On 09/10/2018 04:16 PM, Matthias Blankertz wrote:
Hello,
we are experiencing issues with the R-Car Gen3 SD driver with H3ULCB development kits, in both u-boot 2018.05 and 2018.07. The issues are frequent failures of u-boot to access the eMMC storage. I have traced the issue to u-boot commit 9ac9c341f which switched the default configs to use the UniPhier/Matsushita common driver instead of the previous custom Renesas driver.
With the Matsushita driver we have various H3ULCB starter kits with failures related to the eMMC storage on the ulcb. Either u-boot cannot access it at all (resulting in a failure to load the u-boot environment), or reads corrupted data from the eMMC storage when reading the boot images. These failures occur frequently, but are not reproducible on every boot. Also, the frequency of failure depends on the specific starter kit, with some having no failures at all and others failing almost every boot.
I have tried building u-boot both with the default config, which enables MMC_UHS_SUPPORT and MMC_HS200_SUPPORT, and with UHS and HS200 support disabled. There is no change in the failures.
When I build u-boot with RENESAS_SDHI disabled and instead use the old custom Renesas driver (CONFIG_SH_SDHI) the eMMC failures disappear completely for every starter kit: Both environment loading and reading boot images from eMMC are reliable with the old driver.
I have a vague suspicion that there is some sort of timing issue with the UniPhier/Matsushita common driver and the R-Car Gen3 or the eMMCs. When I add a delay to the u-boot startup (inserting udelay(5) before the mmc device is initialised in env/mmc.c) the failures in u-boot disappear, but then the SD controller (or eMMC card?) is sometimes left in a state where the linux kernel cannot use it: "Timeout waiting for hardware interrupt after CMD...".
This effect, that the controller/card is left in a state that the linux kernel cannot deal with, also does not occur when the old Renesas driver is used in u-boot.
Any ideas?
Try 2018.09 and make sure to use latest ATF, seems to work fine there:
[ 0.000162] NOTICE: BL2: R-Car Gen3 Initial Program Loader(CA57) Rev.1.0.21 [ 0.005724] NOTICE: BL2: PRR is R-Car H3 Ver.2.0 [ 0.010393] NOTICE: BL2: Board is Starter Kit Rev.1.0 [ 0.015506] NOTICE: BL2: Boot device is HyperFlash(80MHz) [ 0.020932] NOTICE: BL2: LCM state is CM [ 0.024974] NOTICE: BL2: AVS setting succeeded. DVFS_SetVID=0x53 [ 0.030962] NOTICE: BL2: DDR3200(rev.0.33)NOTICE: [COLD_BOOT]NOTICE: ..0 [ 0.065954] NOTICE: BL2: DRAM Split is 4ch [ 0.069837] NOTICE: BL2: QoS is default setting(rev.0.19) [ 0.075280] NOTICE: BL2: DRAM refresh interval 1.95 usec [ 0.080701] NOTICE: BL2: v1.4(release):15dba6bb-dirty [ 0.085734] NOTICE: BL2: Built : 15:20:54, Sep 9 2018 [ 0.090921] NOTICE: BL2: Normal boot [ 0.094567] NOTICE: BL2: dst=0xe6322d00 src=0x8180000 len=512(0x200) [ 0.100950] NOTICE: BL2: dst=0x43f00000 src=0x8180400 len=6144(0x1800) [ 0.107570] NOTICE: BL2: dst=0x44000000 src=0x81c0000 len=65536(0x10000) [ 0.114801] NOTICE: BL2: dst=0x44100000 src=0x8200000 len=1048576(0x100000) [ 0.129624] NOTICE: BL2: dst=0x50000000 src=0x8640000 len=1048576(0x100000)
U-Boot 2018.09-rc3 (Sep 09 2018 - 15:21:06 +0000)
CPU: Renesas Electronics R8A7795 rev 2.0 Model: Renesas H3ULCB board based on r8a7795 ES2.0+ DRAM: 3.9 GiB Flash: 64 MiB MMC: sd@ee100000: 0, sd@ee140000: 1 Loading Environment from MMC... OK In: serial@e6e88000 Out: serial@e6e88000 Err: serial@e6e88000 Net: eth0: ethernet@e6800000 Auto-update from TFTP: failed, env. variable 'updatefile' not found Hit any key to stop autoboot: 0 => => => mmc dev 1 switch to partitions #0, OK mmc1(part 0) is current device => mmc info Device: sd@ee140000 Manufacturer ID: 15 OEM: 100 Name: 8GSD4 Bus Speed: 199999992 Mode : HS200 (200MHz) Rd Block Len: 512 MMC version 5.1 High Capacity: Yes Capacity: 7.3 GiB Bus Width: 8-bit Erase Group Size: 512 KiB HC WP Group Size: 8 MiB User Capacity: 7.3 GiB WRREL Boot Capacity: 31.9 MiB ENH RPMB Capacity: 512 KiB ENH