[PATCH] guard mm->rss with page_table_lock (241p11)

From: Rasmus Andersen (rasmus@jaquet.dk)
Date: Mon Jan 29 2001 - 16:23:37 EST

Hi.

This patch tries to fix the potential rss accounting race where we
fiddle with mm->rss without holding the page_table_lock.

In addition to the places this patch touches there are some places
in fs/ where mm->rss is touched. But I am not sure of the finer
points of this code, so perhaps somebody else could have a look?
The files are binfmt_aout.c, binfmt_elf.c, and exec.c.

It applies against ac12 and 241p11.

Please comment. Or else I will continue to sumbit it :)

diff -aur linux-2.4.1-pre11-clean/mm/memory.c linux/mm/memory.c
--- linux-2.4.1-pre11-clean/mm/memory.c Sun Jan 28 20:53:13 2001
+++ linux/mm/memory.c Sun Jan 28 22:43:04 2001
@@ -377,7 +377,6 @@
                 address = (address + PGDIR_SIZE) & PGDIR_MASK;
                 dir++;
         } while (address && (address < end));
- spin_unlock(&mm->page_table_lock);
         /*
          * Update rss for the mm_struct (not necessarily current->mm)
          * Notice that rss is an unsigned long.
@@ -386,6 +385,7 @@
                 mm->rss -= freed;
         else
                 mm->rss = 0;
+ spin_unlock(&mm->page_table_lock);
 }
 
 
@@ -1038,7 +1038,9 @@
                 flush_icache_page(vma, page);
         }
 
+ spin_lock(&mm->page_table_lock);
         mm->rss++;
+ spin_unlock(&mm->page_table_lock);
 
         pte = mk_pte(page, vma->vm_page_prot);
 
@@ -1072,7 +1074,9 @@
                         return -1;
                 clear_user_highpage(page, addr);
                 entry = pte_mkwrite(pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
+ spin_lock(&mm->page_table_lock);
                 mm->rss++;
+ spin_unlock(&mm->page_table_lock);
                 flush_page_to_ram(page);
         }
         set_pte(page_table, entry);
@@ -1111,7 +1115,9 @@
                 return 0;
         if (new_page == NOPAGE_OOM)
                 return -1;
+ spin_lock(&mm->page_table_lock);
         ++mm->rss;
+ spin_unlock(&mm->page_table_lock);
         /*
          * This silly early PAGE_DIRTY setting removes a race
          * due to the bad i386 page protection. But it's valid
diff -aur linux-2.4.1-pre11-clean/mm/mmap.c linux/mm/mmap.c
--- linux-2.4.1-pre11-clean/mm/mmap.c Sat Dec 30 18:35:19 2000
+++ linux/mm/mmap.c Sun Jan 28 22:43:04 2001
@@ -879,8 +879,8 @@
         spin_lock(&mm->page_table_lock);
         mpnt = mm->mmap;
         mm->mmap = mm->mmap_avl = mm->mmap_cache = NULL;
- spin_unlock(&mm->page_table_lock);
         mm->rss = 0;
+ spin_unlock(&mm->page_table_lock);
         mm->total_vm = 0;
         mm->locked_vm = 0;
         while (mpnt) {
diff -aur linux-2.4.1-pre11-clean/mm/swapfile.c linux/mm/swapfile.c
--- linux-2.4.1-pre11-clean/mm/swapfile.c Fri Dec 29 23:07:24 2000
+++ linux/mm/swapfile.c Sun Jan 28 22:43:04 2001
@@ -231,7 +231,9 @@
         set_pte(dir, pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
         swap_free(entry);
         get_page(page);
+ spin_lock(&vma->vm_mm->page_table_lock);
         ++vma->vm_mm->rss;
+ spin_unlock(&vma->vm_mm->page_table_lock);
 }
 
 static inline void unuse_pmd(struct vm_area_struct * vma, pmd_t *dir,
diff -aur linux-2.4.1-pre11-clean/mm/vmscan.c linux/mm/vmscan.c
--- linux-2.4.1-pre11-clean/mm/vmscan.c Sun Jan 28 20:53:13 2001
+++ linux/mm/vmscan.c Mon Jan 29 22:09:18 2001
@@ -72,7 +72,9 @@
                 swap_duplicate(entry);
                 set_pte(page_table, swp_entry_to_pte(entry));
 drop_pte:
+ spin_lock(&mm->page_table_lock);
                 mm->rss--;
+ spin_unlock(&mm->page_table_lock);
                 if (!page->age)
                         deactivate_page(page);
                 UnlockPage(page); 

-- 
Regards,
        Rasmus(rasmus@jaquet.dk)

I've never had major knee surgery on any other part of my body.
-Winston Bennett, University of Kentucky basketball forward
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

This archive was generated by hypermail 2b29 : Wed Jan 31 2001 - 21:00:33 EST