-----Original Message-----
From: devel@edk2.groups.io <devel@edk2.groups.io> On Behalf Of Aiman
Rosli
Sent: Monday, January 17, 2022 8:56 PM
To: devel@edk2.groups.io
Cc: Rosli, Muhammad Aiman <muhammad.aiman.rosli@...>
Subject: [edk2-devel] [PATCH v2] MdeModulePkg: Enabling OS boot from SD
card through UEFI payload

This changes is by adding 50ms delay during voltage switching from 3.3V to
1.8V, plus adding a goto Voltage33Retry for 3.3V checking and retrying.

Signed-off-by: Aiman Rosli <muhammad.aiman.rosli@...>
---
MdeModulePkg/Bus/Pci/SdMmcPciHcDxe/SdDevice.c | 23
++++++++++++++++---
1 file changed, 20 insertions(+), 3 deletions(-)

diff --git a/MdeModulePkg/Bus/Pci/SdMmcPciHcDxe/SdDevice.c
b/MdeModulePkg/Bus/Pci/SdMmcPciHcDxe/SdDevice.c
index 662f9f483c..527fe7dc20 100644
--- a/MdeModulePkg/Bus/Pci/SdMmcPciHcDxe/SdDevice.c
+++ b/MdeModulePkg/Bus/Pci/SdMmcPciHcDxe/SdDevice.c
@@ -1213,9 +1213,14 @@ SdCardIdentification (
UINT32 PresentState;
UINT8 HostCtrl2;
UINTN Retry;
+ BOOLEAN ForceVoltage33;
+
+ ForceVoltage33 = FALSE;

PciIo = Private->PciIo;
PassThru = &Private->PassThru;
+
+Voltage33Retry:
//
// 1. Send Cmd0 to the device
//
@@ -1294,6 +1299,14 @@ SdCardIdentification (
return EFI_UNSUPPORTED;
}

+ //
+ // 1.8V had failed in the previous run, forcing a retry with 3.3V
+ instead // if (ForceVoltage33 == TRUE) {
+ S18r = FALSE;
+ ForceVoltage33 = FALSE;

Like I mentioned in https://github.com/tianocore/edk2/pull/2413#discussion_r785611042
I do not think "ForceVoltage33 = FALSE" is necessary, could you help to double check?

+ }
+
//
// 5. Repeatly send Acmd41 with supply voltage window to the device.
// Note here we only support the cards complied with SD physical
@@ -1362,13 +1375,17 @@ SdCardIdentification (
goto Error;
}

- gBS->Stall (1000);
+ // Workaround to add a delay of 50 ms in order for clock to stabilize
before turning on the SD card again.
+ gBS->Stall (50000);

Sorry for a question.
As I mentioned in:
https://github.com/tianocore/edk2/pull/2413#discussion_r785610179
https://github.com/tianocore/edk2/pull/2413#discussion_r785610742

Can this 50 milliseconds delay be put in the below 1.8V switch failure handling logic?
Enlarging the delay (from 1ms to 50ms) in the normal initialization flow will bring performance impact.

SdMmcHcRwMmio (PciIo, Slot, SD_MMC_HC_PRESENT_STATE, TRUE,
sizeof (PresentState), &PresentState);
if (((PresentState >> 20) & 0xF) != 0xF) {
DEBUG ((DEBUG_ERROR, "SdCardIdentification: SwitchVoltage fails with
PresentState = 0x%x, It should be 0xF\n", PresentState));
- Status = EFI_DEVICE_ERROR;
- goto Error;
+ Status = SdMmcHcReset (Private, Slot);
+ Status = SdMmcHcInitHost (Private, Slot);

Could you help to check my comment in:
https://github.com/tianocore/edk2/pull/2413#discussion_r785610742
to 1) add comments and 2) refine the error handling for the "reset and reinitialize the slot before the 3.3V retry"

Best Regards,
Hao Wu

+ ForceVoltage33 = TRUE;
+ DEBUG ((DEBUG_ERROR, "SdCardIdentification: Switching to 1.8V had
failed in the previous run, forcing a retry with 3.3V instead\n"));
+ goto Voltage33Retry;
}
}

--
2.34.1.windows.1