Re: [PATCH 4/7] hw/riscv: implement CBQRI bandwidth controller

[Drew Fustini <fustini@kernel.org>]

On Mon, Nov 24, 2025 at 06:06:21PM +0100, Radim Krčmář wrote:
> 2025-11-19T16:42:20-08:00, Drew Fustini <fustini@kernel.org>:
> > From: Nicolas Pitre <npitre@baylibre.com>
> >
> > Implement a bandwidth controller according to the Capacity and Bandwidth
> > QoS Register Interface (CBQRI) which supports these capabilities:
> >
> >   - Number of access types: 2 (code and data)
> >   - Usage monitoring operations: CONFIG_EVENT, READ_COUNTER
> >   - Event IDs supported: None, Total read/write byte count, Total
> >                          read byte count, Total write byte count
> >   - Bandwidth allocation operations: CONFIG_LIMIT, READ_LIMIT
> >
> > Link: https://github.com/riscv-non-isa/riscv-cbqri/blob/main/riscv-cbqri.pdf
> > Signed-off-by: Nicolas Pitre <npitre@baylibre.com>
> > [fustini: add fields introduced in the ratified spec: rpfx and p]
> > Signed-off-by: Drew Fustini <fustini@kernel.org>
> > ---
> > diff --git a/hw/riscv/cbqri_bandwidth.c b/hw/riscv/cbqri_bandwidth.c
> > +static uint32_t bandwidth_config(RiscvCbqriBandwidthState *bc,
> > +                                 uint32_t rcid, uint32_t at,
> > +                                 bool *busy)
> > +{
> > +    BandwidthAllocation *bw_alloc = get_bw_alloc(bc, rcid, at);
> > +
> > +    /* for now we only preserve the current BC_BW_ALLOC register content */
> 
> There are a few checks that we could do even without any real
> accounting, like rbwb > 0, and sum_{rcid}(rcid.rbwb) <= mrbwb.

Good point, I'll add basic checks like that.

> Doing so might help us find some bugs in the software management layer.
> What is the level of ISA correctness that you're aiming for?
> (I'll tone down my review if nitpicks like that are not needed.)

The original goal for the Qemu code was for the software proof of
concept needed to freeze the spec. I am working on upstreaming the qemu
code in order to have a "hardware" implementation to reference when
upstreaming the Linux kernel support for resctrl. I do also have access
to CBQRI controllers in RTL emulation but that won't be very useful for
the purposes of public review of my Linux resctrl patches.

Thanks,
Drew