Skip to contentSkip to conversation

dm: Add programmatic generation of ACPI tables (part A)

PATCH Superseded Simon Glass 6 participants 86 messages Mar 30, 2020 – Apr 15, 2020

Subsystems (6)
x86 Kbuild ACPI PCI +2 more subsystems

v3 of 8 Superseded Latest is v8 →

lore
Share & tools
lore
#0 lore raw

v3 cover dm: Add programmatic generation of ACPI tables (part A)

This is split from the original series in an attempt to get things applied in chunks.

This section includes patches up to and including the 'acpi' command.

Changes in v3:
- Add a pointer to information about acpi,compatible
- Add forward declarations for the functions
- Add missing error check in acpi_create_dmar()
- Add new patch to add hexdump.h to the unit test header
- Add new patch to move acpi_s3.h to include/acpi/
- Beef up the comment explaining how the unaligned address is used
- Change the example to ELAN
- Compute ACPI_NAME_MAX from ACPI_NAME_LEN
- Correct description of acpi,probed
- Drop 'Intel' from 'Intel ACPI'
- Drop acpi,name in example
- Drop duplicate assert
- Drop hid-descr-addr
- Drop mention of PRIC
- Fix 'RSDP' typo
- Fix 'XDST' typo
- Fix DMA_ typo
- Fix a few typos
- Fix file comment for acpi_table.c
- Fix indenting error mentioned by Andy Shevchenko
- Fix stray #endif
- Just add the device.txt binding file in this patch
- Make use of BIT()
- Move acpi_align_large() out of dm_test_acpi_setup_base_tables()
- Move acpi_table.h to include/acpi
- Rename acpi,desc to acpi,ddn
- Rename acpi_align_large() to acpi_align64()
- Rename acpi_return_name() to acpi_copy_name()
- Reword commit message to drop the bit about ACPI being complicated
- Split out hid-over-i2c into its own patch
- Update commit message to say that we move most of its contents
- Use strncpy() instead of strcpy() in acpi_copy_name()

Changes in v2:
- Add in the acpi_table.h header file to this patch
- Add the hid-over-i2c binding document
- Add trailing commas to enum
- Don't bracket the definitions with DM_SPI
- Drop definition of ACPI_TABLE_CREATOR
- Drop the Chrome OS pieces
- Drop the other comment change since it is already applied
- Drop two unnecessary __packed
- Fix definition of HID
- Generalise the ACPI function recursion with acpi_recurse_method()
- Infer hid-over-i2c CID value
- Make _acpi_write_dev_tables() static and switch argument order
- Move LOGC_ACPI definition to this patch
- Move __packed to after struct
- Move the sandbox acpi_table.h header file to an earlier patch
- Rename the 'coreboot' console to 'U-Boot'
- Use #defines for MADT and MCFG version numbers

Simon Glass (29):
cpu: Support querying the address width
spi: Add SPI mode enums
tpm: cr50: Release locality on exit
tpm: cr50: Add a comment for cr50_priv
tpm: cr50: Use the correct GPIO binding
tpm: Don't cleanup unless an error happens
dm: pci: Allow disabling auto-config for a device
x86: Correct wording of coreboot source code
x86: apl: Move p2sb ofdata reading to the correct method
pci: Adjust dm_pci_read_bar32() to return errors correctly
x86: apl: Add Global NVS table header
dm: core: Add basic ACPI support
dts: Add a binding for hid-over-i2c
acpi: Add a binding for ACPI settings in the device tree
acpi: Add a simple sandbox test
x86: Move acpi_s3.h to include/acpi/
x86: Move acpi_table header to main include/ directory
acpi: Add an __ACPI__ preprocessor symbol
acpi: Add a central location for table version numbers
acpi: Add support for DMAR
test: Add hexdump.h to the unit test header
acpi: Add a method to write tables for a device
acpi: Convert part of acpi_table to use acpi_ctx
x86: Allow devices to write ACPI tables
acpi: Drop code for missing XSDT from acpi_write_rsdp()
acpi: Move acpi_add_table() to generic code
acpi: Put table-setup code in its own function
acpi: Move the xsdt pointer to acpi_ctx
acpi: Add an acpi command

 arch/sandbox/dts/test.dts                     |   8 +
 arch/sandbox/include/asm/acpi_table.h         |   9 +
 arch/sandbox/include/asm/global_data.h        |   1 +
 arch/x86/cpu/apollolake/cpu_spl.c             |   2 +-
 arch/x86/cpu/apollolake/fsp_s.c               |   2 +-
 arch/x86/cpu/apollolake/pmc.c                 |   2 +-
 arch/x86/cpu/baytrail/acpi.c                  |   6 +-
 arch/x86/cpu/coreboot/timestamp.c             |   4 +-
 arch/x86/cpu/cpu.c                            |   4 +-
 arch/x86/cpu/intel_common/p2sb.c              |  33 +-
 arch/x86/cpu/quark/acpi.c                     |   2 +-
 arch/x86/cpu/tangier/acpi.c                   |   4 +-
 arch/x86/cpu/wakeup.S                         |   2 +-
 arch/x86/dts/chromebook_coral.dts             |   2 +-
 arch/x86/include/asm/acpi_table.h             | 381 +-----------
 .../include/asm/arch-apollolake/global_nvs.h  |  36 ++
 .../x86/include/asm/arch-coreboot/timestamp.h |   4 +-
 arch/x86/include/asm/global_data.h            |   1 +
 arch/x86/include/asm/intel_pinctrl_defs.h     |   2 -
 arch/x86/lib/acpi.c                           |   2 +-
 arch/x86/lib/acpi_s3.c                        |   4 +-
 arch/x86/lib/acpi_table.c                     | 240 ++------
 arch/x86/lib/coreboot_table.c                 |   2 +-
 arch/x86/lib/fsp/fsp_common.c                 |   2 +-
 arch/x86/lib/fsp1/fsp_common.c                |   2 +-
 arch/x86/lib/fsp2/fsp_dram.c                  |   2 +-
 arch/x86/lib/tables.c                         |   2 +-
 arch/x86/lib/zimage.c                         |   2 +-
 cmd/Kconfig                                   |  14 +
 cmd/Makefile                                  |   1 +
 cmd/acpi.c                                    | 179 ++++++
 doc/device-tree-bindings/device.txt           |  37 ++
 .../gpio/intel,apl-gpio.txt                   |   2 +-
 .../input/hid-over-i2c.txt                    |  44 ++
 .../interrupt-controller/intel,acpi-gpe.txt   |   2 +-
 doc/device-tree-bindings/pci/x86-pci.txt      |  23 +
 drivers/core/Kconfig                          |   9 +
 drivers/core/Makefile                         |   1 +
 drivers/core/acpi.c                           |  94 +++
 drivers/cpu/cpu_sandbox.c                     |   1 +
 drivers/pci/pci-uclass.c                      |  11 +-
 drivers/pci/pci_rom.c                         |   4 +-
 drivers/power/acpi_pmc/acpi-pmc-uclass.c      |   2 +-
 drivers/sysreset/sysreset_x86.c               |   2 +-
 drivers/tpm/cr50_i2c.c                        |  24 +-
 drivers/tpm/tpm-uclass.c                      |  13 +-
 include/{ => acpi}/acpi_s3.h                  |   0
 include/acpi/acpi_table.h                     | 581 ++++++++++++++++++
 include/cpu.h                                 |   2 +
 include/dm/acpi.h                             | 114 ++++
 include/dm/device.h                           |   5 +
 include/dm/uclass-id.h                        |   1 +
 include/log.h                                 |   2 +
 include/spi.h                                 |  33 +
 include/test/ut.h                             |   1 +
 lib/Makefile                                  |   1 +
 lib/acpi/Makefile                             |   4 +
 lib/acpi/acpi_table.c                         | 257 ++++++++
 lib/efi_loader/efi_acpi.c                     |   2 +-
 scripts/Makefile.lib                          |   4 +-
 test/dm/Makefile                              |   1 +
 test/dm/acpi.c                                | 317 ++++++++++
 test/dm/cpu.c                                 |   1 +
 63 files changed, 1915 insertions(+), 637 deletions(-)
 create mode 100644 arch/sandbox/include/asm/acpi_table.h
 create mode 100644 arch/x86/include/asm/arch-apollolake/global_nvs.h
 create mode 100644 cmd/acpi.c
 create mode 100644 doc/device-tree-bindings/device.txt
 create mode 100644 doc/device-tree-bindings/input/hid-over-i2c.txt
 create mode 100644 drivers/core/acpi.c
 rename include/{ => acpi}/acpi_s3.h (100%)
 create mode 100644 include/acpi/acpi_table.h
 create mode 100644 include/dm/acpi.h
 create mode 100644 lib/acpi/Makefile
 create mode 100644 lib/acpi/acpi_table.c
 create mode 100644 test/dm/acpi.c
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#1 lore raw

PATCH v3 1/29 cpu: Support querying the address width

Different CPUs may support different address widths, meaning the amount of memory they can address. Add a property for this to the cpu_info struct.

---

Changes in v3: None
Changes in v2: None

 drivers/cpu/cpu_sandbox.c | 1 +
 include/cpu.h             | 2 ++
 test/dm/cpu.c             | 1 +
 3 files changed, 4 insertions(+)
drivers/cpu/cpu_sandbox.c +1 -0
index ff87e8adca3..05b384f6a45 100644
@@ -19,6 +19,7 @@ int cpu_sandbox_get_info(struct udevice *dev, struct cpu_info *info)
 {
 	info->cpu_freq = 42 * 42 * 42 * 42 * 42;
 	info->features = 0x42424242;
+	info->address_width = IS_ENABLED(CONFIG_PHYS_64BIT) ? 64 : 32;
 
 	return 0;
 }
include/cpu.h +2 -0
index 28dd48feb8f..6b1b6b37b3b 100644
@@ -44,10 +44,12 @@ enum {
  *
  * @cpu_freq:	Current CPU frequency in Hz
  * @features:	Flags for supported CPU features
+ * @address_width:	Width of the CPU address space in bits (e.g. 32)
  */
 struct cpu_info {
 	ulong cpu_freq;
 	ulong features;
+	uint address_width;
 };
 
 struct cpu_ops {
test/dm/cpu.c +1 -0
index f5f1caef716..e6dc576ea3c 100644
@@ -33,6 +33,7 @@ static int dm_test_cpu(struct unit_test_state *uts)
 	ut_assertok(cpu_get_info(dev, &info));
 	ut_asserteq(info.cpu_freq, 42 * 42 * 42 * 42 * 42);
 	ut_asserteq(info.features, 0x42424242);
+	ut_asserteq(info.address_width, 32);
 
 	ut_asserteq(cpu_get_count(dev), 42);
 
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#2 lore raw

PATCH v3 2/29 spi: Add SPI mode enums

In reply to Simon Glass

With ACPI we need to describe the settings of the SPI bus. Add enums to handle this.

---

Changes in v3: None
Changes in v2:
- Add trailing commas to enum
- Don't bracket the definitions with DM_SPI

 include/spi.h | 33 +++++++++++++++++++++++++++++++++
 1 file changed, 33 insertions(+)
include/spi.h +33 -0
index 852f570eaa8..2b4929fc79d 100644
@@ -66,6 +66,39 @@ struct dm_spi_slave_platdata {
 
 #endif /* CONFIG_DM_SPI */
 
+/**
+ * enum spi_clock_phase - indicates  the clock phase to use for SPI (CPHA)
+ *
+ * @SPI_CLOCK_PHASE_FIRST: Data sampled on the first phase
+ * @SPI_CLOCK_PHASE_SECOND: Data sampled on the second phase
+ */
+enum spi_clock_phase {
+	SPI_CLOCK_PHASE_FIRST,
+	SPI_CLOCK_PHASE_SECOND,
+};
+
+/**
+ * enum spi_wire_mode - indicates the number of wires used for SPI
+ *
+ * @SPI_4_WIRE_MODE: Normal bidirectional mode with MOSI and MISO
+ * @SPI_3_WIRE_MODE: Unidirectional version with a single data line SISO
+ */
+enum spi_wire_mode {
+	SPI_4_WIRE_MODE,
+	SPI_3_WIRE_MODE,
+};
+
+/**
+ * enum spi_polarity - indicates the polarity of the SPI bus (CPOL)
+ *
+ * @SPI_POLARITY_LOW: Clock is low in idle state
+ * @SPI_POLARITY_HIGH: Clock is high in idle state
+ */
+enum spi_polarity {
+	SPI_POLARITY_LOW,
+	SPI_POLARITY_HIGH,
+};
+
 /**
  * struct spi_slave - Representation of a SPI slave
  *
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#3 lore raw

PATCH v3 3/29 tpm: cr50: Release locality on exit

In reply to Simon Glass

At present the cr50 driver claims the locality and does not release it for Linux. This causes problems. Fix this by tracking what is claimed, and adding a 'remove' method.

---

Changes in v3: None
Changes in v2: None

 drivers/tpm/cr50_i2c.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)
drivers/tpm/cr50_i2c.c +11 -2
index b904a7d426e..b30f55b40d3 100644
@@ -206,7 +206,7 @@ static int release_locality(struct udevice *dev, int force)
 		cr50_i2c_write(dev, addr, &buf, 1);
 	}
 
-	priv->locality = 0;
+	priv->locality = -1;
 
 	return 0;
 }
@@ -499,6 +499,7 @@ static int process_reset(struct udevice *dev)
 static int claim_locality(struct udevice *dev, int loc)
 {
 	const u8 mask = TPM_ACCESS_VALID | TPM_ACCESS_ACTIVE_LOCALITY;
+	struct cr50_priv *priv = dev_get_priv(dev);
 	u8 access;
 	int ret;
 
@@ -525,6 +526,7 @@ static int claim_locality(struct udevice *dev, int loc)
 		return -EPERM;
 	}
 	log_info("Claimed locality %d\n", loc);
+	priv->locality = loc;
 
 	return 0;
 }
@@ -559,7 +561,11 @@ static int cr50_i2c_open(struct udevice *dev)
 
 static int cr50_i2c_cleanup(struct udevice *dev)
 {
-	release_locality(dev, 1);
+	struct cr50_priv *priv = dev_get_priv(dev);
+
+	printf("%s: cleanup %d\n", __func__, priv->locality);
+	if (priv->locality != -1)
+		release_locality(dev, 1);
 
 	return 0;
 }
@@ -631,6 +637,7 @@ static int cr50_i2c_probe(struct udevice *dev)
 		return log_msg_ret("vendor-id", -EXDEV);
 	}
 	priv->vendor = vendor;
+	priv->locality = -1;
 
 	return 0;
 }
@@ -655,5 +662,7 @@ U_BOOT_DRIVER(cr50_i2c) = {
 	.ops    = &cr50_i2c_ops,
 	.ofdata_to_platdata	= cr50_i2c_ofdata_to_platdata,
 	.probe	= cr50_i2c_probe,
+	.remove	= cr50_i2c_cleanup,
 	.priv_auto_alloc_size = sizeof(struct cr50_priv),
+	.flags		= DM_FLAG_OS_PREPARE,
 };
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#4 lore raw

PATCH v3 4/29 tpm: cr50: Add a comment for cr50_priv

In reply to Simon Glass

Add a comment for the private structure

---

Changes in v3: None
Changes in v2:
- Drop the other comment change since it is already applied

 drivers/tpm/cr50_i2c.c | 9 +++++++++
 1 file changed, 9 insertions(+)
drivers/tpm/cr50_i2c.c +9 -0
index b30f55b40d3..c1d2d2fa381 100644
@@ -34,6 +34,15 @@ enum {
 	CR50_MAX_BUF_SIZE = 63,
 };
 
+/**
+ * struct cr50_priv - Private driver data
+ *
+ * @ready_gpio: GPIO to use to check if the TPM is ready
+ * @irq: IRQ to use check if the TPM is ready (has priority over @ready_gpio)
+ * @locality: Currenttly claimed locality (-1 if none)
+ * @vendor: vendor: Vendor ID for TPM
+ * @use_irq: true to use @irq, false to use @ready if available
+ */
 struct cr50_priv {
 	struct gpio_desc ready_gpio;
 	struct irq irq;
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#5 lore raw

PATCH v3 5/29 tpm: cr50: Use the correct GPIO binding

In reply to Simon Glass

This device should use ready-gpios rather than ready-gpio. Fix it.

---

Changes in v3: None
Changes in v2: None

 arch/x86/dts/chromebook_coral.dts                               | 2 +-
 doc/device-tree-bindings/gpio/intel,apl-gpio.txt                | 2 +-
 .../interrupt-controller/intel,acpi-gpe.txt                     | 2 +-
 drivers/tpm/cr50_i2c.c                                          | 2 +-
 4 files changed, 4 insertions(+), 4 deletions(-)
arch/x86/dts/chromebook_coral.dts +1 -1
index af52e11c89a..d48ef3573ec 100644
@@ -292,7 +292,7 @@
 				reg = <0x50>;
 				compatible = "google,cr50";
 				u-boot,i2c-offset-len = <0>;
-				ready-gpio = <&gpio_n 28 GPIO_ACTIVE_LOW>;
+				ready-gpios = <&gpio_n 28 GPIO_ACTIVE_LOW>;
 				interrupts-extended = <&acpi_gpe 0x3c 0>;
 			};
 		};
doc/device-tree-bindings/gpio/intel,apl-gpio.txt +1 -1
index cf0659b70ec..8422ff63abd 100644
@@ -47,7 +47,7 @@ Example:
 			reg = <0x50>;
 			compatible = "google,cr50";
 			u-boot,i2c-offset-len = <0>;
-			ready-gpio = <&gpio_n GPIO_28 GPIO_ACTIVE_LOW>;
+			ready-gpios = <&gpio_n GPIO_28 GPIO_ACTIVE_LOW>;
 		};
 	};
 
doc/device-tree-bindings/interrupt-controller/intel,acpi-gpe.txt +1 -1
index d9252bf29f7..2fe02d8a227 100644
@@ -25,6 +25,6 @@ Example:
 	tpm at 50 {
 		reg = <0x50>;
 		compatible = "google,cr50";
-		ready-gpio = <&gpio_n 0x1c GPIO_ACTIVE_LOW>;
+		ready-gpios = <&gpio_n 0x1c GPIO_ACTIVE_LOW>;
 		interrupts-extended = <&acpi_gpe 0x3c 0>;
 	};
drivers/tpm/cr50_i2c.c +1 -1
index c1d2d2fa381..b67051af26a 100644
@@ -607,7 +607,7 @@ static int cr50_i2c_ofdata_to_platdata(struct udevice *dev)
 		priv->irq = irq;
 		priv->use_irq = true;
 	} else {
-		ret = gpio_request_by_name(dev, "ready-gpio", 0,
+		ret = gpio_request_by_name(dev, "ready-gpios", 0,
 					   &priv->ready_gpio, GPIOD_IS_IN);
 		if (ret) {
 			log_warning("Cr50 does not have an ready GPIO/interrupt (err=%d)\n",
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#6 lore raw

PATCH v3 6/29 tpm: Don't cleanup unless an error happens

In reply to Simon Glass

At present the cleanup() method is called on every transfer. It should only be called on failing transfers. Fix this and tidy up the error handling a little.

---

Changes in v3: None
Changes in v2: None

 drivers/tpm/tpm-uclass.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)
drivers/tpm/tpm-uclass.c +10 -3
index 1b11c93194d..71d58070062 100644
@@ -72,7 +72,7 @@ int tpm_xfer(struct udevice *dev, const uint8_t *sendbuf, size_t send_size,
 	struct tpm_ops *ops = tpm_get_ops(dev);
 	ulong start, stop;
 	uint count, ordinal;
-	int ret, ret2;
+	int ret, ret2 = 0;
 
 	if (ops->xfer)
 		return ops->xfer(dev, sendbuf, send_size, recvbuf, recv_size);
@@ -120,9 +120,16 @@ int tpm_xfer(struct udevice *dev, const uint8_t *sendbuf, size_t send_size,
 		}
 	} while (ret);
 
-	ret2 = ops->cleanup ? ops->cleanup(dev) : 0;
+	if (ret) {
+		if (ops->cleanup) {
+			ret2 = ops->cleanup(dev);
+			if (ret2)
+				return log_msg_ret("cleanup", ret2);
+		}
+		return log_msg_ret("xfer", ret);
+	}
 
-	return ret2 ? ret2 : ret;
+	return 0;
 }
 
 UCLASS_DRIVER(tpm) = {
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#7 lore raw

PATCH v3 7/29 dm: pci: Allow disabling auto-config for a device

In reply to Simon Glass

Add a means to avoid configuring a device when needed. Add an explanation of why this is useful to the binding file.

---

Changes in v3:
- Drop acpi,name in example

Changes in v2: None

 doc/device-tree-bindings/pci/x86-pci.txt | 23 +++++++++++++++++++++++
 drivers/pci/pci-uclass.c                 |  2 ++
 2 files changed, 25 insertions(+)
doc/device-tree-bindings/pci/x86-pci.txt +23 -0
index 3aa5bd9a46c..95e370b3e72 100644
@@ -10,6 +10,17 @@ Optional properties:
 	configuration in TPL/SPL to reduce code size and boot time, since these
 	phases only know about a small subset of PCI devices.
 
+For PCI devices the following optional property is available:
+
+- pci,no-autoconfig : Don't automatically configure this PCI device at all.
+	This is used when the device is statically configured and must maintain
+	this same config throughout the boot process. An example is a serial
+	UART being used to debug PCI configuration, since reconfiguring it stops
+	the UART from working until the driver is re-probed, and this can cause
+	output to be lost. This should not generally be used in production code,
+	although it is often harmless.
+
+
 Example:
 
 pci {
@@ -21,4 +32,16 @@ pci {
 		0x42000000 0x0 0xb0000000 0xb0000000 0 0x10000000
 		0x01000000 0x0 0x1000 0x1000 0 0xefff>;
 	u-boot,skip-auto-config-until-reloc;
+
+
+	serial: serial at 18,2 {
+		reg = <0x0200c210 0 0 0 0>;
+		u-boot,dm-pre-reloc;
+		compatible = "intel,apl-ns16550";
+		early-regs = <0xde000000 0x20>;
+		reg-shift = <2>;
+		clock-frequency = <1843200>;
+		current-speed = <115200>;
+		pci,no-autoconfig;
+	};
 };
drivers/pci/pci-uclass.c +2 -0
index e2882e3b634..ceb64517047 100644
@@ -536,6 +536,8 @@ int pci_auto_config_devices(struct udevice *bus)
 		int ret;
 
 		debug("%s: device %s\n", __func__, dev->name);
+		if (dev_read_bool(dev, "pci,no-autoconfig"))
+			continue;
 		ret = dm_pciauto_config_device(dev);
 		if (ret < 0)
 			return ret;
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#8 lore raw

PATCH v3 8/29 x86: Correct wording of coreboot source code

In reply to Simon Glass

Some files are taken or modified from coreboot, but the files are no-longer part of the coreboot project. Fix the wording in a few places.

---

Changes in v3: None
Changes in v2: None

 arch/x86/cpu/coreboot/timestamp.c              | 4 ++--
 arch/x86/include/asm/arch-coreboot/timestamp.h | 4 ++--
 arch/x86/include/asm/intel_pinctrl_defs.h      | 2 --
 3 files changed, 4 insertions(+), 6 deletions(-)
arch/x86/cpu/coreboot/timestamp.c +2 -2
index e698200d704..e8ccaf22129 100644
@@ -1,8 +1,8 @@
 // SPDX-License-Identifier: GPL-2.0+
 /*
- * This file is part of the coreboot project.
- *
  * Copyright (C) 2011 The ChromiumOS Authors.  All rights reserved.
+ *
+ * Modified from the coreboot version
  */
 
 #include <common.h>
arch/x86/include/asm/arch-coreboot/timestamp.h +2 -2
index 9320afba561..85d42c02c46 100644
@@ -1,8 +1,8 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
- * This file is part of the coreboot project.
- *
  * Copyright (C) 2011 The ChromiumOS Authors.  All rights reserved.
+ *
+ * Taken from the coreboot version
  */
 
 #ifndef __COREBOOT_TIMESTAMP_H__
arch/x86/include/asm/intel_pinctrl_defs.h +0 -2
index 6da06bb52b4..1ea141f082f 100644
@@ -1,7 +1,5 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
- * This file is part of the coreboot project.
- *
  * Copyright (C) 2015-2016 Intel Corp.
  * Copyright 2019 Google LLC
  *
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#9 lore raw

PATCH v3 9/29 x86: apl: Move p2sb ofdata reading to the correct method

In reply to Simon Glass

With P2SB the initial BAR (base-address register) is set up by TPL and this is used unchanged right through U-Boot.

At present the reading of this address is split between the ofdata() and probe() methods. There are a few problems that are unique to the p2sb. One is that its children need to call pcr_read32(), etc. which needs to have the p2sb address correct. Also some of its children are pinctrl devices and pinctrl is used when any device is probed. So p2sb really needs to get its base address set up in ofdata_to_platdata(), before it is probed.

Another point is that reading the p2sb BAR will not work if the p2sb is hidden. The FSP-S seems to hide it, presumably to avoid confusing PCI enumeration.

Reading ofdata in ofdata_to_platdata() is the correct place anyway, so this is easy to fix.

Move the code into one place and use the early-regs property in all cases for simplicity and to avoid needing to probe any PCI devices just to read the BAR.

---

Changes in v3:
- Fix indenting error mentioned by Andy Shevchenko

Changes in v2: None

 arch/x86/cpu/intel_common/p2sb.c | 33 +++++++++++---------------------
 1 file changed, 11 insertions(+), 22 deletions(-)
arch/x86/cpu/intel_common/p2sb.c +11 -22
index d5b4846e0a2..6f3c4416186 100644
@@ -92,46 +92,35 @@ int p2sb_ofdata_to_platdata(struct udevice *dev)
 
 #if !CONFIG_IS_ENABLED(OF_PLATDATA)
 	int ret;
+	u32 base[2];
 
+	ret = dev_read_u32_array(dev, "early-regs", base, ARRAY_SIZE(base));
+	if (ret)
+		return log_msg_ret("Missing/short early-regs", ret);
+	plat->mmio_base = base[0];
+	/* TPL sets up the initial BAR */
 	if (spl_phase() == PHASE_TPL) {
-		u32 base[2];
-
-		/* TPL sets up the initial BAR */
-		ret = dev_read_u32_array(dev, "early-regs", base,
-					 ARRAY_SIZE(base));
-		if (ret)
-			return log_msg_ret("Missing/short early-regs", ret);
-		plat->mmio_base = base[0];
 		plat->bdf = pci_get_devfn(dev);
 		if (plat->bdf < 0)
 			return log_msg_ret("Cannot get p2sb PCI address",
 					   plat->bdf);
 	}
+	upriv->mmio_base = plat->mmio_base;
 #else
 	plat->mmio_base = plat->dtplat.early_regs[0];
 	plat->bdf = pci_ofplat_get_devfn(plat->dtplat.reg[0]);
-#endif
 	upriv->mmio_base = plat->mmio_base;
-	debug("p2sb: mmio_base=%x\n", (uint)plat->mmio_base);
+#endif
 
 	return 0;
 }
 
 static int p2sb_probe(struct udevice *dev)
 {
-	if (spl_phase() == PHASE_TPL) {
+	if (spl_phase() == PHASE_TPL)
 		return p2sb_early_init(dev);
-	} else {
-		struct p2sb_platdata *plat = dev_get_platdata(dev);
-
-		plat->mmio_base = dev_read_addr_pci(dev);
-		/* Don't set BDF since it should not be used */
-		if (!plat->mmio_base || plat->mmio_base == FDT_ADDR_T_NONE)
-			return -EINVAL;
-
-		if (spl_phase() == PHASE_SPL)
-			return p2sb_spl_init(dev);
-	}
+	else if (spl_phase() == PHASE_SPL)
+		return p2sb_spl_init(dev);
 
 	return 0;
 }
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#10 lore raw

PATCH v3 10/29 pci: Adjust dm_pci_read_bar32() to return errors correctly

In reply to Simon Glass

At present if reading a BAR returns 0xffffffff (e.g. the device is not present) then the value is masked and a different value is returned. This makes it harder to detect the problem when debugging.

Update the function to avoid masking in this case.

---

Changes in v3: None
Changes in v2: None

 drivers/pci/pci-uclass.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/pci/pci-uclass.c +8 -1
index ceb64517047..d2e10d6868a 100644
@@ -1213,7 +1213,14 @@ u32 dm_pci_read_bar32(const struct udevice *dev, int barnum)
 
 	bar = PCI_BASE_ADDRESS_0 + barnum * 4;
 	dm_pci_read_config32(dev, bar, &addr);
-	if (addr & PCI_BASE_ADDRESS_SPACE_IO)
+
+	/*
+	 * If we get an invalid address, return this so that comparisons with
+	 * FDT_ADDR_T_NONE work correctly
+	 */
+	if (addr == 0xffffffff)
+		return addr;
+	else if (addr & PCI_BASE_ADDRESS_SPACE_IO)
 		return addr & PCI_BASE_ADDRESS_IO_MASK;
 	else
 		return addr & PCI_BASE_ADDRESS_MEM_MASK;
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#11 lore raw

PATCH v3 10/29 pci: Adjust dm_pci_read_bar32() to return errors correctly

Quoted 3 lines · Simon Glass

At present if reading a BAR returns 0xffffffff (e.g. the device is not present) then the value is masked and a different value is returned. This makes it harder to detect the problem when debugging.

The above ('the device is not present') is actually not correct. BAR is not mandatory register and detection is described in PCI spec.

To get device presence one may have check Vendor ID / Device ID pair rather then BAR.

Quoted 36 lines

Update the function to avoid masking in this case.

---

Changes in v3: None
Changes in v2: None

 drivers/pci/pci-uclass.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/pci/pci-uclass.c +8 -1
index ceb64517047..d2e10d6868a 100644
@@ -1213,7 +1213,14 @@ u32 dm_pci_read_bar32(const struct udevice *dev, int barnum)
 
 	bar = PCI_BASE_ADDRESS_0 + barnum * 4;
 	dm_pci_read_config32(dev, bar, &addr);
-	if (addr & PCI_BASE_ADDRESS_SPACE_IO)
+
+	/*
+	 * If we get an invalid address, return this so that comparisons with
+	 * FDT_ADDR_T_NONE work correctly
+	 */
+	if (addr == 0xffffffff)
+		return addr;
+	else if (addr & PCI_BASE_ADDRESS_SPACE_IO)
 		return addr & PCI_BASE_ADDRESS_IO_MASK;
 	else
 		return addr & PCI_BASE_ADDRESS_MEM_MASK;
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#12 lore raw

PATCH v3 10/29 pci: Adjust dm_pci_read_bar32() to return errors correctly

Hi Andy,

Quoted 8 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

At present if reading a BAR returns 0xffffffff (e.g. the device is not present) then the value is masked and a different value is returned. This makes it harder to detect the problem when debugging.

The above ('the device is not present') is actually not correct. BAR is not mandatory register and detection is described in PCI spec.

What change are you suggesting here? I suggest 'not present' as an example of why this might happen.

Quoted 36 lines

To get device presence one may have check Vendor ID / Device ID pair rather then BAR.

Quoted 32 lines

Update the function to avoid masking in this case.

---

Changes in v3: None
Changes in v2: None

 drivers/pci/pci-uclass.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/pci/pci-uclass.c +8 -1
index ceb64517047..d2e10d6868a 100644
@@ -1213,7 +1213,14 @@ u32 dm_pci_read_bar32(const struct udevice *dev, int barnum)
 
      bar = PCI_BASE_ADDRESS_0 + barnum * 4;
      dm_pci_read_config32(dev, bar, &addr);
-     if (addr & PCI_BASE_ADDRESS_SPACE_IO)
+
+     /*
+      * If we get an invalid address, return this so that comparisons with
+      * FDT_ADDR_T_NONE work correctly
+      */
+     if (addr == 0xffffffff)
+             return addr;
+     else if (addr & PCI_BASE_ADDRESS_SPACE_IO)
              return addr & PCI_BASE_ADDRESS_IO_MASK;
      else
Quoted 10 lines
Quoted 4 lines
drivers/pci/pci-uclass.c +0 -1
              return addr & PCI_BASE_ADDRESS_MEM_MASK;
--

2.26.0.rc2.310.g2932bb562d-goog

--
With Best Regards,
Andy Shevchenko

Regards,
Simon

#13 lore raw

PATCH v3 10/29 pci: Adjust dm_pci_read_bar32() to return errors correctly

Quoted 10 lines · Simon Glass

Hi Andy,

Quoted 6 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

At present if reading a BAR returns 0xffffffff (e.g. the device is not present) then the value is masked and a different value is returned. This makes it harder to detect the problem when debugging.

Quoted 5 lines
Quoted 2 lines

The above ('the device is not present') is actually not correct. BAR is not mandatory register and detection is described in PCI spec.

What change are you suggesting here? I suggest 'not present' as an example of why this might happen.

I suggest to follow PCI spec.
Thus, the code below is fragile and working by luck.

Quoted 48 lines
Quoted 35 lines

To get device presence one may have check Vendor ID / Device ID pair rather then BAR.

Quoted 32 lines

Update the function to avoid masking in this case.

---

Changes in v3: None
Changes in v2: None

 drivers/pci/pci-uclass.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/pci/pci-uclass.c +8 -1
index ceb64517047..d2e10d6868a 100644
@@ -1213,7 +1213,14 @@ u32 dm_pci_read_bar32(const struct udevice *dev, int barnum)
 
      bar = PCI_BASE_ADDRESS_0 + barnum * 4;
      dm_pci_read_config32(dev, bar, &addr);
-     if (addr & PCI_BASE_ADDRESS_SPACE_IO)
+
+     /*
+      * If we get an invalid address, return this so that comparisons with
+      * FDT_ADDR_T_NONE work correctly
+      */
+     if (addr == 0xffffffff)
+             return addr;
+     else if (addr & PCI_BASE_ADDRESS_SPACE_IO)
              return addr & PCI_BASE_ADDRESS_IO_MASK;
      else
Quoted 10 lines
Quoted 4 lines
drivers/pci/pci-uclass.c +0 -1
              return addr & PCI_BASE_ADDRESS_MEM_MASK;
--

2.26.0.rc2.310.g2932bb562d-goog

--
With Best Regards,
Andy Shevchenko

Regards,
Simon

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#14 lore raw

PATCH v3 10/29 pci: Adjust dm_pci_read_bar32() to return errors correctly

HI Andy,

Quoted 12 lines · Andy Shevchenko
Quoted 10 lines · Simon Glass

Hi Andy,

Quoted 6 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

At present if reading a BAR returns 0xffffffff (e.g. the device is not present) then the value is masked and a different value is returned. This makes it harder to detect the problem when debugging.

Quoted 8 lines
Quoted 5 lines
Quoted 2 lines

The above ('the device is not present') is actually not correct. BAR is not mandatory register and detection is described in PCI spec.

What change are you suggesting here? I suggest 'not present' as an example of why this might happen.

I suggest to follow PCI spec.
Thus, the code below is fragile and working by luck.

I don't know what you are suggesting. This allows an error to be reported in the common case and helps people discover mistakes in the driver flow. What is your suggestion?

Rgards,
SImon

Quoted 55 lines
Quoted 47 lines
Quoted 36 lines

To get device presence one may have check Vendor ID / Device ID pair rather then BAR.

Quoted 32 lines

Update the function to avoid masking in this case.

---

Changes in v3: None
Changes in v2: None

 drivers/pci/pci-uclass.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
drivers/pci/pci-uclass.c +8 -1
index ceb64517047..d2e10d6868a 100644
@@ -1213,7 +1213,14 @@ u32 dm_pci_read_bar32(const struct udevice *dev, int barnum)
 
      bar = PCI_BASE_ADDRESS_0 + barnum * 4;
      dm_pci_read_config32(dev, bar, &addr);
-     if (addr & PCI_BASE_ADDRESS_SPACE_IO)
+
+     /*
+      * If we get an invalid address, return this so that comparisons with
+      * FDT_ADDR_T_NONE work correctly
+      */
+     if (addr == 0xffffffff)
+             return addr;
+     else if (addr & PCI_BASE_ADDRESS_SPACE_IO)
              return addr & PCI_BASE_ADDRESS_IO_MASK;
      else
Quoted 10 lines
Quoted 4 lines
drivers/pci/pci-uclass.c +0 -1
              return addr & PCI_BASE_ADDRESS_MEM_MASK;
--

2.26.0.rc2.310.g2932bb562d-goog

--
With Best Regards,
Andy Shevchenko

Quoted 2 lines

Regards,
Simon

--
With Best Regards,
Andy Shevchenko

#15 lore raw

PATCH v3 11/29 x86: apl: Add Global NVS table header

In reply to Simon Glass

Add the C version of this header. It includes a few Chrome OS bits which are disabled for a normal build.

---

Changes in v3:
- Fix stray #endif

Changes in v2:
- Drop the Chrome OS pieces
- Rename the 'coreboot' console to 'U-Boot'

 .../include/asm/arch-apollolake/global_nvs.h  | 36 +++++++++++++++++++
 1 file changed, 36 insertions(+)
 create mode 100644 arch/x86/include/asm/arch-apollolake/global_nvs.h
arch/x86/include/asm/arch-apollolake/global_nvs.h new file +36 -0
new file mode 100644
index 00000000000..5bde837a755
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2015-2017 Intel Corp.
+ * (Written by Lance Zhao <[email protected]> for Intel Corp.)
+ * Copyright Google LLC 2019
+ *
+ * Modified from coreboot apollolake/include/soc/nvs.h
+ */
+
+#ifndef _GLOBAL_NVS_H_
+#define _GLOBAL_NVS_H_
+
+struct __packed acpi_global_nvs {
+	/* Miscellaneous */
+	u8	pcnt; /* 0x00 - Processor Count */
+	u8	ppcm; /* 0x01 - Max PPC State */
+	u8	lids; /* 0x02 - LID State */
+	u8	pwrs; /* 0x03 - AC Power State */
+	u8	dpte; /* 0x04 - Enable DPTF */
+	u32	cbmc; /* 0x05 - 0x08 - U-Boot Console */
+	u64	pm1i; /* 0x09 - 0x10 - System Wake Source - PM1 Index */
+	u64	gpei; /* 0x11 - 0x18 - GPE Wake Source */
+	u64	nhla; /* 0x19 - 0x20 - NHLT Address */
+	u32	nhll; /* 0x21 - 0x24 - NHLT Length */
+	u32	prt0; /* 0x25 - 0x28 - PERST_0 Address */
+	u8	scdp; /* 0x29 - SD_CD GPIO portid */
+	u8	scdo; /* 0x2A - GPIO pad offset relative to the community */
+	u8	uior; /* 0x2B - UART debug controller init on S3 resume */
+	u8	ecps; /* 0x2C - SGX Enabled status */
+	u64	emna; /* 0x2D - 0x34 EPC base address */
+	u64	elng; /* 0x35 - 0x3C EPC Length */
+	u8	unused[195];
+	u8		unused2[0xf00];
+};
+
+#endif /* _GLOBAL_NVS_H_ */
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Reviewed-by on v3 11/29
#16 lore raw

Antwort: [PATCH v3 11/29] x86: apl: Add Global NVS table header

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 58 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>, "Bin Meng" <[email protected]>
Betreff: [PATCH v3 11/29] x86: apl: Add Global NVS table header

Add the C version of this header. It includes a few Chrome OS bits which
are disabled for a normal build.

---

Changes in v3:
- Fix stray #endif

Changes in v2:
- Drop the Chrome OS pieces
- Rename the 'coreboot' console to 'U-Boot'

.../include/asm/arch-apollolake/global_nvs.h | 36
+++++++++++++++++++

1 file changed, 36 insertions(+)

create mode 100644 arch/x86/include/asm/arch-apollolake/global_nvs.h

arch/x86/include/asm/arch-apollolake/global_nvs.h new file +23 -0
new file mode 100644
index 00000000000..5bde837a755
@@ -0,0 +1,36 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (C) 2015-2017 Intel Corp.
+ * (Written by Lance Zhao <[email protected]> for Intel Corp.)
+ * Copyright Google LLC 2019
+ *
+ * Modified from coreboot apollolake/include/soc/nvs.h
+ */
+
+#ifndef _GLOBAL_NVS_H_
+#define _GLOBAL_NVS_H_
+
+struct __packed acpi_global_nvs {
+	/* Miscellaneous */
+	u8	pcnt; /* 0x00 - Processor Count */
+	u8	ppcm; /* 0x01 - Max PPC State */
+	u8	lids; /* 0x02 - LID State */
+	u8	pwrs; /* 0x03 - AC Power State */
+	u8	dpte; /* 0x04 - Enable DPTF */
+	u32	cbmc; /* 0x05 - 0x08 - U-Boot Console */
+	u64	pm1i; /* 0x09 - 0x10 - System Wake Source - PM1 Index */
+	u64	gpei; /* 0x11 - 0x18 - GPE Wake Source */
+	u64	nhla; /* 0x19 - 0x20 - NHLT Address */

arch/x86/include/asm/arch-apollolake/global_nvs.h line 33

Quoted 10 lines
arch/x86/include/asm/arch-apollolake/global_nvs.h new file +10 -0
+	u32	nhll; /* 0x21 - 0x24 - NHLT Length */
+	u32	prt0; /* 0x25 - 0x28 - PERST_0 Address */
+	u8	scdp; /* 0x29 - SD_CD GPIO portid */
+	u8	scdo; /* 0x2A - GPIO pad offset relative to the community */
+	u8	uior; /* 0x2B - UART debug controller init on S3 resume */
+	u8	ecps; /* 0x2C - SGX Enabled status */
+	u64	emna; /* 0x2D - 0x34 EPC base address */
+	u64	elng; /* 0x35 - 0x3C EPC Length */
+	u8	unused[195];
+	u8		unused2[0xf00];

Nit 1: Something is still wrong with the indentation of unused2.

Nit 2: Could you please add comments on why the values 195 and 0xf00 were
chosen? I would assume 195 was selected so that unused2 starts on a 256-byte boundary? But that is only a guess.

Quoted 6 lines
arch/x86/include/asm/arch-apollolake/global_nvs.h new file +3 -0
+};
+
+#endif /* _GLOBAL_NVS_H_ */
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#17 lore raw

Antwort: [PATCH v3 11/29] x86: apl: Add Global NVS table header

Quoted 24 lines · Wolfgang Wallner
Quoted 11 lines
arch/x86/include/asm/arch-apollolake/global_nvs.h new file +11 -0
+struct __packed acpi_global_nvs {
+	/* Miscellaneous */
+	u8	pcnt; /* 0x00 - Processor Count */
+	u8	ppcm; /* 0x01 - Max PPC State */
+	u8	lids; /* 0x02 - LID State */
+	u8	pwrs; /* 0x03 - AC Power State */
+	u8	dpte; /* 0x04 - Enable DPTF */
+	u32	cbmc; /* 0x05 - 0x08 - U-Boot Console */
+	u64	pm1i; /* 0x09 - 0x10 - System Wake Source - PM1 Index */
+	u64	gpei; /* 0x11 - 0x18 - GPE Wake Source */
+	u64	nhla; /* 0x19 - 0x20 - NHLT Address */

arch/x86/include/asm/arch-apollolake/global_nvs.h line 33

Quoted 10 lines
arch/x86/include/asm/arch-apollolake/global_nvs.h new file +10 -0
+	u32	nhll; /* 0x21 - 0x24 - NHLT Length */
+	u32	prt0; /* 0x25 - 0x28 - PERST_0 Address */
+	u8	scdp; /* 0x29 - SD_CD GPIO portid */
+	u8	scdo; /* 0x2A - GPIO pad offset relative to the community */
+	u8	uior; /* 0x2B - UART debug controller init on S3 resume */
+	u8	ecps; /* 0x2C - SGX Enabled status */
+	u64	emna; /* 0x2D - 0x34 EPC base address */
+	u64	elng; /* 0x35 - 0x3C EPC Length */
+	u8	unused[195];
+	u8		unused2[0xf00];

Nit 1: Something is still wrong with the indentation of unused2.

Quoted 3 lines

Nit 2: Could you please add comments on why the values 195 and 0xf00 were
chosen? I would assume 195 was selected so that unused2 starts on a 256-byte boundary? But that is only a guess.

Better to calculate them (I mean 195). In Linux kernel, for instance, a trick is being used for that.

The rest is not simply boundary but rather 256 vs 4096 size of NVS. The previous (small one) is what being used by U-Boot. I don't remember what ACPI spec dictates about this (i.o.w. if it depends to the ACPI version in use).

Quoted 1 line
Quoted 1 line

+};

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#18 lore raw

PATCH v3 12/29 dm: core: Add basic ACPI support

In reply to Simon Glass

ACPI (Advanced Configuration and Power Interface) is a standard for specifying information about a platform. It is a little like device tree but the bindings are part of the specification and it supports an interpreted bytecode language.

Driver model does not use ACPI for U-Boot's configuration, but it is convenient to have it support generation of ACPI tables for passing to Linux, etc.

As a starting point, add an optional set of ACPI operations to each device. Initially only a single operation is available, to obtain the ACPI name for the device. More operations are added later.

Enable ACPI for sandbox to ensure build coverage and so that we can add tests.

---

Changes in v3:
- Compute ACPI_NAME_MAX from ACPI_NAME_LEN
- Drop 'Intel' from 'Intel ACPI'
- Rename acpi_return_name() to acpi_copy_name()
- Reword commit message to drop the bit about ACPI being complicated
- Use strncpy() instead of strcpy() in acpi_copy_name()

Changes in v2:
- Move LOGC_ACPI definition to this patch

 drivers/core/Kconfig  |  9 ++++++
 drivers/core/Makefile |  1 +
 drivers/core/acpi.c   | 33 +++++++++++++++++++
 include/dm/acpi.h     | 73 +++++++++++++++++++++++++++++++++++++++++++
 include/dm/device.h   |  5 +++
 include/log.h         |  2 ++
 6 files changed, 123 insertions(+)
 create mode 100644 drivers/core/acpi.c
 create mode 100644 include/dm/acpi.h
drivers/core/Kconfig +9 -0
index 3b95b5387b9..a3b03993423 100644
@@ -261,4 +261,13 @@ config DM_DEV_READ_INLINE
 	bool
 	default y if !OF_LIVE
 
+config ACPIGEN
+	bool "Support ACPI table generation in driver model"
+	default y if SANDBOX || GENERATE_ACPI_TABLE
+	help
+	  This option enables generation of ACPI tables using driver-model
+	  devices. It adds a new operation struct to each driver, to support
+	  things like generating device-specific tables and returning the ACPI
+	  name of a device.
+
 endmenu
drivers/core/Makefile +1 -0
index bce7467da1d..c707026a3a0 100644
@@ -3,6 +3,7 @@
 # Copyright (c) 2013 Google, Inc
 
 obj-y	+= device.o fdtaddr.o lists.o root.o uclass.o util.o
+obj-$(CONFIG_$(SPL_TPL_)ACPIGEN) += acpi.o
 obj-$(CONFIG_DEVRES) += devres.o
 obj-$(CONFIG_$(SPL_)DM_DEVICE_REMOVE)	+= device-remove.o
 obj-$(CONFIG_$(SPL_)SIMPLE_BUS)	+= simple-bus.o
drivers/core/acpi.c new file +33 -0
new file mode 100644
index 00000000000..0e64c21bf5b
@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Core driver model support for ACPI table generation
+ *
+ * Copyright 2019 Google LLC
+ * Written by Simon Glass <[email protected]>
+ */
+
+#define LOG_CATEOGRY	LOGC_ACPI
+
+#include <common.h>
+#include <dm.h>
+#include <dm/acpi.h>
+#include <dm/root.h>
+
+int acpi_copy_name(char *out_name, const char *name)
+{
+	strncpy(out_name, name, ACPI_NAME_MAX);
+	out_name[ACPI_NAME_LEN] = '\0';
+
+	return 0;
+}
+
+int acpi_get_name(const struct udevice *dev, char *out_name)
+{
+	struct acpi_ops *aops;
+
+	aops = device_get_acpi_ops(dev);
+	if (aops && aops->get_name)
+		return aops->get_name(dev, out_name);
+
+	return -ENOSYS;
+}
include/dm/acpi.h new file +73 -0

73 lines not on this page · raw message

include/dm/device.h +5 -0
index a56164b19bb..35e19d87005 100644
@@ -245,6 +245,8 @@ struct udevice_id {
  * pointers defined by the driver, to implement driver functions required by
  * the uclass.
  * @flags: driver flags - see DM_FLAGS_...
+ * @acpi_ops: Advanced Configuration and Power Interface (ACPI) operations,
+ * allowing the device to add things to the ACPI tables passed to Linux
  */
 struct driver {
 	char *name;
@@ -264,6 +266,9 @@ struct driver {
 	int per_child_platdata_auto_alloc_size;
 	const void *ops;	/* driver-specific operations */
 	uint32_t flags;
+#if CONFIG_IS_ENABLED(ACPIGEN)
+	struct acpi_ops *acpi_ops;
+#endif
 };
 
 /* Declare a new U-Boot driver */
include/log.h +2 -0
index 62fb8afbd0e..fbcbd42fb4b 100644
@@ -51,6 +51,8 @@ enum log_category_t {
 	LOGC_SANDBOX,	/* Related to the sandbox board */
 	LOGC_BLOBLIST,	/* Bloblist */
 	LOGC_DEVRES,	/* Device resources (devres_... functions) */
+	/* Advanced Configuration and Power Interface (ACPI) */
+	LOGC_ACPI,
 
 	LOGC_COUNT,	/* Number of log categories */
 	LOGC_END,	/* Sentinel value for a list of log categories */
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Reviewed-by on v3 12/29
#19 lore raw

Antwort: [PATCH v3 12/29] dm: core: Add basic ACPI support

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 43 lines

Betreff: [PATCH v3 12/29] dm: core: Add basic ACPI support

ACPI (Advanced Configuration and Power Interface) is a standard for specifying information about a platform. It is a little like device tree but the bindings are part of the specification and it supports an
interpreted bytecode language.

Driver model does not use ACPI for U-Boot's configuration, but it is convenient to have it support generation of ACPI tables for passing to
Linux, etc.

As a starting point, add an optional set of ACPI operations to each device. Initially only a single operation is available, to obtain the ACPI name for the device. More operations are added later.

Enable ACPI for sandbox to ensure build coverage and so that we can add
tests.

---

Changes in v3:
- Compute ACPI_NAME_MAX from ACPI_NAME_LEN
- Drop 'Intel' from 'Intel ACPI'
- Rename acpi_return_name() to acpi_copy_name()
- Reword commit message to drop the bit about ACPI being complicated
- Use strncpy() instead of strcpy() in acpi_copy_name()

Changes in v2:
- Move LOGC_ACPI definition to this patch

drivers/core/Kconfig  |  9 ++++++
drivers/core/Makefile |  1 +
drivers/core/acpi.c   | 33 +++++++++++++++++++
include/dm/acpi.h     | 73
+++++++++++++++++++++++++++++++++++++++++++
include/dm/device.h   |  5 +++
include/log.h         |  2 ++
6 files changed, 123 insertions(+)
Quoted 3 lines

create mode 100644 drivers/core/acpi.c
create mode 100644 include/dm/acpi.h

#20 lore raw

PATCH v3 12/29 dm: core: Add basic ACPI support

In reply to Simon Glass
Quoted 15 lines · Simon Glass

ACPI (Advanced Configuration and Power Interface) is a standard for specifying information about a platform. It is a little like device tree but the bindings are part of the specification and it supports an interpreted bytecode language.

Driver model does not use ACPI for U-Boot's configuration, but it is convenient to have it support generation of ACPI tables for passing to Linux, etc.

As a starting point, add an optional set of ACPI operations to each device. Initially only a single operation is available, to obtain the ACPI name for the device. More operations are added later.

Enable ACPI for sandbox to ensure build coverage and so that we can add tests.

...

drivers/core/acpi.c line 18

Quoted 4 lines
drivers/core/acpi.c new file +3 -0
+int acpi_copy_name(char *out_name, const char *name)
+{
 
+	strncpy(out_name, name, ACPI_NAME_MAX);

memcpy()?

drivers/core/acpi.c line 19

Quoted 1 line
drivers/core/acpi.c new file +1 -0
+	out_name[ACPI_NAME_LEN] = '\0';

I dunno if compiler is clever enough to catch this and avoid any warnings.

Also it seems above should also have _LEN, and not _MAX.

Quoted 2 lines
(quoted diff) +2 -0
+	return 0;
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#21 lore raw

PATCH v3 12/29 dm: core: Add basic ACPI support

Hi Andy,

Quoted 19 lines · Andy Shevchenko
Quoted 9 lines · Simon Glass

ACPI (Advanced Configuration and Power Interface) is a standard for specifying information about a platform. It is a little like device tree but the bindings are part of the specification and it supports an interpreted bytecode language.

Driver model does not use ACPI for U-Boot's configuration, but it is convenient to have it support generation of ACPI tables for passing to Linux, etc.

Quoted 6 lines

As a starting point, add an optional set of ACPI operations to each device. Initially only a single operation is available, to obtain the ACPI name for the device. More operations are added later.

Enable ACPI for sandbox to ensure build coverage and so that we can add tests.

Quoted 8 lines

...

drivers/core/acpi.c line 18

Quoted 4 lines
drivers/core/acpi.c new file +3 -0
+int acpi_copy_name(char *out_name, const char *name)
+{
 
+     strncpy(out_name, name, ACPI_NAME_MAX);

memcpy()?

strcpy() seems better since it doesn't rely on the correct string in name.

Quoted 3 lines
Quoted 1 line

+ out_name[ACPI_NAME_LEN] = '\0';

I dunno if compiler is clever enough to catch this and avoid any warnings.

I don't see any warnings.

Quoted 1 line

Also it seems above should also have _LEN, and not _MAX.

Do you mean for the strncpy()? I could do that but I still want to set the terminator to nul so that the string is definitely terminated.

Regards,
Simon

#22 lore raw

PATCH v3 12/29 dm: core: Add basic ACPI support

Quoted 3 lines · Simon Glass
Quoted 1 line · Andy Shevchenko

On Mon, Mar 30, 2020 at 05:12:48PM -0600, Simon Glass wrote:

...

Quoted 5 lines
Quoted 3 lines
Quoted 1 line

+ strncpy(out_name, name, ACPI_NAME_MAX);

memcpy()?

strcpy() seems better since it doesn't rely on the correct string in name.

Definitely not. It has no boundary checks.

Quoted 5 lines
Quoted 3 lines
Quoted 1 line

+ out_name[ACPI_NAME_LEN] = '\0';

I dunno if compiler is clever enough to catch this and avoid any warnings.

I don't see any warnings.

strncpy() fine then.

Quoted 4 lines
Quoted 1 line

Also it seems above should also have _LEN, and not _MAX.

Do you mean for the strncpy()? I could do that but I still want to set the terminator to nul so that the string is definitely terminated.

Terminator can be not present.
I should be

     strncpy(out_name, name, ACPI_NAME_LEN);
     out_name[ACPI_NAME_LEN] = '\0';
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#23 lore raw

PATCH v3 12/29 dm: core: Add basic ACPI support

Hi Andy,

Quoted 26 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass
Quoted 1 line · Andy Shevchenko

On Mon, Mar 30, 2020 at 05:12:48PM -0600, Simon Glass wrote:

...

Quoted 5 lines
Quoted 3 lines
Quoted 1 line

+ strncpy(out_name, name, ACPI_NAME_MAX);

memcpy()?

strcpy() seems better since it doesn't rely on the correct string in name.

Definitely not. It has no boundary checks.

Quoted 5 lines
Quoted 3 lines
Quoted 1 line

+ out_name[ACPI_NAME_LEN] = '\0';

I dunno if compiler is clever enough to catch this and avoid any warnings.

I don't see any warnings.

strncpy() fine then.

Quoted 2 lines
Quoted 1 line

Also it seems above should also have _LEN, and not _MAX.

Quoted 8 lines
Quoted 2 lines

Do you mean for the strncpy()? I could do that but I still want to set the terminator to nul so that the string is definitely terminated.

Terminator can be not present.
I should be

     strncpy(out_name, name, ACPI_NAME_LEN);
     out_name[ACPI_NAME_LEN] = '\0';

Why? The code I have seems to work just fine?

Regards,
Simon

#24 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

In reply to Simon Glass

Add this binding from Linux v5.4.

---

Changes in v3:
- Split out hid-over-i2c into its own patch

Changes in v2: None

 .../input/hid-over-i2c.txt                    | 44 +++++++++++++++++++
 1 file changed, 44 insertions(+)
 create mode 100644 doc/device-tree-bindings/input/hid-over-i2c.txt
doc/device-tree-bindings/input/hid-over-i2c.txt new file +44 -0
new file mode 100644
index 00000000000..c76bafaf98d
@@ -0,0 +1,44 @@
+* HID over I2C Device-Tree bindings
+
+HID over I2C provides support for various Human Interface Devices over the
+I2C bus. These devices can be for example touchpads, keyboards, touch screens
+or sensors.
+
+The specification has been written by Microsoft and is currently available here:
+http://msdn.microsoft.com/en-us/library/windows/hardware/hh852380.aspx
+
+If this binding is used, the kernel module i2c-hid will handle the communication
+with the device and the generic hid core layer will handle the protocol.
+
+Required properties:
+- compatible: must be "hid-over-i2c"
+- reg: i2c slave address
+- hid-descr-addr: HID descriptor address
+- interrupts: interrupt line
+
+Additional optional properties:
+
+Some devices may support additional optional properties to help with, e.g.,
+power sequencing. The following properties can be supported by one or more
+device-specific compatible properties, which should be used in addition to the
+"hid-over-i2c" string.
+
+- compatible:
+  * "wacom,w9013" (Wacom W9013 digitizer). Supports:
+    - vdd-supply (3.3V)
+    - vddl-supply (1.8V)
+    - post-power-on-delay-ms
+
+- vdd-supply: phandle of the regulator that provides the supply voltage.
+- post-power-on-delay-ms: time required by the device after enabling its regulators
+  or powering it on, before it is ready for communication.
+
+Example:
+
+	i2c-hid-dev at 2c {
+		compatible = "hid-over-i2c";
+		reg = <0x2c>;
+		hid-descr-addr = <0x0020>;
+		interrupt-parent = <&gpx3>;
+		interrupts = <3 2>;
+	};
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#25 lore raw

Antwort: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 16 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Devices need to report various identifiers in the ACPI tables. Rather than hard-coding these in drivers it is typically better to put them in the device tree.

Add a binding file to describe this.

Quoted 8 lines

---

Changes in v3:
- Add a pointer to information about acpi,compatible
- Change the example to ELAN
- Correct description of acpi,probed
- Drop hid-descr-addr
- Drop mention of PRIC

I understand now where the term "PRIC" came from.
The property "acpi,has-power-resource" triggers the function acpi_device_add_power_res(), which writes a PowerResource ('PowerResource' as specified in section 7.2 of ACPI 6.3). This function comes from Coreboot, and that implementation uses "PRIC" as the hardcoded device name. That's it, it is an arbitrary chosen device name (probably meaning "power IC" ... ?).

Anyway, dropping the term "PRIC" makes the description easier to understand. The important information is that "acpi,has-power-resource" leads to the addition of a PowerResource entry.

Quoted 23 lines

- Just add the device.txt binding file in this patch
- Rename acpi,desc to acpi,ddn

Changes in v2:
- Add the hid-over-i2c binding document
- Fix definition of HID
- Infer hid-over-i2c CID value

 doc/device-tree-bindings/device.txt | 37 +++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
 create mode 100644 doc/device-tree-bindings/device.txt
doc/device-tree-bindings/device.txt new file +5 -0
new file mode 100644
index 00000000000..06c2b84b6d5
@@ -0,0 +1,37 @@
+Devices
+=======
+
+Device bindings are described by their own individual binding files.
+

doc/device-tree-bindings/device.txt line 15

Quoted 10 lines
doc/device-tree-bindings/device.txt new file +10 -0
+U-Boot provides for some optional properties which are documented here. See
+also hid-over-i2c.txt which describes HID devices. See also
+Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel for
+the acpi,compatible property.
+
+ - acpi,has-power-resource : (boolean) true if this device has a power resource.
+    This causes an ACPI PowerResource to be written containing the properties
+    provided by this binding, to describe how to handle powering the device up
+    and down using GPIOs
+ - acpi,compatible : compatible string to report

I still don't see a use case for a new "acpi,compatible" property. I will take a step back, and explain my understanding of the involved pieces and why I think adding "acpi,compatible" is of no benefit.

Summary:
As far as I understand the proposed "acpi, compatible" property, the following would happen:
We have "acpi,compatible" in Devicetree, which results in a "compatible"-entry in ACPI's _DSD method, which is then again interpreted as the "compatible"-property of Devicetree. IMHO it would be strange for "compatible" and "acpi,compatible" to be different, so we could drop "acpi,compatible" and use the existing "compatible" instead.

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Detailed explanation:

1) ACPI Constraint: Devices need to have either _HID or _ADR

ACPI puts constraints on what properties a device ("device" here means a "Device()" entry in ASL code (DSDT or SSDT)) has to have. It has to have either an _HID or an _ADR property depending on whether it is on an enumerable bus:

* if it is on an enumerable bus, it has to have an _ADR (address) property (e.g. a PCI device on a PCI bus)
* if it is not on an enumerable bus, it has to have a _HID (hardware ID) property (e.g. the GPIO controllers on the Apollo Lake SoC). The legal values for _HID are specified and allow the type of the device to be identified (a similar concept as the "compatible" property in devicetree)

These contraints are specified in section 6.1 of ACPI 6.3 [1].

2) ACPI's _DSD Method

The Device Specific Data (_DSD) method provides a way to provide additional device properties to device drivers. It returns a package of "Device Data Descriptors", each consisting of a UUID and a package in a format defined by the provided UUID.

The details are specified in section 6.2.5 of ACPI 6.3 [1].

3) Special UUID value for _DSD: daffd814-...

The document "Device Properties UUID For _DSD" [2] specifies a special UUID
value: daffd814-6eba-4d8c-8a91-bc9bbf4aa301

When this UUID is used in the package returned by in a _DSD method, it means the format of the package after the UUID consits of packages each of size 2.
The first entry in these packages always has to be a string, the second entry can be an integer, a string, a reference or again a package. The value of the string defines the type and allowed values of the second entry.

The typical use case for this UUID is to return some kind of key/value-pairs.

The specification in [2] does not specify which strings are legal as key names. This depends on the _HID of the device that implements the _DSD method.

4) Special _HID value: PRP0001

When the _HID property has the value "PRP0001", the _DSD method is expected to provide data with the "daffd814"-UUID which contains a "compatible" property.

This interpreted similar to the "compatible" property known from Devicetree.

5) Linux device-property API

Linux provides a "device-property" API (define in include/linux/property.h) which can be used instead of a Devicetree specific API.
E.g. using device_property_read_u32() instead of of_property_read_u32().

Putting these pieces together:

Suppose the following use case:
There is an existing driver for Devicetree, but it should be used under x86, where devices are usually described via ACPI.

To avoid having to register a new _HID that would have the exact same meaning in ACPI as an already existing Devicetree "compatible" string, the possibilities desbribed by 2-4) allow to solve this use case while respecting the contraints described in 1).

Additionally, using the API described in 5) allows to add other Devicetree properties (not only the "compatible" property) to the ACPI description of the device. This allows to have a single device driver that can get its device description from either Devicetree or ACPI, and the description is basically the same in both worlds. This is described e.g. in the presentation in [4].

[1] https://uefi.org/sites/default/files/resources/ACPI_6_3_final_Jan30.pdf
[2] http://www.uefi.org/sites/default/files/resources/_DSD-device-properties-UUID.pdf
[3] Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel
[4] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

doc/device-tree-bindings/device.txt line 22

Quoted 7 lines
doc/device-tree-bindings/device.txt new file +7 -0
+ - acpi,ddn : Contains the string to use as the _DDN (DOS (Disk Operating
+    System) Device Name)
+ - acpi,hid : Contains the string to use as the HID (Hardware ID)
+    identifier _HID
+ - acpi,probed : Tells U-Boot to add 'linux,probed' to the ACPI tables so that
+    Linux will only load the driver if the device can be detected (e.g. on I2C
+    bus)

Will support for 'linux,probed' be mainlined? Otherwise the description should IMHO mention that it is an out-of-tree feature.

doc/device-tree-bindings/device.txt line 30

Quoted 8 lines
doc/device-tree-bindings/device.txt new file +8 -0
+ - acpi,uid : _UID value for device
+
+
+Example
+-------
+
+elan_touchscreen: elan-touchscreen at 10 {
+ compatible = "i2c-chip";

Why would you use a generic compatible string in this case? According to drivers/input/touchscreen/elants_i2c.c in the Linux kernel the ACPI _HID "ELAN0001" refers to the same device as the Devicetree compatible string "elan,ekth3500".

Again, because of the direct relationship between "compatible" string and _HID I think we could move that knowledge into a (stub-) driver for "elan,ekth3500" and then we could avoid the need for "acpi,hid" here.

Quoted 9 lines
doc/device-tree-bindings/device.txt new file +7 -1
+ reg = <0x10>;
+ acpi,hid = "ELAN0001";
+ acpi,ddn = "ELAN Touchscreen";
+ interrupts-extended = <&acpi_gpe GPIO_21_IRQ
+ IRQ_TYPE_EDGE_FALLING>;
+ acpi,probed;
+};
--

2.26.0.rc2.310.g2932bb562d-goog

regards, Wolfgang

#26 lore raw

Antwort: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Hi Simon,

-----"U-Boot" <[email protected]> schrieb: -----

[snip]

Quoted 6 lines
Quoted 1 line

+ - acpi,ddn : Contains the string to use as the _DDN (DOS (Disk

Operating

Review

Quoted 4 lines
(quoted diff) +4 -0
+    System) Device Name)
+ - acpi,hid : Contains the string to use as the HID (Hardware ID)
+    identifier _HID
+ - acpi,probed : Tells U-Boot to add 'linux,probed' to the ACPI
Quoted 8 lines

tables so that

Quoted 1 line

+ Linux will only load the driver if the device can be detected

(e.g. on I2C

Quoted 1 line

+ bus)

Will support for 'linux,probed' be mainlined? Otherwise the description
should IMHO mention that it is an out-of-tree feature.

I have thought some more about this property.
The Chromium discussions [1] mention that "linux,probed" is intended to be a property in both Devicetree and ACPI, so a Linux kernel could handle the device in the expected way whether it boots on a Devicetree platform or an ACPI platform. The proposed "acpi,probed" property would only solve one such use case (ACPI-based platforms): U-Boot boots, reads "acpi,probed" from Devicetree, adds "linux,probed" to the ACPI description and hands that to the Linux kernel.

Why would we not keep the name "linux,probed" for this property? U-Boot could still do the same Devicetree -> ACPI translation, but the same property could then also be useful for Devicetree-based platforms.

[snip]

regards, Wolfgang

[1] https://groups.google.com/a/chromium.org/forum/#!msg/chromium-os-reviews/4HTHl78IGHw/oz82uImnBgAJ

#27 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

Hi Wolfgang,

Quoted 31 lines · Wolfgang Wallner

Hi Simon,

-----"U-Boot" <[email protected]> schrieb: -----

[snip]

Quoted 6 lines
Quoted 1 line

+ - acpi,ddn : Contains the string to use as the _DDN (DOS (Disk

Operating

Review

Quoted 4 lines
(quoted diff) +4 -0
+    System) Device Name)
+ - acpi,hid : Contains the string to use as the HID (Hardware ID)
+    identifier _HID
+ - acpi,probed : Tells U-Boot to add 'linux,probed' to the ACPI
Quoted 8 lines

tables so that

Quoted 1 line

+ Linux will only load the driver if the device can be detected

(e.g. on I2C

Quoted 1 line

+ bus)

Will support for 'linux,probed' be mainlined? Otherwise the description
should IMHO mention that it is an out-of-tree feature.

I have thought some more about this property.
The Chromium discussions [1] mention that "linux,probed" is intended to be a property in both Devicetree and ACPI, so a Linux kernel could handle the device in the expected way whether it boots on a Devicetree platform or an ACPI platform. The proposed "acpi,probed" property would only solve one such use case (ACPI-based platforms): U-Boot boots, reads "acpi,probed" from Devicetree, adds "linux,probed" to the ACPI description and hands that to the Linux kernel.

Quoted 3 lines

Why would we not keep the name "linux,probed" for this property? U-Boot could still do the same Devicetree -> ACPI translation, but the same property could then also be useful for Devicetree-based platforms.

OK...oh dear. I will change it back to linux,probed

Regards,
Simon

#28 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

In reply to Wolfgang Wallner

Hi Wolfgang,

Quoted 79 lines · Wolfgang Wallner

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 16 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Devices need to report various identifiers in the ACPI tables. Rather than hard-coding these in drivers it is typically better to put them in the device tree.

Add a binding file to describe this.

Quoted 8 lines

---

Changes in v3:
- Add a pointer to information about acpi,compatible
- Change the example to ELAN
- Correct description of acpi,probed
- Drop hid-descr-addr
- Drop mention of PRIC

I understand now where the term "PRIC" came from.
The property "acpi,has-power-resource" triggers the function acpi_device_add_power_res(), which writes a PowerResource ('PowerResource' as specified in section 7.2 of ACPI 6.3). This function comes from Coreboot, and that implementation uses "PRIC" as the hardcoded device name. That's it, it is an arbitrary chosen device name (probably meaning "power IC" ... ?).

Anyway, dropping the term "PRIC" makes the description easier to understand. The important information is that "acpi,has-power-resource" leads to the addition of a PowerResource entry.

Quoted 23 lines

- Just add the device.txt binding file in this patch
- Rename acpi,desc to acpi,ddn

Changes in v2:
- Add the hid-over-i2c binding document
- Fix definition of HID
- Infer hid-over-i2c CID value

 doc/device-tree-bindings/device.txt | 37 +++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
 create mode 100644 doc/device-tree-bindings/device.txt
doc/device-tree-bindings/device.txt new file +5 -0
new file mode 100644
index 00000000000..06c2b84b6d5
@@ -0,0 +1,37 @@
+Devices
+=======
+
+Device bindings are described by their own individual binding files.
+

doc/device-tree-bindings/device.txt line 15

Quoted 10 lines
doc/device-tree-bindings/device.txt new file +10 -0
+U-Boot provides for some optional properties which are documented here. See
+also hid-over-i2c.txt which describes HID devices. See also
+Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel for
+the acpi,compatible property.
+
+ - acpi,has-power-resource : (boolean) true if this device has a power resource.
+    This causes an ACPI PowerResource to be written containing the properties
+    provided by this binding, to describe how to handle powering the device up
+    and down using GPIOs
+ - acpi,compatible : compatible string to report

I still don't see a use case for a new "acpi,compatible" property. I will take a step back, and explain my understanding of the involved pieces and why I think adding "acpi,compatible" is of no benefit.

Quoted 8 lines

Summary:
As far as I understand the proposed "acpi, compatible" property, the following would happen:
We have "acpi,compatible" in Devicetree, which results in a "compatible"-entry in ACPI's _DSD method, which is then again interpreted as the "compatible"-property of Devicetree. IMHO it would be strange for "compatible" and "acpi,compatible" to be different, so we could drop "acpi,compatible" and use the existing "compatible" instead.

But the compatible string is "hid-over-i2c". We don't want to have lots of different compatible strings here, different drivers which do the same thing. If we end up wanting a touchscreen drivers in U-Boot that might change, but for now I think a generic driver is easier.

Quoted 58 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Detailed explanation:

1) ACPI Constraint: Devices need to have either _HID or _ADR

ACPI puts constraints on what properties a device ("device" here means a "Device()" entry in ASL code (DSDT or SSDT)) has to have. It has to have either an _HID or an _ADR property depending on whether it is on an enumerable bus:

* if it is on an enumerable bus, it has to have an _ADR (address) property (e.g. a PCI device on a PCI bus)
* if it is not on an enumerable bus, it has to have a _HID (hardware ID) property (e.g. the GPIO controllers on the Apollo Lake SoC). The legal values for _HID are specified and allow the type of the device to be identified (a similar concept as the "compatible" property in devicetree)

These contraints are specified in section 6.1 of ACPI 6.3 [1].

2) ACPI's _DSD Method

The Device Specific Data (_DSD) method provides a way to provide additional device properties to device drivers. It returns a package of "Device Data Descriptors", each consisting of a UUID and a package in a format defined by the provided UUID.

The details are specified in section 6.2.5 of ACPI 6.3 [1].

3) Special UUID value for _DSD: daffd814-...

The document "Device Properties UUID For _DSD" [2] specifies a special UUID
value: daffd814-6eba-4d8c-8a91-bc9bbf4aa301

When this UUID is used in the package returned by in a _DSD method, it means the format of the package after the UUID consits of packages each of size 2.
The first entry in these packages always has to be a string, the second entry can be an integer, a string, a reference or again a package. The value of the string defines the type and allowed values of the second entry.

The typical use case for this UUID is to return some kind of key/value-pairs.

The specification in [2] does not specify which strings are legal as key names. This depends on the _HID of the device that implements the _DSD method.

4) Special _HID value: PRP0001

When the _HID property has the value "PRP0001", the _DSD method is expected to provide data with the "daffd814"-UUID which contains a "compatible" property.

Quoted 8 lines

This interpreted similar to the "compatible" property known from Devicetree.

5) Linux device-property API

Linux provides a "device-property" API (define in include/linux/property.h) which can be used instead of a Devicetree specific API.
E.g. using device_property_read_u32() instead of of_property_read_u32().

Thank you very much for digging into this. I read the ACPI spec years ago but clearly need to read it again.

Quoted 31 lines

Putting these pieces together:

Suppose the following use case:
There is an existing driver for Devicetree, but it should be used under x86, where devices are usually described via ACPI.

To avoid having to register a new _HID that would have the exact same meaning in ACPI as an already existing Devicetree "compatible" string, the possibilities desbribed by 2-4) allow to solve this use case while respecting the contraints described in 1).

Additionally, using the API described in 5) allows to add other Devicetree properties (not only the "compatible" property) to the ACPI description of the device. This allows to have a single device driver that can get its device description from either Devicetree or ACPI, and the description is basically the same in both worlds. This is described e.g. in the presentation in [4].

[1] https://uefi.org/sites/default/files/resources/ACPI_6_3_final_Jan30.pdf
[2] http://www.uefi.org/sites/default/files/resources/_DSD-device-properties-UUID.pdf
[3] Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel
[4] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

doc/device-tree-bindings/device.txt line 22

Quoted 7 lines
doc/device-tree-bindings/device.txt new file +7 -0
+ - acpi,ddn : Contains the string to use as the _DDN (DOS (Disk Operating
+    System) Device Name)
+ - acpi,hid : Contains the string to use as the HID (Hardware ID)
+    identifier _HID
+ - acpi,probed : Tells U-Boot to add 'linux,probed' to the ACPI tables so that
+    Linux will only load the driver if the device can be detected (e.g. on I2C
+    bus)
Quoted 2 lines

Will support for 'linux,probed' be mainlined? Otherwise the description should IMHO mention that it is an out-of-tree feature.

OK

Quoted 15 lines

doc/device-tree-bindings/device.txt line 30

Quoted 9 lines
doc/device-tree-bindings/device.txt new file +8 -0
+ - acpi,uid : _UID value for device
+
+
+Example
+-------
+
+elan_touchscreen: elan-touchscreen at 10 {
+ compatible = "i2c-chip";

Why would you use a generic compatible string in this case? According to drivers/input/touchscreen/elants_i2c.c in the Linux kernel the ACPI _HID "ELAN0001" refers to the same device as the Devicetree compatible string "elan,ekth3500".

Quoted 3 lines

Again, because of the direct relationship between "compatible" string and _HID I think we could move that knowledge into a (stub-) driver for "elan,ekth3500" and then we could avoid the need for "acpi,hid" here.

But then we need a driver for the ELAN touchscreen. My goal here is to have all of these devices serviced by a single driver, using suitable properties in the DT.

Quoted 11 lines
Quoted 10 lines
doc/device-tree-bindings/device.txt new file +7 -1
+ reg = <0x10>;
+ acpi,hid = "ELAN0001";
+ acpi,ddn = "ELAN Touchscreen";
+ interrupts-extended = <&acpi_gpe GPIO_21_IRQ
+ IRQ_TYPE_EDGE_FALLING>;
+ acpi,probed;
+};
--

2.26.0.rc2.310.g2932bb562d-goog

Regards,
SImon

#29 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

Quoted 11 lines · Simon Glass
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#30 lore raw

Antwort: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Hi Andy,

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 20 lines

Betreff: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 11 lines · Simon Glass
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h). There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

regards, Wolfgang

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

#31 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

Quoted 31 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Quoted 2 lines

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#32 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

Quoted 33 lines · Andy Shevchenko
Quoted 31 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Quoted 5 lines
Quoted 2 lines

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Quoted 26 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

--
With Best Regards,
Andy Shevchenko

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#33 lore raw

Antwort: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Hi Andy,

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 40 lines

Betreff: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 36 lines · Andy Shevchenko
Quoted 33 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

Quoted 6 lines
Quoted 2 lines

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Thanks for the references. I have read them carefully, especially the thread in [3].

My understanding up to now was based on presentations by David Woodhouse, so it matches with his viewpoint in the thread at [3]. And as far as I can tell Rafael Wysocki agrees with him in this thread.

What I could not find in either of the references is that PRP0001 is only for debugging, I only know about this constraint from your mail. Could you point me to any source for this?

Quoted 21 lines
Quoted 21 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 5 lines
Quoted 5 lines
Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

--
With Best Regards,
Andy Shevchenko

regards, Wolfgang

PS: I only realized now that I had mixed up the email-subject in my first reply and now the complete thread is under the wrong topic. Sorry for that.

#34 lore raw

PATCH v3 13/29 dts: Add a binding for hid-over-i2c

Quoted 51 lines · Wolfgang Wallner

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 40 lines

Betreff: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 36 lines · Andy Shevchenko
Quoted 33 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

Quoted 6 lines
Quoted 2 lines

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Thanks for the references. I have read them carefully, especially the thread in [3].

Quoted 7 lines

My understanding up to now was based on presentations by David Woodhouse, so it matches with his viewpoint in the thread at [3]. And as far as I can tell Rafael Wysocki agrees with him in this thread.

What I could not find in either of the references is that PRP0001 is only for debugging, I only know about this constraint from your mail. Could you point me to any source for this?

From [3], Rafael said:
"Let alone the fact that PRP0001 is actually quite useful at the prototyping stage when it is premature to allocate a new device ID just yet. Taking that to the extreme, if someone whittles a board in his or her garage and wants to use it to drive their homemade grass watering system, having to invent a new device ID and put it into the existing driver that otherwise doesn't require any modifications is ... you know what I mean."

It implies that the process should have included the allocation of an official ACPI ID.

You always can ask ASWG for the clarification. Maybe I learn something new about PRP0001 :-)

Quoted 22 lines
Quoted 22 lines
Quoted 22 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#35 lore raw

Antwort: Re: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Hi Andy,

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 70 lines
Quoted 51 lines · Wolfgang Wallner

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 40 lines

Betreff: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 36 lines · Andy Shevchenko
Quoted 33 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

Quoted 6 lines
Quoted 2 lines

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Thanks for the references. I have read them carefully, especially the thread in [3].

Quoted 7 lines

My understanding up to now was based on presentations by David Woodhouse, so it matches with his viewpoint in the thread at [3]. And as far as I can tell Rafael Wysocki agrees with him in this thread.

What I could not find in either of the references is that PRP0001 is only for debugging, I only know about this constraint from your mail. Could you point me to any source for this?

From [3], Rafael said:
"Let alone the fact that PRP0001 is actually quite useful at the prototyping stage when it is premature to allocate a new device ID just yet. Taking that to the extreme, if someone whittles a board in his or her garage and wants to use it to drive their homemade grass watering system, having to invent a new device ID and put it into the existing driver that otherwise doesn't require any modifications is ... you know what I mean."

Quoted 2 lines

It implies that the process should have included the allocation of an official ACPI ID.

I understand the quoted paragraph differently. In the paragraphs above Rafael argues that PRP0001 is useful, as with PRP0001 we can avoid registering redundant ACPI IDs. In the quoted paragraph he then gives another example where PRP0001 is also useful: for debugging.
But I don't understand it in a way that PRP0001 would be limited to only debugging.

A few posts before in [4] he explains in more detail, and I think it matches my understanding of the last post, especially the following sentence:

"It would be a failure if people had to write separate drivers for ACPI-based and DT-based platforms to handle the same hardware, at least at the leaf driver level."

[4] https://patchwork.kernel.org/comment/15339311/

Quoted 2 lines

You always can ask ASWG for the clarification. Maybe I learn something new about PRP0001 :-)

I had no interaction with ASWG before so I'm not sure what the process is, but I will look it up.

Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

regards, Wolfgang

#36 lore raw

Antwort: Re: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 80 lines · Wolfgang Wallner
Quoted 70 lines
Quoted 51 lines · Wolfgang Wallner

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 40 lines

Betreff: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 36 lines · Andy Shevchenko
Quoted 33 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

Quoted 6 lines
Quoted 2 lines

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Thanks for the references. I have read them carefully, especially the thread in [3].

Quoted 7 lines

My understanding up to now was based on presentations by David Woodhouse, so it matches with his viewpoint in the thread at [3]. And as far as I can tell Rafael Wysocki agrees with him in this thread.

What I could not find in either of the references is that PRP0001 is only for debugging, I only know about this constraint from your mail. Could you point me to any source for this?

From [3], Rafael said:
"Let alone the fact that PRP0001 is actually quite useful at the prototyping stage when it is premature to allocate a new device ID just yet. Taking that to the extreme, if someone whittles a board in his or her garage and wants to use it to drive their homemade grass watering system, having to invent a new device ID and put it into the existing driver that otherwise doesn't require any modifications is ... you know what I mean."

Quoted 2 lines

It implies that the process should have included the allocation of an official ACPI ID.

I understand the quoted paragraph differently. In the paragraphs above Rafael argues that PRP0001 is useful, as with PRP0001 we can avoid registering redundant ACPI IDs. In the quoted paragraph he then gives another example where PRP0001 is also useful: for debugging.
But I don't understand it in a way that PRP0001 would be limited to only debugging.

Quoted 7 lines

A few posts before in [4] he explains in more detail, and I think it matches my understanding of the last post, especially the following sentence:

"It would be a failure if people had to write separate drivers for ACPI-based and DT-based platforms to handle the same hardware, at least at the leaf driver level."

Maybe better to ask him directly?

Quoted 32 lines

[4] https://patchwork.kernel.org/comment/15339311/

Quoted 2 lines

You always can ask ASWG for the clarification. Maybe I learn something new about PRP0001 :-)

I had no interaction with ASWG before so I'm not sure what the process is, but I will look it up.

Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

regards, Wolfgang

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#37 lore raw

Antwort: Re: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 84 lines · Andy Shevchenko
Quoted 83 lines · Wolfgang Wallner
Quoted 70 lines
Quoted 51 lines · Wolfgang Wallner

-----"Andy Shevchenko" <[email protected]> schrieb: -----

Quoted 40 lines

Betreff: Re: Re: [PATCH v3 13/29] dts: Add a binding for hid-over-i2c

Quoted 36 lines · Andy Shevchenko
Quoted 33 lines · Wolfgang Wallner
Quoted 17 lines · Simon Glass
Quoted 11 lines
Quoted 8 lines · Wolfgang Wallner
Quoted 8 lines

An: u-boot at lists.denx.de
Von: "Simon Glass" <[email protected]>
Datum: 31.03.2020 01:14
Kopie: "Andy Shevchenko" <[email protected]>, "Wolfgang Wallner" <[email protected]>, "Leif Lindholm" <[email protected]>, "Simon Glass" <[email protected]>
Betreff: [PATCH v3 14/29] acpi: Add a binding for ACPI settings in the device tree

Quoted 4 lines
Quoted 4 lines

The _DSD-method for "PRP0001"-devices in ACPI allows to use Devicetree properties inside ACPI, especially it allows to re-use Devicetree's "compatible"-property. But this is for a different use case (using Devicetree properties inside ACPI, not add ACPI properties in Devicetree).

Quoted 6 lines

Before we are going further with this here is a BIG CAVEAT.

PRP0001 MUST NOT be used in production devices.

This has been derived solely for debugging / pre-production testing / etc purposes. The real devices must have an official ACPI _HID.

Thanks for pointing this out! I was not aware of this.
I have tried to understand how the PRP0001 is meant to be used, but could not find sufficient documentation. The best document I could find is Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel, and as far as I can tell the constraint that you mention is also not described there.

Do you know any other resources regarding PRP0001, e.g. some kind of speficiation?

Quoted 6 lines
Quoted 2 lines

I guess the best one is to ask somebody from UEFI Forum / ASWG. PRP is a PNP ID for UEFI Forum.

Basically last message in [3] from Rafael mentions his view on PRP0001. I guess there is still no document, although as I noticed the PRP prefix become official in a mean time.

Thanks for the references. I have read them carefully, especially the thread in [3].

Quoted 7 lines

My understanding up to now was based on presentations by David Woodhouse, so it matches with his viewpoint in the thread at [3]. And as far as I can tell Rafael Wysocki agrees with him in this thread.

What I could not find in either of the references is that PRP0001 is only for debugging, I only know about this constraint from your mail. Could you point me to any source for this?

From [3], Rafael said:
"Let alone the fact that PRP0001 is actually quite useful at the prototyping stage when it is premature to allocate a new device ID just yet. Taking that to the extreme, if someone whittles a board in his or her garage and wants to use it to drive their homemade grass watering system, having to invent a new device ID and put it into the existing driver that otherwise doesn't require any modifications is ... you know what I mean."

Quoted 2 lines

It implies that the process should have included the allocation of an official ACPI ID.

I understand the quoted paragraph differently. In the paragraphs above Rafael argues that PRP0001 is useful, as with PRP0001 we can avoid registering redundant ACPI IDs. In the quoted paragraph he then gives another example where PRP0001 is also useful: for debugging.
But I don't understand it in a way that PRP0001 would be limited to only debugging.

A few posts before in [4] he explains in more detail, and I think it matches my understanding of the last post, especially the following sentence:

Quoted 6 lines
Quoted 4 lines

"It would be a failure if people had to write separate drivers for ACPI-based and DT-based platforms to handle the same hardware, at least at the leaf driver level."

Maybe better to ask him directly?

I guess here is a root of misunderstanding, i.e. PRP0001 as _HID() vs. Device Properties as _DSD() method.

I think you quoted the thoughts WRT latter, and not related to the _HID() part!

Quoted 38 lines
Quoted 32 lines

[4] https://patchwork.kernel.org/comment/15339311/

Quoted 2 lines

You always can ask ASWG for the clarification. Maybe I learn something new about PRP0001 :-)

I had no interaction with ASWG before so I'm not sure what the process is, but I will look it up.

Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 14 lines
Quoted 2 lines

If PRP0001 is only for debugging, then I must also have misunderstood the Linux "device-property" API (define in include/linux/property.h).

Not exactly.

Quoted 3 lines

There are some presentations available on the internet, e.g. [1], that I understand like PRP0001 + "device-property" API provide a way do access data from either Devicetree or ACPI, depending on what kind of platform you are on.

No, these are not hard linked to each other (the relation is that PRP0001 is a way to enumerate devices, which don't have dedicated ACPI _HID, by using compatible property [1]). The _DSD per se (i.o.w. device properties implementation in ACPI) is a different story [2].

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines

And I put [3] here, interesting to read. However, at that time I was quite far from this topic.

[1]: https://www.kernel.org/doc/html/latest/firmware-guide/acpi/enumeration.html#device-tree-namespace-link-device-id
[2]: https://uefi.org/sites/default/files/resources/_DSD-implementation-guide-toplevel-1_2-3.htm.
[3]: https://patchwork.kernel.org/patch/7004241/

Quoted 1 line

[1] https://elinux.org/images/2/2d/Device_tree_acpi_compatibility-david_woodhouse-kernel_recipes_2015.pdf

regards, Wolfgang

Signature · 2 lines
With Best Regards,
Andy Shevchenko
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#38 lore raw

PATCH v3 14/29 acpi: Add a binding for ACPI settings in the device tree

In reply to Simon Glass

Devices need to report various identifiers in the ACPI tables. Rather than hard-coding these in drivers it is typically better to put them in the device tree.

Add a binding file to describe this.

---

Changes in v3:
- Add a pointer to information about acpi,compatible
- Change the example to ELAN
- Correct description of acpi,probed
- Drop hid-descr-addr
- Drop mention of PRIC
- Just add the device.txt binding file in this patch
- Rename acpi,desc to acpi,ddn

Changes in v2:
- Add the hid-over-i2c binding document
- Fix definition of HID
- Infer hid-over-i2c CID value

 doc/device-tree-bindings/device.txt | 37 +++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
 create mode 100644 doc/device-tree-bindings/device.txt
doc/device-tree-bindings/device.txt new file +37 -0
new file mode 100644
index 00000000000..06c2b84b6d5
@@ -0,0 +1,37 @@
+Devices
+=======
+
+Device bindings are described by their own individual binding files.
+
+U-Boot provides for some optional properties which are documented here. See
+also hid-over-i2c.txt which describes HID devices. See also
+Documentation/firmware-guide/acpi/enumeration.rst in the Linux kernel for
+the acpi,compatible property.
+
+ - acpi,has-power-resource : (boolean) true if this device has a power resource.
+    This causes an ACPI PowerResource to be written containing the properties
+    provided by this binding, to describe how to handle powering the device up
+    and down using GPIOs
+ - acpi,compatible : compatible string to report
+ - acpi,ddn : Contains the string to use as the _DDN (DOS (Disk Operating
+    System) Device Name)
+ - acpi,hid : Contains the string to use as the HID (Hardware ID)
+    identifier _HID
+ - acpi,probed : Tells U-Boot to add 'linux,probed' to the ACPI tables so that
+    Linux will only load the driver if the device can be detected (e.g. on I2C
+    bus)
+ - acpi,uid : _UID value for device
+
+
+Example
+-------
+
+elan_touchscreen: elan-touchscreen at 10 {
+	compatible = "i2c-chip";
+	reg = <0x10>;
+	acpi,hid = "ELAN0001";
+	acpi,ddn = "ELAN Touchscreen";
+	interrupts-extended = <&acpi_gpe GPIO_21_IRQ
+		IRQ_TYPE_EDGE_FALLING>;
+	acpi,probed;
+};
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#39 lore raw

PATCH v3 14/29 acpi: Add a binding for ACPI settings in the device tree

Quoted 6 lines · Simon Glass

Devices need to report various identifiers in the ACPI tables. Rather than hard-coding these in drivers it is typically better to put them in the device tree.

Add a binding file to describe this.

doc/device-tree-bindings/device.txt line 35

Quoted 8 lines
doc/device-tree-bindings/device.txt new file +7 -0
+elan_touchscreen: elan-touchscreen at 10 {
+	compatible = "i2c-chip";
+	reg = <0x10>;
+	acpi,hid = "ELAN0001";
+	acpi,ddn = "ELAN Touchscreen";
 
+	interrupts-extended = <&acpi_gpe GPIO_21_IRQ
+		IRQ_TYPE_EDGE_FALLING>;

Can we have this on one line?

Quoted 2 lines
doc/device-tree-bindings/device.txt new file +2 -0
+	acpi,probed;
+};
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#40 lore raw

PATCH v3 15/29 acpi: Add a simple sandbox test

In reply to Simon Glass

Add a sandbox test for the basic ACPI functionality we have so far.

---

Changes in v3: None
Changes in v2:
- Add in the acpi_table.h header file to this patch

 arch/sandbox/dts/test.dts             |  4 ++
 arch/sandbox/include/asm/acpi_table.h |  9 +++++
 include/dm/uclass-id.h                |  1 +
 test/dm/Makefile                      |  1 +
 test/dm/acpi.c                        | 55 +++++++++++++++++++++++++++
 5 files changed, 70 insertions(+)
 create mode 100644 arch/sandbox/include/asm/acpi_table.h
 create mode 100644 test/dm/acpi.c
arch/sandbox/dts/test.dts +4 -0
index 4a277934a71..5fa951ad4b6 100644
@@ -206,6 +206,10 @@
 		compatible = "denx,u-boot-devres-test";
 	};
 
+	acpi-test {
+		compatible = "denx,u-boot-acpi-test";
+	};
+
 	clocks {
 		clk_fixed: clk-fixed {
 			compatible = "fixed-clock";
arch/sandbox/include/asm/acpi_table.h new file +9 -0
new file mode 100644
index 00000000000..921c7f4201d
@@ -0,0 +1,9 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright 2019 Google LLC
+ */
+
+#ifndef __ASM_ACPI_TABLE_H__
+#define __ASM_ACPI_TABLE_H__
+
+#endif /* __ASM_ACPI_TABLE_H__ */
include/dm/uclass-id.h +1 -0
index 598f65ea7a3..37ada51f9f7 100644
@@ -20,6 +20,7 @@ enum uclass_id {
 	UCLASS_TEST_PROBE,
 	UCLASS_TEST_DUMMY,
 	UCLASS_TEST_DEVRES,
+	UCLASS_TEST_ACPI,
 	UCLASS_SPI_EMUL,	/* sandbox SPI device emulator */
 	UCLASS_I2C_EMUL,	/* sandbox I2C device emulator */
 	UCLASS_I2C_EMUL_PARENT,	/* parent for I2C device emulators */
test/dm/Makefile +1 -0
index dd1ceff86c0..3daf8a544ea 100644
@@ -13,6 +13,7 @@ obj-$(CONFIG_UT_DM) += test-uclass.o
 # subsystem you must add sandbox tests here.
 obj-$(CONFIG_UT_DM) += core.o
 ifneq ($(CONFIG_SANDBOX),)
+obj-$(CONFIG_ACPIGEN) += acpi.o
 obj-$(CONFIG_SOUND) += audio.o
 obj-$(CONFIG_BLK) += blk.o
 obj-$(CONFIG_BOARD) += board.o
test/dm/acpi.c new file +55 -0

55 lines not on this page · raw message

Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#41 lore raw

PATCH v3 15/29 acpi: Add a simple sandbox test

test/dm/acpi.c line 35

Quoted 8 lines · Simon Glass

Add a sandbox test for the basic ACPI functionality we have so far.

test/dm/acpi.c new file +5 -0
+U_BOOT_DRIVER(testacpi_drv) = {
+	.name	= "testacpi_drv",
+	.of_match	= testacpi_ids,
+	.id	= UCLASS_TEST_ACPI,
 
+	acpi_ops_ptr(&testacpi_ops)

I have noticed that this is not obvious why no comma here. Perhaps, since apci_ops_ptr is a macro, you should upper case it.

Quoted 1 line

+};

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#42 lore raw

PATCH v3 15/29 acpi: Add a simple sandbox test

Hi Andy,

Quoted 9 lines · Andy Shevchenko
Quoted 6 lines · Simon Glass

Add a sandbox test for the basic ACPI functionality we have so far.

test/dm/acpi.c new file +4 -0
+U_BOOT_DRIVER(testacpi_drv) = {
+     .name   = "testacpi_drv",
+     .of_match       = testacpi_ids,
+     .id     = UCLASS_TEST_ACPI,
Quoted 4 lines
Quoted 1 line

+ acpi_ops_ptr(&testacpi_ops)

I have noticed that this is not obvious why no comma here. Perhaps, since apci_ops_ptr is a macro, you should upper case it.

This is a bit like of_match_ptr() which is a macro used by Linux. Putting them in upper case makes them very hard to read. Admittedly the lack of a comma is odd though. It is because the field doesn't exist until ACPI is enabled (which it is not in SPL, for example).

Regards,
Simon

#43 lore raw

PATCH v3 15/29 acpi: Add a simple sandbox test

Quoted 13 lines · Simon Glass

Hi Andy,

Quoted 9 lines · Andy Shevchenko
Quoted 6 lines · Simon Glass

Add a sandbox test for the basic ACPI functionality we have so far.

test/dm/acpi.c new file +4 -0
+U_BOOT_DRIVER(testacpi_drv) = {
+     .name   = "testacpi_drv",
+     .of_match       = testacpi_ids,
+     .id     = UCLASS_TEST_ACPI,
Quoted 6 lines
Quoted 4 lines
Quoted 1 line

+ acpi_ops_ptr(&testacpi_ops)

I have noticed that this is not obvious why no comma here. Perhaps, since apci_ops_ptr is a macro, you should upper case it.

This is a bit like of_match_ptr() which is a macro used by Linux.

For ACPI there is capitalized, but...

Quoted 3 lines

Putting them in upper case makes them very hard to read. Admittedly the lack of a comma is odd though. It is because the field doesn't exist until ACPI is enabled (which it is not in SPL, for example).

...and this puts them to different categories, like PM ops in Linux kernel, where they are also capitalized, exactly to be used in struct definitions.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#44 lore raw

PATCH v3 16/29 x86: Move acpi_s3.h to include/acpi/

In reply to Simon Glass

This header relates to ACPI and we are about to add some more ACPI headers. Move this one into a new directory so they are together.

---

Changes in v3:
- Add new patch to move acpi_s3.h to include/acpi/

Changes in v2: None

 arch/x86/cpu/apollolake/cpu_spl.c        | 2 +-
 arch/x86/cpu/apollolake/fsp_s.c          | 2 +-
 arch/x86/cpu/apollolake/pmc.c            | 2 +-
 arch/x86/cpu/baytrail/acpi.c             | 4 ++--
 arch/x86/cpu/cpu.c                       | 2 +-
 arch/x86/cpu/wakeup.S                    | 2 +-
 arch/x86/lib/acpi_s3.c                   | 2 +-
 arch/x86/lib/coreboot_table.c            | 2 +-
 arch/x86/lib/fsp/fsp_common.c            | 2 +-
 arch/x86/lib/fsp1/fsp_common.c           | 2 +-
 arch/x86/lib/fsp2/fsp_dram.c             | 2 +-
 drivers/pci/pci_rom.c                    | 4 +---
 drivers/power/acpi_pmc/acpi-pmc-uclass.c | 2 +-
 drivers/sysreset/sysreset_x86.c          | 2 +-
 include/{ => acpi}/acpi_s3.h             | 0
 15 files changed, 15 insertions(+), 17 deletions(-)
 rename include/{ => acpi}/acpi_s3.h (100%)
arch/x86/cpu/apollolake/cpu_spl.c +1 -1
index 8a39c3128e0..e2509e391fa 100644
@@ -6,13 +6,13 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <ec_commands.h>
 #include <log.h>
 #include <spi_flash.h>
 #include <spl.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cpu.h>
 #include <asm/cpu_common.h>
 #include <asm/cpu_x86.h>
arch/x86/cpu/apollolake/fsp_s.c +1 -1
index 1f22c1ea3c6..17cf1682ad0 100644
@@ -5,11 +5,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <binman.h>
 #include <dm.h>
 #include <irq.h>
 #include <malloc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/intel_pinctrl.h>
 #include <asm/io.h>
 #include <asm/intel_regs.h>
arch/x86/cpu/apollolake/pmc.c +1 -1
index aec0c8394c2..4ea7c7447bc 100644
@@ -9,10 +9,10 @@
 #define LOG_CATEGORY UCLASS_ACPI_PMC
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dt-structs.h>
 #include <dm.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/io.h>
 #include <asm/pci.h>
 #include <power/acpi_pmc.h>
arch/x86/cpu/baytrail/acpi.c +2 -2
index f44228e6939..3c27391873c 100644
@@ -4,15 +4,15 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu.h>
 #include <dm.h>
-#include <dm/uclass-internal.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
+#include <dm/uclass-internal.h>
 
 void acpi_create_fadt(struct acpi_fadt *fadt, struct acpi_facs *facs,
 		      void *dsdt)
arch/x86/cpu/cpu.c +1 -1
index dae06949cc6..246ee50948c 100644
@@ -19,7 +19,6 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <command.h>
 #include <cpu_func.h>
 #include <dm.h>
@@ -27,6 +26,7 @@
 #include <init.h>
 #include <malloc.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/control_regs.h>
arch/x86/cpu/wakeup.S +1 -1
index 244ca1276af..093bf3bcc5c 100644
@@ -5,7 +5,7 @@
  * From coreboot src/arch/x86/wakeup.S
  */
 
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/processor.h>
 #include <asm/processor-flags.h>
 
arch/x86/lib/acpi_s3.c +1 -1
index 197636c4b50..c3759ec8492 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/post.h>
arch/x86/lib/coreboot_table.c +1 -1
index 2943e11d2a4..c996fc588df 100644
@@ -4,9 +4,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <malloc.h>
 #include <vbe.h>
+#include <acpi/acpi_s3.h>
 #include <asm/coreboot_tables.h>
 #include <asm/e820.h>
 
arch/x86/lib/fsp/fsp_common.c +1 -1
index 5eff0f99aad..267527eb344 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu_func.h>
 #include <dm.h>
 #include <errno.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp1/fsp_common.c +1 -1
index aee2a05044f..0a726807c2b 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <errno.h>
 #include <malloc.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp2/fsp_dram.c +1 -1
index 90a238a2245..c8f2c09b6a7 100644
@@ -5,9 +5,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <handoff.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/arch/cpu.h>
 #include <asm/fsp/fsp_support.h>
 #include <asm/fsp2/fsp_api.h>
drivers/pci/pci_rom.c +1 -3
index 90f224b0452..6b17f18bc07 100644
@@ -33,12 +33,10 @@
 #include <vbe.h>
 #include <video.h>
 #include <video_fb.h>
+#include <acpi/acpi_s3.h>
 #include <linux/screen_info.h>
 
-#ifdef CONFIG_X86
-#include <acpi_s3.h>
 DECLARE_GLOBAL_DATA_PTR;
-#endif
 
 __weak bool board_should_run_oprom(struct udevice *dev)
 {
drivers/power/acpi_pmc/acpi-pmc-uclass.c +1 -1
index d43de87126c..1c79f835c66 100644
@@ -6,9 +6,9 @@
 #define LOG_CATEGORY UCLASS_ACPI_PMC
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <log.h>
+#include <acpi/acpi_s3.h>
 #ifdef CONFIG_X86
 #include <asm/intel_pinctrl.h>
 #endif
drivers/sysreset/sysreset_x86.c +1 -1
index 8e2d1eaa7a1..bc911435602 100644
@@ -6,11 +6,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <efi_loader.h>
 #include <pch.h>
 #include <sysreset.h>
+#include <acpi/acpi_s3.h>
 #include <asm/io.h>
 #include <asm/processor.h>
 
include/acpi/acpi_s3.h renamed from include/acpi_s3.h +0 -0
similarity index 100%
rename from include/acpi_s3.h
rename to include/acpi/acpi_s3.h
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#45 lore raw

PATCH v3 16/29 x86: Move acpi_s3.h to include/acpi/

Quoted 3 lines · Simon Glass

This header relates to ACPI and we are about to add some more ACPI headers. Move this one into a new directory so they are together.

FWIW,

One nit below.

Quoted 215 lines

---

Changes in v3:
- Add new patch to move acpi_s3.h to include/acpi/

Changes in v2: None

 arch/x86/cpu/apollolake/cpu_spl.c        | 2 +-
 arch/x86/cpu/apollolake/fsp_s.c          | 2 +-
 arch/x86/cpu/apollolake/pmc.c            | 2 +-
 arch/x86/cpu/baytrail/acpi.c             | 4 ++--
 arch/x86/cpu/cpu.c                       | 2 +-
 arch/x86/cpu/wakeup.S                    | 2 +-
 arch/x86/lib/acpi_s3.c                   | 2 +-
 arch/x86/lib/coreboot_table.c            | 2 +-
 arch/x86/lib/fsp/fsp_common.c            | 2 +-
 arch/x86/lib/fsp1/fsp_common.c           | 2 +-
 arch/x86/lib/fsp2/fsp_dram.c             | 2 +-
 drivers/pci/pci_rom.c                    | 4 +---
 drivers/power/acpi_pmc/acpi-pmc-uclass.c | 2 +-
 drivers/sysreset/sysreset_x86.c          | 2 +-
 include/{ => acpi}/acpi_s3.h             | 0
 15 files changed, 15 insertions(+), 17 deletions(-)
 rename include/{ => acpi}/acpi_s3.h (100%)
arch/x86/cpu/apollolake/cpu_spl.c +1 -1
index 8a39c3128e0..e2509e391fa 100644
@@ -6,13 +6,13 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <ec_commands.h>
 #include <log.h>
 #include <spi_flash.h>
 #include <spl.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cpu.h>
 #include <asm/cpu_common.h>
 #include <asm/cpu_x86.h>
arch/x86/cpu/apollolake/fsp_s.c +1 -1
index 1f22c1ea3c6..17cf1682ad0 100644
@@ -5,11 +5,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <binman.h>
 #include <dm.h>
 #include <irq.h>
 #include <malloc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/intel_pinctrl.h>
 #include <asm/io.h>
 #include <asm/intel_regs.h>
arch/x86/cpu/apollolake/pmc.c +1 -1
index aec0c8394c2..4ea7c7447bc 100644
@@ -9,10 +9,10 @@
 #define LOG_CATEGORY UCLASS_ACPI_PMC
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dt-structs.h>
 #include <dm.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/io.h>
 #include <asm/pci.h>
 #include <power/acpi_pmc.h>
arch/x86/cpu/baytrail/acpi.c +2 -2
index f44228e6939..3c27391873c 100644
@@ -4,15 +4,15 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu.h>
 #include <dm.h>
-#include <dm/uclass-internal.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
+#include <dm/uclass-internal.h>
 
 void acpi_create_fadt(struct acpi_fadt *fadt, struct acpi_facs *facs,
 		      void *dsdt)
arch/x86/cpu/cpu.c +1 -1
index dae06949cc6..246ee50948c 100644
@@ -19,7 +19,6 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <command.h>
 #include <cpu_func.h>
 #include <dm.h>
@@ -27,6 +26,7 @@
 #include <init.h>
 #include <malloc.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/control_regs.h>
arch/x86/cpu/wakeup.S +1 -1
index 244ca1276af..093bf3bcc5c 100644
@@ -5,7 +5,7 @@
  * From coreboot src/arch/x86/wakeup.S
  */
 
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/processor.h>
 #include <asm/processor-flags.h>
 
arch/x86/lib/acpi_s3.c +1 -1
index 197636c4b50..c3759ec8492 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/post.h>
arch/x86/lib/coreboot_table.c +1 -1
index 2943e11d2a4..c996fc588df 100644
@@ -4,9 +4,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <malloc.h>
 #include <vbe.h>
+#include <acpi/acpi_s3.h>
 #include <asm/coreboot_tables.h>
 #include <asm/e820.h>
 
arch/x86/lib/fsp/fsp_common.c +1 -1
index 5eff0f99aad..267527eb344 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu_func.h>
 #include <dm.h>
 #include <errno.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp1/fsp_common.c +1 -1
index aee2a05044f..0a726807c2b 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <errno.h>
 #include <malloc.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp2/fsp_dram.c +1 -1
index 90a238a2245..c8f2c09b6a7 100644
@@ -5,9 +5,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <handoff.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/arch/cpu.h>
 #include <asm/fsp/fsp_support.h>
 #include <asm/fsp2/fsp_api.h>
drivers/pci/pci_rom.c +0 -0
index 90f224b0452..6b17f18bc07 100644

drivers/pci/pci_rom.c line 39

Quoted 8 lines
drivers/pci/pci_rom.c +1 -1
@@ -33,12 +33,10 @@
 #include <vbe.h>
 #include <video.h>
 #include <video_fb.h>
+#include <acpi/acpi_s3.h>
 #include <linux/screen_info.h>
 
 
-#ifdef CONFIG_X86

There is nothing in commit message about this change. Is it safe? Don't you get new compiler warnings (like defined-but-not-used)?

Quoted 46 lines
drivers/pci/pci_rom.c +0 -2
-#include <acpi_s3.h>
 DECLARE_GLOBAL_DATA_PTR;
 
-#endif
 

__weak bool board_should_run_oprom(struct udevice *dev) {

drivers/power/acpi_pmc/acpi-pmc-uclass.c +1 -1
index d43de87126c..1c79f835c66 100644
@@ -6,9 +6,9 @@
 #define LOG_CATEGORY UCLASS_ACPI_PMC
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <log.h>
+#include <acpi/acpi_s3.h>
 #ifdef CONFIG_X86
 #include <asm/intel_pinctrl.h>
 #endif
drivers/sysreset/sysreset_x86.c +1 -1
index 8e2d1eaa7a1..bc911435602 100644
@@ -6,11 +6,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <efi_loader.h>
 #include <pch.h>
 #include <sysreset.h>
+#include <acpi/acpi_s3.h>
 #include <asm/io.h>
 #include <asm/processor.h>
 
include/acpi/acpi_s3.h renamed from include/acpi_s3.h +0 -0
similarity index 100%
rename from include/acpi_s3.h
rename to include/acpi/acpi_s3.h
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#46 lore raw

PATCH v3 16/29 x86: Move acpi_s3.h to include/acpi/

Hi Andy,

Quoted 233 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

This header relates to ACPI and we are about to add some more ACPI headers. Move this one into a new directory so they are together.

FWIW,

One nit below.

Quoted 221 lines

---

Changes in v3:
- Add new patch to move acpi_s3.h to include/acpi/

Changes in v2: None

 arch/x86/cpu/apollolake/cpu_spl.c        | 2 +-
 arch/x86/cpu/apollolake/fsp_s.c          | 2 +-
 arch/x86/cpu/apollolake/pmc.c            | 2 +-
 arch/x86/cpu/baytrail/acpi.c             | 4 ++--
 arch/x86/cpu/cpu.c                       | 2 +-
 arch/x86/cpu/wakeup.S                    | 2 +-
 arch/x86/lib/acpi_s3.c                   | 2 +-
 arch/x86/lib/coreboot_table.c            | 2 +-
 arch/x86/lib/fsp/fsp_common.c            | 2 +-
 arch/x86/lib/fsp1/fsp_common.c           | 2 +-
 arch/x86/lib/fsp2/fsp_dram.c             | 2 +-
 drivers/pci/pci_rom.c                    | 4 +---
 drivers/power/acpi_pmc/acpi-pmc-uclass.c | 2 +-
 drivers/sysreset/sysreset_x86.c          | 2 +-
 include/{ => acpi}/acpi_s3.h             | 0
 15 files changed, 15 insertions(+), 17 deletions(-)
 rename include/{ => acpi}/acpi_s3.h (100%)
arch/x86/cpu/apollolake/cpu_spl.c +1 -1
index 8a39c3128e0..e2509e391fa 100644
@@ -6,13 +6,13 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <ec_commands.h>
 #include <log.h>
 #include <spi_flash.h>
 #include <spl.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cpu.h>
 #include <asm/cpu_common.h>
 #include <asm/cpu_x86.h>
arch/x86/cpu/apollolake/fsp_s.c +1 -1
index 1f22c1ea3c6..17cf1682ad0 100644
@@ -5,11 +5,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <binman.h>
 #include <dm.h>
 #include <irq.h>
 #include <malloc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/intel_pinctrl.h>
 #include <asm/io.h>
 #include <asm/intel_regs.h>
arch/x86/cpu/apollolake/pmc.c +1 -1
index aec0c8394c2..4ea7c7447bc 100644
@@ -9,10 +9,10 @@
 #define LOG_CATEGORY UCLASS_ACPI_PMC
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dt-structs.h>
 #include <dm.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/io.h>
 #include <asm/pci.h>
 #include <power/acpi_pmc.h>
arch/x86/cpu/baytrail/acpi.c +2 -2
index f44228e6939..3c27391873c 100644
@@ -4,15 +4,15 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu.h>
 #include <dm.h>
-#include <dm/uclass-internal.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
+#include <dm/uclass-internal.h>
 
 void acpi_create_fadt(struct acpi_fadt *fadt, struct acpi_facs *facs,
                    void *dsdt)
arch/x86/cpu/cpu.c +1 -1
index dae06949cc6..246ee50948c 100644
@@ -19,7 +19,6 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <command.h>
 #include <cpu_func.h>
 #include <dm.h>
@@ -27,6 +26,7 @@
 #include <init.h>
 #include <malloc.h>
 #include <syscon.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/control_regs.h>
arch/x86/cpu/wakeup.S +1 -1
index 244ca1276af..093bf3bcc5c 100644
@@ -5,7 +5,7 @@
  * From coreboot src/arch/x86/wakeup.S
  */
 
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/processor.h>
 #include <asm/processor-flags.h>
 
arch/x86/lib/acpi_s3.c +1 -1
index 197636c4b50..c3759ec8492 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
+#include <acpi/acpi_s3.h>
 #include <asm/acpi.h>
 #include <asm/acpi_table.h>
 #include <asm/post.h>
arch/x86/lib/coreboot_table.c +1 -1
index 2943e11d2a4..c996fc588df 100644
@@ -4,9 +4,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <malloc.h>
 #include <vbe.h>
+#include <acpi/acpi_s3.h>
 #include <asm/coreboot_tables.h>
 #include <asm/e820.h>
 
arch/x86/lib/fsp/fsp_common.c +1 -1
index 5eff0f99aad..267527eb344 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <cpu_func.h>
 #include <dm.h>
 #include <errno.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp1/fsp_common.c +1 -1
index aee2a05044f..0a726807c2b 100644
@@ -4,11 +4,11 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <dm.h>
 #include <errno.h>
 #include <malloc.h>
 #include <rtc.h>
+#include <acpi/acpi_s3.h>
 #include <asm/cmos_layout.h>
 #include <asm/early_cmos.h>
 #include <asm/io.h>
arch/x86/lib/fsp2/fsp_dram.c +1 -1
index 90a238a2245..c8f2c09b6a7 100644
@@ -5,9 +5,9 @@
  */
 
 #include <common.h>
-#include <acpi_s3.h>
 #include <handoff.h>
 #include <spl.h>
+#include <acpi/acpi_s3.h>
 #include <asm/arch/cpu.h>
 #include <asm/fsp/fsp_support.h>
 #include <asm/fsp2/fsp_api.h>
drivers/pci/pci_rom.c +1 -0
index 90f224b0452..6b17f18bc07 100644
@@ -33,12 +33,10 @@
 #include <vbe.h>
 #include <video.h>
 #include <video_fb.h>
+#include <acpi/acpi_s3.h>
 #include <linux/screen_info.h>
 
Quoted 4 lines

drivers/pci/pci_rom.c line 38

Quoted 1 line
drivers/pci/pci_rom.c +0 -1
-#ifdef CONFIG_X86

There is nothing in commit message about this change. Is it safe? Don't you get new compiler warnings (like defined-but-not-used)?

Not that I can see. Since it isn't specific to x86 anymore it seems like a good idea to remove this #ifdef

Regards,
Simon

#47 lore raw

PATCH v3 16/29 x86: Move acpi_s3.h to include/acpi/

Quoted 12 lines · Simon Glass
Quoted 9 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

This header relates to ACPI and we are about to add some more ACPI headers. Move this one into a new directory so they are together.

FWIW,

One nit below.

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

-#ifdef CONFIG_X86

There is nothing in commit message about this change. Is it safe? Don't you get new compiler warnings (like defined-but-not-used)?

Not that I can see. Since it isn't specific to x86 anymore it seems like a good idea to remove this #ifdef

Please mention in commit message.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#48 lore raw

PATCH v3 17/29 x86: Move acpi_table header to main include/ directory

In reply to Simon Glass

This file is potentially useful to other architectures saddled with ACPI so move most of its contents to a common location.

---

Changes in v3:
- Add forward declarations for the functions
- Move acpi_table.h to include/acpi
- Update commit message to say that we move most of its contents

Changes in v2: None

 arch/x86/cpu/baytrail/acpi.c      |   2 +-
 arch/x86/cpu/cpu.c                |   2 +-
 arch/x86/cpu/quark/acpi.c         |   2 +-
 arch/x86/cpu/tangier/acpi.c       |   4 +-
 arch/x86/include/asm/acpi_table.h | 381 +----------------------------
 arch/x86/lib/acpi.c               |   2 +-
 arch/x86/lib/acpi_s3.c            |   2 +-
 arch/x86/lib/acpi_table.c         |   2 +-
 arch/x86/lib/tables.c             |   2 +-
 arch/x86/lib/zimage.c             |   2 +-
 include/acpi/acpi_table.h         | 394 ++++++++++++++++++++++++++++++
 lib/efi_loader/efi_acpi.c         |   2 +-
 12 files changed, 412 insertions(+), 385 deletions(-)
 create mode 100644 include/acpi/acpi_table.h
arch/x86/cpu/baytrail/acpi.c +1 -1
index 3c27391873c..57723109796 100644
@@ -7,7 +7,7 @@
 #include <cpu.h>
 #include <dm.h>
 #include <acpi/acpi_s3.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
arch/x86/cpu/cpu.c +1 -1
index 246ee50948c..cec04b481b9 100644
@@ -27,8 +27,8 @@
 #include <malloc.h>
 #include <syscon.h>
 #include <acpi/acpi_s3.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi.h>
-#include <asm/acpi_table.h>
 #include <asm/control_regs.h>
 #include <asm/coreboot_tables.h>
 #include <asm/cpu.h>
arch/x86/cpu/quark/acpi.c +1 -1
index 7b6fc2f4a53..26cda3b3376 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
arch/x86/cpu/tangier/acpi.c +2 -2
index 8b128138b0d..4ec8fdd6f89 100644
@@ -8,13 +8,13 @@
 #include <common.h>
 #include <cpu.h>
 #include <dm.h>
-#include <dm/uclass-internal.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/ioapic.h>
 #include <asm/mpspec.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
+#include <dm/uclass-internal.h>
 
 void acpi_create_fadt(struct acpi_fadt *fadt, struct acpi_facs *facs,
 		      void *dsdt)
arch/x86/include/asm/acpi_table.h +7 -374

388 lines not on this page · raw message

arch/x86/lib/acpi.c +1 -1
index cba9c24dd45..7431458dcf2 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 
arch/x86/lib/acpi_s3.c +1 -1
index c3759ec8492..52410a515df 100644
@@ -5,8 +5,8 @@
 
 #include <common.h>
 #include <acpi/acpi_s3.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi.h>
-#include <asm/acpi_table.h>
 #include <asm/post.h>
 #include <linux/linkage.h>
 
arch/x86/lib/acpi_table.c +1 -1
index 0d69cf271f4..a00663d4d17 100644
@@ -12,8 +12,8 @@
 #include <dm/uclass-internal.h>
 #include <serial.h>
 #include <version.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi/global_nvs.h>
-#include <asm/acpi_table.h>
 #include <asm/ioapic.h>
 #include <asm/lapic.h>
 #include <asm/mpspec.h>
arch/x86/lib/tables.c +1 -1
index 7aea722d0b8..574d331d76f 100644
@@ -6,10 +6,10 @@
 #include <common.h>
 #include <malloc.h>
 #include <smbios.h>
+#include <acpi/acpi_table.h>
 #include <asm/sfi.h>
 #include <asm/mpspec.h>
 #include <asm/tables.h>
-#include <asm/acpi_table.h>
 #include <asm/coreboot_tables.h>
 
 /**
arch/x86/lib/zimage.c +1 -1
index 9b5e767ccc2..ffc09630b7d 100644
@@ -16,7 +16,7 @@
 #include <env.h>
 #include <irq_func.h>
 #include <malloc.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/ptrace.h>
 #include <asm/zimage.h>
include/acpi/acpi_table.h new file +394 -0

394 lines not on this page · raw message

lib/efi_loader/efi_acpi.c +1 -1
index a4e5e53d15e..9ae9c25c51e 100644
@@ -7,7 +7,7 @@
 
 #include <common.h>
 #include <efi_loader.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 
 static const efi_guid_t acpi_guid = EFI_ACPI_TABLE_GUID;
 
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#49 lore raw

PATCH v3 17/29 x86: Move acpi_table header to main include/ directory

Quoted 2 lines · Simon Glass

This file is potentially useful to other architectures saddled with ACPI so move most of its contents to a common location.

It's not just potentially, it's definitely useful.
But this makes me think, why we don't incorporate ACPICA headers as is?

Quoted 967 lines

---

Changes in v3:
- Add forward declarations for the functions
- Move acpi_table.h to include/acpi
- Update commit message to say that we move most of its contents

Changes in v2: None

 arch/x86/cpu/baytrail/acpi.c      |   2 +-
 arch/x86/cpu/cpu.c                |   2 +-
 arch/x86/cpu/quark/acpi.c         |   2 +-
 arch/x86/cpu/tangier/acpi.c       |   4 +-
 arch/x86/include/asm/acpi_table.h | 381 +----------------------------
 arch/x86/lib/acpi.c               |   2 +-
 arch/x86/lib/acpi_s3.c            |   2 +-
 arch/x86/lib/acpi_table.c         |   2 +-
 arch/x86/lib/tables.c             |   2 +-
 arch/x86/lib/zimage.c             |   2 +-
 include/acpi/acpi_table.h         | 394 ++++++++++++++++++++++++++++++
 lib/efi_loader/efi_acpi.c         |   2 +-
 12 files changed, 412 insertions(+), 385 deletions(-)
 create mode 100644 include/acpi/acpi_table.h
arch/x86/cpu/baytrail/acpi.c +1 -1
index 3c27391873c..57723109796 100644
@@ -7,7 +7,7 @@
 #include <cpu.h>
 #include <dm.h>
 #include <acpi/acpi_s3.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
arch/x86/cpu/cpu.c +1 -1
index 246ee50948c..cec04b481b9 100644
@@ -27,8 +27,8 @@
 #include <malloc.h>
 #include <syscon.h>
 #include <acpi/acpi_s3.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi.h>
-#include <asm/acpi_table.h>
 #include <asm/control_regs.h>
 #include <asm/coreboot_tables.h>
 #include <asm/cpu.h>
arch/x86/cpu/quark/acpi.c +1 -1
index 7b6fc2f4a53..26cda3b3376 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
arch/x86/cpu/tangier/acpi.c +2 -2
index 8b128138b0d..4ec8fdd6f89 100644
@@ -8,13 +8,13 @@
 #include <common.h>
 #include <cpu.h>
 #include <dm.h>
-#include <dm/uclass-internal.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/ioapic.h>
 #include <asm/mpspec.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
 #include <asm/arch/iomap.h>
+#include <dm/uclass-internal.h>
 
 void acpi_create_fadt(struct acpi_fadt *fadt, struct acpi_facs *facs,
 		      void *dsdt)
arch/x86/include/asm/acpi_table.h +7 -374
index 7588913f937..928475cef4e 100644
@@ -9,381 +9,14 @@
 #ifndef __ASM_ACPI_TABLE_H__
 #define __ASM_ACPI_TABLE_H__
 
-#define RSDP_SIG		"RSD PTR "	/* RSDP pointer signature */
-#define OEM_ID			"U-BOOT"	/* U-Boot */
-#define OEM_TABLE_ID		"U-BOOTBL"	/* U-Boot Table */
-#define ASLC_ID			"INTL"		/* Intel ASL Compiler */
-
-#define ACPI_RSDP_REV_ACPI_1_0	0
-#define ACPI_RSDP_REV_ACPI_2_0	2
-
-/*
- * RSDP (Root System Description Pointer)
- * Note: ACPI 1.0 didn't have length, xsdt_address, and ext_checksum
- */
-struct acpi_rsdp {
-	char signature[8];	/* RSDP signature */
-	u8 checksum;		/* Checksum of the first 20 bytes */
-	char oem_id[6];		/* OEM ID */
-	u8 revision;		/* 0 for ACPI 1.0, others 2 */
-	u32 rsdt_address;	/* Physical address of RSDT (32 bits) */
-	u32 length;		/* Total RSDP length (incl. extended part) */
-	u64 xsdt_address;	/* Physical address of XSDT (64 bits) */
-	u8 ext_checksum;	/* Checksum of the whole table */
-	u8 reserved[3];
-};
-
-/* Generic ACPI header, provided by (almost) all tables */
-struct __packed acpi_table_header {
-	char signature[4];	/* ACPI signature (4 ASCII characters) */
-	u32 length;		/* Table length in bytes (incl. header) */
-	u8 revision;		/* Table version (not ACPI version!) */
-	volatile u8 checksum;	/* To make sum of entire table == 0 */
-	char oem_id[6];		/* OEM identification */
-	char oem_table_id[8];	/* OEM table identification */
-	u32 oem_revision;	/* OEM revision number */
-	char aslc_id[4];	/* ASL compiler vendor ID */
-	u32 aslc_revision;	/* ASL compiler revision number */
-};
-
-/* A maximum number of 32 ACPI tables ought to be enough for now */
-#define MAX_ACPI_TABLES		32
-
-/* RSDT (Root System Description Table) */
-struct acpi_rsdt {
-	struct acpi_table_header header;
-	u32 entry[MAX_ACPI_TABLES];
-};
-
-/* XSDT (Extended System Description Table) */
-struct acpi_xsdt {
-	struct acpi_table_header header;
-	u64 entry[MAX_ACPI_TABLES];
-};
-
-/* FADT Preferred Power Management Profile */
-enum acpi_pm_profile {
-	ACPI_PM_UNSPECIFIED = 0,
-	ACPI_PM_DESKTOP,
-	ACPI_PM_MOBILE,
-	ACPI_PM_WORKSTATION,
-	ACPI_PM_ENTERPRISE_SERVER,
-	ACPI_PM_SOHO_SERVER,
-	ACPI_PM_APPLIANCE_PC,
-	ACPI_PM_PERFORMANCE_SERVER,
-	ACPI_PM_TABLET
-};
-
-/* FADT flags for p_lvl2_lat and p_lvl3_lat */
-#define ACPI_FADT_C2_NOT_SUPPORTED	101
-#define ACPI_FADT_C3_NOT_SUPPORTED	1001
-
-/* FADT Boot Architecture Flags */
-#define ACPI_FADT_LEGACY_FREE		0x00
-#define ACPI_FADT_LEGACY_DEVICES	(1 << 0)
-#define ACPI_FADT_8042			(1 << 1)
-#define ACPI_FADT_VGA_NOT_PRESENT	(1 << 2)
-#define ACPI_FADT_MSI_NOT_SUPPORTED	(1 << 3)
-#define ACPI_FADT_NO_PCIE_ASPM_CONTROL	(1 << 4)
-
-/* FADT Feature Flags */
-#define ACPI_FADT_WBINVD		(1 << 0)
-#define ACPI_FADT_WBINVD_FLUSH		(1 << 1)
-#define ACPI_FADT_C1_SUPPORTED		(1 << 2)
-#define ACPI_FADT_C2_MP_SUPPORTED	(1 << 3)
-#define ACPI_FADT_POWER_BUTTON		(1 << 4)
-#define ACPI_FADT_SLEEP_BUTTON		(1 << 5)
-#define ACPI_FADT_FIXED_RTC		(1 << 6)
-#define ACPI_FADT_S4_RTC_WAKE		(1 << 7)
-#define ACPI_FADT_32BIT_TIMER		(1 << 8)
-#define ACPI_FADT_DOCKING_SUPPORTED	(1 << 9)
-#define ACPI_FADT_RESET_REGISTER	(1 << 10)
-#define ACPI_FADT_SEALED_CASE		(1 << 11)
-#define ACPI_FADT_HEADLESS		(1 << 12)
-#define ACPI_FADT_SLEEP_TYPE		(1 << 13)
-#define ACPI_FADT_PCI_EXPRESS_WAKE	(1 << 14)
-#define ACPI_FADT_PLATFORM_CLOCK	(1 << 15)
-#define ACPI_FADT_S4_RTC_VALID		(1 << 16)
-#define ACPI_FADT_REMOTE_POWER_ON	(1 << 17)
-#define ACPI_FADT_APIC_CLUSTER		(1 << 18)
-#define ACPI_FADT_APIC_PHYSICAL		(1 << 19)
-#define ACPI_FADT_HW_REDUCED_ACPI	(1 << 20)
-#define ACPI_FADT_LOW_PWR_IDLE_S0	(1 << 21)
-
-enum acpi_address_space_type {
-	ACPI_ADDRESS_SPACE_MEMORY = 0,	/* System memory */
-	ACPI_ADDRESS_SPACE_IO,		/* System I/O */
-	ACPI_ADDRESS_SPACE_PCI,		/* PCI config space */
-	ACPI_ADDRESS_SPACE_EC,		/* Embedded controller */
-	ACPI_ADDRESS_SPACE_SMBUS,	/* SMBus */
-	ACPI_ADDRESS_SPACE_PCC = 0x0a,	/* Platform Comm. Channel */
-	ACPI_ADDRESS_SPACE_FIXED = 0x7f	/* Functional fixed hardware */
-};
-
-enum acpi_address_space_size {
-	ACPI_ACCESS_SIZE_UNDEFINED = 0,
-	ACPI_ACCESS_SIZE_BYTE_ACCESS,
-	ACPI_ACCESS_SIZE_WORD_ACCESS,
-	ACPI_ACCESS_SIZE_DWORD_ACCESS,
-	ACPI_ACCESS_SIZE_QWORD_ACCESS
-};
-
-struct acpi_gen_regaddr {
-	u8 space_id;	/* Address space ID */
-	u8 bit_width;	/* Register size in bits */
-	u8 bit_offset;	/* Register bit offset */
-	u8 access_size;	/* Access size */
-	u32 addrl;	/* Register address, low 32 bits */
-	u32 addrh;	/* Register address, high 32 bits */
-};
-
-/* FADT (Fixed ACPI Description Table) */
-struct __packed acpi_fadt {
-	struct acpi_table_header header;
-	u32 firmware_ctrl;
-	u32 dsdt;
-	u8 res1;
-	u8 preferred_pm_profile;
-	u16 sci_int;
-	u32 smi_cmd;
-	u8 acpi_enable;
-	u8 acpi_disable;
-	u8 s4bios_req;
-	u8 pstate_cnt;
-	u32 pm1a_evt_blk;
-	u32 pm1b_evt_blk;
-	u32 pm1a_cnt_blk;
-	u32 pm1b_cnt_blk;
-	u32 pm2_cnt_blk;
-	u32 pm_tmr_blk;
-	u32 gpe0_blk;
-	u32 gpe1_blk;
-	u8 pm1_evt_len;
-	u8 pm1_cnt_len;
-	u8 pm2_cnt_len;
-	u8 pm_tmr_len;
-	u8 gpe0_blk_len;
-	u8 gpe1_blk_len;
-	u8 gpe1_base;
-	u8 cst_cnt;
-	u16 p_lvl2_lat;
-	u16 p_lvl3_lat;
-	u16 flush_size;
-	u16 flush_stride;
-	u8 duty_offset;
-	u8 duty_width;
-	u8 day_alrm;
-	u8 mon_alrm;
-	u8 century;
-	u16 iapc_boot_arch;
-	u8 res2;
-	u32 flags;
-	struct acpi_gen_regaddr reset_reg;
-	u8 reset_value;
-	u16 arm_boot_arch;
-	u8 minor_revision;
-	u32 x_firmware_ctl_l;
-	u32 x_firmware_ctl_h;
-	u32 x_dsdt_l;
-	u32 x_dsdt_h;
-	struct acpi_gen_regaddr x_pm1a_evt_blk;
-	struct acpi_gen_regaddr x_pm1b_evt_blk;
-	struct acpi_gen_regaddr x_pm1a_cnt_blk;
-	struct acpi_gen_regaddr x_pm1b_cnt_blk;
-	struct acpi_gen_regaddr x_pm2_cnt_blk;
-	struct acpi_gen_regaddr x_pm_tmr_blk;
-	struct acpi_gen_regaddr x_gpe0_blk;
-	struct acpi_gen_regaddr x_gpe1_blk;
-};
-
-/* FACS flags */
-#define ACPI_FACS_S4BIOS_F	(1 << 0)
-#define ACPI_FACS_64BIT_WAKE_F	(1 << 1)
-
-/* FACS (Firmware ACPI Control Structure) */
-struct acpi_facs {
-	char signature[4];		/* "FACS" */
-	u32 length;			/* Length in bytes (>= 64) */
-	u32 hardware_signature;		/* Hardware signature */
-	u32 firmware_waking_vector;	/* Firmware waking vector */
-	u32 global_lock;		/* Global lock */
-	u32 flags;			/* FACS flags */
-	u32 x_firmware_waking_vector_l;	/* X FW waking vector, low */
-	u32 x_firmware_waking_vector_h;	/* X FW waking vector, high */
-	u8 version;			/* Version 2 */
-	u8 res1[3];
-	u32 ospm_flags;			/* OSPM enabled flags */
-	u8 res2[24];
-};
-
-/* MADT flags */
-#define ACPI_MADT_PCAT_COMPAT	(1 << 0)
-
-/* MADT (Multiple APIC Description Table) */
-struct acpi_madt {
-	struct acpi_table_header header;
-	u32 lapic_addr;			/* Local APIC address */
-	u32 flags;			/* Multiple APIC flags */
-};
-
-/* MADT: APIC Structure Type*/
-enum acpi_apic_types {
-	ACPI_APIC_LAPIC	= 0,		/* Processor local APIC */
-	ACPI_APIC_IOAPIC,		/* I/O APIC */
-	ACPI_APIC_IRQ_SRC_OVERRIDE,	/* Interrupt source override */
-	ACPI_APIC_NMI_SRC,		/* NMI source */
-	ACPI_APIC_LAPIC_NMI,		/* Local APIC NMI */
-	ACPI_APIC_LAPIC_ADDR_OVERRIDE,	/* Local APIC address override */
-	ACPI_APIC_IOSAPIC,		/* I/O SAPIC */
-	ACPI_APIC_LSAPIC,		/* Local SAPIC */
-	ACPI_APIC_PLATFORM_IRQ_SRC,	/* Platform interrupt sources */
-	ACPI_APIC_LX2APIC,		/* Processor local x2APIC */
-	ACPI_APIC_LX2APIC_NMI,		/* Local x2APIC NMI */
-};
-
-/* MADT: Processor Local APIC Structure */
-
-#define LOCAL_APIC_FLAG_ENABLED	(1 << 0)
-
-struct acpi_madt_lapic {
-	u8 type;		/* Type (0) */
-	u8 length;		/* Length in bytes (8) */
-	u8 processor_id;	/* ACPI processor ID */
-	u8 apic_id;		/* Local APIC ID */
-	u32 flags;		/* Local APIC flags */
-};
-
-/* MADT: I/O APIC Structure */
-struct acpi_madt_ioapic {
-	u8 type;		/* Type (1) */
-	u8 length;		/* Length in bytes (12) */
-	u8 ioapic_id;		/* I/O APIC ID */
-	u8 reserved;
-	u32 ioapic_addr;	/* I/O APIC address */
-	u32 gsi_base;		/* Global system interrupt base */
-};
-
-/* MADT: Interrupt Source Override Structure */
-struct __packed acpi_madt_irqoverride {
-	u8 type;		/* Type (2) */
-	u8 length;		/* Length in bytes (10) */
-	u8 bus;			/* ISA (0) */
-	u8 source;		/* Bus-relative int. source (IRQ) */
-	u32 gsirq;		/* Global system interrupt */
-	u16 flags;		/* MPS INTI flags */
-};
-
-/* MADT: Local APIC NMI Structure */
-struct __packed acpi_madt_lapic_nmi {
-	u8 type;		/* Type (4) */
-	u8 length;		/* Length in bytes (6) */
-	u8 processor_id;	/* ACPI processor ID */
-	u16 flags;		/* MPS INTI flags */
-	u8 lint;		/* Local APIC LINT# */
-};
-
-/* MCFG (PCI Express MMIO config space BAR description table) */
-struct acpi_mcfg {
-	struct acpi_table_header header;
-	u8 reserved[8];
-};
-
-struct acpi_mcfg_mmconfig {
-	u32 base_address_l;
-	u32 base_address_h;
-	u16 pci_segment_group_number;
-	u8 start_bus_number;
-	u8 end_bus_number;
-	u8 reserved[4];
-};
-
-/* PM1_CNT bit defines */
-#define PM1_CNT_SCI_EN		(1 << 0)
-
-/* ACPI global NVS structure */
+struct acpi_facs;
+struct acpi_fadt;
 struct acpi_global_nvs;
-
-/* CSRT (Core System Resource Table) */
-struct acpi_csrt {
-	struct acpi_table_header header;
-};
-
-struct acpi_csrt_group {
-	u32 length;
-	u32 vendor_id;
-	u32 subvendor_id;
-	u16 device_id;
-	u16 subdevice_id;
-	u16 revision;
-	u16 reserved;
-	u32 shared_info_length;
-};
-
-struct acpi_csrt_shared_info {
-	u16 major_version;
-	u16 minor_version;
-	u32 mmio_base_low;
-	u32 mmio_base_high;
-	u32 gsi_interrupt;
-	u8 interrupt_polarity;
-	u8 interrupt_mode;
-	u8 num_channels;
-	u8 dma_address_width;
-	u16 base_request_line;
-	u16 num_handshake_signals;
-	u32 max_block_size;
-};
-
-/* DBG2 definitions are partially used for SPCR interface_type */
-
-/* Types for port_type field */
-
-#define ACPI_DBG2_SERIAL_PORT		0x8000
-#define ACPI_DBG2_1394_PORT		0x8001
-#define ACPI_DBG2_USB_PORT		0x8002
-#define ACPI_DBG2_NET_PORT		0x8003
-
-/* Subtypes for port_subtype field */
-
-#define ACPI_DBG2_16550_COMPATIBLE	0x0000
-#define ACPI_DBG2_16550_SUBSET		0x0001
-#define ACPI_DBG2_ARM_PL011		0x0003
-#define ACPI_DBG2_ARM_SBSA_32BIT	0x000D
-#define ACPI_DBG2_ARM_SBSA_GENERIC	0x000E
-#define ACPI_DBG2_ARM_DCC		0x000F
-#define ACPI_DBG2_BCM2835		0x0010
-
-#define ACPI_DBG2_1394_STANDARD		0x0000
-
-#define ACPI_DBG2_USB_XHCI		0x0000
-#define ACPI_DBG2_USB_EHCI		0x0001
-
-#define ACPI_DBG2_UNKNOWN		0x00FF
-
-/* SPCR (Serial Port Console Redirection table) */
-struct __packed acpi_spcr {
-	struct acpi_table_header header;
-	u8 interface_type;
-	u8 reserved[3];
-	struct acpi_gen_regaddr serial_port;
-	u8 interrupt_type;
-	u8 pc_interrupt;
-	u32 interrupt;		/* Global system interrupt */
-	u8 baud_rate;
-	u8 parity;
-	u8 stop_bits;
-	u8 flow_control;
-	u8 terminal_type;
-	u8 reserved1;
-	u16 pci_device_id;	/* Must be 0xffff if not PCI device */
-	u16 pci_vendor_id;	/* Must be 0xffff if not PCI device */
-	u8 pci_bus;
-	u8 pci_device;
-	u8 pci_function;
-	u32 pci_flags;
-	u8 pci_segment;
-	u32 reserved2;
-};
+struct acpi_madt_ioapic;
+struct acpi_madt_irqoverride;
+struct acpi_madt_lapic_nmi;
+struct acpi_mcfg_mmconfig;
+struct acpi_table_header;
 
 /* These can be used by the target port */
 
arch/x86/lib/acpi.c +1 -1
index cba9c24dd45..7431458dcf2 100644
@@ -4,7 +4,7 @@
  */
 
 #include <common.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/tables.h>
 
arch/x86/lib/acpi_s3.c +1 -1
index c3759ec8492..52410a515df 100644
@@ -5,8 +5,8 @@
 
 #include <common.h>
 #include <acpi/acpi_s3.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi.h>
-#include <asm/acpi_table.h>
 #include <asm/post.h>
 #include <linux/linkage.h>
 
arch/x86/lib/acpi_table.c +1 -1
index 0d69cf271f4..a00663d4d17 100644
@@ -12,8 +12,8 @@
 #include <dm/uclass-internal.h>
 #include <serial.h>
 #include <version.h>
+#include <acpi/acpi_table.h>
 #include <asm/acpi/global_nvs.h>
-#include <asm/acpi_table.h>
 #include <asm/ioapic.h>
 #include <asm/lapic.h>
 #include <asm/mpspec.h>
arch/x86/lib/tables.c +1 -1
index 7aea722d0b8..574d331d76f 100644
@@ -6,10 +6,10 @@
 #include <common.h>
 #include <malloc.h>
 #include <smbios.h>
+#include <acpi/acpi_table.h>
 #include <asm/sfi.h>
 #include <asm/mpspec.h>
 #include <asm/tables.h>
-#include <asm/acpi_table.h>
 #include <asm/coreboot_tables.h>
 
 /**
arch/x86/lib/zimage.c +1 -1
index 9b5e767ccc2..ffc09630b7d 100644
@@ -16,7 +16,7 @@
 #include <env.h>
 #include <irq_func.h>
 #include <malloc.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 #include <asm/io.h>
 #include <asm/ptrace.h>
 #include <asm/zimage.h>
include/acpi/acpi_table.h new file +394 -0
new file mode 100644
index 00000000000..b4404a632c0
@@ -0,0 +1,394 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Helpers for ACPI table generation
+ *
+ * Based on acpi.c from coreboot
+ *
+ * Copyright 2019 Google LLC
+ *
+ * Copyright (C) 2015, Saket Sinha <[email protected]>
+ * Copyright (C) 2016, Bin Meng <[email protected]>
+ */
+
+#ifndef __ACPI_TABLE_H__
+#define __ACPI_TABLE_H__
+
+#define RSDP_SIG		"RSD PTR "	/* RSDP pointer signature */
+#define OEM_ID			"U-BOOT"	/* U-Boot */
+#define OEM_TABLE_ID		"U-BOOTBL"	/* U-Boot Table */
+#define ASLC_ID			"INTL"		/* Intel ASL Compiler */
+
+#define ACPI_RSDP_REV_ACPI_1_0	0
+#define ACPI_RSDP_REV_ACPI_2_0	2
+
+/*
+ * RSDP (Root System Description Pointer)
+ * Note: ACPI 1.0 didn't have length, xsdt_address, and ext_checksum
+ */
+struct acpi_rsdp {
+	char signature[8];	/* RSDP signature */
+	u8 checksum;		/* Checksum of the first 20 bytes */
+	char oem_id[6];		/* OEM ID */
+	u8 revision;		/* 0 for ACPI 1.0, others 2 */
+	u32 rsdt_address;	/* Physical address of RSDT (32 bits) */
+	u32 length;		/* Total RSDP length (incl. extended part) */
+	u64 xsdt_address;	/* Physical address of XSDT (64 bits) */
+	u8 ext_checksum;	/* Checksum of the whole table */
+	u8 reserved[3];
+};
+
+/* Generic ACPI header, provided by (almost) all tables */
+struct __packed acpi_table_header {
+	char signature[4];	/* ACPI signature (4 ASCII characters) */
+	u32 length;		/* Table length in bytes (incl. header) */
+	u8 revision;		/* Table version (not ACPI version!) */
+	volatile u8 checksum;	/* To make sum of entire table == 0 */
+	char oem_id[6];		/* OEM identification */
+	char oem_table_id[8];	/* OEM table identification */
+	u32 oem_revision;	/* OEM revision number */
+	char aslc_id[4];	/* ASL compiler vendor ID */
+	u32 aslc_revision;	/* ASL compiler revision number */
+};
+
+/* A maximum number of 32 ACPI tables ought to be enough for now */
+#define MAX_ACPI_TABLES		32
+
+/* RSDT (Root System Description Table) */
+struct acpi_rsdt {
+	struct acpi_table_header header;
+	u32 entry[MAX_ACPI_TABLES];
+};
+
+/* XSDT (Extended System Description Table) */
+struct acpi_xsdt {
+	struct acpi_table_header header;
+	u64 entry[MAX_ACPI_TABLES];
+};
+
+/* FADT Preferred Power Management Profile */
+enum acpi_pm_profile {
+	ACPI_PM_UNSPECIFIED = 0,
+	ACPI_PM_DESKTOP,
+	ACPI_PM_MOBILE,
+	ACPI_PM_WORKSTATION,
+	ACPI_PM_ENTERPRISE_SERVER,
+	ACPI_PM_SOHO_SERVER,
+	ACPI_PM_APPLIANCE_PC,
+	ACPI_PM_PERFORMANCE_SERVER,
+	ACPI_PM_TABLET
+};
+
+/* FADT flags for p_lvl2_lat and p_lvl3_lat */
+#define ACPI_FADT_C2_NOT_SUPPORTED	101
+#define ACPI_FADT_C3_NOT_SUPPORTED	1001
+
+/* FADT Boot Architecture Flags */
+#define ACPI_FADT_LEGACY_FREE		0x00
+#define ACPI_FADT_LEGACY_DEVICES	BIT(0)
+#define ACPI_FADT_8042			BIT(1)
+#define ACPI_FADT_VGA_NOT_PRESENT	BIT(2)
+#define ACPI_FADT_MSI_NOT_SUPPORTED	BIT(3)
+#define ACPI_FADT_NO_PCIE_ASPM_CONTROL	BIT(4)
+
+/* FADT Feature Flags */
+#define ACPI_FADT_WBINVD		BIT(0)
+#define ACPI_FADT_WBINVD_FLUSH		BIT(1)
+#define ACPI_FADT_C1_SUPPORTED		BIT(2)
+#define ACPI_FADT_C2_MP_SUPPORTED	BIT(3)
+#define ACPI_FADT_POWER_BUTTON		BIT(4)
+#define ACPI_FADT_SLEEP_BUTTON		BIT(5)
+#define ACPI_FADT_FIXED_RTC		BIT(6)
+#define ACPI_FADT_S4_RTC_WAKE		BIT(7)
+#define ACPI_FADT_32BIT_TIMER		BIT(8)
+#define ACPI_FADT_DOCKING_SUPPORTED	BIT(9)
+#define ACPI_FADT_RESET_REGISTER	BIT(10)
+#define ACPI_FADT_SEALED_CASE		BIT(11)
+#define ACPI_FADT_HEADLESS		BIT(12)
+#define ACPI_FADT_SLEEP_TYPE		BIT(13)
+#define ACPI_FADT_PCI_EXPRESS_WAKE	BIT(14)
+#define ACPI_FADT_PLATFORM_CLOCK	BIT(15)
+#define ACPI_FADT_S4_RTC_VALID		BIT(16)
+#define ACPI_FADT_REMOTE_POWER_ON	BIT(17)
+#define ACPI_FADT_APIC_CLUSTER		BIT(18)
+#define ACPI_FADT_APIC_PHYSICAL		BIT(19)
+#define ACPI_FADT_HW_REDUCED_ACPI	BIT(20)
+#define ACPI_FADT_LOW_PWR_IDLE_S0	BIT(21)
+
+enum acpi_address_space_type {
+	ACPI_ADDRESS_SPACE_MEMORY = 0,	/* System memory */
+	ACPI_ADDRESS_SPACE_IO,		/* System I/O */
+	ACPI_ADDRESS_SPACE_PCI,		/* PCI config space */
+	ACPI_ADDRESS_SPACE_EC,		/* Embedded controller */
+	ACPI_ADDRESS_SPACE_SMBUS,	/* SMBus */
+	ACPI_ADDRESS_SPACE_PCC = 0x0a,	/* Platform Comm. Channel */
+	ACPI_ADDRESS_SPACE_FIXED = 0x7f	/* Functional fixed hardware */
+};
+
+enum acpi_address_space_size {
+	ACPI_ACCESS_SIZE_UNDEFINED = 0,
+	ACPI_ACCESS_SIZE_BYTE_ACCESS,
+	ACPI_ACCESS_SIZE_WORD_ACCESS,
+	ACPI_ACCESS_SIZE_DWORD_ACCESS,
+	ACPI_ACCESS_SIZE_QWORD_ACCESS
+};
+
+struct acpi_gen_regaddr {
+	u8 space_id;	/* Address space ID */
+	u8 bit_width;	/* Register size in bits */
+	u8 bit_offset;	/* Register bit offset */
+	u8 access_size;	/* Access size */
+	u32 addrl;	/* Register address, low 32 bits */
+	u32 addrh;	/* Register address, high 32 bits */
+};
+
+/* FADT (Fixed ACPI Description Table) */
+struct __packed acpi_fadt {
+	struct acpi_table_header header;
+	u32 firmware_ctrl;
+	u32 dsdt;
+	u8 res1;
+	u8 preferred_pm_profile;
+	u16 sci_int;
+	u32 smi_cmd;
+	u8 acpi_enable;
+	u8 acpi_disable;
+	u8 s4bios_req;
+	u8 pstate_cnt;
+	u32 pm1a_evt_blk;
+	u32 pm1b_evt_blk;
+	u32 pm1a_cnt_blk;
+	u32 pm1b_cnt_blk;
+	u32 pm2_cnt_blk;
+	u32 pm_tmr_blk;
+	u32 gpe0_blk;
+	u32 gpe1_blk;
+	u8 pm1_evt_len;
+	u8 pm1_cnt_len;
+	u8 pm2_cnt_len;
+	u8 pm_tmr_len;
+	u8 gpe0_blk_len;
+	u8 gpe1_blk_len;
+	u8 gpe1_base;
+	u8 cst_cnt;
+	u16 p_lvl2_lat;
+	u16 p_lvl3_lat;
+	u16 flush_size;
+	u16 flush_stride;
+	u8 duty_offset;
+	u8 duty_width;
+	u8 day_alrm;
+	u8 mon_alrm;
+	u8 century;
+	u16 iapc_boot_arch;
+	u8 res2;
+	u32 flags;
+	struct acpi_gen_regaddr reset_reg;
+	u8 reset_value;
+	u16 arm_boot_arch;
+	u8 minor_revision;
+	u32 x_firmware_ctl_l;
+	u32 x_firmware_ctl_h;
+	u32 x_dsdt_l;
+	u32 x_dsdt_h;
+	struct acpi_gen_regaddr x_pm1a_evt_blk;
+	struct acpi_gen_regaddr x_pm1b_evt_blk;
+	struct acpi_gen_regaddr x_pm1a_cnt_blk;
+	struct acpi_gen_regaddr x_pm1b_cnt_blk;
+	struct acpi_gen_regaddr x_pm2_cnt_blk;
+	struct acpi_gen_regaddr x_pm_tmr_blk;
+	struct acpi_gen_regaddr x_gpe0_blk;
+	struct acpi_gen_regaddr x_gpe1_blk;
+};
+
+/* FACS flags */
+#define ACPI_FACS_S4BIOS_F		BIT(0)
+#define ACPI_FACS_64BIT_WAKE_F		BIT(1)
+
+/* FACS (Firmware ACPI Control Structure) */
+struct acpi_facs {
+	char signature[4];		/* "FACS" */
+	u32 length;			/* Length in bytes (>= 64) */
+	u32 hardware_signature;		/* Hardware signature */
+	u32 firmware_waking_vector;	/* Firmware waking vector */
+	u32 global_lock;		/* Global lock */
+	u32 flags;			/* FACS flags */
+	u32 x_firmware_waking_vector_l;	/* X FW waking vector, low */
+	u32 x_firmware_waking_vector_h;	/* X FW waking vector, high */
+	u8 version;			/* Version 2 */
+	u8 res1[3];
+	u32 ospm_flags;			/* OSPM enabled flags */
+	u8 res2[24];
+};
+
+/* MADT flags */
+#define ACPI_MADT_PCAT_COMPAT	BIT(0)
+
+/* MADT (Multiple APIC Description Table) */
+struct acpi_madt {
+	struct acpi_table_header header;
+	u32 lapic_addr;			/* Local APIC address */
+	u32 flags;			/* Multiple APIC flags */
+};
+
+/* MADT: APIC Structure Type*/
+enum acpi_apic_types {
+	ACPI_APIC_LAPIC	= 0,		/* Processor local APIC */
+	ACPI_APIC_IOAPIC,		/* I/O APIC */
+	ACPI_APIC_IRQ_SRC_OVERRIDE,	/* Interrupt source override */
+	ACPI_APIC_NMI_SRC,		/* NMI source */
+	ACPI_APIC_LAPIC_NMI,		/* Local APIC NMI */
+	ACPI_APIC_LAPIC_ADDR_OVERRIDE,	/* Local APIC address override */
+	ACPI_APIC_IOSAPIC,		/* I/O SAPIC */
+	ACPI_APIC_LSAPIC,		/* Local SAPIC */
+	ACPI_APIC_PLATFORM_IRQ_SRC,	/* Platform interrupt sources */
+	ACPI_APIC_LX2APIC,		/* Processor local x2APIC */
+	ACPI_APIC_LX2APIC_NMI,		/* Local x2APIC NMI */
+};
+
+/* MADT: Processor Local APIC Structure */
+
+#define LOCAL_APIC_FLAG_ENABLED		BIT(0)
+
+struct acpi_madt_lapic {
+	u8 type;		/* Type (0) */
+	u8 length;		/* Length in bytes (8) */
+	u8 processor_id;	/* ACPI processor ID */
+	u8 apic_id;		/* Local APIC ID */
+	u32 flags;		/* Local APIC flags */
+};
+
+/* MADT: I/O APIC Structure */
+struct acpi_madt_ioapic {
+	u8 type;		/* Type (1) */
+	u8 length;		/* Length in bytes (12) */
+	u8 ioapic_id;		/* I/O APIC ID */
+	u8 reserved;
+	u32 ioapic_addr;	/* I/O APIC address */
+	u32 gsi_base;		/* Global system interrupt base */
+};
+
+/* MADT: Interrupt Source Override Structure */
+struct __packed acpi_madt_irqoverride {
+	u8 type;		/* Type (2) */
+	u8 length;		/* Length in bytes (10) */
+	u8 bus;			/* ISA (0) */
+	u8 source;		/* Bus-relative int. source (IRQ) */
+	u32 gsirq;		/* Global system interrupt */
+	u16 flags;		/* MPS INTI flags */
+};
+
+/* MADT: Local APIC NMI Structure */
+struct __packed acpi_madt_lapic_nmi {
+	u8 type;		/* Type (4) */
+	u8 length;		/* Length in bytes (6) */
+	u8 processor_id;	/* ACPI processor ID */
+	u16 flags;		/* MPS INTI flags */
+	u8 lint;		/* Local APIC LINT# */
+};
+
+/* MCFG (PCI Express MMIO config space BAR description table) */
+struct acpi_mcfg {
+	struct acpi_table_header header;
+	u8 reserved[8];
+};
+
+struct acpi_mcfg_mmconfig {
+	u32 base_address_l;
+	u32 base_address_h;
+	u16 pci_segment_group_number;
+	u8 start_bus_number;
+	u8 end_bus_number;
+	u8 reserved[4];
+};
+
+/* PM1_CNT bit defines */
+#define PM1_CNT_SCI_EN		BIT(0)
+
+/* ACPI global NVS structure */
+struct acpi_global_nvs;
+
+/* CSRT (Core System Resource Table) */
+struct acpi_csrt {
+	struct acpi_table_header header;
+};
+
+struct acpi_csrt_group {
+	u32 length;
+	u32 vendor_id;
+	u32 subvendor_id;
+	u16 device_id;
+	u16 subdevice_id;
+	u16 revision;
+	u16 reserved;
+	u32 shared_info_length;
+};
+
+struct acpi_csrt_shared_info {
+	u16 major_version;
+	u16 minor_version;
+	u32 mmio_base_low;
+	u32 mmio_base_high;
+	u32 gsi_interrupt;
+	u8 interrupt_polarity;
+	u8 interrupt_mode;
+	u8 num_channels;
+	u8 dma_address_width;
+	u16 base_request_line;
+	u16 num_handshake_signals;
+	u32 max_block_size;
+};
+
+/* DBG2 definitions are partially used for SPCR interface_type */
+
+/* Types for port_type field */
+
+#define ACPI_DBG2_SERIAL_PORT		0x8000
+#define ACPI_DBG2_1394_PORT		0x8001
+#define ACPI_DBG2_USB_PORT		0x8002
+#define ACPI_DBG2_NET_PORT		0x8003
+
+/* Subtypes for port_subtype field */
+
+#define ACPI_DBG2_16550_COMPATIBLE	0x0000
+#define ACPI_DBG2_16550_SUBSET		0x0001
+#define ACPI_DBG2_ARM_PL011		0x0003
+#define ACPI_DBG2_ARM_SBSA_32BIT	0x000D
+#define ACPI_DBG2_ARM_SBSA_GENERIC	0x000E
+#define ACPI_DBG2_ARM_DCC		0x000F
+#define ACPI_DBG2_BCM2835		0x0010
+
+#define ACPI_DBG2_1394_STANDARD		0x0000
+
+#define ACPI_DBG2_USB_XHCI		0x0000
+#define ACPI_DBG2_USB_EHCI		0x0001
+
+#define ACPI_DBG2_UNKNOWN		0x00FF
+
+/* SPCR (Serial Port Console Redirection table) */
+struct __packed acpi_spcr {
+	struct acpi_table_header header;
+	u8 interface_type;
+	u8 reserved[3];
+	struct acpi_gen_regaddr serial_port;
+	u8 interrupt_type;
+	u8 pc_interrupt;
+	u32 interrupt;		/* Global system interrupt */
+	u8 baud_rate;
+	u8 parity;
+	u8 stop_bits;
+	u8 flow_control;
+	u8 terminal_type;
+	u8 reserved1;
+	u16 pci_device_id;	/* Must be 0xffff if not PCI device */
+	u16 pci_vendor_id;	/* Must be 0xffff if not PCI device */
+	u8 pci_bus;
+	u8 pci_device;
+	u8 pci_function;
+	u32 pci_flags;
+	u8 pci_segment;
+	u32 reserved2;
+};
+
+#include <asm/acpi_table.h>
+
+#endif /* __ACPI_TABLE_H__ */
lib/efi_loader/efi_acpi.c +1 -1
index a4e5e53d15e..9ae9c25c51e 100644
@@ -7,7 +7,7 @@
 
 #include <common.h>
 #include <efi_loader.h>
-#include <asm/acpi_table.h>
+#include <acpi/acpi_table.h>
 
 static const efi_guid_t acpi_guid = EFI_ACPI_TABLE_GUID;
 
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#50 lore raw

PATCH v3 17/29 x86: Move acpi_table header to main include/ directory

Hi Andy,

Quoted 7 lines · Andy Shevchenko
Quoted 2 lines · Simon Glass

This file is potentially useful to other architectures saddled with ACPI so move most of its contents to a common location.

It's not just potentially, it's definitely useful.
But this makes me think, why we don't incorporate ACPICA headers as is?

Due to code style and the vast amount of unused code.

Quoted 14 lines
Quoted 14 lines

---

Changes in v3:
- Add forward declarations for the functions
- Move acpi_table.h to include/acpi
- Update commit message to say that we move most of its contents

Changes in v2: None

Quoted 15 lines
Quoted 6 lines
 arch/x86/cpu/baytrail/acpi.c      |   2 +-
 arch/x86/cpu/cpu.c                |   2 +-
 arch/x86/cpu/quark/acpi.c         |   2 +-
 arch/x86/cpu/tangier/acpi.c       |   4 +-
 arch/x86/include/asm/acpi_table.h | 381 +----------------------------
 arch/x86/lib/acpi.c               |   2 +-
Quoted 9 lines
 arch/x86/lib/acpi_s3.c            |   2 +-
 arch/x86/lib/acpi_table.c         |   2 +-
 arch/x86/lib/tables.c             |   2 +-
 arch/x86/lib/zimage.c             |   2 +-
 include/acpi/acpi_table.h         | 394 ++++++++++++++++++++++++++++++
 lib/efi_loader/efi_acpi.c         |   2 +-
 12 files changed, 412 insertions(+), 385 deletions(-)
 create mode 100644 include/acpi/acpi_table.h

Regards,
Simon

#51 lore raw

PATCH v3 17/29 x86: Move acpi_table header to main include/ directory

Quoted 11 lines · Simon Glass
Quoted 7 lines · Andy Shevchenko
Quoted 2 lines · Simon Glass

This file is potentially useful to other architectures saddled with ACPI so move most of its contents to a common location.

It's not just potentially, it's definitely useful.
But this makes me think, why we don't incorporate ACPICA headers as is?

Due to code style and the vast amount of unused code.

So, it will be interesting journey over the same problems then. But it's your problem to maintain this :-)

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#52 lore raw

PATCH v3 17/29 x86: Move acpi_table header to main include/ directory

Hi Andy,

Quoted 9 lines · Andy Shevchenko
Quoted 7 lines · Simon Glass
Quoted 5 lines · Andy Shevchenko
Quoted 2 lines · Simon Glass

This file is potentially useful to other architectures saddled with ACPI so move most of its contents to a common location.

Quoted 7 lines
Quoted 4 lines
Quoted 2 lines

It's not just potentially, it's definitely useful.
But this makes me think, why we don't incorporate ACPICA headers as is?

Due to code style and the vast amount of unused code.

So, it will be interesting journey over the same problems then. But it's your problem to maintain this :-)

If Intel is interested in maintaining this in U-Boot let us know. So long as the code style is right as we don't add dead code it should be fine.

Regards,
Simon

#53 lore raw

PATCH v3 18/29 acpi: Add an __ACPI__ preprocessor symbol

In reply to Simon Glass

The ASL compiler cannot handle C structures and the like so needs some sort of header guard around these.

We already have an __ASSEMBLY__ #define but it seems best to create a new one for ACPI since the rules may be different.

Add the check to a few files that ACPI always includes.

---

Changes in v3: None
Changes in v2: None

 include/acpi/acpi_table.h | 4 ++++
 include/dm/acpi.h         | 4 ++++
 scripts/Makefile.lib      | 4 ++--
 3 files changed, 10 insertions(+), 2 deletions(-)
include/acpi/acpi_table.h +4 -0
index b4404a632c0..dd748958136 100644
@@ -21,6 +21,8 @@
 #define ACPI_RSDP_REV_ACPI_1_0	0
 #define ACPI_RSDP_REV_ACPI_2_0	2
 
+#if !defined(__ACPI__)
+
 /*
  * RSDP (Root System Description Pointer)
  * Note: ACPI 1.0 didn't have length, xsdt_address, and ext_checksum
@@ -389,6 +391,8 @@ struct __packed acpi_spcr {
 	u32 reserved2;
 };
 
+#endif /* !__ACPI__*/
+
 #include <asm/acpi_table.h>
 
 #endif /* __ACPI_TABLE_H__ */
include/dm/acpi.h +4 -0
index c13d0c8448a..ee218b44e85 100644
@@ -22,6 +22,8 @@
 /* Length of an ACPI name string including nul terminator */
 #define ACPI_NAME_MAX	(ACPI_NAME_LEN + 1)
 
+#if !defined(__ACPI__)
+
 /**
  * struct acpi_ops - ACPI operations supported by driver model
  */
@@ -70,4 +72,6 @@ int acpi_get_name(const struct udevice *dev, char *out_name);
  */
 int acpi_copy_name(char *out_name, const char *name);
 
+#endif /* __ACPI__ */
+
 #endif
scripts/Makefile.lib +2 -2
index 30f392fdfb0..aebdb38d0d2 100644
@@ -413,8 +413,8 @@ ASL_TMP = $(patsubst %.c,%.asl.tmp,$@)
 
 quiet_cmd_acpi_c_asl= ASL     $<
 cmd_acpi_c_asl=         \
-	$(CPP) -x assembler-with-cpp -D__ASSEMBLY__ -P $(UBOOTINCLUDE) \
-		-o $(ASL_TMP) $< && \
+	$(CPP) -x assembler-with-cpp -D__ASSEMBLY__ -D__ACPI__ \
+		-P $(UBOOTINCLUDE) -o $(ASL_TMP) $< && \
 	iasl -p $@ -tc $(ASL_TMP) $(if $(KBUILD_VERBOSE:1=), >/dev/null) && \
 	mv $(patsubst %.c,%.hex,$@) $@
 
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#54 lore raw

PATCH v3 19/29 acpi: Add a central location for table version numbers

In reply to Simon Glass

Each ACPI table has its own version number. Add the version numbers in a single function so we can keep them consistent and easily see what versions are supported.

Start a new acpi_table file in a generic directory to house this function. We can move things over to this file from x86 as needed.

---

Changes in v3:
- Fix a few typos
- Fix file comment for acpi_table.c

Changes in v2:
- Move the sandbox acpi_table.h header file to an earlier patch
- Use #defines for MADT and MCFG version numbers

 include/acpi/acpi_table.h | 61 +++++++++++++++++++++++++++++++++++++++
 lib/Makefile              |  1 +
 lib/acpi/Makefile         |  4 +++
 lib/acpi/acpi_table.c     | 60 ++++++++++++++++++++++++++++++++++++++
 test/dm/acpi.c            | 14 +++++++++
 5 files changed, 140 insertions(+)
 create mode 100644 lib/acpi/Makefile
 create mode 100644 lib/acpi/acpi_table.c
include/acpi/acpi_table.h +61 -0
index dd748958136..ccf6fa04dbe 100644
@@ -202,6 +202,26 @@ struct __packed acpi_fadt {
 	struct acpi_gen_regaddr x_gpe1_blk;
 };
 
+/* FADT TABLE Revision values */
+#define ACPI_FADT_REV_ACPI_1_0		1
+#define ACPI_FADT_REV_ACPI_2_0		3
+#define ACPI_FADT_REV_ACPI_3_0		4
+#define ACPI_FADT_REV_ACPI_4_0		4
+#define ACPI_FADT_REV_ACPI_5_0		5
+#define ACPI_FADT_REV_ACPI_6_0		6
+
+/* MADT TABLE Revision values */
+#define ACPI_MADT_REV_ACPI_3_0		2
+#define ACPI_MADT_REV_ACPI_4_0		3
+#define ACPI_MADT_REV_ACPI_5_0		3
+#define ACPI_MADT_REV_ACPI_6_0		5
+
+#define ACPI_MCFG_REV_ACPI_3_0		1
+
+/* IVRS Revision Field */
+#define IVRS_FORMAT_FIXED	0x01	/* Type 10h & 11h only */
+#define IVRS_FORMAT_MIXED	0x02	/* Type 10h, 11h, & 40h */
+
 /* FACS flags */
 #define ACPI_FACS_S4BIOS_F		BIT(0)
 #define ACPI_FACS_64BIT_WAKE_F		BIT(1)
@@ -391,6 +411,47 @@ struct __packed acpi_spcr {
 	u32 reserved2;
 };
 
+/* Tables defined by ACPI and generated by U-Boot */
+enum acpi_tables {
+	ACPITAB_BERT,
+	ACPITAB_DBG2,
+	ACPITAB_DMAR,
+	ACPITAB_DSDT,
+	ACPITAB_FACS,
+	ACPITAB_FADT,
+	ACPITAB_HEST,
+	ACPITAB_HPET,
+	ACPITAB_IVRS,
+	ACPITAB_MADT,
+	ACPITAB_MCFG,
+	ACPITAB_RSDP,
+	ACPITAB_RSDT,
+	ACPITAB_SLIT,
+	ACPITAB_SRAT,
+	ACPITAB_SSDT,
+	ACPITAB_TCPA,
+	ACPITAB_TPM2,
+	ACPITAB_XSDT,
+	ACPITAB_ECDT,
+
+	/* Additional proprietary tables */
+	ACPITAB_VFCT,
+	ACPITAB_NHLT,
+	ACPITAB_SPMI,
+
+	ACPITAB_COUNT,
+};
+
+/**
+ * acpi_get_table_revision() - Get the revision number generated for a table
+ *
+ * This keeps the version-number information in one place
+ *
+ * @table: ACPI table to check
+ * @return version number that U-Boot generates
+ */
+int acpi_get_table_revision(enum acpi_tables table);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
lib/Makefile +1 -0
index 15259d0473c..f737eb559fc 100644
@@ -5,6 +5,7 @@
 
 ifndef CONFIG_SPL_BUILD
 
+obj-$(CONFIG_$(SPL_)ACPIGEN) += acpi/
 obj-$(CONFIG_EFI) += efi/
 obj-$(CONFIG_EFI_LOADER) += efi_driver/
 obj-$(CONFIG_EFI_LOADER) += efi_loader/
lib/acpi/Makefile new file +4 -0
new file mode 100644
index 00000000000..660491ef711
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0+
+#
+
+obj-y += acpi_table.o
lib/acpi/acpi_table.c new file +60 -0

60 lines not on this page · raw message

test/dm/acpi.c +14 -0
index 429993b9cc0..e5510e4323b 100644
@@ -8,6 +8,7 @@
 
 #include <common.h>
 #include <dm.h>
+#include <acpi/acpi_table.h>
 #include <dm/acpi.h>
 #include <dm/test.h>
 #include <test/ut.h>
@@ -53,3 +54,16 @@ static int dm_test_acpi_get_name(struct unit_test_state *uts)
 	return 0;
 }
 DM_TEST(dm_test_acpi_get_name, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
+/* Test acpi_get_table_revision() */
+static int dm_test_acpi_get_table_revision(struct unit_test_state *uts)
+{
+	ut_asserteq(1, acpi_get_table_revision(ACPITAB_MCFG));
+	ut_asserteq(2, acpi_get_table_revision(ACPITAB_RSDP));
+	ut_asserteq(4, acpi_get_table_revision(ACPITAB_TPM2));
+	ut_asserteq(-EINVAL, acpi_get_table_revision(ACPITAB_COUNT));
+
+	return 0;
+}
+DM_TEST(dm_test_acpi_get_table_revision,
+	DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Reviewed-by on v3 19/29
#55 lore raw

Antwort: [PATCH v3 19/29] acpi: Add a central location for table version numbers

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 42 lines

Betreff: [PATCH v3 19/29] acpi: Add a central location for table version numbers

Each ACPI table has its own version number. Add the version numbers in a
single function so we can keep them consistent and easily see what versions are supported.

Start a new acpi_table file in a generic directory to house this function.
We can move things over to this file from x86 as needed.

---

Changes in v3:
- Fix a few typos
- Fix file comment for acpi_table.c

Changes in v2:
- Move the sandbox acpi_table.h header file to an earlier patch
- Use #defines for MADT and MCFG version numbers

include/acpi/acpi_table.h | 61
+++++++++++++++++++++++++++++++++++++++

lib/Makefile              |  1 +
lib/acpi/Makefile         |  4 +++
lib/acpi/acpi_table.c     | 60
++++++++++++++++++++++++++++++++++++++
test/dm/acpi.c            | 14 +++++++++
5 files changed, 140 insertions(+)

create mode 100644 lib/acpi/Makefile
create mode 100644 lib/acpi/acpi_table.c

include/acpi/acpi_table.h +0 -0
index dd748958136..ccf6fa04dbe 100644
@@ -202,6 +202,26 @@ struct __packed acpi_fadt {
 	struct acpi_gen_regaddr x_gpe1_blk;
 };
 

[snip]

Quoted 10 lines

#include <asm/acpi_table.h>

lib/Makefile +0 -0
index 15259d0473c..f737eb559fc 100644
@@ -5,6 +5,7 @@
 

ifndef CONFIG_SPL_BUILD

+obj-$(CONFIG_$(SPL_)ACPIGEN) += acpi/

Could you please help me understand why $(SPL_) is used here?
You have mentioned in your answer of the v2 patch explicitly that you should use $SPL_ here and have changed the patch accordingly in v3, so I trust that it makes sense. But I fail to understand how this works. The newly added line is within "ifndef CONFIG_SPL_BUILD", in which case $SPL_ should alwys be empty anyway. At least that is how I understood the assigment to SPL_ in Makefile.spl.

Quoted 3 lines

obj-$(CONFIG_EFI) += efi/
obj-$(CONFIG_EFI_LOADER) += efi_driver/
obj-$(CONFIG_EFI_LOADER) += efi_loader/

[snip]

#56 lore raw

PATCH v3 19/29 acpi: Add a central location for table version numbers

Hi Wolfgang,

Quoted 60 lines · Wolfgang Wallner

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 42 lines

Betreff: [PATCH v3 19/29] acpi: Add a central location for table version numbers

Each ACPI table has its own version number. Add the version numbers in a
single function so we can keep them consistent and easily see what versions are supported.

Start a new acpi_table file in a generic directory to house this function.
We can move things over to this file from x86 as needed.

---

Changes in v3:
- Fix a few typos
- Fix file comment for acpi_table.c

Changes in v2:
- Move the sandbox acpi_table.h header file to an earlier patch
- Use #defines for MADT and MCFG version numbers

include/acpi/acpi_table.h | 61
+++++++++++++++++++++++++++++++++++++++

lib/Makefile              |  1 +
lib/acpi/Makefile         |  4 +++
lib/acpi/acpi_table.c     | 60
++++++++++++++++++++++++++++++++++++++
test/dm/acpi.c            | 14 +++++++++
5 files changed, 140 insertions(+)

create mode 100644 lib/acpi/Makefile
create mode 100644 lib/acpi/acpi_table.c

include/acpi/acpi_table.h +0 -0
index dd748958136..ccf6fa04dbe 100644
@@ -202,6 +202,26 @@ struct __packed acpi_fadt {
      struct acpi_gen_regaddr x_gpe1_blk;
 };
 

[snip]

Quoted 9 lines

#include <asm/acpi_table.h>

lib/Makefile +0 -0
index 15259d0473c..f737eb559fc 100644
@@ -5,6 +5,7 @@
 

ifndef CONFIG_SPL_BUILD

Quoted 8 lines
Quoted 1 line

+obj-$(CONFIG_$(SPL_)ACPIGEN) += acpi/

Could you please help me understand why $(SPL_) is used here?
You have mentioned in your answer of the v2 patch explicitly that you should use $SPL_ here and have changed the patch accordingly in v3, so I trust that it makes sense. But I fail to understand how this works. The newly added line is within "ifndef CONFIG_SPL_BUILD", in which case $SPL_ should alwys be empty anyway. At least that is how I understood the assigment to SPL_ in Makefile.spl.

Yes you are right. It is unnecessary. I'll move it down further.

Quoted 8 lines
Quoted 4 lines

obj-$(CONFIG_EFI) += efi/
obj-$(CONFIG_EFI_LOADER) += efi_driver/
obj-$(CONFIG_EFI_LOADER) += efi_loader/

[snip]

#57 lore raw

PATCH v3 19/29 acpi: Add a central location for table version numbers

In reply to Simon Glass
Quoted 6 lines · Simon Glass

Each ACPI table has its own version number. Add the version numbers in a single function so we can keep them consistent and easily see what versions are supported.

Start a new acpi_table file in a generic directory to house this function. We can move things over to this file from x86 as needed.

...

include/acpi/acpi_table.h struct __packed acpi_fadt { · line 217

Quoted 13 lines
include/acpi/acpi_table.h +13 -0
+/* FADT TABLE Revision values */
+#define ACPI_FADT_REV_ACPI_1_0		1
+#define ACPI_FADT_REV_ACPI_2_0		3
+#define ACPI_FADT_REV_ACPI_3_0		4
+#define ACPI_FADT_REV_ACPI_4_0		4
+#define ACPI_FADT_REV_ACPI_5_0		5
+#define ACPI_FADT_REV_ACPI_6_0		6
+
+/* MADT TABLE Revision values */
+#define ACPI_MADT_REV_ACPI_3_0		2
+#define ACPI_MADT_REV_ACPI_4_0		3
+#define ACPI_MADT_REV_ACPI_5_0		3
+#define ACPI_MADT_REV_ACPI_6_0		5

Is it for real this fancy numbering? I don't remember spec by heart, perhaps this needs to be elaborated in the comment.

...

Quoted 13 lines
include/acpi/acpi_table.h +12 -0
+/* Tables defined by ACPI and generated by U-Boot */
+enum acpi_tables {
 
+	ACPITAB_BERT,
+	ACPITAB_DBG2,
+	ACPITAB_DMAR,
+	ACPITAB_DSDT,
+	ACPITAB_FACS,
+	ACPITAB_FADT,
+	ACPITAB_HEST,
+	ACPITAB_HPET,
+	ACPITAB_IVRS,
+	ACPITAB_MADT,

include/acpi/acpi_table.h struct __packed acpi_spcr { · line 435

Quoted 10 lines
include/acpi/acpi_table.h +10 -0
+	ACPITAB_MCFG,
+	ACPITAB_RSDP,
+	ACPITAB_RSDT,
+	ACPITAB_SLIT,
+	ACPITAB_SRAT,
+	ACPITAB_SSDT,
+	ACPITAB_TCPA,
+	ACPITAB_TPM2,
+	ACPITAB_XSDT,
+	ACPITAB_ECDT,

This should be cleaned according to uefi.org, i.e. not all above tables are from ACPI spec. And thus, should be rather in the below section.

include/acpi/acpi_table.h struct __packed acpi_spcr { · line 440

Quoted 5 lines
include/acpi/acpi_table.h +5 -0
+
+	/* Additional proprietary tables */
+	ACPITAB_VFCT,
+	ACPITAB_NHLT,
+	ACPITAB_SPMI,

Where is SPCR?

Quoted 3 lines
include/acpi/acpi_table.h +3 -0
+
+	ACPITAB_COUNT,
+};
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#58 lore raw

PATCH v3 19/29 acpi: Add a central location for table version numbers

Hi Andy,

Quoted 11 lines · Andy Shevchenko
Quoted 6 lines · Simon Glass

Each ACPI table has its own version number. Add the version numbers in a single function so we can keep them consistent and easily see what versions are supported.

Start a new acpi_table file in a generic directory to house this function. We can move things over to this file from x86 as needed.

...

Quoted 16 lines

include/acpi/acpi_table.h struct __packed acpi_fadt { · line 217

Quoted 13 lines
include/acpi/acpi_table.h +13 -0
+/* FADT TABLE Revision values */
+#define ACPI_FADT_REV_ACPI_1_0               1
+#define ACPI_FADT_REV_ACPI_2_0               3
+#define ACPI_FADT_REV_ACPI_3_0               4
+#define ACPI_FADT_REV_ACPI_4_0               4
+#define ACPI_FADT_REV_ACPI_5_0               5
+#define ACPI_FADT_REV_ACPI_6_0               6
+
+/* MADT TABLE Revision values */
+#define ACPI_MADT_REV_ACPI_3_0               2
+#define ACPI_MADT_REV_ACPI_4_0               3
+#define ACPI_MADT_REV_ACPI_5_0               3
+#define ACPI_MADT_REV_ACPI_6_0               5

Is it for real this fancy numbering? I don't remember spec by heart, perhaps this needs to be elaborated in the comment.

Will add

Quoted 27 lines

...

Quoted 13 lines
include/acpi/acpi_table.h +12 -0
+/* Tables defined by ACPI and generated by U-Boot */
+enum acpi_tables {
 
+     ACPITAB_BERT,
+     ACPITAB_DBG2,
+     ACPITAB_DMAR,
+     ACPITAB_DSDT,
+     ACPITAB_FACS,
+     ACPITAB_FADT,
+     ACPITAB_HEST,
+     ACPITAB_HPET,
+     ACPITAB_IVRS,
+     ACPITAB_MADT,

include/acpi/acpi_table.h struct __packed acpi_spcr { · line 435

Quoted 10 lines
include/acpi/acpi_table.h +10 -0
+     ACPITAB_MCFG,
+     ACPITAB_RSDP,
+     ACPITAB_RSDT,
+     ACPITAB_SLIT,
+     ACPITAB_SRAT,
+     ACPITAB_SSDT,
+     ACPITAB_TCPA,
+     ACPITAB_TPM2,
+     ACPITAB_XSDT,
+     ACPITAB_ECDT,
Quoted 2 lines

This should be cleaned according to uefi.org, i.e. not all above tables are from ACPI spec. And thus, should be rather in the below section.

I mean things defined or reserved by the spec, so will update the comment.

Quoted 9 lines

include/acpi/acpi_table.h struct __packed acpi_spcr { · line 440

Quoted 6 lines
include/acpi/acpi_table.h +5 -0
+
+     /* Additional proprietary tables */
+     ACPITAB_VFCT,
+     ACPITAB_NHLT,
+     ACPITAB_SPMI,

Where is SPCR?

Will add.

Quoted 9 lines

include/acpi/acpi_table.h struct __packed acpi_spcr { · line 443

Quoted 3 lines
include/acpi/acpi_table.h +3 -0
+
+     ACPITAB_COUNT,
+};

--
With Best Regards,
Andy Shevchenko

#59 lore raw

PATCH v3 20/29 acpi: Add support for DMAR

In reply to Simon Glass

The DMA Remapping Reporting (DMAR) table contains information about DMA remapping.

Add a version simple version of this table with only the minimum fields filled out. i.e. no entries.

---

Changes in v3:
- Add missing error check in acpi_create_dmar()
- Drop duplicate assert
- Fix DMA_ typo
- Make use of BIT()

Changes in v2:
- Drop two unnecessary __packed
- Move __packed to after struct

 include/acpi/acpi_table.h | 57 +++++++++++++++++++++++++++++++++++++++
 lib/acpi/acpi_table.c     | 28 +++++++++++++++++++
 test/dm/acpi.c            | 13 +++++++++
 3 files changed, 98 insertions(+)
include/acpi/acpi_table.h +57 -0
index ccf6fa04dbe..10d6ccdf090 100644
@@ -21,6 +21,9 @@
 #define ACPI_RSDP_REV_ACPI_1_0	0
 #define ACPI_RSDP_REV_ACPI_2_0	2
 
+/* TODO(sjg at chromium.org): Figure out how to get compiler revision */
+#define ASL_REVISION	0
+
 #if !defined(__ACPI__)
 
 /*
@@ -360,6 +363,51 @@ struct acpi_csrt_shared_info {
 	u32 max_block_size;
 };
 
+enum dmar_type {
+	DMAR_DRHD = 0,
+	DMAR_RMRR = 1,
+	DMAR_ATSR = 2,
+	DMAR_RHSA = 3,
+	DMAR_ANDD = 4
+};
+
+enum {
+	DRHD_INCLUDE_PCI_ALL = BIT(0)
+};
+
+enum dmar_flags {
+	DMAR_INTR_REMAP			= BIT(0),
+	DMAR_X2APIC_OPT_OUT		= BIT(1),
+	DMAR_CTRL_PLATFORM_OPT_IN_FLAG	= BIT(2),
+};
+
+struct dmar_entry {
+	u16 type;
+	u16 length;
+	u8 flags;
+	u8 reserved;
+	u16 segment;
+	u64 bar;
+};
+
+struct dmar_rmrr_entry {
+	u16 type;
+	u16 length;
+	u16 reserved;
+	u16 segment;
+	u64 bar;
+	u64 limit;
+};
+
+/* DMAR (DMA Remapping Reporting Structure) */
+struct __packed acpi_dmar {
+	struct acpi_table_header header;
+	u8 host_address_width;
+	u8 flags;
+	u8 reserved[10];
+	struct dmar_entry structure[0];
+};
+
 /* DBG2 definitions are partially used for SPCR interface_type */
 
 /* Types for port_type field */
@@ -452,6 +500,15 @@ enum acpi_tables {
  */
 int acpi_get_table_revision(enum acpi_tables table);
 
+/**
+ * acpi_create_dmar() - Create a DMA Remapping Reporting (DMAR) table
+ *
+ * @dmar: Place to put the table
+ * @flags: DMAR flags to use
+ * @return 0 if OK, -ve on error
+ */
+int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
lib/acpi/acpi_table.c +28 -0
index 30639e4dfbd..3156fb93503 100644
@@ -7,6 +7,34 @@
 
 #include <common.h>
 #include <acpi/acpi_table.h>
+#include <dm.h>
+#include <cpu.h>
+
+int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags)
+{
+	struct acpi_table_header *header = &dmar->header;
+	struct cpu_info info;
+	struct udevice *cpu;
+	int ret;
+
+	ret = uclass_first_device(UCLASS_CPU, &cpu);
+	if (ret)
+		return log_msg_ret("cpu", ret);
+	ret = cpu_get_info(cpu, &info);
+	if (ret)
+		return log_msg_ret("info", ret);
+	memset((void *)dmar, 0, sizeof(struct acpi_dmar));
+
+	/* Fill out header fields. */
+	acpi_fill_header(&dmar->header, "DMAR");
+	header->length = sizeof(struct acpi_dmar);
+	header->revision = acpi_get_table_revision(ACPITAB_DMAR);
+
+	dmar->host_address_width = info.address_width - 1;
+	dmar->flags = flags;
+
+	return 0;
+}
 
 int acpi_get_table_revision(enum acpi_tables table)
 {
test/dm/acpi.c +13 -0
index e5510e4323b..21ed7c74409 100644
@@ -67,3 +67,16 @@ static int dm_test_acpi_get_table_revision(struct unit_test_state *uts)
 }
 DM_TEST(dm_test_acpi_get_table_revision,
 	DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
+/* Test acpi_create_dmar() */
+static int dm_test_acpi_create_dmar(struct unit_test_state *uts)
+{
+	struct acpi_dmar dmar;
+
+	ut_assertok(acpi_create_dmar(&dmar, DMAR_INTR_REMAP));
+	ut_asserteq(DMAR_INTR_REMAP, dmar.flags);
+	ut_asserteq(32 - 1, dmar.host_address_width);
+
+	return 0;
+}
+DM_TEST(dm_test_acpi_create_dmar, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Reviewed-by on v3 20/29
#60 lore raw

Antwort: [PATCH v3 20/29] acpi: Add support for DMAR

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 24 lines

Betreff: [PATCH v3 20/29] acpi: Add support for DMAR

The DMA Remapping Reporting (DMAR) table contains information about DMA
remapping.

Add a version simple version of this table with only the minimum fields
filled out. i.e. no entries.

---

Changes in v3:
- Add missing error check in acpi_create_dmar()
- Drop duplicate assert
- Fix DMA_ typo
- Make use of BIT()

Changes in v2:
- Drop two unnecessary __packed
- Move __packed to after struct

Quoted 5 lines

include/acpi/acpi_table.h | 57
+++++++++++++++++++++++++++++++++++++++
lib/acpi/acpi_table.c | 28 +++++++++++++++++++
test/dm/acpi.c | 13 +++++++++

3 files changed, 98 insertions(+)
#61 lore raw

PATCH v3 20/29 acpi: Add support for DMAR

In reply to Simon Glass

include/acpi/acpi_table.h line 25

Quoted 8 lines · Simon Glass

The DMA Remapping Reporting (DMAR) table contains information about DMA remapping.

Add a version simple version of this table with only the minimum fields filled out. i.e. no entries.

include/acpi/acpi_table.h +2 -0
+/* TODO(sjg at chromium.org): Figure out how to get compiler revision */
+#define ASL_REVISION	0

Why do you need this?

I don't see any users here either.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#62 lore raw

PATCH v3 21/29 test: Add hexdump.h to the unit test header

In reply to Simon Glass

Since ut_asserteq_mem() uses bin2hex() we should include this header in ut.h to avoid errors. Add it.

---

Changes in v3:
- Add new patch to add hexdump.h to the unit test header

Changes in v2: None

 include/test/ut.h | 1 +
 1 file changed, 1 insertion(+)
include/test/ut.h +1 -0
index 04df8ba3af3..39d15953ed3 100644
@@ -8,6 +8,7 @@
 #ifndef __TEST_UT_H
 #define __TEST_UT_H
 
+#include <hexdump.h>
 #include <linux/err.h>
 
 struct unit_test_state;
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
Reviewed-by on v3 21/29
#63 lore raw

Antwort: [PATCH v3 21/29] test: Add hexdump.h to the unit test header

Hi Simon,

-----"Simon Glass" <[email protected]> schrieb: -----

Quoted 8 lines

Since ut_asserteq_mem() uses bin2hex() we should include this header in
ut.h to avoid errors. Add it.

---

Quoted 8 lines

Changes in v3:
- Add new patch to add hexdump.h to the unit test header

Changes in v2: None

include/test/ut.h | 1 +

1 file changed, 1 insertion(+)
#64 lore raw

PATCH v3 22/29 acpi: Add a method to write tables for a device

In reply to Simon Glass

A device may want to write out ACPI tables to describe itself to Linux. Add a method to permit this.

---

Changes in v3: None
Changes in v2:
- Drop definition of ACPI_TABLE_CREATOR
- Generalise the ACPI function recursion with acpi_recurse_method()
- Make _acpi_write_dev_tables() static and switch argument order

 arch/sandbox/dts/test.dts |  4 +++
 arch/x86/lib/acpi_table.c |  9 -----
 drivers/core/acpi.c       | 61 +++++++++++++++++++++++++++++++++
 include/acpi/acpi_table.h | 10 ++++++
 include/dm/acpi.h         | 30 +++++++++++++++++
 lib/acpi/acpi_table.c     | 10 ++++++
 test/dm/acpi.c            | 71 +++++++++++++++++++++++++++++++++++++++
 7 files changed, 186 insertions(+), 9 deletions(-)
arch/sandbox/dts/test.dts +4 -0
index 5fa951ad4b6..1204c14b079 100644
@@ -210,6 +210,10 @@
 		compatible = "denx,u-boot-acpi-test";
 	};
 
+	acpi-test2 {
+		compatible = "denx,u-boot-acpi-test";
+	};
+
 	clocks {
 		clk_fixed: clk-fixed {
 			compatible = "fixed-clock";
arch/x86/lib/acpi_table.c +0 -9
index a00663d4d17..f98d6d17b3e 100644
@@ -60,15 +60,6 @@ static void acpi_write_rsdp(struct acpi_rsdp *rsdp, struct acpi_rsdt *rsdt,
 			sizeof(struct acpi_rsdp));
 }
 
-void acpi_fill_header(struct acpi_table_header *header, char *signature)
-{
-	memcpy(header->signature, signature, 4);
-	memcpy(header->oem_id, OEM_ID, 6);
-	memcpy(header->oem_table_id, OEM_TABLE_ID, 8);
-	header->oem_revision = U_BOOT_BUILD_DATE;
-	memcpy(header->aslc_id, ASLC_ID, 4);
-}
-
 static void acpi_write_rsdt(struct acpi_rsdt *rsdt)
 {
 	struct acpi_table_header *header = &(rsdt->header);
drivers/core/acpi.c +61 -0

72 lines not on this page · raw message

include/acpi/acpi_table.h +10 -0
index 10d6ccdf090..268e36b67e1 100644
@@ -509,6 +509,16 @@ int acpi_get_table_revision(enum acpi_tables table);
  */
 int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags);
 
+/**
+ * acpi_fill_header() - Set up a new table header
+ *
+ * This sets all fields except length, revision, checksum and aslc_revision
+ *
+ * @header: ACPI header to update
+ * @signature: Table signature to use (4 characters)
+ */
+void acpi_fill_header(struct acpi_table_header *header, char *signature);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
include/dm/acpi.h +30 -0
index ee218b44e85..9a0a437dd03 100644
@@ -24,6 +24,17 @@
 
 #if !defined(__ACPI__)
 
+/**
+ * struct acpi_ctx - Context used for writing ACPI tables
+ *
+ * This contains a few useful pieces of information used when writing
+ *
+ * @current: Current address for writing
+ */
+struct acpi_ctx {
+	void *current;
+};
+
 /**
  * struct acpi_ops - ACPI operations supported by driver model
  */
@@ -38,6 +49,15 @@ struct acpi_ops {
 	 *	other error
 	 */
 	int (*get_name)(const struct udevice *dev, char *out_name);
+
+	/**
+	 * write_tables() - Write out any tables required by this device
+	 *
+	 * @dev: Device to write
+	 * @ctx: ACPI context to use
+	 * @return 0 if OK, -ve on error
+	 */
+	int (*write_tables)(const struct udevice *dev, struct acpi_ctx *ctx);
 };
 
 #define device_get_acpi_ops(dev)	((dev)->driver->acpi_ops)
@@ -72,6 +92,16 @@ int acpi_get_name(const struct udevice *dev, char *out_name);
  */
 int acpi_copy_name(char *out_name, const char *name);
 
+/**
+ * acpi_write_dev_tables() - Write ACPI tables required by devices
+ *
+ * This scans through all devices and tells them to write any tables they want
+ * to write.
+ *
+ * @return 0 if OK, -ve if any device returned an error
+ */
+int acpi_write_dev_tables(struct acpi_ctx *ctx);
+
 #endif /* __ACPI__ */
 
 #endif
lib/acpi/acpi_table.c +10 -0
index 3156fb93503..1710cde2c40 100644
@@ -9,6 +9,7 @@
 #include <acpi/acpi_table.h>
 #include <dm.h>
 #include <cpu.h>
+#include <version.h>
 
 int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags)
 {
@@ -86,3 +87,12 @@ int acpi_get_table_revision(enum acpi_tables table)
 		return -EINVAL;
 	}
 }
+
+void acpi_fill_header(struct acpi_table_header *header, char *signature)
+{
+	memcpy(header->signature, signature, 4);
+	memcpy(header->oem_id, OEM_ID, 6);
+	memcpy(header->oem_table_id, OEM_TABLE_ID, 8);
+	header->oem_revision = U_BOOT_BUILD_DATE;
+	memcpy(header->aslc_id, ASLC_ID, 4);
+}
test/dm/acpi.c +71 -0

92 lines not on this page · raw message

Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#65 lore raw

PATCH v3 22/29 acpi: Add a method to write tables for a device

drivers/core/acpi.c int acpi_get_name(const struct udevice *dev, char *out_name) · line 53

Quoted 13 lines · Simon Glass

A device may want to write out ACPI tables to describe itself to Linux. Add a method to permit this.

drivers/core/acpi.c +10 -0
+acpi_method acpi_get_method(struct udevice *dev, enum method_t method)
+{
+	struct acpi_ops *aops;
+
+	aops = device_get_acpi_ops(dev);
+	if (aops) {
+		switch (method) {
+		case METHOD_WRITE_TABLES:
+			return aops->write_tables;
+		}

Where is default?

drivers/core/acpi.c int acpi_get_name(const struct udevice *dev, char *out_name) · line 56

Quoted 3 lines
drivers/core/acpi.c +3 -0
+	}
+
+	return NULL;

Perhaps,

	if (!aops)
		return NULL;

	switch (method) {
		...
	default:
		return NULL;
	}
Quoted 1 line

+}

...

drivers/core/acpi.c int acpi_get_name(const struct udevice *dev, char *out_name) · line 68

Quoted 1 line
drivers/core/acpi.c +1 -0
+		log_debug("\n- %s %p\n", parent->name, func);

Leading '\n' in the messages is not good idea.
It might work nicely in U-Boot, but in general better to avoid.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#66 lore raw

PATCH v3 22/29 acpi: Add a method to write tables for a device

Hi Andy,

Quoted 5 lines · Andy Shevchenko
Quoted 2 lines · Simon Glass

A device may want to write out ACPI tables to describe itself to Linux. Add a method to permit this.

Quoted 12 lines

drivers/core/acpi.c int acpi_get_name(const struct udevice *dev, char *out_name) · line 53

Quoted 10 lines
drivers/core/acpi.c +10 -0
+acpi_method acpi_get_method(struct udevice *dev, enum method_t method)
+{
+     struct acpi_ops *aops;
+
+     aops = device_get_acpi_ops(dev);
+     if (aops) {
+             switch (method) {
+             case METHOD_WRITE_TABLES:
+                     return aops->write_tables;
+             }

Where is default?

There isn't one...that would be an error. We want to get a compiler warning if we leave something out.

Quoted 16 lines

drivers/core/acpi.c int acpi_get_name(const struct udevice *dev, char *out_name) · line 56

Quoted 4 lines
drivers/core/acpi.c +3 -0
+     }
+
+     return NULL;

Perhaps,

        if (!aops)
                return NULL;

        switch (method) {
                ...
        default:
                return NULL;
        }
Quoted 8 lines
Quoted 1 line

+}

...

Quoted 1 line

+ log_debug("\n- %s %p\n", parent->name, func);

Leading '\n' in the messages is not good idea.
It might work nicely in U-Boot, but in general better to avoid.

OK will separate it out.

Regards,
Simon

#67 lore raw

PATCH v3 23/29 acpi: Convert part of acpi_table to use acpi_ctx

In reply to Simon Glass

The current code uses an address but a pointer would result in fewer casts. Also it repeats the alignment code in a lot of places so this would be better done in a helper function.

Update write_acpi_tables() to make use of the new acpi_ctx structure, adding a few helpers to clean things up.

---

Changes in v3:
- Rename acpi_align_large() to acpi_align64()

Changes in v2: None

 arch/x86/lib/acpi_table.c | 88 +++++++++++++++++++--------------------
 include/acpi/acpi_table.h | 36 ++++++++++++++++
 lib/acpi/acpi_table.c     | 24 ++++++++++-
 test/dm/acpi.c            | 28 +++++++++++++
 4 files changed, 130 insertions(+), 46 deletions(-)
arch/x86/lib/acpi_table.c +43 -45
index f98d6d17b3e..afa00bc2f80 100644
@@ -10,6 +10,7 @@
 #include <cpu.h>
 #include <dm.h>
 #include <dm/uclass-internal.h>
+#include <mapmem.h>
 #include <serial.h>
 #include <version.h>
 #include <acpi/acpi_table.h>
@@ -19,6 +20,7 @@
 #include <asm/mpspec.h>
 #include <asm/tables.h>
 #include <asm/arch/global_nvs.h>
+#include <dm/acpi.h>
 
 /*
  * IASL compiles the dsdt entries and writes the hex values
@@ -468,9 +470,9 @@ static void acpi_create_spcr(struct acpi_spcr *spcr)
 /*
  * QEMU's version of write_acpi_tables is defined in drivers/misc/qfw.c
  */
-ulong write_acpi_tables(ulong start)
+ulong write_acpi_tables(ulong start_addr)
 {
-	u32 current;
+	struct acpi_ctx sctx, *ctx = &sctx;
 	struct acpi_rsdp *rsdp;
 	struct acpi_rsdt *rsdt;
 	struct acpi_xsdt *xsdt;
@@ -481,60 +483,61 @@ ulong write_acpi_tables(ulong start)
 	struct acpi_madt *madt;
 	struct acpi_csrt *csrt;
 	struct acpi_spcr *spcr;
+	void *start;
+	ulong addr;
 	int i;
 
-	current = start;
+	start = map_sysmem(start_addr, 0);
+	ctx->current = start;
 
 	/* Align ACPI tables to 16 byte */
-	current = ALIGN(current, 16);
+	acpi_align(ctx);
 
-	debug("ACPI: Writing ACPI tables at %lx\n", start);
+	debug("ACPI: Writing ACPI tables at %lx\n", start_addr);
 
 	/* We need at least an RSDP and an RSDT Table */
-	rsdp = (struct acpi_rsdp *)current;
-	current += sizeof(struct acpi_rsdp);
-	current = ALIGN(current, 16);
-	rsdt = (struct acpi_rsdt *)current;
-	current += sizeof(struct acpi_rsdt);
-	current = ALIGN(current, 16);
-	xsdt = (struct acpi_xsdt *)current;
-	current += sizeof(struct acpi_xsdt);
+	rsdp = ctx->current;
+	acpi_inc_align(ctx, sizeof(struct acpi_rsdp));
+	rsdt = ctx->current;
+	acpi_inc_align(ctx, sizeof(struct acpi_rsdt));
+	xsdt = ctx->current;
+	acpi_inc_align(ctx, sizeof(struct acpi_xsdt));
 	/*
 	 * Per ACPI spec, the FACS table address must be aligned to a 64 byte
 	 * boundary (Windows checks this, but Linux does not).
 	 */
-	current = ALIGN(current, 64);
+	acpi_align64(ctx);
 
 	/* clear all table memory */
-	memset((void *)start, 0, current - start);
+	memset((void *)start, 0, ctx->current - start);
 
 	acpi_write_rsdp(rsdp, rsdt, xsdt);
 	acpi_write_rsdt(rsdt);
 	acpi_write_xsdt(xsdt);
 
 	debug("ACPI:    * FACS\n");
-	facs = (struct acpi_facs *)current;
-	current += sizeof(struct acpi_facs);
-	current = ALIGN(current, 16);
+	facs = ctx->current;
+	acpi_inc_align(ctx, sizeof(struct acpi_facs));
 
 	acpi_create_facs(facs);
 
 	debug("ACPI:    * DSDT\n");
-	dsdt = (struct acpi_table_header *)current;
+	dsdt = ctx->current;
 	memcpy(dsdt, &AmlCode, sizeof(struct acpi_table_header));
-	current += sizeof(struct acpi_table_header);
-	memcpy((char *)current,
+	acpi_inc(ctx, sizeof(struct acpi_table_header));
+	memcpy(ctx->current,
 	       (char *)&AmlCode + sizeof(struct acpi_table_header),
 	       dsdt->length - sizeof(struct acpi_table_header));
-	current += dsdt->length - sizeof(struct acpi_table_header);
-	current = ALIGN(current, 16);
+	acpi_inc_align(ctx, dsdt->length - sizeof(struct acpi_table_header));
 
 	/* Pack GNVS into the ACPI table area */
 	for (i = 0; i < dsdt->length; i++) {
 		u32 *gnvs = (u32 *)((u32)dsdt + i);
 		if (*gnvs == ACPI_GNVS_ADDR) {
-			debug("Fix up global NVS in DSDT to 0x%08x\n", current);
-			*gnvs = current;
+			ulong addr = (ulong)map_to_sysmem(ctx->current);
+
+			debug("Fix up global NVS in DSDT to %#08lx\n", addr);
+			*gnvs = addr;
 			break;
 		}
 	}
@@ -544,51 +547,46 @@ ulong write_acpi_tables(ulong start)
 	dsdt->checksum = table_compute_checksum((void *)dsdt, dsdt->length);
 
 	/* Fill in platform-specific global NVS variables */
-	acpi_create_gnvs((struct acpi_global_nvs *)current);
-	current += sizeof(struct acpi_global_nvs);
-	current = ALIGN(current, 16);
+	acpi_create_gnvs(ctx->current);
+	acpi_inc_align(ctx, sizeof(struct acpi_global_nvs));
 
 	debug("ACPI:    * FADT\n");
-	fadt = (struct acpi_fadt *)current;
-	current += sizeof(struct acpi_fadt);
-	current = ALIGN(current, 16);
+	fadt = ctx->current;
+	acpi_inc_align(ctx, sizeof(struct acpi_fadt));
 	acpi_create_fadt(fadt, facs, dsdt);
 	acpi_add_table(rsdp, fadt);
 
 	debug("ACPI:    * MADT\n");
-	madt = (struct acpi_madt *)current;
+	madt = ctx->current;
 	acpi_create_madt(madt);
-	current += madt->header.length;
+	acpi_inc_align(ctx, madt->header.length);
 	acpi_add_table(rsdp, madt);
-	current = ALIGN(current, 16);
 
 	debug("ACPI:    * MCFG\n");
-	mcfg = (struct acpi_mcfg *)current;
+	mcfg = ctx->current;
 	acpi_create_mcfg(mcfg);
-	current += mcfg->header.length;
+	acpi_inc_align(ctx, mcfg->header.length);
 	acpi_add_table(rsdp, mcfg);
-	current = ALIGN(current, 16);
 
 	debug("ACPI:    * CSRT\n");
-	csrt = (struct acpi_csrt *)current;
+	csrt = ctx->current;
 	acpi_create_csrt(csrt);
-	current += csrt->header.length;
+	acpi_inc_align(ctx, csrt->header.length);
 	acpi_add_table(rsdp, csrt);
-	current = ALIGN(current, 16);
 
 	debug("ACPI:    * SPCR\n");
-	spcr = (struct acpi_spcr *)current;
+	spcr = ctx->current;
 	acpi_create_spcr(spcr);
-	current += spcr->header.length;
+	acpi_inc_align(ctx, spcr->header.length);
 	acpi_add_table(rsdp, spcr);
-	current = ALIGN(current, 16);
 
-	debug("current = %x\n", current);
+	addr = map_to_sysmem(ctx->current);
+	debug("current = %lx\n", addr);
 
 	acpi_rsdp_addr = (unsigned long)rsdp;
 	debug("ACPI: done\n");
 
-	return current;
+	return addr;
 }
 
 ulong acpi_get_rsdp_addr(void)
include/acpi/acpi_table.h +36 -0
index 268e36b67e1..0710324e698 100644
@@ -26,6 +26,8 @@
 
 #if !defined(__ACPI__)
 
+struct acpi_ctx;
+
 /*
  * RSDP (Root System Description Pointer)
  * Note: ACPI 1.0 didn't have length, xsdt_address, and ext_checksum
@@ -519,6 +521,40 @@ int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags);
  */
 void acpi_fill_header(struct acpi_table_header *header, char *signature);
 
+/**
+ * acpi_align() - Align the ACPI output pointer to a 16-byte boundary
+ *
+ * @ctx: ACPI context
+ */
+void acpi_align(struct acpi_ctx *ctx);
+
+/**
+ * acpi_align64() - Align the ACPI output pointer to a 64-byte boundary
+ *
+ * @ctx: ACPI context
+ */
+void acpi_align64(struct acpi_ctx *ctx);
+
+/**
+ * acpi_inc() - Increment the ACPI output pointer by a bit
+ *
+ * The pointer is NOT aligned afterwards.
+ *
+ * @ctx: ACPI context
+ * @amount: Amount to increment by
+ */
+void acpi_inc(struct acpi_ctx *ctx, uint amount);
+
+/**
+ * acpi_inc_align() - Increment the ACPI output pointer by a bit and align
+ *
+ * The pointer is aligned afterwards.
+ *
+ * @ctx: ACPI context
+ * @amount: Amount to increment by
+ */
+void acpi_inc_align(struct acpi_ctx *ctx, uint amount);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
lib/acpi/acpi_table.c +23 -1
index 1710cde2c40..35075a608f7 100644
@@ -6,10 +6,11 @@
  */
 
 #include <common.h>
-#include <acpi/acpi_table.h>
 #include <dm.h>
 #include <cpu.h>
 #include <version.h>
+#include <acpi/acpi_table.h>
+#include <dm/acpi.h>
 
 int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags)
 {
@@ -96,3 +97,24 @@ void acpi_fill_header(struct acpi_table_header *header, char *signature)
 	header->oem_revision = U_BOOT_BUILD_DATE;
 	memcpy(header->aslc_id, ASLC_ID, 4);
 }
+
+void acpi_align(struct acpi_ctx *ctx)
+{
+	ctx->current = (void *)ALIGN((ulong)ctx->current, 16);
+}
+
+void acpi_align64(struct acpi_ctx *ctx)
+{
+	ctx->current = (void *)ALIGN((ulong)ctx->current, 64);
+}
+
+void acpi_inc(struct acpi_ctx *ctx, uint amount)
+{
+	ctx->current += amount;
+}
+
+void acpi_inc_align(struct acpi_ctx *ctx, uint amount)
+{
+	ctx->current += amount;
+	acpi_align(ctx);
+}
test/dm/acpi.c +28 -0
index 5743530a3a5..17e208ee80e 100644
@@ -151,3 +151,31 @@ static int dm_test_acpi_write_tables(struct unit_test_state *uts)
 	return 0;
 }
 DM_TEST(dm_test_acpi_write_tables, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
+/* Test basic ACPI functions */
+static int dm_test_acpi_basic(struct unit_test_state *uts)
+{
+	struct acpi_ctx ctx;
+
+	/* Check align works */
+	ctx.current = (void *)5;
+	acpi_align(&ctx);
+	ut_asserteq_ptr((void *)16, ctx.current);
+
+	/* Check that align does nothing if already aligned */
+	acpi_align(&ctx);
+	ut_asserteq_ptr((void *)16, ctx.current);
+	acpi_align64(&ctx);
+	ut_asserteq_ptr((void *)64, ctx.current);
+	acpi_align64(&ctx);
+	ut_asserteq_ptr((void *)64, ctx.current);
+
+	/* Check incrementing */
+	acpi_inc(&ctx, 3);
+	ut_asserteq_ptr((void *)67, ctx.current);
+	acpi_inc_align(&ctx, 3);
+	ut_asserteq_ptr((void *)80, ctx.current);
+
+	return 0;
+}
+DM_TEST(dm_test_acpi_basic, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#68 lore raw

PATCH v3 23/29 acpi: Convert part of acpi_table to use acpi_ctx

Quoted 6 lines · Simon Glass

The current code uses an address but a pointer would result in fewer casts. Also it repeats the alignment code in a lot of places so this would be better done in a helper function.

Update write_acpi_tables() to make use of the new acpi_ctx structure, adding a few helpers to clean things up.

...

include/acpi/acpi_table.h int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags);

Quoted 3 lines
include/acpi/acpi_table.h +2 -0
+void acpi_align(struct acpi_ctx *ctx);
 
+void acpi_align64(struct acpi_ctx *ctx);

In the code, it will be not understandable the difference. align without number would be good if the function only single one.

So, align16() much better.

Quoted 1 line

+void acpi_inc(struct acpi_ctx *ctx, uint amount);

inc with amount is not inc, it's rather add.

Quoted 1 line

+ * acpi_inc_align() - Increment the ACPI output pointer by a bit and align

Align how?

Quoted 1 line

+void acpi_inc_align(struct acpi_ctx *ctx, uint amount);

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#69 lore raw

PATCH v3 23/29 acpi: Convert part of acpi_table to use acpi_ctx

Quoted 7 lines · Andy Shevchenko
Quoted 6 lines · Simon Glass

The current code uses an address but a pointer would result in fewer casts. Also it repeats the alignment code in a lot of places so this would be better done in a helper function.

Update write_acpi_tables() to make use of the new acpi_ctx structure, adding a few helpers to clean things up.

...

Quoted 3 lines
Quoted 1 line

+void acpi_inc(struct acpi_ctx *ctx, uint amount);

inc with amount is not inc, it's rather add.

...or pad if you wish.

Quoted 2 lines
Quoted 2 lines

+ * acpi_inc_align() - Increment the ACPI output pointer by a bit and align

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#70 lore raw

PATCH v3 23/29 acpi: Convert part of acpi_table to use acpi_ctx

In reply to Andy Shevchenko

Hi Andy,

Quoted 11 lines · Andy Shevchenko
Quoted 6 lines · Simon Glass

The current code uses an address but a pointer would result in fewer casts. Also it repeats the alignment code in a lot of places so this would be better done in a helper function.

Update write_acpi_tables() to make use of the new acpi_ctx structure, adding a few helpers to clean things up.

...

Quoted 8 lines

include/acpi/acpi_table.h int acpi_create_dmar(struct acpi_dmar *dmar, enum dmar_flags flags);

Quoted 3 lines
include/acpi/acpi_table.h +2 -0
+void acpi_align(struct acpi_ctx *ctx);
 
+void acpi_align64(struct acpi_ctx *ctx);

In the code, it will be not understandable the difference. align without number would be good if the function only single one.

So, align16() much better.

We had this discussion previously and changed it to what we have now in v3. I feel it is a matter of preference.

Quoted 4 lines
Quoted 2 lines

+void acpi_inc(struct acpi_ctx *ctx, uint amount);

inc with amount is not inc, it's rather add.

Increment by one, or increment by an amount. The word 'add' would be very confusing here, since we are not adding anything to the ACPI table.

Quoted 4 lines
Quoted 2 lines

+ * acpi_inc_align() - Increment the ACPI output pointer by a bit and align

Align how?

Will update the comment

Quoted 1 line

REgardsm

Simon

#71 lore raw

PATCH v3 24/29 x86: Allow devices to write ACPI tables

In reply to Simon Glass

Call the new core function to permit devices to write their own ACPI tables. These tables will appear after all other tables.

---

Changes in v3: None
Changes in v2: None

 arch/x86/lib/acpi_table.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/lib/acpi_table.c +2 -0
index afa00bc2f80..e5247c64049 100644
@@ -580,6 +580,8 @@ ulong write_acpi_tables(ulong start_addr)
 	acpi_inc_align(ctx, spcr->header.length);
 	acpi_add_table(rsdp, spcr);
 
+	acpi_write_dev_tables(ctx);
+
 	addr = map_to_sysmem(ctx->current);
 	debug("current = %lx\n", addr);
 
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#72 lore raw

PATCH v3 25/29 acpi: Drop code for missing XSDT from acpi_write_rsdp()

In reply to Simon Glass

We don't actually support tables without an XSDT so we can drop this dead code.

---

Changes in v3: None
Changes in v2: None

 arch/x86/lib/acpi_table.c | 15 ++-------------
 1 file changed, 2 insertions(+), 13 deletions(-)
arch/x86/lib/acpi_table.c +2 -13
index e5247c64049..8c169f44938 100644
@@ -42,19 +42,8 @@ static void acpi_write_rsdp(struct acpi_rsdp *rsdp, struct acpi_rsdt *rsdt,
 	rsdp->length = sizeof(struct acpi_rsdp);
 	rsdp->rsdt_address = (u32)rsdt;
 
-	/*
-	 * Revision: ACPI 1.0: 0, ACPI 2.0/3.0/4.0: 2
-	 *
-	 * Some OSes expect an XSDT to be present for RSD PTR revisions >= 2.
-	 * If we don't have an ACPI XSDT, force ACPI 1.0 (and thus RSD PTR
-	 * revision 0)
-	 */
-	if (xsdt == NULL) {
-		rsdp->revision = ACPI_RSDP_REV_ACPI_1_0;
-	} else {
-		rsdp->xsdt_address = (u64)(u32)xsdt;
-		rsdp->revision = ACPI_RSDP_REV_ACPI_2_0;
-	}
+	rsdp->xsdt_address = (u64)(u32)xsdt;
+	rsdp->revision = ACPI_RSDP_REV_ACPI_2_0;
 
 	/* Calculate checksums */
 	rsdp->checksum = table_compute_checksum((void *)rsdp, 20);
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#73 lore raw

PATCH v3 26/29 acpi: Move acpi_add_table() to generic code

In reply to Simon Glass

Move this code to a generic location so that we can test it with sandbox. This requires adding a few new fields to acpi_ctx, so drop the local variables used in the original code.

Also use mapmem to avoid pointer-to-address casts which don't work on sandbox.

---

Changes in v3:
- Fix 'RSDP' typo

Changes in v2: None

 arch/x86/lib/acpi_table.c | 82 +++++----------------------------------
 include/acpi/acpi_table.h |  9 +++++
 include/dm/acpi.h         |  5 +++
 lib/acpi/acpi_table.c     | 61 +++++++++++++++++++++++++++++
 test/dm/acpi.c            |  4 ++
 5 files changed, 89 insertions(+), 72 deletions(-)
arch/x86/lib/acpi_table.c +10 -72
index 8c169f44938..d4af56eabf4 100644
@@ -83,66 +83,6 @@ static void acpi_write_xsdt(struct acpi_xsdt *xsdt)
 			sizeof(struct acpi_xsdt));
 }
 
-/**
- * Add an ACPI table to the RSDT (and XSDT) structure, recalculate length
- * and checksum.
- */
-static void acpi_add_table(struct acpi_rsdp *rsdp, void *table)
-{
-	int i, entries_num;
-	struct acpi_rsdt *rsdt;
-	struct acpi_xsdt *xsdt = NULL;
-
-	/* The RSDT is mandatory while the XSDT is not */
-	rsdt = (struct acpi_rsdt *)rsdp->rsdt_address;
-
-	if (rsdp->xsdt_address)
-		xsdt = (struct acpi_xsdt *)((u32)rsdp->xsdt_address);
-
-	/* This should always be MAX_ACPI_TABLES */
-	entries_num = ARRAY_SIZE(rsdt->entry);
-
-	for (i = 0; i < entries_num; i++) {
-		if (rsdt->entry[i] == 0)
-			break;
-	}
-
-	if (i >= entries_num) {
-		debug("ACPI: Error: too many tables\n");
-		return;
-	}
-
-	/* Add table to the RSDT */
-	rsdt->entry[i] = (u32)table;
-
-	/* Fix RSDT length or the kernel will assume invalid entries */
-	rsdt->header.length = sizeof(struct acpi_table_header) +
-				(sizeof(u32) * (i + 1));
-
-	/* Re-calculate checksum */
-	rsdt->header.checksum = 0;
-	rsdt->header.checksum = table_compute_checksum((u8 *)rsdt,
-			rsdt->header.length);
-
-	/*
-	 * And now the same thing for the XSDT. We use the same index as for
-	 * now we want the XSDT and RSDT to always be in sync in U-Boot
-	 */
-	if (xsdt) {
-		/* Add table to the XSDT */
-		xsdt->entry[i] = (u64)(u32)table;
-
-		/* Fix XSDT length */
-		xsdt->header.length = sizeof(struct acpi_table_header) +
-			(sizeof(u64) * (i + 1));
-
-		/* Re-calculate checksum */
-		xsdt->header.checksum = 0;
-		xsdt->header.checksum = table_compute_checksum((u8 *)xsdt,
-				xsdt->header.length);
-	}
-}
-
 static void acpi_create_facs(struct acpi_facs *facs)
 {
 	memset((void *)facs, 0, sizeof(struct acpi_facs));
@@ -462,8 +402,6 @@ static void acpi_create_spcr(struct acpi_spcr *spcr)
 ulong write_acpi_tables(ulong start_addr)
 {
 	struct acpi_ctx sctx, *ctx = &sctx;
-	struct acpi_rsdp *rsdp;
-	struct acpi_rsdt *rsdt;
 	struct acpi_xsdt *xsdt;
 	struct acpi_facs *facs;
 	struct acpi_table_header *dsdt;
@@ -485,9 +423,9 @@ ulong write_acpi_tables(ulong start_addr)
 	debug("ACPI: Writing ACPI tables at %lx\n", start_addr);
 
 	/* We need at least an RSDP and an RSDT Table */
-	rsdp = ctx->current;
+	ctx->rsdp = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_rsdp));
-	rsdt = ctx->current;
+	ctx->rsdt = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_rsdt));
 	xsdt = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_xsdt));
@@ -500,8 +438,8 @@ ulong write_acpi_tables(ulong start_addr)
 	/* clear all table memory */
 	memset((void *)start, 0, ctx->current - start);
 
-	acpi_write_rsdp(rsdp, rsdt, xsdt);
-	acpi_write_rsdt(rsdt);
+	acpi_write_rsdp(ctx->rsdp, ctx->rsdt, xsdt);
+	acpi_write_rsdt(ctx->rsdt);
 	acpi_write_xsdt(xsdt);
 
 	debug("ACPI:    * FACS\n");
@@ -543,38 +481,38 @@ ulong write_acpi_tables(ulong start_addr)
 	fadt = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_fadt));
 	acpi_create_fadt(fadt, facs, dsdt);
-	acpi_add_table(rsdp, fadt);
+	acpi_add_table(ctx, fadt);
 
 	debug("ACPI:    * MADT\n");
 	madt = ctx->current;
 	acpi_create_madt(madt);
 	acpi_inc_align(ctx, madt->header.length);
-	acpi_add_table(rsdp, madt);
+	acpi_add_table(ctx, madt);
 
 	debug("ACPI:    * MCFG\n");
 	mcfg = ctx->current;
 	acpi_create_mcfg(mcfg);
 	acpi_inc_align(ctx, mcfg->header.length);
-	acpi_add_table(rsdp, mcfg);
+	acpi_add_table(ctx, mcfg);
 
 	debug("ACPI:    * CSRT\n");
 	csrt = ctx->current;
 	acpi_create_csrt(csrt);
 	acpi_inc_align(ctx, csrt->header.length);
-	acpi_add_table(rsdp, csrt);
+	acpi_add_table(ctx, csrt);
 
 	debug("ACPI:    * SPCR\n");
 	spcr = ctx->current;
 	acpi_create_spcr(spcr);
 	acpi_inc_align(ctx, spcr->header.length);
-	acpi_add_table(rsdp, spcr);
+	acpi_add_table(ctx, spcr);
 
 	acpi_write_dev_tables(ctx);
 
 	addr = map_to_sysmem(ctx->current);
 	debug("current = %lx\n", addr);
 
-	acpi_rsdp_addr = (unsigned long)rsdp;
+	acpi_rsdp_addr = (unsigned long)ctx->rsdp;
 	debug("ACPI: done\n");
 
 	return addr;
include/acpi/acpi_table.h +9 -0
index 0710324e698..b86800fdfa4 100644
@@ -555,6 +555,15 @@ void acpi_inc(struct acpi_ctx *ctx, uint amount);
  */
 void acpi_inc_align(struct acpi_ctx *ctx, uint amount);
 
+/**
+ * acpi_add_table() - Add a new table to the RSDP and XSDT
+ *
+ * @ctx: ACPI context
+ * @table: Table to add
+ * @return 0 if OK, -E2BIG if too many tables
+ */
+int acpi_add_table(struct acpi_ctx *ctx, void *table);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
include/dm/acpi.h +5 -0
index 9a0a437dd03..ee6f7a63e79 100644
@@ -30,9 +30,14 @@
  * This contains a few useful pieces of information used when writing
  *
  * @current: Current address for writing
+ * @rsdp: Pointer to the Root System Description Pointer, typically used when
+ *	adding a new table. The RSDP holds pointers to the RSDT and XSDT.
+ * @rsdt: Pointer to the Root System Description Table
  */
 struct acpi_ctx {
 	void *current;
+	struct acpi_rsdp *rsdp;
+	struct acpi_rsdt *rsdt;
 };
 
 /**
lib/acpi/acpi_table.c +61 -0

70 lines not on this page · raw message

test/dm/acpi.c +4 -0
index 17e208ee80e..4dd4a540967 100644
@@ -21,10 +21,14 @@ static int testacpi_write_tables(const struct udevice *dev,
 				 struct acpi_ctx *ctx)
 {
 	struct acpi_dmar *dmar;
+	int ret;
 
 	dmar = (struct acpi_dmar *)ctx->current;
 	acpi_create_dmar(dmar, DMAR_INTR_REMAP);
 	ctx->current += sizeof(struct acpi_dmar);
+	ret = acpi_add_table(ctx, dmar);
+	if (ret)
+		return log_msg_ret("add", ret);
 
 	return 0;
 }
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#74 lore raw

PATCH v3 27/29 acpi: Put table-setup code in its own function

In reply to Simon Glass

We always write three basic tables to ACPI at the start. Move this into its own function, along with acpi_fill_header(), so we can write a test for this code.

---

Changes in v3:
- Beef up the comment explaining how the unaligned address is used
- Fix 'XDST' typo
- Move acpi_align_large() out of dm_test_acpi_setup_base_tables()

Changes in v2: None

 arch/x86/lib/acpi_table.c | 72 +----------------------------------
 include/acpi/acpi_table.h | 10 +++++
 lib/acpi/acpi_table.c     | 79 ++++++++++++++++++++++++++++++++++++++-
 test/dm/acpi.c            | 58 +++++++++++++++++++++++++++-
 4 files changed, 145 insertions(+), 74 deletions(-)
arch/x86/lib/acpi_table.c +1 -71
index d4af56eabf4..4a7b0739394 100644
@@ -31,58 +31,6 @@ extern const unsigned char AmlCode[];
 /* ACPI RSDP address to be used in boot parameters */
 static ulong acpi_rsdp_addr;
 
-static void acpi_write_rsdp(struct acpi_rsdp *rsdp, struct acpi_rsdt *rsdt,
-			    struct acpi_xsdt *xsdt)
-{
-	memset(rsdp, 0, sizeof(struct acpi_rsdp));
-
-	memcpy(rsdp->signature, RSDP_SIG, 8);
-	memcpy(rsdp->oem_id, OEM_ID, 6);
-
-	rsdp->length = sizeof(struct acpi_rsdp);
-	rsdp->rsdt_address = (u32)rsdt;
-
-	rsdp->xsdt_address = (u64)(u32)xsdt;
-	rsdp->revision = ACPI_RSDP_REV_ACPI_2_0;
-
-	/* Calculate checksums */
-	rsdp->checksum = table_compute_checksum((void *)rsdp, 20);
-	rsdp->ext_checksum = table_compute_checksum((void *)rsdp,
-			sizeof(struct acpi_rsdp));
-}
-
-static void acpi_write_rsdt(struct acpi_rsdt *rsdt)
-{
-	struct acpi_table_header *header = &(rsdt->header);
-
-	/* Fill out header fields */
-	acpi_fill_header(header, "RSDT");
-	header->length = sizeof(struct acpi_rsdt);
-	header->revision = 1;
-
-	/* Entries are filled in later, we come with an empty set */
-
-	/* Fix checksum */
-	header->checksum = table_compute_checksum((void *)rsdt,
-			sizeof(struct acpi_rsdt));
-}
-
-static void acpi_write_xsdt(struct acpi_xsdt *xsdt)
-{
-	struct acpi_table_header *header = &(xsdt->header);
-
-	/* Fill out header fields */
-	acpi_fill_header(header, "XSDT");
-	header->length = sizeof(struct acpi_xsdt);
-	header->revision = 1;
-
-	/* Entries are filled in later, we come with an empty set */
-
-	/* Fix checksum */
-	header->checksum = table_compute_checksum((void *)xsdt,
-			sizeof(struct acpi_xsdt));
-}
-
 static void acpi_create_facs(struct acpi_facs *facs)
 {
 	memset((void *)facs, 0, sizeof(struct acpi_facs));
@@ -402,7 +350,6 @@ static void acpi_create_spcr(struct acpi_spcr *spcr)
 ulong write_acpi_tables(ulong start_addr)
 {
 	struct acpi_ctx sctx, *ctx = &sctx;
-	struct acpi_xsdt *xsdt;
 	struct acpi_facs *facs;
 	struct acpi_table_header *dsdt;
 	struct acpi_fadt *fadt;
@@ -415,33 +362,16 @@ ulong write_acpi_tables(ulong start_addr)
 	int i;
 
 	start = map_sysmem(start_addr, 0);
-	ctx->current = start;
-
-	/* Align ACPI tables to 16 byte */
-	acpi_align(ctx);
 
 	debug("ACPI: Writing ACPI tables at %lx\n", start_addr);
 
-	/* We need at least an RSDP and an RSDT Table */
-	ctx->rsdp = ctx->current;
-	acpi_inc_align(ctx, sizeof(struct acpi_rsdp));
-	ctx->rsdt = ctx->current;
-	acpi_inc_align(ctx, sizeof(struct acpi_rsdt));
-	xsdt = ctx->current;
-	acpi_inc_align(ctx, sizeof(struct acpi_xsdt));
+	acpi_setup_base_tables(ctx, start);
 	/*
 	 * Per ACPI spec, the FACS table address must be aligned to a 64 byte
 	 * boundary (Windows checks this, but Linux does not).
 	 */
 	acpi_align64(ctx);
 
-	/* clear all table memory */
-	memset((void *)start, 0, ctx->current - start);
-
-	acpi_write_rsdp(ctx->rsdp, ctx->rsdt, xsdt);
-	acpi_write_rsdt(ctx->rsdt);
-	acpi_write_xsdt(xsdt);
-
 	debug("ACPI:    * FACS\n");
 	facs = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_facs));
include/acpi/acpi_table.h +10 -0
index b86800fdfa4..05b0948b6f6 100644
@@ -564,6 +564,16 @@ void acpi_inc_align(struct acpi_ctx *ctx, uint amount);
  */
 int acpi_add_table(struct acpi_ctx *ctx, void *table);
 
+/**
+ * acpi_setup_base_tables() - Set up context along with RSDP, RSDT and XSDT
+ *
+ * Set up the context with the given start position. Some basic tables are
+ * always needed, so set them up as well.
+ *
+ * @ctx: Context to set up
+ */
+void acpi_setup_base_tables(struct acpi_ctx *ctx, void *start);
+
 #endif /* !__ACPI__*/
 
 #include <asm/acpi_table.h>
lib/acpi/acpi_table.c +78 -1

88 lines not on this page · raw message

test/dm/acpi.c +56 -2

80 lines not on this page · raw message

Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#75 lore raw

PATCH v3 27/29 acpi: Put table-setup code in its own function

Quoted 3 lines · Simon Glass

We always write three basic tables to ACPI at the start. Move this into its own function, along with acpi_fill_header(), so we can write a test for this code.

...

lib/acpi/acpi_table.c int acpi_add_table(struct acpi_ctx *ctx, void *table) · line 160

Quoted 5 lines
lib/acpi/acpi_table.c +1 -1
 	/* Re-calculate checksum */
 	rsdt->header.checksum = 0;
-	rsdt->header.checksum = table_compute_checksum((u8 *)rsdt,
+	rsdt->header.checksum = table_compute_checksum(rsdt,
 						       rsdt->header.length);

Why suddenly casting is not needed in this patch?
Same question to the rest.

(If it's a valid change, it should be in a separate patch)

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#76 lore raw

PATCH v3 27/29 acpi: Put table-setup code in its own function

Hi Andy,

Quoted 6 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

We always write three basic tables to ACPI at the start. Move this into its own function, along with acpi_fill_header(), so we can write a test for this code.

Quoted 7 lines

...

lib/acpi/acpi_table.c int acpi_add_table(struct acpi_ctx *ctx, void *table) · line 160

Quoted 5 lines
      /* Re-calculate checksum */
      rsdt->header.checksum = 0;
lib/acpi/acpi_table.c +1 -1
-     rsdt->header.checksum = table_compute_checksum((u8 *)rsdt,
+     rsdt->header.checksum = table_compute_checksum(rsdt,
                                                     rsdt->header.length);

Please can you keep the filenames / functions in your response? Fragments make it harder to find the code.

Quoted 5 lines

Why suddenly casting is not needed in this patch?
Same question to the rest.

(If it's a valid change, it should be in a separate patch)

It was never needed. See the prototype for table_compute_checksum().

But I can put it back in.

Regards,
Simon

#77 lore raw

PATCH v3 27/29 acpi: Put table-setup code in its own function

Quoted 10 lines · Simon Glass
Quoted 8 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

We always write three basic tables to ACPI at the start. Move this into its own function, along with acpi_fill_header(), so we can write a test for this code.

...

Quoted 8 lines
Quoted 5 lines

lib/acpi/acpi_table.c int acpi_add_table(struct acpi_ctx *ctx, void *table) · line 160

Quoted 5 lines
      /* Re-calculate checksum */
      rsdt->header.checksum = 0;
lib/acpi/acpi_table.c +1 -1
-     rsdt->header.checksum = table_compute_checksum((u8 *)rsdt,
+     rsdt->header.checksum = table_compute_checksum(rsdt,
                                                     rsdt->header.length);

Please can you keep the filenames / functions in your response? Fragments make it harder to find the code.

I thought, obviously mistakenly, that git users know about git grep ...

Quoted 8 lines
Quoted 4 lines

Why suddenly casting is not needed in this patch?
Same question to the rest.

(If it's a valid change, it should be in a separate patch)

It was never needed. See the prototype for table_compute_checksum().

But I can put it back in.

Depends on your preferences, but it's definitely not a material for this change. Separate one?

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#78 lore raw

PATCH v3 27/29 acpi: Put table-setup code in its own function

Hi Andy,

Quoted 18 lines · Andy Shevchenko
Quoted 16 lines · Simon Glass
Quoted 13 lines · Andy Shevchenko
Quoted 3 lines · Simon Glass

We always write three basic tables to ACPI at the start. Move this into its own function, along with acpi_fill_header(), so we can write a test for this code.

...

Quoted 5 lines
      /* Re-calculate checksum */
      rsdt->header.checksum = 0;
lib/acpi/acpi_table.c +1 -1
-     rsdt->header.checksum = table_compute_checksum((u8 *)rsdt,
+     rsdt->header.checksum = table_compute_checksum(rsdt,
                                                     rsdt->header.length);
Quoted 4 lines
Quoted 2 lines

Please can you keep the filenames / functions in your response? Fragments make it harder to find the code.

I thought, obviously mistakenly, that git users know about git grep ...

Well I'd appreciate it if you could keep them in. It is customary, and it avoids grepping as you say.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

Why suddenly casting is not needed in this patch?
Same question to the rest.

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

(If it's a valid change, it should be in a separate patch)

It was never needed. See the prototype for table_compute_checksum().

But I can put it back in.

Depends on your preferences, but it's definitely not a material for this change. Separate one?

It isn't worth worrying about as a later commit drops it. Let's just leave it as you had it, with the cast.

Regards,
Simon

#79 lore raw

PATCH v3 28/29 acpi: Move the xsdt pointer to acpi_ctx

In reply to Simon Glass

Put this in the context along with the other important pointers.

---

Changes in v3: None
Changes in v2: None

 include/dm/acpi.h     |  2 ++
 lib/acpi/acpi_table.c | 10 ++++------
 test/dm/acpi.c        |  5 +++++
 3 files changed, 11 insertions(+), 6 deletions(-)
include/dm/acpi.h +2 -0
index ee6f7a63e79..288310b5128 100644
@@ -33,11 +33,13 @@
  * @rsdp: Pointer to the Root System Description Pointer, typically used when
  *	adding a new table. The RSDP holds pointers to the RSDT and XSDT.
  * @rsdt: Pointer to the Root System Description Table
+ * @xsdt: Pointer to the Extended System Description Table
  */
 struct acpi_ctx {
 	void *current;
 	struct acpi_rsdp *rsdp;
 	struct acpi_rsdt *rsdt;
+	struct acpi_xsdt *xsdt;
 };
 
 /**
lib/acpi/acpi_table.c +4 -6
index 70d29c53881..28003dc8f46 100644
@@ -163,7 +163,7 @@ int acpi_add_table(struct acpi_ctx *ctx, void *table)
 	 * And now the same thing for the XSDT. We use the same index as for
 	 * now we want the XSDT and RSDT to always be in sync in U-Boot
 	 */
-	xsdt = (struct acpi_xsdt *)((u32)rsdp->xsdt_address);
+	xsdt = ctx->xsdt;
 
 	/* Add table to the XSDT */
 	xsdt->entry[i] = map_to_sysmem(table);
@@ -234,8 +234,6 @@ static void acpi_write_xsdt(struct acpi_xsdt *xsdt)
 
 void acpi_setup_base_tables(struct acpi_ctx *ctx, void *start)
 {
-	struct acpi_xsdt *xsdt;
-
 	ctx->current = start;
 
 	/* Align ACPI tables to 16 byte */
@@ -246,13 +244,13 @@ void acpi_setup_base_tables(struct acpi_ctx *ctx, void *start)
 	acpi_inc_align(ctx, sizeof(struct acpi_rsdp));
 	ctx->rsdt = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_rsdt));
-	xsdt = ctx->current;
+	ctx->xsdt = ctx->current;
 	acpi_inc_align(ctx, sizeof(struct acpi_xsdt));
 
 	/* clear all table memory */
 	memset((void *)start, '\0', ctx->current - start);
 
-	acpi_write_rsdp(ctx->rsdp, ctx->rsdt, xsdt);
+	acpi_write_rsdp(ctx->rsdp, ctx->rsdt, ctx->xsdt);
 	acpi_write_rsdt(ctx->rsdt);
-	acpi_write_xsdt(xsdt);
+	acpi_write_xsdt(ctx->xsdt);
 }
test/dm/acpi.c +5 -0
index 5d5e46e2944..8e160d2cf4e 100644
@@ -160,6 +160,10 @@ static int dm_test_acpi_write_tables(struct unit_test_state *uts)
 	ut_asserteq(map_to_sysmem(dmar + 1), ctx.rsdt->entry[1]);
 	ut_asserteq(0, ctx.rsdt->entry[2]);
 
+	ut_asserteq(map_to_sysmem(dmar), ctx.xsdt->entry[0]);
+	ut_asserteq(map_to_sysmem(dmar + 1), ctx.xsdt->entry[1]);
+	ut_asserteq(0, ctx.xsdt->entry[2]);
+
 	return 0;
 }
 DM_TEST(dm_test_acpi_write_tables, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
@@ -223,6 +227,7 @@ static int dm_test_acpi_setup_base_tables(struct unit_test_state *uts)
 	ut_assertok(table_compute_checksum(rsdt, sizeof(*rsdt)));
 
 	xsdt = PTR_ALIGN((void *)rsdt + sizeof(*rsdt), 16);
+	ut_asserteq_ptr(xsdt, ctx.xsdt);
 	ut_assertok(memcmp("XSDT", xsdt->header.signature, ACPI_NAME_LEN));
 	ut_asserteq(sizeof(*xsdt), xsdt->header.length);
 	ut_assertok(table_compute_checksum(xsdt, sizeof(*xsdt)));
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#80 lore raw

PATCH v3 29/29 acpi: Add an acpi command

In reply to Simon Glass

It is useful to dump ACPI tables in U-Boot to see what has been generated. Add a command to handle this.

To allow the command to find the tables, add a position into the global data.

Support subcommands to list and dump the tables.

---

Changes in v3: None
Changes in v2: None

 arch/sandbox/include/asm/global_data.h |   1 +
 arch/x86/include/asm/global_data.h     |   1 +
 cmd/Kconfig                            |  14 ++
 cmd/Makefile                           |   1 +
 cmd/acpi.c                             | 179 +++++++++++++++++++++++++
 lib/acpi/acpi_table.c                  |   1 +
 test/dm/acpi.c                         |  73 ++++++++++
 7 files changed, 270 insertions(+)
 create mode 100644 cmd/acpi.c
arch/sandbox/include/asm/global_data.h +1 -0
index f4ce72d5660..f95ddb058a2 100644
@@ -13,6 +13,7 @@
 struct arch_global_data {
 	uint8_t		*ram_buf;	/* emulated RAM buffer */
 	void		*text_base;	/* pointer to base of text region */
+	ulong acpi_start;		/* Start address of ACPI tables */
 };
 
 #include <asm-generic/global_data.h>
arch/x86/include/asm/global_data.h +1 -0
index f4c1839104e..4aee2f3e8c4 100644
@@ -123,6 +123,7 @@ struct arch_global_data {
 #ifdef CONFIG_FSP_VERSION2
 	struct fsp_header *fsp_s_hdr;	/* Pointer to FSP-S header */
 #endif
+	ulong acpi_start;		/* Start address of ACPI tables */
 };
 
 #endif
cmd/Kconfig +14 -0
index 6403bc45a5e..2d3bfe0ab91 100644
@@ -190,6 +190,20 @@ comment "Commands"
 
 menu "Info commands"
 
+config CMD_ACPI
+	bool "acpi"
+	default y if ACPIGEN
+	help
+	  List and dump ACPI tables. ACPI (Advanced Configuration and Power
+	  Interface) is used mostly on x86 for providing information to the
+	  Operating System about devices in the system. The tables are set up
+	  by the firmware, typically U-Boot but possibly an earlier firmware
+	  module, if U-Boot is chain-loaded from something else. ACPI tables
+	  can also include code, to perform hardware-specific tasks required
+	  by the Operating Systems. This allows some amount of separation
+	  between the firmware and OS, and is particularly useful when you
+	  want to make hardware changes without the OS needing to be adjusted.
+
 config CMD_BDI
 	bool "bdinfo"
 	default y
cmd/Makefile +1 -0
index f1dd513a4b4..15a9693ed0e 100644
@@ -11,6 +11,7 @@ obj-y += help.o
 obj-y += version.o
 
 # command
+obj-$(CONFIG_CMD_ACPI) += acpi.o
 obj-$(CONFIG_CMD_AES) += aes.o
 obj-$(CONFIG_CMD_AB_SELECT) += ab_select.o
 obj-$(CONFIG_CMD_ADC) += adc.o
cmd/acpi.c new file +179 -0

179 lines not on this page · raw message

lib/acpi/acpi_table.c +1 -0
index 28003dc8f46..3b5479eb85b 100644
@@ -238,6 +238,7 @@ void acpi_setup_base_tables(struct acpi_ctx *ctx, void *start)
 
 	/* Align ACPI tables to 16 byte */
 	acpi_align(ctx);
+	gd->arch.acpi_start = map_to_sysmem(ctx->current);
 
 	/* We need at least an RSDP and an RSDT Table */
 	ctx->rsdp = ctx->current;
test/dm/acpi.c +73 -0

91 lines not on this page · raw message

Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#81 lore raw

PATCH v3 29/29 acpi: Add an acpi command

Quoted 164 lines · Simon Glass

It is useful to dump ACPI tables in U-Boot to see what has been generated. Add a command to handle this.

To allow the command to find the tables, add a position into the global data.

Support subcommands to list and dump the tables.

---

Changes in v3: None
Changes in v2: None

 arch/sandbox/include/asm/global_data.h |   1 +
 arch/x86/include/asm/global_data.h     |   1 +
 cmd/Kconfig                            |  14 ++
 cmd/Makefile                           |   1 +
 cmd/acpi.c                             | 179 +++++++++++++++++++++++++
 lib/acpi/acpi_table.c                  |   1 +
 test/dm/acpi.c                         |  73 ++++++++++
 7 files changed, 270 insertions(+)
 create mode 100644 cmd/acpi.c
arch/sandbox/include/asm/global_data.h +1 -0
index f4ce72d5660..f95ddb058a2 100644
@@ -13,6 +13,7 @@
 struct arch_global_data {
 	uint8_t		*ram_buf;	/* emulated RAM buffer */
 	void		*text_base;	/* pointer to base of text region */
+	ulong acpi_start;		/* Start address of ACPI tables */
 };
 
 #include <asm-generic/global_data.h>
arch/x86/include/asm/global_data.h +1 -0
index f4c1839104e..4aee2f3e8c4 100644
@@ -123,6 +123,7 @@ struct arch_global_data {
 #ifdef CONFIG_FSP_VERSION2
 	struct fsp_header *fsp_s_hdr;	/* Pointer to FSP-S header */
 #endif
+	ulong acpi_start;		/* Start address of ACPI tables */
 };
 
 #endif
cmd/Kconfig +14 -0
index 6403bc45a5e..2d3bfe0ab91 100644
@@ -190,6 +190,20 @@ comment "Commands"
 
 menu "Info commands"
 
+config CMD_ACPI
+	bool "acpi"
+	default y if ACPIGEN
+	help
+	  List and dump ACPI tables. ACPI (Advanced Configuration and Power
+	  Interface) is used mostly on x86 for providing information to the
+	  Operating System about devices in the system. The tables are set up
+	  by the firmware, typically U-Boot but possibly an earlier firmware
+	  module, if U-Boot is chain-loaded from something else. ACPI tables
+	  can also include code, to perform hardware-specific tasks required
+	  by the Operating Systems. This allows some amount of separation
+	  between the firmware and OS, and is particularly useful when you
+	  want to make hardware changes without the OS needing to be adjusted.
+
 config CMD_BDI
 	bool "bdinfo"
 	default y
cmd/Makefile +1 -0
index f1dd513a4b4..15a9693ed0e 100644
@@ -11,6 +11,7 @@ obj-y += help.o
 obj-y += version.o
 
 # command
+obj-$(CONFIG_CMD_ACPI) += acpi.o
 obj-$(CONFIG_CMD_AES) += aes.o
 obj-$(CONFIG_CMD_AB_SELECT) += ab_select.o
 obj-$(CONFIG_CMD_ADC) += adc.o
cmd/acpi.c new file +72 -0
new file mode 100644
index 00000000000..8a320435736
@@ -0,0 +1,179 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2019 Google LLC
+ * Written by Simon Glass <[email protected]>
+ */
+#include <common.h>
+#include <command.h>
+#include <mapmem.h>
+#include <acpi/acpi_table.h>
+#include <asm/acpi_table.h>
+#include <dm/acpi.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+static void dump_hdr(struct acpi_table_header *hdr)
+{
+	bool has_hdr = memcmp(hdr->signature, "FACS", ACPI_NAME_LEN);
+
+	printf("%.*s %08lx %06x", ACPI_NAME_LEN, hdr->signature,
+	       (ulong)map_to_sysmem(hdr), hdr->length);
+	if (has_hdr) {
+		printf(" (v%02d %.6s %.8s %u %.4s %d)\n", hdr->revision,
+		       hdr->oem_id, hdr->oem_table_id, hdr->oem_revision,
+		       hdr->aslc_id, hdr->aslc_revision);
+	} else {
+		printf("\n");
+	}
+}
+
+/**
+ * find_table() - Look up an ACPI table
+ *
+ * @sig: Signature of table (4 characters, upper case)
+ * @return pointer to table header, or NULL if not found
+ */
+struct acpi_table_header *find_table(const char *sig)
+{
+	struct acpi_rsdp *rsdp;
+	struct acpi_rsdt *rsdt;
+	int len, i, count;
+
+	rsdp = map_sysmem(gd->arch.acpi_start, 0);
+	if (!rsdp)
+		return NULL;
+	rsdt = map_sysmem(rsdp->rsdt_address, 0);
+	len = rsdt->header.length - sizeof(rsdt->header);
+	count = len / sizeof(u32);
+	for (i = 0; i < count; i++) {
+		struct acpi_table_header *hdr;
+
+		hdr = map_sysmem(rsdt->entry[i], 0);
+		if (!memcmp(hdr->signature, sig, ACPI_NAME_LEN))
+			return hdr;
+		if (!memcmp(hdr->signature, "FACP", ACPI_NAME_LEN)) {
+			struct acpi_fadt *fadt = (struct acpi_fadt *)hdr;
+
+			if (!memcmp(sig, "DSDT", ACPI_NAME_LEN) && fadt->dsdt)
+				return map_sysmem(fadt->dsdt, 0);
+			if (!memcmp(sig, "FACS", ACPI_NAME_LEN) &&
+			    fadt->firmware_ctrl)
+				return map_sysmem(fadt->firmware_ctrl, 0);
+		}
+	}
+
+	return NULL;
+}
+
+static int dump_table_name(const char *sig)
+{
+	struct acpi_table_header *hdr;
+
+	hdr = find_table(sig);

cmd/acpi.c line 82

Quoted 10 lines
cmd/acpi.c new file +10 -0
+	if (!hdr)
+		return -ENOENT;
+	printf("%.*s @ %08lx\n", ACPI_NAME_LEN, hdr->signature,
+	       (ulong)map_to_sysmem(hdr));
+	print_buffer(0, hdr, 1, hdr->length, 0);
+
+	return 0;
+}
+
+static void list_fact(struct acpi_fadt *fadt)

Hmm, should this function be called list_facp or list_fadt? (The wonder that is the table called FADT with the marker FACP.)

/
Leif

Quoted 210 lines
cmd/acpi.c new file +97 -0
+{
+	if (fadt->dsdt)
+		dump_hdr(map_sysmem(fadt->dsdt, 0));
+	if (fadt->firmware_ctrl)
+		dump_hdr(map_sysmem(fadt->firmware_ctrl, 0));
+}
+
+static int list_rsdt(struct acpi_rsdt *rsdt, struct acpi_xsdt *xsdt)
+{
+	int len, i, count;
+
+	dump_hdr(&rsdt->header);
+	if (xsdt)
+		dump_hdr(&xsdt->header);
+	len = rsdt->header.length - sizeof(rsdt->header);
+	count = len / sizeof(u32);
+	for (i = 0; i < count; i++) {
+		struct acpi_table_header *hdr;
+
+		if (!rsdt->entry[i])
+			break;
+		hdr = map_sysmem(rsdt->entry[i], 0);
+		dump_hdr(hdr);
+		if (!memcmp(hdr->signature, "FACP", ACPI_NAME_LEN))
+			list_fact((struct acpi_fadt *)hdr);
+		if (xsdt) {
+			if (xsdt->entry[i] != rsdt->entry[i]) {
+				printf("   (xsdt mismatch %llx)\n",
+				       xsdt->entry[i]);
+			}
+		}
+	}
+
+	return 0;
+}
+
+static int list_rsdp(struct acpi_rsdp *rsdp)
+{
+	struct acpi_rsdt *rsdt;
+	struct acpi_xsdt *xsdt;
+
+	printf("RSDP %08lx %06x (v%02d %.6s)\n", (ulong)map_to_sysmem(rsdp),
+	       rsdp->length, rsdp->revision, rsdp->oem_id);
+	rsdt = map_sysmem(rsdp->rsdt_address, 0);
+	xsdt = map_sysmem(rsdp->xsdt_address, 0);
+	list_rsdt(rsdt, xsdt);
+
+	return 0;
+}
+
+static int do_acpi_list(cmd_tbl_t *cmdtp, int flag, int argc,
+			char *const argv[])
+{
+	struct acpi_rsdp *rsdp;
+
+	rsdp = map_sysmem(gd->arch.acpi_start, 0);
+	if (!rsdp) {
+		printf("No ACPI tables present\n");
+		return 0;
+	}
+	printf("ACPI tables start at %lx\n", gd->arch.acpi_start);
+	list_rsdp(rsdp);
+
+	return 0;
+}
+
+static int do_acpi_dump(cmd_tbl_t *cmdtp, int flag, int argc,
+			char *const argv[])
+{
+	const char *name;
+	char sig[ACPI_NAME_LEN];
+	int ret;
+
+	if (argc < 2)
+		return CMD_RET_USAGE;
+	name = argv[1];
+	if (strlen(name) != ACPI_NAME_LEN) {
+		printf("Table name '%s' must be four characters\n", name);
+		return CMD_RET_FAILURE;
+	}
+	str_to_upper(name, sig, -1);
+	ret = dump_table_name(sig);
+	if (ret) {
+		printf("Table '%.*s' not found\n", ACPI_NAME_LEN, sig);
+		return CMD_RET_FAILURE;
+	}
+
+	return 0;
+}
+
+static char acpi_help_text[] =
+	"list - list ACPI tables\n"
+	"acpi dump <name> - Dump ACPI table";
+
+U_BOOT_CMD_WITH_SUBCMDS(acpi, "ACPI tables", acpi_help_text,
+	U_BOOT_SUBCMD_MKENT(list, 1, 1, do_acpi_list),
+	U_BOOT_SUBCMD_MKENT(dump, 2, 1, do_acpi_dump));
lib/acpi/acpi_table.c +1 -0
index 28003dc8f46..3b5479eb85b 100644
@@ -238,6 +238,7 @@ void acpi_setup_base_tables(struct acpi_ctx *ctx, void *start)
 
 	/* Align ACPI tables to 16 byte */
 	acpi_align(ctx);
+	gd->arch.acpi_start = map_to_sysmem(ctx->current);
 
 	/* We need at least an RSDP and an RSDT Table */
 	ctx->rsdp = ctx->current;
test/dm/acpi.c +73 -0
index 8e160d2cf4e..4eabf53128a 100644
@@ -7,9 +7,11 @@
  */
 
 #include <common.h>
+#include <console.h>
 #include <dm.h>
 #include <malloc.h>
 #include <mapmem.h>
+#include <version.h>
 #include <tables_csum.h>
 #include <version.h>
 #include <acpi/acpi_table.h>
@@ -212,6 +214,7 @@ static int dm_test_acpi_setup_base_tables(struct unit_test_state *uts)
 	buf = memalign(64, BUF_SIZE);
 	ut_assertnonnull(buf);
 	acpi_setup_base_tables(&ctx, buf + 4);
+	ut_asserteq(map_to_sysmem(PTR_ALIGN(buf + 4, 16)), gd->arch.acpi_start);
 
 	rsdp = buf + 16;
 	ut_asserteq_ptr(rsdp, ctx.rsdp);
@@ -242,3 +245,73 @@ static int dm_test_acpi_setup_base_tables(struct unit_test_state *uts)
 }
 DM_TEST(dm_test_acpi_setup_base_tables,
 	DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
+/* Test 'acpi list' command */
+static int dm_test_acpi_cmd_list(struct unit_test_state *uts)
+{
+	struct acpi_ctx ctx;
+	ulong addr;
+	void *buf;
+
+	buf = memalign(16, BUF_SIZE);
+	ut_assertnonnull(buf);
+	acpi_setup_base_tables(&ctx, buf);
+
+	ut_assertok(acpi_write_dev_tables(&ctx));
+
+	console_record_reset();
+	run_command("acpi list", 0);
+	addr = (ulong)map_to_sysmem(buf);
+	ut_assert_nextline("ACPI tables start at %lx", addr);
+	ut_assert_nextline("RSDP %08lx %06lx (v02 U-BOOT)", addr,
+			   sizeof(struct acpi_rsdp));
+	addr = ALIGN(addr + sizeof(struct acpi_rsdp), 16);
+	ut_assert_nextline("RSDT %08lx %06lx (v01 U-BOOT U-BOOTBL %u INTL 0)",
+			   addr, sizeof(struct acpi_table_header) +
+			   2 * sizeof(u32), U_BOOT_BUILD_DATE);
+	addr = ALIGN(addr + sizeof(struct acpi_rsdt), 16);
+	ut_assert_nextline("XSDT %08lx %06lx (v01 U-BOOT U-BOOTBL %u INTL 0)",
+			   addr, sizeof(struct acpi_table_header) +
+			   2 * sizeof(u64), U_BOOT_BUILD_DATE);
+	addr = ALIGN(addr + sizeof(struct acpi_xsdt), 64);
+	ut_assert_nextline("DMAR %08lx %06lx (v01 U-BOOT U-BOOTBL %u INTL 0)",
+			   addr, sizeof(struct acpi_dmar), U_BOOT_BUILD_DATE);
+	addr = ALIGN(addr + sizeof(struct acpi_dmar), 16);
+	ut_assert_nextline("DMAR %08lx %06lx (v01 U-BOOT U-BOOTBL %u INTL 0)",
+			   addr, sizeof(struct acpi_dmar), U_BOOT_BUILD_DATE);
+	ut_assert_console_end();
+
+	return 0;
+}
+DM_TEST(dm_test_acpi_cmd_list, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
+
+/* Test 'acpi dump' command */
+static int dm_test_acpi_cmd_dump(struct unit_test_state *uts)
+{
+	struct acpi_ctx ctx;
+	ulong addr;
+	void *buf;
+
+	buf = memalign(16, BUF_SIZE);
+	ut_assertnonnull(buf);
+	acpi_setup_base_tables(&ctx, buf);
+
+	ut_assertok(acpi_write_dev_tables(&ctx));
+
+	/* First search for a non-existent table */
+	console_record_reset();
+	run_command("acpi dump rdst", 0);
+	ut_assert_nextline("Table 'RDST' not found");
+	ut_assert_console_end();
+
+	/* Now a real table */
+	console_record_reset();
+	run_command("acpi dump dmar", 0);
+	addr = ALIGN(map_to_sysmem(ctx.xsdt) + sizeof(struct acpi_xsdt), 64);
+	ut_assert_nextline("DMAR @ %08lx", addr);
+	ut_assert_nextlines_are_dump(0x30);
+	ut_assert_console_end();
+
+	return 0;
+}
+DM_TEST(dm_test_acpi_cmd_dump, DM_TESTF_SCAN_PDATA | DM_TESTF_SCAN_FDT);
Signature · 1 line
2.26.0.rc2.310.g2932bb562d-goog
#82 lore raw

PATCH v3 29/29 acpi: Add an acpi command

Quoted 6 lines · Leif Lindholm
Quoted 1 line · Simon Glass

+static void list_fact(struct acpi_fadt *fadt)

Hmm, should this function be called list_facp or list_fadt? (The wonder that is the table called FADT with the marker FACP.)

Spec refers to some historical reasons.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#83 lore raw

PATCH v3 29/29 acpi: Add an acpi command

In reply to Simon Glass
Quoted 7 lines · Simon Glass

It is useful to dump ACPI tables in U-Boot to see what has been generated. Add a command to handle this.

To allow the command to find the tables, add a position into the global data.

Support subcommands to list and dump the tables.

...

cmd/acpi.c line 17

Quoted 4 lines
cmd/acpi.c new file +3 -0
+static void dump_hdr(struct acpi_table_header *hdr)
+{
 
+	bool has_hdr = memcmp(hdr->signature, "FACS", ACPI_NAME_LEN);

I believe more than one table has the same header structure. Either this function is incorrectly called (should be dump_facs_hdr() or alike), or you need to make it better, i.e. generic.

Quoted 10 lines
cmd/acpi.c new file +10 -0
+	printf("%.*s %08lx %06x", ACPI_NAME_LEN, hdr->signature,
+	       (ulong)map_to_sysmem(hdr), hdr->length);
+	if (has_hdr) {
+		printf(" (v%02d %.6s %.8s %u %.4s %d)\n", hdr->revision,
+		       hdr->oem_id, hdr->oem_table_id, hdr->oem_revision,
+		       hdr->aslc_id, hdr->aslc_revision);
+	} else {
+		printf("\n");
+	}
+}
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#84 lore raw

v3 cover dm: Add programmatic generation of ACPI tables (part A)

In reply to Simon Glass
Quoted 5 lines · Simon Glass

This is split from the original series in an attempt to get things applied in chunks.

This section includes patches up to and including the 'acpi' command.

Quoted 11 lines

Changes in v3:
- Add a pointer to information about acpi,compatible
- Add forward declarations for the functions
- Add missing error check in acpi_create_dmar()
- Add new patch to add hexdump.h to the unit test header
- Add new patch to move acpi_s3.h to include/acpi/
- Beef up the comment explaining how the unaligned address is used
- Change the example to ELAN
- Compute ACPI_NAME_MAX from ACPI_NAME_LEN
- Correct description of acpi,probed
- Drop 'Intel' from 'Intel ACPI'

Hello Simon,

not only x86 but also ARM can use ACPI tables for booting. Below I see all those files under arch/x86/lib/.

I think we should separate the x86 specific stuff from the general ACPI coding.

This is not meant to stop this series but can be done afterwards.

Best regards

Heinrich

Quoted 150 lines

- Drop acpi,name in example
- Drop duplicate assert
- Drop hid-descr-addr
- Drop mention of PRIC
- Fix 'RSDP' typo
- Fix 'XDST' typo
- Fix DMA_ typo
- Fix a few typos
- Fix file comment for acpi_table.c
- Fix indenting error mentioned by Andy Shevchenko
- Fix stray #endif
- Just add the device.txt binding file in this patch
- Make use of BIT()
- Move acpi_align_large() out of dm_test_acpi_setup_base_tables()
- Move acpi_table.h to include/acpi
- Rename acpi,desc to acpi,ddn
- Rename acpi_align_large() to acpi_align64()
- Rename acpi_return_name() to acpi_copy_name()
- Reword commit message to drop the bit about ACPI being complicated
- Split out hid-over-i2c into its own patch
- Update commit message to say that we move most of its contents
- Use strncpy() instead of strcpy() in acpi_copy_name()

Changes in v2:
- Add in the acpi_table.h header file to this patch
- Add the hid-over-i2c binding document
- Add trailing commas to enum
- Don't bracket the definitions with DM_SPI
- Drop definition of ACPI_TABLE_CREATOR
- Drop the Chrome OS pieces
- Drop the other comment change since it is already applied
- Drop two unnecessary __packed
- Fix definition of HID
- Generalise the ACPI function recursion with acpi_recurse_method()
- Infer hid-over-i2c CID value
- Make _acpi_write_dev_tables() static and switch argument order
- Move LOGC_ACPI definition to this patch
- Move __packed to after struct
- Move the sandbox acpi_table.h header file to an earlier patch
- Rename the 'coreboot' console to 'U-Boot'
- Use #defines for MADT and MCFG version numbers

Simon Glass (29):
cpu: Support querying the address width
spi: Add SPI mode enums
tpm: cr50: Release locality on exit
tpm: cr50: Add a comment for cr50_priv
tpm: cr50: Use the correct GPIO binding
tpm: Don't cleanup unless an error happens
dm: pci: Allow disabling auto-config for a device
x86: Correct wording of coreboot source code
x86: apl: Move p2sb ofdata reading to the correct method
pci: Adjust dm_pci_read_bar32() to return errors correctly
x86: apl: Add Global NVS table header
dm: core: Add basic ACPI support
dts: Add a binding for hid-over-i2c
acpi: Add a binding for ACPI settings in the device tree
acpi: Add a simple sandbox test
x86: Move acpi_s3.h to include/acpi/
x86: Move acpi_table header to main include/ directory
acpi: Add an __ACPI__ preprocessor symbol
acpi: Add a central location for table version numbers
acpi: Add support for DMAR
test: Add hexdump.h to the unit test header
acpi: Add a method to write tables for a device
acpi: Convert part of acpi_table to use acpi_ctx
x86: Allow devices to write ACPI tables
acpi: Drop code for missing XSDT from acpi_write_rsdp()
acpi: Move acpi_add_table() to generic code
acpi: Put table-setup code in its own function
acpi: Move the xsdt pointer to acpi_ctx
acpi: Add an acpi command

  arch/sandbox/dts/test.dts                     |   8 +
  arch/sandbox/include/asm/acpi_table.h         |   9 +
  arch/sandbox/include/asm/global_data.h        |   1 +
  arch/x86/cpu/apollolake/cpu_spl.c             |   2 +-
  arch/x86/cpu/apollolake/fsp_s.c               |   2 +-
  arch/x86/cpu/apollolake/pmc.c                 |   2 +-
  arch/x86/cpu/baytrail/acpi.c                  |   6 +-
  arch/x86/cpu/coreboot/timestamp.c             |   4 +-
  arch/x86/cpu/cpu.c                            |   4 +-
  arch/x86/cpu/intel_common/p2sb.c              |  33 +-
  arch/x86/cpu/quark/acpi.c                     |   2 +-
  arch/x86/cpu/tangier/acpi.c                   |   4 +-
  arch/x86/cpu/wakeup.S                         |   2 +-
  arch/x86/dts/chromebook_coral.dts             |   2 +-
  arch/x86/include/asm/acpi_table.h             | 381 +-----------
  .../include/asm/arch-apollolake/global_nvs.h  |  36 ++
  .../x86/include/asm/arch-coreboot/timestamp.h |   4 +-
  arch/x86/include/asm/global_data.h            |   1 +
  arch/x86/include/asm/intel_pinctrl_defs.h     |   2 -
  arch/x86/lib/acpi.c                           |   2 +-
  arch/x86/lib/acpi_s3.c                        |   4 +-
  arch/x86/lib/acpi_table.c                     | 240 ++------
  arch/x86/lib/coreboot_table.c                 |   2 +-
  arch/x86/lib/fsp/fsp_common.c                 |   2 +-
  arch/x86/lib/fsp1/fsp_common.c                |   2 +-
  arch/x86/lib/fsp2/fsp_dram.c                  |   2 +-
  arch/x86/lib/tables.c                         |   2 +-
  arch/x86/lib/zimage.c                         |   2 +-
  cmd/Kconfig                                   |  14 +
  cmd/Makefile                                  |   1 +
  cmd/acpi.c                                    | 179 ++++++
  doc/device-tree-bindings/device.txt           |  37 ++
  .../gpio/intel,apl-gpio.txt                   |   2 +-
  .../input/hid-over-i2c.txt                    |  44 ++
  .../interrupt-controller/intel,acpi-gpe.txt   |   2 +-
  doc/device-tree-bindings/pci/x86-pci.txt      |  23 +
  drivers/core/Kconfig                          |   9 +
  drivers/core/Makefile                         |   1 +
  drivers/core/acpi.c                           |  94 +++
  drivers/cpu/cpu_sandbox.c                     |   1 +
  drivers/pci/pci-uclass.c                      |  11 +-
  drivers/pci/pci_rom.c                         |   4 +-
  drivers/power/acpi_pmc/acpi-pmc-uclass.c      |   2 +-
  drivers/sysreset/sysreset_x86.c               |   2 +-
  drivers/tpm/cr50_i2c.c                        |  24 +-
  drivers/tpm/tpm-uclass.c                      |  13 +-
  include/{ => acpi}/acpi_s3.h                  |   0
  include/acpi/acpi_table.h                     | 581 ++++++++++++++++++
  include/cpu.h                                 |   2 +
  include/dm/acpi.h                             | 114 ++++
  include/dm/device.h                           |   5 +
  include/dm/uclass-id.h                        |   1 +
  include/log.h                                 |   2 +
  include/spi.h                                 |  33 +
  include/test/ut.h                             |   1 +
  lib/Makefile                                  |   1 +
  lib/acpi/Makefile                             |   4 +
  lib/acpi/acpi_table.c                         | 257 ++++++++
  lib/efi_loader/efi_acpi.c                     |   2 +-
  scripts/Makefile.lib                          |   4 +-
  test/dm/Makefile                              |   1 +
  test/dm/acpi.c                                | 317 ++++++++++
  test/dm/cpu.c                                 |   1 +
  63 files changed, 1915 insertions(+), 637 deletions(-)

create mode 100644 arch/sandbox/include/asm/acpi_table.h create mode 100644 arch/x86/include/asm/arch-apollolake/global_nvs.h create mode 100644 cmd/acpi.c
create mode 100644 doc/device-tree-bindings/device.txt create mode 100644 doc/device-tree-bindings/input/hid-over-i2c.txt create mode 100644 drivers/core/acpi.c
rename include/{ => acpi}/acpi_s3.h (100%)
create mode 100644 include/acpi/acpi_table.h
create mode 100644 include/dm/acpi.h
create mode 100644 lib/acpi/Makefile
create mode 100644 lib/acpi/acpi_table.c
create mode 100644 test/dm/acpi.c

#85 lore raw

v3 cover dm: Add programmatic generation of ACPI tables (part A)

Hi Heinrich,

Quoted 20 lines · Heinrich Schuchardt
Quoted 5 lines · Simon Glass

This is split from the original series in an attempt to get things applied in chunks.

This section includes patches up to and including the 'acpi' command.

Quoted 11 lines

Changes in v3:
- Add a pointer to information about acpi,compatible
- Add forward declarations for the functions
- Add missing error check in acpi_create_dmar()
- Add new patch to add hexdump.h to the unit test header
- Add new patch to move acpi_s3.h to include/acpi/
- Beef up the comment explaining how the unaligned address is used
- Change the example to ELAN
- Compute ACPI_NAME_MAX from ACPI_NAME_LEN
- Correct description of acpi,probed
- Drop 'Intel' from 'Intel ACPI'

Quoted 4 lines

Hello Simon,

not only x86 but also ARM can use ACPI tables for booting. Below I see all those files under arch/x86/lib/.

Not all. Actually quite a few are in drivers/core and lib/acpi.

Quoted 5 lines

I think we should separate the x86 specific stuff from the general ACPI coding.

This is not meant to stop this series but can be done afterwards.

Yes we can easily move things when needed and I've already moved things based on feedback from ARM. I really don't want the location of code to hold up progress.

Regards,
Simon

Quoted 156 lines

Best regards

Heinrich

Quoted 150 lines

- Drop acpi,name in example
- Drop duplicate assert
- Drop hid-descr-addr
- Drop mention of PRIC
- Fix 'RSDP' typo
- Fix 'XDST' typo
- Fix DMA_ typo
- Fix a few typos
- Fix file comment for acpi_table.c
- Fix indenting error mentioned by Andy Shevchenko
- Fix stray #endif
- Just add the device.txt binding file in this patch
- Make use of BIT()
- Move acpi_align_large() out of dm_test_acpi_setup_base_tables()
- Move acpi_table.h to include/acpi
- Rename acpi,desc to acpi,ddn
- Rename acpi_align_large() to acpi_align64()
- Rename acpi_return_name() to acpi_copy_name()
- Reword commit message to drop the bit about ACPI being complicated
- Split out hid-over-i2c into its own patch
- Update commit message to say that we move most of its contents
- Use strncpy() instead of strcpy() in acpi_copy_name()

Changes in v2:
- Add in the acpi_table.h header file to this patch
- Add the hid-over-i2c binding document
- Add trailing commas to enum
- Don't bracket the definitions with DM_SPI
- Drop definition of ACPI_TABLE_CREATOR
- Drop the Chrome OS pieces
- Drop the other comment change since it is already applied
- Drop two unnecessary __packed
- Fix definition of HID
- Generalise the ACPI function recursion with acpi_recurse_method()
- Infer hid-over-i2c CID value
- Make _acpi_write_dev_tables() static and switch argument order
- Move LOGC_ACPI definition to this patch
- Move __packed to after struct
- Move the sandbox acpi_table.h header file to an earlier patch
- Rename the 'coreboot' console to 'U-Boot'
- Use #defines for MADT and MCFG version numbers

Simon Glass (29):
cpu: Support querying the address width
spi: Add SPI mode enums
tpm: cr50: Release locality on exit
tpm: cr50: Add a comment for cr50_priv
tpm: cr50: Use the correct GPIO binding
tpm: Don't cleanup unless an error happens
dm: pci: Allow disabling auto-config for a device
x86: Correct wording of coreboot source code
x86: apl: Move p2sb ofdata reading to the correct method
pci: Adjust dm_pci_read_bar32() to return errors correctly
x86: apl: Add Global NVS table header
dm: core: Add basic ACPI support
dts: Add a binding for hid-over-i2c
acpi: Add a binding for ACPI settings in the device tree
acpi: Add a simple sandbox test
x86: Move acpi_s3.h to include/acpi/
x86: Move acpi_table header to main include/ directory
acpi: Add an __ACPI__ preprocessor symbol
acpi: Add a central location for table version numbers
acpi: Add support for DMAR
test: Add hexdump.h to the unit test header
acpi: Add a method to write tables for a device
acpi: Convert part of acpi_table to use acpi_ctx
x86: Allow devices to write ACPI tables
acpi: Drop code for missing XSDT from acpi_write_rsdp()
acpi: Move acpi_add_table() to generic code
acpi: Put table-setup code in its own function
acpi: Move the xsdt pointer to acpi_ctx
acpi: Add an acpi command

  arch/sandbox/dts/test.dts                     |   8 +
  arch/sandbox/include/asm/acpi_table.h         |   9 +
  arch/sandbox/include/asm/global_data.h        |   1 +
  arch/x86/cpu/apollolake/cpu_spl.c             |   2 +-
  arch/x86/cpu/apollolake/fsp_s.c               |   2 +-
  arch/x86/cpu/apollolake/pmc.c                 |   2 +-
  arch/x86/cpu/baytrail/acpi.c                  |   6 +-
  arch/x86/cpu/coreboot/timestamp.c             |   4 +-
  arch/x86/cpu/cpu.c                            |   4 +-
  arch/x86/cpu/intel_common/p2sb.c              |  33 +-
  arch/x86/cpu/quark/acpi.c                     |   2 +-
  arch/x86/cpu/tangier/acpi.c                   |   4 +-
  arch/x86/cpu/wakeup.S                         |   2 +-
  arch/x86/dts/chromebook_coral.dts             |   2 +-
  arch/x86/include/asm/acpi_table.h             | 381 +-----------
  .../include/asm/arch-apollolake/global_nvs.h  |  36 ++
  .../x86/include/asm/arch-coreboot/timestamp.h |   4 +-
  arch/x86/include/asm/global_data.h            |   1 +
  arch/x86/include/asm/intel_pinctrl_defs.h     |   2 -
  arch/x86/lib/acpi.c                           |   2 +-
  arch/x86/lib/acpi_s3.c                        |   4 +-
  arch/x86/lib/acpi_table.c                     | 240 ++------
  arch/x86/lib/coreboot_table.c                 |   2 +-
  arch/x86/lib/fsp/fsp_common.c                 |   2 +-
  arch/x86/lib/fsp1/fsp_common.c                |   2 +-
  arch/x86/lib/fsp2/fsp_dram.c                  |   2 +-
  arch/x86/lib/tables.c                         |   2 +-
  arch/x86/lib/zimage.c                         |   2 +-
  cmd/Kconfig                                   |  14 +
  cmd/Makefile                                  |   1 +
  cmd/acpi.c                                    | 179 ++++++
  doc/device-tree-bindings/device.txt           |  37 ++
  .../gpio/intel,apl-gpio.txt                   |   2 +-
  .../input/hid-over-i2c.txt                    |  44 ++
  .../interrupt-controller/intel,acpi-gpe.txt   |   2 +-
  doc/device-tree-bindings/pci/x86-pci.txt      |  23 +
  drivers/core/Kconfig                          |   9 +
  drivers/core/Makefile                         |   1 +
  drivers/core/acpi.c                           |  94 +++
  drivers/cpu/cpu_sandbox.c                     |   1 +
  drivers/pci/pci-uclass.c                      |  11 +-
  drivers/pci/pci_rom.c                         |   4 +-
  drivers/power/acpi_pmc/acpi-pmc-uclass.c      |   2 +-
  drivers/sysreset/sysreset_x86.c               |   2 +-
  drivers/tpm/cr50_i2c.c                        |  24 +-
  drivers/tpm/tpm-uclass.c                      |  13 +-
  include/{ => acpi}/acpi_s3.h                  |   0
  include/acpi/acpi_table.h                     | 581 ++++++++++++++++++
  include/cpu.h                                 |   2 +
  include/dm/acpi.h                             | 114 ++++
  include/dm/device.h                           |   5 +
  include/dm/uclass-id.h                        |   1 +
  include/log.h                                 |   2 +
  include/spi.h                                 |  33 +
  include/test/ut.h                             |   1 +
  lib/Makefile                                  |   1 +
  lib/acpi/Makefile                             |   4 +
  lib/acpi/acpi_table.c                         | 257 ++++++++
  lib/efi_loader/efi_acpi.c                     |   2 +-
  scripts/Makefile.lib                          |   4 +-
  test/dm/Makefile                              |   1 +
  test/dm/acpi.c                                | 317 ++++++++++
  test/dm/cpu.c                                 |   1 +
  63 files changed, 1915 insertions(+), 637 deletions(-)

create mode 100644 arch/sandbox/include/asm/acpi_table.h create mode 100644 arch/x86/include/asm/arch-apollolake/global_nvs.h create mode 100644 cmd/acpi.c
create mode 100644 doc/device-tree-bindings/device.txt create mode 100644 doc/device-tree-bindings/input/hid-over-i2c.txt create mode 100644 drivers/core/acpi.c
rename include/{ => acpi}/acpi_s3.h (100%)
create mode 100644 include/acpi/acpi_table.h
create mode 100644 include/dm/acpi.h
create mode 100644 lib/acpi/Makefile
create mode 100644 lib/acpi/acpi_table.c
create mode 100644 test/dm/acpi.c