linux-kernel - Re: [Patch v2 3/4] mm/page_alloc.c: pass all bad reasons to bad

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 21 Jan 2020 09:47:14 +0100
From:   Michal Hocko <mhocko@...nel.org>
To:     Anshuman Khandual <anshuman.khandual@....com>
Cc:     Wei Yang <richardw.yang@...ux.intel.com>,
        akpm@...ux-foundation.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, rientjes@...gle.com
Subject: Re: [Patch v2 3/4] mm/page_alloc.c: pass all bad reasons to
 bad_page()

On Tue 21-01-20 11:38:29, Anshuman Khandual wrote:
> 
> 
> On 01/20/2020 03:52 PM, Michal Hocko wrote:
> > On Mon 20-01-20 11:04:14, Wei Yang wrote:
> >> Now we can pass all bad reasons to __dump_page().
> > And we do we want to do that? The dump of the page will tell us the
> > whole story so a single and the most important reason sounds like a
> > better implementation. The code is also more subtle because each caller
> > of the function has to be aware of how many reasons there might be.
> > Not to mention that you need a room for 5 pointers on the stack and this
> > and page allocator might be called from deeper call chains.
> > 
> 
> Two paths which lead to __dump_page(), dump_page() and bad_page().
> Callers of dump_page() can give a single reason what they consider the
> most important which leads to page dumping. This makes sense but gets
> trickier in bad_page() path. At present, free_pages_check_bad() and
> check_new_page_bad() has a sequence of 'if' statements which decides
> "most important" reason for __dump_page() without much rationale and
> similar in case of free_tail_pages_check() as well. As all information
> about the page for corresponding reasons are printed with __dump_page()
> anyways, do free_pages_check_bad() or check_new_page_bad() really need
> to provide any particular single reason ?

Do you see any particular problem with the existing logic? I find a
single reason sufficient and a good lead for what to check most of the
time.
-- 
Michal Hocko
SUSE Labs