The dynamic MAX_MP_BUSSES fix for v2.4 never made it into 2.5. x440s with
RXE100s and 32-way x445s without can easily overflow it.
Here's a static patch:
diff -pru 2.5.69/include/asm-i386/mpspec.h t69/include/asm-i386/mpspec.h
--- 2.5.69/include/asm-i386/mpspec.h Sun May 4 16:52:48 2003
+++ t69/include/asm-i386/mpspec.h Fri May 23 20:17:59 2003
@@ -191,7 +191,13 @@ struct mpc_config_translation
#define MAX_IRQ_SOURCES 256
#endif /* CONFIG_X86_NUMAQ */
+#ifdef CONFIG_X86_SUMMIT
+/* BIOS reserves an amazing number of PCI busses on x440s and x445s,
+ * especially when RXE100s are attached. */
+#define MAX_MP_BUSSES 258
+#else
#define MAX_MP_BUSSES 32
+#endif
enum mp_bustype {
MP_BUS_ISA = 1,
MP_BUS_EISA,
--
James Cleverdon
IBM xSeries Linux Solutions
{jamesclv(Unix, preferred), cleverdj(Notes)} at us dot ibm dot com
On Fri, May 23, 2003 at 08:27:26PM -0700, James Cleverdon wrote:
> The dynamic MAX_MP_BUSSES fix for v2.4 never made it into 2.5. x440s with
> RXE100s and 32-way x445s without can easily overflow it.
>
> Here's a static patch:
Umm, no. Please forward port the dynamic MAX_MP_BUSSES changes instead of
such a hack.
On Sad, 2003-05-24 at 04:27, James Cleverdon wrote:
> The dynamic MAX_MP_BUSSES fix for v2.4 never made it into 2.5. x440s with
> RXE100s and 32-way x445s without can easily overflow it.
>
> Here's a static patch:
Port the proper fix surely