On 2023-10-27 at 16:30:38 +0300, Ilpo Järvinen wrote:
On Fri, 27 Oct 2023, Maciej Wieczór-Retman wrote:
On 2023-10-24 at 12:26:31 +0300, Ilpo Järvinen wrote:
Resource id is acquired differently depending on CPU. AMD tests use id from L3 cache, whereas CPUs from other vendors base the id on topology package id. In order to support L2 CAT test, this has to be generalized.
The driver side code seems to get the resource ids from cache ids so the approach used by the AMD branch seems to match the kernel-side code. It will also work with L2 resource IDs as long as the cache level is generalized.
Using the topology id was always fragile due to mismatch with the kernel-side way to acquire the resource id. It got incorrect resource id, e.g., when Cluster-on-Die (CoD) is enabled for CPU (but CoD is not well suited for resctrl in the first place so it has not been a big issues if test don't work correctly with it).
"not been a big issues" -> "not been a big issue"?
"if test don't work correctly" -> "if test doesn't work correctly" / "if tests don't work correctly"?
Taking all the above into account, generalize the resource id calculation by taking it from the cache id and do not hard-code the cache level.
In x86/resctrl files group of CPUs sharing a resctrl resource is called a domain. Because of that I think "resource id" terms should be substituted with "domain id" to match core resctrl code.
Okay, I was just using the terminology which pre-existed in the selftest code.
I've really tried to look how I should call it but things were quite inconsistent. The selftest uses resource id and reads it from cache id. Documentation uses "instances of that resource" or "cache id" or "domain x".
Documentation/arch/x86/resctrl.rst is very misleading btw and should be IMHO updated:
Memory bandwidth Allocation (default mode)
Memory b/w domain is L3 cache. ::
MB:<cache_id0>=bandwidth0;<cache_id1>=bandwidth1;...It claims "domain" is L3 cache (and the value we're talking about is called "cache_id" here which is what you just called "domain id"). I suppose it should say "b/w resource is L3 cache"?
I believe so. After some looking around it looks like the "L3 domain" naming is also present in rdtgroup.c:mkdir_mondata_all() and "mon_data" description in the docs:
Docs: "mon_data": This contains a set of files organized by L3 domain and by RDT event. E.g. on a system with two L3 domains there will
rdtgroup.c: /* * This creates a directory mon_data which contains the monitored data. * * mon_data has one directory for each domain which are named * in the format mon_<domain_name>_<domain_id>. For ex: A mon_data * with L3 domain looks as below: * ./mon_data: * mon_L3_00 * mon_L3_01 * mon_L3_02 * ... * * Each domain directory has one file per event: * ./mon_L3_00/: * llc_occupancy * */ static int mkdir_mondata_all(struct kernfs_node *parent_kn, struct rdtgroup *prgrp, struct kernfs_node **dest_kn)
From the reading I've done on resctrl my understanding was that L2/L3 are resources that can be shared between domains.