Re: [PATCH -next] tile/PCI: use for_each_pci_dev to simplify the code

[Bjorn Helgaas <bhelgaas@google.com>]

On Mon, Dec 3, 2012 at 6:28 AM, Wei Yongjun <weiyj.lk@gmail.com> wrote:
> From: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
>
> Use for_each_pci_dev to simplify the code.
>
> Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
> ---
>  arch/tile/kernel/pci.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/arch/tile/kernel/pci.c b/arch/tile/kernel/pci.c
> index 7598226..aac1cd5 100644
> --- a/arch/tile/kernel/pci.c
> +++ b/arch/tile/kernel/pci.c
> @@ -245,7 +245,7 @@ static void __devinit fixup_read_and_payload_sizes(void)
>         u16 new_values;
>
>         /* Scan for the smallest maximum payload size. */
> -       while ((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL) {
> +       for_each_pci_dev(dev) {
>                 u32 devcap;
>                 int max_payload;
>
> @@ -260,7 +260,7 @@ static void __devinit fixup_read_and_payload_sizes(void)
>
>         /* Now, set the max_payload_size for all devices to that value. */
>         new_values = (max_read_size << 12) | (smallest_max_payload << 5);
> -       while ((dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev)) != NULL)
> +       for_each_pci_dev(dev)
>                 pcie_capability_clear_and_set_word(dev, PCI_EXP_DEVCTL,
>                                 PCI_EXP_DEVCTL_PAYLOAD | PCI_EXP_DEVCTL_READRQ,
>                                 new_values);
>
>

This patch looks correct to me, as far as it goes.

Of course, any time we use for_each_pci_dev(), it's a good clue that
this code won't work for hotplug, so presumably you don't care about
hotplug on this system.

tile/kernel/pci_gx.c has a function of the same name, implemented
completely differently, which suggests that some unification or
refactoring might be possible.

In general, I think the PCI_EXP_DEVCAP and PCI_EXP_DEVCTL code for MPS
and MRRS management belongs in the PCI core, not in drivers or
architectures (it's fine if they provide device- or platform-specific
constraints), so anything you can do to move in that direction would
be good.

Bjorn