From: George Dunlap <george.dunlap@eu.citrix.com>
To: "Liu, Jinsong" <jinsong.liu@intel.com>
Cc: Christoph Egger <Christoph.Egger@amd.com>,
"xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>,
"Keir (Xen.org)" <keir@xen.org>,
Ian Campbell <Ian.Campbell@citrix.com>,
Ian Jackson <Ian.Jackson@eu.citrix.com>,
Jan Beulich <JBeulich@suse.com>
Subject: Re: [Patch 4/5] X86/vMCE: handle broken page occurred before migration
Date: Tue, 30 Oct 2012 10:25:07 +0100 [thread overview]
Message-ID: <508F9CF3.3090706@eu.citrix.com> (raw)
In-Reply-To: <DE8DF0795D48FD4CA783C40EC8292335374280@SHSMSX101.ccr.corp.intel.com>
Jinsong,
I'm at UDS now, but I'll try to review the new patches in the next few days.
If you end up sending these patches again, could you please send them in
a more normal "patchbomb-style" format? I.e., with a "00/02" header
describing what's new in the series, and then naming them 01/02 and
02/02 (instead of 4 and 5, when 1-3 have been applied for months)?
The easiest way to do this is to use hg's patchbomb extension; there's a
description of how to set it up here:
http://wiki.xen.org/wiki/SubmittingXenPatches
It's a few minutes to set up, but it's well worth it both for us and for
you.
Thanks,
-George
On 29/10/12 16:21, Liu, Jinsong wrote:
> X86/vMCE: handle broken page occurred before migration
>
> This patch handles guest broken page which occur before migration.
>
> At sender, the broken page would be mapped but not copied to target
> (otherwise it may trigger more serious error, say, SRAR error).
> While its pfn_type and pfn number would be transferred to target
> so that target take appropriate action.
>
> At target, it would set p2m as p2m_ram_broken for broken page, so that
> if guest access the broken page again, it would kill itself as expected.
>
> Signed-off-by: Liu, Jinsong <jinsong.liu@intel.com>
>
> diff -r e27a6d53ac15 tools/libxc/xc_domain.c
> --- a/tools/libxc/xc_domain.c Thu Oct 11 01:52:33 2012 +0800
> +++ b/tools/libxc/xc_domain.c Thu Oct 25 05:49:10 2012 +0800
> @@ -283,6 +283,22 @@
> return ret;
> }
>
> +/* set broken page p2m */
> +int xc_set_broken_page_p2m(xc_interface *xch,
> + uint32_t domid,
> + unsigned long pfn)
> +{
> + int ret;
> + DECLARE_DOMCTL;
> +
> + domctl.cmd = XEN_DOMCTL_set_broken_page_p2m;
> + domctl.domain = (domid_t)domid;
> + domctl.u.set_broken_page_p2m.pfn = pfn;
> + ret = do_domctl(xch, &domctl);
> +
> + return ret ? -1 : 0;
> +}
> +
> /* get info from hvm guest for save */
> int xc_domain_hvm_getcontext(xc_interface *xch,
> uint32_t domid,
> diff -r e27a6d53ac15 tools/libxc/xc_domain_restore.c
> --- a/tools/libxc/xc_domain_restore.c Thu Oct 11 01:52:33 2012 +0800
> +++ b/tools/libxc/xc_domain_restore.c Thu Oct 25 05:49:10 2012 +0800
> @@ -962,9 +962,15 @@
>
> countpages = count;
> for (i = oldcount; i < buf->nr_pages; ++i)
> - if ((buf->pfn_types[i] & XEN_DOMCTL_PFINFO_LTAB_MASK) == XEN_DOMCTL_PFINFO_XTAB
> - ||(buf->pfn_types[i] & XEN_DOMCTL_PFINFO_LTAB_MASK) == XEN_DOMCTL_PFINFO_XALLOC)
> + {
> + unsigned long pagetype;
> +
> + pagetype = buf->pfn_types[i] & XEN_DOMCTL_PFINFO_LTAB_MASK;
> + if ( pagetype == XEN_DOMCTL_PFINFO_XTAB ||
> + pagetype == XEN_DOMCTL_PFINFO_BROKEN ||
> + pagetype == XEN_DOMCTL_PFINFO_XALLOC )
> --countpages;
> + }
>
> if (!countpages)
> return count;
> @@ -1200,6 +1206,17 @@
> /* a bogus/unmapped/allocate-only page: skip it */
> continue;
>
> + if ( pagetype == XEN_DOMCTL_PFINFO_BROKEN )
> + {
> + if ( xc_set_broken_page_p2m(xch, dom, pfn) )
> + {
> + ERROR("Set p2m for broken page failed, "
> + "dom=%d, pfn=%lx\n", dom, pfn);
> + goto err_mapped;
> + }
> + continue;
> + }
> +
> if (pfn_err[i])
> {
> ERROR("unexpected PFN mapping failure pfn %lx map_mfn %lx p2m_mfn %lx",
> diff -r e27a6d53ac15 tools/libxc/xc_domain_save.c
> --- a/tools/libxc/xc_domain_save.c Thu Oct 11 01:52:33 2012 +0800
> +++ b/tools/libxc/xc_domain_save.c Thu Oct 25 05:49:10 2012 +0800
> @@ -1277,6 +1277,13 @@
> if ( !hvm )
> gmfn = pfn_to_mfn(gmfn);
>
> + if ( pfn_type[j] == XEN_DOMCTL_PFINFO_BROKEN )
> + {
> + pfn_type[j] |= pfn_batch[j];
> + ++run;
> + continue;
> + }
> +
> if ( pfn_err[j] )
> {
> if ( pfn_type[j] == XEN_DOMCTL_PFINFO_XTAB )
> @@ -1371,8 +1378,12 @@
> }
> }
>
> - /* skip pages that aren't present or are alloc-only */
> + /*
> + * skip pages that aren't present,
> + * or are broken, or are alloc-only
> + */
> if ( pagetype == XEN_DOMCTL_PFINFO_XTAB
> + || pagetype == XEN_DOMCTL_PFINFO_BROKEN
> || pagetype == XEN_DOMCTL_PFINFO_XALLOC )
> continue;
>
> diff -r e27a6d53ac15 tools/libxc/xenctrl.h
> --- a/tools/libxc/xenctrl.h Thu Oct 11 01:52:33 2012 +0800
> +++ b/tools/libxc/xenctrl.h Thu Oct 25 05:49:10 2012 +0800
> @@ -575,6 +575,17 @@
> xc_domaininfo_t *info);
>
> /**
> + * This function set p2m for broken page
> + * &parm xch a handle to an open hypervisor interface
> + * @parm domid the domain id which broken page belong to
> + * @parm pfn the pfn number of the broken page
> + * @return 0 on success, -1 on failure
> + */
> +int xc_set_broken_page_p2m(xc_interface *xch,
> + uint32_t domid,
> + unsigned long pfn);
> +
> +/**
> * This function returns information about the context of a hvm domain
> * @parm xch a handle to an open hypervisor interface
> * @parm domid the domain to get information from
> diff -r e27a6d53ac15 xen/arch/x86/domctl.c
> --- a/xen/arch/x86/domctl.c Thu Oct 11 01:52:33 2012 +0800
> +++ b/xen/arch/x86/domctl.c Thu Oct 25 05:49:10 2012 +0800
> @@ -209,12 +209,18 @@
> for ( j = 0; j < k; j++ )
> {
> unsigned long type = 0;
> + p2m_type_t t;
>
> - page = get_page_from_gfn(d, arr[j], NULL, P2M_ALLOC);
> + page = get_page_from_gfn(d, arr[j], &t, P2M_ALLOC);
>
> if ( unlikely(!page) ||
> unlikely(is_xen_heap_page(page)) )
> - type = XEN_DOMCTL_PFINFO_XTAB;
> + {
> + if ( p2m_is_broken(t) )
> + type = XEN_DOMCTL_PFINFO_BROKEN;
> + else
> + type = XEN_DOMCTL_PFINFO_XTAB;
> + }
> else
> {
> switch( page->u.inuse.type_info & PGT_type_mask )
> @@ -235,6 +241,9 @@
>
> if ( page->u.inuse.type_info & PGT_pinned )
> type |= XEN_DOMCTL_PFINFO_LPINTAB;
> +
> + if ( page->count_info & PGC_broken )
> + type = XEN_DOMCTL_PFINFO_BROKEN;
> }
>
> if ( page )
> @@ -1568,6 +1577,28 @@
> }
> break;
>
> + case XEN_DOMCTL_set_broken_page_p2m:
> + {
> + struct domain *d;
> + p2m_type_t pt;
> + unsigned long pfn;
> +
> + d = rcu_lock_domain_by_id(domctl->domain);
> + if ( d != NULL )
> + {
> + pfn = domctl->u.set_broken_page_p2m.pfn;
> +
> + get_gfn_query(d, pfn, &pt);
> + p2m_change_type(d, pfn, pt, p2m_ram_broken);
> + put_gfn(d, pfn);
> +
> + rcu_unlock_domain(d);
> + }
> + else
> + ret = -ESRCH;
> + }
> + break;
> +
> default:
> ret = iommu_do_domctl(domctl, u_domctl);
> break;
> diff -r e27a6d53ac15 xen/include/public/domctl.h
> --- a/xen/include/public/domctl.h Thu Oct 11 01:52:33 2012 +0800
> +++ b/xen/include/public/domctl.h Thu Oct 25 05:49:10 2012 +0800
> @@ -136,6 +136,7 @@
> #define XEN_DOMCTL_PFINFO_LPINTAB (0x1U<<31)
> #define XEN_DOMCTL_PFINFO_XTAB (0xfU<<28) /* invalid page */
> #define XEN_DOMCTL_PFINFO_XALLOC (0xeU<<28) /* allocate-only page */
> +#define XEN_DOMCTL_PFINFO_BROKEN (0xdU<<28) /* broken page */
> #define XEN_DOMCTL_PFINFO_PAGEDTAB (0x8U<<28)
> #define XEN_DOMCTL_PFINFO_LTAB_MASK (0xfU<<28)
>
> @@ -835,6 +836,12 @@
> typedef struct xen_domctl_set_access_required xen_domctl_set_access_required_t;
> DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_access_required_t);
>
> +struct xen_domctl_set_broken_page_p2m {
> + uint64_aligned_t pfn;
> +};
> +typedef struct xen_domctl_set_broken_page_p2m xen_domctl_set_broken_page_p2m_t;
> +DEFINE_XEN_GUEST_HANDLE(xen_domctl_set_broken_page_p2m_t);
> +
> struct xen_domctl {
> uint32_t cmd;
> #define XEN_DOMCTL_createdomain 1
> @@ -900,6 +907,7 @@
> #define XEN_DOMCTL_set_access_required 64
> #define XEN_DOMCTL_audit_p2m 65
> #define XEN_DOMCTL_set_virq_handler 66
> +#define XEN_DOMCTL_set_broken_page_p2m 67
> #define XEN_DOMCTL_gdbsx_guestmemio 1000
> #define XEN_DOMCTL_gdbsx_pausevcpu 1001
> #define XEN_DOMCTL_gdbsx_unpausevcpu 1002
> @@ -955,6 +963,7 @@
> struct xen_domctl_audit_p2m audit_p2m;
> struct xen_domctl_set_virq_handler set_virq_handler;
> struct xen_domctl_gdbsx_memio gdbsx_guest_memio;
> + struct xen_domctl_set_broken_page_p2m set_broken_page_p2m;
> struct xen_domctl_gdbsx_pauseunp_vcpu gdbsx_pauseunp_vcpu;
> struct xen_domctl_gdbsx_domstatus gdbsx_domstatus;
> uint8_t pad[128];
next prev parent reply other threads:[~2012-10-30 9:25 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-10-10 14:46 [PATCH 4/5] Xen/MCE: Abort live migration when vMCE occur Liu, Jinsong
2012-10-16 10:42 ` Liu, Jinsong
2012-10-19 14:52 ` Ian Jackson
2012-10-19 20:13 ` Liu, Jinsong
2012-10-22 11:32 ` Liu, Jinsong
2012-10-25 11:21 ` Ian Jackson
2012-10-25 12:32 ` Jan Beulich
2012-10-25 12:38 ` Ian Jackson
2012-10-25 12:44 ` Liu, Jinsong
2012-10-25 12:51 ` Ian Jackson
2012-10-19 16:51 ` George Dunlap
2012-10-19 20:32 ` Liu, Jinsong
2012-10-22 11:32 ` George Dunlap
2012-10-24 14:30 ` Liu, Jinsong
2012-10-29 15:21 ` [Patch 4/5] X86/vMCE: handle broken page occurred before migration Liu, Jinsong
2012-10-29 16:35 ` Jan Beulich
2012-10-29 17:19 ` Liu, Jinsong
2012-10-30 9:02 ` Jan Beulich
2012-10-31 10:55 ` Liu, Jinsong
2012-10-30 9:25 ` George Dunlap [this message]
2012-10-30 9:27 ` George Dunlap
2012-10-31 10:58 ` Liu, Jinsong
2012-10-29 15:22 ` [PATCH 5/5] Xen/MCE: handle broken page occurs during migration Liu, Jinsong
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=508F9CF3.3090706@eu.citrix.com \
--to=george.dunlap@eu.citrix.com \
--cc=Christoph.Egger@amd.com \
--cc=Ian.Campbell@citrix.com \
--cc=Ian.Jackson@eu.citrix.com \
--cc=JBeulich@suse.com \
--cc=jinsong.liu@intel.com \
--cc=keir@xen.org \
--cc=xen-devel@lists.xensource.com \
/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).