All,
I'm wondering if anyone is successfully using sdmmc on stm32h7 in the
current tree (version 13.1.0)
I'm trying to add sdmmc support to newly added stm32h7s8-dk board (based
on the newer stm32h7s3l8) and running into SDMMC initialization issues
that look to originate in mmcsd_cardidentify() where the initial CMD0
command succeeds(since command went on the wire with no expected
response) and the following CMD1 also reports as succeeding (leading
mmcsd_cardidentify() to believe the attached card is an MMC card instead
of an SD card).
The issue looks to originate from the call to mmcsd_sendcmdpoll(priv,
MMC_CMD1, ...) followed by SDIO_RECVR3(...) that returns OK (indicating
a response when the attached trace diagram clearly shows at marker 0
that the SDMMC controller floats the CMD line, but the card does not
pull it down to initiate a response) I believe the problem is that
mmcsd_sendcmdpoll() invokes SDIO_SENDCMD then SDIO_WAITRESPONSE() to get
a repsonse, and on return mmcsd_cardidentify() invokes SDIO_RECVR3() to
determine if the response is valid.
SDIO_WAITRESPONSE() invokes stm32_waitresponse() which waits for
SDMMC_STA to indicate either CTIMEOUT or CCRCFAIL (0x45) and when it
sees 0x40(CTIMEOUT), sets both CTIMEOUT and CRCFAIL bits in SDMMC_ICR to
clear the corresponding bits in SDMMC_STA. Then when SDIO_RECV3()
invokes stm32_recvshort(), it checks for CTIMEOUT in SDMMCR_STA (which
is now cleared!) and not seeing such extracts a response (0x0 in this
case since CMD line never indicated a response) and return OK leading
mmcsd_cardidentify() to believe its an MMC card...
I've gone through the TRM for both stm32h7533 and stm32h7s3l8 and
outside of SDMMC DMA handling the two controller look close enough that
at least for initialization they should act the same.
Any thoughts on how to fix/workaround are most appreciated. Thanks in
advanced!
--
Peter Barada
[email protected]
