Re: [PATCH] check_unwind: Add request/release_firmware to the function table

Hi Dan,

On 31/10/23 8:21 pm, Dan Carpenter wrote:
> On Tue, Oct 31, 2023 at 05:06:47AM -0700, Harshit Mogalapalli wrote:
>> Add request and release firmware to catch bugs where there is an
>> imbalance between release and request firmware.
>>
>> request_firmware(const struct firmware **firmware_p, const char *name,
>> 		 struct device *device)
>>
>> release frees(firmware) so we should pass '0' to the table entry in the
>> ALLOC.
>>
>> Signed-off-by: Harshit Mogalapalli <[email protected]>
>> ---
>>   check_unwind.c | 3 +++
>>   1 file changed, 3 insertions(+)
>>
>> diff --git a/check_unwind.c b/check_unwind.c
>> index 59678a94..0dc8f752 100644
>> --- a/check_unwind.c
>> +++ b/check_unwind.c
>> @@ -95,6 +95,9 @@ static struct ref_func_info func_table[] = {
>>   
>>   	{ "ieee80211_alloc_hw", ALLOC,  -1, "$", &valid_ptr_min_sval, &valid_ptr_max_sval },
>>   	{ "ieee80211_free_hw",  RELEASE, 0, "$" },
>> +
>> +	{ "request_firmware", ALLOC,	0, "$", &valid_ptr_min_sval, &valid_ptr_max_sval },
> 
> Doesn't request_firmware() return zero on success?  &int_zero?
> 

Right, I missed that sorry.

I will fix it in V2.

This also answers why my test complained like this incorrectly, but I 
thought that was due to db:

drivers/usb/typec/tipd/core.c:1009 tps25750_start_patch_burst_mode() 
warn: '&fw' from request_firmware() not released on lines: 960.
drivers/usb/typec/tipd/core.c:1009 tps25750_start_patch_burst_mode() 
warn: '&fw' from request_firmware() not released on lines: 971,976,980,1009.

But when I changed it to &int_zero, it works well and shows this warning 
alone:

drivers/usb/typec/tipd/core.c:1009 tps25750_start_patch_burst_mode() 
warn: '&fw' from request_firmware() not released on lines: 971,976,980,1009.

Thanks i will send a V2.

Regards,
Harshit
> regards,
> dan carpenter
>