pci: pcie-brcmstb: bring over some robustness improvements from Linux

Since the initial U-Boot driver was ported here from Linux, the latter
has had a few changes for robustness/stability. This patch brings over
two of them:
- Do not attempt to access the configuration space of a PCIe device if
  the link has gone down, as that will result in an asynchronous SError
  interrupt which will crash U-Boot.
- Wait for the recommended 100ms after PERST# is deasserted.

I sent this patch while debugging a crash involving PCIe, but these
are unrelated improvements. I do not believe that this patch fixes any
real-world bug.

Signed-off-by: Sam Edwards <CFSworks@gmail.com>

diff --git a/drivers/pci/pcie_brcmstb.c b/drivers/pci/pcie_brcmstb.c
index 1de28021138a7c31587b6336a8dffe843d9b18a3..a159952822bb4495bab42f243fdb848a415a5c55 100644 (file)
--- a/drivers/pci/pcie_brcmstb.c
+++ b/drivers/pci/pcie_brcmstb.c
@@ -223,6 +223,10 @@ static int brcm_pcie_config_address(const struct udevice *dev, pci_dev_t bdf,
                return 0;
        }
 
+       /* An access to our HW w/o link-up will cause a CPU Abort */
+       if (!brcm_pcie_link_up(pcie))
+               return -EINVAL;
+
        /* For devices, write to the config space index register */
        idx = PCIE_ECAM_OFFSET(pci_bus, pci_dev, pci_func, 0);
 
@@ -505,6 +509,12 @@ static int brcm_pcie_probe(struct udevice *dev)
        clrbits_le32(pcie->base + PCIE_RGR1_SW_INIT_1,
                     RGR1_SW_INIT_1_PERST_MASK);
 
+       /*
+        * Wait for 100ms after PERST# deassertion; see PCIe CEM specification
+        * sections 2.2, PCIe r5.0, 6.6.1.
+        */
+       mdelay(100);
+
        /* Give the RC/EP time to wake up, before trying to configure RC.
         * Intermittently check status for link-up, up to a total of 100ms.
         */