diff mbox

[v2] mm: module_alloc: check if size is 0

Message ID 20130627093917.GQ7171@linux-mips.org (mailing list archive)
State New, archived
Headers show

Commit Message

Ralf Baechle June 27, 2013, 9:39 a.m. UTC 
Warming up an ancient thread because the discussion seems to have just
stalled at some point and I still have this patch bitrotting in patchwork.

The original thread can be found at:

  http://www.linux-mips.org/archives/linux-mips/2012-03/msg00006.html
  http://www.linux-mips.org/archives/linux-mips/2012-03/msg00028.html

On Wed, Mar 07, 2012 at 03:09:28PM +0200, Veli-Pekka Peltola wrote:

> After commit de7d2b567d040e3b67fe7121945982f14343213d (mm/vmalloc.c: report
> more vmalloc failures) users will get a warning if vmalloc_node_range() is
> called with size 0. This happens if module's init size equals to 0. This
> patch changes ARM, MIPS and x86 module_alloc() to return NULL before calling
> vmalloc_node_range() that would also return NULL and print a warning.
> 
> Signed-off-by: Veli-Pekka Peltola <veli-pekka.peltola@bluegiga.com>
> Cc: Russell King <linux@arm.linux.org.uk>
> Cc: Thomas Gleixner <tglx@linutronix.de>
> Cc: Ingo Molnar <mingo@redhat.com>
> Cc: "H. Peter Anvin" <hpa@zytor.com>
> Cc: x86@kernel.org
> ---
> I found this with ARM but after checking out various implementations of
> module_alloc() I thought it would be better to fix all at once.
> 
> One way to replicate the warning:
> compile kernel with CONFIG_KALLSYMS=n
> insmod a module without init, I used usb-common.ko

I didn't try to reproduce the issue but the code in question doesn't seem
to have changed so the issue should still persist.

Imho de7d2b567d040e3b67fe7121945982f14343213d [mm/vmalloc.c: report more
vmalloc failures] is overly strict in that it also reports zero-sized
allocations.  I consider such allocations stupid but legitimiate and often
better preferrable over having to scatter checks for zero size all over
place.  So maybe something like below patch?

Thanks,

  Ralf
---

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

 mm/vmalloc.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

Comments

Andrew Morton June 27, 2013, 10:23 p.m. UTC | #1 
On Thu, 27 Jun 2013 11:39:17 +0200 Ralf Baechle <ralf@linux-mips.org> wrote:

> Imho de7d2b567d040e3b67fe7121945982f14343213d [mm/vmalloc.c: report more
> vmalloc failures] is overly strict in that it also reports zero-sized
> allocations.  I consider such allocations stupid but legitimiate and often
> better preferrable over having to scatter checks for zero size all over
> place.  So maybe something like below patch?
> 
> ...
>
> --- a/mm/vmalloc.c
> +++ b/mm/vmalloc.c
> @@ -1679,7 +1679,10 @@ void *__vmalloc_node_range(unsigned long size, unsigned long align,
>  	unsigned long real_size = size;
>  
>  	size = PAGE_ALIGN(size);
> -	if (!size || (size >> PAGE_SHIFT) > totalram_pages)
> +	if (unlikely(!size))
> +		return NULL;
> +
> +	if ((size >> PAGE_SHIFT) > totalram_pages)
>  		goto fail;
>  
>  	area = __get_vm_area_node(size, align, VM_ALLOC | VM_UNLIST,
> @@ -1711,6 +1714,7 @@ fail:
>  	warn_alloc_failed(gfp_mask, 0,
>  			  "vmalloc: allocation failure: %lu bytes\n",
>  			  real_size);
> +
>  	return NULL;
>  }

If the caller actually dereferences the returned pointer the kernel
will go oops, which should provide adequate notification of a
programming error ;) But all callers should be checking the return
value.  So I worry about the by-far-most-common case where code does

	size = some_screwed_up_calculation();
	p = vmalloc(size);
	if (!p)
		return -ENOMEM;

So the mistake gets propagated back to who-knows-where as memory
exhaustion and thereby becomes a lot harder to diagnose.


How many callsites really truly need to be edited to avoid the warning?


Veli-Pekka's original patch would be neater if we were to add a new

void *__vmalloc_node_range_zero_size_ok(<args>)
{
	if (size == 0)
		return NULL;
	return __vmalloc_node_range(<args>);
}

(with a better name than __vmalloc_node_range_zero_size_ok!)
Joe Perches June 27, 2013, 10:46 p.m. UTC | #2 
On Thu, 2013-06-27 at 15:23 -0700, Andrew Morton wrote:
> On Thu, 27 Jun 2013 11:39:17 +0200 Ralf Baechle <ralf@linux-mips.org> wrote:
[]
> Veli-Pekka's original patch would be neater if we were to add a new
> 
> void *__vmalloc_node_range_zero_size_ok(<args>)
> {
> 	if (size == 0)
> 		return NULL;

I believe you mean
		return ZERO_SIZE_PTR;
Rusty Russell July 1, 2013, 3:18 a.m. UTC | #3 
Joe Perches <joe@perches.com> writes:
> On Thu, 2013-06-27 at 15:23 -0700, Andrew Morton wrote:
>> On Thu, 27 Jun 2013 11:39:17 +0200 Ralf Baechle <ralf@linux-mips.org> wrote:
> []
>> Veli-Pekka's original patch would be neater if we were to add a new
>> 
>> void *__vmalloc_node_range_zero_size_ok(<args>)
>> {
>> 	if (size == 0)
>> 		return NULL;
>
> I believe you mean
> 		return ZERO_SIZE_PTR;

Yes, this is the Right Fix.

Thanks,
Rusty.
diff mbox

Patch

diff --git a/mm/vmalloc.c b/mm/vmalloc.c
index d365724..e58ca10 100644
--- a/mm/vmalloc.c
+++ b/mm/vmalloc.c
@@ -1679,7 +1679,10 @@  void *__vmalloc_node_range(unsigned long size, unsigned long align,
 	unsigned long real_size = size;
 
 	size = PAGE_ALIGN(size);
-	if (!size || (size >> PAGE_SHIFT) > totalram_pages)
+	if (unlikely(!size))
+		return NULL;
+
+	if ((size >> PAGE_SHIFT) > totalram_pages)
 		goto fail;
 
 	area = __get_vm_area_node(size, align, VM_ALLOC | VM_UNLIST,
@@ -1711,6 +1714,7 @@  fail:
 	warn_alloc_failed(gfp_mask, 0,
 			  "vmalloc: allocation failure: %lu bytes\n",
 			  real_size);
+
 	return NULL;
 }