[PATCH] xl: Allow shutdown wait for domain death

[PATCH] xl: Allow shutdown wait for domain death

[Jason Andryuk]

`xl shutdown -w` waits for the first of either domain shutdown or death.
Shutdown is the halting of the guest operating system, and death is the
freeing of domain resources.

Allow specifying -w multiple times to wait for only domain death.  This
is useful in scripts so that all resources are free before the script
continues.

Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
---
 docs/man/xl.1.pod.in    |  4 +++-
 tools/xl/xl_vmcontrol.c | 17 +++++++++++------
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/docs/man/xl.1.pod.in b/docs/man/xl.1.pod.in
index 09339282e6..52a47a6fbd 100644
--- a/docs/man/xl.1.pod.in
+++ b/docs/man/xl.1.pod.in
@@ -743,7 +743,9 @@ of a Xen system.
 
 =item B<-w>, B<--wait>
 
-Wait for the domain to complete shutdown before returning.
+Wait for the domain to complete shutdown before returning.  If given once,
+the wait is for domain shutdown or domain death.  If given multiple times,
+the wait is for domain death only.
 
 =item B<-F>
 
diff --git a/tools/xl/xl_vmcontrol.c b/tools/xl/xl_vmcontrol.c
index 17b4514c94..435155a033 100644
--- a/tools/xl/xl_vmcontrol.c
+++ b/tools/xl/xl_vmcontrol.c
@@ -162,7 +162,8 @@ static void shutdown_domain(uint32_t domid,
     }
 }
 
-static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
+static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr,
+                                   int wait_for_shutdown_or_death)
 {
     int rc, count = 0;
     LOG("Waiting for %d domains", nr);
@@ -183,8 +184,12 @@ static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
         case LIBXL_EVENT_TYPE_DOMAIN_SHUTDOWN:
             LOG("Domain %d has been shut down, reason code %d",
                 event->domid, event->u.domain_shutdown.shutdown_reason);
-            libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
-            count++;
+            if (wait_for_shutdown_or_death) {
+                libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
+                count++;
+            } else {
+                LOG("Domain %d continue waiting for death", event->domid);
+            }
             break;
         default:
             LOG("Unexpected event type %d", event->type);
@@ -214,7 +219,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
         all = 1;
         break;
     case 'w':
-        wait_for_it = 1;
+        wait_for_it++;
         break;
     case 'F':
         fallback_trigger = 1;
@@ -246,7 +251,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
         }
 
         if (deathws) {
-            wait_for_domain_deaths(deathws, nrdeathws);
+            wait_for_domain_deaths(deathws, nrdeathws, wait_for_it == 1);
             free(deathws);
         }
 
@@ -258,7 +263,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
         fn(domid, wait_for_it ? &deathw : NULL, 0, fallback_trigger);
 
         if (wait_for_it)
-            wait_for_domain_deaths(&deathw, 1);
+            wait_for_domain_deaths(&deathw, 1, wait_for_it == 1);
     }
 
 
-- 
2.25.1

Re: [PATCH for-4.14] xl: Allow shutdown wait for domain death

[Ian Jackson]

Jason Andryuk writes ("[PATCH] xl: Allow shutdown wait for domain death"):
> `xl shutdown -w` waits for the first of either domain shutdown or death.
> Shutdown is the halting of the guest operating system, and death is the
> freeing of domain resources.
> 
> Allow specifying -w multiple times to wait for only domain death.  This
> is useful in scripts so that all resources are free before the script
> continues.

Thanks!

Reviewed-by: Ian Jackson <ian.jackson@eu.citrix.com>

Paul, I think this is a candidate for 4.14.  Without this patch it is
not possible to reliably wait for a domain, with xl, and then restart
it.  (At least not without a good deal of pratting about and polling
with xl list.)  osstest has a `sleep' as a workaround...

I have reviewed this patch particularly carefully with a view to
understanding what happens if you pass just one `-w' as before.
I have convinced myself that there is definitely no change, so I don't
think this patch can introduce a regression.

Ian.

> Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
> ---
>  docs/man/xl.1.pod.in    |  4 +++-
>  tools/xl/xl_vmcontrol.c | 17 +++++++++++------
>  2 files changed, 14 insertions(+), 7 deletions(-)
> 
> diff --git a/docs/man/xl.1.pod.in b/docs/man/xl.1.pod.in
> index 09339282e6..52a47a6fbd 100644
> --- a/docs/man/xl.1.pod.in
> +++ b/docs/man/xl.1.pod.in
> @@ -743,7 +743,9 @@ of a Xen system.
>  
>  =item B<-w>, B<--wait>
>  
> -Wait for the domain to complete shutdown before returning.
> +Wait for the domain to complete shutdown before returning.  If given once,
> +the wait is for domain shutdown or domain death.  If given multiple times,
> +the wait is for domain death only.
>  
>  =item B<-F>
>  
> diff --git a/tools/xl/xl_vmcontrol.c b/tools/xl/xl_vmcontrol.c
> index 17b4514c94..435155a033 100644
> --- a/tools/xl/xl_vmcontrol.c
> +++ b/tools/xl/xl_vmcontrol.c
> @@ -162,7 +162,8 @@ static void shutdown_domain(uint32_t domid,
>      }
>  }
>  
> -static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
> +static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr,
> +                                   int wait_for_shutdown_or_death)
>  {
>      int rc, count = 0;
>      LOG("Waiting for %d domains", nr);
> @@ -183,8 +184,12 @@ static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
>          case LIBXL_EVENT_TYPE_DOMAIN_SHUTDOWN:
>              LOG("Domain %d has been shut down, reason code %d",
>                  event->domid, event->u.domain_shutdown.shutdown_reason);
> -            libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
> -            count++;
> +            if (wait_for_shutdown_or_death) {
> +                libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
> +                count++;
> +            } else {
> +                LOG("Domain %d continue waiting for death", event->domid);
> +            }
>              break;
>          default:
>              LOG("Unexpected event type %d", event->type);
> @@ -214,7 +219,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
>          all = 1;
>          break;
>      case 'w':
> -        wait_for_it = 1;
> +        wait_for_it++;
>          break;
>      case 'F':
>          fallback_trigger = 1;
> @@ -246,7 +251,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
>          }
>  
>          if (deathws) {
> -            wait_for_domain_deaths(deathws, nrdeathws);
> +            wait_for_domain_deaths(deathws, nrdeathws, wait_for_it == 1);
>              free(deathws);
>          }
>  
> @@ -258,7 +263,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
>          fn(domid, wait_for_it ? &deathw : NULL, 0, fallback_trigger);
>  
>          if (wait_for_it)
> -            wait_for_domain_deaths(&deathw, 1);
> +            wait_for_domain_deaths(&deathw, 1, wait_for_it == 1);
>      }
>  
>  
> -- 
> 2.25.1
> 

RE: [PATCH for-4.14] xl: Allow shutdown wait for domain death

[Paul Durrant]

> -----Original Message-----
> From: Ian Jackson <ian.jackson@citrix.com>
> Sent: 18 June 2020 16:44
> To: Jason Andryuk <jandryuk@gmail.com>; Paul Durrant <xadimgnik@gmail.com>
> Cc: xen-devel@lists.xenproject.org; Wei Liu <wl@xen.org>; Anthony Perard <anthony.perard@citrix.com>
> Subject: Re: [PATCH for-4.14] xl: Allow shutdown wait for domain death
> 
> Jason Andryuk writes ("[PATCH] xl: Allow shutdown wait for domain death"):
> > `xl shutdown -w` waits for the first of either domain shutdown or death.
> > Shutdown is the halting of the guest operating system, and death is the
> > freeing of domain resources.
> >
> > Allow specifying -w multiple times to wait for only domain death.  This
> > is useful in scripts so that all resources are free before the script
> > continues.
> 
> Thanks!
> 
> Reviewed-by: Ian Jackson <ian.jackson@eu.citrix.com>
> 
> Paul, I think this is a candidate for 4.14.  Without this patch it is
> not possible to reliably wait for a domain, with xl, and then restart
> it.  (At least not without a good deal of pratting about and polling
> with xl list.)  osstest has a `sleep' as a workaround...
> 

Yes, it would nice to drop such workarounds.

> I have reviewed this patch particularly carefully with a view to
> understanding what happens if you pass just one `-w' as before.
> I have convinced myself that there is definitely no change, so I don't
> think this patch can introduce a regression.

Ok, I'll trust your judgement.

Release-acked-by: Paul Durrant <paul@xen.org>

> 
> Ian.
> 
> > Signed-off-by: Jason Andryuk <jandryuk@gmail.com>
> > ---
> >  docs/man/xl.1.pod.in    |  4 +++-
> >  tools/xl/xl_vmcontrol.c | 17 +++++++++++------
> >  2 files changed, 14 insertions(+), 7 deletions(-)
> >
> > diff --git a/docs/man/xl.1.pod.in b/docs/man/xl.1.pod.in
> > index 09339282e6..52a47a6fbd 100644
> > --- a/docs/man/xl.1.pod.in
> > +++ b/docs/man/xl.1.pod.in
> > @@ -743,7 +743,9 @@ of a Xen system.
> >
> >  =item B<-w>, B<--wait>
> >
> > -Wait for the domain to complete shutdown before returning.
> > +Wait for the domain to complete shutdown before returning.  If given once,
> > +the wait is for domain shutdown or domain death.  If given multiple times,
> > +the wait is for domain death only.
> >
> >  =item B<-F>
> >
> > diff --git a/tools/xl/xl_vmcontrol.c b/tools/xl/xl_vmcontrol.c
> > index 17b4514c94..435155a033 100644
> > --- a/tools/xl/xl_vmcontrol.c
> > +++ b/tools/xl/xl_vmcontrol.c
> > @@ -162,7 +162,8 @@ static void shutdown_domain(uint32_t domid,
> >      }
> >  }
> >
> > -static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
> > +static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr,
> > +                                   int wait_for_shutdown_or_death)
> >  {
> >      int rc, count = 0;
> >      LOG("Waiting for %d domains", nr);
> > @@ -183,8 +184,12 @@ static void wait_for_domain_deaths(libxl_evgen_domain_death **deathws, int nr)
> >          case LIBXL_EVENT_TYPE_DOMAIN_SHUTDOWN:
> >              LOG("Domain %d has been shut down, reason code %d",
> >                  event->domid, event->u.domain_shutdown.shutdown_reason);
> > -            libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
> > -            count++;
> > +            if (wait_for_shutdown_or_death) {
> > +                libxl_evdisable_domain_death(ctx, deathws[event->for_user]);
> > +                count++;
> > +            } else {
> > +                LOG("Domain %d continue waiting for death", event->domid);
> > +            }
> >              break;
> >          default:
> >              LOG("Unexpected event type %d", event->type);
> > @@ -214,7 +219,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
> >          all = 1;
> >          break;
> >      case 'w':
> > -        wait_for_it = 1;
> > +        wait_for_it++;
> >          break;
> >      case 'F':
> >          fallback_trigger = 1;
> > @@ -246,7 +251,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
> >          }
> >
> >          if (deathws) {
> > -            wait_for_domain_deaths(deathws, nrdeathws);
> > +            wait_for_domain_deaths(deathws, nrdeathws, wait_for_it == 1);
> >              free(deathws);
> >          }
> >
> > @@ -258,7 +263,7 @@ static int main_shutdown_or_reboot(int do_reboot, int argc, char **argv)
> >          fn(domid, wait_for_it ? &deathw : NULL, 0, fallback_trigger);
> >
> >          if (wait_for_it)
> > -            wait_for_domain_deaths(&deathw, 1);
> > +            wait_for_domain_deaths(&deathw, 1, wait_for_it == 1);
> >      }
> >
> >
> > --
> > 2.25.1
> >

RE: [PATCH for-4.14] xl: Allow shutdown wait for domain death

[Ian Jackson]

Paul Durrant writes ("RE: [PATCH for-4.14] xl: Allow shutdown wait for domain death"):
> > -----Original Message-----
> > From: Ian Jackson <ian.jackson@citrix.com>
...
> > I have reviewed this patch particularly carefully with a view to
> > understanding what happens if you pass just one `-w' as before.
> > I have convinced myself that there is definitely no change, so I don't
> > think this patch can introduce a regression.
> 
> Ok, I'll trust your judgement.

Thanks :-).  Pushed.

Ian.