is definitely required, so
> > 
> > Acked-by: Arnd Bergmann <arnd@arndb.de>
> > 
> 
> Thanks, so who's taking that patch? Let me know if I need to resubmit
> targeting a specific tree.

Someone else sent this before you did, and I took that version instead,
sorry.

greg k-h

Re: [PATCH] serial: of-serial: zero-out struct of_serial_info

[Florian Fainelli]

On 11/06/2014 10:33 AM, Greg KH wrote:
> On Tue, Nov 04, 2014 at 10:23:49AM -0800, Florian Fainelli wrote:
>> On 10/30/2014 02:55 PM, Arnd Bergmann wrote:
>>> On Thursday 30 October 2014 14:18:33 Florian Fainelli wrote:
>>>> of_platform_serial_probe() invokes of_platform_serial_setup() with a
>>>> kmalloc'd struct of_serial_info cookie, which is not much of a problem
>>>> for most of the struct of_serial_info fields we access but info->clk.
>>>>
>>>> On platforms which do not provide a clock phandle for their UART nodes
>>>> but do provide a 'clock-frequency' property, we basically leave the
>>>> info->clk variable uninitialized, and this craps out during
>>>> suspend/resume with oopses, crashes or warnings in the clock code.
>>>>
>>>> Fix this by using kzalloc() which also zeroes-out the structure and
>>>> provides an initialized info->clk member set to NULL.
>>>>
>>>> Fixes: 2dea53bf57783 ("serial: of-serial: add PM suspend/resume support")
>>>> Fixes: e34b9c94b6e8d ("[POWERPC] of_serial: add port type checking")
>>>> Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
>>>>
>>>
>>> I thought I'd acked this patch before, maybe someone else sent the same
>>> one. This is definitely required, so
>>>
>>> Acked-by: Arnd Bergmann <arnd@arndb.de>
>>>
>>
>> Thanks, so who's taking that patch? Let me know if I need to resubmit
>> targeting a specific tree.
> 
> Someone else sent this before you did, and I took that version instead,
> sorry.

Not a problem, my message contained much more explanations, but the end
result is the same.

Thanks!
--
Florian