Re: [PATCH 3/4] Char: mxser_new, pci_request_region for pci regions

[Rolf Eike Beer]

Jiri Slaby wrote:
> Rolf Eike Beer wrote:
> > Jiri Slaby wrote:
> >> mxser_new, pci_request_region for pci regions
> >>
> >> Use pci_request_region instead of standard request_region for pci device
> >> regions. More checking, simplier use.
> >>
> >> Signed-off-by: Jiri Slaby <jirislaby@xxxxxxxxx>
> >>
> >> ---
> >> commit 1a717bdb06cef859dfbd426f46ea24a9c740e5c5
> >> tree 85460f01008e9fa2edea675a73b394c48139df4a
> >> parent d4f99406c592fb7ce2a65645d7c1f98ebe599238
> >> author Jiri Slaby <jirislaby@xxxxxxxxx> Sat, 30 Sep 2006 01:20:12 +0200
> >> committer Jiri Slaby <xslaby@xxxxxxxxxxxxxxxxxx> Sat, 30 Sep 2006
> >> 01:20:12 +0200
> >>
> >>  drivers/char/mxser_new.c |   10 ++++------
> >>  1 files changed, 4 insertions(+), 6 deletions(-)
> >>
> >> diff --git a/drivers/char/mxser_new.c b/drivers/char/mxser_new.c
> >> index dfef9ce..c566cd0 100644
> >> --- a/drivers/char/mxser_new.c
> >> +++ b/drivers/char/mxser_new.c
> >> @@ -526,8 +526,8 @@ static void __exit mxser_module_exit(voi
> >>  			pdev = mxser_boards[i].pdev;
> >>  			free_irq(mxser_boards[i].irq, &mxser_boards[i]);
> >>  			if (pdev != NULL) {	/* PCI */
> >> -				release_region(pci_resource_start(pdev, 2), pci_resource_len(pdev,
> >> 2)); -				release_region(pci_resource_start(pdev, 3),
> >> pci_resource_len(pdev, 3)); +				pci_release_region(pdev, 2);
> >> +				pci_release_region(pdev, 3);
> >>  				pci_dev_put(pdev);
> >>  			} else {
> >>  				release_region(mxser_boards[i].ports[0].ioaddr, 8 *
> >> mxser_boards[i].nports);
> >> @@ -627,16 +627,14 @@ static int __init
> >> mxser_get_PCI_conf(int
> >>  	brd->board_type = board_type;
> >>  	brd->nports = mxser_numports[board_type - 1];
> >>  	ioaddress = pci_resource_start(pdev, 2);
> >> -	request_region(pci_resource_start(pdev, 2), pci_resource_len(pdev, 2),
> >> -			"mxser(IO)");
> >> +	pci_request_region(pdev, 2, "mxser(IO)");
> >>
> >>  	for (i = 0; i < brd->nports; i++)
> >>  		brd->ports[i].ioaddr = ioaddress + 8 * i;
> >>
> >>  	/* vector */
> >>  	ioaddress = pci_resource_start(pdev, 3);
> >> -	request_region(pci_resource_start(pdev, 3), pci_resource_len(pdev, 3),
> >> -			"mxser(vector)");
> >> +	pci_request_region(pdev, 3, "mxser(vector)");
> >>  	brd->vector = ioaddress;
> >>
> >>  	/* irq */
> >
> > Correct me if I'm wrong, but that use of ioaddress looks totally wrong to
> > me. Isn't there a pci_iomap() or something missing?
>
> Both brd->vector and brd->ports[i].ioaddr are used in inb and outb function
> calls. They themselves remap a small range (one byte in these cases) and
> require ulong addresses as a parameter, not pointer to virtual space,
> correct?

Possible. I try to ignore that interface completely :) I would feel much more 
comfortable if this would be done the other way. But this would probably 
introduce more code because this one seems to work on non-PCI devices also. 
So just ignore me :)

Eike

Attachment: pgp00000.pgp
Description: PGP signature