[PATCH 0/6] fbdevhw cleanup and pciaccess free

[Mark Kettenis]

> Date: Mon, 31 May 2010 17:42:51 +0300
> From: "Vignatti Tiago (Nokia-D/Helsinki)" <tiago.vignatti at nokia.com>
> 
> On Mon, May 31, 2010 at 04:29:11PM +0200, ext Mark Kettenis wrote:
> > > From: Tiago Vignatti <tiago.vignatti at nokia.com>
> > > 
> > > [0] I'm wondering if it doesn't make sense for someone to free modules from
> > > PCI code? For me it's pretty clear but I'll explain the machiavellian plan
> > > anyway in the next series that I'll send today.
> > 
> > So instead you decide to pollute the generic PCI code with
> > Linux-specific cra^H^H^Hcode?
> 
> can you please point exactly where I'm polluting generic PCI code with Linux
> specific? I found your commentary pretty interesting because this is one of
> the problems I'm trying to address with the cleaning up patches I've been
> sending.

The bit below.

--- a/hw/xfree86/os-support/bus/Pci.c
+++ b/hw/xfree86/os-support/bus/Pci.c
@@ -134,6 +134,7 @@
 #include "Pci.h"
 
 #include <pciaccess.h>
+#include <fcntl.h>
 
 /* Global data */
 
@@ -155,3 +156,24 @@ xf86scanpci(void)
 
     return success;
 }
+
+int
+xf86PciOpenFbDev(struct pci_device *pPci, int num)
+{
+    char filename[256];
+    int fd;
+
+    sprintf(filename,
+        "/sys/bus/pci/devices/%04x:%02x:%02x.%d/graphics/fb%d",
+        pPci->domain, pPci->bus, pPci->dev, pPci->func, num);
+
+    fd = open(filename, O_RDONLY, 0);
+    if (fd < 0) {
+        sprintf(filename,
+            "/sys/bus/pci/devices/%04x:%02x:%02x.%d/graphics:fb%d",
+            pPci->domain, pPci->bus, pPci->dev, pPci->func, num);
+        fd = open(filename, O_RDONLY, 0);
+    }
+
+    return fd;
+}