Re: [netfilter-core] kernel panic: Out of memory and no killable processes... (2)

[Andrew Morton]

On Wed, 7 Feb 2018 18:44:39 +0100 Pablo Neira Ayuso <pablo@xxxxxxxxxxxxx> wrote:

> Hi,
> 
> On Wed, Jan 31, 2018 at 09:19:16AM +0100, Michal Hocko wrote:
> [...]
> > Yeah, we do not BUG but rather fail instead. See __vmalloc_node_range.
> > My excavation tools pointed me to "VM: Rework vmalloc code to support mapping of arbitray pages"
> > by Christoph back in 2002. So yes, we can safely remove it finally. Se
> > below.
> > 
> > 
> > From 8d52e1d939d101b0dafed6ae5c3c1376183e65bb Mon Sep 17 00:00:00 2001
> > From: Michal Hocko <mhocko@xxxxxxxx>
> > Date: Wed, 31 Jan 2018 09:16:56 +0100
> > Subject: [PATCH] net/netfilter/x_tables.c: remove size check
> > 
> > Back in 2002 vmalloc used to BUG on too large sizes. We are much better
> > behaved these days and vmalloc simply returns NULL for those. Remove
> > the check as it simply not needed and the comment even misleading.
> > 
> > Suggested-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
> > ---
> >  net/netfilter/x_tables.c | 4 ----
> >  1 file changed, 4 deletions(-)
> > 
> > diff --git a/net/netfilter/x_tables.c b/net/netfilter/x_tables.c
> > index b55ec5aa51a6..48a6ff620493 100644
> > --- a/net/netfilter/x_tables.c
> > +++ b/net/netfilter/x_tables.c
> > @@ -999,10 +999,6 @@ struct xt_table_info *xt_alloc_table_info(unsigned int size)
> >  	if (sz < sizeof(*info))
> >  		return NULL;
> >  
> > -	/* Pedantry: prevent them from hitting BUG() in vmalloc.c --RR */
> > -	if ((SMP_ALIGN(size) >> PAGE_SHIFT) + 2 > totalram_pages)
> > -		return NULL;
> > -
> >  	/* __GFP_NORETRY is not fully supported by kvmalloc but it should
> >  	 * work reasonably well if sz is too large and bail out rather
> >  	 * than shoot all processes down before realizing there is nothing
> 
> Patchwork didn't catch this patch for some reason, would you mind to
> resend?

From: Michal Hocko <mhocko@xxxxxxxx>
Subject: net/netfilter/x_tables.c: remove size check

Back in 2002 vmalloc used to BUG on too large sizes.  We are much better
behaved these days and vmalloc simply returns NULL for those.  Remove the
check as it simply not needed and the comment is even misleading.

Link: http://lkml.kernel.org/r/20180131081916.GO21609@xxxxxxxxxxxxxx
Suggested-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Michal Hocko <mhocko@xxxxxxxx>
Reviewed-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: Florian Westphal <fw@xxxxxxxxx>
Cc: David S. Miller <davem@xxxxxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---

 net/netfilter/x_tables.c |    4 ----
 1 file changed, 4 deletions(-)

diff -puN net/netfilter/x_tables.c~net-netfilter-x_tablesc-remove-size-check net/netfilter/x_tables.c
--- a/net/netfilter/x_tables.c~net-netfilter-x_tablesc-remove-size-check
+++ a/net/netfilter/x_tables.c
@@ -1004,10 +1004,6 @@ struct xt_table_info *xt_alloc_table_inf
 	if (sz < sizeof(*info))
 		return NULL;
 
-	/* Pedantry: prevent them from hitting BUG() in vmalloc.c --RR */
-	if ((size >> PAGE_SHIFT) + 2 > totalram_pages)
-		return NULL;
-
 	/* __GFP_NORETRY is not fully supported by kvmalloc but it should
 	 * work reasonably well if sz is too large and bail out rather
 	 * than shoot all processes down before realizing there is nothing
_