Re: [PATCH 9/9] hugetlb: add corrupted hugepage counter

[Wu Fengguang <fengguang.wu@intel.com>]

> +void increment_corrupted_huge_page(struct page *page);
> +void decrement_corrupted_huge_page(struct page *page);

nitpick: increment/decrement are not verbs.

> +void increment_corrupted_huge_page(struct page *hpage)
> +{
> +	struct hstate *h = page_hstate(hpage);
> +	spin_lock(&hugetlb_lock);
> +	h->corrupted_huge_pages++;
> +	spin_unlock(&hugetlb_lock);
> +}
> +
> +void decrement_corrupted_huge_page(struct page *hpage)
> +{
> +	struct hstate *h = page_hstate(hpage);
> +	spin_lock(&hugetlb_lock);
> +	BUG_ON(!h->corrupted_huge_pages);

There is no point to have BUG_ON() here:

/*
 * Don't use BUG() or BUG_ON() unless there's really no way out; one
 * example might be detecting data structure corruption in the middle
 * of an operation that can't be backed out of.  If the (sub)system
 * can somehow continue operating, perhaps with reduced functionality,
 * it's probably not BUG-worthy.
 *
 * If you're tempted to BUG(), think again:  is completely giving up
 * really the *only* solution?  There are usually better options, where
 * users don't need to reboot ASAP and can mostly shut down cleanly.
 */


And there is a race case that (corrupted_huge_pages==0)!
Suppose the user space calls unpoison_memory() on a good pfn, and the page
happen to be hwpoisoned between lock_page() and TestClearPageHWPoison(),
corrupted_huge_pages will go negative.

Thanks,
Fengguang

> +	h->corrupted_huge_pages--;
> +	spin_unlock(&hugetlb_lock);
> +}