Re: [RFC PATCH 1/5] mm: print more information about mapping in __dump_page

[Andrew Morton]

On Wed,  7 Nov 2018 11:18:26 +0100 Michal Hocko <mhocko@xxxxxxxxxx> wrote:

> From: Michal Hocko <mhocko@xxxxxxxx>
> 
> __dump_page prints the mapping pointer but that is quite unhelpful
> for many reports because the pointer itself only helps to distinguish
> anon/ksm mappings from other ones (because of lowest bits
> set). Sometimes it would be much more helpful to know what kind of
> mapping that is actually and if we know this is a file mapping then also
> try to resolve the dentry name.
> 
> ...
>
> --- a/mm/debug.c
> +++ b/mm/debug.c
>
> ...
>
> @@ -70,6 +71,18 @@ void __dump_page(struct page *page, const char *reason)
>  	if (PageCompound(page))
>  		pr_cont(" compound_mapcount: %d", compound_mapcount(page));
>  	pr_cont("\n");
> +	if (PageAnon(page))
> +		pr_emerg("anon ");
> +	else if (PageKsm(page))
> +		pr_emerg("ksm ");
> +	else if (mapping) {
> +		pr_emerg("%ps ", mapping->a_ops);
> +		if (mapping->host->i_dentry.first) {
> +			struct dentry *dentry;
> +			dentry = container_of(mapping->host->i_dentry.first, struct dentry, d_u.d_alias);
> +			pr_emerg("name:\"%*s\" ", dentry->d_name.len, dentry->d_name.name);
> +		}
> +	}

There has to be a better way of printing the filename.  It is so often
needed.

The (poorly named and gleefully undocumented)
take_dentry_name_snapshot() looks promising.  However it's unclear that
__dump_page() is always called from contexts where
take_dentry_name_snapshot() and release_dentry_name_snapshot() can be
safely called.  Probably it's OK, but how to guarantee it?