qemu-riscv
[Top][All Lists]
Advanced

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

Re: [PATCH RESEND 1/6] target/riscv: Introduce extension implied rules d


From: Frank Chang
Subject: Re: [PATCH RESEND 1/6] target/riscv: Introduce extension implied rules definition
Date: Tue, 11 Jun 2024 09:56:29 +0800

Hi Alistair,

On Tue, Jun 11, 2024 at 9:35 AM Alistair Francis <alistair23@gmail.com> wrote:
On Wed, Jun 5, 2024 at 4:35 PM <frank.chang@sifive.com> wrote:
>
> From: Frank Chang <frank.chang@sifive.com>
>
> RISCVCPUImpliedExtsRule is created to store the implied rules.
> 'is_misa' flag is used to distinguish whether the rule is derived
> from the MISA or other extensions.
> 'ext' stores the MISA bit if 'is_misa' is true. Otherwise, it stores
> the offset of the extension defined in RISCVCPUConfig. 'ext' will also
> serve as the key of the hash tables to look up the rule in the following
> commit.
>
> Signed-off-by: Frank Chang <frank.chang@sifive.com>
> ---
>  target/riscv/cpu.c |  8 ++++++++
>  target/riscv/cpu.h | 18 ++++++++++++++++++
>  2 files changed, 26 insertions(+)
>
> diff --git a/target/riscv/cpu.c b/target/riscv/cpu.c
> index cee6fc4a9a..c7e5cec7ef 100644
> --- a/target/riscv/cpu.c
> +++ b/target/riscv/cpu.c
> @@ -2242,6 +2242,14 @@ RISCVCPUProfile *riscv_profiles[] = {
>      NULL,
>  };
>
> +RISCVCPUImpliedExtsRule *riscv_misa_implied_rules[] = {
> +    NULL
> +};
> +
> +RISCVCPUImpliedExtsRule *riscv_ext_implied_rules[] = {
> +    NULL
> +};
> +
>  static Property riscv_cpu_properties[] = {
>      DEFINE_PROP_BOOL("debug", RISCVCPU, cfg.debug, true),
>
> diff --git a/target/riscv/cpu.h b/target/riscv/cpu.h
> index 1501868008..b5a036cf27 100644
> --- a/target/riscv/cpu.h
> +++ b/target/riscv/cpu.h
> @@ -122,6 +122,24 @@ typedef enum {
>      EXT_STATUS_DIRTY,
>  } RISCVExtStatus;
>
> +typedef struct riscv_cpu_implied_exts_rule RISCVCPUImpliedExtsRule;
> +
> +struct riscv_cpu_implied_exts_rule {
> +    /* Bitmask indicates the rule enabled status for the harts. */
> +    uint64_t enabled;

I'm not clear why we need this

This is because a rule may be implied more than once.
e.g. Zcf implies RVF, Zfa also implies RVF.
There's no need to check RVF's implied rule again for Zfa after Zcf's implied rules are enabled.

The implied rules are checked recursively, so once the rule has been enabled (per-CPU basis),
the rule (and all its implied rules) will not be rechecked.
 
Regards,
Frank Chang


Alistair

> +    /* True if this is a MISA implied rule. */
> +    bool is_misa;
> +    /* ext is MISA bit if is_misa flag is true, else extension offset. */
> +    const uint32_t ext;
> +    const uint32_t implied_misas;
> +    const uint32_t implied_exts[];
> +};
> +
> +extern RISCVCPUImpliedExtsRule *riscv_misa_implied_rules[];
> +extern RISCVCPUImpliedExtsRule *riscv_ext_implied_rules[];
> +
> +#define RISCV_IMPLIED_EXTS_RULE_END -1
> +
>  #define MMU_USER_IDX 3
>
>  #define MAX_RISCV_PMPS (16)
> --
> 2.43.2
>
>

reply via email to

[Prev in Thread] Current Thread [Next in Thread]