ChangeSet 1.2078, 2005/03/12 09:15:25-08:00, [EMAIL PROTECTED]
[PATCH] PA-RISC PCI update
- Fix PCI-PCI bridges under the first HBA in the system
Signed-off-by: Ryan Bradetich <[EMAIL PROTECTED]>
- fix up the comment to reflect rbrad's change and why it was made.
- fix base_addr warning
From: Grant Grundler <[EMAIL PROTECTED]>
- Reword the explanation about why we renumber PCI busses on all
platforms.
- Remove CONFIG_PARISC64 symbol; use CONFIG_64BIT everywhere instead.
Signed-off-by: Matthew Wilcox <[EMAIL PROTECTED]>
Signed-off-by: Matthew Wilcox <[EMAIL PROTECTED]>
Signed-off-by: Linus Torvalds <[EMAIL PROTECTED]>
pci.h | 39 +++++++++++++++++++++++++--------------
1 files changed, 25 insertions(+), 14 deletions(-)
diff -Nru a/include/asm-parisc/pci.h b/include/asm-parisc/pci.h
--- a/include/asm-parisc/pci.h 2005-03-12 21:43:10 -08:00
+++ b/include/asm-parisc/pci.h 2005-03-12 21:43:10 -08:00
@@ -28,7 +28,7 @@
** Data needed by pcibios layer belongs here.
*/
struct pci_hba_data {
- unsigned long base_addr; /* aka Host Physical Address */
+ void __iomem *base_addr; /* aka Host Physical Address */
const struct parisc_device *dev; /* device from PA bus walk */
struct pci_bus *hba_bus; /* primary PCI bus below HBA */
int hba_num; /* I/O port space access "key" */
@@ -69,7 +69,7 @@
#define PCI_PORT_HBA(a) ((a) >> HBA_PORT_SPACE_BITS)
#define PCI_PORT_ADDR(a) ((a) & (HBA_PORT_SPACE_SIZE - 1))
-#if CONFIG_PARISC64
+#if CONFIG_64BIT
#define PCI_F_EXTEND 0xffffffff00000000UL
#define PCI_IS_LMMIO(hba,a) pci_is_lmmio(hba,a)
@@ -90,14 +90,14 @@
: (a)) /* GMMIO */
#define PCI_HOST_ADDR(hba,a) ((a) + hba->lmmio_space_offset)
-#else /* !CONFIG_PARISC64 */
+#else /* !CONFIG_64BIT */
#define PCI_BUS_ADDR(hba,a) (a)
#define PCI_HOST_ADDR(hba,a) (a)
#define PCI_F_EXTEND 0UL
#define PCI_IS_LMMIO(hba,a) (1) /* 32-bit doesn't support GMMIO */
-#endif /* !CONFIG_PARISC64 */
+#endif /* !CONFIG_64BIT */
/*
** KLUGE: linux/pci.h include asm/pci.h BEFORE declaring struct pci_bus
@@ -199,16 +199,27 @@
#endif
/*
-** used by drivers/pci/pci.c:pci_do_scan_bus()
-** 0 == check if bridge is numbered before re-numbering.
-** 1 == pci_do_scan_bus() should automatically number all PCI-PCI bridges.
-**
-** REVISIT:
-** To date, only alpha sets this to one. We'll need to set this
-** to zero for legacy platforms and one for PAT platforms.
-*/
-#define pcibios_assign_all_busses() (pdc_type == PDC_TYPE_PAT)
-#define pcibios_scan_all_fns(a, b) 0
+ * pcibios_assign_all_busses() is used in drivers/pci/pci.c:pci_do_scan_bus()
+ * 0 == check if bridge is numbered before re-numbering.
+ * 1 == pci_do_scan_bus() should automatically number all PCI-PCI bridges.
+ *
+ * We *should* set this to zero for "legacy" platforms and one
+ * for PAT platforms.
+ *
+ * But legacy platforms also need to renumber the busses below a Host
+ * Bus controller. Adding a 4-port Tulip card on the first PCI root
+ * bus of a C200 resulted in the secondary bus being numbered as 1.
+ * The second PCI host bus controller's root bus had already been
+ * assigned bus number 1 by firmware and sysfs complained.
+ *
+ * Firmware isn't doing anything wrong here since each controller
+ * is its own PCI domain. It's simpler and easier for us to renumber
+ * the busses rather than treat each Dino as a separate PCI domain.
+ * Eventually, we may want to introduce PCI domains for Superdome or
+ * rp7420/8420 boxes and then revisit this issue.
+ */
+#define pcibios_assign_all_busses() (1)
+#define pcibios_scan_all_fns(a, b) (0)
#define PCIBIOS_MIN_IO 0x10
#define PCIBIOS_MIN_MEM 0x1000 /* NBPG - but pci/setup-res.c dies */
-
To unsubscribe from this list: send the line "unsubscribe bk-commits-head" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
