Re: [PATCH] PCI: Mediatek: Use resource_size function on resource object

From: Honghui Zhang
Date: Thu Jan 31 2019 - 02:52:46 EST


On Thu, 2019-01-31 at 11:03 +0800, Honghui Zhang wrote:
> On Wed, 2019-01-30 at 09:49 -0600, Bjorn Helgaas wrote:
> > On Wed, Jan 02, 2019 at 02:03:53PM +0800, honghui.zhang@xxxxxxxxxxxx wrote:
> > > From: Honghui Zhang <honghui.zhang@xxxxxxxxxxxx>
> > >
> > > drivers/pci/pcie-mediatek.c:720:13-16: WARNING: Suspicious code. resource_size is maybe missing with mem
> > >
> > > Generated by: scripts/coccinelle/api/resource_size.cocci
> > >
> > > Signed-off-by: Honghui Zhang <honghui.zhang@xxxxxxxxxxxx>
> > > ---
> > > drivers/pci/controller/pcie-mediatek.c | 4 +---
> > > 1 file changed, 1 insertion(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/pci/controller/pcie-mediatek.c b/drivers/pci/controller/pcie-mediatek.c
> > > index e307166..0168376 100644
> > > --- a/drivers/pci/controller/pcie-mediatek.c
> > > +++ b/drivers/pci/controller/pcie-mediatek.c
> > > @@ -654,7 +654,6 @@ static int mtk_pcie_startup_port_v2(struct mtk_pcie_port *port)
> > > struct resource *mem = &pcie->mem;
> > > const struct mtk_pcie_soc *soc = port->pcie->soc;
> > > u32 val;
> > > - size_t size;
> > > int err;
> > >
> > > /* MT7622 platforms need to enable LTSSM and ASPM from PCIe subsys */
> > > @@ -706,8 +705,7 @@ static int mtk_pcie_startup_port_v2(struct mtk_pcie_port *port)
> > > mtk_pcie_enable_msi(port);
> > >
> > > /* Set AHB to PCIe translation windows */
> > > - size = mem->end - mem->start;
> > > - val = lower_32_bits(mem->start) | AHB2PCIE_SIZE(fls(size));
> > > + val = lower_32_bits(mem->start) | AHB2PCIE_SIZE(fls(resource_size(mem)));
> >
> > This is actually a fairly interesting change because it effectively
> > changes this:
> >
> > fls(mem->end - mem->start)
> >
> > to this:
> >
> > fls(mem->end - mem->start + 1)
> >
> > And mem->end is the last valid address, so it changes something like
> > this:
> >
> > fls(0xffff) # == 15
> >
> > to this:
> >
> > fls(0x10000) # == 16
> >
> > So while this *looks* like a trivial warning fix, it likely fixes an
> > important bug, and it's worth pointing out what that bug is in the
> > changelog.
> >
> > > writel(val, port->base + PCIE_AHB_TRANS_BASE0_L);
> > >
> > > val = upper_32_bits(mem->start);
>
> This size will set the MMIO size, which means that the RC will translate
> the MMIO access from mem->start to mem->end.
> The real MMIO size is specified in devicetree, which is 0x1000_0000 for
> both mt2712 and mt7622.
>
> This change make the size from fls(0x1000_0000) to fls(0x1000_0001), not
> really change the values.
>
> I will update the commit message and add the information mentioned
> above.
>
> Thanks for your kindly review.

I was wrong, after take a look at the resource parser function, that it
will initialize the res->end as res->start + res->size - 1.

But this change is still Ok since it will change the MMIO from
fls(0xfff_ffff) to fls(0x1000_0000), this just enlarge the MMIO
translate window size. The HW assigned MMIO is 0x1000_0000, but original
code set this translate window to fls(0xfff_ffff) = 0x800_0000 is fine
in most case since all the EP device we connect never asked a MMIO
window bigger than 0x800_0000. (As a matter of fact, most EP device will
only ask for 4MB MMIO window size.)

I will put this in the commit message.

thanks.
>
> > > --
> > > 2.6.4
> > >
>
>