Messages in this thread | | | From | Jolly Shah <> | Subject | RE: [PATCH v5 1/4] dt-bindings: firmware: Add bindings for ZynqMP firmware | Date | Tue, 13 Mar 2018 18:56:38 +0000 |
| |
> -----Original Message----- > From: Sudeep Holla [mailto:sudeep.holla@arm.com] > Sent: Tuesday, March 13, 2018 3:16 AM > To: Jolly Shah <JOLLYS@xilinx.com> > Cc: Sudeep Holla <sudeep.holla@arm.com>; gregkh@linuxfoundation.org; > matt@codeblueprint.co.uk; hkallweit1@gmail.com; michal.simek@xilinx.com; > robh+dt@kernel.org; mark.rutland@arm.com; ard.biesheuvel@linaro.org; > mingo@kernel.org; keescook@chromium.org; dmitry.torokhov@gmail.com; > Rajan Vaja <RAJANV@xilinx.com>; linux-arm-kernel@lists.infradead.org; linux- > kernel@vger.kernel.org; devicetree@vger.kernel.org > Subject: Re: [PATCH v5 1/4] dt-bindings: firmware: Add bindings for ZynqMP > firmware > > > > On 12/03/18 23:07, Jolly Shah wrote: > > Hi Sudeep, > > >>>> Do you foresee using SMC/HVC for this firmware even on future > platforms? > >>>> If not, I suggest to keep the protocol part separate from the transport i.e. > >>>> smc/hvc via ATF. It could be replaced with mailbox or some h/w > >>>> mechanism in future ? > >>>> > >>> > >>> We have PSCI and EEMI interfaces exposed to linux from ATF. PSCI is > >>> an EEMI client. We do not have current plans to switch to mailbox as > >>> it will require 2 communication channels to PMU as PSCI is through ATF. > >>> > >> > >> OK, but I just saw some bindings that has mailbox interface, honestly > >> it's getting too confusing with multiple series on the same thing > >> floating and hence I requested to put it together as one series. > > > > Mailbox binding is used for power management driver. Mailbox is only > > used for PMU->APU communication. APU->PMU communication is always > > through EEMI firmware interface which is using SMC/HVC. > > > > Ah OK, is it because there's no non-secure mailbox or to avoid races, all non- > secure EEMI is channeled through SMC ? >
2 reasons: 1> Avoid multiple EEMI communication channels as PSCI is through ATF. 2> We have some secure operations handled in ATF because of memory constraints on PMU
> -- > Regards, > Sudeep
| |