Re: [PATCH v7 10/50] powerpc/powernv: Simplify pnv_ioda_setup_pe_seg()

On Mon, Nov 16, 2015 at 07:01:18PM +1100, Alexey Kardashevskiy wrote:

On 11/06/2015 10:52 AM, Gavin Shan wrote:

quoted

On Fri, Nov 06, 2015 at 09:56:06AM +1100, Daniel Axtens wrote:

quoted

Gavin Shan [off-list ref] writes:

quoted

The original implementation of pnv_ioda_setup_pe_seg() configures
IO and M32 segments by separate logics, which can be merged by
by caching @segmap, @seg_size, @win in advance. This shouldn't
cause any behavioural changes.

Signed-off-by: Gavin Shan <redacted>
---
 arch/powerpc/platforms/powernv/pci-ioda.c | 62 ++++++++++++++-----------------
 1 file changed, 28 insertions(+), 34 deletions(-)

diff --git a/arch/powerpc/platforms/powernv/pci-ioda.c b/arch/powerpc/platforms/powernv/pci-ioda.c
index 7ee7cfe..553d3f3 100644
--- a/arch/powerpc/platforms/powernv/pci-ioda.c
+++ b/arch/powerpc/platforms/powernv/pci-ioda.c

@@ -2752,8 +2752,10 @@ static void pnv_ioda_setup_pe_seg(struct pci_controller *hose,
 	struct pnv_phb *phb = hose->private_data;
 	struct pci_bus_region region;
 	struct resource *res;
-	int i, index;
-	int rc;
+	unsigned int segsize;
+	int *segmap, index, i;
+	uint16_t win;
+	int64_t rc;

Good catch! Opal return codes are 64 bit and that should be explicit
in the type. However, I seem to remember that we preferred a different
type for 64 bit ints in the kernel. I think it's s64, and there are some
other uses of that in pci_ioda.c for return codes.

Both int64_t and s64 are fine. I used s64 for the OPAL return value, but
Alexey likes "int64_t", which is ok to me as well. I won't change it back
to s64 :-)

quoted

(I'm actually surprised that's not picked up as a compiler
warning. Maybe that's something to look at in future.)

Indeed, I didn't see a warning from gcc.

quoted

The rest of the patch looks good on casual inspection - to be sure I'll
test the entire series on a machine. (hopefully, time permitting!)

I run scripts/checkpatch.pl on the patchset. Only one warning came from
[PATCH 44/50], but I won't bother to change that as the warning was
brought by original code.

None of these patches failed checkpatch.pl check, what was the error in 44/50?

You're right that none of those patches failed with checkpatch.pl. I had revison
6.1, 6.2, 6.3 before this revision (v7) was posted. There was a warning for 44/50
in 6.3 (perhaps).

I run checkpatch.pl on all (v7) patches, no warning reported.

Thanks,
Gavin