RE: [PATCH] cpqarray update

[Wiran, Francis]

Hmm... cpqarray supports both pci and eisa ctrls.

What would happen if there are no pci ctrl detected, but there is one
eisa ctrl detected? What do we return in cpqarray_init()? The return
code of pci_module_init is either 0 or -ENODEV. What happen if
cpqarray_init returns -ENODEV? Would the driver stay loaded? How about
0, or positive number?


thanks
-francis-

> -----Original Message-----
> From: Jeff Garzik [mailto:jgarzik@xxxxxxxxx] 
> Sent: Monday, January 26, 2004 7:52 PM
> To: Wiran, Francis
> Cc: Linux Kernel Mailing List
> Subject: Re: [PATCH] cpqarray update
> 
> 
> Wiran, Francis wrote:
> >>You need to check the return value of pci_module_init() for errors.
> > 
> > No, because the return value is determined from number of 
> ctrls found,
> > and not from function return.
> > 
> > int __init cpqarray_init(void)
> > {
> > ...
> > 	pci_module_init(&cpqarray_pci_driver);
> > 	cpqarray_eisa_detect();
> > 
> > 	for(i=0;i<MAX_CTLR;i++) {
> > 		if(hba[i] != NULL)
> > 			num_ctlrs_reg++
> > 	}
> > 
> > 	return (num_ctlrs_reg);
> > }
> > 
> > int __init cpqarray_init_module(void)
> > {
> > 	if (cpqarray_init() == 0)
> > 		return -ENODEV;
> > 	return 0;
> > }
> 
> 
> Nope, this needs to be turned inside out.  The proper PCI 
> driver looks like
> 
> static int __init cp_init (void)
> {
>          return pci_module_init (&cp_driver);
> }
> 
> static void __exit cp_exit (void)
> {
>          pci_unregister_driver (&cp_driver);
> }
> 
> We already handle the cases you describe.  The cpqarray code -breaks- 
> the API design by doing it this way.
> 
> cpqarray does not fully support the pci_ids features and 
> hotplug without 
> this.
> 
> 	Jeff
> 
> 
> 
> 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/