Re: WARNING in sk_stream_kill_queues (5)

[Marco Elver <elver@google.com>]

On Thu, Dec 03, 2020 at 07:01PM +0100, Eric Dumazet wrote:
> On 12/3/20 6:41 PM, Marco Elver wrote:
> 
> > One more experiment -- simply adding
> > 
> > --- a/net/core/skbuff.c
> > +++ b/net/core/skbuff.c
> > @@ -207,7 +207,21 @@ struct sk_buff *__alloc_skb(unsigned int size, gfp_t gfp_mask,
> >  	 */
> >  	size = SKB_DATA_ALIGN(size);
> >  	size += SKB_DATA_ALIGN(sizeof(struct skb_shared_info));
> > +	size = 1 << kmalloc_index(size); /* HACK */
> >  	data = kmalloc_reserve(size, gfp_mask, node, &pfmemalloc);
> > 
> > 
> > also got rid of the warnings. Something must be off with some value that
> > is computed in terms of ksize(). If not, I don't have any explanation
> > for why the above hides the problem.
> 
> Maybe the implementations of various macros (SKB_DATA_ALIGN and friends)
> hae some kind of assumptions, I will double check this.


Some more data; removing all uses of ksize() fixes the warnings:

| --- a/net/core/skbuff.c
| +++ b/net/core/skbuff.c
| @@ -214,7 +214,7 @@ struct sk_buff *__alloc_skb(unsigned int size, gfp_t gfp_mask,
|  	 * Put skb_shared_info exactly at the end of allocated zone,
|  	 * to allow max possible filling before reallocation.
|  	 */
| -	size = SKB_WITH_OVERHEAD(ksize(data));
| +	size = SKB_WITH_OVERHEAD(size);
|  	prefetchw(data + size);
|  
|  	/*
| @@ -1628,7 +1628,7 @@ int pskb_expand_head(struct sk_buff *skb, int nhead, int ntail,
|  			       gfp_mask, NUMA_NO_NODE, NULL);
|  	if (!data)
|  		goto nodata;
| -	size = SKB_WITH_OVERHEAD(ksize(data));
| +	size = SKB_WITH_OVERHEAD(size + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
^^ Reverting *only* this to 'ksize(data)' triggers the warning.
|  	/* Copy only real data... and, alas, header. This should be
|  	 * optimized for the cases when header is void.
| @@ -5901,7 +5901,7 @@ static int pskb_carve_inside_header(struct sk_buff *skb, const u32 off,
|  	if (!data)
|  		return -ENOMEM;
|  
| -	size = SKB_WITH_OVERHEAD(ksize(data));
| +	size = SKB_WITH_OVERHEAD(size + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
|  
|  	/* Copy real data, and all frags */
|  	skb_copy_from_linear_data_offset(skb, off, data, new_hlen);
| @@ -6025,7 +6025,7 @@ static int pskb_carve_inside_nonlinear(struct sk_buff *skb, const u32 off,
|  	if (!data)
|  		return -ENOMEM;
|  
| -	size = SKB_WITH_OVERHEAD(ksize(data));
| +	size = SKB_WITH_OVERHEAD(size + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
|  
|  	memcpy((struct skb_shared_info *)(data + size),
|  	       skb_shinfo(skb), offsetof(struct skb_shared_info, frags[0]));


Conversely, only doing this also fixes the warnings:

| --- a/net/core/skbuff.c
| +++ b/net/core/skbuff.c
| @@ -1628,7 +1628,7 @@ int pskb_expand_head(struct sk_buff *skb, int nhead, int ntail,
|  			       gfp_mask, NUMA_NO_NODE, NULL);
|  	if (!data)
|  		goto nodata;
| -	size = SKB_WITH_OVERHEAD(ksize(data));
| +	size = SKB_WITH_OVERHEAD(size + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)));
|  
|  	/* Copy only real data... and, alas, header. This should be
|  	 * optimized for the cases when header is void.

But not sure if any of this is helpful, since in the end what we want is
to make a bunch of subtractions reach precisely 0, and any deviation
somewhere might, by chance, achieve that.

Thanks,
-- Marco