[U-Boot] [PATCH 1/5 v2] OMAP3: Fix SDRC init

[Steve Sakoman]

On Mon, Oct 19, 2009 at 7:55 AM, Dirk Behme <dirk.behme at googlemail.com> wrote:
> Steve Sakoman wrote:
>>
>> On Tue, Oct 6, 2009 at 7:17 PM, Nishanth Menon <nm at ti.com> wrote:
>>>
>>> Defaults are for Infineon DDR timings.
>>> Since none of the supported boards currently do
>>> XIP boot, these seem to be faulty. fix the values
>>> as per the calculations(ACTIMA,B), conf
>>> the sdrc power with pwdnen and wakeupproc bits
>>>
>>> Signed-off-by: Nishanth Menon <nm at ti.com>
>>> Cc: David B <david-b at pacbell.net>
>>> Cc: Vikram Pandita <vikram.pandita at ti.com>
>>> Cc: Richard Woodruff <r-woodruff2 at ti.com>
>>> Cc: Sandeep Paulraj <s-paulraj at ti.com>
>>> Cc: Tom Rix <tom.rix at windriver.com>
>>> Cc: Dirk Behme <dirk.behme at googlemail.com>
>>> ---
>>>  cpu/arm_cortexa8/omap3/mem.c     |    3 ++-
>>>  include/asm-arm/arch-omap3/cpu.h |    1 +
>>>  include/asm-arm/arch-omap3/mem.h |    8 ++++----
>>>  3 files changed, 7 insertions(+), 5 deletions(-)
>>>
>>> diff --git a/cpu/arm_cortexa8/omap3/mem.c b/cpu/arm_cortexa8/omap3/mem.c
>>> index 079c848..8731c9d 100644
>>> --- a/cpu/arm_cortexa8/omap3/mem.c
>>> +++ b/cpu/arm_cortexa8/omap3/mem.c
>>> @@ -164,7 +164,8 @@ void do_sdrc_init(u32 cs, u32 early)
>>>               writel(SDP_SDRC_SHARING, &sdrc_base->sharing);
>>>
>>>               /* Disable Power Down of CKE cuz of 1 CKE on combo part */
>>> -               writel(SRFRONRESET | PAGEPOLICY_HIGH, &sdrc_base->power);
>>> +               writel(WAKEUPPROC | PWDNEN | SRFRONRESET |
>>> PAGEPOLICY_HIGH,
>>> +                               &sdrc_base->power);
>>>
>>>               writel(ENADLL | DLLPHASE_90, &sdrc_base->dlla_ctrl);
>>>               sdelay(0x20000);
>>> diff --git a/include/asm-arm/arch-omap3/cpu.h
>>> b/include/asm-arm/arch-omap3/cpu.h
>>> index 8ab2e39..e51c4f3 100644
>>> --- a/include/asm-arm/arch-omap3/cpu.h
>>> +++ b/include/asm-arm/arch-omap3/cpu.h
>>> @@ -222,6 +222,7 @@ struct sdrc {
>>>
>>>  #define PAGEPOLICY_HIGH                (0x1 << 0)
>>>  #define SRFRONRESET            (0x1 << 7)
>>> +#define PWDNEN                 (0x1 << 2)
>>>  #define WAKEUPPROC             (0x1 << 26)
>>>
>>>  #define DDR_SDRAM              (0x1 << 0)
>>> diff --git a/include/asm-arm/arch-omap3/mem.h
>>> b/include/asm-arm/arch-omap3/mem.h
>>> index 5b9ac75..31cbdef 100644
>>> --- a/include/asm-arm/arch-omap3/mem.h
>>> +++ b/include/asm-arm/arch-omap3/mem.h
>>> @@ -78,16 +78,16 @@ enum {
>>>  #define TRP_165                3
>>>  #define TRAS_165       7
>>>  #define TRC_165                10
>>> -#define TRFC_165       21
>>> +#define TRFC_165       12
>>>  #define V_ACTIMA_165   ((TRFC_165 << 27) | (TRC_165 << 22) | \
>>>                       (TRAS_165 << 18) | (TRP_165 << 15) |  \
>>>                       (TRCD_165 << 12) | (TRRD_165 << 9) |  \
>>>                       (TDPL_165 << 6) | (TDAL_165))
>>>
>>>  #define TWTR_165       1
>>> -#define TCKE_165       1
>>> -#define TXP_165                5
>>> -#define XSR_165                23
>>> +#define TCKE_165       2
>>> +#define TXP_165                2
>>> +#define XSR_165                20
>>>  #define V_ACTIMB_165   (((TCKE_165 << 12) | (XSR_165 << 0)) |  \
>>>                       (TXP_165 << 8) | (TWTR_165 << 16))
>>>
>>> --
>>> 1.6.0.4
>>
>> I see issues after applying this patch (Overo/Beagle).
>>
>> In about half of my boot attempts I get a hang after Uncompressing
>> Linux....
>>
>> In the other half I get many many errors of this type:
>>
>> SLAB: cache with size 192 has lost its name
>>
>> Reverting the patch restores normal operation.
>
> What's about removing it from recent ARM pull request than and do some
> further testing?

That sounds like a good plan to me.

Steve