Skip to contentSkip to conversation

misc: fs_loader: reorg and split to FS and FW loader + FIP loader

PATCH Superseded Christian Marangi 2 participants 16 messages Oct 22 – Oct 23 Char and misc drivers

v8 of 12 Superseded Latest is v12 →

lore
Share & tools
lore
#0 lore raw

v8 cover misc: fs_loader: reorg and split to FS and FW loader + FIP loader

This series mainly rework the FS loader to permit reading firmware from container that are not exactly a readable filesystem.

Also splitting out the generic firmware loader infra from the filesystem-specific code will make adding new loader types easier.

One scenario is when a firmware is shipped in a FIP container as a generic blob with an UUID.

FIP are mostly used on ARM in the context of ATF.
In such context U-Boot is loaded as BL31 and the PHY firmware can't be stored in the FIT image for the kernel as U-Boot should not depend on the presence of the kernel to correctly enable PHY for Ethernet port.

To handle such case the PHY firmware is stored in FIP with a predictable UUID.

One example is with Airoha 8811H firmware where the FIP can have a blob with UUID "d39d2cf2-9bd0-3ca7-93e9-e71b4f9250b2". (generated from command "uuidgen -n @dns --md5 --name en8811h.bin")

With these example DTS entry:

fs_loader0: fip-loader {
	bootph-all;
	compatible = "u-boot,fip-loader";
	phandlepart = <&mmc0 0>;
	partoffset = <0x100>;
};

mdio {

  en8811: ethernet-phy@f {
    reg = <0xf>;
    firmware-name = "d39d2cf2-9bd0-3ca7-93e9-e71b4f9250b2";
    firmware-loader = <&fs_loader0>;

};
};

And PHY driver using the get_fw_loader_from_node() and the common request_firmware_into_buf() it's possible to load the PHY firmware transparently by just declaring the required entry in the DTS.

get_fw_loader_from_node() is implemented to actual get the loader from DT. This was something that was already in mind from when the FS loader was implemented but then it was never implemented in favor of a single loader per device.

The first patch is a minor fixup for something that probably won't ever happen.

The second one is a good reworking of the FS and FW loader moving the internal struct to a dedicated header. (this is really to enforce what drivers should use and what driver should not mess with)

Then there is the request_firmware_size() new OP to get only the size of the firmware. Useful for case where the firmware size is not always the same and change across different version. (the patter might be get size -> alloc buffer -> get firmware).

Then the FIP loader as a basic parser of FIP. This only read the FIP header, loop all the entry and search for a matching UUID. If nothing is found then no firmware blob. Very simple implementation.

This is being CI tested on [0]

[0] https://github.com/u-boot/u-boot/pull/884

Changes v8:
- Address all the comments from Simon
- Massively improve test coverage for both FS and FIP
- Enable mmc node when needed for the tests
- Change FIP implementation to follow FS one (read up to buf size or end of entry size) (complain when offset is bigger than size of entry)
- Update related documentation for fw loader
- Rebase on top of next
Changes v7:
- Add tests
- Rebase on top of next
- Fix compilation for vpl
Changes v6:
- Rework to generic FW loader UCLASS
- Account for firmware offset in FIP loader
- Add UBI example in documentation
- Improve variables order
Changes v5:
- Fix unment dependency for config flag
- Move request_firmware_size before FIP loader
- Add documentation patch
- Improve get_fw_loader_from_node() implementation
- use min() instead of custom MIN()
- Fix handling of some return value
Changes v4:
- Rebase on top of next
Changes v3:
- Add review tag where possible
- Check blk_dread ret
- Generalize mount_ubifs with generic_fw_loader_ubi_select
- Fix some typo
- Drop useless ubifs umount (nothing is mounted in FIP) Changes v2:
- Better handle header include to fix compilation error on some devices
- Fix typo for ubifs fix commit

Christian Marangi (7):
misc: fs_loader: fix ubifs not unmounted on dev_get_priv error
misc: fs_loader: reorganize and split to FS loader and FW UCLASS
misc: fw_loader: implement generic get_fw_loader_from_node()
misc: fw_loader: implement request_firmware_size() OP
misc: fw_loader: introduce FIP loader driver
doc: dtbinding: Update documentation for Generic Firmware loader
test: dm: Add test for FW Loader feature

 arch/sandbox/dts/test.dts                     |  38 ++
 configs/sandbox64_defconfig                   |   2 +
 configs/sandbox_defconfig                     |   2 +
 configs/sandbox_flattree_defconfig            |   2 +
 configs/sandbox_noinst_defconfig              |   1 +
 configs/sandbox_spl_defconfig                 |   2 +
 configs/sandbox_vpl_defconfig                 |   2 +
 ...irmware_loader.rst => firmware_loader.rst} |   6 +-
 doc/develop/driver-model/index.rst            |   2 +-
 doc/device-tree-bindings/misc/fs_loader.txt   |  48 --
 doc/device-tree-bindings/misc/fw_loader.yaml  | 126 ++++
 drivers/misc/Kconfig                          |  31 +
 drivers/misc/Makefile                         |   2 +-
 drivers/misc/fs_loader.c                      | 326 ----------
 drivers/misc/fw_loader/Makefile               |   5 +
 drivers/misc/fw_loader/fip_loader.c           | 556 ++++++++++++++++++
 drivers/misc/fw_loader/fs_loader.c            | 220 +++++++
 drivers/misc/fw_loader/fw_loader-uclass.c     | 231 ++++++++
 drivers/misc/fw_loader/internal.h             | 100 ++++
 include/dm/uclass-id.h                        |   2 +-
 include/fs_loader.h                           |  47 +-
 include/fw_loader.h                           |  42 ++
 test/dm/Makefile                              |   1 +
 test/dm/fw_loader.c                           | 218 +++++++
 test/py/tests/fip/test.fip                    | Bin 0 -> 101 bytes
 test/py/tests/test_ut.py                      |  26 +
 26 files changed, 1613 insertions(+), 425 deletions(-)
 rename doc/develop/driver-model/{fs_firmware_loader.rst => firmware_loader.rst} (95%)
 delete mode 100644 doc/device-tree-bindings/misc/fs_loader.txt
 create mode 100644 doc/device-tree-bindings/misc/fw_loader.yaml
 delete mode 100644 drivers/misc/fs_loader.c
 create mode 100644 drivers/misc/fw_loader/Makefile
 create mode 100644 drivers/misc/fw_loader/fip_loader.c
 create mode 100644 drivers/misc/fw_loader/fs_loader.c
 create mode 100644 drivers/misc/fw_loader/fw_loader-uclass.c
 create mode 100644 drivers/misc/fw_loader/internal.h
 create mode 100644 test/dm/fw_loader.c
 create mode 100644 test/py/tests/fip/test.fip
Signature · 1 line
2.55.0
#1 lore raw

PATCH v8 1/7 misc: fs_loader: fix ubifs not unmounted on dev_get_priv error

When dev_get_priv errors out, the ubifs is not umounted (if used).

Correctly handle this handle condition and while at it also return -EINVAL instead of -ENOMEM as a better error since no memory is allocated but is actually an invalid scenario for fw_get_filesystem_firmware().

---
 drivers/misc/fs_loader.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/misc/fs_loader.c +4 -2
index 6af4c7f15e7b..cd4695b08eea 100644
@@ -175,8 +175,10 @@ static int fw_get_filesystem_firmware(struct udevice *dev)
 
 	struct firmware *firmwarep = dev_get_priv(dev);
 
-	if (!firmwarep)
-		return -ENOMEM;
+	if (!firmwarep) {
+		ret = -EINVAL;
+		goto out;
+	}
 
 	ret = fs_read(firmwarep->name, (ulong)map_to_sysmem(firmwarep->data),
 			firmwarep->offset, firmwarep->size, &actread);
Signature · 1 line
2.55.0
#2 lore raw

PATCH v8 2/7 misc: fs_loader: reorganize and split to FS loader and FW UCLASS

In reply to Christian Marangi

In preparation to the introduction of variant of the FS loader, reorganize and split the driver to generic fw_loader UCLASS and specific fs_loader function. Create a dedicated directory for the loader and move the internal structs and functions to a dedicated header file. While at it also drop some redundant check and rename some variables as we are updating the code for the UCLASS rework.
Doc are renamed and updated to a more generic name and for the new UCLASS.

This will permit to reuse all the property and logic of FS loader with container that are not exactly a readable filesystem.

---
 ...irmware_loader.rst => firmware_loader.rst} |   6 +-
 doc/develop/driver-model/index.rst            |   2 +-
 drivers/misc/Kconfig                          |   8 +
 drivers/misc/Makefile                         |   2 +-
 drivers/misc/fs_loader.c                      | 328 ------------------
 drivers/misc/fw_loader/Makefile               |   4 +
 drivers/misc/fw_loader/fs_loader.c            | 176 ++++++++++
 drivers/misc/fw_loader/fw_loader-uclass.c     | 188 ++++++++++
 drivers/misc/fw_loader/internal.h             |  91 +++++
 include/dm/uclass-id.h                        |   2 +-
 include/fs_loader.h                           |  47 +--
 include/fw_loader.h                           |  19 +
 12 files changed, 494 insertions(+), 379 deletions(-)
 rename doc/develop/driver-model/{fs_firmware_loader.rst => firmware_loader.rst} (95%)
 delete mode 100644 drivers/misc/fs_loader.c
 create mode 100644 drivers/misc/fw_loader/Makefile
 create mode 100644 drivers/misc/fw_loader/fs_loader.c
 create mode 100644 drivers/misc/fw_loader/fw_loader-uclass.c
 create mode 100644 drivers/misc/fw_loader/internal.h
doc/develop/driver-model/firmware_loader.rst renamed from doc/develop/driver-model/fs_firmware_loader.rst +3 -3
similarity index 95%
rename from doc/develop/driver-model/fs_firmware_loader.rst
rename to doc/develop/driver-model/firmware_loader.rst
index 630daf636f18..2d11a74d5e2f 100644
@@ -4,7 +4,7 @@
 File System Firmware Loader
 ===========================
 
-This is file system firmware loader for U-Boot framework, which has very close
+This is firmware loader for U-Boot framework, which has very close
 to some Linux Firmware API. For the details of Linux Firmware API, you can refer
 to https://docs.kernel.org/driver-api/firmware/index.html.
 
@@ -87,7 +87,7 @@ For example of getting DT phandle from /chosen and creating instance:
 	}
 
 	phandle = fdt32_to_cpu(*phandle_p);
-	ret = uclass_get_device_by_phandle_id(UCLASS_FS_FIRMWARE_LOADER,
+	ret = uclass_get_device_by_phandle_id(UCLASS_FIRMWARE_LOADER,
 					     phandle, &dev);
 	if (ret)
 		return ret;
@@ -145,7 +145,7 @@ instance:
 
 .. code-block:: c
 
-	ret = uclass_get_device_by_phandle_id(UCLASS_FS_FIRMWARE_LOADER,
+	ret = uclass_get_device_by_phandle_id(UCLASS_FIRMWARE_LOADER,
 					     phandle, &dev);
 	if (ret)
 		return ret;
doc/develop/driver-model/index.rst +1 -1
index 8e12bbd9366a..b2d46da4e239 100644
@@ -15,7 +15,7 @@ subsystems
    design
    ethernet
    fdt-fixup
-   fs_firmware_loader
+   firmware_loader
    i2c-howto
    livetree
    migration
drivers/misc/Kconfig +8 -0
index 5f22bcb330ee..1cab469a3434 100644
@@ -621,8 +621,12 @@ config MPC83XX_SERDES
 	help
 	  Support for serdes found on MPC83xx SoCs.
 
+config FW_LOADER_UCLASS
+	bool
+
 config FS_LOADER
 	bool "Enable loader driver for file system"
+	select FW_LOADER_UCLASS
 	help
 	  This is file system generic loader which can be used to load
 	  the file image from the storage into target such as memory.
@@ -630,8 +634,12 @@ config FS_LOADER
 	  The consumer driver would then use this loader to program whatever,
 	  ie. the FPGA device.
 
+config SPL_FW_LOADER_UCLASS
+	bool
+
 config SPL_FS_LOADER
 	bool "Enable loader driver for file system in SPL"
+	select SPL_FW_LOADER_UCLASS
 	depends on SPL
 	help
 	  This is file system generic loader which can be used to load
drivers/misc/Makefile +1 -1
index 9ddbe9cd1142..f138ff9126f1 100644
@@ -36,7 +36,7 @@ obj-$(CONFIG_FSL_DEVICE_DISABLE) += fsl_devdis.o
 obj-$(CONFIG_FSL_IFC) += fsl_ifc.o
 obj-$(CONFIG_FSL_IIM) += fsl_iim.o
 obj-$(CONFIG_FSL_SEC_MON) += fsl_sec_mon.o
-obj-$(CONFIG_$(PHASE_)FS_LOADER) += fs_loader.o
+obj-$(CONFIG_$(PHASE_)FW_LOADER_UCLASS) += fw_loader/
 obj-$(CONFIG_GATEWORKS_SC) += gsc.o
 obj-$(CONFIG_GXP_EEPROM) += gxp_eeprom.o
 obj-$(CONFIG_GDSYS_IOEP) += gdsys_ioep.o
drivers/misc/fs_loader.c deleted +0 -328

328 lines not on this page · raw message

drivers/misc/fw_loader/Makefile new file +4 -0
new file mode 100644
index 000000000000..f5df0db61c22
@@ -0,0 +1,4 @@
+# SPDX-License-Identifier: GPL-2.0+
+
+obj-$(CONFIG_$(PHASE_)FW_LOADER_UCLASS) += fw_loader-uclass.o
+obj-$(CONFIG_$(PHASE_)FS_LOADER) += fs_loader.o
drivers/misc/fw_loader/fs_loader.c new file +176 -0

176 lines not on this page · raw message

drivers/misc/fw_loader/fw_loader-uclass.c new file +188 -0

188 lines not on this page · raw message

drivers/misc/fw_loader/internal.h new file +91 -0

91 lines not on this page · raw message

include/dm/uclass-id.h +1 -1
index 7de9312cde7a..48f36e88fee9 100644
@@ -69,7 +69,7 @@ enum uclass_id {
 	UCLASS_FIRMWARE,	/* Firmware */
 	UCLASS_FPGA,		/* FPGA device */
 	UCLASS_FUZZING_ENGINE,	/* Fuzzing engine */
-	UCLASS_FS_FIRMWARE_LOADER,		/* Generic loader */
+	UCLASS_FIRMWARE_LOADER,	/* Firmware loader */
 	UCLASS_FWU_MDATA,	/* FWU Metadata Access */
 	UCLASS_GPIO,		/* Bank of general-purpose I/O pins */
 	UCLASS_HASH,		/* Hash device */
include/fs_loader.h +2 -45

54 lines not on this page · raw message

include/fw_loader.h +19 -0
index 35574482b2b9..56f5e3be6195 100644
@@ -1,10 +1,29 @@
 /* SPDX-License-Identifier: GPL-2.0 */
 /*
+ * Copyright (C) 2018-2019 Intel Corporation <www.intel.com>
  * Copyright (C) 2025 Lucien Jheng <[email protected]>
  */
 #ifndef _FW_LOADER_H_
 #define _FW_LOADER_H_
 
+struct udevice;
+
+/**
+ * request_firmware_into_buf - Load firmware into a previously allocated buffer.
+ * @dev: An instance of a driver.
+ * @name: Name of firmware file.
+ * @buf: Address of buffer to load firmware into.
+ * @size: Size of buffer.
+ * @offset: Offset of a file for start reading into buffer.
+ *
+ * The firmware is loaded directly into the buffer pointed to by @buf.
+ *
+ * Return: Size of total read, negative value when error.
+ */
+int request_firmware_into_buf(struct udevice *dev,
+			      const char *name,
+			      void *buf, size_t size, u32 offset);
+
 /**
  * request_firmware_into_buf_via_script() -
  * Load firmware using a U-Boot script and copy to buffer
Signature · 1 line
2.55.0
#3 lore raw

Re: [PATCH v8 2/7] misc: fs_loader: reorganize and split to FS loader and FW UCLASS

Hi Christian,

Quoted 31 lines · Christian Marangi (Ansuel)

misc: fs_loader: reorganize and split to FS loader and FW UCLASS

In preparation to the introduction of variant of the FS loader, reorganize and split the driver to generic fw_loader UCLASS and specific fs_loader function. Create a dedicated directory for the loader and move the internal structs and functions to a dedicated header file. While at it also drop some redundant check and rename some variables as we are updating the code for the UCLASS rework.
Doc are renamed and updated to a more generic name and for the new UCLASS.

This will permit to reuse all the property and logic of FS loader with container that are not exactly a readable filesystem.

 ...{fs_firmware_loader.rst => firmware_loader.rst} |   6 +-
 doc/develop/driver-model/index.rst                 |   2 +-
 drivers/misc/Kconfig                               |   8 +
 drivers/misc/Makefile                              |   2 +-
 drivers/misc/fs_loader.c                           | 328 ---------------------
 drivers/misc/fw_loader/Makefile                    |   4 +
 drivers/misc/fw_loader/fs_loader.c                 | 176 +++++++++++
 drivers/misc/fw_loader/fw_loader-uclass.c          | 188 ++++++++++++
 drivers/misc/fw_loader/internal.h                  |  91 ++++++
 include/dm/uclass-id.h                             |   2 +-
 include/fs_loader.h                                |  47 +--
 include/fw_loader.h                                |  19 ++
 12 files changed, 494 insertions(+), 379 deletions(-)

drivers/misc/fw_loader/fs_loader.c line 6

Quoted 6 lines
drivers/misc/fw_loader/fs_loader.c +6 -0
@@ -0,0 +1,176 @@
+	if (plat->phandlepart.phandle) {
+		struct blk_desc *desc;
+
+		ret = generic_fw_loader_get_phandle_desc(plat, &desc);
+		if (!desc)
+			debug("%s: No device found\n", __func__);

generic_fw_loader_get_phandle_desc() only sets *desc on success, so if device_get_global_by_ofnode() or blk_get_by_device() fails, 'desc' is read uninitialised. Please can you check 'ret' instead, e.g.

   ret = generic_fw_loader_get_phandle_desc(plat, &desc);
   if (ret)
           debug("%s: No device found\n", __func__);
   else
           ret = fs_set_blk_dev_with_part(desc,
                                          plat->phandlepart.partition);

Regards,
Simon

#4 lore raw

Re: [PATCH v8 2/7] misc: fs_loader: reorganize and split to FS loader and FW UCLASS

Quoted 45 lines · Simon Glass

Hi Christian,

Quoted 31 lines · Christian Marangi (Ansuel)

misc: fs_loader: reorganize and split to FS loader and FW UCLASS

In preparation to the introduction of variant of the FS loader, reorganize and split the driver to generic fw_loader UCLASS and specific fs_loader function. Create a dedicated directory for the loader and move the internal structs and functions to a dedicated header file. While at it also drop some redundant check and rename some variables as we are updating the code for the UCLASS rework.
Doc are renamed and updated to a more generic name and for the new UCLASS.

This will permit to reuse all the property and logic of FS loader with container that are not exactly a readable filesystem.

 ...{fs_firmware_loader.rst => firmware_loader.rst} |   6 +-
 doc/develop/driver-model/index.rst                 |   2 +-
 drivers/misc/Kconfig                               |   8 +
 drivers/misc/Makefile                              |   2 +-
 drivers/misc/fs_loader.c                           | 328 ---------------------
 drivers/misc/fw_loader/Makefile                    |   4 +
 drivers/misc/fw_loader/fs_loader.c                 | 176 +++++++++++
 drivers/misc/fw_loader/fw_loader-uclass.c          | 188 ++++++++++++
 drivers/misc/fw_loader/internal.h                  |  91 ++++++
 include/dm/uclass-id.h                             |   2 +-
 include/fs_loader.h                                |  47 +--
 include/fw_loader.h                                |  19 ++
 12 files changed, 494 insertions(+), 379 deletions(-)

drivers/misc/fw_loader/fs_loader.c line 6

Quoted 6 lines
drivers/misc/fw_loader/fs_loader.c +6 -0
@@ -0,0 +1,176 @@
+	if (plat->phandlepart.phandle) {
+		struct blk_desc *desc;
+
+		ret = generic_fw_loader_get_phandle_desc(plat, &desc);
+		if (!desc)
+			debug("%s: No device found\n", __func__);

generic_fw_loader_get_phandle_desc() only sets *desc on success, so if device_get_global_by_ofnode() or blk_get_by_device() fails, 'desc' is read uninitialised. Please can you check 'ret' instead, e.g.

Quoted 7 lines
   ret = generic_fw_loader_get_phandle_desc(plat, &desc);
   if (ret)
           debug("%s: No device found\n", __func__);
   else
           ret = fs_set_blk_dev_with_part(desc,
                                          plat->phandlepart.partition);

Hi Simon,

isn't the current logic the exact implementation you are suggesting?

But yes I will update with the compacted if-else since it's more readable.

Signature · 1 line
	Ansuel
#5 lore raw

PATCH v8 3/7 misc: fw_loader: implement generic get_fw_loader_from_node()

In reply to Christian Marangi

Implement a generic function to get a FW loader dev from a node.

There is currently only get_fs_loader() but that is limited to chosen node and is limited only to FS loader.

Introduce get_fw_loader_from_node() that will parse the "firmware-loader" from a specified node and will search and probe a matching FW loader device, if found.

---
 drivers/misc/fw_loader/fw_loader-uclass.c | 24 +++++++++++++++++++++++
 include/fw_loader.h                       | 14 +++++++++++++
 2 files changed, 38 insertions(+)
drivers/misc/fw_loader/fw_loader-uclass.c +24 -0
index 0d95fac8f42b..7f19d6c4af4c 100644
@@ -7,6 +7,8 @@
 #include <blk.h>
 #include <dm.h>
 #include <dm/device.h>
+#include <dm/device-internal.h>
+#include <dm/uclass.h>
 #include <env.h>
 #include <errno.h>
 #include <fw_loader.h>
@@ -137,6 +139,28 @@ UCLASS_DRIVER(fw_loader) = {
 	.per_device_auto = sizeof(struct firmware),
 };
 
+int get_fw_loader_from_node(ofnode node, struct udevice **dev)
+{
+	struct udevice *fw_dev;
+	int ret;
+
+	node = ofnode_parse_phandle(node, "firmware-loader", 0);
+	if (!ofnode_valid(node))
+		return -ENODEV;
+
+	ret = uclass_get_device_by_ofnode(UCLASS_FIRMWARE_LOADER,
+					  node, &fw_dev);
+	if (ret)
+		return ret;
+
+	ret = device_probe(fw_dev);
+	if (ret)
+		return ret;
+
+	*dev = fw_dev;
+	return 0;
+}
+
 /**
  * _request_firmware_prepare - Prepare firmware struct.
  *
include/fw_loader.h +14 -0
index 56f5e3be6195..99c47380a172 100644
@@ -6,8 +6,22 @@
 #ifndef _FW_LOADER_H_
 #define _FW_LOADER_H_
 
+#include <dm/ofnode_decl.h>
+
 struct udevice;
 
+/**
+ * get_fw_loader_from_node - Get FW loader dev from @node.
+ *
+ * @node: ofnode where "firmware-loader" phandle is stored.
+ * @dev: pointer where to store the FW loader dev.
+ *
+ * Search and probe a matching FW loader device, if found.
+ *
+ * Return: Negative value if fail, 0 for successful.
+ */
+int get_fw_loader_from_node(ofnode node, struct udevice **dev);
+
 /**
  * request_firmware_into_buf - Load firmware into a previously allocated buffer.
  * @dev: An instance of a driver.
Signature · 1 line
2.55.0
#6 lore raw

PATCH v8 4/7 misc: fw_loader: implement request_firmware_size() OP

In reply to Christian Marangi

It might be useful to get the firmware size before reading it to allocate the correct size. This is needed for case where the firmware size change across different firmware version and is not always the same. In such case the only way to handle this scenario is to allocate a big-enough buffer to read the firmware.

To better handle this, introduce the request_firmware_size() OP to get the firmware size before reading it. This way proper buffer can be allocated and dynamic firmware size can be better supported.

FS loader is updated to handle the new .get_size() handler.

---
 drivers/misc/fw_loader/fs_loader.c        | 68 +++++++++++++++++++----
 drivers/misc/fw_loader/fw_loader-uclass.c | 19 +++++++
 drivers/misc/fw_loader/internal.h         |  7 +++
 include/fw_loader.h                       |  9 +++
 4 files changed, 91 insertions(+), 12 deletions(-)
drivers/misc/fw_loader/fs_loader.c +56 -12
index 9923fb5e123d..c7aef99be8bb 100644
@@ -80,18 +80,10 @@ static int select_fs_dev(struct fw_loader_plat *plat)
 	return ret;
 }
 
-/**
- * fw_get_filesystem_firmware - load firmware into an allocated buffer.
- * @dev: An instance of a driver.
- *
- * Return: Size of total read, negative value when error.
- */
-static int fw_get_filesystem_firmware(struct udevice *dev)
+static int fw_get_filesystem_prepare(struct udevice *dev)
 {
-	struct firmware *upriv = dev_get_uclass_priv(dev);
 	struct fw_loader_plat *plat = dev_get_uclass_plat(dev);
 	struct fw_loader_env_info info = { };
-	loff_t actread = 0;
 	int ret;
 
 	generic_fw_loader_env_info(&info);
@@ -114,6 +106,29 @@ static int fw_get_filesystem_firmware(struct udevice *dev)
 		ret = select_fs_dev(plat);
 	}
 
+	return ret;
+}
+
+static void fw_get_filesystem_release(struct udevice *dev)
+{
+#ifdef CONFIG_CMD_UBIFS
+	umount_ubifs();
+#endif
+}
+
+/**
+ * fw_get_filesystem_firmware - load firmware into an allocated buffer.
+ * @dev: An instance of a driver.
+ *
+ * Return: Size of total read, negative value when error.
+ */
+static int fw_get_filesystem_firmware(struct udevice *dev)
+{
+	struct firmware *upriv = dev_get_uclass_priv(dev);
+	loff_t actread = 0;
+	int ret;
+
+	ret = fw_get_filesystem_prepare(dev);
 	if (ret)
 		goto out;
 
@@ -128,14 +143,43 @@ static int fw_get_filesystem_firmware(struct udevice *dev)
 	}
 
 out:
-#ifdef CONFIG_CMD_UBIFS
-	umount_ubifs();
-#endif
+	fw_get_filesystem_release(dev);
+	return ret;
+}
+
+/**
+ * fw_get_filesystem_firmware_size - get firmware size.
+ * @dev: An instance of a driver.
+ *
+ * Return: Size of firmware, negative value when error.
+ */
+static int fw_get_filesystem_firmware_size(struct udevice *dev)
+{
+	struct firmware *upriv = dev_get_uclass_priv(dev);
+	loff_t size = 0;
+	int ret;
+
+	ret = fw_get_filesystem_prepare(dev);
+	if (ret)
+		goto out;
+
+	ret = fs_size(upriv->name, &size);
+	if (ret) {
+		debug("Error: %d Failed to get size for %s.\n",
+		      ret, upriv->name);
+		goto out;
+	}
+
+	ret = size;
+
+out:
+	fw_get_filesystem_release(dev);
 	return ret;
 }
 
 static const struct fw_loader_ops fs_loader_ops = {
 	.get_firmware = fw_get_filesystem_firmware,
+	.get_size = fw_get_filesystem_firmware_size,
 };
 
 static const struct udevice_id fs_loader_ids[] = {
drivers/misc/fw_loader/fw_loader-uclass.c +19 -0
index 7f19d6c4af4c..4e67f2559fa2 100644
@@ -210,3 +210,22 @@ int request_firmware_into_buf(struct udevice *dev,
 
 	return ops->get_firmware(dev);
 }
+
+int request_firmware_size(struct udevice *dev, const char *name)
+{
+	const struct fw_loader_ops *ops;
+	int ret;
+
+	if (!dev)
+		return -EINVAL;
+
+	ret = _request_firmware_prepare(dev, name, NULL, 0, 0);
+	if (ret < 0) /* error */
+		return ret;
+
+	ops = fw_loader_get_ops(dev);
+	if (!ops->get_size)
+		return -EOPNOTSUPP;
+
+	return ops->get_size(dev);
+}
drivers/misc/fw_loader/internal.h +7 -0
index 9ff43e872554..1abe7504a81e 100644
@@ -32,6 +32,13 @@ struct fw_loader_ops {
 	 * @dev:	Firmware Loader device to read firmware from
 	 */
 	int (*get_firmware)(struct udevice *dev);
+
+	/**
+	 * get_size() - get firmware size from Firmware Loader driver
+	 *
+	 * @dev:	Firmware Loader device to get firmware size from
+	 */
+	int (*get_size)(struct udevice *dev);
 };
 
 #define fw_loader_get_ops(dev)	((const struct fw_loader_ops *)(dev)->driver->ops)
include/fw_loader.h +9 -0
index 99c47380a172..01f242dcbefd 100644
@@ -38,6 +38,15 @@ int request_firmware_into_buf(struct udevice *dev,
 			      const char *name,
 			      void *buf, size_t size, u32 offset);
 
+/**
+ * request_firmware_size - Get firmware size.
+ * @dev: An instance of a driver.
+ * @name: Name of firmware file.
+ *
+ * Return: Size of firmware, negative value when error.
+ */
+int request_firmware_size(struct udevice *dev, const char *name);
+
 /**
  * request_firmware_into_buf_via_script() -
  * Load firmware using a U-Boot script and copy to buffer
Signature · 1 line
2.55.0
#7 lore raw

PATCH v8 5/7 misc: fw_loader: introduce FIP loader driver

In reply to Christian Marangi

Introduce a variant of the FS loader driver to extract images from FIP image. These image can contain additional binary used to init Network accelerator or PHY firmware blob.

The way FIP handle image type is with the usage of UUID.

This FIP loader driver implement a simple FIP image parser that check every entry for a matching UUID.

Similar to FS loader, this driver also support both UBI and Block devices.

Also an additional property is added to handle special case with eMMC that doesn't have a GPT partition and require a global offset to reference the FIP partition.

An example usage of this driver is the following:

Entry in DTS:

	fs_loader0: fip-loader {
		bootph-all;
		compatible = "u-boot,fip-loader";
		phandlepart = <&mmc0 0>;
		partoffset = <0x100>;
	};

	ethernet@1fb50000 {
		firmware-loader = <&fs_loader0>;
	}

FIP loader user:

	/* get the FW loader from the ethernet node */
	get_fw_loader_from_node(dev_ofnode(dev), &fw_loader);

	/* read the blob identified by "58704aef-389f-3e52-b475-e0bf2234a6a2" UUID */
	request_firmware_into_buf(fw_loader, "58704aef-389f-3e52-b475-e0bf2234a6a2",
				  buf, 261400, 0);
---
 drivers/misc/Kconfig                |  23 ++
 drivers/misc/fw_loader/Makefile     |   1 +
 drivers/misc/fw_loader/fip_loader.c | 556 ++++++++++++++++++++++++++++
 drivers/misc/fw_loader/internal.h   |   2 +
 4 files changed, 582 insertions(+)
 create mode 100644 drivers/misc/fw_loader/fip_loader.c
drivers/misc/Kconfig +23 -0
index 1cab469a3434..f1091e96775e 100644
@@ -624,6 +624,17 @@ config MPC83XX_SERDES
 config FW_LOADER_UCLASS
 	bool
 
+config FIP_LOADER
+	bool "Enable loader driver from FIP partition"
+	select LIB_UUID
+	select FW_LOADER_UCLASS
+	help
+	  This is FIP partition generic loader which can be used to load
+	  the file image from the FIP image into target such as memory.
+
+	  The consumer driver would then use this loader to program whatever,
+	  ie. the FPGA device/PHY firmware.
+
 config FS_LOADER
 	bool "Enable loader driver for file system"
 	select FW_LOADER_UCLASS
@@ -637,6 +648,18 @@ config FS_LOADER
 config SPL_FW_LOADER_UCLASS
 	bool
 
+config SPL_FIP_LOADER
+	bool "Enable loader driver from FIP partition in SPL"
+	select LIB_UUID
+	select SPL_FW_LOADER_UCLASS
+	depends on SPL
+	help
+	  This is FIP partition generic loader which can be used to load
+	  the file image from the FIP image into target such as memory.
+
+	  The consumer driver would then use this loader to program whatever,
+	  ie. the FPGA device/PHY firmware.
+
 config SPL_FS_LOADER
 	bool "Enable loader driver for file system in SPL"
 	select SPL_FW_LOADER_UCLASS
drivers/misc/fw_loader/Makefile +1 -0
index f5df0db61c22..120722c4ed17 100644
@@ -1,4 +1,5 @@
 # SPDX-License-Identifier: GPL-2.0+
 
 obj-$(CONFIG_$(PHASE_)FW_LOADER_UCLASS) += fw_loader-uclass.o
+obj-$(CONFIG_$(PHASE_)FIP_LOADER) += fip_loader.o
 obj-$(CONFIG_$(PHASE_)FS_LOADER) += fs_loader.o
drivers/misc/fw_loader/fip_loader.c new file +556 -0

556 lines not on this page · raw message

drivers/misc/fw_loader/internal.h +2 -0
index 1abe7504a81e..5af061321512 100644
@@ -49,11 +49,13 @@ struct fw_loader_ops {
  * This holds information about all supported storage devices for driver use.
  *
  * @phandlepart: Attribute data for block device.
+ * @partoffset: Global offset for BLK partition.
  * @mtdpart: MTD partition for ubi partition.
  * @ubivol: UBI volume-name for ubifsmount.
  */
 struct fw_loader_plat {
 	struct phandle_part phandlepart;
+	u32 partoffset;
 	char *mtdpart;
 	char *ubivol;
 };
Signature · 1 line
2.55.0
#8 lore raw

Re: [PATCH v8 5/7] misc: fw_loader: introduce FIP loader driver

Hi Christian,

Quoted 30 lines · Christian Marangi (Ansuel)

misc: fw_loader: introduce FIP loader driver

Introduce a variant of the FS loader driver to extract images from FIP image. These image can contain additional binary used to init Network accelerator or PHY firmware blob.

The way FIP handle image type is with the usage of UUID.

This FIP loader driver implement a simple FIP image parser that check every entry for a matching UUID.

Similar to FS loader, this driver also support both UBI and Block devices.

Also an additional property is added to handle special case with eMMC that doesn't have a GPT partition and require a global offset to reference the FIP partition.

An example usage of this driver is the following:

[...]

 drivers/misc/Kconfig                |  23 ++
 drivers/misc/fw_loader/Makefile     |   1 +
 drivers/misc/fw_loader/fip_loader.c | 556 ++++++++++++++++++++++++++++++++++++
 drivers/misc/fw_loader/internal.h   |   2 +
 4 files changed, 582 insertions(+)

drivers/misc/fw_loader/fip_loader.c line 7

Quoted 7 lines
drivers/misc/fw_loader/fip_loader.c +7 -0
@@ -0,0 +1,556 @@
+static int blk_read_fip_firmware(struct firmware *firmwarep,
+				 struct blk_desc *desc,
+				 struct disk_partition *part_info,
+				 unsigned int part_offset,
+				 const struct fip_toc_entry *ent)
+{
+	unsigned int offset = part_info->start + part_offset;

The parse path now uses lbaint_t, but this read path still truncates part_info->start to unsigned int, so on a large device the header is found but the data is read from the wrong block. Please can you use lbaint_t here too (and for blkcnt/blkstart, to match blk_parse_fip_firmware())?

drivers/misc/fw_loader/fip_loader.c line 9

Quoted 11 lines
drivers/misc/fw_loader/fip_loader.c +9 -0
@@ -0,0 +1,556 @@
+	size = min(firmwarep->size, (size_t)ent->size - offset);
+	ret = ubi_volume_read(ubi_vol,
+			      (u8 *)firmwarep->data,
+			      ent->offset_address + offset,
+			      size);
+	if (ret)
+		return ret;
+
+	return size - firmwarep->offset;

'size' already has the offset taken off, so this subtracts it twice. With a non-zero offset the UBI path returns less than it actually read, whereas the block path returns the full 'read' count. I think this should just be 'return size;'

Also, ubi_volume_read() treats size == 0 as 'read the whole volume', which overruns the buffer. That happens if offset == ent->size (which the check in fw_get_fip_firmware() allows) or if the caller passes a zero-sized buffer. Please can you return early when size is 0?

Regards,
Simon

#9 lore raw

Re: [PATCH v8 5/7] misc: fw_loader: introduce FIP loader driver

Quoted 64 lines · Simon Glass

Hi Christian,

Quoted 30 lines · Christian Marangi (Ansuel)

misc: fw_loader: introduce FIP loader driver

Introduce a variant of the FS loader driver to extract images from FIP image. These image can contain additional binary used to init Network accelerator or PHY firmware blob.

The way FIP handle image type is with the usage of UUID.

This FIP loader driver implement a simple FIP image parser that check every entry for a matching UUID.

Similar to FS loader, this driver also support both UBI and Block devices.

Also an additional property is added to handle special case with eMMC that doesn't have a GPT partition and require a global offset to reference the FIP partition.

An example usage of this driver is the following:

[...]

 drivers/misc/Kconfig                |  23 ++
 drivers/misc/fw_loader/Makefile     |   1 +
 drivers/misc/fw_loader/fip_loader.c | 556 ++++++++++++++++++++++++++++++++++++
 drivers/misc/fw_loader/internal.h   |   2 +
 4 files changed, 582 insertions(+)

drivers/misc/fw_loader/fip_loader.c line 7

Quoted 7 lines
drivers/misc/fw_loader/fip_loader.c +7 -0
@@ -0,0 +1,556 @@
+static int blk_read_fip_firmware(struct firmware *firmwarep,
+				 struct blk_desc *desc,
+				 struct disk_partition *part_info,
+				 unsigned int part_offset,
+				 const struct fip_toc_entry *ent)
+{
+	unsigned int offset = part_info->start + part_offset;

The parse path now uses lbaint_t, but this read path still truncates part_info->start to unsigned int, so on a large device the header is found but the data is read from the wrong block. Please can you use lbaint_t here too (and for blkcnt/blkstart, to match blk_parse_fip_firmware())?

drivers/misc/fw_loader/fip_loader.c line 9

Quoted 11 lines
drivers/misc/fw_loader/fip_loader.c +9 -0
@@ -0,0 +1,556 @@
+	size = min(firmwarep->size, (size_t)ent->size - offset);
+	ret = ubi_volume_read(ubi_vol,
+			      (u8 *)firmwarep->data,
+			      ent->offset_address + offset,
+			      size);
+	if (ret)
+		return ret;
+
+	return size - firmwarep->offset;

'size' already has the offset taken off, so this subtracts it twice. With a non-zero offset the UBI path returns less than it actually read, whereas the block path returns the full 'read' count. I think this should just be 'return size;'

Quoted 5 lines

Also, ubi_volume_read() treats size == 0 as 'read the whole volume', which overruns the buffer. That happens if offset == ent->size (which the check in fw_get_fip_firmware() allows) or if the caller passes a zero-sized buffer. Please can you return early when size is 0?

!!! No idea ubi_volume_read() 0 is read all. Thanks a lot for noticing this.

Signature · 1 line
	Ansuel
#10 lore raw

PATCH v8 6/7 doc: dtbinding: Update documentation for Generic Firmware loader

In reply to Christian Marangi

Update documentation for Generic Firmware loader to YAML schema, generalize it from FS specific and add new property and example for FIP loader.

Also add details on the usage of 'phandlepart', 'mtdpart' and 'ubivol' property.

---
 doc/device-tree-bindings/misc/fs_loader.txt  |  48 -------
 doc/device-tree-bindings/misc/fw_loader.yaml | 126 +++++++++++++++++++
 2 files changed, 126 insertions(+), 48 deletions(-)
 delete mode 100644 doc/device-tree-bindings/misc/fs_loader.txt
 create mode 100644 doc/device-tree-bindings/misc/fw_loader.yaml
doc/device-tree-bindings/misc/fs_loader.txt deleted +0 -48
deleted file mode 100644
index 542be4b25a0a..000000000000
@@ -1,48 +0,0 @@
-* File system firmware loader
-
-Required properties:
---------------------
-
-- compatible: should contain "u-boot,fs-loader"
-- phandlepart: which block storage device and partition the image loading from,
-	       this property is required for mmc, usb and sata. This is unsigned
-	       32-bit array. For example phandlepart=<&mmc_0 1>, meaning use
-	       that MMC0 node pointer, partition 1.
-- mdtpart: which partition of ubi the image loading from, this property is
-	   required for ubi and mounting.
-- ubivol: which volume of ubi the image loading from, this property is required
-	  for ubi and mounting.
-
-Example of storage device and partition search set for mmc, usb, sata and
-ubi in device tree source as shown in below:
-
-	Example of storage type and device partition search set for mmc, usb,
-	sata and ubi as shown in below:
-	Example for mmc:
-	fs_loader0: fs-loader@0 {
-		bootph-all;
-		compatible = "u-boot,fs-loader";
-		phandlepart = <&mmc_0 1>;
-	};
-
-	Example for usb:
-	fs_loader1: fs-loader@1 {
-		bootph-all;
-		compatible = "u-boot,fs-loader";
-		phandlepart = <&usb0 1>;
-	};
-
-	Example for sata:
-	fs_loader2: fs-loader@2 {
-		bootph-all;
-		compatible = "u-boot,fs-loader";
-		phandlepart = <&sata0 1>;
-	};
-
-	Example for ubi:
-	fs_loader3: fs-loader@3 {
-		bootph-all;
-		compatible = "u-boot,fs-loader";
-		mtdpart = "UBI",
-		ubivol = "ubi0";
-	};
doc/device-tree-bindings/misc/fw_loader.yaml new file +126 -0

126 lines not on this page · raw message

Signature · 1 line
2.55.0
#11 lore raw

Re: [PATCH v8 6/7] doc: dtbinding: Update documentation for Generic Firmware loader

Hi Christian,

Quoted 17 lines · Christian Marangi (Ansuel)

doc: dtbinding: Update documentation for Generic Firmware loader

Update documentation for Generic Firmware loader to YAML schema, generalize it from FS specific and add new property and example for FIP loader.

Also add details on the usage of 'phandlepart', 'mtdpart' and 'ubivol' property.

 doc/device-tree-bindings/misc/fs_loader.txt  |  48 ----------
 doc/device-tree-bindings/misc/fw_loader.yaml | 126 +++++++++++++++++++++++++++
 2 files changed, 126 insertions(+), 48 deletions(-)

doc/device-tree-bindings/misc/fw_loader.yaml line 5

Quoted 5 lines
doc/device-tree-bindings/misc/fw_loader.yaml +5 -0
@@ -0,0 +1,126 @@
+  mtdpart:
+    $ref: /schemas/types.yaml#/definitions/string
+    description: |
+      Specifies the MTD partition from which the image is loaded.
+      Required for MTD and UBI devices.

Both loaders only look at mtdpart when ubivol is also set (plat->mtdpart && plat->ubivol in fs_loader.c and fip_loader.c), so there is no MTD-only path. Please can you say 'Required for UBI devices' here, and make ubivol required alongside mtdpart in the schema? The 'MTD/UBI storage' wording in the top-level description has the same problem.

Quoted 29 lines
doc/device-tree-bindings/misc/fw_loader.yaml +27 -0
@@ -0,0 +1,126 @@
+oneOf:
+  - required:
+      - phandlepart
+
+    properties:
+      mtdpart: false
+
+  - required:
+      - mtdpart
+
+    properties:
+      phandlepart: false
+
+allOf:
+  - if:
+      properties:
+        compatible:
+          const: u-boot,fs-loader
+    then:
+      properties:
+        partoffset: false
+
+  - if:
+      required:
+        - phandlepart
+    then:
+      properties:

doc/device-tree-bindings/misc/fw_loader.yaml line 37

Quoted 10 lines
doc/device-tree-bindings/misc/fw_loader.yaml +10 -0
+        mtdpart: false
+
+        ubivol: false
+
+  - if:
+      required:
+        - mtdpart
+    then:
+      properties:
+        phandlepart: false

The second and third allOf entries repeat the oneOf. The third adds nothing and the second only adds 'ubivol: false'. I suggest putting it all in the oneOf, along these lines:

oneOf:
- required:
- phandlepart
properties:
mtdpart: false
ubivol: false
- required:
- mtdpart
- ubivol
properties:
phandlepart: false

and keeping only the partoffset rule in the allOf list. What do you think?

Regards,
Simon

#12 lore raw

PATCH v8 7/7 test: dm: Add test for FW Loader feature

In reply to Christian Marangi

Add basic test for FW Loader feature for the Filesystem way and FIP.

Additional MMC testing image are added that loads a dummy firmware.bin in a fat partition or a simple FIP blob.

---
 arch/sandbox/dts/test.dts          |  38 +++++
 configs/sandbox64_defconfig        |   2 +
 configs/sandbox_defconfig          |   2 +
 configs/sandbox_flattree_defconfig |   2 +
 configs/sandbox_noinst_defconfig   |   1 +
 configs/sandbox_spl_defconfig      |   2 +
 configs/sandbox_vpl_defconfig      |   2 +
 test/dm/Makefile                   |   1 +
 test/dm/fw_loader.c                | 218 +++++++++++++++++++++++++++++
 test/py/tests/fip/test.fip         | Bin 0 -> 101 bytes
 test/py/tests/test_ut.py           |  26 ++++
 11 files changed, 294 insertions(+)
 create mode 100644 test/dm/fw_loader.c
 create mode 100644 test/py/tests/fip/test.fip
arch/sandbox/dts/test.dts +38 -0
index 25c99ef81472..772d8dd548bc 100644
@@ -189,6 +189,18 @@
 		};
 	};
 
+	fs_loader: fs-loader {
+		bootph-all;
+		compatible = "u-boot,fs-loader";
+		phandlepart = <&mmc12 1>;
+	};
+
+	fip_loader: fip-loader {
+		bootph-all;
+		compatible = "u-boot,fip-loader";
+		phandlepart = <&mmc13 1>;
+	};
+
 	cedit: cedit {
 	};
 
@@ -1398,6 +1410,20 @@
 		filename = "mmc11.img";
 	};
 
+	/* This is used for FS Loader tests */
+	mmc12: mmc12 {
+		status = "disabled";
+		compatible = "sandbox,mmc";
+		filename = "mmc12.img";
+	};
+
+	/* This is used for FIP Loader tests */
+	mmc13: mmc13 {
+		status = "disabled";
+		compatible = "sandbox,mmc";
+		filename = "mmc13.img";
+	};
+
 	pch {
 		compatible = "sandbox,pch";
 	};
@@ -2143,6 +2169,18 @@
 		ethphy1: ethernet-phy@1 {
 			reg = <1>;
 		};
+
+		ethernet-phy@2 {
+			reg = <2>;
+
+			firmware-loader = <&fs_loader>;
+		};
+
+		ethernet-phy@3 {
+			reg = <3>;
+
+			firmware-loader = <&fip_loader>;
+		};
 	};
 
 	pm-bus-test {
configs/sandbox64_defconfig +2 -0
index 59e08d03f776..7d56ab984c1d 100644
@@ -266,9 +266,11 @@ CONFIG_WDT=y
 CONFIG_WDT_GPIO=y
 CONFIG_WDT_SANDBOX=y
 CONFIG_WDT_ALARM_SANDBOX=y
+CONFIG_FIP_LOADER=y
 CONFIG_FS_CBFS=y
 CONFIG_FS_EXFAT=y
 CONFIG_FS_CRAMFS=y
+CONFIG_FS_LOADER=y
 CONFIG_CMD_DHRYSTONE=y
 CONFIG_TPM=y
 CONFIG_ERRNO_STR=y
configs/sandbox_defconfig +2 -0
index f307972a762d..cf3c858b4b72 100644
@@ -383,9 +383,11 @@ CONFIG_WDT_GPIO=y
 CONFIG_WDT_SANDBOX=y
 CONFIG_WDT_ALARM_SANDBOX=y
 CONFIG_WDT_FTWDT010=y
+CONFIG_FIP_LOADER=y
 CONFIG_FS_CBFS=y
 CONFIG_FS_EXFAT=y
 CONFIG_FS_CRAMFS=y
+CONFIG_FS_LOADER=y
 CONFIG_ADDR_MAP=y
 CONFIG_PANIC_HANG=y
 CONFIG_PMBUS=y
configs/sandbox_flattree_defconfig +2 -0
index 57f915d4ba9a..3488036ddf8f 100644
@@ -12,10 +12,12 @@ CONFIG_EFI_CAPSULE_FIRMWARE_FIT=y
 CONFIG_EFI_CAPSULE_AUTHENTICATE=y
 CONFIG_EFI_CAPSULE_CRT_FILE="board/sandbox/capsule_pub_key_good.crt"
 CONFIG_BUTTON_CMD=y
+CONFIG_FIP_LOADER=y
 CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERITY=y
 CONFIG_FIT_VERBOSE=y
+CONFIG_FS_LOADER=y
 CONFIG_LEGACY_IMAGE_FORMAT=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
configs/sandbox_noinst_defconfig +1 -0
index 7a7550127956..941bf2d69431 100644
@@ -21,6 +21,7 @@ CONFIG_SYS_MEMTEST_END=0x00101000
 CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERBOSE=y
+CONFIG_FS_LOADER=y
 CONFIG_SPL_LOAD_FIT=y
 CONFIG_SPL_HAS_LOAD_FIT_ADDRESS=y
 CONFIG_SPL_LOAD_FIT_ADDRESS=0x0
configs/sandbox_spl_defconfig +2 -0
index 60b497f48886..183611632ef8 100644
@@ -223,8 +223,10 @@ CONFIG_I2C_EDID=y
 CONFIG_VIDEO_SANDBOX_SDL=y
 CONFIG_OSD=y
 CONFIG_SANDBOX_OSD=y
+CONFIG_FIP_LOADER=y
 CONFIG_FS_CBFS=y
 CONFIG_FS_CRAMFS=y
+CONFIG_FS_LOADER=y
 # CONFIG_SPL_USE_TINY_PRINTF is not set
 CONFIG_CMD_DHRYSTONE=y
 CONFIG_RSA_VERIFY_WITH_PKEY=y
configs/sandbox_vpl_defconfig +2 -0
index c85bfbfb3b15..445e979915e8 100644
@@ -231,8 +231,10 @@ CONFIG_CONSOLE_TRUETYPE_CANTORAONE=y
 CONFIG_VIDEO_SANDBOX_SDL=y
 CONFIG_OSD=y
 CONFIG_SANDBOX_OSD=y
+CONFIG_FIP_LOADER=y
 CONFIG_FS_CBFS=y
 CONFIG_FS_CRAMFS=y
+CONFIG_FS_LOADER=y
 # CONFIG_SPL_USE_TINY_PRINTF is not set
 CONFIG_CMD_DHRYSTONE=y
 CONFIG_RSA_VERIFY_WITH_PKEY=y
test/dm/Makefile +1 -0
index 5dd6b57bbe84..2f095ceedd8c 100644
@@ -59,6 +59,7 @@ obj-$(CONFIG_EXTCON) += extcon.o
 ifneq ($(CONFIG_EFI_PARTITION),)
 obj-$(CONFIG_FASTBOOT_FLASH_MMC) += fastboot.o
 endif
+obj-$(CONFIG_FW_LOADER_UCLASS) += fw_loader.o
 obj-$(CONFIG_FIRMWARE) += firmware.o
 obj-$(CONFIG_DM_FPGA) += fpga.o
 obj-$(CONFIG_FWU_MDATA_GPT_BLK) += fwu_mdata.o
test/dm/fw_loader.c new file +218 -0

218 lines not on this page · raw message

test/py/tests/fip/test.fip new file binary +0 -0
new file mode 100644
index 0000000000000000000000000000000000000000..a793dfd05c016b2bfe693a0b908877689abe8e5b
GIT binary patch
test/py/tests/test_ut.py +26 -0
index 2a5acc18c612..f3da9f3ef005 100644
@@ -685,6 +685,31 @@ def test_ut_dm_init(ubman):
     with open(fn, 'wb') as fh:
         fh.write(data)
 
+def setup_fw_loader_image(ubman):
+    """Create a 2MB disk image with a dummy firmware on it"""
+    mmc_dev = 12
+    fname, mnt = fs_helper.setup_image(ubman, mmc_dev, 0xc, img_size=2)
+
+    fw = os.path.join(mnt, 'firmware.bin')
+    data = b'\xaa' * 256
+    with open(fw, 'wb') as fh:
+        fh.write(data)
+
+    fsfile = 'vfat1M.img'
+    utils.run_and_log(ubman, f'fallocate -l 1M {fsfile}')
+    utils.run_and_log(ubman, f'mkfs.vfat {fsfile}')
+    utils.run_and_log(ubman, ['sh', '-c', f'mcopy -i {fsfile} {mnt}/* ::/'])
+    copy_partition(ubman, fsfile, fname)
+    utils.run_and_log(ubman, f'rm -rf {mnt}')
+    utils.run_and_log(ubman, f'rm -f {fsfile}')
+
+    mmc_dev = 13
+    fname, mnt = fs_helper.setup_image(ubman, mmc_dev, 0x83, img_size=2)
+
+    fipfile = os.path.join(ubman.config.source_dir,
+                            'test/py/tests/fip/test.fip')
+    copy_partition(ubman, fipfile, fname)
+    utils.run_and_log(ubman, f'rm -rf {mnt}')
 
 def setup_efi_image(ubman):
     """Create a 20MB disk image with an EFI app on it"""
@@ -752,6 +777,7 @@ def test_ut_dm_init_bootstd(ubman):
     setup_cedit_file(ubman)
     setup_cros_image(ubman)
     setup_android_image(ubman)
+    setup_fw_loader_image(ubman)
     setup_efi_image(ubman)
     setup_rauc_image(ubman)
     setup_bls_prefix_image(ubman)
Signature · 1 line
2.55.0
#13 lore raw

Re: [PATCH v8 7/7] test: dm: Add test for FW Loader feature

Hi Christian,

Quoted 24 lines · Christian Marangi (Ansuel)

test: dm: Add test for FW Loader feature

Add basic test for FW Loader feature for the Filesystem way and FIP.

Additional MMC testing image are added that loads a dummy firmware.bin in a fat partition or a simple FIP blob.

 arch/sandbox/dts/test.dts          |  38 +++++++
 configs/sandbox64_defconfig        |   2 +
 configs/sandbox_defconfig          |   2 +
 configs/sandbox_flattree_defconfig |   2 +
 configs/sandbox_noinst_defconfig   |   1 +
 configs/sandbox_spl_defconfig      |   2 +
 configs/sandbox_vpl_defconfig      |   2 +
 test/dm/Makefile                   |   1 +
 test/dm/fw_loader.c                | 218 +++++++++++++++++++++++++++++++++++++
 test/py/tests/fip/test.fip         | Bin 0 -> 101 bytes
 test/py/tests/test_ut.py           |  26 +++++
 11 files changed, 294 insertions(+)

test/dm/fw_loader.c line 3

Quoted 3 lines
test/dm/fw_loader.c +3 -0
@@ -0,0 +1,218 @@
+	ret = request_firmware_into_buf(fw_loader_dev, "firmware.bin",
+					buf, firmware_size, firmware_size + 1);
+	ut_asserteq(0, ret);

These tests show that the two loaders behave differently when the offset is past the end: FS returns 0 here, but FIP returns -EINVAL for the same request (see below). Callers go through the uclass and should not need to know which loader is behind the phandle, so please can you pick one behaviour for both drivers? I think -EINVAL is better, and the cover letter says FIP is meant to match FS anyway.

test/dm/fw_loader.c line 4

Quoted 6 lines
test/dm/fw_loader.c +4 -0
@@ -0,0 +1,218 @@
+	ret = request_firmware_into_buf(fw_loader_dev,
+					"45a113ac-c7f2-30b0-90a5-a399ab912716",
+					buf, firmware_size, firmware_size + 1);
+	ut_asserteq(-EINVAL, ret);

This is the FIP case mentioned above.

test/dm/fw_loader.c line 2

Quoted 4 lines
test/dm/fw_loader.c +2 -0
@@ -0,0 +1,218 @@
+	/* Enable the mmc12 node for this test */
+	root = oftree_root(oftree_default());

This helper is used for mmc13 too, so the comment should not name a specific node.

test/py/tests/test_ut.py def test_ut_dm_init(ubman): · line 686

Quoted 4 lines
test/py/tests/test_ut.py +2 -0
@@ -685,6 +685,31 @@ def test_ut_dm_init(ubman):
+    fipfile = os.path.join(ubman.config.source_dir,
+                            'test/py/tests/fip/test.fip')

Please can you explain in the commit message how test.fip was created (e.g. the fiptool command line), so the next person can update it? Better still, since the format is just a header plus one ToC entry, you could build the 101 bytes in Python here and avoid a binary blob in the tree. What do you think?

Binman has a atf_fip.py but you can't use that here, unfortunately.

Regards,
Simon

#14 lore raw

Re: [PATCH v8 7/7] test: dm: Add test for FW Loader feature

Quoted 25 lines · Simon Glass

Hi Christian,

Quoted 21 lines · Christian Marangi (Ansuel)

test: dm: Add test for FW Loader feature

Add basic test for FW Loader feature for the Filesystem way and FIP.

Additional MMC testing image are added that loads a dummy firmware.bin in a fat partition or a simple FIP blob.

 arch/sandbox/dts/test.dts          |  38 +++++++
 configs/sandbox64_defconfig        |   2 +
 configs/sandbox_defconfig          |   2 +
 configs/sandbox_flattree_defconfig |   2 +
 configs/sandbox_noinst_defconfig   |   1 +
 configs/sandbox_spl_defconfig      |   2 +
 configs/sandbox_vpl_defconfig      |   2 +
 test/dm/Makefile                   |   1 +
 test/dm/fw_loader.c                | 218 +++++++++++++++++++++++++++++++++++++
 test/py/tests/fip/test.fip         | Bin 0 -> 101 bytes
 test/py/tests/test_ut.py           |  26 +++++
 11 files changed, 294 insertions(+)
Quoted 13 lines

test/dm/fw_loader.c line 3

Quoted 5 lines
test/dm/fw_loader.c +3 -0
@@ -0,0 +1,218 @@
+	ret = request_firmware_into_buf(fw_loader_dev, "firmware.bin",
+					buf, firmware_size, firmware_size + 1);
+	ut_asserteq(0, ret);

These tests show that the two loaders behave differently when the offset is past the end: FS returns 0 here, but FIP returns -EINVAL for the same request (see below). Callers go through the uclass and should not need to know which loader is behind the phandle, so please can you pick one behaviour for both drivers? I think -EINVAL is better, and the cover letter says FIP is meant to match FS anyway.

Yes exactly... the problem is that I don't want to cause regression in the use of FS loader with this change of error handling.

Ok I will change FS loader to return -EINVAL as a separate commit.

Quoted 17 lines

test/dm/fw_loader.c line 4

Quoted 6 lines
test/dm/fw_loader.c +4 -0
@@ -0,0 +1,218 @@
+	ret = request_firmware_into_buf(fw_loader_dev,
+					"45a113ac-c7f2-30b0-90a5-a399ab912716",
+					buf, firmware_size, firmware_size + 1);
+	ut_asserteq(-EINVAL, ret);

This is the FIP case mentioned above.

test/dm/fw_loader.c line 2

Quoted 4 lines
test/dm/fw_loader.c +2 -0
@@ -0,0 +1,218 @@
+	/* Enable the mmc12 node for this test */
+	root = oftree_root(oftree_default());

This helper is used for mmc13 too, so the comment should not name a specific node.

Quoted 10 lines

test/py/tests/test_ut.py def test_ut_dm_init(ubman): · line 686

Quoted 4 lines
test/py/tests/test_ut.py +2 -0
@@ -685,6 +685,31 @@ def test_ut_dm_init(ubman):
+    fipfile = os.path.join(ubman.config.source_dir,
+                            'test/py/tests/fip/test.fip')

Please can you explain in the commit message how test.fip was created (e.g. the fiptool command line), so the next person can update it? Better still, since the format is just a header plus one ToC entry, you could build the 101 bytes in Python here and avoid a binary blob in the tree. What do you think?

Ehhhhh I checked and it's a lot of effort on creating a python shim to simulate the format of a FIP blob. I will check it again but providing the 3kb blob was more practical than polluting the python script with something used for just one task.

Quoted 5 lines

Binman has a atf_fip.py but you can't use that here, unfortunately.

Regards,
Simon

Signature · 1 line
	Ansuel
#15 lore raw
In reply to Christian Marangi

Hi Christian,

Quoted 2 lines · Christian Marangi (Ansuel)

Changes v8:
- Address all the comments from Simon

Just to check, two series-level points from v7 still seem to be open. The binding update (patch 6) still comes after the FIP loader (patch 5), so u-boot,fip-loader and partoffset are used before they are documented. Please can you move the binding before the driver, or fold it into the driver patch?

Quoted 1 line

fs_loader0: fip-loader {

The 'fs_loader0' label is still used for the fip-loader node, here and in the commit message of patch 5, and the binding example in patch 6 has 'fs_loader4' and 'fs_loader5' for its fip-loader nodes. Now that FS and FW are separate, please can you rename these to something like 'fw_loader0'?

Regards,
Simon