On 02/05/2018 14:41, Jagan Teki wrote:
Hi Stefano,
On Thu, Apr 26, 2018 at 1:16 PM, Stefano Babic sbabic@denx.de wrote:
On 26/04/2018 09:33, Jagan Teki wrote:
On Thu, Apr 26, 2018 at 12:47 PM, Stefano Babic sbabic@denx.de wrote:
Hi Jagan,
[snip]
+static int mx6dl_dcd_table[] = {
0x020e0774, 0x000C0000, /* MX6_IOM_GRP_DDR_TYPE */0x020e0754, 0x00000000, /* MX6_IOM_GRP_DDRPKE */0x020e04ac, 0x00000028, /* MX6_IOM_DRAM_SDCLK_0 */0x020e04b0, 0x00000028, /* MX6_IOM_DRAM_SDCLK_1 */0x020e0464, 0x00000028, /* MX6_IOM_DRAM_CAS */0x020e0490, 0x00000028, /* MX6_IOM_DRAM_RAS */0x020e074c, 0x00000028, /* MX6_IOM_GRP_ADDDS */0x020e0494, 0x00000028, /* MX6_IOM_DRAM_RESET */0x020e04a0, 0x00000000, /* MX6_IOM_DRAM_SDBA2 */0x020e04b4, 0x00000028, /* MX6_IOM_DRAM_SDODT0 */0x020e04b8, 0x00000028, /* MX6_IOM_DRAM_SDODT1 */0x020e076c, 0x00000028, /* MX6_IOM_GRP_CTLDS */0x020e0750, 0x00020000, /* MX6_IOM_GRP_DDRMODE_CTL */0x020e04bc, 0x00000028, /* MX6_IOM_DRAM_SDQS0 */0x020e04c0, 0x00000028, /* MX6_IOM_DRAM_SDQS1 */0x020e04c4, 0x00000028, /* MX6_IOM_DRAM_SDQS2 */0x020e04c8, 0x00000028, /* MX6_IOM_DRAM_SDQS3 */0x020e0760, 0x00020000, /* MX6_IOM_GRP_DDRMODE */0x020e0764, 0x00000028, /* MX6_IOM_GRP_B0DS */0x020e0770, 0x00000028, /* MX6_IOM_GRP_B1DS */0x020e0778, 0x00000028, /* MX6_IOM_GRP_B2DS */0x020e077c, 0x00000028, /* MX6_IOM_GRP_B3DS */0x020e0470, 0x00000028, /* MX6_IOM_DRAM_DQM0 */0x020e0474, 0x00000028, /* MX6_IOM_DRAM_DQM1 */0x020e0478, 0x00000028, /* MX6_IOM_DRAM_DQM2 */0x020e047c, 0x00000028, /* MX6_IOM_DRAM_DQM3 */0x021b001c, 0x00008000, /* MMDC0_MDSCR */0x021b0800, 0xA1390003, /* DDR_PHY_P0_MPZQHWCTRL */0x021b080c, 0x0042004b, /* MMDC1_MPWLDECTRL0 */0x021b0810, 0x0038003c, /* MMDC1_MPWLDECTRL1 */0x021b083c, 0x42340230, /* MPDGCTRL0 PHY0 */0x021b0840, 0x0228022c, /* MPDGCTRL1 PHY0 */0x021b0848, 0x42444646, /* MPRDDLCTL PHY0 */0x021b0850, 0x38382e2e, /* MPWRDLCTL PHY0 */0x021b081c, 0x33333333, /* DDR_PHY_P0_MPREDQBY0DL3 */0x021b0820, 0x33333333, /* DDR_PHY_P0_MPREDQBY1DL3 */0x021b0824, 0x33333333, /* DDR_PHY_P0_MPREDQBY2DL3 */0x021b0828, 0x33333333, /* DDR_PHY_P0_MPREDQBY3DL3 */0x021b08b8, 0x00000800, /* DDR_PHY_P0_MPMUR0 */0x021b0004, 0x0002002D, /* MMDC0_MDPDC */0x021b0008, 0x00333040, /* MMDC0_MDOTC */0x021b000c, 0x3F4352F3, /* MMDC0_MDCFG0 */0x021b0010, 0xB66D8B63, /* MMDC0_MDCFG1 */0x021b0014, 0x01FF00DB, /* MMDC0_MDCFG2 */0x021b0018, 0x00011740, /* MMDC0_MDMISC */0x021b001c, 0x00008000, /* MMDC0_MDSCR */0x021b002c, 0x000026D2, /* MMDC0_MDRWD */0x021b0030, 0x00431023, /* MMDC0_MDOR */0x021b0040, 0x00000017, /* Chan0 CS0_END */0x021b0000, 0x83190000, /* MMDC0_MDCTL */0x021b001c, 0x02008032, /* MMDC0_MDSCR MR2 write, CS0 */0x021b001c, 0x00008033, /* MMDC0_MDSCR, MR3 write, CS0 */0x021b001c, 0x00048031, /* MMDC0_MDSCR, MR1 write, CS0 */0x021b001c, 0x15208030, /* MMDC0_MDSCR, MR0write, CS0 */0x021b001c, 0x04008040, /* MMDC0_MDSCR */0x021b0020, 0x00007800, /* MMDC0_MDREF */0x021b0818, 0x00022227, /* DDR_PHY_P0_MPODTCTRL */0x021b0004, 0x0002556D, /* MMDC0_MDPDC */0x021b0404, 0x00011006, /* MMDC0_MAPSR ADOPT */0x021b001c, 0x00000000, /* MMDC0_MDSCR */+};
Sorry to have not raised this before. I saw this the first time, but I did not understand. I tried to find a reason for it, but still I had no answer.
This is a DL, and DDR support is full integrated in U-Boot, including dynamic calibration if desired. What is the reason to dump the DCD table into code and push it with ddr_init, instead of setting structures for chip, gpr and calibration as most of boards are doing ? If you dump the table, you could this in the .cfg file where the DCD table belongs to, of course adding entries for DDR initialisation. I do not see after long thoughts any reason to go in this way.
Like initializing ddr config and calibration using mx6sdl_dram_iocfg and mx6_dram_cfg? yes I usually does the same instead of hot codding hex values.
Yes, this is what I mean.
ddr calibration here has reinitialized few of same registers multiple times with a sequence, using mx6_dram_cfg I can't achieve the same.
Do you mean dynamic calibration CONFIG_MX6_DDRCAL ? If this is the case, you can still use the static values passing the parameters you got from ddr_stress.
I need to dig further on to this for proper working register values meanwhile can it be possible to apply this sequence as of now?
I do not think this is a good idea - I really prefer to get in the final solution.
Thanks, Stefano