[PATCH 3/3] drm/xe/vf: Add proper detection of the SR-IOV VF mode

[Matt Roper]

On Wed, Mar 27, 2024 at 07:27:40PM +0100, Michal Wajdeczko wrote:
> SR-IOV VF mode detection is based on testing VF capability bit on
> the register that is accessible from both the PF and enabled VFs.
> 
> Bspec: 49904, 53227
> Signed-off-by: Michal Wajdeczko <michal.wajdeczko at intel.com>
> ---
>  drivers/gpu/drm/xe/regs/xe_sriov_regs.h |  3 +++
>  drivers/gpu/drm/xe/xe_sriov.c           | 17 +++++++++++++++--
>  2 files changed, 18 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/gpu/drm/xe/regs/xe_sriov_regs.h b/drivers/gpu/drm/xe/regs/xe_sriov_regs.h
> index 58a4e0fad1e1..617ddb84b7fa 100644
> --- a/drivers/gpu/drm/xe/regs/xe_sriov_regs.h
> +++ b/drivers/gpu/drm/xe/regs/xe_sriov_regs.h
> @@ -14,4 +14,7 @@
>  #define   LMEM_EN			REG_BIT(31)
>  #define   LMTT_DIR_PTR			REG_GENMASK(30, 0) /* in multiples of 64KB */
>  
> +#define VF_CAP_REG			XE_REG(0x1901f8, XE_REG_OPTION_VF)
> +#define   VF_CAP			REG_BIT(0)
> +
>  #endif
> diff --git a/drivers/gpu/drm/xe/xe_sriov.c b/drivers/gpu/drm/xe/xe_sriov.c
> index 2bcef998c8a9..3e103edf7174 100644
> --- a/drivers/gpu/drm/xe/xe_sriov.c
> +++ b/drivers/gpu/drm/xe/xe_sriov.c
> @@ -5,7 +5,11 @@
>  
>  #include <drm/drm_managed.h>
>  
> +#include "regs/xe_sriov_regs.h"
> +
>  #include "xe_assert.h"
> +#include "xe_device.h"
> +#include "xe_mmio.h"
>  #include "xe_sriov.h"
>  
>  /**
> @@ -28,6 +32,13 @@ const char *xe_sriov_mode_to_string(enum xe_sriov_mode mode)
>  	}
>  }
>  
> +static bool test_is_vf(struct xe_device *xe)
> +{
> +	u32 value = xe_mmio_read32(xe_root_mmio_gt(xe), VF_CAP_REG);
> +
> +	return value & VF_CAP;
> +}
> +
>  /**
>   * xe_sriov_probe_early - Probe a SR-IOV mode.
>   * @xe: the &xe_device to probe mode on
> @@ -44,8 +55,10 @@ void xe_sriov_probe_early(struct xe_device *xe)
>  	enum xe_sriov_mode mode = XE_SRIOV_MODE_NONE;
>  	bool has_sriov = xe->info.has_sriov;
>  
> -	/* TODO: replace with proper mode detection */
> -	xe_assert(xe, !has_sriov);
> +	if (has_sriov) {
> +		if (test_is_vf(xe))

Seems like we could && these conditions together and avoid the nested
if's.  But up to you; either way,

Reviewed-by: Matt Roper <matthew.d.roper at intel.com>

> +			mode = XE_SRIOV_MODE_VF;
> +	}
>  
>  	xe_assert(xe, !xe->sriov.__mode);
>  	xe->sriov.__mode = mode;
> -- 
> 2.43.0
> 
> 

-- 
Matt Roper
Graphics Software Engineer
Linux GPU Platform Enablement
Intel Corporation