| Message ID | 1425953334-42249-1-git-send-email-fenghua@phytium.com.cn |
|---|---|
| State | Changes Requested |
| Delegated to: | Tom Rini |
| Headers | show |
On Tue, Mar 10, 2015 at 3:08 AM, <fenghua@phytium.com.cn> wrote: > From: David Feng <fenghua@phytium.com.cn> > > CPU_RELEASE_ADDR should be defined when CONFIG_ARMV8_MULTIENTRY is used. > > Signed-off-by: David Feng <fenghua@phytium.com.cn> As asked earlier: how can I test this with FVP or the base model? I'm very interested in doing this as I guess it involves starting U-Boot at EL3 on bare metal and I really want to try this. > +/* SMP Spin Table Definitions */ > +#ifdef CONFIG_BASE_FVP > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) > +#else > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) > +#endif Where are these address defines coming from? Do these spin tables exist in a standard ARM FVP or base model? I get the impression that a secondary operating system is being booted on the secondary CPU at one of these addresses, but why is it running at these addresses specifically, and is that something coming with these simulators, or is it some image that is loaded on the side, that the community does not have access to? Yours, Linus Walleij
> -----Original Messages----- > From: "Linus Walleij" <linus.walleij@linaro.org> > Sent Time: 2015-03-10 18:08:03 (Tuesday) > To: "David Feng" <fenghua@phytium.com.cn> > Cc: "U-Boot Mailing List" <u-boot@lists.denx.de>, "Tom Rini" <trini@ti.com>, "Albert ARIBAUD" <albert.u.boot@aribaud.net> > Subject: Re: [PATCH] Vexpress64: Fix the compiling error when CONFIG_ARMV8_MULTIENTRY defined > > On Tue, Mar 10, 2015 at 3:08 AM, <fenghua@phytium.com.cn> wrote: > > > From: David Feng <fenghua@phytium.com.cn> > > > > CPU_RELEASE_ADDR should be defined when CONFIG_ARMV8_MULTIENTRY is used. > > > > Signed-off-by: David Feng <fenghua@phytium.com.cn> > > As asked earlier: how can I test this with FVP or the base model? > I usually use Foundation Model. > I'm very interested in doing this as I guess it involves starting U-Boot > at EL3 on bare metal and I really want to try this. > > > +/* SMP Spin Table Definitions */ > > +#ifdef CONFIG_BASE_FVP > > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) > > +#else > > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) > > +#endif > > Where are these address defines coming from? It's just hard coded and should be the same value with that in DTS. > > Do these spin tables exist in a standard ARM FVP or base model? > > I get the impression that a secondary operating system is being booted > on the secondary CPU at one of these addresses, but why is it running > at these addresses specifically, and is that something coming with these > simulators, or is it some image that is loaded on the side, that the > community does not have access to? > PSCI is not implemented in uboot-armv8. If booting u-boot on bare-metal only spin table can be used. All we do is describing booting method(spin table) and cpu release address in DTS. Linux kernel get cpu release address from DTS also. Yours, David.
> -----Original Message----- > From: U-Boot [mailto:u-boot-bounces@lists.denx.de] On Behalf Of FengHua > Sent: Wednesday, March 11, 2015 6:38 PM > To: Linus Walleij > Cc: U-Boot Mailing List > Subject: Re: [U-Boot] [PATCH] Vexpress64: Fix the compiling error when > CONFIG_ARMV8_MULTIENTRY defined > > > > > > -----Original Messages----- > > From: "Linus Walleij" <linus.walleij@linaro.org> Sent Time: 2015-03-10 > > 18:08:03 (Tuesday) > > To: "David Feng" <fenghua@phytium.com.cn> > > Cc: "U-Boot Mailing List" <u-boot@lists.denx.de>, "Tom Rini" > > <trini@ti.com>, "Albert ARIBAUD" <albert.u.boot@aribaud.net> > > Subject: Re: [PATCH] Vexpress64: Fix the compiling error when > > CONFIG_ARMV8_MULTIENTRY defined > > > > On Tue, Mar 10, 2015 at 3:08 AM, <fenghua@phytium.com.cn> wrote: > > > > > From: David Feng <fenghua@phytium.com.cn> > > > > > > CPU_RELEASE_ADDR should be defined when CONFIG_ARMV8_MULTIENTRY is > used. > > > > > > Signed-off-by: David Feng <fenghua@phytium.com.cn> > > > > As asked earlier: how can I test this with FVP or the base model? > > > I usually use Foundation Model. > > > I'm very interested in doing this as I guess it involves starting > > U-Boot at EL3 on bare metal and I really want to try this. > > > > > +/* SMP Spin Table Definitions */ > > > +#ifdef CONFIG_BASE_FVP > > > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + > 0x03f00000) > > > +#else > > > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + > 0x7fff0) > > > +#endif > > > > Where are these address defines coming from? > It's just hard coded and should be the same value with that in DTS. > > > > > Do these spin tables exist in a standard ARM FVP or base model? > > > > I get the impression that a secondary operating system is being booted > > on the secondary CPU at one of these addresses, but why is it running > > at these addresses specifically, and is that something coming with > > these simulators, or is it some image that is loaded on the side, that > > the community does not have access to? > > > PSCI is not implemented in uboot-armv8. If booting u-boot on bare-metal > only spin table can be used. All we do is describing booting method(spin > table) and cpu release address in DTS. Linux kernel get cpu release > address from DTS also. > Arnab's patches enable PSCI for amrv8 on u-boot: http://comments.gmane.org/gmane.comp.boot-loaders.u-boot/207882 We have tested the same on ARMv8 foundation model as well. Regards, Bhupesh
On Wed, Mar 11, 2015 at 2:08 PM, FengHua <fenghua@phytium.com.cn> wrote: (...) >> As asked earlier: how can I test this with FVP or the base model? >> > I usually use Foundation Model. OK... That is the same as the FVP I'm using I guess: $ Foundation_v8pkg/models/Linux64_GCC-4.1/Foundation_v8 --version ARM V8 Foundation Model r0p0 (model build 9.0.24) Copyright 2013 ARM Limited. All Rights Reserved. Correct? >> I'm very interested in doing this as I guess it involves starting U-Boot >> at EL3 on bare metal and I really want to try this. >> >> > +/* SMP Spin Table Definitions */ >> > +#ifdef CONFIG_BASE_FVP >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) >> > +#else >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) >> > +#endif >> >> Where are these address defines coming from? > > It's just hard coded and should be the same value with that in DTS. I look in the DTS from the Linux kernel: arch/arm64/boot/dts/arm/foundation-v8.dts: cpu@0 { device_type = "cpu"; compatible = "arm,armv8"; reg = <0x0 0x0>; enable-method = "spin-table"; cpu-release-addr = <0x0 0x8000fff8>; next-level-cache = <&L2_0>; }; cpu@1 { device_type = "cpu"; compatible = "arm,armv8"; reg = <0x0 0x1>; enable-method = "spin-table"; cpu-release-addr = <0x0 0x8000fff8>; next-level-cache = <&L2_0>; }; (...) It's not the same addres for what I can tell, CONFIG_SYS_SDRAM_BASE + 0x03f00000 = 0x83f00000 but the DTS cpu-release-addr is 0x8000fff8... Curiously we also have an ontology problem here: the DTS in the Linux kernel does use spin tables, but there is another set of DTS files in the ARM Trusted Firmware distribution, for the same simulator, stating PSCI as CPU release mechanism. These are the only ones that work properly when using ARM TF. (Well obviously you don't use these...) >> Do these spin tables exist in a standard ARM FVP or base model? >> >> I get the impression that a secondary operating system is being booted >> on the secondary CPU at one of these addresses, but why is it running >> at these addresses specifically, and is that something coming with these >> simulators, or is it some image that is loaded on the side, that the >> community does not have access to? >> > PSCI is not implemented in uboot-armv8. Nope. But it is implemented in ARM Trusted Firmware for ARMv8. ARM TF install the PSCI handlers before U-Boot is executed. > If booting u-boot on bare-metal > only spin table can be used. All we do is describing booting > method(spin table) and cpu release > address in DTS. Linux kernel get cpu release address from DTS also. Yep, I want to try this method... I just cannot even get U-Boot to run on the foundation model. I alter CONFIG_SYS_TEXT_BASE to 0x0: #define CONFIG_SYS_TEXT_BASE 0x00000000 Then I run the simulator like so: Foundation_v8pkg/models/Linux64_GCC-4.1/Foundation_v8 \ --cores=4 \ --no-secure-memory \ --visualization \ --gicv3 \ --data="u-boot-fvp-semi.bin"@0x00000000 Do you do this as well? Or how do you kick your simulator to run U-Boot on bare metal? I must be missing something. Yours, Linus Walleij
hi Linus, > -----Original Messages----- > From: "Linus Walleij" <linus.walleij@linaro.org> > Sent Time: 2015-03-20 17:39:48 (Friday) > To: FengHua <fenghua@phytium.com.cn> > Cc: "U-Boot Mailing List" <u-boot@lists.denx.de>, "albert.u.boot" <albert.u.boot@aribaud.net> > Subject: Re: Re: [PATCH] Vexpress64: Fix the compiling error when CONFIG_ARMV8_MULTIENTRY defined > > On Wed, Mar 11, 2015 at 2:08 PM, FengHua <fenghua@phytium.com.cn> wrote: > > (...) > >> As asked earlier: how can I test this with FVP or the base model? > >> > > I usually use Foundation Model. > > OK... That is the same as the FVP I'm using I guess: > > $ Foundation_v8pkg/models/Linux64_GCC-4.1/Foundation_v8 --version > ARM V8 Foundation Model r0p0 (model build 9.0.24) > Copyright 2013 ARM Limited. > All Rights Reserved. > > Correct? Yes > > >> I'm very interested in doing this as I guess it involves starting U-Boot > >> at EL3 on bare metal and I really want to try this. > >> > >> > +/* SMP Spin Table Definitions */ > >> > +#ifdef CONFIG_BASE_FVP > >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) > >> > +#else > >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) > >> > +#endif > >> > >> Where are these address defines coming from? > > > > It's just hard coded and should be the same value with that in DTS. > > I look in the DTS from the Linux kernel: > > arch/arm64/boot/dts/arm/foundation-v8.dts: > > cpu@0 { > device_type = "cpu"; > compatible = "arm,armv8"; > reg = <0x0 0x0>; > enable-method = "spin-table"; > cpu-release-addr = <0x0 0x8000fff8>; > next-level-cache = <&L2_0>; > }; > cpu@1 { > device_type = "cpu"; > compatible = "arm,armv8"; > reg = <0x0 0x1>; > enable-method = "spin-table"; > cpu-release-addr = <0x0 0x8000fff8>; > next-level-cache = <&L2_0>; > }; > (...) > > It's not the same addres for what I can tell, > > CONFIG_SYS_SDRAM_BASE + 0x03f00000 = 0x83f00000 > > but the DTS cpu-release-addr is 0x8000fff8... > > Curiously we also have an ontology problem here: the DTS in > the Linux kernel does use spin tables, but there is another set of > DTS files in the ARM Trusted Firmware distribution, for the same > simulator, stating PSCI as CPU release mechanism. These are > the only ones that work properly when using ARM TF. > > (Well obviously you don't use these...) PSCI is prefered by Linux. > > >> Do these spin tables exist in a standard ARM FVP or base model? > >> > >> I get the impression that a secondary operating system is being booted > >> on the secondary CPU at one of these addresses, but why is it running > >> at these addresses specifically, and is that something coming with these > >> simulators, or is it some image that is loaded on the side, that the > >> community does not have access to? > >> > > PSCI is not implemented in uboot-armv8. > > Nope. But it is implemented in ARM Trusted Firmware for ARMv8. > ARM TF install the PSCI handlers before U-Boot is executed. > > > If booting u-boot on bare-metal > > only spin table can be used. All we do is describing booting > > method(spin table) and cpu release > > address in DTS. Linux kernel get cpu release address from DTS also. > > Yep, I want to try this method... > > I just cannot even get U-Boot to run on the foundation model. > > I alter CONFIG_SYS_TEXT_BASE to 0x0: > > #define CONFIG_SYS_TEXT_BASE 0x00000000 > > Then I run the simulator like so: > > Foundation_v8pkg/models/Linux64_GCC-4.1/Foundation_v8 \ > --cores=4 \ > --no-secure-memory \ > --visualization \ > --gicv3 \ > --data="u-boot-fvp-semi.bin"@0x00000000 > > Do you do this as well? Or how do you kick your simulator to > run U-Boot on bare metal? > CONFIG_SYS_TEXT_BASE should be defined as 0x80000000. The reset PC value is 0x80000000 on Foundation Model. and I use "--image=u-boot.bin" instead of "--data=...". Yours, David.
> >> > +/* SMP Spin Table Definitions */ > >> > +#ifdef CONFIG_BASE_FVP > >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) > >> > +#else > >> > +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) > >> > +#endif > >> > >> Where are these address defines coming from? > > > > It's just hard coded and should be the same value with that in DTS. > > I look in the DTS from the Linux kernel: > > arch/arm64/boot/dts/arm/foundation-v8.dts: > > cpu@0 { > device_type = "cpu"; > compatible = "arm,armv8"; > reg = <0x0 0x0>; > enable-method = "spin-table"; > cpu-release-addr = <0x0 0x8000fff8>; > next-level-cache = <&L2_0>; > }; > cpu@1 { > device_type = "cpu"; > compatible = "arm,armv8"; > reg = <0x0 0x1>; > enable-method = "spin-table"; > cpu-release-addr = <0x0 0x8000fff8>; > next-level-cache = <&L2_0>; > }; > (...) > > It's not the same addres for what I can tell, > > CONFIG_SYS_SDRAM_BASE + 0x03f00000 = 0x83f00000 > > but the DTS cpu-release-addr is 0x8000fff8... > > Curiously we also have an ontology problem here: the DTS in > the Linux kernel does use spin tables, but there is another set of > DTS files in the ARM Trusted Firmware distribution, for the same > simulator, stating PSCI as CPU release mechanism. These are > the only ones that work properly when using ARM TF. FWIW in the bootwrapper we inject the relevant PSCI properties into the DTB if the bootwrapper is configured to use PSCI, and we should really do the same for spin-table. Given the enable-method is entirely dependent on the FW, it would be better for FW to fill in an appropriate value (where possible), leaving those out of the dts. Mark.
diff --git a/board/armltd/vexpress64/vexpress64.c b/board/armltd/vexpress64/vexpress64.c index de62864..bb6b5f6 100644 --- a/board/armltd/vexpress64/vexpress64.c +++ b/board/armltd/vexpress64/vexpress64.c @@ -22,6 +22,14 @@ int board_init(void) int dram_init(void) { +#ifdef CONFIG_ARMV8_MULTIENTRY + /* + * Clear spin table so that secondary processors + * observe the correct value after waken up from wfe. + */ + *(unsigned long *)CPU_RELEASE_ADDR = 0; +#endif + gd->ram_size = PHYS_SDRAM_1_SIZE; return 0; } diff --git a/include/configs/vexpress_aemv8a.h b/include/configs/vexpress_aemv8a.h index 810eef1..6fa651a 100644 --- a/include/configs/vexpress_aemv8a.h +++ b/include/configs/vexpress_aemv8a.h @@ -54,6 +54,15 @@ /* Flat Device Tree Definitions */ #define CONFIG_OF_LIBFDT +#ifdef CONFIG_ARMV8_MULTIENTRY +/* SMP Spin Table Definitions */ +#ifdef CONFIG_BASE_FVP +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x03f00000) +#else +#define CPU_RELEASE_ADDR (CONFIG_SYS_SDRAM_BASE + 0x7fff0) +#endif +#endif + /* CS register bases for the original memory map. */ #define V2M_PA_CS0 0x00000000 #define V2M_PA_CS1 0x14000000