On 11/10/2018 06:58, Mike Rapoport wrote: > From: David Hildenbrand <david@xxxxxxxxxx> > > Let's document the magic a bit, especially why device_hotplug_lock is > required when adding/removing memory and how it all play together with > requests to online/offline memory from user space. > > [ rppt: moved the text to Documentation/core-api/memory-hotplug.rst ] > > Link: http://lkml.kernel.org/r/20180925091457.28651-7-david@xxxxxxxxxx > Signed-off-by: David Hildenbrand <david@xxxxxxxxxx> > Reviewed-by: Pavel Tatashin <pavel.tatashin@xxxxxxxxxxxxx> > Reviewed-by: Rashmica Gupta <rashmica.g@xxxxxxxxx> > Cc: Jonathan Corbet <corbet@xxxxxxx> > Cc: Michal Hocko <mhocko@xxxxxxxx> > Cc: Balbir Singh <bsingharora@xxxxxxxxx> > Cc: Benjamin Herrenschmidt <benh@xxxxxxxxxxxxxxxxxxx> > Cc: Boris Ostrovsky <boris.ostrovsky@xxxxxxxxxx> > Cc: Dan Williams <dan.j.williams@xxxxxxxxx> > Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> > Cc: Haiyang Zhang <haiyangz@xxxxxxxxxxxxx> > Cc: Heiko Carstens <heiko.carstens@xxxxxxxxxx> > Cc: John Allen <jallen@xxxxxxxxxxxxxxxxxx> > Cc: Joonsoo Kim <iamjoonsoo.kim@xxxxxxx> > Cc: Juergen Gross <jgross@xxxxxxxx> > Cc: Kate Stewart <kstewart@xxxxxxxxxxxxxxxxxxx> > Cc: "K. Y. Srinivasan" <kys@xxxxxxxxxxxxx> > Cc: Len Brown <lenb@xxxxxxxxxx> > Cc: Martin Schwidefsky <schwidefsky@xxxxxxxxxx> > Cc: Mathieu Malaterre <malat@xxxxxxxxxx> > Cc: Michael Ellerman <mpe@xxxxxxxxxxxxxx> > Cc: Michael Neuling <mikey@xxxxxxxxxxx> > Cc: Nathan Fontenot <nfont@xxxxxxxxxxxxxxxxxx> > Cc: Oscar Salvador <osalvador@xxxxxxx> > Cc: Paul Mackerras <paulus@xxxxxxxxx> > Cc: Philippe Ombredanne <pombredanne@xxxxxxxx> > Cc: Rafael J. Wysocki <rafael.j.wysocki@xxxxxxxxx> > Cc: "Rafael J. Wysocki" <rjw@xxxxxxxxxxxxx> > Cc: Stephen Hemminger <sthemmin@xxxxxxxxxxxxx> > Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx> > Cc: Vlastimil Babka <vbabka@xxxxxxx> > Cc: YASUAKI ISHIMATSU <yasu.isimatu@xxxxxxxxx> > Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> > Signed-off-by: Mike Rapoport <rppt@xxxxxxxxxxxxx> > --- > Documentation/core-api/memory-hotplug.rst | 38 +++++++++++++++++++++++++++++++ > 1 file changed, 38 insertions(+) > > diff --git a/Documentation/core-api/memory-hotplug.rst b/Documentation/core-api/memory-hotplug.rst > index a99f2f2..de7467e 100644 > --- a/Documentation/core-api/memory-hotplug.rst > +++ b/Documentation/core-api/memory-hotplug.rst > @@ -85,3 +85,41 @@ MEM_ONLINE, or MEM_OFFLINE action to cancel hotplugging. It stops > further processing of the notification queue. > > NOTIFY_STOP stops further processing of the notification queue. > + > +Locking Internals > +================= > + > +When adding/removing memory that uses memory block devices (i.e. ordinary RAM), > +the device_hotplug_lock should be held to: > + > +- synchronize against online/offline requests (e.g. via sysfs). This way, memory > + block devices can only be accessed (.online/.state attributes) by user > + space once memory has been fully added. And when removing memory, we > + know nobody is in critical sections. > +- synchronize against CPU hotplug and similar (e.g. relevant for ACPI and PPC) > + > +Especially, there is a possible lock inversion that is avoided using > +device_hotplug_lock when adding memory and user space tries to online that > +memory faster than expected: > + > +- device_online() will first take the device_lock(), followed by > + mem_hotplug_lock > +- add_memory_resource() will first take the mem_hotplug_lock, followed by > + the device_lock() (while creating the devices, during bus_add_device()). > + > +As the device is visible to user space before taking the device_lock(), this > +can result in a lock inversion. > + > +onlining/offlining of memory should be done via device_online()/ > +device_offline() - to make sure it is properly synchronized to actions > +via sysfs. Holding device_hotplug_lock is advised (to e.g. protect online_type) > + > +When adding/removing/onlining/offlining memory or adding/removing > +heterogeneous/device memory, we should always hold the mem_hotplug_lock in > +write mode to serialise memory hotplug (e.g. access to global/zone > +variables). > + > +In addition, mem_hotplug_lock (in contrast to device_hotplug_lock) in read > +mode allows for a quite efficient get_online_mems/put_online_mems > +implementation, so code accessing memory can protect from that memory > +vanishing. > Looks good to me. -- Thanks, David / dhildenb
