[uml-devel] Replacement of A-04 patch for mm/memory.c for UML/2.6

[BlaisorBlade <bla...@ya...>]

It turned out that replacing that patch is probably easier than I thought. 
This is the B-02 patch that I will post on my web space:

--- linux-2.6.0/mm/memory.c.memFix	2004-01-03 13:39:37.000000000 +0100
+++ linux-2.6.0/mm/memory.c	2004-01-03 13:42:17.000000000 +0100
@@ -44,6 +44,7 @@
 #include <linux/highmem.h>
 #include <linux/pagemap.h>
 #include <linux/rmap-locking.h>
+#include <linux/init.h>
 #include <linux/module.h>
 
 #include <asm/pgalloc.h>
@@ -680,6 +681,24 @@
 }
 
 
+static struct vm_area_struct fixmap_vma = {
+	/* Catch users - if there are any valid
+	   ones, we can make this be "&init_mm" or
+	   something.  */
+	.vm_mm = NULL,
+	.vm_page_prot = PAGE_READONLY,
+	.vm_flags = VM_READ | VM_EXEC,
+};
+
+static int init_fixmap_vma(void)
+{
+	fixmap_vma.vm_start = FIXADDR_USER_START;
+	fixmap_vma.vm_end = FIXADDR_USER_END;
+	return(0);
+}
+
+__initcall(init_fixmap_vma);
+
 int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
 		unsigned long start, int len, int write, int force,
 		struct page **pages, struct vm_area_struct **vmas)
@@ -703,16 +722,6 @@
 #ifdef FIXADDR_USER_START
 		if (!vma &&
 		    start >= FIXADDR_USER_START && start < FIXADDR_USER_END) {
-			static struct vm_area_struct fixmap_vma = {
-				/* Catch users - if there are any valid
-				   ones, we can make this be "&init_mm" or
-				   something.  */
-				.vm_mm = NULL,
-				.vm_start = FIXADDR_USER_START,
-				.vm_end = FIXADDR_USER_END,
-				.vm_page_prot = PAGE_READONLY,
-				.vm_flags = VM_READ | VM_EXEC,
-			};
 			unsigned long pg = start & PAGE_MASK;
 			pgd_t *pgd;
 			pmd_t *pmd;

The difference with the A-04 patch is that while that one replaces 
FIXADDR_USER_* with FIXADDR_*, while this one doesn't apply this change.
I've done this because I've seen that old 2.5 kernels used FIXADDR_* macros; 
then FIXADDR_USER_* was introduced but Jeff Dike, probably, didn't notice 
this.

The change was (and in this version it is) only a syntactic one.
Looking at the change itself, I think that this version makes more sense: that 
code is used to map in the address space of a process the page containing the 
vsyscall code. But it should not be possible to map the other "fixed" virtual 
pages inside a process's address space, right?

Obviously all in my very humble opinion. I'm posting this patch here for 
comments and corrections (as I'm still a newbie about Linux internals, and 
too careless at times).

Besides, after applying it, I got a non-reproducible segmentation fault the 
first time I ran the resulting binary (I've not been able to reproduce that 
crash). I'm not sure the fault was related to my code, but however this is 
the output:

$ ./linux ubd0=~/Uml/root_fs_toms1.7.205 root=/dev/ubd0

Checking for the skas3 patch in the host...found
Checking for /proc/mm...found
Linux version 2.6.0-1um (paolo@blaisorblade) (gcc version 3.2.2 (Mandrake 
Linux 9.1 3.2.2-3mdk)) #12 Sat Jan 3 13:50:46 CET 2004
On node 0 totalpages: 8192
  DMA zone: 8192 pages, LIFO batch:2
  Normal zone: 0 pages, LIFO batch:1
  HighMem zone: 0 pages, LIFO batch:1
Building zonelist for node : 0
Kernel command line: ubd0=/home/paolo/Uml/root_fs_toms1.7.205 root=/dev/ubd0
PID hash table entries: 16 (order 4: 128 bytes)
Memory: 30680k available
Calibrating delay loop... 681.57 BogoMIPS
Dentry cache hash table entries: 4096 (order: 2, 16384 bytes)
Inode-cache hash table entries: 2048 (order: 1, 8192 bytes)
Mount-cache hash table entries: 512 (order: 0, 4096 bytes)
Checking for host processor cmov support...Yes
Checking for host processor xmm support...No
Checking that ptrace can change system call numbers...OK
Segmentation fault

-- 
cat <<EOSIGN
Paolo Giarrusso, aka Blaisorblade
Linux Kernel 2.4.23/2.6.0 on an i686; Linux registered user n. 292729
EOSIGN