lkml.org 
[lkml]   [2023]   [Dec]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v10 16/50] x86/sev: Introduce snp leaked pages list
From
Hello Vlastimil,

On 12/7/2023 10:20 AM, Vlastimil Babka wrote:

>> +
>> +void snp_leak_pages(u64 pfn, unsigned int npages)
>> +{
>> + struct page *page = pfn_to_page(pfn);
>> +
>> + pr_debug("%s: leaking PFN range 0x%llx-0x%llx\n", __func__, pfn, pfn + npages);
>> +
>> + spin_lock(&snp_leaked_pages_list_lock);
>> + while (npages--) {
>> + /*
>> + * Reuse the page's buddy list for chaining into the leaked
>> + * pages list. This page should not be on a free list currently
>> + * and is also unsafe to be added to a free list.
>> + */
>> + list_add_tail(&page->buddy_list, &snp_leaked_pages_list);
>> + sev_dump_rmpentry(pfn);
>> + pfn++;
>
> You increment pfn, but not page, which is always pointing to the page of the
> initial pfn, so need to do page++ too.

Yes, that is a bug and needs to be fixed.

> But that assumes it's all order-0 pages (hard to tell for me whether that's
> true as we start with a pfn), if there can be compound pages, it would be
> best to only add the head page and skip the tail pages - it's not expected
> to use page->buddy_list of tail pages.

Can't we use PageCompound() to check if the page is a compound page and
then use page->compound_head to get and add the head page to leaked
pages list. I understand the tail pages for compound pages are really
limited for usage.

Thanks,
Ashish

\
 
 \ /
  Last update: 2023-12-08 23:11    [W:8.208 / U:1.464 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site