On Wed, Jun 21, 2017 at 11:52:59PM +0200, Arnd Bergmann wrote:
> The last rework returned a variable that the compiler cannot prove
> to be initialized:
>
> arch/arm/mach-iop13xx/pci.c: In function 'iop13xx_scan_bus':
> arch/arm/mach-iop13xx/pci.c:509:17: error: 'ret' may be used uninitialized in this function [-Werror=maybe-uninitialized]
>
> In the switch statement that sets the return value, this adds
> a default returning an error.
>
> Fixes: ace27646da8e ("ARM/PCI: Convert PCI scan API to pci_scan_root_bus_bridge()")
> Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
> ---
> arch/arm/mach-iop13xx/pci.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/arch/arm/mach-iop13xx/pci.c b/arch/arm/mach-iop13xx/pci.c
> index b97879bfb81a..3c51a9bb9b49 100644
> --- a/arch/arm/mach-iop13xx/pci.c
> +++ b/arch/arm/mach-iop13xx/pci.c
> @@ -553,6 +553,8 @@ int iop13xx_scan_bus(int nr, struct pci_host_bridge *bridge)
> if (!ret)
> pci_bus_atue = bridge->bus;
> break;
> + default:
> + ret = -EINVAL;
> }
>
> return ret;
Ok I managed to miss these ones somehow, apologies. This series can be
squashed into the initial series but there is also Linus' fix for faraday
to address:
https://patchwork.ozlabs.org/patch/778999/
I can respin a v3 but I am off the radar next week so timing for the
pci_fixup_irqs() removal looks bad, I am not sure what's the next course
of action but I think it is better to drop it for this cycle and wait
for 4.14 (hopefully there are not many PCI host bridges updates to
rebase but I will cope with that anyway - that's the best I can do).
Thanks,
Lorenzo
