[PATCH 1/2] remap file pages MAP_NONBLOCK fix

[PATCH 1/2] remap file pages MAP_NONBLOCK fix

[Rajesh Venkatasubramanian]

Hi Andrew, Ingo,

The remap_file_pages system call with MAP_NONBLOCK flag does not
install file-ptes when the required pages are not found in the
page cache. Modify the populate functions to install file-ptes
if the mapping is non-linear and the required pages are not found
in the page cache.

Patch is for test4-mm6. Compiles and boots. Patch tested using the
programs at:

http://www-personal.engin.umich.edu/~vrajesh/linux/remap-file-pages/

Please apply if appropriate.

Thanks,
Rajesh



 include/linux/mm.h |    1 +
 mm/filemap.c       |   14 ++++++++++++++
 mm/fremap.c        |   39 +++++++++++++++++++++++++++++++++++++++
 mm/shmem.c         |   15 +++++++++++++++
 4 files changed, 69 insertions(+)

diff -puN mm/fremap.c~fremap-MAP_NONBLOCK mm/fremap.c
--- dev-2.6.0-test4/mm/fremap.c~fremap-MAP_NONBLOCK	Fri Sep  5 12:57:52 2003
+++ dev-2.6.0-test4-vrajesh/mm/fremap.c	Fri Sep  5 12:57:52 2003
@@ -100,6 +100,45 @@ err:
 EXPORT_SYMBOL(install_page);
 
 
+/*
+ * Install a file pte to a given virtual memory address, release any
+ * previously existing mapping.
+ */
+int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma,
+		unsigned long addr, unsigned long pgoff, pgprot_t prot)
+{
+	int err = -ENOMEM, flush;
+	pte_t *pte;
+	pgd_t *pgd;
+	pmd_t *pmd;
+
+	pgd = pgd_offset(mm, addr);
+	spin_lock(&mm->page_table_lock);
+
+	pmd = pmd_alloc(mm, pgd, addr);
+	if (!pmd)
+		goto err_unlock;
+
+	pte = pte_alloc_map(mm, pmd, addr);
+	if (!pte)
+		goto err_unlock;
+
+	flush = zap_pte(mm, vma, addr, pte);
+
+	set_pte(pte, pgoff_to_pte(pgoff));
+	pte_unmap(pte);
+	if (flush)
+		flush_tlb_page(vma, addr);
+	update_mmu_cache(vma, addr, *pte);
+	spin_unlock(&mm->page_table_lock);
+	return 0;
+
+err_unlock:
+	spin_unlock(&mm->page_table_lock);
+	return err;
+}
+
+
 /***
  * sys_remap_file_pages - remap arbitrary pages of a shared backing store
  *                        file within an existing vma.
diff -puN include/linux/mm.h~fremap-MAP_NONBLOCK include/linux/mm.h
--- dev-2.6.0-test4/include/linux/mm.h~fremap-MAP_NONBLOCK	Fri Sep  5 12:57:52 2003
+++ dev-2.6.0-test4-vrajesh/include/linux/mm.h	Fri Sep  5 12:57:52 2003
@@ -431,6 +431,7 @@ extern pmd_t *FASTCALL(__pmd_alloc(struc
 extern pte_t *FASTCALL(pte_alloc_kernel(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
 extern pte_t *FASTCALL(pte_alloc_map(struct mm_struct *mm, pmd_t *pmd, unsigned long address));
 extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot);
+extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot);
 extern int handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma, unsigned long address, int write_access);
 extern int make_pages_present(unsigned long addr, unsigned long end);
 extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
diff -puN mm/filemap.c~fremap-MAP_NONBLOCK mm/filemap.c
--- dev-2.6.0-test4/mm/filemap.c~fremap-MAP_NONBLOCK	Fri Sep  5 12:57:52 2003
+++ dev-2.6.0-test4-vrajesh/mm/filemap.c	Fri Sep  5 12:57:52 2003
@@ -1363,6 +1363,20 @@ repeat:
 			page_cache_release(page);
 			return err;
 		}
+	} else {
+	    	/*
+		 * If a nonlinear mapping then store the file page offset
+		 * in the pte.
+		 */
+	    	unsigned long pgidx;
+		pgidx = (addr - vma->vm_start) >> PAGE_SHIFT;
+		pgidx += vma->vm_pgoff;
+		pgidx >>= PAGE_CACHE_SHIFT - PAGE_SHIFT;
+		if (pgoff != pgidx) {
+	    		err = install_file_pte(mm, vma, addr, pgoff, prot);
+			if (err)
+		    		return err;
+		}
 	}
 
 	len -= PAGE_SIZE;
diff -puN mm/shmem.c~fremap-MAP_NONBLOCK mm/shmem.c
--- dev-2.6.0-test4/mm/shmem.c~fremap-MAP_NONBLOCK	Fri Sep  5 12:57:52 2003
+++ dev-2.6.0-test4-vrajesh/mm/shmem.c	Fri Sep  5 12:57:52 2003
@@ -984,7 +984,22 @@ static int shmem_populate(struct vm_area
 				page_cache_release(page);
 				return err;
 			}
+		} else if (nonblock) {
+	    		/*
+		 	 * If a nonlinear mapping then store the file page 
+			 * offset in the pte.  
+			 */
+	    		unsigned long pgidx;
+			pgidx = (addr - vma->vm_start) >> PAGE_SHIFT;
+			pgidx += vma->vm_pgoff;
+			pgidx >>= PAGE_CACHE_SHIFT - PAGE_SHIFT; 
+			if (pgoff != pgidx) {
+	    			err = install_file_pte(mm, vma, addr, pgoff, prot);
+				if (err)
+		    			return err;
+			}
 		}
+	
 		len -= PAGE_SIZE;
 		addr += PAGE_SIZE;
 		pgoff++;

_


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: [PATCH 1/2] remap file pages MAP_NONBLOCK fix

[Jeff Garzik]

On Fri, Sep 05, 2003 at 02:31:37PM -0400, Rajesh Venkatasubramanian wrote:
> Hi Andrew, Ingo,
> 
> The remap_file_pages system call with MAP_NONBLOCK flag does not
> install file-ptes when the required pages are not found in the
> page cache. Modify the populate functions to install file-ptes
> if the mapping is non-linear and the required pages are not found
> in the page cache.

Or we could just kill remap_file_pages(), because it's a PITA to
maintain and it has maybe 10 legitimate users in the entire world...

	Jeff



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: [PATCH 1/2] remap file pages MAP_NONBLOCK fix

[Martin J. Bligh]

>> Or we could just kill remap_file_pages(), because it's a PITA to
>> maintain and it has maybe 10 legitimate users in the entire world...
> 
> It has its uses.  I just don't think it should support
> a non-mlocked VMA or deal with files being truncated.
> 
> The remap_file_pages we have in Taroon is pretty light
> weight. I hope the implementation in 2.6 will be
> simplified too ;)

We did actually agree to do this at Kernel Summit too ....

M.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: [PATCH 1/2] remap file pages MAP_NONBLOCK fix

[Rik van Riel]

On Fri, 5 Sep 2003, Jeff Garzik wrote:

> Or we could just kill remap_file_pages(), because it's a PITA to
> maintain and it has maybe 10 legitimate users in the entire world...

It has its uses.  I just don't think it should support
a non-mlocked VMA or deal with files being truncated.

The remap_file_pages we have in Taroon is pretty light
weight. I hope the implementation in 2.6 will be
simplified too ;)

-- 
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: [PATCH 1/2] remap file pages MAP_NONBLOCK fix

[Ingo Molnar]

On Fri, 5 Sep 2003, Martin J. Bligh wrote:

> >> Or we could just kill remap_file_pages(), because it's a PITA to
> >> maintain and it has maybe 10 legitimate users in the entire world...
> > 
> > It has its uses.  I just don't think it should support
> > a non-mlocked VMA or deal with files being truncated.
> > 
> > The remap_file_pages we have in Taroon is pretty light
> > weight. I hope the implementation in 2.6 will be
> > simplified too ;)
> 
> We did actually agree to do this at Kernel Summit too ....

yep - i did most of the simplification for 2.4 and it worked out well.  
Having fremap a non-swappable thing lessens the impact on the rest of the
VM quite significantly.

	Ingo
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>