[linuxsh-dev] SH-4 cache problem?

[NIIBE Y. <gn...@m1...>]

SUGIOKA Toshinobu wrote:
 > It seems that current cvs kernel have some cache handling problem.
 > When it boots with NFS root file system, user process does not run at all.
 > 
 > If I remove checking of PG_mapped_with_alias in flush_dcache_page(),
 > it works.

Thanks for your report.  Yup, I've found a issue for NFS.  It's not that
cache alias issue, but I-cache/D-cache coherency issue.  

We write-back the data in the GD-ROM or IDE driver.  However, NFS
doesn't do that.  This causes problem for the archtecture which has
write-back cache and I-cache/D-cache coherency issue.

Here's the patch (for me, this is work around).  Please try.  It works
for my environment.

	* arch/sh/mm/fault.c (update_mmu_cache): Flush the cache when first
	mapped, even if it has no alias.  We need this to use NFS.

Index: arch/sh/mm/fault.c
===================================================================
RCS file: /cvsroot/linuxsh/kernel/arch/sh/mm/fault.c,v
retrieving revision 1.45
diff -u -r1.45 fault.c
--- arch/sh/mm/fault.c	2001/08/07 05:20:02	1.45
+++ arch/sh/mm/fault.c	2001/08/07 08:25:41
@@ -290,14 +290,14 @@
 		return;
 
 #if defined(__SH4__)
-	if ((address ^ pte_val(pte)) & CACHE_ALIAS) {
-		page = pte_page(pte);
-		if (VALID_PAGE(page) &&
-		    !test_bit(PG_mapped_with_alias, &page->flags)) {
-			unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
-			__flush_cache_page(phys, 1);
+	page = pte_page(pte);
+	if (VALID_PAGE(page) &&
+	    !test_bit(PG_mapped_with_alias, &page->flags)) {
+		unsigned long phys = pte_val(pte) & PTE_PHYS_MASK;
+
+		__flush_cache_page(phys, 1);
+		if ((address ^ pte_val(pte)) & CACHE_ALIAS)
 			set_bit(PG_mapped_with_alias, &page->flags);
-		}
 	}
 #endif
 
--