Re: [PATCH 1/3] mm: Print head flags in dump_page

[John Hubbard <jhubbard@nvidia.com>]

On 2020-06-29 15:51, Matthew Wilcox wrote:
> On Mon, Jun 29, 2020 at 03:38:13PM -0700, John Hubbard wrote:
>> On 2020-06-29 08:19, Matthew Wilcox (Oracle) wrote:
>>> Tail page flags contain very little useful information.  Print the head
>>> page's flags instead (even though PageHead is a little misleading).
>>
>> You are right about the tail page. And the raw output provides the tail
>> page flags, in case someone *really* needs to dig into tail page problems,
>> so that's all good.
>>
>> However, I just gave this a spin, and seeing the "|head" in the list for
>> my "tail page: dump_page test" is also slightly misleading for me, too.
> 
> We could also do ...
> 
> @@ -48,6 +48,8 @@ void __dump_page(struct page *page, const char *reason)
>          struct address_space *mapping;
>          bool page_poisoned = PagePoisoned(page);
>          bool compound = PageCompound(page);
> +       unsigned long flags = page->flags;
> +
>          /*
>           * Accessing the pageblock without the zone lock. It could change to
>           * "isolate" again in the meantime, but since we are just dumping the
> @@ -165,7 +162,9 @@ void __dump_page(struct page *page, const char *reason)
>   out_mapping:
>          BUILD_BUG_ON(ARRAY_SIZE(pageflag_names) != __NR_PAGEFLAGS + 1);
>   
> -       pr_warn("%sflags: %#lx(%pGp)%s\n", type, head->flags, &head->flags,
> +       if (head != page)
> +               flags = head->flags & ~PG_head;
> +       pr_warn("%sflags: %#lx(%pGp)%s\n", type, flags, &flags,
>                  page_cma ? " CMA" : "");
>   

OK, I'll give this a spin, along with your line break approach...

thanks,
-- 
John Hubbard
NVIDIA