On 25.02.2019 15:47, Leonard Crestez wrote:
> On Mon, 2019-02-25 at 15:25 +0100, Stefan Agner wrote:
>> Define the length of the DBI registers and limit config space to its
>> length. This makes sure that the kernel does not access registers
>> beyond that point, avoiding the following abort on a i.MX 6Quad:
>> # cat /sys/devices/soc0/soc/1ffc000.pcie/pci0000\:00/0000\:00\:00.0/config
>> [ 100.021433] Unhandled fault: imprecise external abort (0x1406) at 0xb6ea7000
>> ...
>> [ 100.056423] PC is at dw_pcie_read+0x50/0x84
>> [ 100.060790] LR is at dw_pcie_rd_own_conf+0x44/0x48
>>
>> +static void imx6_pcie_quirk(struct pci_dev *dev)
>> +{
>> + struct pci_bus *bus = dev->bus;
>> + struct pcie_port *pp = bus->sysdata;
>> +
>> + if (bus->number == pp->root_bus_nr) {
>> + struct dw_pcie *pci = to_dw_pcie_from_pp(pp);
>> + struct imx6_pcie *imx6_pcie = to_imx6_pcie(pci);
>> +
>> + /*
>> + * Limit config length to avoid the kernel reading beyond
>> + * the register set and causing an abort on i.MX 6Quad
>> + */
>> + dev->cfg_size = imx6_pcie->drvdata->dbi_length;
>> + }
>> +}
>> +DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, imx6_pcie_quirk);
>
> Are you sure that this quirk is only applied to imx6-pcie?
> Superficially it looks like it would apply to all PCI roots and this
> could be a problem on configs which enable all drivers, like arm64.
That was inspired by the FIXUP in keystone, which uses
PCI_ANY_ID/PCI_ANY_ID too.
However, keystone also does not have arm64 variants it seems.
Furthermore, the keystone fixup checks PCI ids in code.
>
> Maybe the linker magic inside DECLARE_PCI_FIXUP deals with that, in
> that case I apologize.
I don't think there is special linker magic, if CONFIG_PCI_IMX6 is
enabled this will get called.
I agree, I should use PCI_VENDOR_ID_SYNOPSYS, 0xabcd here.
Btw, while looking at the patch again I realized that I need to check
imx6_pcie->drvdata->dbi_length for 0 and not apply in this case to avoid
breaking other i.MX designs.
I will send a v7.
--
Stefan
