Re: [Minios-devel] [UNIKRAFT PATCHv4 35/43] plat/common: Implement CPU suspend for arm64

[Wei Chen <Wei.Chen@xxxxxxx>]

Hi Julien,

> -----Original Message-----
> From: Julien Grall <julien.grall@xxxxxxx>
> Sent: 2018年7月18日 21:28
> To: Wei Chen <Wei.Chen@xxxxxxx>; minios-devel@xxxxxxxxxxxxxxxxxxxx;
> simon.kuenzer@xxxxxxxxx
> Cc: Kaly Xin <Kaly.Xin@xxxxxxx>; nd <nd@xxxxxxx>
> Subject: Re: [Minios-devel] [UNIKRAFT PATCHv4 35/43] plat/common: Implement
> CPU suspend for arm64
> 
> 
> 
> On 18/07/18 10:11, Wei Chen wrote:
> > Hi Julien,
> 
> Hi Wei,
> 
> >
> >> -----Original Message-----
> >> From: Julien Grall <julien.grall@xxxxxxx>
> >> Sent: 2018年7月16日 21:48
> >> To: Wei Chen <Wei.Chen@xxxxxxx>; minios-devel@xxxxxxxxxxxxxxxxxxxx;
> >> simon.kuenzer@xxxxxxxxx
> >> Cc: Kaly Xin <Kaly.Xin@xxxxxxx>; nd <nd@xxxxxxx>
> >> Subject: Re: [Minios-devel] [UNIKRAFT PATCHv4 35/43] plat/common: Implement
> >> CPU suspend for arm64
> >>
> >> Hi Wei,
> >>
> >> On 06/07/18 10:03, Wei Chen wrote:
> >>> Using PSCI despatch function to call PSCI_CPU_SUSPEND
> >>> function to halt CPU.
> >>>
> >>> Signed-off-by: Wei Chen <Wei.Chen@xxxxxxx>
> >>> ---
> >>>    plat/common/arm/cpu_native.c        | 44 +++++++++++++++++++++++++++++
> >>>    plat/common/include/arm/arm64/cpu.h |  2 ++
> >>>    2 files changed, 46 insertions(+)
> >>>    create mode 100644 plat/common/arm/cpu_native.c
> >>>
> >>> diff --git a/plat/common/arm/cpu_native.c b/plat/common/arm/cpu_native.c
> >>> new file mode 100644
> >>> index 0000000..07548bf
> >>> --- /dev/null
> >>> +++ b/plat/common/arm/cpu_native.c
> >>> @@ -0,0 +1,44 @@
> >>> +/* SPDX-License-Identifier: BSD-3-Clause */
> >>> +/*
> >>> + * Authors: Wei Chen <wei.chen@xxxxxxx>
> >>> + *
> >>> + * Copyright (c) 2018, Arm Ltd., All rights reserved.
> >>> + *
> >>> + * Redistribution and use in source and binary forms, with or without
> >>> + * modification, are permitted provided that the following conditions
> >>> + * are met:
> >>> + *
> >>> + * 1. Redistributions of source code must retain the above copyright
> >>> + *    notice, this list of conditions and the following disclaimer.
> >>> + * 2. Redistributions in binary form must reproduce the above copyright
> >>> + *    notice, this list of conditions and the following disclaimer in the
> >>> + *    documentation and/or other materials provided with the 
> >>> distribution.
> >>> + * 3. Neither the name of the copyright holder nor the names of its
> >>> + *    contributors may be used to endorse or promote products derived
> from
> >>> + *    this software without specific prior written permission.
> >>> + *
> >>> + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> "AS
> >> IS"
> >>> + * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
> >> THE
> >>> + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
> >> PURPOSE
> >>> + * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
> >> BE
> >>> + * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
> >>> + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
> >>> + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
> BUSINESS
> >>> + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
> IN
> >>> + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR 
> >>> OTHERWISE)
> >>> + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
> >> THE
> >>> + * POSSIBILITY OF SUCH DAMAGE.
> >>> + *
> >>> + * THIS HEADER MAY NOT BE EXTRACTED OR MODIFIED IN ANY WAY.
> >>> + */
> >>> +
> >>> +#include <arm/cpu.h>
> >>> +#include <arm/cpu_defs.h>
> >>> +
> >>> +void halt(void)
> >>
> >> Looking at the usage of halt, I think PSCI CPU suspend is too expensive
> >> for brief idling.
> >>
> >> I am not entirely sure of the semantic expected, but it looks like "wfi"
> >> would be more suitable here.
> >
> > Does timer interrupt can wake up wfi? If yes, I think I can't use wfi here.
> 
> Yes interrupt will wake-up the CPU. This is the same with PSCI suspend.
> 
> The main difference with the two is the latter will allow you to go in
> deeper state (depending on the arguments you passed).
> 
> You don't need that at the moment.
> 

I checked the code again. Before call halt(), we will call local_disable_irq().
So I think I will use wfi here.

> >>
> >>> +{
> >>> + if (psci_method == PSCI_METHOD_HVC)
> >>> +         psci_hvc_call(PSCI_FNID_CPU_SUSPEND);
> >>> + else if (psci_method == PSCI_METHOD_SMC)
> >>> +         psci_smc_call(PSCI_FNID_CPU_SUSPEND);
> >>
> >> Regardless my comment above, I don't think this code follow the PSCI
> >> spec. CPU_SUSPEND takes another parameter that is left unknown. Also,
> >>
> >
> > Yes, I have ignored other parameters. I will use another patch series to
> > improve the PSCI library. Let's record it in TODO list.
> 
> I don't think this should be a TODO. That function takes 3 arguments,
> the first one will be used to decide how deep the processor will sleep.
> 
> If you don't specify it, you will get whatever value will be in w1 when
> calling CPU_SUSPEND. This could be anything from standby to powerdown
> the CPU.
> 
> So you should at least pass a default values to all the parameters of
> PSCI calls.
> 

OK

> 
> >
> >> Furthermore, you may want to introduce a function pointer and detect at
> >> initialization which call you want to use rather than checking in every
> >> function.
> >>
> >
> > That's sensible. I will use such a function pointer.
> >
> >> Lastly, in case of non-PSCI platform function will just do nothing. This
> >> looks slightly strange.
> >>
> >
> > Yes, because in the code context, while cpu_suspend/cpu_halt could not take
> > effect, the code will enter dead loop.
> 
> There are multiple place where halt can be called. Not all of them will
> end up with a dead loop. I would at least at a comment explaining
> nothing is done with non-PSCI platform.

Ok

> 
> Cheers,
> 
> --
> Julien Grall
_______________________________________________
Minios-devel mailing list
Minios-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/minios-devel