[PATCH] netxen: fix pci bar mapping

Use phys_addr_t data type for pci resource address
instead of unsigned long. This fixes mapping of
pci resources where resource addresses can be larger
than word size (e.g. PAE).

Signed-off-by: Dhananjay Phadke <redacted>
---
 drivers/net/netxen/netxen_nic_main.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

diff --git a/drivers/net/netxen/netxen_nic_main.c b/drivers/net/netxen/netxen_nic_main.c
index 9b9eab1..194b3e3 100644
--- a/drivers/net/netxen/netxen_nic_main.c
+++ b/drivers/net/netxen/netxen_nic_main.c

@@ -595,7 +595,8 @@ netxen_setup_pci_map(struct netxen_adapter *adapter)
 	void __iomem *mem_ptr2 = NULL;
 	void __iomem *db_ptr = NULL;
 
-	unsigned long mem_base, mem_len, db_base, db_len = 0, pci_len0 = 0;
+	phys_addr_t mem_base, db_base;
+	unsigned long mem_len, db_len = 0, pci_len0 = 0;
 
 	struct pci_dev *pdev = adapter->pdev;
 	int pci_func = adapter->ahw.pci_func;

-- 
1.6.0.2

From: Dhananjay Phadke <redacted>
Date: Sun, 11 Oct 2009 19:43:22 -0700

Use phys_addr_t data type for pci resource address
instead of unsigned long. This fixes mapping of
pci resources where resource addresses can be larger
than word size (e.g. PAE).

Signed-off-by: Dhananjay Phadke <redacted>

The correct type to use is 'resource_size_t' not 'phys_addr_t'.

Look at the "struct resource" member types in linux/ioport.h, that's
what these pci_resource_start() et al. routines return.

Please fix this up and resubmit.

Ok, I am resending the patch.

Thanks,
Dhananjay

David Miller wrote:

From: Dhananjay Phadke <redacted>
Date: Sun, 11 Oct 2009 19:43:22 -0700

quoted

Use phys_addr_t data type for pci resource address
instead of unsigned long. This fixes mapping of
pci resources where resource addresses can be larger
than word size (e.g. PAE).

Signed-off-by: Dhananjay Phadke <redacted>

The correct type to use is 'resource_size_t' not 'phys_addr_t'.

Look at the "struct resource" member types in linux/ioport.h, that's
what these pci_resource_start() et al. routines return.

Please fix this up and resubmit.