| To: | Alan Cox <alan@lxorguk.ukuu.org.uk> |
|---|---|
| Subject: | Re: MIPS ioport_map and Au1xxx PCMCIA |
| From: | Manuel Lauss <mano@roarinelk.homelinux.net> |
| Date: | Thu, 26 Apr 2007 19:45:08 +0200 |
| Cc: | linux-mips@linux-mips.org |
| In-reply-to: | <20070426165525.27bd8a77@the-village.bc.nu> |
| Original-recipient: | rfc822;linux-mips@linux-mips.org |
| References: | <20070426093515.GA26434@roarinelk.homelinux.net> <20070426165525.27bd8a77@the-village.bc.nu> |
| Sender: | linux-mips-bounce@linux-mips.org |
| User-agent: | Mutt/1.5.11 |
On Thu, Apr 26, 2007 at 04:55:25PM +0100, Alan Cox wrote: > iomap uses PIO_MASK to idenfiy ports within the I/O space range. The mips > code looks somewhat incomplete in the base kernel which is probably much > of the problem. > > The basic idea is that ioread/iowrite and friends work generically > regardless of MMIO v PIO so you can just do > > pci_iomap(some bar); > iowrite(blah); > pci_iounmap(blah) > > and not worry about anything. > > Now if your platform doesn't have seperate IO space you could probably > implement iowrite/ioremap/ioport_map to be pretty much no-ops depending > upon how your PCI bars are encoded. This Au1200 doesn't have PCI; to access the PCMCIA IO space one has to ioremap() a part from the top of the 36bit space. I took the cheap route and simply removed the PIO_MASK check from my local kernel. Thank you, -- Manuel Lauss |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | Re: MIPS ioport_map and Au1xxx PCMCIA, Alan Cox |
|---|---|
| Next by Date: | bug in arch/mips/sibyte/cfe/cfe_api.c ?, van Tassell, Eric |
| Previous by Thread: | Re: MIPS ioport_map and Au1xxx PCMCIA, Alan Cox |
| Next by Thread: | [PATCH][MIPS] rename VR41XX to VR4100 series, Yoichi Yuasa |
| Indexes: | [Date] [Thread] [Top] [All Lists] |