Re: [PATCH v2] MdeModulePkg/AtaAtapiPassThru: relax PHY detect timeout

[Ard Biesheuvel <ard.biesheuvel@linaro.org>]

On 28 June 2017 at 08:31, Zeng, Star <star.zeng@intel.com> wrote:
> The updated comments "Wait at least 10 ms" seems not correct.
>

That is the whole point of the change. The SATA spec mandates that the
PHY respond within 10 ms. It does *not* mandate that the software wait
10 ms or less, rather the opposite, i.e., that the software should
wait 10 ms or *more*, so 'at least 10ms'. The original comment said
that the software *must* wait for no more than 10 ms, but this is not
what the spec says.



> -----Original Message-----
> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of Ard Biesheuvel
> Sent: Wednesday, June 28, 2017 4:23 PM
> To: edk2-devel@lists.01.org; Zeng, Star <star.zeng@intel.com>
> Cc: Tian, Feng <feng.tian@intel.com>; Dong, Eric <eric.dong@intel.com>; leif.lindholm@linaro.org; Ard Biesheuvel <ard.biesheuvel@linaro.org>
> Subject: [edk2] [PATCH v2] MdeModulePkg/AtaAtapiPassThru: relax PHY detect timeout
>
> The SATA spec mandates that link detection by the PHY completes within
> 10 ms after receiving a reset signal. However, there is no obligation to uphold this requirement at the driver end as strictly as we do, and as it turns out, some combinations of host and device (e.g., Samsung
> 850 EVO connected to a LeMaker Cello) are only borderline compliant, which means the device is not detected reliably.
>
> So let's allow for a bit of margin, and increase the PHY detect timeout value to 15 ms.
>
> Contributed-under: TianoCore Contribution Agreement 1.0
> Signed-off-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
> ---
> v2: update comment in AhciModeInitialization() as well
>
>  MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.c | 5 +++--  MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.h | 3 ++-
>  2 files changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.c b/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.c
> index 4d01c1dd7fca..4418e5c3763e 100644
> --- a/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.c
> +++ b/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.c
> @@ -2376,8 +2376,9 @@ AhciModeInitialization (
>        AhciOrReg (PciIo, Offset, EFI_AHCI_PORT_CMD_FRE);
>
>        //
> -      // Wait no longer than 10 ms to wait the Phy to detect the presence of a device.
> -      // It's the requirment from SATA1.0a spec section 5.2.
> +      // Wait at least 10 ms for the Phy to detect the presence of a device.
> +      // It's the requirement from SATA1.0a spec section 5.2.
> +      // Add a bit of margin for robustness.
>        //
>        PhyDetectDelay = EFI_AHCI_BUS_PHY_DETECT_TIMEOUT;
>        Offset = EFI_AHCI_PORT_START + Port * EFI_AHCI_PORT_REG_WIDTH + EFI_AHCI_PORT_SSTS; diff --git a/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.h b/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.h
> index 6401fb2e9fcd..809bcc307fc4 100644
> --- a/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.h
> +++ b/MdeModulePkg/Bus/Ata/AtaAtapiPassThru/AhciMode.h
> @@ -41,8 +41,9 @@ typedef union {
>
>  //
>  // Refer SATA1.0a spec section 5.2, the Phy detection time should be less than 10ms.
> +// Add a bit of margin for robustness.
>  //
> -#define  EFI_AHCI_BUS_PHY_DETECT_TIMEOUT       10
> +#define  EFI_AHCI_BUS_PHY_DETECT_TIMEOUT       15
>  //
>  // Refer SATA1.0a spec, the FIS enable time should be less than 500ms.
>  //
> --
> 2.9.3
>
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org
> https://lists.01.org/mailman/listinfo/edk2-devel