On Wed, Jul 05, 2017 at 04:27:08PM +0800, Jun Nie wrote:
> Adjust FIFO threshold according to FIFO depth. Skip
> the adjustment if we do not have FIFO depth info.
> 

So, this is a big improvement in readability - but some of my generic
style comments do not appear to have been addressed.

> Contributed-under: TianoCore Contribution Agreement 1.0
> Signed-off-by: Jun Nie <jun....@linaro.org>
> ---
>  EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmc.h      |  6 ++++
>  EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.c   | 50 
> +++++++++++++++++++++++++++++
>  EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.inf |  1 +
>  EmbeddedPkg/EmbeddedPkg.dec                 |  1 +
>  4 files changed, 58 insertions(+)
> 
> diff --git a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmc.h 
> b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmc.h
> index 055f1e0..90c7676 100644
> --- a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmc.h
> +++ b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmc.h
> @@ -38,7 +38,10 @@
>  #define DWEMMC_RINTSTS          ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x044)
>  #define DWEMMC_STATUS           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x048)
>  #define DWEMMC_FIFOTH           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x04c)
> +#define DWEMMC_TCBCNT           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x05c)
> +#define DWEMMC_TBBCNT           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x060)
>  #define DWEMMC_DEBNCE           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x064)
> +#define DWEMMC_HCON             ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x070)
>  #define DWEMMC_UHSREG           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x074)
>  #define DWEMMC_BMOD             ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x080)
>  #define DWEMMC_DBADDR           ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x088)
> @@ -47,6 +50,7 @@
>  #define DWEMMC_DSCADDR          ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x094)
>  #define DWEMMC_BUFADDR          ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0x098)
>  #define DWEMMC_CARDTHRCTL       ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0X100)
> +#define DWEMMC_DATA             ((UINT32)PcdGet32 (PcdDwEmmcDxeBaseAddress) 
> + 0X200)
>  
>  #define CMD_UPDATE_CLK                          0x80202000
>  #define CMD_START_BIT                           (1 << 31)
> @@ -124,4 +128,6 @@
>  #define DWEMMC_CARD_RD_THR(x)                   ((x & 0xfff) << 16)
>  #define DWEMMC_CARD_RD_THR_EN                   (1 << 0)
>  
> +#define DWEMMC_GET_HDATA_WIDTH(x)               (((x) >> 7) & 0x7)
> +
>  #endif  // __DWEMMC_H__
> diff --git a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.c 
> b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.c
> index bb26b69..70e064d 100644
> --- a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.c
> +++ b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.c
> @@ -415,6 +415,55 @@ DwEmmcReceiveResponse (
>    return EFI_SUCCESS;
>  }
>  
> +VOID DwEmmcAdjustFifothreshold (

This still needs to be:
VOID
DwEmmcAdjustFifoThreshold (

> +  VOID
> +  )
> +{
> +  /* DMA multiple transaction size map to reg value as array index */
> +  CONST UINT32 BurstSize[] = {1, 4, 8, 16, 32, 64, 128, 256};
> +  UINT32 BlkDepthInFifo, Fifoth, FifoWidth, FifoDepth;

Fifoth -> FifoThreshold.

> +  UINT32 BlkSize = DWEMMC_BLOCK_SIZE, Idx = 0, RxWmark = 1, TxWmark, 
> TxWmarkInvers;

RxWmark -> RxWaterMark.
TxWmark -> TxWaterMark.
TxWmarkInvers -> TxWaterMarkInverse.

> +
> +  /* Skip FIFO adjustment if we do not have platform FIFO depth info */
> +  FifoDepth = PcdGet32 (PcdDwEmmcDxeFifoDepth);
> +  if (!FifoDepth) {
> +    return;
> +  }
> +
> +  TxWmark = FifoDepth / 2;
> +  TxWmarkInvers = FifoDepth - TxWmark;
> +
> +  FifoWidth = DWEMMC_GET_HDATA_WIDTH (MmioRead32 (DWEMMC_HCON));
> +  if (!FifoWidth) {
> +    FifoWidth = 2;
> +  } else if (FifoWidth == 2) {
> +    FifoWidth = 8;
> +  } else {
> +    FifoWidth = 4;
> +  }
> +
> +  BlkDepthInFifo = BlkSize / FifoWidth;
> +
> +  /* if BlkSize is not a multiple of the FIFO width */

This comment still only says exactly what the code says.
Why are we done if BlkSize is not a multiple of FifoWidth?
Is this situation possible? What does it indicate?

/
    Leif

> +  if (BlkSize % FifoWidth) {
> +    goto done;
> +  }
> +
> +  Idx = ARRAY_SIZE (BurstSize) - 1;
> +  while (Idx && ((BlkDepthInFifo % BurstSize[Idx]) || (TxWmarkInvers % 
> BurstSize[Idx]))) {
> +    Idx--;
> +  }
> +  RxWmark = BurstSize[Idx] - 1;
> +  /*
> +   * If Idx is '0', it won't be tried
> +   * Thus, initial values are used
> +   */
> +done:
> +  Fifoth = DWEMMC_DMA_BURST_SIZE (Idx) | DWEMMC_FIFO_TWMARK (TxWmark)
> +           | DWEMMC_FIFO_RWMARK (RxWmark);
> +  MmioWrite32 (DWEMMC_FIFOTH, Fifoth);
> +}
> +
>  EFI_STATUS
>  PrepareDmaData (
>    IN DWEMMC_IDMAC_DESCRIPTOR*    IdmacDesc,
> @@ -633,6 +682,7 @@ DwEmmcDxeInitialize (
>  
>    Handle = NULL;
>  
> +  DwEmmcAdjustFifothreshold ();
>    gpIdmacDesc = (DWEMMC_IDMAC_DESCRIPTOR *)AllocatePages 
> (DWEMMC_MAX_DESC_PAGES);
>    if (gpIdmacDesc == NULL) {
>      return EFI_BUFFER_TOO_SMALL;
> diff --git a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.inf 
> b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.inf
> index 99b4f99..bc4413e 100644
> --- a/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.inf
> +++ b/EmbeddedPkg/Drivers/DwEmmcDxe/DwEmmcDxe.inf
> @@ -49,6 +49,7 @@
>    gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeBaseAddress
>    gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeClockFrequencyInHz
>    gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeMaxClockFreqInHz
> +  gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeFifoDepth
>  
>  [Depex]
>    TRUE
> diff --git a/EmbeddedPkg/EmbeddedPkg.dec b/EmbeddedPkg/EmbeddedPkg.dec
> index 2da9b2f..5f39d9d 100644
> --- a/EmbeddedPkg/EmbeddedPkg.dec
> +++ b/EmbeddedPkg/EmbeddedPkg.dec
> @@ -170,6 +170,7 @@
>    gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeBaseAddress|0x0|UINT32|0x00000035
>    
> gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeClockFrequencyInHz|0x0|UINT32|0x00000036
>    gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeMaxClockFreqInHz|0x0|UINT32|0x00000037
> +  gEmbeddedTokenSpaceGuid.PcdDwEmmcDxeFifoDepth|0x0|UINT32|0x00000038
>  
>    #
>    # Android FastBoot
> -- 
> 1.9.1
> 
_______________________________________________
edk2-devel mailing list
edk2-devel@lists.01.org
https://lists.01.org/mailman/listinfo/edk2-devel

Reply via email to