This is a requirement of the scheduler interface, violating this
causes for example with the RT scheduler:
(XEN) Assertion 'local_irq_is_enabled()' failed at spinlock.c:137
(XEN) ----[ Xen-4.5.0 arm32 debug=y Not tainted ]----
(XEN) CPU: 0
(XEN) PC:
0022a074 _spin_lock_irq+0x18/0x94
(XEN) Xen call trace:
(XEN) [<
0022a074>] _spin_lock_irq+0x18/0x94 (PC)
(XEN) [<
002256b8>] rt_context_saved+0x3c/0x144 (LR)
(XEN) [<
002256b8>] rt_context_saved+0x3c/0x144
(XEN) [<
00228ed8>] context_saved+0x4c/0x80
(XEN) [<
002524b4>] schedule_tail+0x148/0x2f0
(XEN) [<
00252668>] continue_new_vcpu+0xc/0x70
(XEN) [<
00252a18>] context_switch+0x74/0x7c
Signed-off-by: denys drozdov <denys.drozdov@globallogic.com>
Acked-by: Ian Campbell <ian.campbell@citrix.com>
Reviewed-by: Dario Faggioli <dario.faggioli@citrix.com>
Reviewed-by: Julien Grall <julien.grall@linaro.org>
* mode. Therefore we don't need to save the context of an idle VCPU.
*/
if ( is_idle_vcpu(p) )
- goto end_context;
+ return;
p2m_save_state(p);
gic_save_state(p);
isb();
-
-end_context:
- context_saved(p);
}
static void ctxt_switch_to(struct vcpu *n)
local_irq_enable();
+ context_saved(prev);
+
if ( prev != current )
update_runstate_area(current);