xen: cpupool: fix shutdown with cpupools with different schedulers

PATCH In review Dario Faggioli 3 participants 3 messages Jun 11, 2015

Share & tools

Dario Faggioli

author

2015-06-11 12:31 UTC #0 lore raw

PATCH 1/1 xen: cpupool: fix shutdown with cpupools with different schedulers

trying to shutdown the host when a cpupool exists, has pCPUs, and has a scheduler different than the Xen's default one, produces this:

 root@Zhaman:~# xl cpupool-cpu-remove Pool-0 8
 root@Zhaman:~# xl cpupool-create name=\"Pool-1\" sched=\"credit2\"
 Using config file "command line"
 cpupool name:   Pool-1
 scheduler:      credit2
 number of cpus: 0
 root@Zhaman:~# xl cpupool-cpu-add Pool-1 8
 root@Zhaman:~# shutdown -h now

(XEN) ----[ Xen-4.6-unstable x86_64 debug=y Not tainted ]----
(XEN) CPU: 0
(XEN) RIP: e008:[<ffff82d080133bdf>] kill_timer+0x56/0x298
(XEN) RFLAGS: 0000000000010002 CONTEXT: hypervisor
... ... ...
(XEN) Xen call trace:

 (XEN)    [<ffff82d080133bdf>] kill_timer+0x56/0x298
 (XEN)    [<ffff82d08012233f>] csched_free_pdata+0x9b/0xcf
 (XEN)    [<ffff82d08012c30c>] cpu_schedule_callback+0x64/0x8b
 (XEN)    [<ffff82d08011bc7a>] notifier_call_chain+0x67/0x87
 (XEN)    [<ffff82d08010153e>] cpu_down+0xd9/0x12c
 (XEN)    [<ffff82d080101744>] disable_nonboot_cpus+0x93/0x138
 (XEN)    [<ffff82d0801aa6e7>] enter_state_helper+0xbd/0x365
 (XEN)    [<ffff82d0801061e5>] continue_hypercall_tasklet_handler+0x4a/0xb1
 (XEN)    [<ffff82d080132387>] do_tasklet_work+0x78/0xab
 (XEN)    [<ffff82d0801326bd>] do_tasklet+0x5e/0x8a
 (XEN)    [<ffff82d0801646d2>] idle_loop+0x56/0x6b

... ... ...
(XEN) ****************************************
(XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0000]
(XEN) Faulting linear address: 0000000000000041
(XEN) ****************************************

The fix is, when tearing down a pCPU, call the free_pdata() hook from the scheduler of the cpupool the pCPU belongs to, not always the one from the default scheduler.

Signed-off-by Dario Faggioli <[email protected]>

---
Cc: Juergen Gross <[email protected]>
Cc: George Dunlap <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Tim Deegan <[email protected]>
Cc: Keir Fraser <[email protected]>

---
 xen/common/schedule.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

xen/common/schedule.c +2 -1

index 6b02f98..3325ccd 100644
@@ -1303,9 +1303,10 @@ static int cpu_schedule_up(unsigned int cpu)
 static void cpu_schedule_down(unsigned int cpu)
 {
     struct schedule_data *sd = &per_cpu(schedule_data, cpu);
+    struct scheduler *sched = per_cpu(scheduler, cpu);
 
     if ( sd->sched_priv != NULL )
-        SCHED_OP(&ops, free_pdata, sd->sched_priv, cpu);
+        SCHED_OP(sched, free_pdata, sd->sched_priv, cpu);
 
     kill_timer(&sd->s_timer);
 }

Juergen Gross

Reviewed-by on v1 1/1

2015-06-11 12:46 UTC #1 lore raw

Quoted 38 lines · Dario Faggioli

trying to shutdown the host when a cpupool exists, has pCPUs, and has a scheduler different than the Xen's default one, produces this:
  root@Zhaman:~# xl cpupool-cpu-remove Pool-0 8
  root@Zhaman:~# xl cpupool-create name=\"Pool-1\" sched=\"credit2\"
  Using config file "command line"
  cpupool name:   Pool-1
  scheduler:      credit2
  number of cpus: 0
  root@Zhaman:~# xl cpupool-cpu-add Pool-1 8
  root@Zhaman:~# shutdown -h now
(XEN) ----[ Xen-4.6-unstable x86_64 debug=y Not tainted ]----
(XEN) CPU: 0
(XEN) RIP: e008:[<ffff82d080133bdf>] kill_timer+0x56/0x298
(XEN) RFLAGS: 0000000000010002 CONTEXT: hypervisor
... ... ...
(XEN) Xen call trace:
  (XEN)    [<ffff82d080133bdf>] kill_timer+0x56/0x298
  (XEN)    [<ffff82d08012233f>] csched_free_pdata+0x9b/0xcf
  (XEN)    [<ffff82d08012c30c>] cpu_schedule_callback+0x64/0x8b
  (XEN)    [<ffff82d08011bc7a>] notifier_call_chain+0x67/0x87
  (XEN)    [<ffff82d08010153e>] cpu_down+0xd9/0x12c
  (XEN)    [<ffff82d080101744>] disable_nonboot_cpus+0x93/0x138
  (XEN)    [<ffff82d0801aa6e7>] enter_state_helper+0xbd/0x365
  (XEN)    [<ffff82d0801061e5>] continue_hypercall_tasklet_handler+0x4a/0xb1
  (XEN)    [<ffff82d080132387>] do_tasklet_work+0x78/0xab
  (XEN)    [<ffff82d0801326bd>] do_tasklet+0x5e/0x8a
  (XEN)    [<ffff82d0801646d2>] idle_loop+0x56/0x6b
... ... ...
(XEN) ****************************************
(XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0000]
(XEN) Faulting linear address: 0000000000000041
(XEN) ****************************************

Quoted 5 lines

The fix is, when tearing down a pCPU, call the free_pdata() hook from the scheduler of the cpupool the pCPU belongs to, not always the one from the default scheduler.
Signed-off-by Dario Faggioli <[email protected]>

Reviewed-by Juergen Gross <[email protected]>

Quoted 34 lines

---
Cc: Juergen Gross <[email protected]>
Cc: George Dunlap <[email protected]>
Cc: Jan Beulich <[email protected]>
Cc: Tim Deegan <[email protected]>
Cc: Keir Fraser <[email protected]>
---
  xen/common/schedule.c |    3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
xen/common/schedule.c +2 -1
index 6b02f98..3325ccd 100644
@@ -1303,9 +1303,10 @@ static int cpu_schedule_up(unsigned int cpu)
  static void cpu_schedule_down(unsigned int cpu)
  {
      struct schedule_data *sd = &per_cpu(schedule_data, cpu);
+    struct scheduler *sched = per_cpu(scheduler, cpu);
 
      if ( sd->sched_priv != NULL )
-        SCHED_OP(&ops, free_pdata, sd->sched_priv, cpu);
+        SCHED_OP(sched, free_pdata, sd->sched_priv, cpu);
 
      kill_timer(&sd->s_timer);
  }
_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xen.org/xen-devel

George Dunlap

Acked-by on v1 1/1

2015-06-11 14:25 UTC #2 lore raw

In reply to Dario Faggioli

Quoted 38 lines · Dario Faggioli

trying to shutdown the host when a cpupool exists, has pCPUs, and has a scheduler different than the Xen's default one, produces this:
 root@Zhaman:~# xl cpupool-cpu-remove Pool-0 8
 root@Zhaman:~# xl cpupool-create name=\"Pool-1\" sched=\"credit2\"
 Using config file "command line"
 cpupool name:   Pool-1
 scheduler:      credit2
 number of cpus: 0
 root@Zhaman:~# xl cpupool-cpu-add Pool-1 8
 root@Zhaman:~# shutdown -h now
(XEN) ----[ Xen-4.6-unstable x86_64 debug=y Not tainted ]----
(XEN) CPU: 0
(XEN) RIP: e008:[<ffff82d080133bdf>] kill_timer+0x56/0x298
(XEN) RFLAGS: 0000000000010002 CONTEXT: hypervisor
... ... ...
(XEN) Xen call trace:
 (XEN)    [<ffff82d080133bdf>] kill_timer+0x56/0x298
 (XEN)    [<ffff82d08012233f>] csched_free_pdata+0x9b/0xcf
 (XEN)    [<ffff82d08012c30c>] cpu_schedule_callback+0x64/0x8b
 (XEN)    [<ffff82d08011bc7a>] notifier_call_chain+0x67/0x87
 (XEN)    [<ffff82d08010153e>] cpu_down+0xd9/0x12c
 (XEN)    [<ffff82d080101744>] disable_nonboot_cpus+0x93/0x138
 (XEN)    [<ffff82d0801aa6e7>] enter_state_helper+0xbd/0x365
 (XEN)    [<ffff82d0801061e5>] continue_hypercall_tasklet_handler+0x4a/0xb1
 (XEN)    [<ffff82d080132387>] do_tasklet_work+0x78/0xab
 (XEN)    [<ffff82d0801326bd>] do_tasklet+0x5e/0x8a
 (XEN)    [<ffff82d0801646d2>] idle_loop+0x56/0x6b
... ... ...
(XEN) **************************************** (XEN) Panic on CPU 0:
(XEN) FATAL PAGE FAULT
(XEN) [error_code=0000]
(XEN) Faulting linear address: 0000000000000041 (XEN) ****************************************

Quoted 5 lines

The fix is, when tearing down a pCPU, call the free_pdata() hook from the scheduler of the cpupool the pCPU belongs to, not always the one from the default scheduler.
Signed-off-by Dario Faggioli <[email protected]>

Acked-by George Dunlap <[email protected]>