xen-devel.lists.xenproject.org archive mirror
 help / color / mirror / Atom feed
From: Lin Ming <mlin@ss.pku.edu.cn>
To: Ian Campbell <Ian.Campbell@citrix.com>
Cc: "xen-devel@lists.xensource.com" <xen-devel@lists.xensource.com>,
	Teck Choon Giam <giamteckchoon@gmail.com>
Subject: Re: Upgrade from xen-4.1-testing to xen-unstable report
Date: Thu, 12 Apr 2012 22:42:53 +0800	[thread overview]
Message-ID: <1334241773.4001.5.camel@hp6530s> (raw)
In-Reply-To: <1334239381.16387.106.camel@zakaz.uk.xensource.com>

On Thu, 2012-04-12 at 15:03 +0100, Ian Campbell wrote:
> On Thu, 2012-04-12 at 13:50 +0100, Lin Ming wrote:
> > On Thu, 2012-04-12 at 07:08 +0000, Ian Campbell wrote:
> > > On Wed, 2012-04-11 at 22:23 +0100, Teck Choon Giam wrote:
> > > > Hi,
> > > > 
> > > > This is just my experience about issues I encountered when upgrade
> > > > from xen-4.1-testing changeset 23277:80130491806f to xen-unstable
> > > > changeset 25191:a95fc7decc83.
> > > > 
> > > > 1. Immediately after upgrade, xl list show such error:
> > > > 
> > > > # xl list
> > > > libxl: error: libxl.c:506:libxl_list_domain: geting domain info list:
> > > > Permission denied
> > > > libxl_domain_infolist failed.
> > > > 
> > > > After a reboot, it is fine.  Any idea why such behaviour?  Imagine if
> > > > there are running domUs... this might cause issues to shutdown?  I
> > > > will downgrade and repeat such test to confirm.  Might be worth a note
> > > > in upgrading note about this if this is intended?
> > > 
> > > The tools and the hypervisor are a matched pair so you would need to
> > > reboot the system in order to use the new tools. This has always been
> > > the case with Xen upgrades.
> > > 
> > > > 2. localtime setting not working.  Set to localtime=1 doesn't seems to
> > > > work whereby setting rtc_timeoffset works.  Any idea?
> > > 
> > > I've CC'd Lin Ming who implemented both of those.
> > 
> > Just did a quick gdb debug, seems libxl__domain_build_info_setdefault
> > was called 3 times. So below statement was executed 3 times too.
> > 
> > b_info->rtc_timeoffset += tm->tm_gmtoff;
> 
> Oh, that sounds like the problem. The setdefault functions are supposed
> to be idempotent, IOW calling it two or more times should give the same
> result as calling it just once.
> 
> I think you need to move this logic from setdefault
>     if (libxl_defbool_val(b_info->localtime)) {
>         time_t t;
>         struct tm *tm;
> 
>         t = time(NULL);
>         tm = localtime(&t);
> 
>         b_info->rtc_timeoffset += tm->tm_gmtoff;
>     }
> 
> into libxl__build_pre and do that calculation on a temporary variable
> whjich you pass to xc_domain_set_time_offset rather than modifying
> b_info.

How about below patch?

>From e06589cb5f1efd885f9589405d0f4ecc97427ad6 Mon Sep 17 00:00:00 2001
From: Lin Ming <mlin@ss.pku.edu.cn>
Date: Thu, 12 Apr 2012 22:36:24 +0800
Subject: [PATCH] libxl: fix rtc_timeoffset setting

libxl__domain_build_info_setdefault may be called several times, so
rtc_timeoffset can't be set in it.

Move rtc_timeoffset setting logic to libxl__build_pre.

Reported-by: Teck Choon Giam <giamteckchoon@gmail.com>
Signed-off-by: Lin Ming <mlin@ss.pku.edu.cn>
---
 tools/libxl/libxl_create.c |    9 ---------
 tools/libxl/libxl_dom.c    |   17 +++++++++++++++--
 2 files changed, 15 insertions(+), 11 deletions(-)

diff --git a/tools/libxl/libxl_create.c b/tools/libxl/libxl_create.c
index e63c7bd..e706124 100644
--- a/tools/libxl/libxl_create.c
+++ b/tools/libxl/libxl_create.c
@@ -127,15 +127,6 @@ int libxl__domain_build_info_setdefault(libxl__gc *gc,
         b_info->target_memkb = b_info->max_memkb;
 
     libxl_defbool_setdefault(&b_info->localtime, false);
-    if (libxl_defbool_val(b_info->localtime)) {
-        time_t t;
-        struct tm *tm;
-
-        t = time(NULL);
-        tm = localtime(&t);
-
-        b_info->rtc_timeoffset += tm->tm_gmtoff;
-    }
 
     libxl_defbool_setdefault(&b_info->disable_migrate, false);
 
diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c
index 0bdd654..91c4bd8 100644
--- a/tools/libxl/libxl_dom.c
+++ b/tools/libxl/libxl_dom.c
@@ -65,6 +65,8 @@ int libxl__build_pre(libxl__gc *gc, uint32_t domid,
     libxl_ctx *ctx = libxl__gc_owner(gc);
     int tsc_mode;
     char *xs_domid, *con_domid;
+    uint32_t rtc_timeoffset;
+
     xc_domain_max_vcpus(ctx->xch, domid, info->max_vcpus);
     libxl_set_vcpuaffinity_all(ctx, domid, info->max_vcpus, &info->cpumap);
     xc_domain_setmaxmem(ctx->xch, domid, info->target_memkb + LIBXL_MAXMEM_CONSTANT);
@@ -91,8 +93,19 @@ int libxl__build_pre(libxl__gc *gc, uint32_t domid,
     if (libxl_defbool_val(info->disable_migrate))
         xc_domain_disable_migrate(ctx->xch, domid);
 
-    if (info->rtc_timeoffset)
-        xc_domain_set_time_offset(ctx->xch, domid, info->rtc_timeoffset);
+    rtc_timeoffset = info->rtc_timeoffset;
+    if (libxl_defbool_val(info->localtime)) {
+        time_t t;
+        struct tm *tm;
+
+        t = time(NULL);
+        tm = localtime(&t);
+
+        rtc_timeoffset += tm->tm_gmtoff;
+    }
+
+    if (rtc_timeoffset)
+        xc_domain_set_time_offset(ctx->xch, domid, rtc_timeoffset);
 
     if (info->type == LIBXL_DOMAIN_TYPE_HVM) {
         unsigned long shadow;
-- 
1.7.2.5



> 
> Ian.
> 
> 

  reply	other threads:[~2012-04-12 14:42 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-04-11 21:23 Upgrade from xen-4.1-testing to xen-unstable report Teck Choon Giam
2012-04-12  7:08 ` Ian Campbell
2012-04-12 11:44   ` Teck Choon Giam
2012-04-12 12:50   ` Lin Ming
2012-04-12 14:03     ` Ian Campbell
2012-04-12 14:42       ` Lin Ming [this message]
2012-04-12 14:50         ` Ian Campbell
2012-04-12 15:25           ` Teck Choon Giam
2012-04-12 16:27             ` Teck Choon Giam

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=1334241773.4001.5.camel@hp6530s \
    --to=mlin@ss.pku.edu.cn \
    --cc=Ian.Campbell@citrix.com \
    --cc=giamteckchoon@gmail.com \
    --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).