Re: [PATCH v4 03/16] riscv: vector: Use vlenb from DT

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Apr 26, 2024 at 02:29:17PM -0700, Charlie Jenkins wrote:
> If vlenb is provided in the device tree, prefer that over reading the
> vlenb csr.
> 
> Signed-off-by: Charlie Jenkins <charlie@xxxxxxxxxxxx>
> ---
>  arch/riscv/include/asm/cpufeature.h |  2 ++
>  arch/riscv/kernel/cpufeature.c      | 43 +++++++++++++++++++++++++++++++++++++
>  arch/riscv/kernel/vector.c          | 12 ++++++++++-
>  3 files changed, 56 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/riscv/include/asm/cpufeature.h b/arch/riscv/include/asm/cpufeature.h
> index 347805446151..0c4f08577015 100644
> --- a/arch/riscv/include/asm/cpufeature.h
> +++ b/arch/riscv/include/asm/cpufeature.h
> @@ -31,6 +31,8 @@ DECLARE_PER_CPU(struct riscv_cpuinfo, riscv_cpuinfo);
>  /* Per-cpu ISA extensions. */
>  extern struct riscv_isainfo hart_isa[NR_CPUS];
>  
> +extern u32 riscv_vlenb_of;
> +
>  void riscv_user_isa_enable(void);
>  
>  #if defined(CONFIG_RISCV_MISALIGNED)
> diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> index 3ed2359eae35..8158f34c3e36 100644
> --- a/arch/riscv/kernel/cpufeature.c
> +++ b/arch/riscv/kernel/cpufeature.c
> @@ -35,6 +35,8 @@ static DECLARE_BITMAP(riscv_isa, RISCV_ISA_EXT_MAX) __read_mostly;
>  /* Per-cpu ISA extensions. */
>  struct riscv_isainfo hart_isa[NR_CPUS];
>  
> +u32 riscv_vlenb_of;
> +
>  /**
>   * riscv_isa_extension_base() - Get base extension word
>   *
> @@ -648,6 +650,42 @@ static int __init riscv_isa_fallback_setup(char *__unused)
>  early_param("riscv_isa_fallback", riscv_isa_fallback_setup);
>  #endif
>  
> +static int riscv_homogeneous_vlenb(void)

Without a verb, this function name is rather odd.

> +{
> +	int cpu;
> +	u32 prev_vlenb = 0;
> +	u32 vlenb;
> +
> +	for_each_possible_cpu(cpu) {
> +		struct device_node *cpu_node;
> +
> +		cpu_node = of_cpu_device_node_get(cpu);
> +		if (!cpu_node) {
> +			pr_warn("Unable to find cpu node\n");
> +			continue;

Hmm, if we fail to find the cpu node, then shouldn't we be returning an
error?

> +		}
> +
> +		if (of_property_read_u32(cpu_node, "riscv,vlenb", &vlenb)) {
> +			of_node_put(cpu_node);
> +
> +			if (prev_vlenb)
> +				return -1;

Can you return an errno here and below please?

> +			continue;
> +		}
> +
> +		if (prev_vlenb && vlenb != prev_vlenb) {
> +			of_node_put(cpu_node);
> +			return -1;
> +		}
> +
> +		prev_vlenb = vlenb;
> +		of_node_put(cpu_node);
> +	}
> +
> +	riscv_vlenb_of = vlenb;
> +	return 0;
> +}
> +
>  void __init riscv_fill_hwcap(void)
>  {
>  	char print_str[NUM_ALPHA_EXTS + 1];
> @@ -671,6 +709,11 @@ void __init riscv_fill_hwcap(void)
>  			pr_info("Falling back to deprecated \"riscv,isa\"\n");
>  			riscv_fill_hwcap_from_isa_string(isa2hwcap);
>  		}
> +
> +		if (riscv_homogeneous_vlenb() < 0) {
> +			pr_warn("RISCV_ISA_V only supports one vlenb on SMP systems. Please ensure that the riscv,vlenb devicetree property is the same across all CPUs. Either all CPUs must have the riscv,vlenb property, or none. If no CPUs in the devicetree use riscv,vlenb then vlenb will be probed from the vlenb CSR. Disabling vector.\n");

Oh dear, that's a bit unwieldy... I think you could get away with a far
more basic message - and you should be able to break this over lines,
adjacent string literals should get concatenated.
I'd probably say something like "unsupported heterogeneous vlen detected,
vector extension disabled", however we should actually check that the
vector extension has been detected on all CPUs and that kernel support
for vector is enabled before emitting a warning for this.

Cheers,
Conor.

> +			elf_hwcap &= ~COMPAT_HWCAP_ISA_V;
> +		}
>  	}
>  
>  	/*
> diff --git a/arch/riscv/kernel/vector.c b/arch/riscv/kernel/vector.c
> index 6727d1d3b8f2..e04586cdb7f0 100644
> --- a/arch/riscv/kernel/vector.c
> +++ b/arch/riscv/kernel/vector.c
> @@ -33,7 +33,17 @@ int riscv_v_setup_vsize(void)
>  {
>  	unsigned long this_vsize;
>  
> -	/* There are 32 vector registers with vlenb length. */
> +	/*
> +	 * There are 32 vector registers with vlenb length.
> +	 *
> +	 * If the riscv,vlenb property was provided by the firmware, use that
> +	 * instead of probing the CSRs.
> +	 */
> +	if (riscv_vlenb_of) {
> +		this_vsize = riscv_vlenb_of * 32;
> +		return 0;
> +	}
> +
>  	riscv_v_enable();
>  	this_vsize = csr_read(CSR_VLENB) * 32;
>  	riscv_v_disable();
> 
> -- 
> 2.44.0
> 

Attachment: signature.asc
Description: PGP signature


[Index of Archives]     [Device Tree Compilter]     [Device Tree Spec]     [Linux Driver Backports]     [Video for Linux]     [Linux USB Devel]     [Linux PCI Devel]     [Linux Audio Users]     [Linux Kernel]     [Linux SCSI]     [XFree86]     [Yosemite Backpacking]


  Powered by Linux