[PATCHv5 1/2] ARM: socfpga: support suspend to ram

[atull]

On Fri, 29 May 2015, Dinh Nguyen wrote:

> 
> 
> On 5/28/15 4:19 PM, Alan Tull wrote:
> > Add code that requests that the sdr controller go into
> > self-refresh mode.  This code is run from ocram.
> > 
> > Suspend-to-RAM and EDAC support are mutually exclusive on
> > SOCFPGA.  If the EDAC is enabled, it will prevent the
> > platform from going into suspend.
> > 
> > Example of how to request to suspend to ram:
> >  $ echo enabled > \
> > /sys/devices/soc/ffc02000.serial0/tty/ttyS0/power/wakeup
> > 
> >  $ echo -n mem > /sys/power/state
> > 
> > Signed-off-by: Alan Tull <atull at opensource.altera.com>
> > Cc: Pavel Machek <pavel at denx.de>
> > Cc: Arnd Bergmann <arnd at arndb.de>
> > Cc: Dinh Nguyen <dinguyen at opensource.altera.com>
> > Cc: Steffen Trumtrar <s.trumtrar at pengutronix.de>
> > ---
> > v2: use Generic on-chip SRAM driver to allocate ocram
> >     rm fncpy_align since generic allocator handles alignment
> >     check __arm_ioremap_exec return code
> >     check for NULL pointers
> >     add a comment regarding sdram controller configuration
> > v3: fix renamed #define
> >     propagate socfpga_setup_ocram_self_refresh error code
> > v4: Kconfig: don't need to select GENERIC_ALLOCATER
> >     add CONFIG_SOCFPGA_SUSPEND
> >     make s2r and EDAC support mutually exclusive
> >     socfpga.c: add sdr_ctl_base_addr
> >     return error if ocram not available in device tree
> >     update copyright years
> > v5: remove compile time dependency
> >     edac driver will prevent suspend
> >     don't configure scu standby mode; done in enable_scu().
> >     fix comments about required sdram controller configuration
> > ---
> >  arch/arm/mach-socfpga/Kconfig        |   10 ++-
> >  arch/arm/mach-socfpga/Makefile       |    1 +
> >  arch/arm/mach-socfpga/core.h         |    6 +-
> >  arch/arm/mach-socfpga/pm.c           |  149 ++++++++++++++++++++++++++++++++++
> >  arch/arm/mach-socfpga/self-refresh.S |  136 +++++++++++++++++++++++++++++++
> >  arch/arm/mach-socfpga/socfpga.c      |    6 +-
> >  drivers/edac/altera_edac.c           |   20 +++++
> >  7 files changed, 325 insertions(+), 3 deletions(-)
> >  create mode 100644 arch/arm/mach-socfpga/pm.c
> >  create mode 100644 arch/arm/mach-socfpga/self-refresh.S
> > 
> 
> Which branch are you basing this patch on? I got a trivial patch
> conflict in Kconfig when I applied to arm-soc/for-next and v4.1-rc5.
> 
> Also, I'm getting these sparse warnings:
> 
> arch/arm/mach-socfpga/pm.c:86:25: warning: incorrect type in argument 1
> (different address spaces)
> arch/arm/mach-socfpga/pm.c:86:25:    expected void *<noident>
> arch/arm/mach-socfpga/pm.c:86:25:    got void [noderef]
> <asn:2>*[assigned] suspend_ocram_base
> arch/arm/mach-socfpga/pm.c:108:52: warning: cast removes address space
> of expression
> 
> Dinh
> 

The problem is that fncpy wants parameter 1 to be void* and 
__arm_ioremap_exec wants 'void __ioram *'.

I get other warnings if I declare 'void *suspend_ocram_base' or cast it as
(void *) for fncpy().  It looks like we are stuck with this warning.

However sparse kindly pointed out that I needed to fix a declaration in
altera_edac.c:

-const struct dev_pm_ops altr_sdram_pm_ops = {
+static const struct dev_pm_ops altr_sdram_pm_ops = {

I'll fix that in the next rev.

Alan