Re: [PATCH v12 5/5] riscv_cbqri: Add CBQRI capacity allocation platform driver

From: Jonathan Cameron

Date: Wed Nov 18 2026 - 19:00:22 EST


On Sat, 14 Nov 2026 14:06:25 -0800
Drew Fustini <fustini@xxxxxxxxxx> wrote:

> Add a platform driver for CBQRI capacity controllers described in
> device tree with the riscv,cbqri-capacity-controller compatible.
>
> The controller node points at the cache it governs with a
> riscv,cbqri-cache phandle. The driver maps the register block, looks
> that cache up in cacheinfo and uses its cache id as the resctrl domain
> id, so the binding does not need to carry a CPU mask. The controller is
> then handed to the CBQRI core, which probes its capabilities and decides
> whether to expose it through resctrl. Only an L2 or L3 cache is
> accepted, since those are the cache levels resctrl has resources for.
>
> Nothing is vendor-specific and reg is the CBQRI register block itself,
> so any SoC that describes a CBQRI capacity controller in device tree can
> use the driver unchanged.
>
> Tested-by: Chen Pei <cp0613@xxxxxxxxxxxxxxxxx>
> Tested-by: Zeng Heng <zengheng@xxxxxxxxxxxxx>
> Reviewed-by: Zeng Heng <zengheng@xxxxxxxxxxxxx>
> Signed-off-by: Drew Fustini <fustini@xxxxxxxxxx>

Sorry, late to the game and pretty much a drive by review.

Still a few things below that will leave you with more readable
and probably less fragile code.

Jonathan

> diff --git a/drivers/resctrl/cbqri_capacity.c b/drivers/resctrl/cbqri_capacity.c
> new file mode 100644
> index 000000000000..6f3d06c4f74b
> --- /dev/null
> +++ b/drivers/resctrl/cbqri_capacity.c

> +static int cbqri_capacity_probe(struct platform_device *pdev)
> +{

Noted in the ACPI patch that you can save some effort there
by a few tweaks that probably make sense here anyway.

> + struct device *dev = &pdev->dev;
> + struct cbqri_controller_info info = {};
> + struct device_node *cache_np;
> + struct resource *res;
> + void __iomem *base;
> + u32 rcid_count, cache_level, cache_id;
> + bool found = false;
> + int cpu, ret;
> +
> + base = devm_platform_get_and_ioremap_resource(pdev, 0, &res);
> + if (IS_ERR(base))
> + return PTR_ERR(base);
> +
> + ret = of_property_read_u32(dev->of_node, "riscv,cbqri-rcid", &rcid_count);
> + if (ret) {
> + dev_err(dev, "missing riscv,cbqri-rcid\n");
It's a probe, use
return dev_err_probe(dev, ret, "missing riscv,cbqri-rcid\n);
same for all these paths. Sure we won't defer but it does nice formatting
and tends to save a couple of lines every time!

With the __free() magic, you can do this for all the returns in this
function.

> + return ret;
> + }
> +
> + cache_np = of_parse_phandle(dev->of_node, "riscv,cbqri-cache", 0);
Follow on from glancing at the acpi and then finding this elsewhere in my
inbox.
struct device_node *cache_np __free(device_node) =
of_parse_phandle(dev->of_node, "riscv,cbqri-cache", 0);
and drop all the goto iritiations.

> + if (!cache_np) {
> + dev_err(dev, "missing riscv,cbqri-cache phandle\n");
> + return -EINVAL;
> + }
> +
> + ret = of_property_read_u32(cache_np, "cache-level", &cache_level);
> + if (ret) {
> + dev_err(dev, "%pOF: missing cache-level\n", cache_np);
> + goto out_put;
> + }
> +
> + if (cache_level != 2 && cache_level != 3) {
> + dev_err(dev, "%pOF: unsupported cache-level %u\n",
> + cache_np, cache_level);
> + ret = -EINVAL;
> + goto out_put;
> + }
> +
> + /*
> + * Associate the controller with its cache instance via cacheinfo.
> + * The matching cache provides the cache id used as the resctrl
> + * domain id.
> + *
> + * Taking the first leaf at the level is sufficient because a
> + * unified cache node yields one cacheinfo leaf and
> + * cache_setup_of_node() walks one cache node per leaf past L1, so
> + * the first leaf at the level has that level's node as fw_token.
> + *
> + * get_cpu_cacheinfo_level() returns the leaf only when cacheinfo
> + * assigned it an id (CACHE_ID), which the resctrl domain id needs.
> + */
> + cpus_read_lock();
> + for_each_online_cpu(cpu) {
> + struct cacheinfo *ci = get_cpu_cacheinfo_level(cpu, cache_level);
> +
> + if (ci && ci->fw_token == cache_np) {
> + cache_id = ci->id;
> + found = true;
> + break;
> + }
> + }
> + cpus_read_unlock();
> +
> + if (!found) {
> + dev_err(dev, "%pOF: no online hart reports an L%u cache with a cache id for this node\n",
> + cache_np, cache_level);
> + ret = -ENODEV;
> + goto out_put;
> + }
> +
> + info.type = CBQRI_CONTROLLER_TYPE_CAPACITY;
> + info.base = base;
> + info.addr = res->start;
> + info.size = resource_size(res);
> + info.rcid_count = rcid_count;
> + info.cache_id = cache_id;
> + info.cache_level = cache_level;
info = (struct cbqri_controller_info) {
.type = CBQRI_CONTROLLER_TYPE_CAPACITY,
.base = base,
.addr = res->start,
.size = resource_size,
...

Might as well as you nicely set everything here.
It's more readable and also makes it clear anything not set
is zeroed without anyone having to go check the = { }; above.


};
> +
> + ret = riscv_cbqri_register_cc(&info);
> + if (ret) {
> + dev_err(dev, "failed to register capacity controller: %d\n", ret);
> + goto out_put;
> + }
> +
> + dev_info(dev, "registered L%u capacity controller at %pa (cache_id=%u, rcid=%u)\n",
> + cache_level, &info.addr, cache_id, rcid_count);
> +
> +out_put:
> + of_node_put(cache_np);
> + return ret;
> +}