[coreboot-gerrit] New patch to review for coreboot: soc/intel/common: don't infinitely recurse in busmaster_disable_on_bus()

[Aaron Durbin (adurbin@chromium.org)]

Aaron Durbin (adurbin at chromium.org) just uploaded a new patch set to gerrit, which you can find at https://review.coreboot.org/15157

-gerrit

commit ceaacd5632c33114c5181b84da1993e6075f9c2c
Author: Aaron Durbin <adurbin at chromium.org>
Date:   Fri Jun 10 21:44:26 2016 -0500

    soc/intel/common: don't infinitely recurse in busmaster_disable_on_bus()
    
    If a bridge has the primary bus equal to the secondary bus the
    busmaster_disable_on_bus() will infinitely call itself. Avoid the
    inifinite recursion by checking current bus number against the
    secondary bus number.
    
    BUG=chrome-os-partner:54262
    TEST=Ran on reef. Able to actually get the chipset to assert SLP_Sx
         signals which means no more infinite recursion.
    
    Change-Id: I52b21fbba24e6a652ea8f9f87f5f49f60109c8f2
    Signed-off-by: Aaron Durbin <adurbin at chromium.org>
---
 src/soc/intel/common/smihandler.c | 23 ++++++++++++++++-------
 1 file changed, 16 insertions(+), 7 deletions(-)

diff --git a/src/soc/intel/common/smihandler.c b/src/soc/intel/common/smihandler.c
index 0b220e2..5915587 100644
--- a/src/soc/intel/common/smihandler.c
+++ b/src/soc/intel/common/smihandler.c
@@ -104,15 +104,24 @@ static void busmaster_disable_on_bus(int bus)
 			reg32 &= ~PCI_COMMAND_MASTER;
 			pci_write_config32(dev, PCI_COMMAND, reg32);
 
-			/* If this is a bridge, then follow it. */
+			/* If it's not a bridge, move on. */
 			hdr = pci_read_config8(dev, PCI_HEADER_TYPE);
 			hdr &= 0x7f;
-			if (hdr == PCI_HEADER_TYPE_BRIDGE ||
-			    hdr == PCI_HEADER_TYPE_CARDBUS) {
-				unsigned int buses;
-				buses = pci_read_config32(dev, PCI_PRIMARY_BUS);
-				busmaster_disable_on_bus((buses >> 8) & 0xff);
-			}
+			if (hdr != PCI_HEADER_TYPE_BRIDGE &&
+			    hdr != PCI_HEADER_TYPE_CARDBUS)
+				continue;
+
+			/*
+			 * If secondary bus is equal to current bus bypass
+			 * the bridge because it's likely unconfigured and
+			 * would cause infinite recursion.
+			 */
+			int secbus = pci_read_config8(dev, PCI_SECONDARY_BUS);
+
+			if (secbus == bus)
+				continue;
+
+			busmaster_disable_on_bus(secbus);
 		}
 	}
 }