As I think I mentioned in my original email, when I patched the driver to limit the size to 256M, it did prevent the crash. I'm somewhat ignorant of how PCI devices are handled by X and the OS. but I expected it should be use the smaller of RADEON_CONFIG_MEMSIZE and the BAR (although if you know the BAR is never too big, that would be a good simplification).
<br><br>Thanks,<br>Robert<br><br><div><span class="gmail_quote">On 7/29/07, <b class="gmail_sendername">Daniel Stone</b> <<a href="mailto:daniel@fooishbar.org">daniel@fooishbar.org</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On Sun, Jul 29, 2007 at 10:45:41AM -0400, Alex Deucher wrote:<br>> On 7/28/07, Robert Tate <<a href="mailto:wrtate@users.sourceforge.net">wrtate@users.sourceforge.net</a>> wrote:<br>> > The gzipped radeondumps are attached. Also with respect to the crashing problem, I forgot to mention that the line "avivo->fb_size = INREG(RADEON_CONFIG_MEMSIZE);" set fb_size to 536870912 (512M), but fglrx only reports 256M.
<br>><br>> I think the max pci BAR size is 256 MB, so you'd need to clip the<br>> videoram to 256 MB. the additional vram is GPU only. Try adding a<br>> check to the driver to clip vram to 256 MB.<br><br>
Well, it should really be keying off the size of the BAR, not off<br>RADEON_CONFIG_MEMSIZE. Sorry.<br><br>Cheers,<br>Daniel<br><br>-----BEGIN PGP SIGNATURE-----<br>Version: GnuPG v1.4.6 (GNU/Linux)<br><br>iD8DBQFGrQfbUVYB1rKAgJQRAuEMAJ94idUvw1ELX60bqoEQKpM/IDs+EACglDFy
<br>3Hl6l8gtlR0KasymAElXcN0=<br>=oUBw<br>-----END PGP SIGNATURE-----<br><br></blockquote></div><br>