[U-Boot] [PATCH V2 2/3] mx7: psci: add basic psci support

[Albert ARIBAUD]

Hello Li,

(sorry for my dropping other receipients from the discussion; restoring
them

On Tue, 20 Oct 2015 15:04:40 +0000, Li Frank <Frank.Li at freescale.com>
wrote:
> 
> 
> > -----Original Message-----
> > From: Albert ARIBAUD [mailto:albert.u.boot at aribaud.net]
> > Sent: Tuesday, October 20, 2015 9:55 AM
> > To: Li Frank-B20596 <Frank.Li at freescale.com>
> > Subject: Re: [U-Boot] [PATCH V2 2/3] mx7: psci: add basic psci support
> > 
> > Hello Li,
> > 
> > On Tue, 20 Oct 2015 14:29:46 +0000, Li Frank <Frank.Li at freescale.com>
> > wrote:
> > >
> > >
> > > > -----Original Message-----
> > > > From: Albert ARIBAUD [mailto:albert.u.boot at aribaud.net]
> > > > Sent: Tuesday, October 20, 2015 9:25 AM
> > > > To: Li Frank-B20596 <Frank.Li at freescale.com>
> > > > Cc: Fan Peng-B51431 <Peng.Fan at freescale.com>; u-boot at lists.denx.de;
> > > > Estevam Fabio-R49496 <Fabio.Estevam at freescale.com>
> > > > Subject: Re: [U-Boot] [PATCH V2 2/3] mx7: psci: add basic psci
> > > > support
> > > >
> > > > Hello Li,
> > > >
> > > > On Tue, 20 Oct 2015 14:05:45 +0000, Li Frank
> > > > <Frank.Li at freescale.com>
> > > > wrote:
> > > > >
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Peng Fan [mailto:Peng.Fan at freescale.com]
> > > > > > Sent: Tuesday, October 20, 2015 1:00 AM
> > > > > > To: u-boot at lists.denx.de
> > > > > > Cc: Fan Peng-B51431 <Peng.Fan at freescale.com>; Li Frank-B20596
> > > > > > <Frank.Li at freescale.com>; Stefano Babic <sbabic at denx.de>;
> > > > > > Estevam
> > > > > > Fabio-
> > > > > > R49496 <Fabio.Estevam at freescale.com>
> > > > > > Subject: [PATCH V2 2/3] mx7: psci: add basic psci support
> > > > > >
> > > > > > 1. add basic psci support for imx7 chip.
> > > > > > 2. support cpu_on and cpu_off.
> > > > > > 3. switch to non-secure mode when boot linux kernel.
> > > > > > 4. set csu allow accessing all peripherial register in non-secure mode.
> > > > > >
> > > > > > Signed-off-by: Frank Li <Frank.Li at freescale.com>
> > > > > > Signed-off-by: Peng Fan <Peng.Fan at freescale.com>
> > > > > > Cc: Stefano Babic <sbabic at denx.de>
> > > > > > Cc: Fabio Estevam <fabio.estevam at freescale.com>
> > > > > > ---
> > > > > >
> > > > > > Changes V2:
> > > > > >  Refine commit msg.
> > > > > >
> > > > > >  arch/arm/cpu/armv7/mx7/Makefile   |  4 ++
> > > > > >  arch/arm/cpu/armv7/mx7/psci-mx7.c | 78
> > > > > > +++++++++++++++++++++++++++++++++++++++
> > > > > >  arch/arm/cpu/armv7/mx7/psci.S     | 54
> > +++++++++++++++++++++++++++
> > > > > >  arch/arm/cpu/armv7/mx7/soc.c      |  9 +++++
> > > > > >  4 files changed, 145 insertions(+)  create mode 100644
> > > > > > arch/arm/cpu/armv7/mx7/psci-mx7.c  create mode
> > > > > > 100644 arch/arm/cpu/armv7/mx7/psci.S
> > > > > >
> > > > > > diff --git a/arch/arm/cpu/armv7/mx7/Makefile
> > > > > > b/arch/arm/cpu/armv7/mx7/Makefile index e6ecef0..f25461c 100644
> > > > > > --- a/arch/arm/cpu/armv7/mx7/Makefile
> > > > > > +++ b/arch/arm/cpu/armv7/mx7/Makefile
> > > > > > @@ -6,3 +6,7 @@
> > > > > >  #
> > > > > >
> > > > > >  obj-y	:= soc.o clock.o clock_slice.o
> > > > > > +
> > > > > > +ifdef CONFIG_ARMV7_PSCI
> > > > > > +obj-y  += psci.o psci-mx7.o
> > > > >
> > > > > Obj-y += psci-mx7.o psci.o
> > > > > The otherwise psci_text_end will not be last one.
> > > >
> > > > I don't like this object module order sensitivity.
> > > >
> > > > The object module order of secure code modules should not affect the
> > > > resulting binary to the point of possibly preventing it from working
> > > > -- after all, the object module order of 'vanilla' image code does
> > > > not matter (1). We don't have this kind of problem when defining the
> > > > image start and end, why would we have it with the secure code start and
> > end?
> > >
> > > Secure code use psci_text_end to calculate stack top point.
> > 
> > Yes, and that does not require defining psci_text_end in an object module, nor
> > does it require specific object module ordering.
> 
> If psci_text_end in mid of secure text, 
> And psci_text_end + 1k < total secure text size,   stack will overwrite the code. 

This is a good reason to allocate the PSCCI stack and create a symbol
for its top in the linker script.

> > > > IOW, psci_text_end could (and should) be defined in the linker
> > > > script, not in an object module.
> > >
> > > I agree.  The other platform put psci_text_end to object module.
> > 
> > Which 'other platform' is this?
> 
> arch/arm/cpu/armv7/sunxi
> arch/arm/cpu/armv7/ls102xa

Ok, but they all use the same routine for moving the PSCI code and
setting its stack, right?

Amicalement,
-- 
Albert.