Skip to contentSkip to conversation

J72xx: HSM rearch support series

PATCH Superseded Tero Kristo 3 participants 30 messages May 11, 2021 – Jun 2, 2021

Subsystems (7)

v4 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Tero Kristo

author maintainer
#0 lore raw

v4 cover J72xx: HSM rearch support series

Hello,

Couple of small changes in v4:
- re-worked patch #14 to include review comments from Jaehoon Chung
* changed code to use iopoll version instead of hand crafted loops for timeout handling
* other mostly cosmetic changes
- patch #19/#21 changed to allow RM init to happen based on comment from Vignesh

-Tero

author maintainer
#1 lore raw

PATCH v4 1/26 lib: rational: copy the rational fraction lib routines from Linux

From: Tero Kristo <[email protected]>

Copy the best rational approximation calculation routines from Linux. Typical usecase for these routines is to calculate the M/N divider values for PLLs to reach a specific clock rate.

This is based on linux kernel commit:
"lib/math/rational.c: fix possible incorrect result from rational fractions helper"
(sha1: 323dd2c3ed0641f49e89b4e420f9eef5d3d5a881)

---
 include/linux/rational.h | 20 ++++++++
 lib/Kconfig              |  7 +++
 lib/Makefile             |  2 +
 lib/rational.c           | 99 ++++++++++++++++++++++++++++++++++++++++
 4 files changed, 128 insertions(+)
 create mode 100644 include/linux/rational.h
 create mode 100644 lib/rational.c
include/linux/rational.h new file +20 -0
new file mode 100644
index 0000000000..33f5f5fc3e
@@ -0,0 +1,20 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * rational fractions
+ *
+ * Copyright (C) 2009 emlix GmbH, Oskar Schirmer <[email protected]>
+ *
+ * helper functions when coping with rational numbers,
+ * e.g. when calculating optimum numerator/denominator pairs for
+ * pll configuration taking into account restricted register size
+ */
+
+#ifndef _LINUX_RATIONAL_H
+#define _LINUX_RATIONAL_H
+
+void rational_best_approximation(
+	unsigned long given_numerator, unsigned long given_denominator,
+	unsigned long max_numerator, unsigned long max_denominator,
+	unsigned long *best_numerator, unsigned long *best_denominator);
+
+#endif /* _LINUX_RATIONAL_H */
lib/Kconfig +7 -0
index 6d2d41de30..68d58aa90d 100644
@@ -663,6 +663,13 @@ config GENERATE_SMBIOS_TABLE
 	  See also SMBIOS_SYSINFO which allows SMBIOS values to be provided in
 	  the devicetree.
 
+config LIB_RATIONAL
+	bool "enable continued fraction calculation routines"
+
+config SPL_LIB_RATIONAL
+	bool "enable continued fraction calculation routines for SPL"
+	depends on SPL
+
 endmenu
 
 config ASN1_COMPILER
lib/Makefile +2 -0
index 6825671955..ea6bd0cb55 100644
@@ -73,6 +73,8 @@ obj-$(CONFIG_$(SPL_)LZO) += lzo/
 obj-$(CONFIG_$(SPL_)LZMA) += lzma/
 obj-$(CONFIG_$(SPL_)LZ4) += lz4_wrapper.o
 
+obj-$(CONFIG_$(SPL_)LIB_RATIONAL) += rational.o
+
 obj-$(CONFIG_LIBAVB) += libavb/
 
 obj-$(CONFIG_$(SPL_TPL_)OF_LIBFDT) += libfdt/
lib/rational.c new file +99 -0

99 lines not on this page · raw message

Signature · 1 line
2.17.1
author maintainer
#2 lore raw

PATCH v4 2/26 arm: mach-k3: introduce new config option for sysfw split

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

On J7 family of SoCs (J721E and J7200), sysfw is being split to be run under two cores, TIFS portion on DMSC core, and DM firmware under MCU R5. As MCU R5 is also used to run one phase of the bootloader, we must prevent access from here towards sysfw services. To support this, add new config option which can be used to detect presence of RM/PM sysfw services.

---
 arch/arm/mach-k3/Kconfig | 11 +++++++++++
 1 file changed, 11 insertions(+)
arch/arm/mach-k3/Kconfig +11 -0
index c7d186149b..c92722eab6 100644
@@ -141,6 +141,17 @@ config SYS_K3_SPL_ATF
 	  Enabling this will try to start Cortex-A (typically with ATF)
 	  after SPL from R5.
 
+config K3_DM_FW
+	bool "Separate DM firmware image"
+	depends on SPL && CPU_V7R && SOC_K3_J721E && !CLK_TI_SCI && !TI_SCI_POWER_DOMAIN
+	default y
+	help
+	  Enabling this will indicate that the system has separate DM
+	  and TIFS firmware images in place, instead of a single SYSFW
+	  firmware. Due to DM being executed on the same core as R5 SPL
+	  bootloader, it makes RM and PM services not being available
+	  during R5 SPL execution time.
+
 source "board/ti/am65x/Kconfig"
 source "board/ti/j721e/Kconfig"
 endif
Signature · 1 line
2.17.1
author maintainer
#3 lore raw

PATCH v4 3/26 remoteproc: k3-r5: remove sysfw PM calls if not supported

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

With the sysfw rearch, sysfw PM calls are no longer available from SPL level. To properly support this, remove the is_on checks and the reset assertion from the R5 remoteproc driver as these are not supported. Attempting to access unavailable services will cause the device to hang.

---
 drivers/remoteproc/ti_k3_r5f_rproc.c | 9 +++++++++
 1 file changed, 9 insertions(+)
drivers/remoteproc/ti_k3_r5f_rproc.c +9 -0
index 3c569a3b7b..c8c30b9020 100644
@@ -781,7 +781,9 @@ static int k3_r5f_probe(struct udevice *dev)
 {
 	struct k3_r5f_cluster *cluster = dev_get_priv(dev->parent);
 	struct k3_r5f_core *core = dev_get_priv(dev);
+#ifndef CONFIG_K3_DM_FW
 	bool r_state;
+#endif
 	int ret;
 
 	dev_dbg(dev, "%s\n", __func__);
@@ -804,6 +806,12 @@ static int k3_r5f_probe(struct udevice *dev)
 		return ret;
 	}
 
+	/*
+	 * The PM functionality is not supported by the firmware during
+	 * SPL execution with the separated DM firmware image. The following
+	 * piece of code is not compiled in that case.
+	 */
+#ifndef CONFIG_K3_DM_FW
 	ret = core->tsp.sci->ops.dev_ops.is_on(core->tsp.sci, core->tsp.dev_id,
 					       &r_state, &core->in_use);
 	if (ret)
@@ -817,6 +825,7 @@ static int k3_r5f_probe(struct udevice *dev)
 
 	/* Make sure Local reset is asserted. Redundant? */
 	reset_assert(&core->reset);
+#endif
 
 	ret = k3_r5f_rproc_configure(core);
 	if (ret) {
Signature · 1 line
2.17.1
author maintainer
#4 lore raw

PATCH v4 4/26 common: fit: Update board_fit_image_post_process() to pass fit and node_offset

In reply to Tero Kristo

From: Lokesh Vutla <[email protected]>

board_fit_image_post_process() passes only start and size of the image, but type of the image is not passed. So pass fit and node_offset, to derive information about image to be processed.

---
 arch/arm/mach-k3/security.c  | 3 ++-
 arch/arm/mach-keystone/mon.c | 3 ++-
 board/ti/am335x/board.c      | 3 ++-
 board/ti/am43xx/board.c      | 3 ++-
 board/ti/am57xx/board.c      | 3 ++-
 board/ti/dra7xx/evm.c        | 3 ++-
 common/image-fit.c           | 2 +-
 common/spl/spl_fit.c         | 2 +-
 include/image.h              | 5 ++++-
 9 files changed, 18 insertions(+), 9 deletions(-)
arch/arm/mach-k3/security.c +2 -1
index 66f90a5a34..5b5ff9ba7b 100644
@@ -18,7 +18,8 @@
 #include <spl.h>
 #include <asm/arch/sys_proto.h>
 
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	struct ti_sci_handle *ti_sci = get_ti_sci_handle();
 	struct ti_sci_proc_ops *proc_ops = &ti_sci->ops.proc_ops;
arch/arm/mach-keystone/mon.c +2 -1
index 58995d73ac..b863bab196 100644
@@ -103,7 +103,8 @@ static int k2_hs_bm_auth(int cmd, void *arg1)
 	return  result;
 }
 
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	int result = 0;
 	void *image = *p_image;
board/ti/am335x/board.c +2 -1
index bc1657e88f..6952484d01 100644
@@ -957,7 +957,8 @@ int board_fit_config_name_match(const char *name)
 #endif
 
 #ifdef CONFIG_TI_SECURE_DEVICE
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	secure_boot_verify_image(p_image, p_size);
 }
board/ti/am43xx/board.c +2 -1
index e9febb9592..a71b588efc 100644
@@ -896,7 +896,8 @@ int embedded_dtb_select(void)
 #endif
 
 #ifdef CONFIG_TI_SECURE_DEVICE
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	secure_boot_verify_image(p_image, p_size);
 }
board/ti/am57xx/board.c +2 -1
index 73063faee6..99d1ad3215 100644
@@ -1198,7 +1198,8 @@ static int board_bootmode_has_emmc(void)
 #endif
 
 #ifdef CONFIG_TI_SECURE_DEVICE
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	secure_boot_verify_image(p_image, p_size);
 }
board/ti/dra7xx/evm.c +2 -1
index 05f251f778..23e8005991 100644
@@ -1065,7 +1065,8 @@ int fastboot_set_reboot_flag(enum fastboot_reboot_reason reason)
 #endif
 
 #ifdef CONFIG_TI_SECURE_DEVICE
-void board_fit_image_post_process(void **p_image, size_t *p_size)
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
 {
 	secure_boot_verify_image(p_image, p_size);
 }
common/image-fit.c +1 -1
index e614643fe3..0c5a05948d 100644
@@ -2143,7 +2143,7 @@ int fit_image_load(bootm_headers_t *images, ulong addr,
 
 	/* perform any post-processing on the image data */
 	if (!host_build() && IS_ENABLED(CONFIG_FIT_IMAGE_POST_PROCESS))
-		board_fit_image_post_process(&buf, &size);
+		board_fit_image_post_process(fit, noffset, &buf, &size);
 
 	len = (ulong)size;
 
common/spl/spl_fit.c +1 -1
index 4288f571fc..9f17146f78 100644
@@ -316,7 +316,7 @@ static int spl_load_fit_image(struct spl_load_info *info, ulong sector,
 	}
 
 	if (CONFIG_IS_ENABLED(FIT_IMAGE_POST_PROCESS))
-		board_fit_image_post_process(&src, &length);
+		board_fit_image_post_process(fit, node, &src, &length);
 
 	load_ptr = map_sysmem(load_addr, length);
 	if (IS_ENABLED(CONFIG_SPL_GZIP) && image_comp == IH_COMP_GZIP) {
include/image.h +4 -1
index 459685d4d4..0c24bf6f35 100644
@@ -1581,11 +1581,14 @@ int board_fit_config_name_match(const char *name);
  * into the FIT creation (i.e. the binary blobs would have been pre-processed
  * before being added to the FIT image).
  *
+ * @fit: pointer to fit image
+ * @node: offset of image node
  * @image: pointer to the image start pointer
  * @size: pointer to the image size
  * @return no return value (failure should be handled internally)
  */
-void board_fit_image_post_process(void **p_image, size_t *p_size);
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size);
 
 #define FDT_ERROR	((ulong)(-1))
 
Signature · 1 line
2.17.1
author maintainer
#5 lore raw

PATCH v4 5/26 clk: fixed_rate: add API for directly registering fixed rate clocks

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Current driver only supports registering fixed rate clocks from DT. Add new API which makes it possible to register fixed rate clocks directly from e.g. platform specific clock drivers.

---
 drivers/clk/clk_fixed_rate.c | 43 ++++++++++++++++++++++++++++++++++++
 include/linux/clk-provider.h |  3 +++
 2 files changed, 46 insertions(+)
drivers/clk/clk_fixed_rate.c +43 -0
index 09f9ef26a4..c591abf685 100644
@@ -9,6 +9,9 @@
 #include <dm/device-internal.h>
 #include <linux/clk-provider.h>
 
+#define UBOOT_DM_CLK_FIXED_RATE "fixed_rate_clock"
+#define UBOOT_DM_CLK_FIXED_RATE_RAW "fixed_rate_raw_clock"
+
 static ulong clk_fixed_rate_get_rate(struct clk *clk)
 {
 	return to_clk_fixed_rate(clk->dev)->fixed_rate;
@@ -40,6 +43,15 @@ void clk_fixed_rate_ofdata_to_plat_(struct udevice *dev,
 	clk->enable_count = 0;
 }
 
+static ulong clk_fixed_rate_raw_get_rate(struct clk *clk)
+{
+	return container_of(clk, struct clk_fixed_rate, clk)->fixed_rate;
+}
+
+const struct clk_ops clk_fixed_rate_raw_ops = {
+	.get_rate = clk_fixed_rate_raw_get_rate,
+};
+
 static int clk_fixed_rate_of_to_plat(struct udevice *dev)
 {
 	clk_fixed_rate_ofdata_to_plat_(dev, to_clk_fixed_rate(dev));
@@ -47,6 +59,30 @@ static int clk_fixed_rate_of_to_plat(struct udevice *dev)
 	return 0;
 }
 
+struct clk *clk_register_fixed_rate(struct device *dev, const char *name,
+				    ulong rate)
+{
+	struct clk *clk;
+	struct clk_fixed_rate *fixed;
+	int ret;
+
+	fixed = kzalloc(sizeof(*fixed), GFP_KERNEL);
+	if (!fixed)
+		return ERR_PTR(-ENOMEM);
+
+	fixed->fixed_rate = rate;
+
+	clk = &fixed->clk;
+
+	ret = clk_register(clk, UBOOT_DM_CLK_FIXED_RATE_RAW, name, NULL);
+	if (ret) {
+		kfree(fixed);
+		return ERR_PTR(ret);
+	}
+
+	return clk;
+}
+
 static const struct udevice_id clk_fixed_rate_match[] = {
 	{
 		.compatible = "fixed-clock",
@@ -63,3 +99,10 @@ U_BOOT_DRIVER(fixed_clock) = {
 	.ops = &clk_fixed_rate_ops,
 	.flags = DM_FLAG_PRE_RELOC,
 };
+
+U_BOOT_DRIVER(clk_fixed_rate_raw) = {
+	.name = UBOOT_DM_CLK_FIXED_RATE_RAW,
+	.id = UCLASS_CLK,
+	.ops = &clk_fixed_rate_raw_ops,
+	.flags = DM_FLAG_PRE_RELOC,
+};
include/linux/clk-provider.h +3 -0
index 6fda14f5fe..9d296f240a 100644
@@ -247,6 +247,9 @@ struct clk *clk_register_mux(struct device *dev, const char *name,
 		void __iomem *reg, u8 shift, u8 width,
 		u8 clk_mux_flags);
 
+struct clk *clk_register_fixed_rate(struct device *dev, const char *name,
+				    ulong rate);
+
 const char *clk_hw_get_name(const struct clk *hw);
 ulong clk_generic_get_rate(struct clk *clk);
 
Signature · 1 line
2.17.1
author maintainer
#6 lore raw

PATCH v4 6/26 clk: fix clock tree dump to properly dump out every registered clock

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Some clocks are not associated to a DM node, so just parsing the DM is not enough. This is especially true for root clocks, which typically don't have any parents. Instead, fetch every registered UCLASS_CLK instance, and dump these out.

---
 cmd/clk.c | 27 +++++++++++++++++++--------
 1 file changed, 19 insertions(+), 8 deletions(-)
cmd/clk.c +19 -8
index e3c3d2f9bb..0245b97136 100644
@@ -18,11 +18,14 @@ static void show_clks(struct udevice *dev, int depth, int last_flag)
 {
 	int i, is_last;
 	struct udevice *child;
-	struct clk *clkp;
+	struct clk *clkp, *parent;
 	u32 rate;
 
 	clkp = dev_get_clk_ptr(dev);
 	if (device_get_uclass_id(dev) == UCLASS_CLK && clkp) {
+		parent = clk_get_parent(clkp);
+		if (!IS_ERR(parent) && depth == -1)
+			return;
 		depth++;
 		rate = clk_get_rate(clkp);
 
@@ -47,6 +50,9 @@ static void show_clks(struct udevice *dev, int depth, int last_flag)
 	}
 
 	list_for_each_entry(child, &dev->child_head, sibling_node) {
+		if (child == dev)
+			continue;
+
 		is_last = list_is_last(&child->sibling_node, &dev->child_head);
 		show_clks(child, depth, (last_flag << 1) | is_last);
 	}
@@ -54,14 +60,19 @@ static void show_clks(struct udevice *dev, int depth, int last_flag)
 
 int __weak soc_clk_dump(void)
 {
-	struct udevice *root;
+	struct udevice *dev;
+	struct uclass *uc;
+	int ret;
 
-	root = dm_root();
-	if (root) {
-		printf(" Rate               Usecnt      Name\n");
-		printf("------------------------------------------\n");
-		show_clks(root, -1, 0);
-	}
+	ret = uclass_get(UCLASS_CLK, &uc);
+	if (ret)
+		return ret;
+
+	printf(" Rate               Usecnt      Name\n");
+	printf("------------------------------------------\n");
+
+	uclass_foreach_dev(dev, uc)
+		show_clks(dev, -1, 0);
 
 	return 0;
 }
Signature · 1 line
2.17.1
author maintainer
#7 lore raw

PATCH v4 7/26 clk: do not attempt to fetch clock pointer with null device

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Bail out early if device returned for the parent clock is null. This avoids warning prints like this when doing clk dump:

dev_get_uclass_priv: null device

---
 drivers/clk/clk-uclass.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/clk/clk-uclass.c +2 -0
index 4ab3c402ed..ac2b7ae84e 100644
@@ -499,6 +499,8 @@ struct clk *clk_get_parent(struct clk *clk)
 		return NULL;
 
 	pdev = dev_get_parent(clk->dev);
+	if (!pdev)
+		return ERR_PTR(-ENODEV);
 	pclk = dev_get_clk_ptr(pdev);
 	if (!pclk)
 		return ERR_PTR(-ENODEV);
Signature · 1 line
2.17.1
author maintainer
#8 lore raw

PATCH v4 8/26 clk: add support for setting clk rate from cmdline

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add new clk subcommand "clk setfreq", for setting up a clock rate directly from u-boot cmdline. This is handy for any debugging purposes towards clocks.

---
 cmd/clk.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 47 insertions(+), 2 deletions(-)
cmd/clk.c +47 -2
index 0245b97136..fd7944c02e 100644
@@ -98,8 +98,52 @@ static int do_clk_dump(struct cmd_tbl *cmdtp, int flag, int argc,
 	return ret;
 }
 
+struct udevice *clk_lookup(const char *name)
+{
+	int i = 0;
+	struct udevice *dev;
+
+	do {
+		uclass_get_device(UCLASS_CLK, i++, &dev);
+		if (!strcmp(name, dev->name))
+			return dev;
+	} while (dev);
+
+	return NULL;
+}
+
+static int do_clk_setfreq(struct cmd_tbl *cmdtp, int flag, int argc,
+			  char *const argv[])
+{
+	struct clk *clk = NULL;
+	s32 freq;
+	struct udevice *dev;
+
+	freq = simple_strtoul(argv[2], NULL, 10);
+
+	dev = clk_lookup(argv[1]);
+
+	if (dev)
+		clk = dev_get_clk_ptr(dev);
+
+	if (!clk) {
+		printf("clock '%s' not found.\n", argv[1]);
+		return -EINVAL;
+	}
+
+	freq = clk_set_rate(clk, freq);
+	if (freq < 0) {
+		printf("set_rate failed: %d\n", freq);
+		return CMD_RET_FAILURE;
+	}
+
+	printf("set_rate returns %u\n", freq);
+	return 0;
+}
+
 static struct cmd_tbl cmd_clk_sub[] = {
 	U_BOOT_CMD_MKENT(dump, 1, 1, do_clk_dump, "", ""),
+	U_BOOT_CMD_MKENT(setfreq, 3, 1, do_clk_setfreq, "", ""),
 };
 
 static int do_clk(struct cmd_tbl *cmdtp, int flag, int argc,
@@ -124,7 +168,8 @@ static int do_clk(struct cmd_tbl *cmdtp, int flag, int argc,
 
 #ifdef CONFIG_SYS_LONGHELP
 static char clk_help_text[] =
-	"dump - Print clock frequencies";
+	"dump - Print clock frequencies\n"
+	"setfreq [clk] [freq] - Set clock frequency";
 #endif
 
-U_BOOT_CMD(clk, 2, 1, do_clk, "CLK sub-system", clk_help_text);
+U_BOOT_CMD(clk, 4, 1, do_clk, "CLK sub-system", clk_help_text);
Signature · 1 line
2.17.1
author maintainer
#9 lore raw

PATCH v4 9/26 clk: sci-clk: fix return value of set_rate

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Set rate should return the new clock rate on success, and negative error value on failure. Fix this, as currently set_rate returns 0 on success.

---
 drivers/clk/ti/clk-sci.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/clk/ti/clk-sci.c +4 -2
index 6f0fdaa111..acb9eadf03 100644
@@ -111,10 +111,12 @@ static ulong ti_sci_clk_set_rate(struct clk *clk, ulong rate)
 #endif
 
 	ret = cops->set_freq(sci, clk->id, clk->data, 0, rate, ULONG_MAX);
-	if (ret)
+	if (ret) {
 		dev_err(clk->dev, "%s: set_freq failed (%d)\n", __func__, ret);
+		return ret;
+	}
 
-	return ret;
+	return rate;
 }
 
 static int ti_sci_clk_set_parent(struct clk *clk, struct clk *parent)
Signature · 1 line
2.17.1
author maintainer
#10 lore raw

PATCH v4 10/26 clk: fix assigned-clocks to pass with deferring provider

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

If a clock provider is not ready for assigning default rates/parents during its probe, it may return -EPROBE_DEFER directly from xlate. Handle this special case properly by skipping the entry and adjusting the return value to pass. The defaults will be handled properly in post probe phase then.

---
 drivers/clk/clk-uclass.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)
drivers/clk/clk-uclass.c +18 -0
index ac2b7ae84e..5ebf7a33cb 100644
@@ -240,6 +240,15 @@ static int clk_set_default_parents(struct udevice *dev, int stage)
 
 		ret = clk_get_by_indexed_prop(dev, "assigned-clocks",
 					      index, &clk);
+		/*
+		 * If the clock provider is not ready yet, let it handle
+		 * the re-programming later.
+		 */
+		if (ret == -EPROBE_DEFER) {
+			ret = 0;
+			continue;
+		}
+
 		if (ret) {
 			debug("%s: could not get assigned clock %d for %s\n",
 			      __func__, index, dev_read_name(dev));
@@ -308,6 +317,15 @@ static int clk_set_default_rates(struct udevice *dev, int stage)
 
 		ret = clk_get_by_indexed_prop(dev, "assigned-clocks",
 					      index, &clk);
+		/*
+		 * If the clock provider is not ready yet, let it handle
+		 * the re-programming later.
+		 */
+		if (ret == -EPROBE_DEFER) {
+			ret = 0;
+			continue;
+		}
+
 		if (ret) {
 			debug("%s: could not get assigned clock %d for %s\n",
 			      __func__, index, dev_read_name(dev));
Signature · 1 line
2.17.1
author maintainer
#11 lore raw

PATCH v4 11/26 clk: fix set_rate to clean up cached rates for the hierarchy

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Clock rates are cached within the individual clock nodes, and right now if one changes a clock rate somewhere in the middle of the tree, none of its child clocks notice the change. To fix this, clear up all the cached rates for us and our child clocks.

---
 drivers/clk/clk-uclass.c | 19 +++++++++++++++++++
 1 file changed, 19 insertions(+)
drivers/clk/clk-uclass.c +19 -0
index 5ebf7a33cb..0e495a6a2e 100644
@@ -565,6 +565,22 @@ ulong clk_round_rate(struct clk *clk, ulong rate)
 	return ops->round_rate(clk, rate);
 }
 
+static void clk_clean_rate_cache(struct clk *clk)
+{
+	struct udevice *child_dev;
+	struct clk *clkp;
+
+	if (!clk)
+		return;
+
+	clk->rate = 0;
+
+	list_for_each_entry(child_dev, &clk->dev->child_head, sibling_node) {
+		clkp = dev_get_clk_ptr(child_dev);
+		clk_clean_rate_cache(clkp);
+	}
+}
+
 ulong clk_set_rate(struct clk *clk, ulong rate)
 {
 	const struct clk_ops *ops;
@@ -577,6 +593,9 @@ ulong clk_set_rate(struct clk *clk, ulong rate)
 	if (!ops->set_rate)
 		return -ENOSYS;
 
+	/* Clean up cached rates for us and all child clocks */
+	clk_clean_rate_cache(clk);
+
 	return ops->set_rate(clk, rate);
 }
 
Signature · 1 line
2.17.1
author maintainer
#12 lore raw

PATCH v4 12/26 clk: add support for TI K3 SoC PLL

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add support for TI K3 SoC PLLs. This clock type supports enabling/disabling/setting and querying the clock rate for the PLL. The euclidean library routine is used to calculate divider/multiplier rates for the PLLs.

---
 drivers/clk/ti/Kconfig      |  12 ++
 drivers/clk/ti/Makefile     |   1 +
 drivers/clk/ti/clk-k3-pll.c | 283 ++++++++++++++++++++++++++++++++++++
 include/k3-clk.h            |  15 ++
 4 files changed, 311 insertions(+)
 create mode 100644 drivers/clk/ti/clk-k3-pll.c
 create mode 100644 include/k3-clk.h
drivers/clk/ti/Kconfig +12 -0
index 2dc86d44a9..a8ec4f541a 100644
@@ -41,3 +41,15 @@ config CLK_TI_SCI
 	  This enables the clock driver support over TI System Control Interface
 	  available on some new TI's SoCs. If you wish to use clock resources
 	  managed by the TI System Controller, say Y here. Otherwise, say N.
+
+config CLK_K3_PLL
+	bool "PLL clock support for K3 SoC family of devices"
+	depends on CLK && LIB_RATIONAL
+	help
+	  Enables PLL clock support for K3 SoC family of devices.
+
+config SPL_CLK_K3_PLL
+	bool "PLL clock support for K3 SoC family of devices"
+	depends on CLK && LIB_RATIONAL && SPL
+	help
+	  Enables PLL clock support for K3 SoC family of devices.
drivers/clk/ti/Makefile +1 -0
index 9f56b47736..47839213e5 100644
@@ -11,3 +11,4 @@ obj-$(CONFIG_CLK_TI_DIVIDER) += clk-divider.o
 obj-$(CONFIG_CLK_TI_GATE) += clk-gate.o
 obj-$(CONFIG_CLK_TI_MUX) += clk-mux.o
 obj-$(CONFIG_CLK_TI_SCI) += clk-sci.o
+obj-$(CONFIG_$(SPL_TPL_)CLK_K3_PLL) += clk-k3-pll.o
drivers/clk/ti/clk-k3-pll.c new file +283 -0

283 lines not on this page · raw message

include/k3-clk.h new file +15 -0
new file mode 100644
index 0000000000..fc84378d03
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * (C) Copyright 2020 - Texas Instruments Incorporated - http://www.ti.com
+ *      Tero Kristo <[email protected]>
+ */
+
+#ifndef __K3_CLK_H__
+#define __K3_CLK_H__
+
+#include <linux/clk-provider.h>
+
+struct clk *clk_register_ti_pll(const char *name, const char *parent_name,
+				void __iomem *reg);
+
+#endif /* __K3_CLK_H__ */
Signature · 1 line
2.17.1
author maintainer
#13 lore raw

PATCH v4 13/26 clk: add support for TI K3 SoC clocks

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add driver to support TI K3 generation SoC clocks. This driver registers the clocks provided via platform data, and adds support for controlling the clocks via DT handles.

---
 drivers/clk/ti/Kconfig  |  12 ++
 drivers/clk/ti/Makefile |   1 +
 drivers/clk/ti/clk-k3.c | 374 ++++++++++++++++++++++++++++++++++++++++
 include/k3-clk.h        | 162 +++++++++++++++++
 4 files changed, 549 insertions(+)
 create mode 100644 drivers/clk/ti/clk-k3.c
drivers/clk/ti/Kconfig +12 -0
index a8ec4f541a..fbcdefd889 100644
@@ -53,3 +53,15 @@ config SPL_CLK_K3_PLL
 	depends on CLK && LIB_RATIONAL && SPL
 	help
 	  Enables PLL clock support for K3 SoC family of devices.
+
+config CLK_K3
+	bool "Clock support for K3 SoC family of devices"
+	depends on CLK
+	help
+	  Enables the clock translation layer from DT to device clocks.
+
+config SPL_CLK_K3
+	bool "Clock support for K3 SoC family of devices"
+	depends on CLK && SPL
+	help
+	  Enables the clock translation layer from DT to device clocks.
drivers/clk/ti/Makefile +1 -0
index 47839213e5..07aa9a53e0 100644
@@ -12,3 +12,4 @@ obj-$(CONFIG_CLK_TI_GATE) += clk-gate.o
 obj-$(CONFIG_CLK_TI_MUX) += clk-mux.o
 obj-$(CONFIG_CLK_TI_SCI) += clk-sci.o
 obj-$(CONFIG_$(SPL_TPL_)CLK_K3_PLL) += clk-k3-pll.o
+obj-$(CONFIG_$(SPL_TPL_)CLK_K3) += clk-k3.o
drivers/clk/ti/clk-k3.c new file +374 -0

374 lines not on this page · raw message

include/k3-clk.h +162 -0

169 lines not on this page · raw message

Signature · 1 line
2.17.1
author maintainer
#14 lore raw

PATCH v4 14/26 power: domain: Introduce driver for raw TI K3 PDs

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Normally, power domains are handled via TI-SCI in K3 SoCs. However, SPL is not going to have access to sysfw resources, so it must control them directly. Add driver for supporting this.

---
 drivers/power/domain/Kconfig           |   7 +
 drivers/power/domain/Makefile          |   1 +
 drivers/power/domain/ti-power-domain.c | 368 +++++++++++++++++++++++++
 include/k3-dev.h                       |  76 +++++
 4 files changed, 452 insertions(+)
 create mode 100644 drivers/power/domain/ti-power-domain.c
 create mode 100644 include/k3-dev.h
drivers/power/domain/Kconfig +7 -0
index a0fd980752..99b3f9ae71 100644
@@ -72,4 +72,11 @@ config TI_SCI_POWER_DOMAIN
 	help
 	  Generic power domain implementation for TI devices implementing the
 	  TI SCI protocol.
+
+config TI_POWER_DOMAIN
+	bool "Enable the TI K3 Power domain driver"
+	depends on POWER_DOMAIN && ARCH_K3
+	help
+	  Generic power domain implementation for TI K3 devices.
+
 endmenu
drivers/power/domain/Makefile +1 -0
index 45bf9f6383..3d1e5f073c 100644
@@ -14,3 +14,4 @@ obj-$(CONFIG_SANDBOX_POWER_DOMAIN) += sandbox-power-domain.o
 obj-$(CONFIG_SANDBOX_POWER_DOMAIN) += sandbox-power-domain-test.o
 obj-$(CONFIG_TEGRA186_POWER_DOMAIN) += tegra186-power-domain.o
 obj-$(CONFIG_TI_SCI_POWER_DOMAIN) += ti-sci-power-domain.o
+obj-$(CONFIG_TI_POWER_DOMAIN) += ti-power-domain.o
drivers/power/domain/ti-power-domain.c new file +368 -0

368 lines not on this page · raw message

include/k3-dev.h new file +76 -0

76 lines not on this page · raw message

Signature · 1 line
2.17.1
#15 lore raw

PATCH v4 14/26 power: domain: Introduce driver for raw TI K3 PDs

Quoted 8 lines · Tero Kristo

From: Tero Kristo <[email protected]>

Normally, power domains are handled via TI-SCI in K3 SoCs. However, SPL is not going to have access to sysfw resources, so it must control them directly. Add driver for supporting this.

Best Regards,
Jaehoon Chung

Quoted 491 lines
---
 drivers/power/domain/Kconfig           |   7 +
 drivers/power/domain/Makefile          |   1 +
 drivers/power/domain/ti-power-domain.c | 368 +++++++++++++++++++++++++
 include/k3-dev.h                       |  76 +++++
 4 files changed, 452 insertions(+)
 create mode 100644 drivers/power/domain/ti-power-domain.c
 create mode 100644 include/k3-dev.h
drivers/power/domain/Kconfig +7 -0
index a0fd980752..99b3f9ae71 100644
@@ -72,4 +72,11 @@ config TI_SCI_POWER_DOMAIN
 	help
 	  Generic power domain implementation for TI devices implementing the
 	  TI SCI protocol.
+
+config TI_POWER_DOMAIN
+	bool "Enable the TI K3 Power domain driver"
+	depends on POWER_DOMAIN && ARCH_K3
+	help
+	  Generic power domain implementation for TI K3 devices.
+
 endmenu
drivers/power/domain/Makefile +1 -0
index 45bf9f6383..3d1e5f073c 100644
@@ -14,3 +14,4 @@ obj-$(CONFIG_SANDBOX_POWER_DOMAIN) += sandbox-power-domain.o
 obj-$(CONFIG_SANDBOX_POWER_DOMAIN) += sandbox-power-domain-test.o
 obj-$(CONFIG_TEGRA186_POWER_DOMAIN) += tegra186-power-domain.o
 obj-$(CONFIG_TI_SCI_POWER_DOMAIN) += ti-sci-power-domain.o
+obj-$(CONFIG_TI_POWER_DOMAIN) += ti-power-domain.o
drivers/power/domain/ti-power-domain.c new file +368 -0
new file mode 100644
index 0000000000..e418a7b996
@@ -0,0 +1,368 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Texas Instruments power domain driver
+ *
+ * Copyright (C) 2020-2021 Texas Instruments Incorporated - http://www.ti.com/
+ *	Tero Kristo <[email protected]>
+ */
+
+#include <asm/io.h>
+#include <common.h>
+#include <dm.h>
+#include <errno.h>
+#include <power-domain-uclass.h>
+#include <soc.h>
+#include <k3-dev.h>
+#include <linux/iopoll.h>
+
+#define PSC_PTCMD		0x120
+#define PSC_PTSTAT		0x128
+#define PSC_PDSTAT		0x200
+#define PSC_PDCTL		0x300
+#define PSC_MDSTAT		0x800
+#define PSC_MDCTL		0xa00
+
+#define PDCTL_STATE_MASK		0x1
+#define PDCTL_STATE_OFF			0x0
+#define PDCTL_STATE_ON			0x1
+
+#define MDSTAT_STATE_MASK		0x3f
+#define MDSTAT_BUSY_MASK		0x30
+#define MDSTAT_STATE_SWRSTDISABLE	0x0
+#define MDSTAT_STATE_ENABLE		0x3
+
+#define LPSC_TIMEOUT		1000
+#define PD_TIMEOUT		1000
+
+static u32 psc_read(struct ti_psc *psc, u32 reg)
+{
+	u32 val;
+
+	val = readl(psc->base + reg);
+	debug("%s: 0x%x from %p\n", __func__, val, psc->base + reg);
+	return val;
+}
+
+static void psc_write(u32 val, struct ti_psc *psc, u32 reg)
+{
+	debug("%s: 0x%x to %p\n", __func__, val, psc->base + reg);
+	writel(val, psc->base + reg);
+}
+
+static u32 pd_read(struct ti_pd *pd, u32 reg)
+{
+	return psc_read(pd->psc, reg + 4 * pd->id);
+}
+
+static void pd_write(u32 val, struct ti_pd *pd, u32 reg)
+{
+	psc_write(val, pd->psc, reg + 4 * pd->id);
+}
+
+static u32 lpsc_read(struct ti_lpsc *lpsc, u32 reg)
+{
+	return psc_read(lpsc->psc, reg + 4 * lpsc->id);
+}
+
+static void lpsc_write(u32 val, struct ti_lpsc *lpsc, u32 reg)
+{
+	psc_write(val, lpsc->psc, reg + 4 * lpsc->id);
+}
+
+static const struct soc_attr ti_k3_soc_pd_data[] = {
+#ifdef CONFIG_SOC_K3_J721E
+	{
+		.family = "J721E",
+		.data = &j721e_pd_platdata,
+	},
+	{
+		.family = "J7200",
+		.data = &j7200_pd_platdata,
+	},
+#endif
+	{ /* sentinel */ }
+};
+
+static int ti_power_domain_probe(struct udevice *dev)
+{
+	struct ti_k3_pd_platdata *data = dev_get_priv(dev);
+	const struct soc_attr *soc_match_data;
+	const struct ti_k3_pd_platdata *pdata;
+
+	printf("%s(dev=%p)\n", __func__, dev);
+
+	if (!data)
+		return -ENOMEM;
+
+	soc_match_data = soc_device_match(ti_k3_soc_pd_data);
+	if (!soc_match_data)
+		return -ENODEV;
+
+	pdata = (const struct ti_k3_pd_platdata *)soc_match_data->data;
+
+	data->psc = pdata->psc;
+	data->pd = pdata->pd;
+	data->lpsc = pdata->lpsc;
+	data->devs = pdata->devs;
+	data->num_psc = pdata->num_psc;
+	data->num_pd = pdata->num_pd;
+	data->num_lpsc = pdata->num_lpsc;
+	data->num_devs = pdata->num_devs;
+
+	return 0;
+}
+
+static int ti_pd_wait(struct ti_pd *pd)
+{
+	u32 ptstat;
+	int ret;
+
+	ret = readl_poll_timeout(pd->psc->base + PSC_PTSTAT, ptstat,
+				 !(ptstat & BIT(pd->id)), PD_TIMEOUT);
+
+	if (ret)
+		printf("%s: psc%d, pd%d failed to transition.\n", __func__,
+		       pd->psc->id, pd->id);
+
+	return ret;
+}
+
+static void ti_pd_transition(struct ti_pd *pd)
+{
+	psc_write(BIT(pd->id), pd->psc, PSC_PTCMD);
+}
+
+static u8 ti_pd_state(struct ti_pd *pd)
+{
+	return pd_read(pd, PSC_PDCTL) & PDCTL_STATE_MASK;
+}
+
+static int ti_pd_get(struct ti_pd *pd)
+{
+	u32 pdctl;
+	int ret;
+
+	pd->usecount++;
+
+	if (pd->usecount > 1)
+		return 0;
+
+	if (pd->depend) {
+		ret = ti_pd_get(pd->depend);
+		if (ret)
+			return ret;
+		ti_pd_transition(pd->depend);
+		ret = ti_pd_wait(pd->depend);
+		if (ret)
+			return ret;
+	}
+
+	pdctl = pd_read(pd, PSC_PDCTL);
+
+	if ((pdctl & PDCTL_STATE_MASK) == PDCTL_STATE_ON)
+		return 0;
+
+	debug("%s: enabling psc:%d, pd:%d\n", __func__, pd->psc->id, pd->id);
+
+	pdctl &= ~PDCTL_STATE_MASK;
+	pdctl |= PDCTL_STATE_ON;
+
+	pd_write(pdctl, pd, PSC_PDCTL);
+
+	return 0;
+}
+
+static int ti_pd_put(struct ti_pd *pd)
+{
+	u32 pdctl;
+	int ret;
+
+	pd->usecount--;
+
+	if (pd->usecount > 0)
+		return 0;
+
+	pdctl = pd_read(pd, PSC_PDCTL);
+	if ((pdctl & PDCTL_STATE_MASK) == PDCTL_STATE_OFF)
+		return 0;
+
+	pdctl &= ~PDCTL_STATE_MASK;
+	pdctl |= PDCTL_STATE_OFF;
+
+	debug("%s: disabling psc:%d, pd:%d\n", __func__, pd->psc->id, pd->id);
+
+	pd_write(pdctl, pd, PSC_PDCTL);
+
+	if (pd->depend) {
+		ti_pd_transition(pd);
+		ret = ti_pd_wait(pd);
+		if (ret)
+			return ret;
+
+		ret = ti_pd_put(pd->depend);
+		if (ret)
+			return ret;
+		ti_pd_transition(pd->depend);
+		ret = ti_pd_wait(pd->depend);
+		if (ret)
+			return ret;
+	}
+
+	return 0;
+}
+
+static int ti_lpsc_wait(struct ti_lpsc *lpsc)
+{
+	u32 mdstat;
+	int ret;
+
+	ret = readl_poll_timeout(lpsc->psc->base + PSC_MDSTAT + lpsc->id * 4,
+				 mdstat,
+				 !(mdstat & MDSTAT_BUSY_MASK), LPSC_TIMEOUT);
+
+	if (ret)
+		printf("%s: module %d failed to transition.\n", __func__,
+		       lpsc->id);
+
+	return ret;
+}
+
+static u8 lpsc_get_state(struct ti_lpsc *lpsc)
+{
+	return lpsc_read(lpsc, PSC_MDCTL) & MDSTAT_STATE_MASK;
+}
+
+static int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state)
+{
+	struct ti_pd *psc_pd;
+	int ret;
+	u32 mdctl;
+
+	psc_pd = lpsc->pd;
+
+	if (state == MDSTAT_STATE_ENABLE) {
+		lpsc->usecount++;
+		if (lpsc->usecount > 1)
+			return 0;
+	} else {
+		lpsc->usecount--;
+		if (lpsc->usecount >= 1)
+			return 0;
+	}
+
+	debug("%s: transitioning psc:%d, lpsc:%d to %x\n", __func__,
+	      lpsc->psc->id, lpsc->id, state);
+
+	if (lpsc->depend)
+		ti_lpsc_transition(lpsc->depend, state);
+
+	mdctl = lpsc_read(lpsc, PSC_MDCTL);
+	if ((mdctl & MDSTAT_STATE_MASK) == state)
+		return 0;
+
+	if (state == MDSTAT_STATE_ENABLE)
+		ti_pd_get(psc_pd);
+	else
+		ti_pd_put(psc_pd);
+
+	mdctl &= ~MDSTAT_STATE_MASK;
+	mdctl |= state;
+
+	lpsc_write(mdctl, lpsc, PSC_MDCTL);
+
+	ti_pd_transition(psc_pd);
+	ret = ti_pd_wait(psc_pd);
+	if (ret)
+		return ret;
+
+	return ti_lpsc_wait(lpsc);
+}
+
+static int ti_power_domain_transition(struct power_domain *pd, u8 state)
+{
+	struct ti_lpsc *lpsc = pd->priv;
+
+	return ti_lpsc_transition(lpsc, state);
+}
+
+static int ti_power_domain_on(struct power_domain *pd)
+{
+	debug("%s(pd=%p, id=%lu)\n", __func__, pd, pd->id);
+
+	return ti_power_domain_transition(pd, MDSTAT_STATE_ENABLE);
+}
+
+static int ti_power_domain_off(struct power_domain *pd)
+{
+	debug("%s(pd=%p, id=%lu)\n", __func__, pd, pd->id);
+
+	return ti_power_domain_transition(pd, MDSTAT_STATE_SWRSTDISABLE);
+}
+
+static struct ti_lpsc *lpsc_lookup(struct ti_k3_pd_platdata *data, int id)
+{
+	int idx;
+
+	for (idx = 0; idx < data->num_devs; idx++)
+		if (data->devs[idx].id == id)
+			return data->devs[idx].lpsc;
+
+	return NULL;
+}
+
+static int ti_power_domain_of_xlate(struct power_domain *pd,
+				    struct ofnode_phandle_args *args)
+{
+	struct ti_k3_pd_platdata *data = dev_get_priv(pd->dev);
+	struct ti_lpsc *lpsc;
+
+	debug("%s(power_domain=%p, id=%d)\n", __func__, pd, args->args[0]);
+
+	if (args->args_count < 1) {
+		printf("Invalid args_count: %d\n", args->args_count);
+		return -EINVAL;
+	}
+
+	lpsc = lpsc_lookup(data, args->args[0]);
+	if (!lpsc) {
+		printf("%s: invalid dev-id: %d\n", __func__, args->args[0]);
+		return -ENOENT;
+	}
+
+	pd->id = lpsc->id;
+	pd->priv = lpsc;
+
+	return 0;
+}
+
+static int ti_power_domain_request(struct power_domain *pd)
+{
+	return 0;
+}
+
+static int ti_power_domain_free(struct power_domain *pd)
+{
+	return 0;
+}
+
+static const struct udevice_id ti_power_domain_of_match[] = {
+	{ .compatible = "ti,sci-pm-domain" },
+	{ /* sentinel */ }
+};
+
+static struct power_domain_ops ti_power_domain_ops = {
+	.on = ti_power_domain_on,
+	.off = ti_power_domain_off,
+	.of_xlate = ti_power_domain_of_xlate,
+	.request = ti_power_domain_request,
+	.rfree = ti_power_domain_free,
+};
+
+U_BOOT_DRIVER(ti_pm_domains) = {
+	.name = "ti-pm-domains",
+	.id = UCLASS_POWER_DOMAIN,
+	.of_match = ti_power_domain_of_match,
+	.probe = ti_power_domain_probe,
+	.priv_auto = sizeof(struct ti_k3_pd_platdata),
+	.ops = &ti_power_domain_ops,
+};
include/k3-dev.h new file +76 -0
new file mode 100644
index 0000000000..de3a8bdf9e
@@ -0,0 +1,76 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Texas Instruments K3 Device Platform Data
+ *
+ * Copyright (C) 2020 Texas Instruments Incorporated - http://www.ti.com/
+ */
+#ifndef __K3_DEV_H__
+#define __K3_DEV_H__
+
+#include <asm/io.h>
+#include <linux/types.h>
+#include <stdint.h>
+
+#define LPSC_MODULE_EXISTS      BIT(0)
+#define LPSC_NO_CLOCK_GATING    BIT(1)
+#define LPSC_DEPENDS            BIT(2)
+#define LPSC_HAS_RESET_ISO      BIT(3)
+#define LPSC_HAS_LOCAL_RESET    BIT(4)
+#define LPSC_NO_MODULE_RESET    BIT(5)
+
+#define PSC_PD_EXISTS           BIT(0)
+#define PSC_PD_ALWAYSON         BIT(1)
+#define PSC_PD_DEPENDS          BIT(2)
+
+struct ti_psc {
+	int id;
+	void __iomem *base;
+};
+
+struct ti_pd;
+
+struct ti_pd {
+	int id;
+	int usecount;
+	struct ti_psc *psc;
+	struct ti_pd *depend;
+};
+
+struct ti_lpsc;
+
+struct ti_lpsc {
+	int id;
+	int usecount;
+	struct ti_psc *psc;
+	struct ti_pd *pd;
+	struct ti_lpsc *depend;
+};
+
+struct ti_dev {
+	struct ti_lpsc *lpsc;
+	int id;
+};
+
+/**
+ * struct ti_k3_pd_platdata - pm domain controller information structure
+ */
+struct ti_k3_pd_platdata {
+	struct ti_psc *psc;
+	struct ti_pd *pd;
+	struct ti_lpsc *lpsc;
+	struct ti_dev *devs;
+	int num_psc;
+	int num_pd;
+	int num_lpsc;
+	int num_devs;
+};
+
+#define PSC(_id, _base) { .id = _id, .base = (void *)_base, }
+#define PSC_PD(_id, _psc, _depend) { .id = _id, .psc = _psc, .depend = _depend }
+#define PSC_LPSC(_id, _psc, _pd, _depend) { .id = _id, .psc = _psc, .pd = _pd, .depend = _depend }
+#define PSC_DEV(_id, _lpsc) { .id = _id, .lpsc = _lpsc }
+
+extern const struct ti_k3_pd_platdata j721e_pd_platdata;
+extern const struct ti_k3_pd_platdata j7200_pd_platdata;
+
+#endif
author maintainer
#16 lore raw

PATCH v4 15/26 cmd: ti: pd: Add debug command for K3 power domains

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add support command for debugging K3 power domains. This is useful with the HSM rearch setup, where power domains are directly controlled by SPL instead of going through the TI SCI layer. The debugging support is only available in the u-boot codebase though, so the raw register access power domain layer must be enabled on u-boot side for this to work. By default, u-boot side uses the TI SCI layer, and R5 SPL only uses the direct access methods.

---
 cmd/ti/Kconfig                         |   8 ++
 cmd/ti/Makefile                        |   1 +
 cmd/ti/pd.c                            | 185 +++++++++++++++++++++++++
 drivers/power/domain/ti-power-domain.c |   6 +-
 include/k3-dev.h                       |   9 ++
 5 files changed, 206 insertions(+), 3 deletions(-)
 create mode 100644 cmd/ti/pd.c
cmd/ti/Kconfig +8 -0
index efeff0d482..db557445a8 100644
@@ -7,4 +7,12 @@ config CMD_DDR3
 	   supports memory verification, memory comapre and ecc
 	   verification if supported.
 
+config CMD_PD
+	bool "command for verifying power domains"
+	depends on TI_POWER_DOMAIN
+	help
+	   Debug command for K3 power domains. For this to work, the
+	   K3 power domain driver must be enabled for the u-boot; by
+	   default it is only enabled for SPL.
+
 endmenu
cmd/ti/Makefile +1 -0
index 16fbade9ed..045593396b 100644
@@ -5,4 +5,5 @@ obj- += dummy.o
 
 ifndef CONFIG_SPL_BUILD
 obj-$(CONFIG_CMD_DDR3) += ddr3.o
+obj-$(CONFIG_CMD_PD) += pd.o
 endif
cmd/ti/pd.c new file +185 -0

185 lines not on this page · raw message

drivers/power/domain/ti-power-domain.c +3 -3
index e418a7b996..567e7e1e8e 100644
@@ -132,7 +132,7 @@ static void ti_pd_transition(struct ti_pd *pd)
 	psc_write(BIT(pd->id), pd->psc, PSC_PTCMD);
 }
 
-static u8 ti_pd_state(struct ti_pd *pd)
+u8 ti_pd_state(struct ti_pd *pd)
 {
 	return pd_read(pd, PSC_PDCTL) & PDCTL_STATE_MASK;
 }
@@ -227,12 +227,12 @@ static int ti_lpsc_wait(struct ti_lpsc *lpsc)
 	return ret;
 }
 
-static u8 lpsc_get_state(struct ti_lpsc *lpsc)
+u8 lpsc_get_state(struct ti_lpsc *lpsc)
 {
 	return lpsc_read(lpsc, PSC_MDCTL) & MDSTAT_STATE_MASK;
 }
 
-static int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state)
+int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state)
 {
 	struct ti_pd *psc_pd;
 	int ret;
include/k3-dev.h +9 -0
index de3a8bdf9e..55c5057db3 100644
@@ -22,6 +22,11 @@
 #define PSC_PD_ALWAYSON         BIT(1)
 #define PSC_PD_DEPENDS          BIT(2)
 
+#define MDSTAT_STATE_MASK		0x3f
+#define MDSTAT_BUSY_MASK		0x30
+#define MDSTAT_STATE_SWRSTDISABLE	0x0
+#define MDSTAT_STATE_ENABLE		0x3
+
 struct ti_psc {
 	int id;
 	void __iomem *base;
@@ -73,4 +78,8 @@ struct ti_k3_pd_platdata {
 extern const struct ti_k3_pd_platdata j721e_pd_platdata;
 extern const struct ti_k3_pd_platdata j7200_pd_platdata;
 
+u8 ti_pd_state(struct ti_pd *pd);
+u8 lpsc_get_state(struct ti_lpsc *lpsc);
+int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state);
+
 #endif
Signature · 1 line
2.17.1
#17 lore raw

PATCH v4 15/26 cmd: ti: pd: Add debug command for K3 power domains

Quoted 10 lines · Tero Kristo

From: Tero Kristo <[email protected]>

Add support command for debugging K3 power domains. This is useful with the HSM rearch setup, where power domains are directly controlled by SPL instead of going through the TI SCI layer. The debugging support is only available in the u-boot codebase though, so the raw register access power domain layer must be enabled on u-boot side for this to work. By default, u-boot side uses the TI SCI layer, and R5 SPL only uses the direct access methods.

Quoted 2 lines

Best Regards,
Jaehoon Chung

Quoted 281 lines
---
 cmd/ti/Kconfig                         |   8 ++
 cmd/ti/Makefile                        |   1 +
 cmd/ti/pd.c                            | 185 +++++++++++++++++++++++++
 drivers/power/domain/ti-power-domain.c |   6 +-
 include/k3-dev.h                       |   9 ++
 5 files changed, 206 insertions(+), 3 deletions(-)
 create mode 100644 cmd/ti/pd.c
cmd/ti/Kconfig +8 -0
index efeff0d482..db557445a8 100644
@@ -7,4 +7,12 @@ config CMD_DDR3
 	   supports memory verification, memory comapre and ecc
 	   verification if supported.
 
+config CMD_PD
+	bool "command for verifying power domains"
+	depends on TI_POWER_DOMAIN
+	help
+	   Debug command for K3 power domains. For this to work, the
+	   K3 power domain driver must be enabled for the u-boot; by
+	   default it is only enabled for SPL.
+
 endmenu
cmd/ti/Makefile +1 -0
index 16fbade9ed..045593396b 100644
@@ -5,4 +5,5 @@ obj- += dummy.o
 
 ifndef CONFIG_SPL_BUILD
 obj-$(CONFIG_CMD_DDR3) += ddr3.o
+obj-$(CONFIG_CMD_PD) += pd.o
 endif
cmd/ti/pd.c new file +185 -0
new file mode 100644
index 0000000000..a53ccdcc40
@@ -0,0 +1,185 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Power Domain test commands
+ *
+ * Copyright (C) 2020 Texas Instruments Incorporated, <www.ti.com>
+ */
+
+#include <common.h>
+#include <command.h>
+#include <dm.h>
+#include <k3-dev.h>
+
+static const struct udevice_id ti_pd_of_match[] = {
+	{ .compatible = "ti,sci-pm-domain" },
+	{ /* sentinel */ }
+};
+
+static struct ti_k3_pd_platdata *ti_pd_find_data(void)
+{
+	struct udevice *dev;
+	int i = 0;
+
+	while (1) {
+		uclass_get_device(UCLASS_POWER_DOMAIN, i++, &dev);
+		if (!dev)
+			return NULL;
+
+		if (device_is_compatible(dev,
+					 ti_pd_of_match[0].compatible))
+			return  dev_get_priv(dev);
+	}
+
+	return NULL;
+}
+
+static void dump_lpsc(struct ti_k3_pd_platdata *data, struct ti_pd *pd)
+{
+	int i;
+	struct ti_lpsc *lpsc;
+	u8 state;
+	static const char * const lpsc_states[] = {
+		"swrstdis", "syncrst", "disable", "enable", "autosleep",
+		"autowake", "unknown",
+	};
+
+	for (i = 0; i < data->num_lpsc; i++) {
+		lpsc = &data->lpsc[i];
+		if (lpsc->pd != pd)
+			continue;
+		state = lpsc_get_state(lpsc);
+		if (state > ARRAY_SIZE(lpsc_states))
+			state = ARRAY_SIZE(lpsc_states) - 1;
+		printf("    LPSC%d: state=%s, usecount=%d\n",
+		       lpsc->id, lpsc_states[state], lpsc->usecount);
+	}
+}
+
+static void dump_pd(struct ti_k3_pd_platdata *data, struct ti_psc *psc)
+{
+	int i;
+	struct ti_pd *pd;
+	u8 state;
+	static const char * const pd_states[] = {
+		"off", "on", "unknown"
+	};
+
+	for (i = 0; i < data->num_pd; i++) {
+		pd = &data->pd[i];
+		if (pd->psc != psc)
+			continue;
+		state = ti_pd_state(pd);
+		if (state > ARRAY_SIZE(pd_states))
+			state = ARRAY_SIZE(pd_states) - 1;
+		printf("  PD%d: state=%s, usecount=%d:\n",
+		       pd->id, pd_states[state], pd->usecount);
+		dump_lpsc(data, pd);
+	}
+}
+
+static void dump_psc(struct ti_k3_pd_platdata *data)
+{
+	int i;
+	struct ti_psc *psc;
+
+	for (i = 0; i < data->num_psc; i++) {
+		psc = &data->psc[i];
+		printf("PSC%d [%p]:\n", psc->id, psc->base);
+		dump_pd(data, psc);
+	}
+}
+
+static int do_pd_dump(struct cmd_tbl *cmdtp, int flag, int argc,
+		      char *const argv[])
+{
+	struct ti_k3_pd_platdata *data;
+
+	data = ti_pd_find_data();
+	if (!data)
+		return CMD_RET_FAILURE;
+
+	dump_psc(data);
+
+	return 0;
+}
+
+static int do_pd_endis(int argc, char *const argv[], u8 state)
+{
+	u32 psc_id;
+	u32 lpsc_id;
+	int i;
+	struct ti_k3_pd_platdata *data;
+	struct ti_lpsc *lpsc;
+	int ret;
+
+	if (argc < 3)
+		return CMD_RET_FAILURE;
+
+	data = ti_pd_find_data();
+	if (!data)
+		return CMD_RET_FAILURE;
+
+	psc_id = simple_strtoul(argv[1], NULL, 10);
+	lpsc_id = simple_strtoul(argv[2], NULL, 10);
+
+	for (i = 0; i < data->num_lpsc; i++) {
+		lpsc = &data->lpsc[i];
+		if (lpsc->pd->psc->id != psc_id)
+			continue;
+		if (lpsc->id != lpsc_id)
+			continue;
+		printf("%s pd [PSC:%d,LPSC:%d]...\n",
+		       state == MDSTAT_STATE_ENABLE ? "Enabling" : "Disabling",
+		       psc_id, lpsc_id);
+		ret = ti_lpsc_transition(lpsc, state);
+		if (ret)
+			return CMD_RET_FAILURE;
+		else
+			return 0;
+	}
+
+	printf("No matching psc/lpsc found.\n");
+
+	return CMD_RET_FAILURE;
+}
+
+static int do_pd_enable(struct cmd_tbl *cmdtp, int flag, int argc,
+			char *const argv[])
+{
+	return do_pd_endis(argc, argv, MDSTAT_STATE_ENABLE);
+}
+
+static int do_pd_disable(struct cmd_tbl *cmdtp, int flag, int argc,
+			 char *const argv[])
+{
+	return do_pd_endis(argc, argv, MDSTAT_STATE_SWRSTDISABLE);
+}
+
+static struct cmd_tbl cmd_pd[] = {
+	U_BOOT_CMD_MKENT(dump, 1, 0, do_pd_dump, "", ""),
+	U_BOOT_CMD_MKENT(enable, 3, 0, do_pd_enable, "", ""),
+	U_BOOT_CMD_MKENT(disable, 3, 0, do_pd_disable, "", ""),
+};
+
+static int ti_do_pd(struct cmd_tbl *cmdtp, int flag, int argc, char * const argv[])
+{
+	struct cmd_tbl *c;
+
+	argc--;
+	argv++;
+
+	c = find_cmd_tbl(argv[0], cmd_pd, ARRAY_SIZE(cmd_pd));
+	if (c)
+		return c->cmd(cmdtp, flag, argc, argv);
+	else
+		return CMD_RET_USAGE;
+}
+
+U_BOOT_CMD(pd, 4, 1, ti_do_pd,
+	   "TI power domain control",
+#ifdef CONFIG_SYS_LONGHELP
+	   "dump                 - show power domain status\n"
+	   "enable [psc] [lpsc]  - enable power domain\n"
+	   "disable [psc] [lpsc] - disable power domain\n"
+#endif
+);
drivers/power/domain/ti-power-domain.c +3 -3
index e418a7b996..567e7e1e8e 100644
@@ -132,7 +132,7 @@ static void ti_pd_transition(struct ti_pd *pd)
 	psc_write(BIT(pd->id), pd->psc, PSC_PTCMD);
 }
 
-static u8 ti_pd_state(struct ti_pd *pd)
+u8 ti_pd_state(struct ti_pd *pd)
 {
 	return pd_read(pd, PSC_PDCTL) & PDCTL_STATE_MASK;
 }
@@ -227,12 +227,12 @@ static int ti_lpsc_wait(struct ti_lpsc *lpsc)
 	return ret;
 }
 
-static u8 lpsc_get_state(struct ti_lpsc *lpsc)
+u8 lpsc_get_state(struct ti_lpsc *lpsc)
 {
 	return lpsc_read(lpsc, PSC_MDCTL) & MDSTAT_STATE_MASK;
 }
 
-static int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state)
+int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state)
 {
 	struct ti_pd *psc_pd;
 	int ret;
include/k3-dev.h +9 -0
index de3a8bdf9e..55c5057db3 100644
@@ -22,6 +22,11 @@
 #define PSC_PD_ALWAYSON         BIT(1)
 #define PSC_PD_DEPENDS          BIT(2)
 
+#define MDSTAT_STATE_MASK		0x3f
+#define MDSTAT_BUSY_MASK		0x30
+#define MDSTAT_STATE_SWRSTDISABLE	0x0
+#define MDSTAT_STATE_ENABLE		0x3
+
 struct ti_psc {
 	int id;
 	void __iomem *base;
@@ -73,4 +78,8 @@ struct ti_k3_pd_platdata {
 extern const struct ti_k3_pd_platdata j721e_pd_platdata;
 extern const struct ti_k3_pd_platdata j7200_pd_platdata;
 
+u8 ti_pd_state(struct ti_pd *pd);
+u8 lpsc_get_state(struct ti_lpsc *lpsc);
+int ti_lpsc_transition(struct ti_lpsc *lpsc, u8 state);
+
 #endif
author maintainer
#18 lore raw

PATCH v4 16/26 tools: k3_fit_atf: add DM binary to the FIT image

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add DM (device manager) firmware image to the fit image that is loaded by R5 SPL. This is needed with the HSM rearch where the firmware allocation has been changed slightly.

---
 arch/arm/mach-k3/config.mk |  4 ++++
 tools/k3_fit_atf.sh        | 19 ++++++++++++++++++-
 2 files changed, 22 insertions(+), 1 deletion(-)
arch/arm/mach-k3/config.mk +4 -0
index 41fee2b5a1..503ece4520 100644
@@ -49,6 +49,10 @@ endif
 
 ifdef CONFIG_ARM64
 
+ifeq ($(CONFIG_SOC_K3_J721E),)
+export DM := /dev/null
+endif
+
 ifeq ($(CONFIG_TI_SECURE_DEVICE),y)
 SPL_ITS := u-boot-spl-k3_HS.its
 $(SPL_ITS): export IS_HS=1
tools/k3_fit_atf.sh +18 -1
index 4e9f69c087..c0940a2fcc 100755
@@ -21,6 +21,13 @@ if [ ! -f $TEE ]; then
 	TEE=/dev/null
 fi
 
+[ -z "$DM" ] && DM="dm.bin"
+
+if [ ! -e $DM ]; then
+	echo "WARNING DM file $DM NOT found, resulting might be non-functional" >&2
+	DM=/dev/null
+fi
+
 if [ ! -z "$IS_HS" ]; then
 	HS_APPEND=_HS
 fi
@@ -53,6 +60,16 @@ cat << __HEADER_EOF
 			load = <0x9e800000>;
 			entry = <0x9e800000>;
 		};
+		dm {
+			description = "DM binary";
+			data = /incbin/("$DM");
+			type = "firmware";
+			arch = "arm32";
+			compression = "none";
+			os = "DM";
+			load = <0xa0000000>;
+			entry = <0xa0000000>;
+		};
 		spl {
 			description = "SPL (64-bit)";
 			data = /incbin/("spl/u-boot-spl-nodtb.bin$HS_APPEND");
@@ -91,7 +108,7 @@ do
 		$(basename $dtname) {
 			description = "$(basename $dtname .dtb)";
 			firmware = "atf";
-			loadables = "tee", "spl";
+			loadables = "tee", "dm", "spl";
 			fdt = "$(basename $dtname)";
 		};
 __CONF_SECTION_EOF
Signature · 1 line
2.17.1
author maintainer
#19 lore raw

PATCH v4 17/26 arm: mach-k3: Add platform data for j721e and j7200

In reply to Tero Kristo

From: Dave Gerlach <[email protected]>

Add platform clock and powerdomain data for J721e and J7200. This data is used by the corresponding drivers to register all the required device clocks and powerdomains.

---
 arch/arm/mach-k3/Makefile         |   2 +-
 arch/arm/mach-k3/j7200/Makefile   |   5 +
 arch/arm/mach-k3/j7200/clk-data.c | 547 +++++++++++++++++++++
 arch/arm/mach-k3/j7200/dev-data.c |  77 +++
 arch/arm/mach-k3/j721e/Makefile   |   5 +
 arch/arm/mach-k3/j721e/clk-data.c | 781 ++++++++++++++++++++++++++++++
 arch/arm/mach-k3/j721e/dev-data.c |  75 +++
 7 files changed, 1491 insertions(+), 1 deletion(-)
 create mode 100644 arch/arm/mach-k3/j7200/Makefile
 create mode 100644 arch/arm/mach-k3/j7200/clk-data.c
 create mode 100644 arch/arm/mach-k3/j7200/dev-data.c
 create mode 100644 arch/arm/mach-k3/j721e/Makefile
 create mode 100644 arch/arm/mach-k3/j721e/clk-data.c
 create mode 100644 arch/arm/mach-k3/j721e/dev-data.c
arch/arm/mach-k3/Makefile +1 -1
index 7572f56925..534ddfcd49 100644
@@ -4,7 +4,7 @@
 #	Lokesh Vutla <[email protected]>
 
 obj-$(CONFIG_SOC_K3_AM6) += am6_init.o
-obj-$(CONFIG_SOC_K3_J721E) += j721e_init.o
+obj-$(CONFIG_SOC_K3_J721E) += j721e_init.o j721e/ j7200/
 obj-$(CONFIG_ARM64) += arm64-mmu.o
 obj-$(CONFIG_CPU_V7R) += r5_mpu.o lowlevel_init.o
 obj-$(CONFIG_TI_SECURE_DEVICE) += security.o
arch/arm/mach-k3/j7200/Makefile new file +5 -0
new file mode 100644
index 0000000000..ff9abd78ea
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier:	GPL-2.0+
+#
+# Copyright (C) 2020 Texas Instruments Incorporated - http://www.ti.com/
+obj-y += clk-data.o
+obj-y += dev-data.o
arch/arm/mach-k3/j7200/clk-data.c new file +547 -0

547 lines not on this page · raw message

arch/arm/mach-k3/j7200/dev-data.c new file +77 -0

77 lines not on this page · raw message

arch/arm/mach-k3/j721e/Makefile new file +5 -0
new file mode 100644
index 0000000000..ff9abd78ea
@@ -0,0 +1,5 @@
+# SPDX-License-Identifier:	GPL-2.0+
+#
+# Copyright (C) 2020 Texas Instruments Incorporated - http://www.ti.com/
+obj-y += clk-data.o
+obj-y += dev-data.o
arch/arm/mach-k3/j721e/clk-data.c new file +781 -0

781 lines not on this page · raw message

arch/arm/mach-k3/j721e/dev-data.c new file +75 -0

75 lines not on this page · raw message

Signature · 1 line
2.17.1
author maintainer
#20 lore raw

PATCH v4 18/26 arm: mach-k3: add support for detecting firmware images from FIT

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Add callback routines for parsing the firmware info from FIT image, and use the data to boot up ATF and the MCU R5 firmware.

---
 arch/arm/mach-k3/common.c   | 80 +++++++++++++++++++++++++++++++++----
 arch/arm/mach-k3/common.h   |  1 +
 arch/arm/mach-k3/security.c |  3 +-
 3 files changed, 75 insertions(+), 9 deletions(-)
arch/arm/mach-k3/common.c +73 -7
index 9191f686f0..a5d6a10dc7 100644
@@ -28,6 +28,27 @@
 #include <elf.h>
 #include <soc.h>
 
+#ifdef CONFIG_SYS_K3_SPL_ATF
+enum {
+	IMAGE_ID_ATF,
+	IMAGE_ID_OPTEE,
+	IMAGE_ID_SPL,
+	IMAGE_ID_DM_FW,
+	IMAGE_AMT,
+};
+
+#ifdef CONFIG_SPL_FIT_IMAGE_POST_PROCESS
+static const char *image_os_match[IMAGE_AMT] = {
+	"arm-trusted-firmware",
+	"tee",
+	"U-Boot",
+	"DM",
+};
+#endif
+
+static struct image_info fit_image_info[IMAGE_AMT];
+#endif
+
 struct ti_sci_handle *get_ti_sci_handle(void)
 {
 	struct udevice *dev;
@@ -181,7 +202,7 @@ void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
 	typedef void __noreturn (*image_entry_noargs_t)(void);
 	struct ti_sci_handle *ti_sci = get_ti_sci_handle();
 	u32 loadaddr = 0;
-	int ret, size;
+	int ret, size = 0;
 
 	/* Release all the exclusive devices held by SPL before starting ATF */
 	ti_sci->ops.dev_ops.release_exclusive_devices(ti_sci);
@@ -192,15 +213,20 @@ void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
 
 	init_env();
 	start_non_linux_remote_cores();
-	size = load_firmware("name_mcur5f0_0fw", "addr_mcur5f0_0load",
-			     &loadaddr);
+	if (!fit_image_info[IMAGE_ID_DM_FW].image_start)
+		size = load_firmware("name_mcur5f0_0fw", "addr_mcur5f0_0load",
+				     &loadaddr);
 
 
 	/*
 	 * It is assumed that remoteproc device 1 is the corresponding
 	 * Cortex-A core which runs ATF. Make sure DT reflects the same.
 	 */
-	ret = rproc_load(1, spl_image->entry_point, 0x200);
+	if (!fit_image_info[IMAGE_ID_ATF].image_start)
+		fit_image_info[IMAGE_ID_ATF].image_start =
+			spl_image->entry_point;
+
+	ret = rproc_load(1, fit_image_info[IMAGE_ID_ATF].image_start, 0x200);
 	if (ret)
 		panic("%s: ATF failed to load on rproc (%d)\n", __func__, ret);
 
@@ -210,7 +236,8 @@ void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
 	ret = rproc_start(1);
 	if (ret)
 		panic("%s: ATF failed to start on rproc (%d)\n", __func__, ret);
-	if (!(size > 0 && valid_elf_image(loadaddr))) {
+	if (!fit_image_info[IMAGE_ID_DM_FW].image_len &&
+	    !(size > 0 && valid_elf_image(loadaddr))) {
 		debug("Shutting down...\n");
 		release_resources_for_core_shutdown();
 
@@ -218,13 +245,52 @@ void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
 			asm volatile("wfe");
 	}
 
-	image_entry_noargs_t image_entry =
-		(image_entry_noargs_t)load_elf_image_phdr(loadaddr);
+	if (!fit_image_info[IMAGE_ID_DM_FW].image_start) {
+		loadaddr = load_elf_image_phdr(loadaddr);
+	} else {
+		loadaddr = fit_image_info[IMAGE_ID_DM_FW].image_start;
+		if (valid_elf_image(loadaddr))
+			loadaddr = load_elf_image_phdr(loadaddr);
+	}
+
+	debug("%s: jumping to address %x\n", __func__, loadaddr);
+
+	image_entry_noargs_t image_entry = (image_entry_noargs_t)loadaddr;
 
 	image_entry();
 }
 #endif
 
+#ifdef CONFIG_SPL_FIT_IMAGE_POST_PROCESS
+void board_fit_image_post_process(const void *fit, int node, void **p_image,
+				  size_t *p_size)
+{
+	int len;
+	int i;
+	const char *os;
+	u32 addr;
+
+	os = fdt_getprop(fit, node, "os", &len);
+	addr = fdt_getprop_u32_default_node(fit, node, 0, "entry", -1);
+
+	debug("%s: processing image: addr=%x, size=%d, os=%s\n", __func__,
+	      addr, *p_size, os);
+
+	for (i = 0; i < IMAGE_AMT; i++) {
+		if (!strcmp(os, image_os_match[i])) {
+			fit_image_info[i].image_start = addr;
+			fit_image_info[i].image_len = *p_size;
+			debug("%s: matched image for ID %d\n", __func__, i);
+			break;
+		}
+	}
+
+#ifdef CONFIG_TI_SECURE_DEVICE
+	ti_secure_image_post_process(p_image, p_size);
+#endif
+}
+#endif
+
 #if defined(CONFIG_OF_LIBFDT)
 int fdt_fixup_msmc_ram(void *blob, char *parent_path, char *node_name)
 {
arch/arm/mach-k3/common.h +1 -0
index a6dbc7808b..f421ed1bb1 100644
@@ -28,3 +28,4 @@ void k3_sysfw_print_ver(void);
 void spl_enable_dcache(void);
 void mmr_unlock(phys_addr_t base, u32 partition);
 bool is_rom_loaded_sysfw(struct rom_extended_boot_data *data);
+void ti_secure_image_post_process(void **p_image, size_t *p_size);
arch/arm/mach-k3/security.c +1 -2
index 5b5ff9ba7b..8de9739a40 100644
@@ -18,8 +18,7 @@
 #include <spl.h>
 #include <asm/arch/sys_proto.h>
 
-void board_fit_image_post_process(const void *fit, int node, void **p_image,
-				  size_t *p_size)
+void ti_secure_image_post_process(void **p_image, size_t *p_size)
 {
 	struct ti_sci_handle *ti_sci = get_ti_sci_handle();
 	struct ti_sci_proc_ops *proc_ops = &ti_sci->ops.proc_ops;
Signature · 1 line
2.17.1
author maintainer
#21 lore raw

PATCH v4 19/26 arm: mach-k3: do board config for PM only if supported

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

If the raw PM support is built in, we are operating in the split firmware approach mode where PM support is not available. In this case, skip the board config for this.

---
 arch/arm/mach-k3/sysfw-loader.c | 2 ++
 1 file changed, 2 insertions(+)
arch/arm/mach-k3/sysfw-loader.c +2 -0
index 0bacfc4d07..c6af205163 100644
@@ -159,11 +159,13 @@ static void k3_sysfw_configure_using_fit(void *fit,
 		      ret);
 
 	/* Apply power/clock (PM) specific configuration to SYSFW */
+#ifndef CONFIG_K3_DM_FW
 	ret = board_ops->board_config_pm(ti_sci,
 					 (u64)(u32)cfg_fragment_addr,
 					 (u32)cfg_fragment_size);
 	if (ret)
 		panic("Failed to set board PM configuration (%d)\n", ret);
+#endif
 
 	/* Extract resource management (RM) specific configuration from FIT */
 	ret = fit_get_data_by_name(fit, images, SYSFW_CFG_RM,
Signature · 1 line
2.17.1
author maintainer
#22 lore raw

PATCH v4 20/26 arm: mach-k3: common: Drop main r5 start

In reply to Tero Kristo

From: Dave Gerlach <[email protected]>

Only start-up the non-linux remote cores if we are running in legacy boot mode. HSM rearch is not yet supporting this.

---
 arch/arm/mach-k3/common.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
arch/arm/mach-k3/common.c +4 -3
index a5d6a10dc7..2b8c09a650 100644
@@ -212,11 +212,12 @@ void __noreturn jump_to_image_no_args(struct spl_image_info *spl_image)
 		panic("rproc failed to be initialized (%d)\n", ret);
 
 	init_env();
-	start_non_linux_remote_cores();
-	if (!fit_image_info[IMAGE_ID_DM_FW].image_start)
+
+	if (!fit_image_info[IMAGE_ID_DM_FW].image_start) {
+		start_non_linux_remote_cores();
 		size = load_firmware("name_mcur5f0_0fw", "addr_mcur5f0_0load",
 				     &loadaddr);
-
+	}
 
 	/*
 	 * It is assumed that remoteproc device 1 is the corresponding
Signature · 1 line
2.17.1
author maintainer
#23 lore raw

PATCH v4 21/26 arm: mach-k3: sysfw-loader: pass boardcfg to sciserver

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Copy the contents of the board config loaded from sysfw.itb into an EXTBOOT shared memory buffer that gets passed to sciserver. This only needs to be done if EXTBOOT area has not been populated by ROM code yet.

---
 arch/arm/mach-k3/sysfw-loader.c | 103 ++++++++++++++++++++++++++++++++
 1 file changed, 103 insertions(+)
arch/arm/mach-k3/sysfw-loader.c +103 -0
index c6af205163..ac719cd46f 100644
@@ -40,6 +40,46 @@ DECLARE_GLOBAL_DATA_PTR;
  */
 #define K3_SYSTEM_CONTROLLER_RPROC_ID	0
 
+#define COMMON_HEADER_ADDRESS		0x41cffb00
+#define BOARDCFG_ADDRESS		0x41c80000
+
+#define COMP_TYPE_SBL_DATA		0x11
+#define DESC_TYPE_BOARDCFG_PM_INDEX	0x2
+#define DESC_TYPE_BOARDCFG_RM_INDEX	0x3
+
+#define BOARD_CONFIG_RM_DESC_TYPE	0x000c
+#define BOARD_CONFIG_PM_DESC_TYPE	0x000e
+
+struct extboot_comp {
+	u32 comp_type;
+	u32 boot_core;
+	u32 comp_opts;
+	u64 dest_addr;
+	u32 comp_size;
+};
+
+struct extboot_header {
+	u8 magic[8];
+	u32 num_comps;
+	struct extboot_comp comps[5];
+	u32 reserved;
+};
+
+struct bcfg_desc {
+	u16 type;
+	u16 offset;
+	u16 size;
+	u8 devgrp;
+	u8 reserved;
+} __packed;
+
+struct bcfg_header {
+	u8 num_elems;
+	u8 sw_rev;
+	struct bcfg_desc descs[4];
+	u16 reserved;
+} __packed;
+
 static bool sysfw_loaded;
 static void *sysfw_load_address;
 
@@ -131,6 +171,15 @@ static void k3_sysfw_configure_using_fit(void *fit,
 	const void *cfg_fragment_addr;
 	size_t cfg_fragment_size;
 	int ret;
+#ifdef CONFIG_K3_DM_FW
+	u8 *buf;
+	struct extboot_header *common_header;
+	struct bcfg_header *bcfg_header;
+	struct extboot_comp *comp;
+	struct bcfg_desc *desc;
+	u32 addr;
+	bool copy_bcfg = false;
+#endif
 
 	/* Find the node holding the images information */
 	images = fdt_path_offset(fit, FIT_IMAGES_PATH);
@@ -165,6 +214,46 @@ static void k3_sysfw_configure_using_fit(void *fit,
 					 (u32)cfg_fragment_size);
 	if (ret)
 		panic("Failed to set board PM configuration (%d)\n", ret);
+#else
+	/* Initialize shared memory boardconfig buffer */
+	buf = (u8 *)COMMON_HEADER_ADDRESS;
+	common_header = (struct extboot_header *)buf;
+
+	/* Check if we have a struct populated by ROM in memory already */
+	if (strcmp((char *)common_header->magic, "EXTBOOT"))
+		copy_bcfg = true;
+
+	if (copy_bcfg) {
+		strcpy((char *)common_header->magic, "EXTBOOT");
+		common_header->num_comps = 1;
+
+		comp = &common_header->comps[0];
+
+		comp->comp_type = COMP_TYPE_SBL_DATA;
+		comp->boot_core = 0x10;
+		comp->comp_opts = 0;
+		addr = (u32)BOARDCFG_ADDRESS;
+		comp->dest_addr = addr;
+		comp->comp_size = sizeof(*bcfg_header);
+
+		bcfg_header = (struct bcfg_header *)addr;
+
+		bcfg_header->num_elems = 2;
+		bcfg_header->sw_rev = 0;
+
+		desc = &bcfg_header->descs[0];
+
+		desc->type = BOARD_CONFIG_PM_DESC_TYPE;
+		desc->offset = sizeof(*bcfg_header);
+		desc->size = cfg_fragment_size;
+		comp->comp_size += desc->size;
+		desc->devgrp = 0;
+		desc->reserved = 0;
+		memcpy((u8 *)bcfg_header + desc->offset, cfg_fragment_addr,
+		       cfg_fragment_size);
+
+		bcfg_header->descs[1].offset = desc->offset + desc->size;
+	}
 #endif
 
 	/* Extract resource management (RM) specific configuration from FIT */
@@ -174,6 +263,20 @@ static void k3_sysfw_configure_using_fit(void *fit,
 		panic("Error accessing %s node in FIT (%d)\n", SYSFW_CFG_RM,
 		      ret);
 
+#ifdef CONFIG_K3_DM_FW
+	if (copy_bcfg) {
+		desc = &bcfg_header->descs[1];
+
+		desc->type = BOARD_CONFIG_RM_DESC_TYPE;
+		desc->size = cfg_fragment_size;
+		comp->comp_size += desc->size;
+		desc->devgrp = 0;
+		desc->reserved = 0;
+		memcpy((u8 *)bcfg_header + desc->offset, cfg_fragment_addr,
+		       cfg_fragment_size);
+	}
+#endif
+
 	/* Apply resource management (RM) configuration to SYSFW */
 	ret = board_ops->board_config_rm(ti_sci,
 					 (u64)(u32)cfg_fragment_addr,
Signature · 1 line
2.17.1
author maintainer
#24 lore raw

PATCH v4 22/26 arm: mach-k3: j721e_init: Force early probe of clk-k3 driver

In reply to Tero Kristo

From: Dave Gerlach <[email protected]>

Force the clk-k3 driver to probe early during R5 SPL boot to ensure the default system clock configuration is completed. Many other drivers assume a default state of the clock tree and it is currently possible for them to probe before clk-k3 depending on the exact system configuration.

---
 arch/arm/mach-k3/j721e_init.c | 12 ++++++++++++
 1 file changed, 12 insertions(+)
arch/arm/mach-k3/j721e_init.c +12 -0
index 1a4f796e5e..b5268372dd 100644
@@ -180,6 +180,18 @@ void board_init_f(ulong dummy)
 	k3_sysfw_loader(is_rom_loaded_sysfw(&bootdata),
 			k3_mmc_stop_clock, k3_mmc_restart_clock);
 
+#ifdef CONFIG_SPL_CLK_K3
+	/*
+	 * Force probe of clk_k3 driver here to ensure basic default clock
+	 * configuration is always done.
+	 */
+	ret = uclass_get_device_by_driver(UCLASS_CLK,
+					  DM_DRIVER_GET(ti_clk),
+					  &dev);
+	if (ret)
+		panic("Failed to initialize clk-k3!\n");
+#endif
+
 	/* Prepare console output */
 	preloader_console_init();
 
Signature · 1 line
2.17.1
author maintainer
#25 lore raw

PATCH v4 23/26 configs: j721e_evm_r5: Enable raw access power management features

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Sysfw is not going to provide access to power management features in the new architecture, so SPL must implement these itself. Enable all the raw register access based clock + power domain drivers.

---
 configs/j721e_evm_r5_defconfig | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
configs/j721e_evm_r5_defconfig +10 -2
index 7b4f0afce6..0d21904a00 100644
@@ -72,7 +72,7 @@ CONFIG_SPL_REGMAP=y
 CONFIG_SPL_OF_TRANSLATE=y
 CONFIG_CLK=y
 CONFIG_SPL_CLK=y
-CONFIG_CLK_TI_SCI=y
+# CONFIG_CLK_TI_SCI is not set
 CONFIG_DMA_CHANNELS=y
 CONFIG_TI_K3_NAVSS_UDMA=y
 CONFIG_TI_SCI_PROTOCOL=y
@@ -103,7 +103,7 @@ CONFIG_SPL_PINCTRL=y
 # CONFIG_SPL_PINCTRL_GENERIC is not set
 CONFIG_PINCTRL_SINGLE=y
 CONFIG_POWER_DOMAIN=y
-CONFIG_TI_SCI_POWER_DOMAIN=y
+# CONFIG_TI_SCI_POWER_DOMAIN is not set
 CONFIG_DM_PMIC=y
 CONFIG_PMIC_TPS65941=y
 CONFIG_DM_REGULATOR=y
@@ -141,3 +141,11 @@ CONFIG_USB_GADGET_PRODUCT_NUM=0x6163
 CONFIG_USB_GADGET_DOWNLOAD=y
 CONFIG_FS_EXT4=y
 CONFIG_FS_FAT_MAX_CLUSTSIZE=16384
+CONFIG_TI_POWER_DOMAIN=y
+CONFIG_SPL_CLK_CCF=y
+CONFIG_LIB_RATIONAL=y
+CONFIG_SPL_LIB_RATIONAL=y
+CONFIG_SPL_CLK_K3_PLL=y
+CONFIG_SPL_CLK_K3=y
+CONFIG_K3_DM_FW=y
+CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y
Signature · 1 line
2.17.1
author maintainer
#26 lore raw

PATCH v4 24/26 configs: j7200_evm_r5: Enable raw access power management features

In reply to Tero Kristo

From: Dave Gerlach <[email protected]>

Sysfw is not going to provide access to power management features in the new architecture, so SPL must implement these itself. Enable all the raw register access based clock + power domain drivers.

---
 configs/j7200_evm_r5_defconfig | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)
configs/j7200_evm_r5_defconfig +12 -2
index b20698fd0e..97eb1b3c9a 100644
@@ -75,7 +75,7 @@ CONFIG_SPL_SYSCON=y
 CONFIG_SPL_OF_TRANSLATE=y
 CONFIG_CLK=y
 CONFIG_SPL_CLK=y
-CONFIG_CLK_TI_SCI=y
+# CONFIG_CLK_TI_SCI is not set
 CONFIG_DMA_CHANNELS=y
 CONFIG_TI_K3_NAVSS_UDMA=y
 CONFIG_TI_SCI_PROTOCOL=y
@@ -111,7 +111,7 @@ CONFIG_SPL_PINCTRL=y
 # CONFIG_SPL_PINCTRL_GENERIC is not set
 CONFIG_PINCTRL_SINGLE=y
 CONFIG_POWER_DOMAIN=y
-CONFIG_TI_SCI_POWER_DOMAIN=y
+# CONFIG_TI_SCI_POWER_DOMAIN is not set
 CONFIG_K3_SYSTEM_CONTROLLER=y
 CONFIG_REMOTEPROC_TI_K3_ARM64=y
 CONFIG_DM_RESET=y
@@ -143,3 +143,13 @@ CONFIG_USB_GADGET_PRODUCT_NUM=0x6164
 CONFIG_USB_GADGET_DOWNLOAD=y
 CONFIG_FS_EXT4=y
 CONFIG_FS_FAT_MAX_CLUSTSIZE=16384
+CONFIG_SOC_DEVICE=y
+CONFIG_SOC_DEVICE_TI_K3=y
+CONFIG_TI_POWER_DOMAIN=y
+CONFIG_SPL_CLK_CCF=y
+CONFIG_LIB_RATIONAL=y
+CONFIG_SPL_LIB_RATIONAL=y
+CONFIG_SPL_CLK_K3_PLL=y
+CONFIG_SPL_CLK_K3=y
+CONFIG_K3_DM_FW=y
+CONFIG_SPL_FIT_IMAGE_POST_PROCESS=y
Signature · 1 line
2.17.1
author maintainer
#27 lore raw

PATCH v4 25/26 board: ti: j72xx: README: update build instructions and image formats

In reply to Tero Kristo

From: Tero Kristo <[email protected]>

Update build instructions and image formats based on HSM rearch. A new DM image is added into the build, which gets executed right after R5 SPL finishes its job.

---
 board/ti/j721e/README | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)
board/ti/j721e/README +6 -3
index c33afa496e..b1c9145c92 100644
@@ -73,12 +73,12 @@ support. Below is the pictorial representation of boot flow:
 |    |         |         |   +-------------+     |                       |                       |
 |    |         |<--------|---| Start A72   |     |                       |                       |
 |    |         |         |   | and jump to |     |                       |                       |
-|    |         |         |   | next image  |     |                       |                       |
+|    |         |         |   | DM fw image |     |                       |                       |
 |    |         |         |   +-------------+     |                       |                       |
 |    |         |         |                       |     +-----------+     |                       |
 |    |         |---------|-----------------------|---->| Reset rls |     |                       |
 |    |         |         |                       |     +-----------+     |                       |
-|    |  DMSC   |         |                       |          :            |                       |
+|    |  TIFS   |         |                       |          :            |                       |
 |    |Services |         |                       |     +-----------+     |                       |
 |    |         |<--------|-----------------------|---->|*ATF/OPTEE*|     |                       |
 |    |         |         |                       |     +-----------+     |                       |
@@ -154,7 +154,7 @@ $ make CROSS_COMPILE=arm-linux-gnueabihf- O=/tmp/r5
 
 4.2. A72:
 $ make CROSS_COMPILE=aarch64-linux-gnu- j721e_evm_a72_defconfig O=/tmp/a72
-$ make CROSS_COMPILE=aarch64-linux-gnu- ATF=<path to ATF dir>/build/k3/generic/release/bl31.bin TEE=<path to OPTEE OS dir>/out/arm-plat-k3/core/tee-pager_v2.bin O=/tmp/a72
+$ make CROSS_COMPILE=aarch64-linux-gnu- ATF=<path to ATF dir>/build/k3/generic/release/bl31.bin TEE=<path to OPTEE OS dir>/out/arm-plat-k3/core/tee-pager_v2.bin DM=<path to DM firmware image> O=/tmp/a72
 
 Target Images
 --------------
@@ -197,6 +197,9 @@ Image formats:
                 | |     A72 OPTEE     | |
                 | +-------------------+ |
                 | |                   | |
+                | |      R5 DM FW     | |
+                | +-------------------+ |
+                | |                   | |
                 | |      A72 SPL      | |
                 | +-------------------+ |
                 | |                   | |
Signature · 1 line
2.17.1
author maintainer
#28 lore raw

PATCH v4 26/26 arm: dts: k3-j72xx: correct MCU timer1 frequency

In reply to Tero Kristo

MCU timer1 is used as the tick timer for MCU R5 SPL, and the clock-frequency defined in DT appears to be incorrect at the moment. Actual clock source for the timer is MCU_SYSCLK0 / 4 which is 250MHz.

Earlier setup of 25MHz went unnoticed, as there was a separate issue with omap-timer, which caused an error to the clock by a factor of 8 with j7 devices. This problem surfaced once the omap-timer was fixed.

---
 arch/arm/dts/k3-j7200-common-proc-board-u-boot.dtsi | 2 +-
 arch/arm/dts/k3-j721e-common-proc-board-u-boot.dtsi | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
arch/arm/dts/k3-j7200-common-proc-board-u-boot.dtsi +1 -1
index bd037be350..c3aae65b39 100644
@@ -33,7 +33,7 @@
 		compatible = "ti,omap5430-timer";
 		reg = <0x0 0x40400000 0x0 0x80>;
 		ti,timer-alwon;
-		clock-frequency = <25000000>;
+		clock-frequency = <250000000>;
 		u-boot,dm-spl;
 	};
 
arch/arm/dts/k3-j721e-common-proc-board-u-boot.dtsi +1 -1
index 3384ed9f3a..1135de5a92 100644
@@ -46,7 +46,7 @@
 		compatible = "ti,omap5430-timer";
 		reg = <0x0 0x40400000 0x0 0x80>;
 		ti,timer-alwon;
-		clock-frequency = <25000000>;
+		clock-frequency = <250000000>;
 		u-boot,dm-spl;
 	};
 
Signature · 1 line
2.17.1
#29 lore raw
In reply to Tero Kristo

Hi Tero,

Quoted 9 lines · Tero Kristo

Hello,

Couple of small changes in v4:
- re-worked patch #14 to include review comments from Jaehoon Chung
* changed code to use iopoll version instead of hand crafted loops for timeout handling
* other mostly cosmetic changes
- patch #19/#21 changed to allow RM init to happen based on comment from Vignesh

Can you please rebase on top of latest u-boot and re-post?

Thanks and regards,
Lokesh

Quoted 4 lines

-Tero