From: David Hildenbrand <david-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
To: "Barret Rhoden" <brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>,
"Dan Williams"
<dan.j.williams-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>,
"Dave Jiang" <dave.jiang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>,
"Ross Zwisler" <zwisler-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>,
"Vishal Verma"
<vishal.l.verma-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>,
"Paolo Bonzini"
<pbonzini-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
"Radim Krčmář" <rkrcmar-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
"Thomas Gleixner" <tglx-hfZtesqFncYOwBW4kG4KsQ@public.gmane.org>,
"Ingo Molnar" <mingo-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
"Borislav Petkov" <bp-Gina5bIWoIWzQB+pC5nmwQ@public.gmane.org>
Cc: kvm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
yu.c.zhang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org,
linux-nvdimm-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org,
x86-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
"H. Peter Anvin" <hpa-YMNOUZJC4hwAvxtiuMwx3w@public.gmane.org>,
yi.z.zhang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org
Subject: Re: [PATCH 2/2] kvm: Use huge pages for DAX-backed files
Date: Tue, 13 Nov 2018 10:36:13 +0100 [thread overview]
Message-ID: <043a592d-6592-3053-15a0-68cc54a26deb@redhat.com> (raw)
In-Reply-To: <20181109203921.178363-3-brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
On 09.11.18 21:39, Barret Rhoden wrote:
> This change allows KVM to map DAX-backed files made of huge pages with
> huge mappings in the EPT/TDP.
>
> DAX pages are not PageTransCompound. The existing check is trying to
> determine if the mapping for the pfn is a huge mapping or not. For
> non-DAX maps, e.g. hugetlbfs, that means checking PageTransCompound.
> For DAX, we can check the page table itself.
>
> Note that KVM already faulted in the page (or huge page) in the host's
> page table, and we hold the KVM mmu spinlock (grabbed before checking
> the mmu seq).
I wonder if the KVM mmu spinlock is enough for walking (not KVM
exclusive) host page tables. Can you elaborate?
>
> Signed-off-by: Barret Rhoden <brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
> ---
> arch/x86/kvm/mmu.c | 34 ++++++++++++++++++++++++++++++++--
> 1 file changed, 32 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
> index cf5f572f2305..2df8c459dc6a 100644
> --- a/arch/x86/kvm/mmu.c
> +++ b/arch/x86/kvm/mmu.c
> @@ -3152,6 +3152,36 @@ static int kvm_handle_bad_page(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn)
> return -EFAULT;
> }
>
> +static bool pfn_is_huge_mapped(struct kvm *kvm, gfn_t gfn, kvm_pfn_t pfn)
> +{
> + struct page *page = pfn_to_page(pfn);
> + unsigned long hva, map_shift;
> +
> + if (!is_zone_device_page(page))
> + return PageTransCompoundMap(page);
> +
> + /*
> + * DAX pages do not use compound pages. The page should have already
> + * been mapped into the host-side page table during try_async_pf(), so
> + * we can check the page tables directly.
> + */
> + hva = gfn_to_hva(kvm, gfn);
> + if (kvm_is_error_hva(hva))
> + return false;
> +
> + /*
> + * Our caller grabbed the KVM mmu_lock with a successful
> + * mmu_notifier_retry, so we're safe to walk the page table.
> + */
> + map_shift = dev_pagemap_mapping_shift(hva, current->mm);
You could get rid of that local variable map_shift.
> + switch (map_shift) {
> + case PMD_SHIFT:
> + case PUD_SIZE:
> + return true;
> + }
> + return false;
> +}
> +
> static void transparent_hugepage_adjust(struct kvm_vcpu *vcpu,
> gfn_t *gfnp, kvm_pfn_t *pfnp,
> int *levelp)
> @@ -3168,7 +3198,7 @@ static void transparent_hugepage_adjust(struct kvm_vcpu *vcpu,
> */
> if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn) &&
> level == PT_PAGE_TABLE_LEVEL &&
> - PageTransCompoundMap(pfn_to_page(pfn)) &&
> + pfn_is_huge_mapped(vcpu->kvm, gfn, pfn) &&
> !mmu_gfn_lpage_is_disallowed(vcpu, gfn, PT_DIRECTORY_LEVEL)) {
> unsigned long mask;
> /*
> @@ -5678,7 +5708,7 @@ static bool kvm_mmu_zap_collapsible_spte(struct kvm *kvm,
> */
> if (sp->role.direct &&
> !kvm_is_reserved_pfn(pfn) &&
> - PageTransCompoundMap(pfn_to_page(pfn))) {
> + pfn_is_huge_mapped(kvm, sp->gfn, pfn)) {
> pte_list_remove(rmap_head, sptep);
> need_tlb_flush = 1;
> goto restart;
>
This looks surprisingly simple to me :)
--
Thanks,
David / dhildenb
next prev parent reply other threads:[~2018-11-13 9:36 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-09 20:39 [PATCH 0/2] kvm: Use huge pages for DAX-backed files Barret Rhoden
2018-11-09 20:39 ` [PATCH 1/2] mm: make dev_pagemap_mapping_shift() externally visible Barret Rhoden
[not found] ` <20181109203921.178363-2-brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2018-11-13 9:28 ` David Hildenbrand
2018-11-09 20:39 ` [PATCH 2/2] kvm: Use huge pages for DAX-backed files Barret Rhoden
2018-11-12 19:31 ` Paolo Bonzini
[not found] ` <861c4adb-e2f0-2caf-8f6e-9f09ecb0b624-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2018-11-13 16:21 ` Barret Rhoden
2018-11-13 18:22 ` Paolo Bonzini
[not found] ` <20181109203921.178363-3-brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2018-11-13 9:36 ` David Hildenbrand [this message]
[not found] ` <043a592d-6592-3053-15a0-68cc54a26deb-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2018-11-13 10:02 ` Pankaj Gupta
2018-11-13 12:41 ` Paolo Bonzini
[not found] ` <286665658.33247363.1542103353780.JavaMail.zimbra-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
2018-11-13 15:56 ` Barret Rhoden
2018-11-14 9:09 ` Pankaj Gupta
2018-11-13 15:50 ` Barret Rhoden
[not found] ` <20181109203921.178363-1-brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2018-11-14 21:51 ` [PATCH v2 0/3] " Barret Rhoden
[not found] ` <20181114215155.259978-1-brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org>
2018-11-14 21:51 ` [PATCH v2 1/3] mm: make dev_pagemap_mapping_shift() externally visible Barret Rhoden
2018-11-26 16:50 ` Paolo Bonzini
2018-11-26 18:32 ` Dan Williams
2018-11-14 21:51 ` [PATCH v2 2/3] kvm: Use huge pages for DAX-backed files Barret Rhoden
2018-11-14 21:51 ` [PATCH v2 3/3] kvm: remove redundant PageReserved() check Barret Rhoden
2018-11-27 13:31 ` David Hildenbrand
2018-11-15 0:55 ` [PATCH v2 0/3] kvm: Use huge pages for DAX-backed files Dan Williams
2018-12-03 17:40 ` Barret Rhoden
[not found] ` <20181203124051.05c1d2ce-5rj//V2oKXqT+5UN0cW69VLMcqb5oVE02SarAXORi/o@public.gmane.org>
2018-12-03 18:32 ` Alexander Duyck
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=043a592d-6592-3053-15a0-68cc54a26deb@redhat.com \
--to=david-h+wxahxf7alqt0dzr+alfa@public.gmane.org \
--cc=bp-Gina5bIWoIWzQB+pC5nmwQ@public.gmane.org \
--cc=brho-hpIqsD4AKlfQT0dZR+AlfA@public.gmane.org \
--cc=dan.j.williams-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
--cc=dave.jiang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
--cc=hpa-YMNOUZJC4hwAvxtiuMwx3w@public.gmane.org \
--cc=kvm-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
--cc=linux-nvdimm-hn68Rpc1hR1g9hUCZPvPmw@public.gmane.org \
--cc=mingo-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
--cc=pbonzini-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
--cc=rkrcmar-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
--cc=tglx-hfZtesqFncYOwBW4kG4KsQ@public.gmane.org \
--cc=vishal.l.verma-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
--cc=x86-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org \
--cc=yi.z.zhang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
--cc=yu.c.zhang-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org \
--cc=zwisler-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.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).