On Tue, 2016-09-27 at 11:05 +0000, C.H. Zhao wrote:

From: Scott Wood <oss@buserror.net>
Sent: Sunday, September 25, 2016 3:24 PM
To: C.H. Zhao
Cc: linuxppc-dev@lists.ozlabs.org; linux-kernel@vger.kernel.org; z.chenhui@g
mail.com; Jason Jin
Subject: Re: [v3,4/5] powerpc/pm: support deep sleep feature on T104x
    
On Tue, Aug 02, 2016 at 07:59:31PM +0800, Chenhui Zhao wrote:

quoted

T104x has deep sleep feature, which can switch off most parts of
the SoC when it is in deep sleep mode. This way, it becomes more
energy-efficient.

The DDR controller will also be powered off in deep sleep. Therefore,
the last stage (the latter part of fsl_dp_enter_low) will run without DDR
access. This piece of code and related TLBs are prefetched in advance.

Due to the different initialization code between 32-bit and 64-bit, they
have separate resume entry and precedure.

The feature supports 32-bit and 64-bit kernel mode.

Signed-off-by: Chenhui Zhao <redacted>
---
  arch/powerpc/include/asm/fsl_pm.h             |  24 ++
  arch/powerpc/kernel/asm-offsets.c             |  12 +
  arch/powerpc/kernel/fsl_booke_entry_mapping.S |  10 +
  arch/powerpc/kernel/head_64.S                 |   2 +-
  arch/powerpc/platforms/85xx/Makefile          |   1 +
  arch/powerpc/platforms/85xx/deepsleep.c       | 278 ++++++++++++++
  arch/powerpc/platforms/85xx/qoriq_pm.c        |  25 ++
  arch/powerpc/platforms/85xx/t104x_deepsleep.S | 531
++++++++++++++++++++++++++
  arch/powerpc/sysdev/fsl_rcpm.c                |   8 +-
  9 files changed, 889 insertions(+), 2 deletions(-)
  create mode 100644 arch/powerpc/platforms/85xx/deepsleep.c
  create mode 100644 arch/powerpc/platforms/85xx/t104x_deepsleep.S

diff --git a/arch/powerpc/include/asm/fsl_pm.h

b/arch/powerpc/include/asm/fsl_pm.h
index e05049b..48c2631 100644

--- a/arch/powerpc/include/asm/fsl_pm.h
+++ b/arch/powerpc/include/asm/fsl_pm.h

@@ -20,6 +20,7 @@

  
  #define PLAT_PM_SLEEP        20
  #define PLAT_PM_LPM20        30
+#define PLAT_PM_LPM35        40
  
  #define FSL_PM_SLEEP         (1 << 0)
  #define FSL_PM_DEEP_SLEEP    (1 << 1)

@@ -48,4 +49,27 @@ extern const struct fsl_pm_ops *qoriq_pm_ops;

  
  int __init fsl_rcpm_init(void);
  
+#ifdef CONFIG_FSL_QORIQ_PM
+int fsl_enter_deepsleep(void);
+int fsl_deepsleep_init(void);
+#else
+static inline int fsl_enter_deepsleep(void) { return -1; }
+static inline int fsl_deepsleep_init(void) { return -1; }
+#endif

Please return proper error codes.

Where can fsl_deepsleep_init() be called without CONFIG_FSL_QORIQ_PM?

[Chenhui] I can get rid of the ifdef here. And add it
in arch/powerpc/sysdev/fsl_rcpm.c.

No, this is the right place for the ifdef for functions that are called from
code that doesn't depend on CONFIG_FSL_QORIQ_PM.  But fsl_deepsleep_init() is
called from deepsleep.c which is only built with CONFIG_FSL_QORIQ_PM, and it's
hard to picture a scenario where it would be called from elsewhere.

quoted

diff --git a/arch/powerpc/kernel/fsl_booke_entry_mapping.S

b/arch/powerpc/kernel/fsl_booke_entry_mapping.S
index 83dd0f6..659b059 100644

--- a/arch/powerpc/kernel/fsl_booke_entry_mapping.S
+++ b/arch/powerpc/kernel/fsl_booke_entry_mapping.S

@@ -173,6 +173,10 @@ skpinv:  addi    r6,r6,1                         /*

Increment */
        lis     r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_64M, M_IF_NEEDED)@h
        ori     r6,r6,MAS2_VAL(PAGE_OFFSET, BOOK3E_PAGESZ_64M,
M_IF_NEEDED)@l
        mtspr   SPRN_MAS2,r6
+#ifdef ENTRY_DEEPSLEEP_SETUP
+     LOAD_REG_IMMEDIATE(r8, MEMORY_START)
+     ori     r8,r8,(MAS3_SX|MAS3_SW|MAS3_SR)
+#endif
        mtspr   SPRN_MAS3,r8
        tlbwe
  

Have you tried this with a relocatable kernel?

[Chenhui] Not yet. Not sure whether it has been supported on QorIQ platform.

It is supported, and deep sleep needs to work with it.

quoted

+static void fsl_dp_set_resume_pointer(void)
+{
+     u32 resume_addr;
+
+     /* the bootloader will finally jump to this address to return kernel
*/
+#ifdef CONFIG_PPC32
+     resume_addr = (u32)(__pa(fsl_booke_deep_sleep_resume));
+#else
+     resume_addr = (u32)(__pa(*(u64 *)fsl_booke_deep_sleep_resume)
+                         & 0xffffffff);
+#endif

Why are you masking the physical address by 0xffffffff?  Besides the
(u32) cast accomplishing the same thing, wouldn't it be a problem if
(e.g. due to a relocatable kernel) the address is above 4 GiB?

[Chenhui] Here, I assumed kernel is below 4 GiB. Maybe I should add a
comment here.

It needs a fix rather than a comment, unless you can show that the relocatable
mechanism doesn't support kernels over 4 GiB (I don't remember of the top of
my head whether it does).

-Scott