Re: [Xen-devel] [PATCH 5/6] xen/domain: Call sched_destroy_domain() in the domain_create() error path

2018-03-01 Thread Dario Faggioli
On Wed, 2018-02-28 at 14:14 +, Andrew Cooper wrote: > If domain_create() fails, complete_domain_destroy() doesn't get > called, > meaning that sched_destroy_domain() is missed. In practice, this can > only > fail because of exceptional late_hwdom_init() issues at the moment. > > Make

Re: [Xen-devel] [PATCH 5/6] xen/domain: Call sched_destroy_domain() in the domain_create() error path

2018-03-01 Thread Dario Faggioli
On Wed, 2018-02-28 at 14:14 +, Andrew Cooper wrote: > If domain_create() fails, complete_domain_destroy() doesn't get > called, > meaning that sched_destroy_domain() is missed. In practice, this can > only > fail because of exceptional late_hwdom_init() issues at the moment. > > Make

Re: [Xen-devel] [PATCH 5/6] xen/domain: Call sched_destroy_domain() in the domain_create() error path

2018-02-28 Thread George Dunlap
On 02/28/2018 02:14 PM, Andrew Cooper wrote: > If domain_create() fails, complete_domain_destroy() doesn't get called, > meaning that sched_destroy_domain() is missed. In practice, this can only > fail because of exceptional late_hwdom_init() issues at the moment. > > Make sched_destroy_domain()

[Xen-devel] [PATCH 5/6] xen/domain: Call sched_destroy_domain() in the domain_create() error path

2018-02-28 Thread Andrew Cooper
If domain_create() fails, complete_domain_destroy() doesn't get called, meaning that sched_destroy_domain() is missed. In practice, this can only fail because of exceptional late_hwdom_init() issues at the moment. Make sched_destroy_domain() idempotent, and call it in the fail path.