Hi all,
working on a Linux driver for this module,
on CMD 18 (and other), from the upper layer, i am getting a value of 0x1200 (4608) for block size (BLKSZE field).
As a bit-width, this value can be applied, but wondering if it is legal, and what it can cause, since datasheet shows a maximum of 4096.
Thanks
Regards,
Angelo
Angelo, nowhere near enough information. What are the "upper layers" and what is a "CMD 18"?
Isn't there already a driver for this chip in the linux sources?
Linux source code: drivers/mmc/host/sdhci-of-esdhc.c (v5.2.14) - Bootlin
There some strange code in there that looks to be related to this. Reading through it may give you hints as to how this should work on this chip:
if (spec_reg == SDHCI_BLOCK_SIZE) {
/*
* Two last DMA bits are reserved, and first one is used for
* non-standard blksz of 4096 bytes that we don't support
* yet. So clear the DMA boundary bits.
*/
ret &= (~SDHCI_MAKE_BLKSZ(0x7, 0));
}
(Edit) Apparently there is a "standard SDHCI register set" that isn't quite the same as these parts use. So as in the above code, it "intercepts" all register reads and writes and then changes the accesses where it has to.
More research indicates that "multi sector reads" (CMD18) should be setting the "BLOCK SIZE" to "512" and the block count to whatever multiple of that reads enough data. There's meant to be a "default block size" in the stack somewhere, so I can't guess where those large numbers are coming from.
Here's some previous pattern-matches...
SDHC - How to properly terminate a Multi-Sector Read
There's over 40 matches for "CMD18" in these forums.
Tom
Hi Tom,
many thanks,
i reply so late since there is always something that delay my coldfire activity
Yes the same seems to be actually done by
host->sdma_boundary = 0;
I checked sdhci-of-esdhc.c and also tried to have working sdhci-esdhc-imx.c, it seems both are for esdhc but there are some issues with them,
- sdhci-of-esdhc.c is for devicetree (of) and m68k is still stuck on platform board.c, of course i can add platform support in it, but not sure it's welcome, then there was some other reasons i don't remember right now,
- coldfire controller has some oddities and different bitfields from imx, and it is actually resultied easier, and less risky, to issue a new < 200 lines mcf driver. I am trying this way, but of course community is able to block it. Until now, they seems to agree my approach.
Hopefully in few days i should be able to come back on this.
Thanks
Angelo
Hi Angelo,
Does this issue always happen?
Have you tried with different board? If another board still with the same issue?
Thanks for more detailed info.
Have a great day,
Mike
-------------------------------------------------------------------------------
Note:
- If this post answers your question, please click the "Mark Correct" button. Thank you!
- We are following threads for 7 weeks after the last post, later replies are ignored
Please open a new thread and refer to the closed one, if you have a related question at a later point in time.
-------------------------------------------------------------------------------