Re: [PATCH] x86: uv: Fix potential NULL pointer dereference of kmalloc_node

[Peter Zijlstra]

On Sat, Mar 02, 2019 at 03:09:04PM -0600, Aditya Pakki wrote:
> kmalloc_node might fail to allocate memory for thp field. This fix
> attempts to avoid a potential NULL pointer dereference.
> 
> Signed-off-by: Aditya Pakki <pakki001@xxxxxxx>
> ---
>  arch/x86/platform/uv/tlb_uv.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/arch/x86/platform/uv/tlb_uv.c b/arch/x86/platform/uv/tlb_uv.c
> index a4130b84d1ff..5a6d51e30a36 100644
> --- a/arch/x86/platform/uv/tlb_uv.c
> +++ b/arch/x86/platform/uv/tlb_uv.c
> @@ -2011,6 +2011,9 @@ static void make_per_cpu_thp(struct bau_control *smaster)
>  	size_t hpsz = sizeof(struct hub_and_pnode) * num_possible_cpus();
>  
>  	smaster->thp = kmalloc_node(hpsz, GFP_KERNEL, smaster->osnode);
> +	if (!smaster->thp)
> +		return;

This is init code; memeory allocation is 'unlikely' to fail. If it were
to fail, we'd have gotten a nice crash pinpointing the failure.

Now, we boot but get weird crashes later. Note how the rest of the code
assumes smaster->thp to be set. How is that any better?