[patch] bogus ramdisk sanity check on ip27

[patch] bogus ramdisk sanity check on ip27

[Andrew Clausen]

Hi all,

There is currently a check that the ramdisk image resides at a sane
memory address, below "max_pfn".  However, max_pfn isn't initialized
(and apparently isn't relevant) for ip27.  The only assignments to
max_pfn lie inside #ifndef CONFIG_SGI_IP27.

Therefore, this test is bogus, so here's a patch to
#ifndef CONFIG_SGI_IP27 it.

This patch applies cleanly on 2.4.x and 2.5.x (at a different offset).

Cheers,
Andrew

diff -u -r1.31.2.25 setup.c
--- arch/mips64/kernel/setup.c	22 Jan 2003 05:11:38 -0000	1.31.2.25
+++ arch/mips64/kernel/setup.c	20 Feb 2003 00:05:41 -0000
@@ -358,6 +358,8 @@
 		printk("Initial ramdisk at: 0x%p (%lu bytes)\n",
 		       (void *)initrd_start,
 		       initrd_size);
+/* FIXME: is this right? */
+#ifndef CONFIG_SGI_IP27
 		if (CPHYSADDR(initrd_end) > PFN_PHYS(max_pfn)) {
 			printk("initrd extends beyond end of memory "
 			       "(0x%p > 0x%p)\ndisabling initrd\n",
@@ -365,6 +367,7 @@
 			       (void *)PFN_PHYS(max_pfn));
 			initrd_start = 0;
 		}
+#endif /* !CONFIG_SGI_IP27 */
 	}
 #endif
 }

Re: [patch] bogus ramdisk sanity check on ip27

[Ralf Baechle]

On Thu, Feb 20, 2003 at 11:38:39AM +1100, Andrew Clausen wrote:

> There is currently a check that the ramdisk image resides at a sane
> memory address, below "max_pfn".  However, max_pfn isn't initialized
> (and apparently isn't relevant) for ip27.  The only assignments to
> max_pfn lie inside #ifndef CONFIG_SGI_IP27.
> 
> Therefore, this test is bogus, so here's a patch to
> #ifndef CONFIG_SGI_IP27 it.
> 
> This patch applies cleanly on 2.4.x and 2.5.x (at a different offset).

No winner in a beauty contest but I think it's ok until we sort the
underlying NUMA issues which we'd have to do anyway at some point -
more ccNUMA stuff will come eventually ...

  Ralf