Re: [PATCH] dma,debug: avoid deadlock between dma debug vs printk and netcons

[Christoph Hellwig]

On Mon, Aug 05, 2024 at 02:08:44PM -0400, Rik van Riel wrote:
> index a6e3792b15f8..dc96e414be38 100644
> --- a/kernel/dma/debug.c
> +++ b/kernel/dma/debug.c
> @@ -417,7 +417,7 @@ static unsigned long long phys_addr(struct dma_debug_entry *entry)
>   * other hand, consumes a single dma_debug_entry, but inserts 'nents'
>   * entries into the tree.
>   */
> -static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC);
> +static RADIX_TREE(dma_active_cacheline, GFP_ATOMIC | __GFP_NOWARN);

This looks reasonable, but please add a comment here to explain why
this is using __GFP_NOWARN so that future readers don't have to wonder
about it and look into git history.