[PATCH 0/2] remoteproc: ti_k3_arm64: Program GTC FID0 register

Hi,
ARMv8's generic timer[1] picks up it's graycode from GTC. However, the frequency of the GTC is supposed to be programmed in CNTFID0[2] register prior to enabling the GTC in CNTCR[3] register. U-boot for K3 is currently missing that programming sequence. This series introduces the clock nodes necessary for the lookup and writes the frequency to the corresponding register for TF-A to pick up [4].
Based on: 62b07b5173e3 (tag: v2021.01-rc5, upstream/master) Prepare v2021.01-rc5
J721e: https://pastebin.ubuntu.com/p/Wcxfd3BNSv/ J7200: https://pastebin.ubuntu.com/p/VsVqn7y8pv/ I dont have an AM654-evm handy at the moment, but the programming sequences are similar in nature.
[1] https://developer.arm.com/documentation/100095/0002/generic-timer/generic-ti... [2] https://developer.arm.com/docs/ddi0595/h/external-system-registers/cntfid0 [3] https://developer.arm.com/docs/ddi0595/h/external-system-registers/cntcr [4] https://github.com/ARM-software/arm-trusted-firmware/commit/6a22d9ea3c7fa28d...
Nishanth Menon (2): arm: dts: k3-*-r5-*-board: Add GTC clock remoteproc: ti_k3_arm64: Program CNTFID0 register in GTC
arch/arm/dts/k3-am654-r5-base-board.dts | 1 + arch/arm/dts/k3-j7200-r5-common-proc-board.dts | 1 + arch/arm/dts/k3-j721e-r5-common-proc-board.dts | 1 + drivers/remoteproc/ti_k3_arm64_rproc.c | 15 +++++++++++++++ 4 files changed, 18 insertions(+)

Add GTC Clock definition as index 0 clock so that we can use the clock node in the driver later on.
Signed-off-by: Nishanth Menon nm@ti.com --- arch/arm/dts/k3-am654-r5-base-board.dts | 1 + arch/arm/dts/k3-j7200-r5-common-proc-board.dts | 1 + arch/arm/dts/k3-j721e-r5-common-proc-board.dts | 1 + 3 files changed, 3 insertions(+)
diff --git a/arch/arm/dts/k3-am654-r5-base-board.dts b/arch/arm/dts/k3-am654-r5-base-board.dts index d43a4edc7189..f12510259da4 100644 --- a/arch/arm/dts/k3-am654-r5-base-board.dts +++ b/arch/arm/dts/k3-am654-r5-base-board.dts @@ -36,6 +36,7 @@ power-domains = <&k3_pds 61 TI_SCI_PD_EXCLUSIVE>, <&k3_pds 202 TI_SCI_PD_EXCLUSIVE>; resets = <&k3_reset 202 0>; + clocks = <&k3_clks 61 0>; assigned-clocks = <&k3_clks 202 0>; assigned-clock-rates = <800000000>; ti,sci = <&dmsc>; diff --git a/arch/arm/dts/k3-j7200-r5-common-proc-board.dts b/arch/arm/dts/k3-j7200-r5-common-proc-board.dts index db63d93777e0..2ac887b180c9 100644 --- a/arch/arm/dts/k3-j7200-r5-common-proc-board.dts +++ b/arch/arm/dts/k3-j7200-r5-common-proc-board.dts @@ -32,6 +32,7 @@ power-domains = <&k3_pds 61 TI_SCI_PD_EXCLUSIVE>, <&k3_pds 202 TI_SCI_PD_EXCLUSIVE>; resets = <&k3_reset 202 0>; + clocks = <&k3_clks 61 1>; assigned-clocks = <&k3_clks 202 2>, <&k3_clks 61 1>; assigned-clock-rates = <2000000000>, <200000000>; ti,sci = <&dmsc>; diff --git a/arch/arm/dts/k3-j721e-r5-common-proc-board.dts b/arch/arm/dts/k3-j721e-r5-common-proc-board.dts index 2dde65d9681d..1f621676785e 100644 --- a/arch/arm/dts/k3-j721e-r5-common-proc-board.dts +++ b/arch/arm/dts/k3-j721e-r5-common-proc-board.dts @@ -28,6 +28,7 @@ power-domains = <&k3_pds 61 TI_SCI_PD_EXCLUSIVE>, <&k3_pds 202 TI_SCI_PD_EXCLUSIVE>; resets = <&k3_reset 202 0>; + clocks = <&k3_clks 61 1>; assigned-clocks = <&k3_clks 202 2>, <&k3_clks 61 1>; assigned-clock-rates = <2000000000>, <200000000>; ti,sci = <&dmsc>;

ARMv8's generic timer[1] picks up it's graycode from GTC. However, the frequency of the GTC is supposed to be programmed in CNTFID0[2] register prior to enabling the GTC in CNTCR[3] register.
In K3 architecture, GTC provides a central time to many parts of the SoC including graycode to the generic timer in the ARMv8 subsystem. However, due to the central nature and the need to enable the counter early in the boot process, the R5 based u-boot enables GTC and programs it's frequency based on central needs of the system. This may not be a constant 200MHz based on the system. The bootloader is supposed to program the FID0 register with the correct frequency it has sourced for GTC from the central system controller OR from PLLs as appropriate, and TF-A is supposed[4] to use that as the frequency for it's local timer.
Currently we are programming just the CNTCR[3] register to enable the GTC, however we dont let TF-A know the frequency that GTC is actually running at. A mismatch in programmed frequency and what we program for generic timer will, as we can imagine, all kind of weird mayhem.
So, program the CNTFID0 register with the clock frequency. Note: assigned-clock-rates should have set the clock frequency, so the only operation we need to explicitly do is to retrieve the frequency and program it in FID0 register.
Since the valid in K3 for GTC clock frequencies are < U32_MAX, we can just cast the ulong and continue.
[1] https://developer.arm.com/documentation/100095/0002/generic-timer/generic-ti... [2] https://developer.arm.com/docs/ddi0595/h/external-system-registers/cntfid0 [3] https://developer.arm.com/docs/ddi0595/h/external-system-registers/cntcr [4] https://github.com/ARM-software/arm-trusted-firmware/commit/6a22d9ea3c7fa28d...
Signed-off-by: Nishanth Menon nm@ti.com --- drivers/remoteproc/ti_k3_arm64_rproc.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+)
diff --git a/drivers/remoteproc/ti_k3_arm64_rproc.c b/drivers/remoteproc/ti_k3_arm64_rproc.c index 28c6ddb69191..395b74192b06 100644 --- a/drivers/remoteproc/ti_k3_arm64_rproc.c +++ b/drivers/remoteproc/ti_k3_arm64_rproc.c @@ -23,6 +23,7 @@ #define INVALID_ID 0xffff
#define GTC_CNTCR_REG 0x0 +#define GTC_CNTFID0_REG 0x20 #define GTC_CNTR_EN 0x3
/** @@ -31,6 +32,7 @@ * @rproc_rst: rproc reset control data * @sci: Pointer to TISCI handle * @tsp: TISCI processor control helper structure + * @gtc_clk: GTC clock description * @gtc_base: Timer base address. */ struct k3_arm64_privdata { @@ -38,6 +40,7 @@ struct k3_arm64_privdata { struct power_domain gtc_pwrdmn; struct reset_ctl rproc_rst; struct ti_sci_proc tsp; + struct clk gtc_clk; void *gtc_base; };
@@ -73,6 +76,7 @@ static int k3_arm64_load(struct udevice *dev, ulong addr, ulong size) static int k3_arm64_start(struct udevice *dev) { struct k3_arm64_privdata *rproc = dev_get_priv(dev); + ulong gtc_rate; int ret;
dev_dbg(dev, "%s\n", __func__); @@ -83,6 +87,11 @@ static int k3_arm64_start(struct udevice *dev) return ret; }
+ gtc_rate = clk_get_rate(&rproc->gtc_clk); + dev_dbg(dev, "GTC RATE= %d\n", (u32) gtc_rate); + /* Store the clock frequency down for GTC users to pick up */ + writel((u32)gtc_rate, rproc->gtc_base + GTC_CNTFID0_REG); + /* Enable the timer before starting remote core */ writel(GTC_CNTR_EN, rproc->gtc_base + GTC_CNTCR_REG);
@@ -169,6 +178,12 @@ static int k3_arm64_of_to_priv(struct udevice *dev, return ret; }
+ ret = clk_get_by_index(dev, 0, &rproc->gtc_clk); + if (ret) { + dev_err(dev, "clk_get failed: %d\n", ret); + return ret; + } + ret = reset_get_by_index(dev, 0, &rproc->rproc_rst); if (ret) { dev_err(dev, "reset_get() failed: %d\n", ret);

On 07/01/21 12:50 am, Nishanth Menon wrote:
Hi,
ARMv8's generic timer[1] picks up it's graycode from GTC. However, the frequency of the GTC is supposed to be programmed in CNTFID0[2] register prior to enabling the GTC in CNTCR[3] register. U-boot for K3 is currently missing that programming sequence. This series introduces the clock nodes necessary for the lookup and writes the frequency to the corresponding register for TF-A to pick up [4].
Based on: 62b07b5173e3 (tag: v2021.01-rc5, upstream/master) Prepare v2021.01-rc5
J721e: https://pastebin.ubuntu.com/p/Wcxfd3BNSv/ J7200: https://pastebin.ubuntu.com/p/VsVqn7y8pv/ I dont have an AM654-evm handy at the moment, but the programming sequences are similar in nature.
Applied to u-boot-ti/for-next branch.
Thanks and regards, Lokesh
participants (2)
-
Lokesh Vutla
-
Nishanth Menon