On Mon, Oct 20, 2014 at 02:40:13AM -0400, Andrew Hills wrote:
> On 10/19/14, 11:15 AM, Gerke M. Preussner wrote:
> > I haven't had any time yet to look into this. Were you able to make any
> > progress on your end?
>
> I'm not sure I'd call it progress; I'm not familiar with the MMC
> protocol, so I'm spending most of my time learning rather than debugging.
>
> During the initialization, I see sys/dev/sdmmc roughly follow this order:
>
> sdmmc_card_attach()
> . sdmmc_enable()
> .. sdmmc_mem_enable()
> ... sdmmc_mem_send_op_cond() reads memory OCR
> ... sdmmc_mem_send_op_cond() attempts to send memory OCR
> .... sdmmc_mmc_command() doesn't fail, but doesn't succeed
> (MMC_OCR_MEM_READY is not set)
> .... command times out after 100 attempts (sdmmc_mem.c:519-538)
>
> What I haven't been able to figure out is why the command doesn't have
> the intended effect. Unfortunately, I enabled debug statements in
> sys/arch/armv7/omap/ommmc.c without thinking, so the system is unusable
> until I get the old bsd.umg back. Hopefully more to come soon.
>
I tracked the problem down, it seems u-boot only sets up CAPA for emmc.
With the diff below I can access the sd card on my bbb rev C (sd0) in
addition to the emmc (sd1). Both are still quite slow though, with
reads to the raw device around 780 KB/s on sd and 800 KB/s on emmc.
Besides DMA, it sounds like opting into a higher bus width
(4 bit on sd, 8 bit on emmc) and clock speeds might help there?
Then there is also the HSPE/High Speed Enable bit in HCTL.
scsibus0 at sdmmc0: 2 targets, initiator 0
sd0 at scsibus0 targ 1 lun 0: <SD/MMC, Drive #01, > SCSI2 0/direct fixed
sd0: 15271MB, 512 bytes/sector, 31275008 sectors
scsibus1 at sdmmc1: 2 targets, initiator 0
sd1 at scsibus1 targ 1 lun 0: <SD/MMC, Drive #01, > SCSI2 0/direct fixed
sd1: 3688MB, 512 bytes/sector, 7553024 sectors
Index: ommmc.c
===================================================================
RCS file: /cvs/src/sys/arch/armv7/omap/ommmc.c,v
retrieving revision 1.12
diff -u -p -r1.12 ommmc.c
--- ommmc.c 4 Dec 2013 12:31:14 -0000 1.12
+++ ommmc.c 2 Nov 2014 09:45:39 -0000
@@ -300,11 +300,8 @@ ommmc_attach(struct device *parent, stru
goto err;
}
-#if 0
- /* XXX - IIRC firmware should set this */
/* Controller Voltage Capabilities Initialization */
HSET4(sc, MMCHS_CAPA, MMCHS_CAPA_VS18 | MMCHS_CAPA_VS30);
-#endif
#ifdef SDHC_DEBUG
ommmc_dump_regs(sc);
