Bug 198840

Created attachment 132404 [details]
Xorg configuration.

Video card's PCI information:

01:00.0 0300: 1002:5961 (rev 01)
        Subsystem: 174b:7c13
        Flags: bus master, 66MHz, medium devsel, latency 64, IRQ 17
        Memory at f0000000 (32-bit, prefetchable) [size=128M]
        I/O ports at ec00 [size=256]
        Memory at ff8f0000 (32-bit, non-prefetchable) [size=64K]
        Expansion ROM at ff800000 [disabled] [size=128K]
        Capabilities: <access denied>

01:00.1 0380: 1002:5941 (rev 01)
        Subsystem: 174b:7c12
        Flags: bus master, 66MHz, medium devsel, latency 64
        Memory at e8000000 (32-bit, prefetchable) [size=128M]
        Memory at ff8e0000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: <access denied>

Created attachment 132405 [details]
Xorg log file

Dave, did this PCI ID get removed from the kernel along with R300 too?

What's the difference between this RPM and the "vanilla" 6.5.8.0 ATI driver?

Could you try downloading and installing (not upgrading but installing) the
latest kernel from rawhide and see if that helps?

Oops, I forgot to mention that I'd since contacted the Xorg people, who gave me
this patch against my Radeon DRI kernel module. The basic problem was that the
u32 fb_end variable was overflowing on my system, such that fb_start + fb_size = 
0.

diff -u -u -r1.72 radeon_state.c
--- drm/shared-core/radeon_state.c      24 May 2006 18:36:24 -0000      1.72
+++ drm/shared-core/radeon_state.c      23 Aug 2006 20:20:50 -0000
@@ -43,6 +43,10 @@
                                                    u32 * offset)
 {
        u32 off = *offset;
+       u32 fb_start = dev_priv->fb_location;
+       u32 fb_end = fb_start + dev_priv->fb_size - 1;
+       u32 gart_start = dev_priv->gart_vm_start;
+       u32 gart_end = gart_start + dev_priv->gart_size - 1;
        struct drm_radeon_driver_file_fields *radeon_priv;

        /* Hrm ... the story of the offset ... So this function converts
@@ -62,10 +66,8 @@
        /* First, the best case, the offset already lands in either the
         * framebuffer or the GART mapped space
         */
-       if ((off >= dev_priv->fb_location &&
-            off < (dev_priv->fb_location + dev_priv->fb_size)) ||
-           (off >= dev_priv->gart_vm_start &&
-            off < (dev_priv->gart_vm_start + dev_priv->gart_size)))
+       if ((off >= fb_start && off <= fb_end) ||
+           (off >= gart_start && off <= gart_end))
                return 0;

        /* Ok, that didn't happen... now check if we have a zero based
@@ -78,15 +80,12 @@
        }

        /* Finally, assume we aimed at a GART offset if beyond the fb */
-       if (off > (dev_priv->fb_location + dev_priv->fb_size))
-               off = off - (dev_priv->fb_location + dev_priv->fb_size) +
-                       dev_priv->gart_vm_start;
+       if (off > fb_end)
+               off = off - (fb_end + 1) + gart_start;

        /* Now recheck and fail if out of bounds */
-       if ((off >= dev_priv->fb_location &&
-            off < (dev_priv->fb_location + dev_priv->fb_size)) ||
-           (off >= dev_priv->gart_vm_start &&
-            off < (dev_priv->gart_vm_start + dev_priv->gart_size))) {
+       if ((off >= fb_start && off <= fb_end) ||
+           (off >= gart_start && off <= gart_end)) {
                DRM_DEBUG("offset fixed up to 0x%x\n", off);
                *offset = off;
                return 0;

Good,

Can you verify that this is fixed by the rawhide kernel version? If not we
should incorperate this into the Fedora kernel so that this is fixed for other
users too before FC-6 gets released.