Den 2024-10-23 kl. 17:26, skrev Waiman Long:
On 10/23/24 3:52 AM, Maarten Lankhorst wrote:
The initial version was based roughly on the rdma and misc cgroup
controllers, with a lot of the accounting code borrowed from rdma.
The current version is a complete rewrite with page counter; it uses
the same min/low/max semantics as the memory cgroup as a result.
There's a small mismatch as TTM uses u64, and page_counter long pages.
In practice it's not a problem. 32-bits systems don't really come with
=4GB cards and as long as we're consistently wrong with units, it's
fine. The device page size may not be in the same units as kernel page
size, and each region might also have a different page size (VRAM vs GART
for example).
The interface is simple:
- populate dev_cgroup_try_charge->regions[..] name and size for each
active
region, set num_regions accordingly.
- Call (dev,drmm)_cgroup_register_device()
- Use dev_cgroup_try_charge to check if you can allocate a chunk of
memory,
use dev_cgroup__uncharge when freeing it. This may return an error
code,
or -EAGAIN when the cgroup limit is reached. In that case a reference
to the limiting pool is returned.
- The limiting cs can be used as compare function for
dev_cgroup_state_evict_valuable.
- After having evicted enough, drop reference to limiting cs with
dev_cgroup_pool_state_put.
This API allows you to limit device resources with cgroups.
You can see the supported cards in /sys/fs/cgroup/dev.region.capacity
You need to echo +dev to cgroup.subtree_control, and then you can
partition memory.
Co-developed-by: Friedrich Vock <friedrich.vock@xxxxxx>
Signed-off-by: Friedrich Vock <friedrich.vock@xxxxxx>
Co-developed-by: Maxime Ripard <mripard@xxxxxxxxxx>
Signed-off-by: Maxime Ripard <mripard@xxxxxxxxxx>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@xxxxxxxxxxxxxxx>
---
Documentation/admin-guide/cgroup-v2.rst | 51 ++
Documentation/core-api/cgroup.rst | 9 +
Documentation/core-api/index.rst | 1 +
Documentation/gpu/drm-compute.rst | 54 ++
include/linux/cgroup_dev.h | 91 +++
include/linux/cgroup_subsys.h | 4 +
include/linux/page_counter.h | 2 +-
init/Kconfig | 7 +
kernel/cgroup/Makefile | 1 +
kernel/cgroup/dev.c | 893 ++++++++++++++++++++++++
mm/page_counter.c | 4 +-
11 files changed, 1114 insertions(+), 3 deletions(-)
create mode 100644 Documentation/core-api/cgroup.rst
create mode 100644 Documentation/gpu/drm-compute.rst
create mode 100644 include/linux/cgroup_dev.h
create mode 100644 kernel/cgroup/dev.c
Just a general comment.
Cgroup v1 has a legacy device controller in security/device_cgroup.c
which is no longer available in cgroup v2. So if you use the name device
controller, the documentation must be clear that it is completely
different and have no relationship from the device controller in cgroup v1.
Hey,
Thanks for noticing. I didn't know there was such a controller. Seems
weird to have one for managing access to opening devnodes instead of a
security module.
I'll update the documentation in the next version to make it more clear.
Cheers,
~Maarten
