From: David Woodhouse <dwmw2@infradead.org>
To: George Dunlap <george.dunlap@citrix.com>,
Jan Beulich <jbeulich@suse.com>,
Stewart Hildebrand <stewart.hildebrand@dornerworks.com>
Cc: Stefano Stabellini <sstabellini@kernel.org>,
Julien Grall <julien@xen.org>, Wei Liu <wl@xen.org>,
Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>,
George Dunlap <George.Dunlap@eu.citrix.com>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Ian Jackson <ian.jackson@eu.citrix.com>,
Jeff Kubascik <jeff.kubascik@dornerworks.com>,
xen-devel@lists.xenproject.org
Subject: Re: [Xen-devel] [XEN PATCH v2 1/2] Check zone before merging adjacent blocks in heap
Date: Wed, 05 Feb 2020 09:50:21 +0000 [thread overview]
Message-ID: <81c41a686f6a520936342f04b62d7c27c137f57e.camel@infradead.org> (raw)
In-Reply-To: <428eb589-ec3d-1527-02ce-92e2d4367a1a@citrix.com>
[-- Attachment #1.1: Type: text/plain, Size: 2121 bytes --]
On Tue, 2020-02-04 at 15:37 +0000, George Dunlap wrote:
> At very least it's more robust this way; the algorithm is also less
> "magic". We just had a long discussion this morning trying to re-create
> the logic for why "Remove MFN 0" was sufficient to prevent this issue,
> and even then David wasn't sure it was correct at first.
Right. So the real reason I'm staring hard at this is because I can't
convince myself there aren't places where memory allocated by the boot
allocator is later freed with free_xenheap_pages().
We have a few pieces of code which decide whether to use the boot
allocator vs. heap based on system_state >= SYS_STATE_boot, and *if*
the rule is "thou shalt not allocate with boot allocator and free
later" then it's *extremely* fragile and probably being violated —
especially because it actually *works* most of the time, except in some
esoteric corner cases like MFN#0, boot allocations which cross
zones/regions, etc.
So because we want to make that *more* likely by allowing vmap() to
happen earlier, I'd like to clean things up by addressing those corner
cases and making it unconditionally OK to free boot-allocated pages
into the heap.
I think might be as simple as checking for (first_pg)->count_info == 0
in free_xenheap_pages(). That's quick enough, and if the count_info is
zero then I think it does indicate a boot-allocated page, because pages
from alloc_xenheap_pages() would have PGC_xen_heap set?
It would suffice just to pass such pages to init_heap_pages() instead
of directly to free_heap_pages(), I think. Julien?
The straw man version of that looks a bit like this...
--- a/xen/common/page_alloc.c
+++ b/xen/common/page_alloc.c
@@ -2304,6 +2304,12 @@ void free_xenheap_pages(void *v, unsigned int order)
pg = virt_to_page(v);
+ /* Pages from the boot allocator need to pass through init_heap_pages() */
+ if ( unlikely(!pg->count_info) )
+ {
+ init_heap_pages(pg, 1 << order);
+ return;
+ }
for ( i = 0; i < (1u << order); i++ )
pg[i].count_info &= ~PGC_xen_heap;
[-- Attachment #1.2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 5174 bytes --]
[-- Attachment #2: Type: text/plain, Size: 157 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xenproject.org
https://lists.xenproject.org/mailman/listinfo/xen-devel
next prev parent reply other threads:[~2020-02-05 9:51 UTC|newest]
Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-04 15:14 [Xen-devel] [XEN PATCH v2 1/2] Check zone before merging adjacent blocks in heap Stewart Hildebrand
2020-02-04 15:14 ` [Xen-devel] [DO NOT APPLY XEN PATCH v2 2/2] Test case for buddy allocator merging issue Stewart Hildebrand
2020-02-04 15:22 ` [Xen-devel] [XEN PATCH v2 1/2] Check zone before merging adjacent blocks in heap Jan Beulich
2020-02-04 15:37 ` George Dunlap
2020-02-05 9:50 ` David Woodhouse [this message]
2020-02-05 10:02 ` Jan Beulich
2020-02-05 10:24 ` David Woodhouse
2020-02-05 10:49 ` Jan Beulich
2020-02-05 11:23 ` David Woodhouse
2020-02-05 13:37 ` Jan Beulich
2020-02-05 14:12 ` David Woodhouse
2020-02-07 15:49 ` David Woodhouse
2020-02-07 15:57 ` [Xen-devel] [PATCH 1/2] xen/mm: fold PGC_broken into PGC_state bits David Woodhouse
2020-02-07 20:27 ` Julien Grall
2020-02-09 13:22 ` David Woodhouse
2020-02-09 17:59 ` Julien Grall
2020-03-17 21:39 ` David Woodhouse
2020-02-20 11:10 ` Jan Beulich
2020-03-17 21:52 ` David Woodhouse
2020-03-18 9:56 ` Jan Beulich
2020-03-18 12:31 ` Julien Grall
2020-03-18 13:23 ` Jan Beulich
2020-03-18 17:13 ` David Woodhouse
2020-03-19 8:49 ` Jan Beulich
2020-03-19 10:26 ` David Woodhouse
2020-03-19 11:59 ` Jan Beulich
2020-03-19 13:54 ` David Woodhouse
2020-03-19 14:46 ` Jan Beulich
2020-02-07 15:57 ` [Xen-devel] [PATCH 2/2] xen/mm: Introduce PG_state_uninitialised David Woodhouse
2020-02-07 16:30 ` Xia, Hongyan
2020-02-07 16:32 ` David Woodhouse
2020-02-07 16:40 ` Xia, Hongyan
2020-02-07 17:06 ` David Woodhouse
2020-02-07 18:04 ` David Woodhouse
2020-02-20 11:59 ` Jan Beulich
2020-02-20 13:27 ` Julien Grall
2020-03-17 22:15 ` David Woodhouse
2020-03-18 8:53 ` Paul Durrant
2020-03-18 10:10 ` Jan Beulich
2020-03-18 10:41 ` Paul Durrant
2020-03-18 11:12 ` Jan Beulich
2020-03-18 10:03 ` Jan Beulich
2020-03-18 12:11 ` David Woodhouse
2020-03-18 13:27 ` Jan Beulich
2020-02-05 10:22 ` [Xen-devel] [XEN PATCH v2 1/2] Check zone before merging adjacent blocks in heap Julien Grall
2020-02-05 10:32 ` David Woodhouse
2020-02-05 11:36 ` David Woodhouse
2020-02-04 15:37 ` Stewart Hildebrand
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=81c41a686f6a520936342f04b62d7c27c137f57e.camel@infradead.org \
--to=dwmw2@infradead.org \
--cc=George.Dunlap@eu.citrix.com \
--cc=andrew.cooper3@citrix.com \
--cc=george.dunlap@citrix.com \
--cc=ian.jackson@eu.citrix.com \
--cc=jbeulich@suse.com \
--cc=jeff.kubascik@dornerworks.com \
--cc=julien@xen.org \
--cc=konrad.wilk@oracle.com \
--cc=sstabellini@kernel.org \
--cc=stewart.hildebrand@dornerworks.com \
--cc=wl@xen.org \
--cc=xen-devel@lists.xenproject.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).