Re: [PATCH RFC v4 18/18] riscv: enable resctrl filesystem for Ssqosid

On Tue, May 12, 2026 at 10:56:48PM +0000, [email protected] wrote:
> This commit updates the RISCV_ISA_SSQOSID Kconfig option to select
> RISCV_CBQRI_DRIVER. It conditionally selects the driver only if
> RESCTRL_FS is enabled, aiming to keep the filesystem as a user choice.
> 
> > RISCV_ISA_SSQOSID selects RISCV_CBQRI_DRIVER, which in turn depends
> > on RESCTRL_FS. Enabling the resctrl filesystem itself stays a user
> 
> Is this description accurate?
> 
> The driver configuration shows that RISCV_CBQRI_DRIVER only depends on
> RISCV && RISCV_ISA_SSQOSID. The filesystem integration is decoupled into
> a separate silent config, so the driver does not actually depend on
> the filesystem.

The commit message is wrong. I will rewrite it to describe the actual
wiring.

> > diff --git a/arch/riscv/Kconfig b/arch/riscv/Kconfig
> > --- a/arch/riscv/Kconfig
> > +++ b/arch/riscv/Kconfig
> > @@ -596,6 +596,7 @@ config RISCV_ISA_SSQOSID
> >  	depends on 64BIT
> >  	default n
> >  	select ARCH_HAS_CPU_RESCTRL
> > +	select RISCV_CBQRI_DRIVER if RESCTRL_FS
> 
> Could this conditional selection lead to unexpected configurations?
> 
> Because the driver does not actually depend on the filesystem, if a user
> enables the SSQOSID extension but leaves RESCTRL_FS disabled, the
> underlying hardware driver will unexpectedly not be automatically
> selected. 
> 
> This differs from platforms like ARM64, where ARM64_MPAM unconditionally
> selects ARM64_MPAM_DRIVER regardless of the filesystem configuration.

The 'if RESCTRL_FS' guard is wrong. I will drop it so that enabling
SSQOSID always pulls in the controller driver.

-Drew