Re: PCI driver

[Manu Abraham <manu@kromtek.com>]

Jiri Slaby wrote:
> Manu Abraham napsal(a):
> 
>> Jiri Slaby wrote:
>>
>>> Manu Abraham napsal(a):
>>>
>>>> Jiri Slaby wrote:
>>>>
>>>>> you do NOT do this at all, because you have pdev already (the param 
>>>>> of the probe function)
>>>>>
>>>>
>>>> I rewrote the entire thing like this including the pci_remove 
>>>> function too, but now it so seems that in the remove function, 
>>>> pci_get_drvdata(pdev) returns NULL, and hence i get an Oops at 
>>>> module removal.
>>>
>>>
>>>
>>> Maybe because this is badly written driver.
>>
>>
>>
>> I have not written the driver, but this is my first go at it ..
>>
>>>
>>>> static int mantis_pci_probe(struct pci_dev *pdev, const struct 
>>>> pci_device_id *mantis_pci_table)
>>>> {
>>>>     struct mantis_pci *mantis;
>>>>     struct mantis_eeprom eeprom;
>>>>     u8 revision, latency;
>>>>     u8 data[2];      if (pci_enable_device(pdev)) {               
>>>> dprintk(verbose, MANTIS_DEBUG, 1, "Found a mantis chip");
>>>>         if ((mantis = (struct mantis_pci *) kmalloc(sizeof (struct 
>>>> mantis_pci), GFP_KERNEL)) == NULL) {
>>>>             dprintk(verbose, MANTIS_ERROR, 1, "Out of memory");
>>>>             return -ENOMEM;
>>>>         }
>>>>         pci_set_master(pdev);
>>>>         mantis->mantis_addr = pci_resource_start(pdev, 0);
>>>>         if (!request_mem_region(pci_resource_start(pdev, 0),
>>>>             pci_resource_len(pdev, 0), DRIVER_NAME)) {
>>>>             kfree(mantis);
>>>>             return -EBUSY;
>>>>         }
>>>>         pci_read_config_byte(pdev, PCI_CLASS_REVISION, &revision);
>>>>         pci_read_config_byte(pdev, PCI_LATENCY_TIMER, &latency);
>>>>         mantis->mantis_mmio = ioremap(mantis->mantis_addr, 0x1000);
>>>>         pci_set_drvdata(pdev, mantis);       
>>>
>>>
>>>
>>> if pci_enable_device fails, you set this?? Maybe you haven't read the 
>>> doc enough.
>>
>>
>>
>>
>> I just found that, pci_enable_device() fails. So what's the way to go 
>> ahead ?
> 
> JESUS.


What i meant is i do have to enable the device, not just exit.
I understood that i have to exit if pci_enable_device() fails, but what 
i am looking for is why i can't enable it in the first place.


> int retval = 0;
> 
> if ((retval = pci_enable_device()))
>     goto end;
> 
> ...
> pci_set_drvdata(pdev, mantis);
> ...
> 
> end:
>     return retval;
> 
> not
> if (pci_enable_device())
>     do something


Regards,
Manu