[igt-dev] [PATCH i-g-t v7 2/3] lib/intel_mmio: add funtions for read/write register funtions

[Chris Wilson]

Quoting Daniel Mrzyglod (2019-08-16 15:53:16)
>     This patch is first move to extend functionality of intel_mmio library.
>     There was limitation for 1 device, adding pointer for IO functions to
>     mmaped area gives us possibility to use those IO functions for other mmaped
>     devices.
> 
>     v4: reword commitmsg, spelling errors
> 
>     Cc: Antonio Argenziano <antonio.argenziano at intel.com>
>     Cc: Daniele Spurio Ceraolo <daniele.ceraolospurio at intel.com>
>     Cc: Katarzyna Dec <katarzyna.dec at intel.com>
>     Cc: Chris Wilson <chris at chris-wilson.co.uk>
>     Cc: Petri Latvala <petri.latvala at intel.com>
>     Cc: Zbigniew Kempczyński <zbigniew.kempczynski at intel.com>
> 
>     Reviewed-by: Zbigniew Kempczyński <zbigniew.kempczynski at intel.com>
>     Signed-off-by: Daniel Mrzyglod <daniel.t.mrzyglod at intel.com>
> ---
>  lib/intel_io.h   |   7 ++++
>  lib/intel_mmio.c | 107 +++++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 114 insertions(+)
> 
> diff --git a/lib/intel_io.h b/lib/intel_io.h
> index 6014c485..99cb1829 100644
> --- a/lib/intel_io.h
> +++ b/lib/intel_io.h
> @@ -49,6 +49,13 @@ void OUTREG(uint32_t reg, uint32_t val);
>  void OUTREG16(uint32_t reg, uint16_t val);
>  void OUTREG8(uint32_t reg, uint8_t val);
>  
> +uint32_t INREG_DEV(void *igt_mmio, uint32_t reg);
> +uint16_t INREG16_DEV(void *igt_mmio, uint32_t reg);
> +uint8_t INREG8_DEV(void *igt_mmio, uint32_t reg);
> +void OUTREG_DEV(void *igt_mmio, uint32_t reg, uint32_t val);
> +void OUTREG16_DEV(void *igt_mmio, uint32_t reg, uint16_t val);
> +void OUTREG8_DEV(void *igt_mmio, uint32_t reg, uint8_t val);
> +
>  /* sideband access functions from intel_iosf.c */
>  uint32_t intel_dpio_reg_read(uint32_t reg, int phy);
>  void intel_dpio_reg_write(uint32_t reg, uint32_t val, int phy);
> diff --git a/lib/intel_mmio.c b/lib/intel_mmio.c
> index a5458aeb..01411c37 100644
> --- a/lib/intel_mmio.c
> +++ b/lib/intel_mmio.c
> @@ -405,3 +405,110 @@ void OUTREG8(uint32_t reg, uint8_t val)
>  {
>         *((volatile uint8_t *)igt_global_mmio + reg) = val;
>  }
> +
> +
> +/**
> + * INREG_DEV:
> + * @igt_mmio maped memory pointer
> + * @reg: register offset
> + *
> + * 32-bit read of the register at offset @reg. This function only works when the
> + * new register access helper is initialized with intel_register_access_init().
> + *
> + * This function directly accesses the igt_mmio without safety checks.
> + *
> + * Returns:
> + * The value read from the register.
> + */
> +uint32_t INREG_DEV(void *igt_mmio, uint32_t reg)
> +{
> +       return *(volatile uint32_t *)((volatile char *)igt_mmio + reg);

We allow arithmetic on void*, the first cast is unnecessary.

Is there a good reason why these are not inlines? The cost of function
call looks to be greater than the body?
-Chris