[PATCH v3 00/32] Tidy up use of CONFIG_CMDLINE

[PATCH v3 00/32] Tidy up use of CONFIG_CMDLINE

[Simon Glass]

It should be possible to disable CONFIG_CMDLINE and have all commands
and related functionality dropped from U-Boot. This is useful when
trying to reduce the size of U-Boot.

Recent changes have stopped this from working.

This series repairs the feature for sandbox and adds a test to stop it
breaking again.

Note that quite a lot of functionality is lost if CONFIG_CMDLINE is
disabled, e.g. networking and most booting options. Further work is
needed to make the option more generally useful.

Changes in v3:
- Reword commit message slightly
- Add an explation as to why this patch is here
- Fix 'ut' typo
- Just drop the condition instead
- Add new patch to rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR
- Add new patch with a Kconfig for command-line entry
- Add new patch with a Kconfig for command-line entry
- Reorder the Kconfig options a little
- Add new patch to tidy up semicolon after command macros
- Rebase on Tom's LONGHELP series
- Correct 'of' typo

Changes in v2:
- Move AUTOBOOT_USE_MENUKEY under AUTOBOOT
- Change this to use a Kconfig dependency instead of code failing
- Adjust the depends on the commands, instead of EFI_LOADER
- Add new patch to update EFI_LOADER to depend on DM_ETH
- Add new patch to drop discarding of command linker-lists
- Add new patch to unify the U_BOOT_ENV_LOCATION conditions

Simon Glass (32):
  buildman: Use oldconfig when adjusting the config
  bootstd: Correct dependencies on CMDLINE
  autoboot: Correct dependencies on CMDLINE
  cmd: Add a few more dependencies on CMDLINE
  test: Make UNIT_TEST depend on CMDLINE
  sifive: Drop an unnecessary #ifdef
  fastboot: Declare a dependency on CMDLINE
  cli: Always build cli_getch
  cmd: Use an #ifdef around run_commandf()
  Move bootmenu_conv_key() into its own file
  pxe: Depend on CMDLINE
  env: Split out non-command code into a new file
  console: Move SYS_PBSIZE into common/
  bootm: Allow building when cleanup functions are missing
  fdt: Move working_fdt into fdt_support
  net: Depend on CONFIG_CMDLINE
  log: Allow use without CONFIG_CMDLINE
  video: Allow use without CONFIG_CMDLINE
  video: Dont require the font command
  efi: Make commands depend on CMDLINE
  efi: Rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR
  efi: Update EFI_LOADER to depend on DM_ETH
  cli: Split command-line-editing into its own file
  Add a new Kconfig for command-line entry
  Add a new Kconfig for command-line history
  sandbox: Disable CONFIG_DISTRO_DEFAULTS
  cmd: Make all commands depend on CMDLINE
  sandbox: Avoid requiring cmdline
  arm: x86: Drop discarding of command linker-lists
  mmc: env: Unify the U_BOOT_ENV_LOCATION conditions
  treewide: Tidy up semicolon after command macros
  sandbox: Add a test for disabling CONFIG_CMDLINE

 arch/Kconfig                       |  11 +-
 arch/arm/cpu/u-boot.lds            |   3 -
 arch/arm/lib/bootm.c               |   2 +
 arch/x86/cpu/u-boot-64.lds         |   4 -
 arch/x86/cpu/u-boot-spl.lds        |   4 -
 arch/x86/cpu/u-boot.lds            |   4 -
 board/freescale/common/vid.c       |   2 +-
 board/xilinx/common/fru.c          |   2 +-
 board/xilinx/versal/cmds.c         |   2 +-
 board/xilinx/zynqmp/cmds.c         |   2 +-
 boot/Kconfig                       |  58 ++--
 boot/bootm.c                       |  10 +-
 boot/fdt_support.c                 |   5 +
 cmd/Kconfig                        |  50 ++--
 cmd/Makefile                       |   2 +-
 cmd/btrfs.c                        |   2 +-
 cmd/eeprom.c                       |   2 +-
 cmd/ext2.c                         |   4 +-
 cmd/fdt.c                          |   5 -
 cmd/fs.c                           |   8 +-
 cmd/nvedit.c                       | 122 +--------
 cmd/pinmux.c                       |   2 +-
 cmd/qfw.c                          |   2 +-
 common/Kconfig                     |  12 +
 common/Makefile                    |  10 +-
 common/cli.c                       |   2 +
 common/cli_cread.c                 | 407 +++++++++++++++++++++++++++++
 common/cli_getch.c                 |   1 +
 common/cli_readline.c              | 384 +--------------------------
 common/log.c                       |   4 +-
 common/menu.c                      |  40 ---
 common/menu_key.c                  |  49 ++++
 configs/sandbox64_defconfig        |   1 -
 configs/sandbox_defconfig          |   1 -
 configs/sandbox_flattree_defconfig |   1 -
 configs/sandbox_noinst_defconfig   |   1 -
 configs/sandbox_spl_defconfig      |   1 -
 configs/sandbox_vpl_defconfig      |   1 -
 drivers/fastboot/Kconfig           |   1 +
 drivers/video/Kconfig              |   2 +-
 drivers/video/console_truetype.c   |   4 +
 env/Makefile                       |   1 +
 env/env_set.c                      | 132 ++++++++++
 env/mmc.c                          |   2 +-
 include/bootm.h                    |  15 +-
 include/cli.h                      |  31 +++
 include/command.h                  |   2 +-
 include/env_internal.h             |  23 ++
 include/k210/pll.h                 |   2 +-
 lib/efi_loader/Kconfig             |   7 +-
 lib/efi_loader/Makefile            |   2 +-
 net/Kconfig                        |   1 +
 test/Kconfig                       |   1 +
 test/py/tests/test_sandbox_opts.py |  20 ++
 tools/buildman/builder.py          |   2 +-
 tools/buildman/builderthread.py    |   6 +
 tools/buildman/func_test.py        |   4 +-
 57 files changed, 833 insertions(+), 648 deletions(-)
 create mode 100644 common/cli_cread.c
 create mode 100644 common/menu_key.c
 create mode 100644 env/env_set.c
 create mode 100644 test/py/tests/test_sandbox_opts.py

-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 01/32] buildman: Use oldconfig when adjusting the config

[Simon Glass]

We cannot be sure that the new config is consistent, particularly when
changing a major item like CONFIG_CMDLINE. Use 'make oldconfig' to
check that and avoid any such problems.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 tools/buildman/builder.py       | 2 +-
 tools/buildman/builderthread.py | 6 ++++++
 tools/buildman/func_test.py     | 4 +++-
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/tools/buildman/builder.py b/tools/buildman/builder.py
index 5305477c5be6..782e59dd5cca 100644
--- a/tools/buildman/builder.py
+++ b/tools/buildman/builder.py
@@ -476,7 +476,7 @@ class Builder:
         Args:
             commit: Commit object that is being built
             brd: Board object that is being built
-            stage: Stage that we are at (mrproper, config, build)
+            stage: Stage that we are at (mrproper, config, oldconfig, build)
             cwd: Directory where make should be run
             args: Arguments to pass to make
             kwargs: Arguments to pass to command.run_pipe()
diff --git a/tools/buildman/builderthread.py b/tools/buildman/builderthread.py
index 6a61f64da1d4..a8599c0bb2a8 100644
--- a/tools/buildman/builderthread.py
+++ b/tools/buildman/builderthread.py
@@ -426,6 +426,12 @@ class BuilderThread(threading.Thread):
 
         # Now do the build, if everything looks OK
         if result.return_code == 0:
+            if adjust_cfg:
+                oldc_args = list(args) + ['oldconfig']
+                oldc_result = self.make(commit, brd, 'oldconfig', cwd,
+                                        *oldc_args, env=env)
+                if oldc_result.return_code:
+                    return oldc_result
             result = self._build(commit, brd, cwd, args, env, cmd_list,
                                  config_only)
             if adjust_cfg:
diff --git a/tools/buildman/func_test.py b/tools/buildman/func_test.py
index 55dd494fe8ee..6b88ed815d65 100644
--- a/tools/buildman/func_test.py
+++ b/tools/buildman/func_test.py
@@ -439,6 +439,8 @@ class TestFunctional(unittest.TestCase):
             tools.write_file(fname, b'CONFIG_SOMETHING=1')
             return command.CommandResult(return_code=0,
                     combined='Test configuration complete')
+        elif stage == 'oldconfig':
+            return command.CommandResult(return_code=0)
         elif stage == 'build':
             stderr = ''
             fname = os.path.join(cwd or '', out_dir, 'u-boot')
@@ -461,7 +463,7 @@ Some images are invalid'''
             return command.CommandResult(return_code=0)
 
         # Not handled, so abort
-        print('make', stage)
+        print('_HandleMake failure: make', stage)
         sys.exit(1)
 
     # Example function to print output lines
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Simon Glass]

With recent changes over the last few years in boot/Kconfig it is
no-longer possible to disable CMDLINE. It results in various link
errors because some options which require CMDLINE are enabled
regardless of whether it is available.

Add the necessary conditions to fix this.

Note that it would be better to have all commands depend on CMDLINE,
but that is extremely difficult at present, since some functions use
CMD_xxx to enable feature xxx. For example networking and environment
have a number of problems to tease apart.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Reword commit message slightly

 boot/Kconfig           | 19 ++++++++++++-------
 lib/efi_loader/Kconfig |  1 +
 2 files changed, 13 insertions(+), 7 deletions(-)

diff --git a/boot/Kconfig b/boot/Kconfig
index a01e6cb8aafe..f74ac7e9cc72 100644
--- a/boot/Kconfig
+++ b/boot/Kconfig
@@ -342,6 +342,7 @@ endif # FIT
 
 config PXE_UTILS
 	bool
+	depends on CMDLINE
 	select MENU
 	help
 	  Utilities for parsing PXE file formats.
@@ -357,7 +358,7 @@ config BOOT_DEFAULTS
 	select CMD_PART if PARTITIONS
 	select CMD_DHCP if CMD_NET
 	select CMD_PING if CMD_NET
-	select CMD_PXE if CMD_NET
+	select CMD_PXE if CMDLINE && CMD_NET
 	select SUPPORT_RAW_INITRD
 	select ENV_VARS_UBOOT_CONFIG
 	select CMD_BOOTI if ARM64
@@ -461,6 +462,7 @@ config BOOTMETH_GLOBAL
 
 config BOOTMETH_CROS
 	bool "Bootdev support for Chromium OS"
+	depends on CMDLINE
 	depends on X86 || ARM || SANDBOX
 	default y if !ARM
 	select EFI_PARTITION
@@ -475,6 +477,7 @@ config BOOTMETH_CROS
 
 config BOOTMETH_EXTLINUX
 	bool "Bootdev support for extlinux boot"
+	depends on CMDLINE
 	select PXE_UTILS
 	default y
 	help
@@ -490,7 +493,7 @@ config BOOTMETH_EXTLINUX
 
 config BOOTMETH_EXTLINUX_PXE
 	bool "Bootdev support for extlinux boot over network"
-	depends on CMD_PXE && CMD_NET && DM_ETH
+	depends on CMDLINE && CMD_PXE && CMD_NET && DM_ETH
 	default y
 	help
 	  Enables support for extlinux boot using bootdevs. This makes the
@@ -504,7 +507,7 @@ config BOOTMETH_EXTLINUX_PXE
 
 config BOOTMETH_EFILOADER
 	bool "Bootdev support for EFI boot"
-	depends on EFI_LOADER
+	depends on EFI_LOADER && CMDLINE
 	default y
 	help
 	  Enables support for EFI boot using bootdevs. This makes the
@@ -536,10 +539,10 @@ config BOOTMETH_VBE
 
 config BOOTMETH_DISTRO
 	bool  # Options needed to boot any distro
-	select BOOTMETH_SCRIPT  # E.g. Armbian uses scripts
-	select BOOTMETH_EXTLINUX  # E.g. Debian uses these
-	select BOOTMETH_EXTLINUX_PXE if CMD_PXE && CMD_NET && DM_ETH
-	select BOOTMETH_EFILOADER if EFI_LOADER # E.g. Ubuntu uses this
+	select BOOTMETH_SCRIPT if CMDLINE # E.g. Armbian uses scripts
+	select BOOTMETH_EXTLINUX if CMDLINE # E.g. Debian uses these
+	select BOOTMETH_EXTLINUX_PXE if CMDLINE && CMD_PXE && CMD_NET && DM_ETH
+	select BOOTMETH_EFILOADER if CMDLINE && EFI_LOADER # E.g. Ubuntu uses this
 
 config SPL_BOOTMETH_VBE
 	bool "Bootdev support for Verified Boot for Embedded (SPL)"
@@ -663,6 +666,7 @@ config BOOTMETH_SANDBOX
 
 config BOOTMETH_SCRIPT
 	bool "Bootdev support for U-Boot scripts"
+	depends on CMDLINE
 	default y if BOOTSTD_FULL
 	select HUSH_PARSER
 	help
@@ -777,6 +781,7 @@ endmenu		# Boot images
 
 config DISTRO_DEFAULTS
 	bool "(deprecated) Script-based booting of Linux distributions"
+	depends on CMDLINE
 	select BOOT_DEFAULTS
 	select AUTO_COMPLETE
 	select CMDLINE_EDITING
diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
index d20aaab6dba4..621ed5e5b0fb 100644
--- a/lib/efi_loader/Kconfig
+++ b/lib/efi_loader/Kconfig
@@ -34,6 +34,7 @@ if EFI_LOADER
 
 config CMD_BOOTEFI_BOOTMGR
 	bool "UEFI Boot Manager"
+	depends on CMDLINE
 	default y
 	select BOOTMETH_GLOBAL if BOOTSTD
 	help
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 03/32] autoboot: Correct dependencies on CMDLINE

[Simon Glass]

Make AUTOBOOT depend on CMDLINE since it is mostly meaningless without it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
---

(no changes since v2)

Changes in v2:
- Move AUTOBOOT_USE_MENUKEY under AUTOBOOT

 boot/Kconfig | 29 +++++++++++++++++++----------
 1 file changed, 19 insertions(+), 10 deletions(-)

diff --git a/boot/Kconfig b/boot/Kconfig
index f74ac7e9cc72..6461f7ebd040 100644
--- a/boot/Kconfig
+++ b/boot/Kconfig
@@ -1167,14 +1167,16 @@ menu "Autoboot options"
 
 config AUTOBOOT
 	bool "Autoboot"
+	depends on CMDLINE
 	default y
 	help
 	  This enables the autoboot.  See doc/README.autoboot for detail.
 
+if AUTOBOOT
+
 config BOOTDELAY
 	int "delay in seconds before automatically booting"
 	default 2
-	depends on AUTOBOOT
 	help
 	  Delay before automatically running bootcmd;
 	  set to 0 to autoboot with no delay, but you can stop it by key input.
@@ -1196,9 +1198,11 @@ config AUTOBOOT_KEYED
 	  U-Boot automatic booting process and bring the device
 	  to the U-Boot prompt for user input.
 
+if AUTOBOOT_KEYED
+
 config AUTOBOOT_FLUSH_STDIN
 	bool "Enable flushing stdin before starting to read the password"
-	depends on AUTOBOOT_KEYED && !SANDBOX
+	depends on !SANDBOX
 	help
 	  When this option is enabled stdin buffer will be flushed before
 	  starting to read the password.
@@ -1207,7 +1211,6 @@ config AUTOBOOT_FLUSH_STDIN
 
 config AUTOBOOT_PROMPT
 	string "Autoboot stop prompt"
-	depends on AUTOBOOT_KEYED
 	default "Autoboot in %d seconds\\n"
 	help
 	  This string is displayed before the boot delay selected by
@@ -1223,7 +1226,6 @@ config AUTOBOOT_PROMPT
 
 config AUTOBOOT_ENCRYPTION
 	bool "Enable encryption in autoboot stopping"
-	depends on AUTOBOOT_KEYED
 	help
 	  This option allows a string to be entered into U-Boot to stop the
 	  autoboot.
@@ -1250,7 +1252,7 @@ config AUTOBOOT_SHA256_FALLBACK
 
 config AUTOBOOT_DELAY_STR
 	string "Delay autobooting via specific input key / string"
-	depends on AUTOBOOT_KEYED && !AUTOBOOT_ENCRYPTION
+	depends on !AUTOBOOT_ENCRYPTION
 	help
 	  This option delays the automatic boot feature by issuing
 	  a specific input key or string. If CONFIG_AUTOBOOT_DELAY_STR
@@ -1262,7 +1264,7 @@ config AUTOBOOT_DELAY_STR
 
 config AUTOBOOT_STOP_STR
 	string "Stop autobooting via specific input key / string"
-	depends on AUTOBOOT_KEYED && !AUTOBOOT_ENCRYPTION
+	depends on !AUTOBOOT_ENCRYPTION
 	help
 	  This option enables stopping (aborting) of the automatic
 	  boot feature only by issuing a specific input key or
@@ -1274,7 +1276,7 @@ config AUTOBOOT_STOP_STR
 
 config AUTOBOOT_KEYED_CTRLC
 	bool "Enable Ctrl-C autoboot interruption"
-	depends on AUTOBOOT_KEYED && !AUTOBOOT_ENCRYPTION
+	depends on !AUTOBOOT_ENCRYPTION
 	help
 	  This option allows for the boot sequence to be interrupted
 	  by ctrl-c, in addition to the "bootdelaykey" and "bootstopkey".
@@ -1283,7 +1285,7 @@ config AUTOBOOT_KEYED_CTRLC
 
 config AUTOBOOT_NEVER_TIMEOUT
 	bool "Make the password entry never time-out"
-	depends on AUTOBOOT_KEYED && AUTOBOOT_ENCRYPTION && CRYPT_PW
+	depends on AUTOBOOT_ENCRYPTION && CRYPT_PW
 	help
 	  This option removes the timeout from the password entry
 	  when the user first presses the <Enter> key before entering
@@ -1291,7 +1293,7 @@ config AUTOBOOT_NEVER_TIMEOUT
 
 config AUTOBOOT_STOP_STR_ENABLE
 	bool "Enable fixed string to stop autobooting"
-	depends on AUTOBOOT_KEYED && AUTOBOOT_ENCRYPTION
+	depends on AUTOBOOT_ENCRYPTION
 	help
 	  This option enables the feature to add a fixed stop
 	  string that is defined at compile time.
@@ -1322,9 +1324,12 @@ config AUTOBOOT_STOP_STR_SHA256
 	  includes a ":", the portion prior to the ":" will be treated
 	  as a salt value.
 
+endif  # AUTOBOOT_KEYED
+
+if !AUTOBOOT_KEYED
+
 config AUTOBOOT_USE_MENUKEY
 	bool "Allow a specify key to run a menu from the environment"
-	depends on !AUTOBOOT_KEYED
 	help
 	  If a specific key is pressed to stop autoboot, then the commands in
 	  the environment variable 'menucmd' are executed before boot starts.
@@ -1339,6 +1344,10 @@ config AUTOBOOT_MENUKEY
 	  For example, 33 means "!" in ASCII, so pressing ! at boot would take
 	  this action.
 
+endif
+
+endif  # AUTOBOOT
+
 config AUTOBOOT_MENU_SHOW
 	bool "Show a menu on boot"
 	depends on CMD_BOOTMENU
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Simon Glass]

Add this to some more commands to avoid build errors with sandbox.

Note that this is a temporary solution to expose more problems. A later
patch puts these behind an 'if CMDLINE'

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Add an explation as to why this patch is here

 cmd/Kconfig | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 5bc0a92d57ad..00a7f84bce99 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -231,6 +231,7 @@ menu "Boot commands"
 
 config CMD_BOOTD
 	bool "bootd"
+	depends on CMDLINE
 	default y
 	help
 	  Run the command stored in the environment "bootcmd", i.e.
@@ -420,6 +421,7 @@ source lib/efi_selftest/Kconfig
 
 config CMD_BOOTMENU
 	bool "bootmenu"
+	depends on CMDLINE
 	select MENU
 	select CHARSET
 	help
@@ -486,6 +488,7 @@ config CMD_GO
 
 config CMD_RUN
 	bool "run"
+	depends on CMDLINE
 	default y
 	help
 	  Run the command in the given environment variable.
@@ -576,6 +579,7 @@ menu "Environment commands"
 
 config CMD_ASKENV
 	bool "ask for env variable"
+	depends on CMDLINE
 	help
 	  Ask for environment variable
 
@@ -2132,6 +2136,7 @@ config CMD_EFICONFIG
 
 config CMD_EXCEPTION
 	bool "exception - raise exception"
+	depends on CMDLINE
 	depends on ARM || RISCV || SANDBOX || X86
 	help
 	  Enable the 'exception' command which allows to raise an exception.
@@ -2238,6 +2243,7 @@ config CMD_SYSBOOT
 
 config CMD_QFW
 	bool "qfw"
+	depends on CMDLINE
 	select QFW
 	help
 	  This provides access to the QEMU firmware interface.  The main
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 05/32] test: Make UNIT_TEST depend on CMDLINE

[Simon Glass]

Many tests make some use of the command line, so require it for all test
code.

This could be teased apart, perhaps with a test flag indicating that it
uses the command line. Leave that for later.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
---

Changes in v3:
- Fix 'ut' typo

 test/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/test/Kconfig b/test/Kconfig
index 830245b6f9a9..2b4036704f91 100644
--- a/test/Kconfig
+++ b/test/Kconfig
@@ -2,6 +2,7 @@ menu "Testing"
 
 config UNIT_TEST
 	bool "Unit tests"
+	depends on CMDLINE
 	help
 	  Select this to compile in unit tests for various parts of
 	  U-Boot. Test suites will be subcommands of the "ut" command.
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 06/32] sifive: Drop an unnecessary #ifdef

[Simon Glass]

This code is normally compiled for sifive, but sandbox can also compile
it. We should not use UNIT_TEST as a synonym for SANDBOX, since it is
possible to disable UNIT_TEST for sandbox.

Drop the condition since it isn't needed.

Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: Sean Anderson <seanga2@gmail.com>
---

Changes in v3:
- Just drop the condition instead

 include/k210/pll.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/k210/pll.h b/include/k210/pll.h
index fd16a89cb203..6dd60b2eb4fc 100644
--- a/include/k210/pll.h
+++ b/include/k210/pll.h
@@ -13,7 +13,7 @@ struct k210_pll_config {
 	u8 od;
 };
 
-#ifdef CONFIG_UNIT_TEST
+#ifdef CONFIG_SANDBOX
 TEST_STATIC int k210_pll_calc_config(u32 rate, u32 rate_in,
 				     struct k210_pll_config *best);
 #ifndef nop
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 07/32] fastboot: Declare a dependency on CMDLINE

[Simon Glass]

When CMDLINE is not enabled, this code fails to build. Correct this by
adding a condition.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v2)

Changes in v2:
- Change this to use a Kconfig dependency instead of code failing

 drivers/fastboot/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/fastboot/Kconfig b/drivers/fastboot/Kconfig
index 837c6f1180da..24b100f381fe 100644
--- a/drivers/fastboot/Kconfig
+++ b/drivers/fastboot/Kconfig
@@ -2,6 +2,7 @@ menu "Fastboot support"
 
 config FASTBOOT
 	bool
+	depends on CMDLINE
 	imply ANDROID_BOOT_IMAGE
 	imply CMD_FASTBOOT
 	help
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 08/32] cli: Always build cli_getch

[Simon Glass]

This module is used for user input with menus, not just with the command
line. Compile it always, so it is available even when CMDLINE is disabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 common/Makefile | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/common/Makefile b/common/Makefile
index cdeadf72026c..b21916f15340 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -8,6 +8,7 @@ ifndef CONFIG_SPL_BUILD
 obj-y += init/
 obj-y += main.o
 obj-y += exports.o
+obj-y += cli_getch.o
 obj-$(CONFIG_HUSH_PARSER) += cli_hush.o
 obj-$(CONFIG_AUTOBOOT) += autoboot.o
 
@@ -37,7 +38,7 @@ obj-$(CONFIG_SPLASH_SOURCE) += splash_source.o
 obj-$(CONFIG_MENU) += menu.o
 obj-$(CONFIG_UPDATE_COMMON) += update.o
 obj-$(CONFIG_USB_KEYBOARD) += usb_kbd.o
-obj-$(CONFIG_CMDLINE) += cli_getch.o cli_readline.o cli_simple.o
+obj-$(CONFIG_CMDLINE) += cli_readline.o cli_simple.o
 
 endif # !CONFIG_SPL_BUILD
 
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 09/32] cmd: Use an #ifdef around run_commandf()

[Simon Glass]

This is not available if CMDLINE is disabled, so add an #ifdef to correct
this.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 common/cli.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/common/cli.c b/common/cli.c
index 3916a7b10a7d..4d0fea4387f2 100644
--- a/common/cli.c
+++ b/common/cli.c
@@ -129,6 +129,7 @@ int run_command_list(const char *cmd, int len, int flag)
 	return rcode;
 }
 
+#ifdef CONFIG_CMDLINE
 int run_commandf(const char *fmt, ...)
 {
 	va_list args;
@@ -153,6 +154,7 @@ int run_commandf(const char *fmt, ...)
 	}
 	return run_command(console_buffer, 0);
 }
+#endif /* CMDLINE */
 
 /****************************************************************************/
 
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 10/32] Move bootmenu_conv_key() into its own file

[Simon Glass]

This conversion function is used by expo which does not require CMDLINE.
The menu feature does require CMDLINE.

Move the function into a separate file so that it can be used even when
CMDLINE is not enabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 common/Makefile    |  2 +-
 common/cli_getch.c |  1 +
 common/menu.c      | 40 -------------------------------------
 common/menu_key.c  | 49 ++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 51 insertions(+), 41 deletions(-)
 create mode 100644 common/menu_key.c

diff --git a/common/Makefile b/common/Makefile
index b21916f15340..637066ae6682 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -8,7 +8,7 @@ ifndef CONFIG_SPL_BUILD
 obj-y += init/
 obj-y += main.o
 obj-y += exports.o
-obj-y += cli_getch.o
+obj-y += cli_getch.o menu_key.o
 obj-$(CONFIG_HUSH_PARSER) += cli_hush.o
 obj-$(CONFIG_AUTOBOOT) += autoboot.o
 
diff --git a/common/cli_getch.c b/common/cli_getch.c
index 61d4cb261b81..c3332dc27fae 100644
--- a/common/cli_getch.c
+++ b/common/cli_getch.c
@@ -8,6 +8,7 @@
 
 #include <common.h>
 #include <cli.h>
+#include <menu.h>
 
 /**
  * enum cli_esc_state_t - indicates what to do with an escape character
diff --git a/common/menu.c b/common/menu.c
index b55cf7b99967..844d0ec52af3 100644
--- a/common/menu.c
+++ b/common/menu.c
@@ -483,46 +483,6 @@ enum bootmenu_key bootmenu_autoboot_loop(struct bootmenu_data *menu,
 	return key;
 }
 
-enum bootmenu_key bootmenu_conv_key(int ichar)
-{
-	enum bootmenu_key key;
-
-	switch (ichar) {
-	case '\n':
-		/* enter key was pressed */
-		key = BKEY_SELECT;
-		break;
-	case CTL_CH('c'):
-	case '\e':
-		/* ^C was pressed */
-		key = BKEY_QUIT;
-		break;
-	case CTL_CH('p'):
-		key = BKEY_UP;
-		break;
-	case CTL_CH('n'):
-		key = BKEY_DOWN;
-		break;
-	case CTL_CH('s'):
-		key = BKEY_SAVE;
-		break;
-	case '+':
-		key = BKEY_PLUS;
-		break;
-	case '-':
-		key = BKEY_MINUS;
-		break;
-	case ' ':
-		key = BKEY_SPACE;
-		break;
-	default:
-		key = BKEY_NONE;
-		break;
-	}
-
-	return key;
-}
-
 enum bootmenu_key bootmenu_loop(struct bootmenu_data *menu,
 				struct cli_ch_state *cch)
 {
diff --git a/common/menu_key.c b/common/menu_key.c
new file mode 100644
index 000000000000..4e9c3b426b0c
--- /dev/null
+++ b/common/menu_key.c
@@ -0,0 +1,49 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright 2010-2011 Calxeda, Inc.
+ * Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
+ */
+
+#include <common.h>
+#include <cli.h>
+#include <menu.h>
+
+enum bootmenu_key bootmenu_conv_key(int ichar)
+{
+	enum bootmenu_key key;
+
+	switch (ichar) {
+	case '\n':
+		/* enter key was pressed */
+		key = BKEY_SELECT;
+		break;
+	case CTL_CH('c'):
+	case '\e':
+		/* ^C was pressed */
+		key = BKEY_QUIT;
+		break;
+	case CTL_CH('p'):
+		key = BKEY_UP;
+		break;
+	case CTL_CH('n'):
+		key = BKEY_DOWN;
+		break;
+	case CTL_CH('s'):
+		key = BKEY_SAVE;
+		break;
+	case '+':
+		key = BKEY_PLUS;
+		break;
+	case '-':
+		key = BKEY_MINUS;
+		break;
+	case ' ':
+		key = BKEY_SPACE;
+		break;
+	default:
+		key = BKEY_NONE;
+		break;
+	}
+
+	return key;
+}
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 11/32] pxe: Depend on CMDLINE

[Simon Glass]

We cannot use PXE or sysboot commands without CONFIG_CMDLINE so add the
required condition.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 cmd/Kconfig | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 00a7f84bce99..18be3da972d2 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -2002,6 +2002,7 @@ config CMD_ETHSW
 
 config CMD_PXE
 	bool "pxe"
+	depends on CMDLINE
 	select PXE_UTILS
 	help
 	  Boot image via network using PXE protocol
@@ -2237,6 +2238,7 @@ config CMD_SOUND
 
 config CMD_SYSBOOT
 	bool "sysboot"
+	depends on CMDLINE
 	select PXE_UTILS
 	help
 	  Boot image via local extlinux.conf file
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 12/32] env: Split out non-command code into a new file

[Simon Glass]

It is not possible to set environment variables without having
CONFIG_CMD_NVEDIT enabled. When CONFIG_CMDLINE is disabled, we need a
way to set variables.

Split the setting code out into its own file, so that env_set() is
available even when CONFIG_CMDLINE is not. If it is never called, the
code will be dropped at link time.

Update the Makefile rule to only include the env commands when
CONFIG_CMD_NVEDIT is enabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 cmd/Makefile           |   2 +-
 cmd/nvedit.c           | 122 ++-----------------------------------
 env/Makefile           |   1 +
 env/env_set.c          | 132 +++++++++++++++++++++++++++++++++++++++++
 include/env_internal.h |  23 +++++++
 5 files changed, 161 insertions(+), 119 deletions(-)
 create mode 100644 env/env_set.c

diff --git a/cmd/Makefile b/cmd/Makefile
index 44db5f22861e..29d7c500c66e 100644
--- a/cmd/Makefile
+++ b/cmd/Makefile
@@ -245,7 +245,7 @@ endif # !CONFIG_SPL_BUILD
 obj-$(CONFIG_$(SPL_)CMD_TLV_EEPROM) += tlv_eeprom.o
 
 # core command
-obj-y += nvedit.o
+obj-$(CONFIG_$(SPL_)CMDLINE) += nvedit.o
 
 obj-$(CONFIG_CMD_BCM_EXT_UTILS) += broadcom/
 
diff --git a/cmd/nvedit.c b/cmd/nvedit.c
index daf1ad37f9be..812d82b157f4 100644
--- a/cmd/nvedit.c
+++ b/cmd/nvedit.c
@@ -49,20 +49,6 @@ DECLARE_GLOBAL_DATA_PTR;
  */
 #define	MAX_ENV_SIZE	(1 << 20)	/* 1 MiB */
 
-/*
- * This variable is incremented on each do_env_set(), so it can
- * be used via env_get_id() as an indication, if the environment
- * has changed or not. So it is possible to reread an environment
- * variable only if the environment was changed ... done so for
- * example in NetInitLoop()
- */
-static int env_id = 1;
-
-int env_get_id(void)
-{
-	return env_id;
-}
-
 #ifndef CONFIG_SPL_BUILD
 /*
  * Command interface: print one or all environment variables
@@ -195,108 +181,8 @@ DONE:
 
 	return 0;
 }
-#endif
-#endif /* CONFIG_SPL_BUILD */
-
-/*
- * Set a new environment variable,
- * or replace or delete an existing one.
- */
-static int _do_env_set(int flag, int argc, char *const argv[], int env_flag)
-{
-	int   i, len;
-	char  *name, *value, *s;
-	struct env_entry e, *ep;
-
-	debug("Initial value for argc=%d\n", argc);
-
-#if !IS_ENABLED(CONFIG_SPL_BUILD) && IS_ENABLED(CONFIG_CMD_NVEDIT_EFI)
-	if (argc > 1 && argv[1][0] == '-' && argv[1][1] == 'e')
-		return do_env_set_efi(NULL, flag, --argc, ++argv);
-#endif
-
-	while (argc > 1 && **(argv + 1) == '-') {
-		char *arg = *++argv;
-
-		--argc;
-		while (*++arg) {
-			switch (*arg) {
-			case 'f':		/* force */
-				env_flag |= H_FORCE;
-				break;
-			default:
-				return CMD_RET_USAGE;
-			}
-		}
-	}
-	debug("Final value for argc=%d\n", argc);
-	name = argv[1];
-
-	if (strchr(name, '=')) {
-		printf("## Error: illegal character '='"
-		       "in variable name \"%s\"\n", name);
-		return 1;
-	}
-
-	env_id++;
-
-	/* Delete only ? */
-	if (argc < 3 || argv[2] == NULL) {
-		int rc = hdelete_r(name, &env_htab, env_flag);
-
-		/* If the variable didn't exist, don't report an error */
-		return rc && rc != -ENOENT ? 1 : 0;
-	}
-
-	/*
-	 * Insert / replace new value
-	 */
-	for (i = 2, len = 0; i < argc; ++i)
-		len += strlen(argv[i]) + 1;
-
-	value = malloc(len);
-	if (value == NULL) {
-		printf("## Can't malloc %d bytes\n", len);
-		return 1;
-	}
-	for (i = 2, s = value; i < argc; ++i) {
-		char *v = argv[i];
-
-		while ((*s++ = *v++) != '\0')
-			;
-		*(s - 1) = ' ';
-	}
-	if (s != value)
-		*--s = '\0';
-
-	e.key	= name;
-	e.data	= value;
-	hsearch_r(e, ENV_ENTER, &ep, &env_htab, env_flag);
-	free(value);
-	if (!ep) {
-		printf("## Error inserting \"%s\" variable, errno=%d\n",
-			name, errno);
-		return 1;
-	}
-
-	return 0;
-}
-
-int env_set(const char *varname, const char *varvalue)
-{
-	const char * const argv[4] = { "setenv", varname, varvalue, NULL };
-
-	/* before import into hashtable */
-	if (!(gd->flags & GD_FLG_ENV_READY))
-		return 1;
-
-	if (varvalue == NULL || varvalue[0] == '\0')
-		return _do_env_set(0, 2, (char * const *)argv, H_PROGRAMMATIC);
-	else
-		return _do_env_set(0, 3, (char * const *)argv, H_PROGRAMMATIC);
-}
+#endif /* CONFIG_CMD_GREPENV */
 
-#ifndef CONFIG_SPL_BUILD
 static int do_env_set(struct cmd_tbl *cmdtp, int flag, int argc,
 		      char *const argv[])
 {
@@ -456,7 +342,7 @@ int do_env_callback(struct cmd_tbl *cmdtp, int flag, int argc,
 	hwalk_r(&env_htab, print_active_callback);
 	return 0;
 }
-#endif
+#endif /* CONFIG_CMD_ENV_CALLBACK */
 
 #if defined(CONFIG_CMD_ENV_FLAGS)
 static int print_static_flags(const char *var_name, const char *flags,
@@ -529,7 +415,7 @@ int do_env_flags(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[])
 	hwalk_r(&env_htab, print_active_flags);
 	return 0;
 }
-#endif
+#endif /* CONFIG_CMD_ENV_FLAGS */
 
 /*
  * Interactively edit an environment variable
@@ -679,7 +565,7 @@ static int do_env_delete(struct cmd_tbl *cmdtp, int flag,
 	}
 	debug("Final value for argc=%d\n", argc);
 
-	env_id++;
+	env_inc_id();
 
 	while (--argc > 0) {
 		char *name = *++argv;
diff --git a/env/Makefile b/env/Makefile
index 673b979fdfa9..5250b6df2cfc 100644
--- a/env/Makefile
+++ b/env/Makefile
@@ -5,6 +5,7 @@
 
 obj-$(CONFIG_$(SPL_TPL_)ENV_SUPPORT) += common.o
 obj-$(CONFIG_$(SPL_TPL_)ENV_SUPPORT) += env.o
+obj-$(CONFIG_$(SPL_TPL_)ENV_SUPPORT) += env_set.o
 obj-$(CONFIG_$(SPL_TPL_)ENV_SUPPORT) += attr.o
 obj-$(CONFIG_$(SPL_TPL_)ENV_SUPPORT) += flags.o
 
diff --git a/env/env_set.c b/env/env_set.c
new file mode 100644
index 000000000000..eccbda6a791c
--- /dev/null
+++ b/env/env_set.c
@@ -0,0 +1,132 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2000-2013
+ * Wolfgang Denk, DENX Software Engineering, wd@denx.de.
+ *
+ * (C) Copyright 2001 Sysgo Real-Time Solutions, GmbH <www.elinos.com>
+ * Andreas Heppel <aheppel@sysgo.de>
+ *
+ * Copyright 2011 Freescale Semiconductor, Inc.
+ */
+
+#include <common.h>
+#include <command.h>
+#include <env.h>
+#include <env_internal.h>
+#include <malloc.h>
+#include <asm/global_data.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+/*
+ * This variable is incremented on each do_env_set(), so it can
+ * be used via env_get_id() as an indication, if the environment
+ * has changed or not. So it is possible to reread an environment
+ * variable only if the environment was changed ... done so for
+ * example in NetInitLoop()
+ */
+static int env_id = 1;
+
+int env_get_id(void)
+{
+	return env_id;
+}
+
+void env_inc_id(void)
+{
+	env_id++;
+}
+
+int _do_env_set(int flag, int argc, char *const argv[], int env_flag)
+{
+	int   i, len;
+	char  *name, *value, *s;
+	struct env_entry e, *ep;
+
+	debug("Initial value for argc=%d\n", argc);
+
+#if !IS_ENABLED(CONFIG_SPL_BUILD) && IS_ENABLED(CONFIG_CMD_NVEDIT_EFI)
+	if (argc > 1 && argv[1][0] == '-' && argv[1][1] == 'e')
+		return do_env_set_efi(NULL, flag, --argc, ++argv);
+#endif
+
+	while (argc > 1 && **(argv + 1) == '-') {
+		char *arg = *++argv;
+
+		--argc;
+		while (*++arg) {
+			switch (*arg) {
+			case 'f':		/* force */
+				env_flag |= H_FORCE;
+				break;
+			default:
+				return CMD_RET_USAGE;
+			}
+		}
+	}
+	debug("Final value for argc=%d\n", argc);
+	name = argv[1];
+
+	if (strchr(name, '=')) {
+		printf("## Error: illegal character '=' in variable name \"%s\"\n",
+		       name);
+		return 1;
+	}
+
+	env_id++;
+
+	/* Delete only ? */
+	if (argc < 3 || !argv[2]) {
+		int rc = hdelete_r(name, &env_htab, env_flag);
+
+		/* If the variable didn't exist, don't report an error */
+		return rc && rc != -ENOENT ? 1 : 0;
+	}
+
+	/*
+	 * Insert / replace new value
+	 */
+	for (i = 2, len = 0; i < argc; ++i)
+		len += strlen(argv[i]) + 1;
+
+	value = malloc(len);
+	if (!value) {
+		printf("## Can't malloc %d bytes\n", len);
+		return 1;
+	}
+	for (i = 2, s = value; i < argc; ++i) {
+		char *v = argv[i];
+
+		while ((*s++ = *v++) != '\0')
+			;
+		*(s - 1) = ' ';
+	}
+	if (s != value)
+		*--s = '\0';
+
+	e.key	= name;
+	e.data	= value;
+	hsearch_r(e, ENV_ENTER, &ep, &env_htab, env_flag);
+	free(value);
+	if (!ep) {
+		printf("## Error inserting \"%s\" variable, errno=%d\n",
+		       name, errno);
+		return 1;
+	}
+
+	return 0;
+}
+
+int env_set(const char *varname, const char *varvalue)
+{
+	const char * const argv[4] = { "setenv", varname, varvalue, NULL };
+
+	/* before import into hashtable */
+	if (!(gd->flags & GD_FLG_ENV_READY))
+		return 1;
+
+	if (!varvalue || !varvalue[0])
+		return _do_env_set(0, 2, (char * const *)argv, H_PROGRAMMATIC);
+	else
+		return _do_env_set(0, 3, (char * const *)argv, H_PROGRAMMATIC);
+}
diff --git a/include/env_internal.h b/include/env_internal.h
index 6a6949464689..99e358b6d8a5 100644
--- a/include/env_internal.h
+++ b/include/env_internal.h
@@ -260,6 +260,29 @@ const char *env_fat_get_intf(void);
  * Return: string of device and partition
  */
 char *env_fat_get_dev_part(void);
+
+/*
+ * _do_env_set() - Add / replace / delete an environment variable
+ *
+ * This implements the bulk of the 'env set' command:
+ *
+ *	env set [-f] name [value]
+ *
+ * Sets the value of variable <name>
+ * If <value> is NULL, it removes the variable
+ * Use the -f flag to overwrite read-only/write-once variables
+ *
+ * @flag: CMD_FLAG_... value
+ * @argc: Number of arguments
+ * @args: List of arguments
+ * @env_flag: H_... flags from search.h
+ * Return: 0 if OK, 1 on failure, or CMD_RET_USAGE for invalid flag
+ */
+int _do_env_set(int flag, int argc, char *const argv[], int env_flag);
+
+/** env_inc_id() - Increment the environment ID */
+void env_inc_id(void);
+
 #endif /* DO_DEPS_ONLY */
 
 #endif /* _ENV_INTERNAL_H_ */
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 13/32] console: Move SYS_PBSIZE into common/

[Simon Glass]

This relates to printing output and does not need a command line. Move
it next to the other console-related options.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 cmd/Kconfig    | 5 -----
 common/Kconfig | 5 +++++
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 18be3da972d2..148414f011df 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -84,11 +84,6 @@ config SYS_CBSIZE
 	default 256 if M68K || PPC
 	default 1024
 
-config SYS_PBSIZE
-	int "Buffer size for console output"
-	default 1024 if ARCH_SUNXI
-	default 1044
-
 config SYS_XTRACE
 	bool "Command execution tracer"
 	depends on CMDLINE
diff --git a/common/Kconfig b/common/Kconfig
index 93c96f23b013..1ffb055744d9 100644
--- a/common/Kconfig
+++ b/common/Kconfig
@@ -52,6 +52,11 @@ config CONSOLE_RECORD_IN_SIZE
 	  The buffer is allocated immediately after the malloc() region is
 	  ready.
 
+config SYS_PBSIZE
+	int "Buffer size for console output"
+	default 1024 if ARCH_SUNXI
+	default 1044
+
 config DISABLE_CONSOLE
 	bool "Add functionality to disable console completely"
 	help
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Simon Glass]

There are two cleanup functions needed during boot which depend on
CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()

Provide static inline versions of these for when commands are not
enabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 arch/arm/lib/bootm.c |  2 ++
 boot/bootm.c         | 10 ++++------
 include/bootm.h      | 15 +++++++++++++--
 3 files changed, 19 insertions(+), 8 deletions(-)

diff --git a/arch/arm/lib/bootm.c b/arch/arm/lib/bootm.c
index c56285738a26..db8df57cb56e 100644
--- a/arch/arm/lib/bootm.c
+++ b/arch/arm/lib/bootm.c
@@ -42,9 +42,11 @@ DECLARE_GLOBAL_DATA_PTR;
 
 static struct tag *params;
 
+#ifdef CONFIG_CMD_BOOTM
 __weak void board_quiesce_devices(void)
 {
 }
+#endif
 
 /**
  * announce_and_cleanup() - Print message and prepare for kernel boot
diff --git a/boot/bootm.c b/boot/bootm.c
index 8f96a80d4259..7deae058552a 100644
--- a/boot/bootm.c
+++ b/boot/bootm.c
@@ -47,9 +47,11 @@ static const void *boot_get_kernel(struct cmd_tbl *cmdtp, int flag, int argc,
 				   char *const argv[], struct bootm_headers *images,
 				   ulong *os_data, ulong *os_len);
 
+#ifdef CONFIG_CMD_BOOTM
 __weak void board_quiesce_devices(void)
 {
 }
+#endif
 
 #ifdef CONFIG_LMB
 static void boot_start_lmb(struct bootm_headers *images)
@@ -470,12 +472,7 @@ static int bootm_load_os(struct bootm_headers *images, int boot_progress)
 	return 0;
 }
 
-/**
- * bootm_disable_interrupts() - Disable interrupts in preparation for load/boot
- *
- * Return: interrupt flag (0 if interrupts were disabled, non-zero if they were
- *	enabled)
- */
+#ifdef CONFIG_CMD_BOOTM
 ulong bootm_disable_interrupts(void)
 {
 	ulong iflag;
@@ -505,6 +502,7 @@ ulong bootm_disable_interrupts(void)
 #endif
 	return iflag;
 }
+#endif
 
 #define CONSOLE_ARG		"console="
 #define NULL_CONSOLE		(CONSOLE_ARG "ttynull")
diff --git a/include/bootm.h b/include/bootm.h
index c3c7336207b1..17c740449efd 100644
--- a/include/bootm.h
+++ b/include/bootm.h
@@ -50,8 +50,6 @@ int bootm_host_load_images(const void *fit, int cfg_noffset);
 int boot_selected_os(int argc, char *const argv[], int state,
 		     struct bootm_headers *images, boot_os_fn *boot_fn);
 
-ulong bootm_disable_interrupts(void);
-
 /* This is a special function used by booti/bootz */
 int bootm_find_images(int flag, int argc, char *const argv[], ulong start,
 		      ulong size);
@@ -62,6 +60,15 @@ int do_bootm_states(struct cmd_tbl *cmdtp, int flag, int argc,
 
 void arch_preboot_os(void);
 
+#ifdef CONFIG_CMD_BOOTM
+/**
+ * bootm_disable_interrupts() - Disable interrupts, stop Ethernet and USB
+ *
+ * Return: interrupt flag (0 if interrupts were disabled, non-zero if they were
+ *	enabled)
+ */
+ulong bootm_disable_interrupts(void);
+
 /*
  * boards should define this to disable devices when EFI exits from boot
  * services.
@@ -69,6 +76,10 @@ void arch_preboot_os(void);
  * TODO(sjg@chromium.org>): Update this to use driver model's device_remove().
  */
 void board_quiesce_devices(void);
+#else
+static inline ulong bootm_disable_interrupts(void) { return 0; }
+static inline void board_quiesce_devices(void) {}
+#endif
 
 /**
  * switch_to_non_secure_mode() - switch to non-secure mode
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 15/32] fdt: Move working_fdt into fdt_support

[Simon Glass]

This can be accessed even when commands are not enabled. Move it into
the fdt_support.c file, which is where most of the FDT helpers are.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 boot/fdt_support.c | 5 +++++
 cmd/fdt.c          | 5 -----
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/boot/fdt_support.c b/boot/fdt_support.c
index 5e49078f8c35..6ae7b8e20f65 100644
--- a/boot/fdt_support.c
+++ b/boot/fdt_support.c
@@ -23,6 +23,11 @@
 #include <fdtdec.h>
 #include <version.h>
 
+/*
+ * The working_fdt points to our working flattened device tree.
+ */
+struct fdt_header *working_fdt;
+
 /**
  * fdt_getprop_u32_default_node - Return a node's property or a default
  *
diff --git a/cmd/fdt.c b/cmd/fdt.c
index 331564c13be9..86e847a41826 100644
--- a/cmd/fdt.c
+++ b/cmd/fdt.c
@@ -31,11 +31,6 @@ static int fdt_parse_prop(char *const*newval, int count, char *data, int *len);
 static int fdt_print(const char *pathp, char *prop, int depth);
 static int is_printable_string(const void *data, int len);
 
-/*
- * The working_fdt points to our working flattened device tree.
- */
-struct fdt_header *working_fdt;
-
 static void set_working_fdt_addr_quiet(ulong addr)
 {
 	void *buf;
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 16/32] net: Depend on CONFIG_CMDLINE

[Simon Glass]

At present it isn't possible to use networking without the command line
enabled. Add this as a condition.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
---

(no changes since v1)

 cmd/Kconfig | 1 +
 net/Kconfig | 1 +
 2 files changed, 2 insertions(+)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 148414f011df..c6ea5c860e33 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -1694,6 +1694,7 @@ if NET
 
 menuconfig CMD_NET
 	bool "Network commands"
+	depends on CMDLINE
 	default y
 	imply NETDEVICES
 
diff --git a/net/Kconfig b/net/Kconfig
index 4215889127c9..25d494e1db46 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -4,6 +4,7 @@
 
 menuconfig NET
 	bool "Networking support"
+	depends on CMDLINE
 	default y
 
 if NET
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 17/32] log: Allow use without CONFIG_CMDLINE

[Simon Glass]

When CONFIG_SYS_CBSIZE is not used we need an alternative. For logging
it seems that CONFIG_SYS_PBSIZE is a better choice anyway, so update
this.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 common/log.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/common/log.c b/common/log.c
index b2de57fcb3b8..72a4de3274c7 100644
--- a/common/log.c
+++ b/common/log.c
@@ -206,7 +206,7 @@ static bool log_passes_filters(struct log_device *ldev, struct log_rec *rec)
 static int log_dispatch(struct log_rec *rec, const char *fmt, va_list args)
 {
 	struct log_device *ldev;
-	char buf[CONFIG_SYS_CBSIZE];
+	char buf[CONFIG_SYS_PBSIZE];
 
 	/*
 	 * When a log driver writes messages (e.g. via the network stack) this
@@ -268,7 +268,7 @@ int _log(enum log_category_t cat, enum log_level_t level, const char *file,
 		/* display dropped traces with console puts and DEBUG_UART */
 		if (rec.level <= CONFIG_LOG_DEFAULT_LEVEL ||
 		    rec.flags & LOGRECF_FORCE_DEBUG) {
-			char buf[CONFIG_SYS_CBSIZE];
+			char buf[CONFIG_SYS_PBSIZE];
 
 			va_start(args, fmt);
 			vsnprintf(buf, sizeof(buf), fmt, args);
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 18/32] video: Allow use without CONFIG_CMDLINE

[Simon Glass]

Provide a fallback for when CONFIG_SYS_CBSIZE is not provided, so that
the console can still be used.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 drivers/video/console_truetype.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/video/console_truetype.c b/drivers/video/console_truetype.c
index 14fb81e9563c..e4dad3f9a191 100644
--- a/drivers/video/console_truetype.c
+++ b/drivers/video/console_truetype.c
@@ -125,7 +125,11 @@ struct pos_info {
  * Allow one for each character on the command line plus one for each newline.
  * This is just an estimate, but it should not be exceeded.
  */
+#ifdef CONFIG_SYS_CBSIZE
 #define POS_HISTORY_SIZE	(CONFIG_SYS_CBSIZE * 11 / 10)
+#else
+#define POS_HISTORY_SIZE	(250 * 11 / 10)
+#endif
 
 /**
  * struct console_tt_metrics - Information about a font / size combination
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 19/32] video: Dont require the font command

[Simon Glass]

While it is nice to have the font command, using 'select' makes it
impossible to build the console code without it. Change this to use
'imply' instead.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 drivers/video/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
index ab927641bb7a..21ea5c860cca 100644
--- a/drivers/video/Kconfig
+++ b/drivers/video/Kconfig
@@ -180,7 +180,7 @@ config CONSOLE_ROTATION
 
 config CONSOLE_TRUETYPE
 	bool "Support a console that uses TrueType fonts"
-	select CMD_SELECT_FONT
+	imply CMD_SELECT_FONT
 	help
 	  TrueTrype fonts can provide outline-drawing capability rather than
 	  needing to provide a bitmap for each font and size that is needed.
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 20/32] efi: Make commands depend on CMDLINE

[Simon Glass]

Most of the EFI functionality requires the command line, so make this
explicit. Future work could adjust this, but it needs effort within
the booting support first, like the bootm command.

Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
---

(no changes since v2)

Changes in v2:
- Adjust the depends on the commands, instead of EFI_LOADER

 cmd/Kconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index c6ea5c860e33..46e484fc08b6 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -384,7 +384,7 @@ config SYS_BOOTM_LEN
 
 config CMD_BOOTEFI
 	bool "bootefi"
-	depends on EFI_LOADER
+	depends on EFI_LOADER && CMDLINE
 	default y
 	help
 	  Boot an EFI image from memory.
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 21/32] efi: Rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR

[Simon Glass]

The command should not be used to enable library functionality. Add a
new BOOTEFI_BOOTMGR Kconfig for that. Adjust the conditions so that the
same code is built.

Signed-off-by: Simon Glass <sjg@chromium.org>
Suggested-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
---

Changes in v3:
- Add new patch to rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR

 cmd/Kconfig             | 9 +++++++++
 lib/efi_loader/Kconfig  | 7 +++----
 lib/efi_loader/Makefile | 2 +-
 3 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 46e484fc08b6..3b4112d9f319 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -306,6 +306,15 @@ config CMD_BOOTI
 	help
 	  Boot an AArch64 Linux Kernel image from memory.
 
+config CMD_BOOTEFI_BOOTMGR
+	bool "UEFI Boot Manager command"
+	depends on BOOTEFI_BOOTMGR && CMD_BOOTEFI
+	default y
+	help
+	  Select this option if you want to select the UEFI binary to be booted
+	  via UEFI variables Boot####, BootOrder, and BootNext. This enables the
+	  'bootefi bootmgr' command.
+
 config BOOTM_LINUX
 	bool "Support booting Linux OS images"
 	depends on CMD_BOOTM || CMD_BOOTZ || CMD_BOOTI
diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
index 621ed5e5b0fb..13cad6342c36 100644
--- a/lib/efi_loader/Kconfig
+++ b/lib/efi_loader/Kconfig
@@ -32,15 +32,14 @@ config EFI_LOADER
 
 if EFI_LOADER
 
-config CMD_BOOTEFI_BOOTMGR
+config BOOTEFI_BOOTMGR
 	bool "UEFI Boot Manager"
-	depends on CMDLINE
 	default y
 	select BOOTMETH_GLOBAL if BOOTSTD
 	help
 	  Select this option if you want to select the UEFI binary to be booted
-	  via UEFI variables Boot####, BootOrder, and BootNext. This enables the
-	  'bootefi bootmgr' command.
+	  via UEFI variables Boot####, BootOrder, and BootNext. You should also
+	  normally enable CMD_BOOTEFI_BOOTMGR so that the command is available.
 
 choice
 	prompt "Store for non-volatile UEFI variables"
diff --git a/lib/efi_loader/Makefile b/lib/efi_loader/Makefile
index 8d31fc61c601..0a2cb6e3c476 100644
--- a/lib/efi_loader/Makefile
+++ b/lib/efi_loader/Makefile
@@ -42,7 +42,7 @@ targets += initrddump.o
 endif
 
 obj-$(CONFIG_CMD_BOOTEFI_HELLO) += helloworld_efi.o
-obj-$(CONFIG_CMD_BOOTEFI_BOOTMGR) += efi_bootmgr.o
+obj-$(CONFIG_BOOTEFI_BOOTMGR) += efi_bootmgr.o
 obj-y += efi_boottime.o
 obj-y += efi_helper.o
 obj-$(CONFIG_EFI_HAVE_CAPSULE_SUPPORT) += efi_capsule.o
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
available, add it as an explicit dependency.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v2)

Changes in v2:
- Add new patch to update EFI_LOADER to depend on DM_ETH

 lib/efi_loader/Kconfig | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
index 13cad6342c36..fca4b3eef270 100644
--- a/lib/efi_loader/Kconfig
+++ b/lib/efi_loader/Kconfig
@@ -11,6 +11,7 @@ config EFI_LOADER
 	# We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
 	depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
 	depends on BLK
+	depends on DM_ETH
 	depends on !EFI_APP
 	default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
 	select CHARSET
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 23/32] cli: Split command-line-editing into its own file

[Simon Glass]

This features has buffers and code which is behind an #ifdef at
present. Move it into its own file, exporting functions as needed.

Drop add_idx_minus_one() since it is not used.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 boot/Kconfig          |   3 +-
 common/Makefile       |   7 +-
 common/cli_cread.c    | 403 ++++++++++++++++++++++++++++++++++++++++++
 common/cli_readline.c | 384 +---------------------------------------
 include/cli.h         |  30 ++++
 5 files changed, 441 insertions(+), 386 deletions(-)
 create mode 100644 common/cli_cread.c

diff --git a/boot/Kconfig b/boot/Kconfig
index 6461f7ebd040..2fbe70245ec9 100644
--- a/boot/Kconfig
+++ b/boot/Kconfig
@@ -349,7 +349,8 @@ config PXE_UTILS
 
 config BOOT_DEFAULTS
 	bool  # Common defaults for standard boot and distroboot
-	imply USE_BOOTCOMMAND
+	depends on CMDLINE
+	imply USE_BOOTCOMMAND if !SANDBOX
 	select CMD_ENV_EXISTS
 	select CMD_EXT2
 	select CMD_EXT4
diff --git a/common/Makefile b/common/Makefile
index 637066ae6682..e22ced0c507f 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -38,6 +38,7 @@ obj-$(CONFIG_SPLASH_SOURCE) += splash_source.o
 obj-$(CONFIG_MENU) += menu.o
 obj-$(CONFIG_UPDATE_COMMON) += update.o
 obj-$(CONFIG_USB_KEYBOARD) += usb_kbd.o
+obj-$(CONFIG_CMDLINE_EDITING) += cli_cread.o
 obj-$(CONFIG_CMDLINE) += cli_readline.o cli_simple.o
 
 endif # !CONFIG_SPL_BUILD
@@ -90,8 +91,10 @@ obj-y += eeprom/eeprom_field.o eeprom/eeprom_layout.o
 endif
 
 obj-y += cli.o
-obj-$(CONFIG_FSL_DDR_INTERACTIVE) += cli_getch.o cli_simple.o cli_readline.o
-obj-$(CONFIG_STM32MP1_DDR_INTERACTIVE) += cli_getch.o cli_simple.o cli_readline.o
+obj-$(CONFIG_FSL_DDR_INTERACTIVE) += cli_getch.o cli_simple.o cli_readline.o \
+	cli_cread.o
+obj-$(CONFIG_STM32MP1_DDR_INTERACTIVE) += cli_getch.o cli_simple.o \
+	cli_readline.o cli_cread.o
 obj-$(CONFIG_DFU_OVER_USB) += dfu.o
 obj-y += command.o
 obj-$(CONFIG_$(SPL_TPL_)LOG) += log.o
diff --git a/common/cli_cread.c b/common/cli_cread.c
new file mode 100644
index 000000000000..19af27303cfc
--- /dev/null
+++ b/common/cli_cread.c
@@ -0,0 +1,403 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * (C) Copyright 2000
+ * Wolfgang Denk, DENX Software Engineering, wd@denx.de.
+ *
+ * Add to readline cmdline-editing by
+ * (C) Copyright 2005
+ * JinHua Luo, GuangDong Linux Center, <luo.jinhua@gd-linux.com>
+ */
+
+#include <common.h>
+#include <bootretry.h>
+#include <cli.h>
+#include <command.h>
+#include <time.h>
+#include <watchdog.h>
+#include <asm/global_data.h>
+
+DECLARE_GLOBAL_DATA_PTR;
+
+/*
+ * cmdline-editing related codes from vivi.
+ * Author: Janghoon Lyu <nandy@mizi.com>
+ */
+
+#define putnstr(str, n)	printf("%.*s", (int)n, str)
+
+#define CTL_BACKSPACE		('\b')
+#define DEL			((char)255)
+#define DEL7			((char)127)
+#define CREAD_HIST_CHAR		('!')
+
+#define getcmd_putch(ch)	putc(ch)
+#define getcmd_cbeep()		getcmd_putch('\a')
+
+#ifdef CONFIG_SPL_BUILD
+#define HIST_MAX		3
+#define HIST_SIZE		32
+#else
+#define HIST_MAX		20
+#define HIST_SIZE		CONFIG_SYS_CBSIZE
+#endif
+
+static int hist_max;
+static int hist_add_idx;
+static int hist_cur = -1;
+static uint hist_num;
+
+static char *hist_list[HIST_MAX];
+static char hist_lines[HIST_MAX][HIST_SIZE + 1];	/* Save room for NULL */
+
+static void getcmd_putchars(int count, int ch)
+{
+	int i;
+
+	for (i = 0; i < count; i++)
+		getcmd_putch(ch);
+}
+
+void hist_init(void)
+{
+	int i;
+
+	hist_max = 0;
+	hist_add_idx = 0;
+	hist_cur = -1;
+	hist_num = 0;
+
+	for (i = 0; i < HIST_MAX; i++) {
+		hist_list[i] = hist_lines[i];
+		hist_list[i][0] = '\0';
+	}
+}
+
+void cread_add_to_hist(char *line)
+{
+	if (line[0] && line[0] != CREAD_HIST_CHAR) {
+		strcpy(hist_list[hist_add_idx], line);
+
+		if (++hist_add_idx >= HIST_MAX)
+			hist_add_idx = 0;
+
+		if (hist_add_idx > hist_max)
+			hist_max = hist_add_idx;
+
+		hist_num++;
+	}
+
+	hist_cur = hist_add_idx;
+}
+
+char *hist_prev(void)
+{
+	char *ret;
+	int old_cur;
+
+	if (hist_cur < 0)
+		return NULL;
+
+	old_cur = hist_cur;
+	if (--hist_cur < 0)
+		hist_cur = hist_max;
+
+	if (hist_cur == hist_add_idx) {
+		hist_cur = old_cur;
+		ret = NULL;
+	} else {
+		ret = hist_list[hist_cur];
+	}
+
+	return ret;
+}
+
+char *hist_next(void)
+{
+	char *ret;
+
+	if (hist_cur < 0)
+		return NULL;
+
+	if (hist_cur == hist_add_idx)
+		return NULL;
+
+	if (++hist_cur > hist_max)
+		hist_cur = 0;
+
+	if (hist_cur == hist_add_idx)
+		ret = "";
+	else
+		ret = hist_list[hist_cur];
+
+	return ret;
+}
+
+void cread_print_hist_list(void)
+{
+	int i;
+	uint n;
+
+	n = hist_num - hist_max;
+
+	i = hist_add_idx + 1;
+	while (1) {
+		if (i > hist_max)
+			i = 0;
+		if (i == hist_add_idx)
+			break;
+		printf("%s\n", hist_list[i]);
+		n++;
+		i++;
+	}
+}
+
+#define BEGINNING_OF_LINE() {			\
+	while (cls->num) {			\
+		getcmd_putch(CTL_BACKSPACE);	\
+		cls->num--;			\
+	}					\
+}
+
+#define ERASE_TO_EOL() {				\
+	if (cls->num < cls->eol_num) {		\
+		printf("%*s", (int)(cls->eol_num - cls->num), ""); \
+		do {					\
+			getcmd_putch(CTL_BACKSPACE);	\
+		} while (--cls->eol_num > cls->num);	\
+	}						\
+}
+
+#define REFRESH_TO_EOL() {				\
+	if (cls->num < cls->eol_num) {			\
+		uint wlen = cls->eol_num - cls->num;	\
+		putnstr(buf + cls->num, wlen);		\
+		cls->num = cls->eol_num;		\
+	}						\
+}
+
+static void cread_add_char(char ichar, int insert, uint *num,
+			   uint *eol_num, char *buf, uint len)
+{
+	uint wlen;
+
+	/* room ??? */
+	if (insert || *num == *eol_num) {
+		if (*eol_num > len - 1) {
+			getcmd_cbeep();
+			return;
+		}
+		(*eol_num)++;
+	}
+
+	if (insert) {
+		wlen = *eol_num - *num;
+		if (wlen > 1)
+			memmove(&buf[*num + 1], &buf[*num], wlen - 1);
+
+		buf[*num] = ichar;
+		putnstr(buf + *num, wlen);
+		(*num)++;
+		while (--wlen)
+			getcmd_putch(CTL_BACKSPACE);
+	} else {
+		/* echo the character */
+		wlen = 1;
+		buf[*num] = ichar;
+		putnstr(buf + *num, wlen);
+		(*num)++;
+	}
+}
+
+static void cread_add_str(char *str, int strsize, int insert,
+			  uint *num, uint *eol_num, char *buf, uint len)
+{
+	while (strsize--) {
+		cread_add_char(*str, insert, num, eol_num, buf, len);
+		str++;
+	}
+}
+
+int cread_line_process_ch(struct cli_line_state *cls, char ichar)
+{
+	char *buf = cls->buf;
+
+	/* ichar=0x0 when error occurs in U-Boot getc */
+	if (!ichar)
+		return -EAGAIN;
+
+	if (ichar == '\n') {
+		putc('\n');
+		buf[cls->eol_num] = '\0';	/* terminate the string */
+		return 0;
+	}
+
+	switch (ichar) {
+	case CTL_CH('a'):
+		BEGINNING_OF_LINE();
+		break;
+	case CTL_CH('c'):	/* ^C - break */
+		*buf = '\0';	/* discard input */
+		return -EINTR;
+	case CTL_CH('f'):
+		if (cls->num < cls->eol_num) {
+			getcmd_putch(buf[cls->num]);
+			cls->num++;
+		}
+		break;
+	case CTL_CH('b'):
+		if (cls->num) {
+			getcmd_putch(CTL_BACKSPACE);
+			cls->num--;
+		}
+		break;
+	case CTL_CH('d'):
+		if (cls->num < cls->eol_num) {
+			uint wlen;
+
+			wlen = cls->eol_num - cls->num - 1;
+			if (wlen) {
+				memmove(&buf[cls->num], &buf[cls->num + 1],
+					wlen);
+				putnstr(buf + cls->num, wlen);
+			}
+
+			getcmd_putch(' ');
+			do {
+				getcmd_putch(CTL_BACKSPACE);
+			} while (wlen--);
+			cls->eol_num--;
+		}
+		break;
+	case CTL_CH('k'):
+		ERASE_TO_EOL();
+		break;
+	case CTL_CH('e'):
+		REFRESH_TO_EOL();
+		break;
+	case CTL_CH('o'):
+		cls->insert = !cls->insert;
+		break;
+	case CTL_CH('w'):
+		if (cls->num) {
+			uint base, wlen;
+
+			for (base = cls->num - 1;
+			     base >= 0 && buf[base] == ' ';)
+				base--;
+			for (; base > 0 && buf[base - 1] != ' ';)
+				base--;
+
+			/* now delete chars from base to cls->num */
+			wlen = cls->num - base;
+			cls->eol_num -= wlen;
+			memmove(&buf[base], &buf[cls->num],
+				cls->eol_num - base + 1);
+			cls->num = base;
+			getcmd_putchars(wlen, CTL_BACKSPACE);
+			puts(buf + base);
+			getcmd_putchars(wlen, ' ');
+			getcmd_putchars(wlen + cls->eol_num - cls->num,
+					CTL_BACKSPACE);
+		}
+		break;
+	case CTL_CH('x'):
+	case CTL_CH('u'):
+		BEGINNING_OF_LINE();
+		ERASE_TO_EOL();
+		break;
+	case DEL:
+	case DEL7:
+	case 8:
+		if (cls->num) {
+			uint wlen;
+
+			wlen = cls->eol_num - cls->num;
+			cls->num--;
+			memmove(&buf[cls->num], &buf[cls->num + 1], wlen);
+			getcmd_putch(CTL_BACKSPACE);
+			putnstr(buf + cls->num, wlen);
+			getcmd_putch(' ');
+			do {
+				getcmd_putch(CTL_BACKSPACE);
+			} while (wlen--);
+			cls->eol_num--;
+		}
+		break;
+	case CTL_CH('p'):
+	case CTL_CH('n'):
+		if (cls->history) {
+			char *hline;
+
+			if (ichar == CTL_CH('p'))
+				hline = hist_prev();
+			else
+				hline = hist_next();
+
+			if (!hline) {
+				getcmd_cbeep();
+				break;
+			}
+
+			/* nuke the current line */
+			/* first, go home */
+			BEGINNING_OF_LINE();
+
+			/* erase to end of line */
+			ERASE_TO_EOL();
+
+			/* copy new line into place and display */
+			strcpy(buf, hline);
+			cls->eol_num = strlen(buf);
+			REFRESH_TO_EOL();
+			break;
+		}
+		break;
+	case '\t':
+		if (IS_ENABLED(CONFIG_AUTO_COMPLETE) && cls->cmd_complete) {
+			int num2, col;
+
+			/* do not autocomplete when in the middle */
+			if (cls->num < cls->eol_num) {
+				getcmd_cbeep();
+				break;
+			}
+
+			buf[cls->num] = '\0';
+			col = strlen(cls->prompt) + cls->eol_num;
+			num2 = cls->num;
+			if (cmd_auto_complete(cls->prompt, buf, &num2, &col)) {
+				col = num2 - cls->num;
+				cls->num += col;
+				cls->eol_num += col;
+			}
+			break;
+		}
+		fallthrough;
+	default:
+		cread_add_char(ichar, cls->insert, &cls->num, &cls->eol_num,
+			       buf, cls->len);
+		break;
+	}
+
+	/*
+	 * keep the string terminated...if we added a char at the end then we
+	 * want a \0 after it
+	 */
+	buf[cls->eol_num] = '\0';
+
+	return -EAGAIN;
+}
+
+void cli_cread_init(struct cli_line_state *cls, char *buf, uint buf_size)
+{
+	int init_len = strlen(buf);
+
+	memset(cls, '\0', sizeof(struct cli_line_state));
+	cls->insert = true;
+	cls->buf = buf;
+	cls->len = buf_size;
+
+	if (init_len)
+		cread_add_str(buf, init_len, 0, &cls->num, &cls->eol_num, buf,
+			      buf_size);
+}
diff --git a/common/cli_readline.c b/common/cli_readline.c
index 06b8d4650447..ebbc979af6a9 100644
--- a/common/cli_readline.c
+++ b/common/cli_readline.c
@@ -60,383 +60,7 @@ static char *delete_char (char *buffer, char *p, int *colp, int *np, int plen)
  * Author: Janghoon Lyu <nandy@mizi.com>
  */
 
-#define putnstr(str, n)	printf("%.*s", (int)n, str)
-
-#define CTL_BACKSPACE		('\b')
-#define DEL			((char)255)
-#define DEL7			((char)127)
-#define CREAD_HIST_CHAR		('!')
-
-#define getcmd_putch(ch)	putc(ch)
 #define getcmd_getch()		getchar()
-#define getcmd_cbeep()		getcmd_putch('\a')
-
-#ifdef CONFIG_SPL_BUILD
-#define HIST_MAX		3
-#define HIST_SIZE		32
-#else
-#define HIST_MAX		20
-#define HIST_SIZE		CONFIG_SYS_CBSIZE
-#endif
-
-static int hist_max;
-static int hist_add_idx;
-static int hist_cur = -1;
-static unsigned hist_num;
-
-static char *hist_list[HIST_MAX];
-static char hist_lines[HIST_MAX][HIST_SIZE + 1];	/* Save room for NULL */
-
-#define add_idx_minus_one() ((hist_add_idx == 0) ? hist_max : hist_add_idx-1)
-
-static void getcmd_putchars(int count, int ch)
-{
-	int i;
-
-	for (i = 0; i < count; i++)
-		getcmd_putch(ch);
-}
-
-static void hist_init(void)
-{
-	int i;
-
-	hist_max = 0;
-	hist_add_idx = 0;
-	hist_cur = -1;
-	hist_num = 0;
-
-	for (i = 0; i < HIST_MAX; i++) {
-		hist_list[i] = hist_lines[i];
-		hist_list[i][0] = '\0';
-	}
-}
-
-static void cread_add_to_hist(char *line)
-{
-	strcpy(hist_list[hist_add_idx], line);
-
-	if (++hist_add_idx >= HIST_MAX)
-		hist_add_idx = 0;
-
-	if (hist_add_idx > hist_max)
-		hist_max = hist_add_idx;
-
-	hist_num++;
-}
-
-static char *hist_prev(void)
-{
-	char *ret;
-	int old_cur;
-
-	if (hist_cur < 0)
-		return NULL;
-
-	old_cur = hist_cur;
-	if (--hist_cur < 0)
-		hist_cur = hist_max;
-
-	if (hist_cur == hist_add_idx) {
-		hist_cur = old_cur;
-		ret = NULL;
-	} else {
-		ret = hist_list[hist_cur];
-	}
-
-	return ret;
-}
-
-static char *hist_next(void)
-{
-	char *ret;
-
-	if (hist_cur < 0)
-		return NULL;
-
-	if (hist_cur == hist_add_idx)
-		return NULL;
-
-	if (++hist_cur > hist_max)
-		hist_cur = 0;
-
-	if (hist_cur == hist_add_idx)
-		ret = "";
-	else
-		ret = hist_list[hist_cur];
-
-	return ret;
-}
-
-void cread_print_hist_list(void)
-{
-	int i;
-	uint n;
-
-	n = hist_num - hist_max;
-
-	i = hist_add_idx + 1;
-	while (1) {
-		if (i > hist_max)
-			i = 0;
-		if (i == hist_add_idx)
-			break;
-		printf("%s\n", hist_list[i]);
-		n++;
-		i++;
-	}
-}
-
-#define BEGINNING_OF_LINE() {			\
-	while (cls->num) {			\
-		getcmd_putch(CTL_BACKSPACE);	\
-		cls->num--;			\
-	}					\
-}
-
-#define ERASE_TO_EOL() {				\
-	if (cls->num < cls->eol_num) {		\
-		printf("%*s", (int)(cls->eol_num - cls->num), ""); \
-		do {					\
-			getcmd_putch(CTL_BACKSPACE);	\
-		} while (--cls->eol_num > cls->num);	\
-	}						\
-}
-
-#define REFRESH_TO_EOL() {				\
-	if (cls->num < cls->eol_num) {			\
-		uint wlen = cls->eol_num - cls->num;	\
-		putnstr(buf + cls->num, wlen);		\
-		cls->num = cls->eol_num;		\
-	}						\
-}
-
-static void cread_add_char(char ichar, int insert, uint *num,
-			   uint *eol_num, char *buf, uint len)
-{
-	uint wlen;
-
-	/* room ??? */
-	if (insert || *num == *eol_num) {
-		if (*eol_num > len - 1) {
-			getcmd_cbeep();
-			return;
-		}
-		(*eol_num)++;
-	}
-
-	if (insert) {
-		wlen = *eol_num - *num;
-		if (wlen > 1)
-			memmove(&buf[*num+1], &buf[*num], wlen-1);
-
-		buf[*num] = ichar;
-		putnstr(buf + *num, wlen);
-		(*num)++;
-		while (--wlen)
-			getcmd_putch(CTL_BACKSPACE);
-	} else {
-		/* echo the character */
-		wlen = 1;
-		buf[*num] = ichar;
-		putnstr(buf + *num, wlen);
-		(*num)++;
-	}
-}
-
-static void cread_add_str(char *str, int strsize, int insert,
-			  uint *num, uint *eol_num, char *buf, uint len)
-{
-	while (strsize--) {
-		cread_add_char(*str, insert, num, eol_num, buf, len);
-		str++;
-	}
-}
-
-int cread_line_process_ch(struct cli_line_state *cls, char ichar)
-{
-	char *buf = cls->buf;
-
-	/* ichar=0x0 when error occurs in U-Boot getc */
-	if (!ichar)
-		return -EAGAIN;
-
-	if (ichar == '\n') {
-		putc('\n');
-		buf[cls->eol_num] = '\0';	/* terminate the string */
-		return 0;
-	}
-
-	switch (ichar) {
-	case CTL_CH('a'):
-		BEGINNING_OF_LINE();
-		break;
-	case CTL_CH('c'):	/* ^C - break */
-		*buf = '\0';	/* discard input */
-		return -EINTR;
-	case CTL_CH('f'):
-		if (cls->num < cls->eol_num) {
-			getcmd_putch(buf[cls->num]);
-			cls->num++;
-		}
-		break;
-	case CTL_CH('b'):
-		if (cls->num) {
-			getcmd_putch(CTL_BACKSPACE);
-			cls->num--;
-		}
-		break;
-	case CTL_CH('d'):
-		if (cls->num < cls->eol_num) {
-			uint wlen;
-
-			wlen = cls->eol_num - cls->num - 1;
-			if (wlen) {
-				memmove(&buf[cls->num], &buf[cls->num + 1],
-					wlen);
-				putnstr(buf + cls->num, wlen);
-			}
-
-			getcmd_putch(' ');
-			do {
-				getcmd_putch(CTL_BACKSPACE);
-			} while (wlen--);
-			cls->eol_num--;
-		}
-		break;
-	case CTL_CH('k'):
-		ERASE_TO_EOL();
-		break;
-	case CTL_CH('e'):
-		REFRESH_TO_EOL();
-		break;
-	case CTL_CH('o'):
-		cls->insert = !cls->insert;
-		break;
-	case CTL_CH('w'):
-		if (cls->num) {
-			uint base, wlen;
-
-			for (base = cls->num - 1;
-			     base >= 0 && buf[base] == ' ';)
-				base--;
-			for (; base > 0 && buf[base - 1] != ' ';)
-				base--;
-
-			/* now delete chars from base to cls->num */
-			wlen = cls->num - base;
-			cls->eol_num -= wlen;
-			memmove(&buf[base], &buf[cls->num],
-				cls->eol_num - base + 1);
-			cls->num = base;
-			getcmd_putchars(wlen, CTL_BACKSPACE);
-			puts(buf + base);
-			getcmd_putchars(wlen, ' ');
-			getcmd_putchars(wlen + cls->eol_num - cls->num,
-					CTL_BACKSPACE);
-		}
-		break;
-	case CTL_CH('x'):
-	case CTL_CH('u'):
-		BEGINNING_OF_LINE();
-		ERASE_TO_EOL();
-		break;
-	case DEL:
-	case DEL7:
-	case 8:
-		if (cls->num) {
-			uint wlen;
-
-			wlen = cls->eol_num - cls->num;
-			cls->num--;
-			memmove(&buf[cls->num], &buf[cls->num + 1], wlen);
-			getcmd_putch(CTL_BACKSPACE);
-			putnstr(buf + cls->num, wlen);
-			getcmd_putch(' ');
-			do {
-				getcmd_putch(CTL_BACKSPACE);
-			} while (wlen--);
-			cls->eol_num--;
-		}
-		break;
-	case CTL_CH('p'):
-	case CTL_CH('n'):
-		if (cls->history) {
-			char *hline;
-
-			if (ichar == CTL_CH('p'))
-				hline = hist_prev();
-			else
-				hline = hist_next();
-
-			if (!hline) {
-				getcmd_cbeep();
-				break;
-			}
-
-			/* nuke the current line */
-			/* first, go home */
-			BEGINNING_OF_LINE();
-
-			/* erase to end of line */
-			ERASE_TO_EOL();
-
-			/* copy new line into place and display */
-			strcpy(buf, hline);
-			cls->eol_num = strlen(buf);
-			REFRESH_TO_EOL();
-			break;
-		}
-		break;
-	case '\t':
-		if (IS_ENABLED(CONFIG_AUTO_COMPLETE) && cls->cmd_complete) {
-			int num2, col;
-
-			/* do not autocomplete when in the middle */
-			if (cls->num < cls->eol_num) {
-				getcmd_cbeep();
-				break;
-			}
-
-			buf[cls->num] = '\0';
-			col = strlen(cls->prompt) + cls->eol_num;
-			num2 = cls->num;
-			if (cmd_auto_complete(cls->prompt, buf, &num2, &col)) {
-				col = num2 - cls->num;
-				cls->num += col;
-				cls->eol_num += col;
-			}
-			break;
-		}
-		fallthrough;
-	default:
-		cread_add_char(ichar, cls->insert, &cls->num, &cls->eol_num,
-			       buf, cls->len);
-		break;
-	}
-
-	/*
-	 * keep the string terminated...if we added a char at the end then we
-	 * want a \0 after it
-	 */
-	buf[cls->eol_num] = '\0';
-
-	return -EAGAIN;
-}
-
-void cli_cread_init(struct cli_line_state *cls, char *buf, uint buf_size)
-{
-	int init_len = strlen(buf);
-
-	memset(cls, '\0', sizeof(struct cli_line_state));
-	cls->insert = true;
-	cls->buf = buf;
-	cls->len = buf_size;
-
-	if (init_len)
-		cread_add_str(buf, init_len, 0, &cls->num, &cls->eol_num, buf,
-			      buf_size);
-}
 
 static int cread_line(const char *const prompt, char *buf, unsigned int *len,
 		      int timeout)
@@ -484,19 +108,13 @@ static int cread_line(const char *const prompt, char *buf, unsigned int *len,
 	}
 	*len = cls->eol_num;
 
-	if (buf[0] && buf[0] != CREAD_HIST_CHAR)
-		cread_add_to_hist(buf);
-	hist_cur = hist_add_idx;
+	cread_add_to_hist(buf);
 
 	return 0;
 }
 
 #else /* !CONFIG_CMDLINE_EDITING */
 
-static inline void hist_init(void)
-{
-}
-
 static int cread_line(const char *const prompt, char *buf, unsigned int *len,
 		      int timeout)
 {
diff --git a/include/cli.h b/include/cli.h
index e183d5613697..bca24c8c565f 100644
--- a/include/cli.h
+++ b/include/cli.h
@@ -280,4 +280,34 @@ void cli_cread_init(struct cli_line_state *cls, char *buf, uint buf_size);
 /** cread_print_hist_list() - Print the command-line history list */
 void cread_print_hist_list(void);
 
+/**
+ * hist_prev() - Get the previous line of history
+ *
+ * Returns: Previous history line, or NULL if none
+ */
+char *hist_prev(void);
+
+/**
+ * hist_next() - Get the next line of history
+ *
+ * Returns: Next history line, or NULL if none
+ */
+char *hist_next(void);
+
+/**
+ * cread_add_to_hist() - Add a line to the history buffer
+ *
+ * The line is added if it is non-empty and doesn't start with CREAD_HIST_CHAR
+ * This also sets hist_cur to hist_add_idx whether or not anything was added
+ *
+ * @line: Line to add
+ */
+void cread_add_to_hist(char *line);
+
+#ifdef CONFIG_CMDLINE_EDITING
+void hist_init(void);
+#else
+static inline void hist_init(void) {}
+#endif
+
 #endif
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 24/32] Add a new Kconfig for command-line entry

[Simon Glass]

At present EXPO requires CMDLINE since it uses cli_readline, which is
enabled by CMDLINE. It should be possible to enter lines of text to an
expo without having CMDLINE available.

Fix this dependency by creating a new Kconfig for cli_readline

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Add new patch with a Kconfig for command-line entry

 boot/Kconfig    | 1 +
 cmd/Kconfig     | 1 +
 common/Kconfig  | 7 +++++++
 common/Makefile | 2 +-
 4 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/boot/Kconfig b/boot/Kconfig
index 2fbe70245ec9..e71de0647bc5 100644
--- a/boot/Kconfig
+++ b/boot/Kconfig
@@ -649,6 +649,7 @@ config EXPO
 	bool "Support for expos - groups of scenes displaying a UI"
 	depends on VIDEO
 	default y if BOOTMETH_VBE
+	select CLI_READLINE
 	help
 	  An expo is a way of presenting and collecting information from the
 	  user. It consists of a collection of 'scenes' of which only one is
diff --git a/cmd/Kconfig b/cmd/Kconfig
index 3b4112d9f319..5cb45f9c025e 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -3,6 +3,7 @@ menu "Command line interface"
 config CMDLINE
 	bool "Support U-Boot commands"
 	default y
+	select CLI_READLINE
 	help
 	  Enable U-Boot's command-line functions. This provides a means
 	  to enter commands into U-Boot for a wide variety of purposes. It
diff --git a/common/Kconfig b/common/Kconfig
index 1ffb055744d9..df5bac646491 100644
--- a/common/Kconfig
+++ b/common/Kconfig
@@ -1164,3 +1164,10 @@ endif
 
 config IO_TRACE
 	bool
+
+config CLI_READLINE
+	bool
+	help
+	  Enables support for reading a line of text from the user, This
+	  feature is used by the command-line interpreter and also by expo,
+	  which needs to read text when textline objects are used.
diff --git a/common/Makefile b/common/Makefile
index e22ced0c507f..a9c18c61895c 100644
--- a/common/Makefile
+++ b/common/Makefile
@@ -38,7 +38,7 @@ obj-$(CONFIG_SPLASH_SOURCE) += splash_source.o
 obj-$(CONFIG_MENU) += menu.o
 obj-$(CONFIG_UPDATE_COMMON) += update.o
 obj-$(CONFIG_USB_KEYBOARD) += usb_kbd.o
-obj-$(CONFIG_CMDLINE_EDITING) += cli_cread.o
+obj-$(CONFIG_CLI_READLINE) += cli_cread.o
 obj-$(CONFIG_CMDLINE) += cli_readline.o cli_simple.o
 
 endif # !CONFIG_SPL_BUILD
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 25/32] Add a new Kconfig for command-line history

[Simon Glass]

At present EXPO requires CMDLINE since it uses cli_readline, which
needs history enabled. It should be possible to enter lines of text
into an expo without having history available.

Fix this dependency by creating a new Kconfig for cmdline history.
Adjust the code to use the correct condiition.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Add new patch with a Kconfig for command-line entry

 cmd/Kconfig        | 15 +++++++++++++--
 common/cli_cread.c | 24 ++++++++++++++----------
 include/cli.h      |  3 ++-
 3 files changed, 29 insertions(+), 13 deletions(-)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 5cb45f9c025e..491737ca8ba7 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -29,8 +29,19 @@ config CMDLINE_EDITING
 	depends on CMDLINE
 	default y
 	help
-	  Enable editing and History functions for interactive command line
-	  input operations
+	  Enable an editing function for interactive, command-line-input
+	  operations. This allows moving the cursor back and forth within
+	  the line, inserting and deleting characters, etc.
+
+config CMDLINE_HISTORY
+	bool "Enable command-line history"
+	depends on CMDLINE_EDITING
+	default y
+	help
+	  Enable a history function for interactive, command-line-input
+	  operations. This maintains a small buffer of previously entered
+	  commands, allowing the user to enter or edit old commands without
+	  having to retype them.
 
 config CMDLINE_PS_SUPPORT
 	bool "Enable support for changing the command prompt string at run-time"
diff --git a/common/cli_cread.c b/common/cli_cread.c
index 19af27303cfc..33c678e89f3d 100644
--- a/common/cli_cread.c
+++ b/common/cli_cread.c
@@ -28,11 +28,12 @@ DECLARE_GLOBAL_DATA_PTR;
 #define CTL_BACKSPACE		('\b')
 #define DEL			((char)255)
 #define DEL7			((char)127)
-#define CREAD_HIST_CHAR		('!')
 
 #define getcmd_putch(ch)	putc(ch)
 #define getcmd_cbeep()		getcmd_putch('\a')
 
+#define CREAD_HIST_CHAR		('!')
+
 #ifdef CONFIG_SPL_BUILD
 #define HIST_MAX		3
 #define HIST_SIZE		32
@@ -41,14 +42,6 @@ DECLARE_GLOBAL_DATA_PTR;
 #define HIST_SIZE		CONFIG_SYS_CBSIZE
 #endif
 
-static int hist_max;
-static int hist_add_idx;
-static int hist_cur = -1;
-static uint hist_num;
-
-static char *hist_list[HIST_MAX];
-static char hist_lines[HIST_MAX][HIST_SIZE + 1];	/* Save room for NULL */
-
 static void getcmd_putchars(int count, int ch)
 {
 	int i;
@@ -57,6 +50,16 @@ static void getcmd_putchars(int count, int ch)
 		getcmd_putch(ch);
 }
 
+#ifdef CONFIG_CMDLINE_HISTORY
+
+static int hist_max;
+static int hist_add_idx;
+static int hist_cur = -1;
+static uint hist_num;
+
+static char *hist_list[HIST_MAX];
+static char hist_lines[HIST_MAX][HIST_SIZE + 1];	/* Save room for NULL */
+
 void hist_init(void)
 {
 	int i;
@@ -150,6 +153,7 @@ void cread_print_hist_list(void)
 		i++;
 	}
 }
+#endif /* CMDLINE_HISTORY */
 
 #define BEGINNING_OF_LINE() {			\
 	while (cls->num) {			\
@@ -325,7 +329,7 @@ int cread_line_process_ch(struct cli_line_state *cls, char ichar)
 		break;
 	case CTL_CH('p'):
 	case CTL_CH('n'):
-		if (cls->history) {
+		if (IS_ENABLED(CONFIG_CMDLINE_HISTORY) && cls->history) {
 			char *hline;
 
 			if (ichar == CTL_CH('p'))
diff --git a/include/cli.h b/include/cli.h
index bca24c8c565f..45a338f709f7 100644
--- a/include/cli.h
+++ b/include/cli.h
@@ -294,6 +294,7 @@ char *hist_prev(void);
  */
 char *hist_next(void);
 
+#ifdef CONFIG_CMDLINE_HISTORY
 /**
  * cread_add_to_hist() - Add a line to the history buffer
  *
@@ -304,10 +305,10 @@ char *hist_next(void);
  */
 void cread_add_to_hist(char *line);
 
-#ifdef CONFIG_CMDLINE_EDITING
 void hist_init(void);
 #else
 static inline void hist_init(void) {}
+static inline void cread_add_to_hist(char *line) {}
 #endif
 
 #endif
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 26/32] sandbox: Disable CONFIG_DISTRO_DEFAULTS

[Simon Glass]

This is not used for sandbox, so drop it. Enable the things that it
controls to avoid dstrastic changes in the config settings for
sandbox builds.

The end result is that these are enabled:

   BOOTMETH_DISTRO
   BOOTSTD_DEFAULTS

and these are disabled:

   USE_BOOTCOMMAND
   BOOTCOMMAND (was "run distro_bootcmd")
   DISTRO_DEFAULTS

Note that the tools-only build has already disabled DISTRO_DEFAULTS
and BOOTSTD_FULL

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 arch/Kconfig                       | 5 +++++
 boot/Kconfig                       | 6 +++---
 configs/sandbox64_defconfig        | 1 -
 configs/sandbox_defconfig          | 1 -
 configs/sandbox_flattree_defconfig | 1 -
 configs/sandbox_noinst_defconfig   | 1 -
 configs/sandbox_spl_defconfig      | 1 -
 configs/sandbox_vpl_defconfig      | 1 -
 8 files changed, 8 insertions(+), 9 deletions(-)

diff --git a/arch/Kconfig b/arch/Kconfig
index 19f2891ba1c5..64d0a5548942 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -208,6 +208,11 @@ config SANDBOX
 	imply PHYSMEM
 	imply GENERATE_ACPI_TABLE
 	imply BINMAN
+	imply SYSRESET_CMD_POWEROFF
+	imply SUPPORT_EXTENSION_SCAN
+	imply BOOTSTD_DEFAULTS if BOOTSTD_FULL && CMDLINE
+	imply BOOTMETH_DISTRO if BOOTSTD_FULL && CMDLINE
+	imply CMD_SYSBOOT if BOOTSTD_FULL
 
 config SH
 	bool "SuperH architecture"
diff --git a/boot/Kconfig b/boot/Kconfig
index e71de0647bc5..bab646bcac34 100644
--- a/boot/Kconfig
+++ b/boot/Kconfig
@@ -433,8 +433,8 @@ config BOOTSTD_FULL
 config BOOTSTD_DEFAULTS
 	bool "Select some common defaults for standard boot"
 	depends on BOOTSTD
-	imply USE_BOOTCOMMAND
-	select BOOT_DEFAULTS
+	imply USE_BOOTCOMMAND if !SANDBOX
+	select BOOT_DEFAULTS if CMDLINE
 	select BOOTMETH_DISTRO
 	help
 	  These are not required but are commonly needed to support a good
@@ -443,7 +443,7 @@ config BOOTSTD_DEFAULTS
 
 config BOOTSTD_BOOTCOMMAND
 	bool "Use bootstd to boot"
-	default y if !DISTRO_DEFAULTS
+	default y if !DISTRO_DEFAULTS && !SANDBOX
 	help
 	  Enable this to select a default boot-command suitable for booting
 	  with standard boot. This can be overridden by the board if needed,
diff --git a/configs/sandbox64_defconfig b/configs/sandbox64_defconfig
index 1a033b22018b..ff895b930170 100644
--- a/configs/sandbox64_defconfig
+++ b/configs/sandbox64_defconfig
@@ -14,7 +14,6 @@ CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_LEGACY_IMAGE_FORMAT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
diff --git a/configs/sandbox_defconfig b/configs/sandbox_defconfig
index 47417cb0391d..5230b81be2c4 100644
--- a/configs/sandbox_defconfig
+++ b/configs/sandbox_defconfig
@@ -14,7 +14,6 @@ CONFIG_FIT_RSASSA_PSS=y
 CONFIG_FIT_CIPHER=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_LEGACY_IMAGE_FORMAT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
diff --git a/configs/sandbox_flattree_defconfig b/configs/sandbox_flattree_defconfig
index 29ae4532c508..8df2a82c521c 100644
--- a/configs/sandbox_flattree_defconfig
+++ b/configs/sandbox_flattree_defconfig
@@ -12,7 +12,6 @@ CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_LEGACY_IMAGE_FORMAT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
diff --git a/configs/sandbox_noinst_defconfig b/configs/sandbox_noinst_defconfig
index d39e54f98d25..263cc013d68e 100644
--- a/configs/sandbox_noinst_defconfig
+++ b/configs/sandbox_noinst_defconfig
@@ -20,7 +20,6 @@ CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_SPL_LOAD_FIT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
diff --git a/configs/sandbox_spl_defconfig b/configs/sandbox_spl_defconfig
index 4a67af2f088f..a84c6d2ba248 100644
--- a/configs/sandbox_spl_defconfig
+++ b/configs/sandbox_spl_defconfig
@@ -20,7 +20,6 @@ CONFIG_FIT=y
 CONFIG_FIT_SIGNATURE=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_SPL_LOAD_FIT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
diff --git a/configs/sandbox_vpl_defconfig b/configs/sandbox_vpl_defconfig
index 8d76f19729b9..60fb1701708a 100644
--- a/configs/sandbox_vpl_defconfig
+++ b/configs/sandbox_vpl_defconfig
@@ -27,7 +27,6 @@ CONFIG_FIT=y
 CONFIG_FIT_VERBOSE=y
 CONFIG_FIT_BEST_MATCH=y
 CONFIG_SPL_LOAD_FIT=y
-CONFIG_DISTRO_DEFAULTS=y
 CONFIG_BOOTSTAGE=y
 CONFIG_BOOTSTAGE_REPORT=y
 CONFIG_BOOTSTAGE_FDT=y
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 27/32] cmd: Make all commands depend on CMDLINE

[Simon Glass]

If this option is disabled, commands should not be available. Convert
the CMDLINE option into a menuconfig and make every command in
cmd/Kconfig depend on it.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v1)

 cmd/Kconfig | 31 +++++++++----------------------
 1 file changed, 9 insertions(+), 22 deletions(-)

diff --git a/cmd/Kconfig b/cmd/Kconfig
index 491737ca8ba7..b6fdbc6ff3b1 100644
--- a/cmd/Kconfig
+++ b/cmd/Kconfig
@@ -1,7 +1,5 @@
-menu "Command line interface"
-
-config CMDLINE
-	bool "Support U-Boot commands"
+menuconfig CMDLINE
+	bool "Command line interface"
 	default y
 	select CLI_READLINE
 	help
@@ -12,9 +10,10 @@ config CMDLINE
 	  Depending on the number of commands enabled, this can add
 	  substantially to the size of U-Boot.
 
+if CMDLINE
+
 config HUSH_PARSER
 	bool "Use hush shell"
-	depends on CMDLINE
 	help
 	  This option enables the "hush" shell (from Busybox) as command line
 	  interpreter, thus enabling powerful command line syntax like
@@ -26,7 +25,6 @@ config HUSH_PARSER
 
 config CMDLINE_EDITING
 	bool "Enable command line editing"
-	depends on CMDLINE
 	default y
 	help
 	  Enable an editing function for interactive, command-line-input
@@ -52,15 +50,13 @@ config CMDLINE_PS_SUPPORT
 
 config AUTO_COMPLETE
 	bool "Enable auto complete using TAB"
-	depends on CMDLINE
 	default y
 	help
 	  Enable auto completion of commands using TAB.
 
 config SYS_LONGHELP
 	bool "Enable long help messages"
-	depends on CMDLINE
-	default y if CMDLINE
+	default y
 	help
 	  Defined when you want long help messages included
 	  Do not set this option when short of memory.
@@ -98,8 +94,7 @@ config SYS_CBSIZE
 
 config SYS_XTRACE
 	bool "Command execution tracer"
-	depends on CMDLINE
-	default y if CMDLINE
+	default y
 	help
 	  This option enables the possiblity to print all commands before
 	  executing them and after all variables are evaluated (similar
@@ -238,7 +233,6 @@ menu "Boot commands"
 
 config CMD_BOOTD
 	bool "bootd"
-	depends on CMDLINE
 	default y
 	help
 	  Run the command stored in the environment "bootcmd", i.e.
@@ -405,7 +399,7 @@ config SYS_BOOTM_LEN
 
 config CMD_BOOTEFI
 	bool "bootefi"
-	depends on EFI_LOADER && CMDLINE
+	depends on EFI_LOADER
 	default y
 	help
 	  Boot an EFI image from memory.
@@ -437,7 +431,6 @@ source lib/efi_selftest/Kconfig
 
 config CMD_BOOTMENU
 	bool "bootmenu"
-	depends on CMDLINE
 	select MENU
 	select CHARSET
 	help
@@ -504,7 +497,6 @@ config CMD_GO
 
 config CMD_RUN
 	bool "run"
-	depends on CMDLINE
 	default y
 	help
 	  Run the command in the given environment variable.
@@ -595,7 +587,6 @@ menu "Environment commands"
 
 config CMD_ASKENV
 	bool "ask for env variable"
-	depends on CMDLINE
 	help
 	  Ask for environment variable
 
@@ -1715,7 +1706,6 @@ if NET
 
 menuconfig CMD_NET
 	bool "Network commands"
-	depends on CMDLINE
 	default y
 	imply NETDEVICES
 
@@ -2019,7 +2009,6 @@ config CMD_ETHSW
 
 config CMD_PXE
 	bool "pxe"
-	depends on CMDLINE
 	select PXE_UTILS
 	help
 	  Boot image via network using PXE protocol
@@ -2154,7 +2143,6 @@ config CMD_EFICONFIG
 
 config CMD_EXCEPTION
 	bool "exception - raise exception"
-	depends on CMDLINE
 	depends on ARM || RISCV || SANDBOX || X86
 	help
 	  Enable the 'exception' command which allows to raise an exception.
@@ -2255,14 +2243,12 @@ config CMD_SOUND
 
 config CMD_SYSBOOT
 	bool "sysboot"
-	depends on CMDLINE
 	select PXE_UTILS
 	help
 	  Boot image via local extlinux.conf file
 
 config CMD_QFW
 	bool "qfw"
-	depends on CMDLINE
 	select QFW
 	help
 	  This provides access to the QEMU firmware interface.  The main
@@ -2911,4 +2897,5 @@ config CMD_MESON
 	default y
 	help
 	  Enable useful commands for the Meson Soc family developed by Amlogic Inc.
-endmenu
+
+endif # CMDLINE
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 28/32] sandbox: Avoid requiring cmdline

[Simon Glass]

Use 'imply' rather than 'select' for command-related options, so that
it is possible to build sandbox without CONFIG_CMDLINE enabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Reorder the Kconfig options a little

 arch/Kconfig | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/Kconfig b/arch/Kconfig
index 64d0a5548942..e85134e9346c 100644
--- a/arch/Kconfig
+++ b/arch/Kconfig
@@ -134,7 +134,6 @@ config SANDBOX
 	select ARCH_SUPPORTS_LTO
 	select BOARD_LATE_INIT
 	select BZIP2
-	select CMD_POWEROFF
 	select DM
 	select DM_EVENT
 	select DM_FUZZING_ENGINE
@@ -152,10 +151,8 @@ config SANDBOX
 	select PCI_ENDPOINT
 	select SPI
 	select SUPPORT_OF_CONTROL
-	select SYSRESET_CMD_POWEROFF
 	select SYS_CACHE_SHIFT_4
 	select IRQ
-	select SUPPORT_EXTENSION_SCAN
 	select SUPPORT_ACPI
 	imply BITREVERSE
 	select BLOBLIST
@@ -167,6 +164,9 @@ config SANDBOX
 	imply CMD_IO
 	imply CMD_IOTRACE
 	imply CMD_LZMADEC
+	imply CMD_POWEROFF
+	imply SYSRESET_CMD_POWEROFF
+	imply SUPPORT_EXTENSION_SCAN
 	imply CMD_SF
 	imply CMD_SF_TEST
 	imply CRC32_VERIFY
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 29/32] arm: x86: Drop discarding of command linker-lists

[Simon Glass]

Since we can now cleanly disable CMDLINE when needed, drop the rules
which discard the command code.  It will not be built in the first
place.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v2)

Changes in v2:
- Add new patch to drop discarding of command linker-lists

 arch/arm/cpu/u-boot.lds     | 3 ---
 arch/x86/cpu/u-boot-64.lds  | 4 ----
 arch/x86/cpu/u-boot-spl.lds | 4 ----
 arch/x86/cpu/u-boot.lds     | 4 ----
 4 files changed, 15 deletions(-)

diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds
index fc4f63d83489..7724c9332c3b 100644
--- a/arch/arm/cpu/u-boot.lds
+++ b/arch/arm/cpu/u-boot.lds
@@ -14,9 +14,6 @@ OUTPUT_ARCH(arm)
 ENTRY(_start)
 SECTIONS
 {
-#ifndef CONFIG_CMDLINE
-	/DISCARD/ : { *(__u_boot_list_2_cmd_*) }
-#endif
 #if defined(CONFIG_ARMV7_SECURE_BASE) && defined(CONFIG_ARMV7_NONSEC)
 	/*
 	 * If CONFIG_ARMV7_SECURE_BASE is true, secure code will not
diff --git a/arch/x86/cpu/u-boot-64.lds b/arch/x86/cpu/u-boot-64.lds
index d0398ff00d71..00a6d8691211 100644
--- a/arch/x86/cpu/u-boot-64.lds
+++ b/arch/x86/cpu/u-boot-64.lds
@@ -11,10 +11,6 @@ ENTRY(_start)
 
 SECTIONS
 {
-#ifndef CONFIG_CMDLINE
-	/DISCARD/ : { *(__u_boot_list_2_cmd_*) }
-#endif
-
 #ifdef CONFIG_TEXT_BASE
 	. = CONFIG_TEXT_BASE;	/* Location of bootcode in flash */
 #endif
diff --git a/arch/x86/cpu/u-boot-spl.lds b/arch/x86/cpu/u-boot-spl.lds
index a0a2a06a18cd..50b4b1608552 100644
--- a/arch/x86/cpu/u-boot-spl.lds
+++ b/arch/x86/cpu/u-boot-spl.lds
@@ -11,10 +11,6 @@ ENTRY(_start)
 
 SECTIONS
 {
-#ifndef CONFIG_CMDLINE
-	/DISCARD/ : { *(__u_boot_list_2_cmd_*) }
-#endif
-
 	. = IMAGE_TEXT_BASE;	/* Location of bootcode in flash */
 	__text_start = .;
 	.text  : {
diff --git a/arch/x86/cpu/u-boot.lds b/arch/x86/cpu/u-boot.lds
index a31f4220a000..c418ff44aa08 100644
--- a/arch/x86/cpu/u-boot.lds
+++ b/arch/x86/cpu/u-boot.lds
@@ -11,10 +11,6 @@ ENTRY(_start)
 
 SECTIONS
 {
-#ifndef CONFIG_CMDLINE
-	/DISCARD/ : { *(__u_boot_list_2_cmd_*) }
-#endif
-
 	. = CONFIG_TEXT_BASE;	/* Location of bootcode in flash */
 	__text_start = .;
 
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 30/32] mmc: env: Unify the U_BOOT_ENV_LOCATION conditions

[Simon Glass]

The declaration of U_BOOT_ENV_LOCATION() uses a different #ifdef
condition from the code it calls. Use the same condition to avoid a
build warning if CONFIG_CMD_SAVEENV is disabled.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

(no changes since v2)

Changes in v2:
- Add new patch to unify the U_BOOT_ENV_LOCATION conditions

 env/mmc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/env/mmc.c b/env/mmc.c
index cb14bbb58f13..da84cddd74f0 100644
--- a/env/mmc.c
+++ b/env/mmc.c
@@ -495,7 +495,7 @@ U_BOOT_ENV_LOCATION(mmc) = {
 	.location	= ENVL_MMC,
 	ENV_NAME("MMC")
 	.load		= env_mmc_load,
-#ifndef CONFIG_SPL_BUILD
+#if defined(CONFIG_CMD_SAVEENV) && !defined(CONFIG_SPL_BUILD)
 	.save		= env_save_ptr(env_mmc_save),
 	.erase		= ENV_ERASE_PTR(env_mmc_erase)
 #endif
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 31/32] treewide: Tidy up semicolon after command macros

[Simon Glass]

The U_BOOT_CMD_COMPLETE() macro has a semicolon at the end, perhaps
inadvertently. Some code has taken advantage of this.

Tidy this up by dropping the semicolon from the macro and adding it to
macro invocations as required.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Add new patch to tidy up semicolon after command macros

 board/freescale/common/vid.c | 2 +-
 board/xilinx/common/fru.c    | 2 +-
 board/xilinx/versal/cmds.c   | 2 +-
 board/xilinx/zynqmp/cmds.c   | 2 +-
 cmd/btrfs.c                  | 2 +-
 cmd/eeprom.c                 | 2 +-
 cmd/ext2.c                   | 4 ++--
 cmd/fs.c                     | 8 ++++----
 cmd/pinmux.c                 | 2 +-
 cmd/qfw.c                    | 2 +-
 include/command.h            | 2 +-
 11 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/board/freescale/common/vid.c b/board/freescale/common/vid.c
index 5ec3f2a76b19..fc5d400cfe18 100644
--- a/board/freescale/common/vid.c
+++ b/board/freescale/common/vid.c
@@ -793,4 +793,4 @@ U_BOOT_CMD(
 	vdd_read, 1, 0, do_vdd_read,
 	"read VDD",
 	" - Read the voltage specified in mV"
-)
+);
diff --git a/board/xilinx/common/fru.c b/board/xilinx/common/fru.c
index c916c3d6b4c8..12b21317496a 100644
--- a/board/xilinx/common/fru.c
+++ b/board/xilinx/common/fru.c
@@ -85,4 +85,4 @@ U_BOOT_CMD(
 	fru, 8, 1, do_fru,
 	"FRU table info",
 	fru_help_text
-)
+);
diff --git a/board/xilinx/versal/cmds.c b/board/xilinx/versal/cmds.c
index 9cc2cdcebf1c..2a74e49aedec 100644
--- a/board/xilinx/versal/cmds.c
+++ b/board/xilinx/versal/cmds.c
@@ -98,4 +98,4 @@ U_BOOT_LONGHELP(versal,
 U_BOOT_CMD(versal, 4, 1, do_versal,
 	   "versal sub-system",
 	   versal_help_text
-)
+);
diff --git a/board/xilinx/zynqmp/cmds.c b/board/xilinx/zynqmp/cmds.c
index f1f3eff501e1..9524688f27d9 100644
--- a/board/xilinx/zynqmp/cmds.c
+++ b/board/xilinx/zynqmp/cmds.c
@@ -427,4 +427,4 @@ U_BOOT_CMD(
 	zynqmp, 9, 1, do_zynqmp,
 	"ZynqMP sub-system",
 	zynqmp_help_text
-)
+);
diff --git a/cmd/btrfs.c b/cmd/btrfs.c
index 98daea99e9ed..2843835d08b8 100644
--- a/cmd/btrfs.c
+++ b/cmd/btrfs.c
@@ -24,4 +24,4 @@ U_BOOT_CMD(btrsubvol, 3, 1, do_btrsubvol,
 	"list subvolumes of a BTRFS filesystem",
 	"<interface> <dev[:part]>\n"
 	"     - List subvolumes of a BTRFS filesystem."
-)
+);
diff --git a/cmd/eeprom.c b/cmd/eeprom.c
index 0b6ca8c505fb..322765ad02a0 100644
--- a/cmd/eeprom.c
+++ b/cmd/eeprom.c
@@ -435,4 +435,4 @@ U_BOOT_CMD(
 	"The values which can be provided with the -l option are:\n"
 	CONFIG_EEPROM_LAYOUT_HELP_STRING"\n"
 #endif
-)
+);
diff --git a/cmd/ext2.c b/cmd/ext2.c
index 57a99516a6ac..a0ce0cf5796b 100644
--- a/cmd/ext2.c
+++ b/cmd/ext2.c
@@ -42,7 +42,7 @@ U_BOOT_CMD(
 	"list files in a directory (default /)",
 	"<interface> <dev[:part]> [directory]\n"
 	"    - list files from 'dev' on 'interface' in a 'directory'"
-)
+);
 
 U_BOOT_CMD(
 	ext2load,	6,	0,	do_ext2load,
@@ -50,4 +50,4 @@ U_BOOT_CMD(
 	"<interface> [<dev[:part]> [addr [filename [bytes [pos]]]]]\n"
 	"    - load binary file 'filename' from 'dev' on 'interface'\n"
 	"      to address 'addr' from ext2 filesystem."
-)
+);
diff --git a/cmd/fs.c b/cmd/fs.c
index 6044f73af5b4..46cb43dcdb5b 100644
--- a/cmd/fs.c
+++ b/cmd/fs.c
@@ -39,7 +39,7 @@ U_BOOT_CMD(
 	"      If 'bytes' is 0 or omitted, the file is read until the end.\n"
 	"      'pos' gives the file byte position to start reading from.\n"
 	"      If 'pos' is 0 or omitted, the file is read from the start."
-)
+);
 
 static int do_save_wrapper(struct cmd_tbl *cmdtp, int flag, int argc,
 			   char *const argv[])
@@ -56,7 +56,7 @@ U_BOOT_CMD(
 	"      'bytes' gives the size to save in bytes and is mandatory.\n"
 	"      'pos' gives the file byte position to start writing to.\n"
 	"      If 'pos' is 0 or omitted, the file is written from the start."
-)
+);
 
 static int do_ls_wrapper(struct cmd_tbl *cmdtp, int flag, int argc,
 			 char *const argv[])
@@ -70,7 +70,7 @@ U_BOOT_CMD(
 	"<interface> [<dev[:part]> [directory]]\n"
 	"    - List files in directory 'directory' of partition 'part' on\n"
 	"      device type 'interface' instance 'dev'."
-)
+);
 
 static int do_ln_wrapper(struct cmd_tbl *cmdtp, int flag, int argc,
 			 char *const argv[])
@@ -84,7 +84,7 @@ U_BOOT_CMD(
 	"<interface> <dev[:part]> target linkname\n"
 	"    - create a symbolic link to 'target' with the name 'linkname' on\n"
 	"      device type 'interface' instance 'dev'."
-)
+);
 
 static int do_fstype_wrapper(struct cmd_tbl *cmdtp, int flag, int argc,
 			     char *const argv[])
diff --git a/cmd/pinmux.c b/cmd/pinmux.c
index f17cf4110d9f..105f01eaafff 100644
--- a/cmd/pinmux.c
+++ b/cmd/pinmux.c
@@ -178,4 +178,4 @@ U_BOOT_CMD(pinmux, CONFIG_SYS_MAXARGS, 1, do_pinmux,
 	   "list                     - list UCLASS_PINCTRL devices\n"
 	   "pinmux dev [pincontroller-name] - select pin-controller device\n"
 	   "pinmux status [-a | pin-name]   - print pin-controller muxing [for all | for pin-name]\n"
-)
+);
diff --git a/cmd/qfw.c b/cmd/qfw.c
index d6ecfa60d5a7..1b8c775ebf5a 100644
--- a/cmd/qfw.c
+++ b/cmd/qfw.c
@@ -121,4 +121,4 @@ U_BOOT_CMD(
 	"    - list                             : print firmware(s) currently loaded\n"
 	"    - cpus                             : print online cpu number\n"
 	"    - load <kernel addr> <initrd addr> : load kernel and initrd (if any), and setup for zboot\n"
-)
+);
diff --git a/include/command.h b/include/command.h
index 6262365e128f..5bd3ecbe8f91 100644
--- a/include/command.h
+++ b/include/command.h
@@ -390,7 +390,7 @@ int cmd_source_script(ulong addr, const char *fit_uname, const char *confname);
 #define U_BOOT_CMD_COMPLETE(_name, _maxargs, _rep, _cmd, _usage, _help, _comp) \
 	ll_entry_declare(struct cmd_tbl, _name, cmd) =			\
 		U_BOOT_CMD_MKENT_COMPLETE(_name, _maxargs, _rep, _cmd,	\
-						_usage, _help, _comp);
+						_usage, _help, _comp)
 
 #define U_BOOT_CMDREP_COMPLETE(_name, _maxargs, _cmd_rep, _usage,	\
 			       _help, _comp)				\
-- 
2.42.0.655.g421f12c284-goog

[PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Simon Glass]

Now that everything is working, add a test to make sure that this
builds correctly.

Signed-off-by: Simon Glass <sjg@chromium.org>
---

Changes in v3:
- Rebase on Tom's LONGHELP series
- Correct 'of' typo

 test/py/tests/test_sandbox_opts.py | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
 create mode 100644 test/py/tests/test_sandbox_opts.py

diff --git a/test/py/tests/test_sandbox_opts.py b/test/py/tests/test_sandbox_opts.py
new file mode 100644
index 000000000000..91790b3374b4
--- /dev/null
+++ b/test/py/tests/test_sandbox_opts.py
@@ -0,0 +1,20 @@
+# SPDX-License-Identifier: GPL-2.0
+# Copyright 2022 Google LLC
+# Written by Simon Glass <sjg@chromium.org>
+
+import pytest
+
+import u_boot_utils as util
+
+# This is needed for Azure, since the default '..' directory is not writeable
+TMPDIR = '/tmp/test_cmdline'
+
+@pytest.mark.slow
+@pytest.mark.boardspec('sandbox')
+def test_sandbox_cmdline(u_boot_console):
+    """Test building sandbox without CONFIG_CMDLINE"""
+    cons = u_boot_console
+
+    out = util.run_and_log(
+        cons, ['./tools/buildman/buildman', '-m', '--board', 'sandbox',
+               '-a', '~CMDLINE', '-o', TMPDIR])
-- 
2.42.0.655.g421f12c284-goog

Re: [PATCH v3 21/32] efi: Rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR

[AKASHI Takahiro]

Hi Simon,

Thank you for taking my idea here.

On Mon, Oct 16, 2023 at 04:28:12PM -0600, Simon Glass wrote:
> The command should not be used to enable library functionality. Add a
> new BOOTEFI_BOOTMGR Kconfig for that. Adjust the conditions so that the
> same code is built.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> Suggested-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
> ---
> 
> Changes in v3:
> - Add new patch to rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR
> 
>  cmd/Kconfig             | 9 +++++++++
>  lib/efi_loader/Kconfig  | 7 +++----
>  lib/efi_loader/Makefile | 2 +-
>  3 files changed, 13 insertions(+), 5 deletions(-)
> 
> diff --git a/cmd/Kconfig b/cmd/Kconfig
> index 46e484fc08b6..3b4112d9f319 100644
> --- a/cmd/Kconfig
> +++ b/cmd/Kconfig
> @@ -306,6 +306,15 @@ config CMD_BOOTI
>  	help
>  	  Boot an AArch64 Linux Kernel image from memory.
>  
> +config CMD_BOOTEFI_BOOTMGR
> +	bool "UEFI Boot Manager command"
> +	depends on BOOTEFI_BOOTMGR && CMD_BOOTEFI
> +	default y
> +	help
> +	  Select this option if you want to select the UEFI binary to be booted
> +	  via UEFI variables Boot####, BootOrder, and BootNext. This enables the
> +	  'bootefi bootmgr' command.
> +
>  config BOOTM_LINUX

Do you have any intention to put this configuration here?
Otherwise, it should be placed just after CMD_BOOTEFI.

In addition, CMD_EFICONFIG should have a dependency on BOOTEFI_BOOTMGR
rather than CMD_BOOTEFI_BOOTMGR as it is a separate command.

Thanks,
-Takahiro Akashi

>  	bool "Support booting Linux OS images"
>  	depends on CMD_BOOTM || CMD_BOOTZ || CMD_BOOTI
> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> index 621ed5e5b0fb..13cad6342c36 100644
> --- a/lib/efi_loader/Kconfig
> +++ b/lib/efi_loader/Kconfig
> @@ -32,15 +32,14 @@ config EFI_LOADER
>  
>  if EFI_LOADER
>  
> -config CMD_BOOTEFI_BOOTMGR
> +config BOOTEFI_BOOTMGR
>  	bool "UEFI Boot Manager"
> -	depends on CMDLINE
>  	default y
>  	select BOOTMETH_GLOBAL if BOOTSTD
>  	help
>  	  Select this option if you want to select the UEFI binary to be booted
> -	  via UEFI variables Boot####, BootOrder, and BootNext. This enables the
> -	  'bootefi bootmgr' command.
> +	  via UEFI variables Boot####, BootOrder, and BootNext. You should also
> +	  normally enable CMD_BOOTEFI_BOOTMGR so that the command is available.
>  
>  choice
>  	prompt "Store for non-volatile UEFI variables"
> diff --git a/lib/efi_loader/Makefile b/lib/efi_loader/Makefile
> index 8d31fc61c601..0a2cb6e3c476 100644
> --- a/lib/efi_loader/Makefile
> +++ b/lib/efi_loader/Makefile
> @@ -42,7 +42,7 @@ targets += initrddump.o
>  endif
>  
>  obj-$(CONFIG_CMD_BOOTEFI_HELLO) += helloworld_efi.o
> -obj-$(CONFIG_CMD_BOOTEFI_BOOTMGR) += efi_bootmgr.o
> +obj-$(CONFIG_BOOTEFI_BOOTMGR) += efi_bootmgr.o
>  obj-y += efi_boottime.o
>  obj-y += efi_helper.o
>  obj-$(CONFIG_EFI_HAVE_CAPSULE_SUPPORT) += efi_capsule.o
> -- 
> 2.42.0.655.g421f12c284-goog
> 

Re: [PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 916 bytes --]

On Mon, Oct 16, 2023 at 04:27:53PM -0600, Simon Glass wrote:

> With recent changes over the last few years in boot/Kconfig it is
> no-longer possible to disable CMDLINE. It results in various link
> errors because some options which require CMDLINE are enabled
> regardless of whether it is available.
> 
> Add the necessary conditions to fix this.
> 
> Note that it would be better to have all commands depend on CMDLINE,
> but that is extremely difficult at present, since some functions use
> CMD_xxx to enable feature xxx. For example networking and environment
> have a number of problems to tease apart.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>

Since you later rework things to tease apart most, if not all of this,
please split out the patches which tease things apart in to their own
series so that the rest of this becomes reviewable and also reasonable
looking.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 500 bytes --]

On Mon, Oct 16, 2023 at 04:27:55PM -0600, Simon Glass wrote:

> Add this to some more commands to avoid build errors with sandbox.
> 
> Note that this is a temporary solution to expose more problems. A later
> patch puts these behind an 'if CMDLINE'
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>

We're doing a bunch of churn here that I'd rather just get done at once,
make the current #27 which guards all commands with if CMDLINE happen
earlier in the series instead.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 09/32] cmd: Use an #ifdef around run_commandf()

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 429 bytes --]

On Mon, Oct 16, 2023 at 04:28:00PM -0600, Simon Glass wrote:

> This is not available if CMDLINE is disabled, so add an #ifdef to correct
> this.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v1)
> 
>  common/cli.c | 2 ++
>  1 file changed, 2 insertions(+)

NAK, please tease this file apart in to CMDLINE code, and whatever
inadvertent utility code is also in there.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 12/32] env: Split out non-command code into a new file

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1192 bytes --]

On Mon, Oct 16, 2023 at 04:28:03PM -0600, Simon Glass wrote:

> It is not possible to set environment variables without having
> CONFIG_CMD_NVEDIT enabled. When CONFIG_CMDLINE is disabled, we need a
> way to set variables.
> 
> Split the setting code out into its own file, so that env_set() is
> available even when CONFIG_CMDLINE is not. If it is never called, the
> code will be dropped at link time.
> 
> Update the Makefile rule to only include the env commands when
> CONFIG_CMD_NVEDIT is enabled.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v1)
> 
>  cmd/Makefile           |   2 +-
>  cmd/nvedit.c           | 122 ++-----------------------------------
>  env/Makefile           |   1 +
>  env/env_set.c          | 132 +++++++++++++++++++++++++++++++++++++++++
>  include/env_internal.h |  23 +++++++
>  5 files changed, 161 insertions(+), 119 deletions(-)
>  create mode 100644 env/env_set.c

This feels like it's partly but not entirely correct.  We need to split
cmd/nvedit.c in to cmd/env.c and env/nvedit.c instead I think.  And keep
in mind that SPL + ENV (and so - CMDLINE I believe) does work today.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 506 bytes --]

On Mon, Oct 16, 2023 at 04:28:05PM -0600, Simon Glass wrote:

> There are two cleanup functions needed during boot which depend on
> CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()
> 
> Provide static inline versions of these for when commands are not
> enabled.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>

NAK, these functions need to be available to boot the OS, regardless of
command line existing or not.  Unwind things in the other direction
please.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 16/32] net: Depend on CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1065 bytes --]

On Mon, Oct 16, 2023 at 04:28:07PM -0600, Simon Glass wrote:

> At present it isn't possible to use networking without the command line
> enabled. Add this as a condition.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
> ---
> 
> (no changes since v1)
> 
>  cmd/Kconfig | 1 +
>  net/Kconfig | 1 +
>  2 files changed, 2 insertions(+)
> 
> diff --git a/cmd/Kconfig b/cmd/Kconfig
> index 148414f011df..c6ea5c860e33 100644
> --- a/cmd/Kconfig
> +++ b/cmd/Kconfig
> @@ -1694,6 +1694,7 @@ if NET
>  
>  menuconfig CMD_NET
>  	bool "Network commands"
> +	depends on CMDLINE
>  	default y
>  	imply NETDEVICES
>  
> diff --git a/net/Kconfig b/net/Kconfig
> index 4215889127c9..25d494e1db46 100644
> --- a/net/Kconfig
> +++ b/net/Kconfig
> @@ -4,6 +4,7 @@
>  
>  menuconfig NET
>  	bool "Networking support"
> +	depends on CMDLINE
>  	default y
>  
>  if NET

Please unwind this in the other direction.  What in network is depending
on cmdline in a non-trivial way?

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 17/32] log: Allow use without CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 330 bytes --]

On Mon, Oct 16, 2023 at 04:28:08PM -0600, Simon Glass wrote:

> When CONFIG_SYS_CBSIZE is not used we need an alternative. For logging
> it seems that CONFIG_SYS_PBSIZE is a better choice anyway, so update
> this.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>

Reviewed-by: Tom Rini <trini@konsulko.com>

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 18/32] video: Allow use without CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1230 bytes --]

On Mon, Oct 16, 2023 at 04:28:09PM -0600, Simon Glass wrote:

> Provide a fallback for when CONFIG_SYS_CBSIZE is not provided, so that
> the console can still be used.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v1)
> 
>  drivers/video/console_truetype.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/video/console_truetype.c b/drivers/video/console_truetype.c
> index 14fb81e9563c..e4dad3f9a191 100644
> --- a/drivers/video/console_truetype.c
> +++ b/drivers/video/console_truetype.c
> @@ -125,7 +125,11 @@ struct pos_info {
>   * Allow one for each character on the command line plus one for each newline.
>   * This is just an estimate, but it should not be exceeded.
>   */
> +#ifdef CONFIG_SYS_CBSIZE
>  #define POS_HISTORY_SIZE	(CONFIG_SYS_CBSIZE * 11 / 10)
> +#else
> +#define POS_HISTORY_SIZE	(250 * 11 / 10)
> +#endif
>  
>  /**
>   * struct console_tt_metrics - Information about a font / size combination

NAK, this either should be SYS_PBSIZE (output buffer) instead which you
move out from under CMDLINE or SYS_CBSIZE (input buffer) should also be
outside, and both get a help text to explain them a bit better.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 19/32] video: Dont require the font command

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1010 bytes --]

On Mon, Oct 16, 2023 at 04:28:10PM -0600, Simon Glass wrote:
> While it is nice to have the font command, using 'select' makes it
> impossible to build the console code without it. Change this to use
> 'imply' instead.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v1)
> 
>  drivers/video/Kconfig | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
> index ab927641bb7a..21ea5c860cca 100644
> --- a/drivers/video/Kconfig
> +++ b/drivers/video/Kconfig
> @@ -180,7 +180,7 @@ config CONSOLE_ROTATION
>  
>  config CONSOLE_TRUETYPE
>  	bool "Support a console that uses TrueType fonts"
> -	select CMD_SELECT_FONT
> +	imply CMD_SELECT_FONT
>  	help
>  	  TrueTrype fonts can provide outline-drawing capability rather than
>  	  needing to provide a bitmap for each font and size that is needed.

This is one of those cases where "if CMDLINE" makes sense to add
somewhere.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1068 bytes --]

On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:

> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> available, add it as an explicit dependency.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v2)
> 
> Changes in v2:
> - Add new patch to update EFI_LOADER to depend on DM_ETH
> 
>  lib/efi_loader/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> index 13cad6342c36..fca4b3eef270 100644
> --- a/lib/efi_loader/Kconfig
> +++ b/lib/efi_loader/Kconfig
> @@ -11,6 +11,7 @@ config EFI_LOADER
>  	# We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
>  	depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
>  	depends on BLK
> +	depends on DM_ETH
>  	depends on !EFI_APP
>  	default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
>  	select CHARSET

Does this work for you Heinrich, or do you want to clarify the
dependencies (and re-organize the code as needed) around networking?

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 265 bytes --]

On Mon, Oct 16, 2023 at 04:28:23PM -0600, Simon Glass wrote:

> Now that everything is working, add a test to make sure that this
> builds correctly.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>

Reviewed-by: Tom Rini <trini@konsulko.com>

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 21/32] efi: Rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 948 bytes --]

On Mon, Oct 16, 2023 at 04:28:12PM -0600, Simon Glass wrote:
> The command should not be used to enable library functionality. Add a
> new BOOTEFI_BOOTMGR Kconfig for that. Adjust the conditions so that the
> same code is built.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> Suggested-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
[snip]
> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> index 621ed5e5b0fb..13cad6342c36 100644
> --- a/lib/efi_loader/Kconfig
> +++ b/lib/efi_loader/Kconfig
> @@ -32,15 +32,14 @@ config EFI_LOADER
>  
>  if EFI_LOADER
>  
> -config CMD_BOOTEFI_BOOTMGR
> +config BOOTEFI_BOOTMGR
>  	bool "UEFI Boot Manager"
> -	depends on CMDLINE

This is another example of why I'm asking for re-ordering things so that
first you clean / re-order things then you make all of CMD depend on
CMDLINE.  This patch, aside from other feedback, is standalone, if you
do that.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 07:32, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:27:53PM -0600, Simon Glass wrote:
>
> > With recent changes over the last few years in boot/Kconfig it is
> > no-longer possible to disable CMDLINE. It results in various link
> > errors because some options which require CMDLINE are enabled
> > regardless of whether it is available.
> >
> > Add the necessary conditions to fix this.
> >
> > Note that it would be better to have all commands depend on CMDLINE,
> > but that is extremely difficult at present, since some functions use
> > CMD_xxx to enable feature xxx. For example networking and environment
> > have a number of problems to tease apart.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
>
> Since you later rework things to tease apart most, if not all of this,
> please split out the patches which tease things apart in to their own
> series so that the rest of this becomes reviewable and also reasonable
> looking.

At this point I fear I have lost track of things, since there is just
too much going on. Just in this version I have dealt with:

- some EFI stuff (which needs tweaking)
- teasing apart the CLI handling (which apparently doesn't go far enough?)

Before this version I already looked at environment (which you say
doesn't go far enough, and fair enough, but please review the code as
sent, not what you wish it was).

Perhaps you could apply whatever seems reasonable and then I can take
another look? Or suggest some small changes that would allow progress
to be made.

But do note that supporting booting without CONFIG_CMDLINE is a large
effort, the subject of 5-10 series, perhaps more. It is not my
intention to undertake that in this series, or before this series
lands

The problem, also, is that without this series, it isn't possible to
start that work. With this series, one can look at what code is
missing and find a way to call boot functions (for example) bypassing
the command line. Once you get it booting you are good. Without this
series, you don't even know where to look, since calls to the command
line happen silently.

So I think this series is a big step forward and urge another look, please.

Regards,
Simon

Re: [PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 07:32, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:27:55PM -0600, Simon Glass wrote:
>
> > Add this to some more commands to avoid build errors with sandbox.
> >
> > Note that this is a temporary solution to expose more problems. A later
> > patch puts these behind an 'if CMDLINE'
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
>
> We're doing a bunch of churn here that I'd rather just get done at once,
> make the current #27 which guards all commands with if CMDLINE happen
> earlier in the series instead.

Yes I did consider this when reviewing this patch for v3, but got
build-bisect failures. I cannot move #27 earlier, since all the builds
fail. Allowing #27 to pass is the point of this series, in fact.

If we can make progress on the thrust of this series then I will see
if I can somehow reorder some other patches to drop this patch, or
move it later, or cut it down, or something else...

I also don't agree with 'bunch of'. This is 6 lines, which enables
another 20 patches which clean everything up. Are there other patches
with churn?

Regards,
Simon

Re: [PATCH v3 12/32] env: Split out non-command code into a new file

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 07:40, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:03PM -0600, Simon Glass wrote:
>
> > It is not possible to set environment variables without having
> > CONFIG_CMD_NVEDIT enabled. When CONFIG_CMDLINE is disabled, we need a
> > way to set variables.
> >
> > Split the setting code out into its own file, so that env_set() is
> > available even when CONFIG_CMDLINE is not. If it is never called, the
> > code will be dropped at link time.
> >
> > Update the Makefile rule to only include the env commands when
> > CONFIG_CMD_NVEDIT is enabled.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > (no changes since v1)
> >
> >  cmd/Makefile           |   2 +-
> >  cmd/nvedit.c           | 122 ++-----------------------------------
> >  env/Makefile           |   1 +
> >  env/env_set.c          | 132 +++++++++++++++++++++++++++++++++++++++++
> >  include/env_internal.h |  23 +++++++
> >  5 files changed, 161 insertions(+), 119 deletions(-)
> >  create mode 100644 env/env_set.c
>
> This feels like it's partly but not entirely correct.  We need to split
> cmd/nvedit.c in to cmd/env.c and env/nvedit.c instead I think.  And keep
> in mind that SPL + ENV (and so - CMDLINE I believe) does work today.

OK, fine, but can I do it later? That seems like an excellent
follow-on but is it essential for the purpose of this already-large
series?

Regards,
Simon

Re: [PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 08:02, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:05PM -0600, Simon Glass wrote:
>
> > There are two cleanup functions needed during boot which depend on
> > CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()
> >
> > Provide static inline versions of these for when commands are not
> > enabled.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
>
> NAK, these functions need to be available to boot the OS, regardless of
> command line existing or not.  Unwind things in the other direction
> please.

That won't be a successful strategy. See my other reply on this. I
have had this idea since 2016 and have looked at it every now and
then. With bootstd it makes more sense since we actually have a
framework to boot without the command line. This series enables such
work, but cannot include it.

Regards,
Simon

Re: [PATCH v3 16/32] net: Depend on CONFIG_CMDLINE

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 08:04, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:07PM -0600, Simon Glass wrote:
>
> > At present it isn't possible to use networking without the command line
> > enabled. Add this as a condition.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > Reviewed-by: Ramon Fried <rfried.dev@gmail.com>
> > ---
> >
> > (no changes since v1)
> >
> >  cmd/Kconfig | 1 +
> >  net/Kconfig | 1 +
> >  2 files changed, 2 insertions(+)
> >
> > diff --git a/cmd/Kconfig b/cmd/Kconfig
> > index 148414f011df..c6ea5c860e33 100644
> > --- a/cmd/Kconfig
> > +++ b/cmd/Kconfig
> > @@ -1694,6 +1694,7 @@ if NET
> >
> >  menuconfig CMD_NET
> >       bool "Network commands"
> > +     depends on CMDLINE
> >       default y
> >       imply NETDEVICES
> >
> > diff --git a/net/Kconfig b/net/Kconfig
> > index 4215889127c9..25d494e1db46 100644
> > --- a/net/Kconfig
> > +++ b/net/Kconfig
> > @@ -4,6 +4,7 @@
> >
> >  menuconfig NET
> >       bool "Networking support"
> > +     depends on CMDLINE
> >       default y
> >
> >  if NET
>
> Please unwind this in the other direction.  What in network is depending
> on cmdline in a non-trivial way?

I can look at this as a follow-on, but I need a starting point. See my
other comment.

Regards,
Simon

Re: [PATCH v3 18/32] video: Allow use without CONFIG_CMDLINE

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 08:07, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:09PM -0600, Simon Glass wrote:
>
> > Provide a fallback for when CONFIG_SYS_CBSIZE is not provided, so that
> > the console can still be used.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > (no changes since v1)
> >
> >  drivers/video/console_truetype.c | 4 ++++
> >  1 file changed, 4 insertions(+)
> >
> > diff --git a/drivers/video/console_truetype.c b/drivers/video/console_truetype.c
> > index 14fb81e9563c..e4dad3f9a191 100644
> > --- a/drivers/video/console_truetype.c
> > +++ b/drivers/video/console_truetype.c
> > @@ -125,7 +125,11 @@ struct pos_info {
> >   * Allow one for each character on the command line plus one for each newline.
> >   * This is just an estimate, but it should not be exceeded.
> >   */
> > +#ifdef CONFIG_SYS_CBSIZE
> >  #define POS_HISTORY_SIZE     (CONFIG_SYS_CBSIZE * 11 / 10)
> > +#else
> > +#define POS_HISTORY_SIZE     (250 * 11 / 10)
> > +#endif
> >
> >  /**
> >   * struct console_tt_metrics - Information about a font / size combination
>
> NAK, this either should be SYS_PBSIZE (output buffer) instead which you
> move out from under CMDLINE or SYS_CBSIZE (input buffer) should also be
> outside, and both get a help text to explain them a bit better.

I think PBSIZE would be better. Even better (eventually) will be to
drop this truetype buffer if input is not needed.

Regards,
Simon

Re: [PATCH v3 19/32] video: Dont require the font command

[Simon Glass]

Hi Tom,

On Tue, 17 Oct 2023 at 08:07, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:10PM -0600, Simon Glass wrote:
> > While it is nice to have the font command, using 'select' makes it
> > impossible to build the console code without it. Change this to use
> > 'imply' instead.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > (no changes since v1)
> >
> >  drivers/video/Kconfig | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
> > index ab927641bb7a..21ea5c860cca 100644
> > --- a/drivers/video/Kconfig
> > +++ b/drivers/video/Kconfig
> > @@ -180,7 +180,7 @@ config CONSOLE_ROTATION
> >
> >  config CONSOLE_TRUETYPE
> >       bool "Support a console that uses TrueType fonts"
> > -     select CMD_SELECT_FONT
> > +     imply CMD_SELECT_FONT
> >       help
> >         TrueTrype fonts can provide outline-drawing capability rather than
> >         needing to provide a bitmap for each font and size that is needed.
>
> This is one of those cases where "if CMDLINE" makes sense to add
> somewhere.

Maybe, if you can explain it a bit more. I want boards to be able to
enable or disable this command, independently of whether truetype
fonts are supported. Using 'select' here seems quite inflexible.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi,

On Tue, 17 Oct 2023 at 08:09, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
>
> > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > available, add it as an explicit dependency.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > (no changes since v2)
> >
> > Changes in v2:
> > - Add new patch to update EFI_LOADER to depend on DM_ETH
> >
> >  lib/efi_loader/Kconfig | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > index 13cad6342c36..fca4b3eef270 100644
> > --- a/lib/efi_loader/Kconfig
> > +++ b/lib/efi_loader/Kconfig
> > @@ -11,6 +11,7 @@ config EFI_LOADER
> >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> >       depends on BLK
> > +     depends on DM_ETH
> >       depends on !EFI_APP
> >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> >       select CHARSET
>
> Does this work for you Heinrich, or do you want to clarify the
> dependencies (and re-organize the code as needed) around networking?

It would be great to tidy up networking in lots of ways...perhaps we
should add ~CMDLINE support to the list of post-lwip tasks if it lands

Regards,
Simon

Re: [PATCH v3 21/32] efi: Rearrange the Kconfig for CMD_BOOTEFI_BOOTMGR

[Simon Glass]

Hi Tom,

On Tue, Oct 17, 2023, 07:13 Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:12PM -0600, Simon Glass wrote:
> > The command should not be used to enable library functionality. Add a
> > new BOOTEFI_BOOTMGR Kconfig for that. Adjust the conditions so that the
> > same code is built.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > Suggested-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
> [snip]
> > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > index 621ed5e5b0fb..13cad6342c36 100644
> > --- a/lib/efi_loader/Kconfig
> > +++ b/lib/efi_loader/Kconfig
> > @@ -32,15 +32,14 @@ config EFI_LOADER
> >
> >  if EFI_LOADER
> >
> > -config CMD_BOOTEFI_BOOTMGR
> > +config BOOTEFI_BOOTMGR
> >       bool "UEFI Boot Manager"
> > -     depends on CMDLINE
>
> This is another example of why I'm asking for re-ordering things so that
> first you clean / re-order things then you make all of CMD depend on
> CMDLINE.  This patch, aside from other feedback, is standalone, if you
> do that.

This patch is before the one that makes all of CMD depend on CMDLINE.

Regards,
Simon

Re: [PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1295 bytes --]

On Tue, Oct 17, 2023 at 09:31:04PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Oct 2023 at 08:02, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:05PM -0600, Simon Glass wrote:
> >
> > > There are two cleanup functions needed during boot which depend on
> > > CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()
> > >
> > > Provide static inline versions of these for when commands are not
> > > enabled.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> >
> > NAK, these functions need to be available to boot the OS, regardless of
> > command line existing or not.  Unwind things in the other direction
> > please.
> 
> That won't be a successful strategy. See my other reply on this. I
> have had this idea since 2016 and have looked at it every now and
> then. With bootstd it makes more sense since we actually have a
> framework to boot without the command line. This series enables such
> work, but cannot include it.

I don't understand.  The "cleanup" functions here are to ensure that any
devices that we've touched are in the state the OS needs them to be.
It's not "we ran a command", it's "we loaded something off of USB (or
network or MMC or ..)" and so we must have those called.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 18/32] video: Allow use without CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1678 bytes --]

On Tue, Oct 17, 2023 at 09:31:22PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Oct 2023 at 08:07, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:09PM -0600, Simon Glass wrote:
> >
> > > Provide a fallback for when CONFIG_SYS_CBSIZE is not provided, so that
> > > the console can still be used.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v1)
> > >
> > >  drivers/video/console_truetype.c | 4 ++++
> > >  1 file changed, 4 insertions(+)
> > >
> > > diff --git a/drivers/video/console_truetype.c b/drivers/video/console_truetype.c
> > > index 14fb81e9563c..e4dad3f9a191 100644
> > > --- a/drivers/video/console_truetype.c
> > > +++ b/drivers/video/console_truetype.c
> > > @@ -125,7 +125,11 @@ struct pos_info {
> > >   * Allow one for each character on the command line plus one for each newline.
> > >   * This is just an estimate, but it should not be exceeded.
> > >   */
> > > +#ifdef CONFIG_SYS_CBSIZE
> > >  #define POS_HISTORY_SIZE     (CONFIG_SYS_CBSIZE * 11 / 10)
> > > +#else
> > > +#define POS_HISTORY_SIZE     (250 * 11 / 10)
> > > +#endif
> > >
> > >  /**
> > >   * struct console_tt_metrics - Information about a font / size combination
> >
> > NAK, this either should be SYS_PBSIZE (output buffer) instead which you
> > move out from under CMDLINE or SYS_CBSIZE (input buffer) should also be
> > outside, and both get a help text to explain them a bit better.
> 
> I think PBSIZE would be better. Even better (eventually) will be to
> drop this truetype buffer if input is not needed.

Then lets switch this to PBSIZE.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 19/32] video: Dont require the font command

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1675 bytes --]

On Tue, Oct 17, 2023 at 09:31:29PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Oct 2023 at 08:07, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:10PM -0600, Simon Glass wrote:
> > > While it is nice to have the font command, using 'select' makes it
> > > impossible to build the console code without it. Change this to use
> > > 'imply' instead.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v1)
> > >
> > >  drivers/video/Kconfig | 2 +-
> > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > >
> > > diff --git a/drivers/video/Kconfig b/drivers/video/Kconfig
> > > index ab927641bb7a..21ea5c860cca 100644
> > > --- a/drivers/video/Kconfig
> > > +++ b/drivers/video/Kconfig
> > > @@ -180,7 +180,7 @@ config CONSOLE_ROTATION
> > >
> > >  config CONSOLE_TRUETYPE
> > >       bool "Support a console that uses TrueType fonts"
> > > -     select CMD_SELECT_FONT
> > > +     imply CMD_SELECT_FONT
> > >       help
> > >         TrueTrype fonts can provide outline-drawing capability rather than
> > >         needing to provide a bitmap for each font and size that is needed.
> >
> > This is one of those cases where "if CMDLINE" makes sense to add
> > somewhere.
> 
> Maybe, if you can explain it a bit more. I want boards to be able to
> enable or disable this command, independently of whether truetype
> fonts are supported. Using 'select' here seems quite inflexible.

Ah, OK.  Checking the command itself, we should drop the select here and
make CMD_SELECT_FONT default y if CONSOLE_TRUETYPE (and drop the default
n line it has).

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1677 bytes --]

On Tue, Oct 17, 2023 at 09:31:34PM -0600, Simon Glass wrote:
> Hi,
> 
> On Tue, 17 Oct 2023 at 08:09, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> >
> > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > available, add it as an explicit dependency.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v2)
> > >
> > > Changes in v2:
> > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > >
> > >  lib/efi_loader/Kconfig | 1 +
> > >  1 file changed, 1 insertion(+)
> > >
> > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > index 13cad6342c36..fca4b3eef270 100644
> > > --- a/lib/efi_loader/Kconfig
> > > +++ b/lib/efi_loader/Kconfig
> > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > >       depends on BLK
> > > +     depends on DM_ETH
> > >       depends on !EFI_APP
> > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > >       select CHARSET
> >
> > Does this work for you Heinrich, or do you want to clarify the
> > dependencies (and re-organize the code as needed) around networking?
> 
> It would be great to tidy up networking in lots of ways...perhaps we
> should add ~CMDLINE support to the list of post-lwip tasks if it lands

But this isn't even directly networking.  It's I believe that the
network related portion of EFI loader isn't optional today.  But perhaps
it could / should be?

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Heinrich Schuchardt]

On 10/17/23 16:09, Tom Rini wrote:
> On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
>
>> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
>> available, add it as an explicit dependency.
>>
>> Signed-off-by: Simon Glass <sjg@chromium.org>
>> ---
>>
>> (no changes since v2)
>>
>> Changes in v2:
>> - Add new patch to update EFI_LOADER to depend on DM_ETH
>>
>>   lib/efi_loader/Kconfig | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
>> index 13cad6342c36..fca4b3eef270 100644
>> --- a/lib/efi_loader/Kconfig
>> +++ b/lib/efi_loader/Kconfig
>> @@ -11,6 +11,7 @@ config EFI_LOADER
>>   	# We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
>>   	depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
>>   	depends on BLK
>> +	depends on DM_ETH
>>   	depends on !EFI_APP
>>   	default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
>>   	select CHARSET
>
> Does this work for you Heinrich, or do you want to clarify the
> dependencies (and re-organize the code as needed) around networking?
>

We should be able to boot via EFI on devices without U-Boot network support.

We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
CONFIG_DM_ETH.

Why is this not sufficient?
Is there a configuration that does not build?

Best regards

Heinrich

Re: [PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1476 bytes --]

On Tue, Oct 17, 2023 at 09:30:46PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Oct 2023 at 07:32, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:27:55PM -0600, Simon Glass wrote:
> >
> > > Add this to some more commands to avoid build errors with sandbox.
> > >
> > > Note that this is a temporary solution to expose more problems. A later
> > > patch puts these behind an 'if CMDLINE'
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> >
> > We're doing a bunch of churn here that I'd rather just get done at once,
> > make the current #27 which guards all commands with if CMDLINE happen
> > earlier in the series instead.
> 
> Yes I did consider this when reviewing this patch for v3, but got
> build-bisect failures. I cannot move #27 earlier, since all the builds
> fail. Allowing #27 to pass is the point of this series, in fact.

Yes, I suspect a number of the other patches need to be shuffled around
as well.

> If we can make progress on the thrust of this series then I will see
> if I can somehow reorder some other patches to drop this patch, or
> move it later, or cut it down, or something else...
> 
> I also don't agree with 'bunch of'. This is 6 lines, which enables
> another 20 patches which clean everything up. Are there other patches
> with churn?

Other patches in the series also added if CMDLINE or similar sets of
dependencies that then get pulled out in #27.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Heinrich Schuchardt]

On 10/17/23 00:27, Simon Glass wrote:
> Add this to some more commands to avoid build errors with sandbox.
>
> Note that this is a temporary solution to expose more problems. A later
> patch puts these behind an 'if CMDLINE'
>
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
>
> Changes in v3:
> - Add an explation as to why this patch is here
>
>   cmd/Kconfig | 6 ++++++
>   1 file changed, 6 insertions(+)
>
> diff --git a/cmd/Kconfig b/cmd/Kconfig
> index 5bc0a92d57ad..00a7f84bce99 100644
> --- a/cmd/Kconfig
> +++ b/cmd/Kconfig
> @@ -231,6 +231,7 @@ menu "Boot commands"
>
>   config CMD_BOOTD
>   	bool "bootd"
> +	depends on CMDLINE

Why don't we simply use a single "if" for all commands?
Adding the same dependency to 200+ commands does not feel right.

If there is any CONFIG_CMD* that is needed without CMDLINE, then that
code should move to lib/ or drivers/.

Best regards

Heinrich

>   	default y
>   	help
>   	  Run the command stored in the environment "bootcmd", i.e.
> @@ -420,6 +421,7 @@ source lib/efi_selftest/Kconfig
>
>   config CMD_BOOTMENU
>   	bool "bootmenu"
> +	depends on CMDLINE
>   	select MENU
>   	select CHARSET
>   	help
> @@ -486,6 +488,7 @@ config CMD_GO
>
>   config CMD_RUN
>   	bool "run"
> +	depends on CMDLINE
>   	default y
>   	help
>   	  Run the command in the given environment variable.
> @@ -576,6 +579,7 @@ menu "Environment commands"
>
>   config CMD_ASKENV
>   	bool "ask for env variable"
> +	depends on CMDLINE
>   	help
>   	  Ask for environment variable
>
> @@ -2132,6 +2136,7 @@ config CMD_EFICONFIG
>
>   config CMD_EXCEPTION
>   	bool "exception - raise exception"
> +	depends on CMDLINE
>   	depends on ARM || RISCV || SANDBOX || X86
>   	help
>   	  Enable the 'exception' command which allows to raise an exception.
> @@ -2238,6 +2243,7 @@ config CMD_SYSBOOT
>
>   config CMD_QFW
>   	bool "qfw"
> +	depends on CMDLINE
>   	select QFW
>   	help
>   	  This provides access to the QEMU firmware interface.  The main

Re: [PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 3938 bytes --]

On Tue, Oct 17, 2023 at 09:30:30PM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Tue, 17 Oct 2023 at 07:32, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:27:53PM -0600, Simon Glass wrote:
> >
> > > With recent changes over the last few years in boot/Kconfig it is
> > > no-longer possible to disable CMDLINE. It results in various link
> > > errors because some options which require CMDLINE are enabled
> > > regardless of whether it is available.
> > >
> > > Add the necessary conditions to fix this.
> > >
> > > Note that it would be better to have all commands depend on CMDLINE,
> > > but that is extremely difficult at present, since some functions use
> > > CMD_xxx to enable feature xxx. For example networking and environment
> > > have a number of problems to tease apart.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> >
> > Since you later rework things to tease apart most, if not all of this,
> > please split out the patches which tease things apart in to their own
> > series so that the rest of this becomes reviewable and also reasonable
> > looking.
> 
> At this point I fear I have lost track of things, since there is just
> too much going on.

Yes, there is too much going on, which is why I wanted you to split this
up further.  I think we both agree that's a problem with this series.

> Just in this version I have dealt with:
> 
> - some EFI stuff (which needs tweaking)
> - teasing apart the CLI handling (which apparently doesn't go far enough?)

I don't think it's fair to call out EFI here, it's fairly well behaved
and no worse / oddly using CONFIG symbols than video or networking or
just about everything else is.

> Before this version I already looked at environment (which you say
> doesn't go far enough, and fair enough, but please review the code as
> sent, not what you wish it was).

Well, that's just it.  I did, and I wasn't entirely sure how you were
splitting things was right.  There was more stuff that either didn't
belong or should also have been moved, it wasn't clear.  The split I'm
suggesting there should make it clear.

> Perhaps you could apply whatever seems reasonable and then I can take
> another look? Or suggest some small changes that would allow progress
> to be made.

I guess I'll see what I can make some sense out of, yes.

> But do note that supporting booting without CONFIG_CMDLINE is a large
> effort, the subject of 5-10 series, perhaps more. It is not my
> intention to undertake that in this series, or before this series
> lands

Well, one of my issues here, especially with the idea of cycling back
and cleaning things up later is that I kind of assume you had intended
to do that in some places already, having done the good and important
task of breaking up the old common directory.

> The problem, also, is that without this series, it isn't possible to
> start that work. With this series, one can look at what code is
> missing and find a way to call boot functions (for example) bypassing
> the command line. Once you get it booting you are good. Without this
> series, you don't even know where to look, since calls to the command
> line happen silently.

That's where I disagree.  Step one is working towards being able to
build with CMDLINE disabled, and doing it in small reviewable chunks.
This is one big series that gives you "can build with CMDLINE disabled",
but while I'm usually not a fan of the Linux kernel's strict "break it
down per subsystem", I'd be a lot happier here, and we'd have quicker
progress too I bet, if instead of 1 30 part series we had 5 or 6 smaller
series.

> So I think this series is a big step forward and urge another look, please.

That's what I did with v3 yesterday and why I had a bunch of other
feedback.

But, yes, I'll take some parts of v3 and see what I can come up with.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1124 bytes --]

On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:

> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> available, add it as an explicit dependency.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> (no changes since v2)
> 
> Changes in v2:
> - Add new patch to update EFI_LOADER to depend on DM_ETH
> 
>  lib/efi_loader/Kconfig | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> index 13cad6342c36..fca4b3eef270 100644
> --- a/lib/efi_loader/Kconfig
> +++ b/lib/efi_loader/Kconfig
> @@ -11,6 +11,7 @@ config EFI_LOADER
>  	# We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
>  	depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
>  	depends on BLK
> +	depends on DM_ETH
>  	depends on !EFI_APP
>  	default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
>  	select CHARSET

I don't think this is needed.  After reconfiguring qemu_arm64 to be able
to disable networking entirely, we still are able to build with
EFI_LOADER enabled, and no warning / link errors.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 06/32] sifive: Drop an unnecessary #ifdef

[Heinrich Schuchardt]

On 10/17/23 00:27, Simon Glass wrote:
> This code is normally compiled for sifive, but sandbox can also compile
> it. We should not use UNIT_TEST as a synonym for SANDBOX, since it is
> possible to disable UNIT_TEST for sandbox.
>
> Drop the condition since it isn't needed.

This is not what the patch does. It introduces another condition.

>
> Signed-off-by: Simon Glass <sjg@chromium.org>
> Suggested-by: Sean Anderson <seanga2@gmail.com>
> ---
>
> Changes in v3:
> - Just drop the condition instead
>
>   include/k210/pll.h | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/include/k210/pll.h b/include/k210/pll.h
> index fd16a89cb203..6dd60b2eb4fc 100644
> --- a/include/k210/pll.h
> +++ b/include/k210/pll.h
> @@ -13,7 +13,7 @@ struct k210_pll_config {
>   	u8 od;
>   };
>
> -#ifdef CONFIG_UNIT_TEST
> +#ifdef CONFIG_SANDBOX

We should be able to compile with and without unit tests on the MAIX
boards. Why should CONFIG_SANDBOX have any relevance here?

Why don't we make k210_pll_calc_config() non-static in all cases?

Best regards

Heinrich

>   TEST_STATIC int k210_pll_calc_config(u32 rate, u32 rate_in,
>   				     struct k210_pll_config *best);
>   #ifndef nop

Re: [PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1305 bytes --]

On Mon, Oct 16, 2023 at 04:27:53PM -0600, Simon Glass wrote:
> With recent changes over the last few years in boot/Kconfig it is
> no-longer possible to disable CMDLINE. It results in various link
> errors because some options which require CMDLINE are enabled
> regardless of whether it is available.
> 
> Add the necessary conditions to fix this.
> 
> Note that it would be better to have all commands depend on CMDLINE,
> but that is extremely difficult at present, since some functions use
> CMD_xxx to enable feature xxx. For example networking and environment
> have a number of problems to tease apart.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> ---
> 
> Changes in v3:
> - Reword commit message slightly
> 
>  boot/Kconfig           | 19 ++++++++++++-------
>  lib/efi_loader/Kconfig |  1 +
>  2 files changed, 13 insertions(+), 7 deletions(-)
> 
> diff --git a/boot/Kconfig b/boot/Kconfig
> index a01e6cb8aafe..f74ac7e9cc72 100644
> --- a/boot/Kconfig
> +++ b/boot/Kconfig
> @@ -342,6 +342,7 @@ endif # FIT
>  
>  config PXE_UTILS
>  	bool
> +	depends on CMDLINE
>  	select MENU
>  	help
>  	  Utilities for parsing PXE file formats.

So, this part here is because of CONFIG_SYS_PBSIZE, which I think we can
/ should solve another way.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Tom,

On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
>
> > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > available, add it as an explicit dependency.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > (no changes since v2)
> >
> > Changes in v2:
> > - Add new patch to update EFI_LOADER to depend on DM_ETH
> >
> >  lib/efi_loader/Kconfig | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > index 13cad6342c36..fca4b3eef270 100644
> > --- a/lib/efi_loader/Kconfig
> > +++ b/lib/efi_loader/Kconfig
> > @@ -11,6 +11,7 @@ config EFI_LOADER
> >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> >       depends on BLK
> > +     depends on DM_ETH
> >       depends on !EFI_APP
> >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> >       select CHARSET
>
> I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> to disable networking entirely, we still are able to build with
> EFI_LOADER enabled, and no warning / link errors.

Fair enough.

As of this patch the following errors are generated with -a ~CMDLINE :

+lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
reference to `eth_get_dev'
+/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
undefined reference to `eth_get_dev'
+/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
undefined reference to `eth_get_dev'
+/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
+lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
reference to `efi_get_image_parameters'

which is why I added this patch. A later patch disables networking
entirely with ~CMDLINE so that is why this problem goes away.

This series was basically built up by disabling CMDLINE and then
fixing the errors one by one. I didn't go back and check whether (at
the end) all the patches were needed.

I could do that and send a v4, if your more general concerns can be sorted out.

Regards,
Simon

Re: [PATCH v3 02/32] bootstd: Correct dependencies on CMDLINE

[Simon Glass]

Hi Tom,

On Wed, 18 Oct 2023 at 08:42, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:27:53PM -0600, Simon Glass wrote:
> > With recent changes over the last few years in boot/Kconfig it is
> > no-longer possible to disable CMDLINE. It results in various link
> > errors because some options which require CMDLINE are enabled
> > regardless of whether it is available.
> >
> > Add the necessary conditions to fix this.
> >
> > Note that it would be better to have all commands depend on CMDLINE,
> > but that is extremely difficult at present, since some functions use
> > CMD_xxx to enable feature xxx. For example networking and environment
> > have a number of problems to tease apart.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > Changes in v3:
> > - Reword commit message slightly
> >
> >  boot/Kconfig           | 19 ++++++++++++-------
> >  lib/efi_loader/Kconfig |  1 +
> >  2 files changed, 13 insertions(+), 7 deletions(-)
> >
> > diff --git a/boot/Kconfig b/boot/Kconfig
> > index a01e6cb8aafe..f74ac7e9cc72 100644
> > --- a/boot/Kconfig
> > +++ b/boot/Kconfig
> > @@ -342,6 +342,7 @@ endif # FIT
> >
> >  config PXE_UTILS
> >       bool
> > +     depends on CMDLINE
> >       select MENU
> >       help
> >         Utilities for parsing PXE file formats.
>
> So, this part here is because of CONFIG_SYS_PBSIZE, which I think we can
> / should solve another way.

Perhaps it would help if you just clarify your goal.

My goal with this series is to allow CMDLINE to be disabled without
build errors, so we can start the work booting via bootstd (with
CMDLINE disabled).

This isn't a straight-line activity. It involves changes that will
become redundant later, simply because I haven't written the 100s of
patches needed to figure it all out. Even if I did it would be
unreviewable and require lots of rework based on feedback.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 2729 bytes --]

On Wed, Oct 18, 2023 at 09:23:42AM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> >
> > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > available, add it as an explicit dependency.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v2)
> > >
> > > Changes in v2:
> > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > >
> > >  lib/efi_loader/Kconfig | 1 +
> > >  1 file changed, 1 insertion(+)
> > >
> > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > index 13cad6342c36..fca4b3eef270 100644
> > > --- a/lib/efi_loader/Kconfig
> > > +++ b/lib/efi_loader/Kconfig
> > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > >       depends on BLK
> > > +     depends on DM_ETH
> > >       depends on !EFI_APP
> > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > >       select CHARSET
> >
> > I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> > to disable networking entirely, we still are able to build with
> > EFI_LOADER enabled, and no warning / link errors.
> 
> Fair enough.
> 
> As of this patch the following errors are generated with -a ~CMDLINE :
> 
> +lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
> reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> undefined reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> undefined reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> +lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
> reference to `efi_get_image_parameters'
> 
> which is why I added this patch. A later patch disables networking
> entirely with ~CMDLINE so that is why this problem goes away.
> 
> This series was basically built up by disabling CMDLINE and then
> fixing the errors one by one. I didn't go back and check whether (at
> the end) all the patches were needed.

Yes, part of the issue with the series is that you didn't go and rework
things after getting CMDLINE to link for sandbox, to see what else is or
is not still needed.

> I could do that and send a v4, if your more general concerns can be sorted out.

Don't worry about a v4 currently please, I'm working through the issues
now picking out parts of v3 and re-working others.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Tom,

On Wed, 18 Oct 2023 at 09:27, Tom Rini <trini@konsulko.com> wrote:
>
> On Wed, Oct 18, 2023 at 09:23:42AM -0600, Simon Glass wrote:
> > Hi Tom,
> >
> > On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > >
> > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > available, add it as an explicit dependency.
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > ---
> > > >
> > > > (no changes since v2)
> > > >
> > > > Changes in v2:
> > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > >
> > > >  lib/efi_loader/Kconfig | 1 +
> > > >  1 file changed, 1 insertion(+)
> > > >
> > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > index 13cad6342c36..fca4b3eef270 100644
> > > > --- a/lib/efi_loader/Kconfig
> > > > +++ b/lib/efi_loader/Kconfig
> > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > >       depends on BLK
> > > > +     depends on DM_ETH
> > > >       depends on !EFI_APP
> > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > >       select CHARSET
> > >
> > > I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> > > to disable networking entirely, we still are able to build with
> > > EFI_LOADER enabled, and no warning / link errors.
> >
> > Fair enough.
> >
> > As of this patch the following errors are generated with -a ~CMDLINE :
> >
> > +lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
> > reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> > +lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
> > reference to `efi_get_image_parameters'
> >
> > which is why I added this patch. A later patch disables networking
> > entirely with ~CMDLINE so that is why this problem goes away.
> >
> > This series was basically built up by disabling CMDLINE and then
> > fixing the errors one by one. I didn't go back and check whether (at
> > the end) all the patches were needed.
>
> Yes, part of the issue with the series is that you didn't go and rework
> things after getting CMDLINE to link for sandbox, to see what else is or
> is not still needed.
>
> > I could do that and send a v4, if your more general concerns can be sorted out.
>
> Don't worry about a v4 currently please, I'm working through the issues
> now picking out parts of v3 and re-working others.

OK, thanks.

Regards,
Simon

Re: [PATCH v3 06/32] sifive: Drop an unnecessary #ifdef

[Sean Anderson]

On 10/18/23 09:51, Heinrich Schuchardt wrote:
> On 10/17/23 00:27, Simon Glass wrote:
>> This code is normally compiled for sifive, but sandbox can also compile
>> it. We should not use UNIT_TEST as a synonym for SANDBOX, since it is
>> possible to disable UNIT_TEST for sandbox.
>>
>> Drop the condition since it isn't needed.
> 
> This is not what the patch does. It introduces another condition.
> 
>>
>> Signed-off-by: Simon Glass <sjg@chromium.org>
>> Suggested-by: Sean Anderson <seanga2@gmail.com>
>> ---
>>
>> Changes in v3:
>> - Just drop the condition instead
>>
>>   include/k210/pll.h | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/include/k210/pll.h b/include/k210/pll.h
>> index fd16a89cb203..6dd60b2eb4fc 100644
>> --- a/include/k210/pll.h
>> +++ b/include/k210/pll.h
>> @@ -13,7 +13,7 @@ struct k210_pll_config {
>>       u8 od;
>>   };
>>
>> -#ifdef CONFIG_UNIT_TEST
>> +#ifdef CONFIG_SANDBOX
> 
> We should be able to compile with and without unit tests on the MAIX
> boards. Why should CONFIG_SANDBOX have any relevance here?
> 
> Why don't we make k210_pll_calc_config() non-static in all cases?

U-Boot is smaller when it is static. But the forward declaration can be made
unconditionally, as long as it is unused.

--Sean

Re: [PATCH v3 06/32] sifive: Drop an unnecessary #ifdef

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1954 bytes --]

On Wed, Oct 18, 2023 at 01:23:13PM -0400, Sean Anderson wrote:
> On 10/18/23 09:51, Heinrich Schuchardt wrote:
> > On 10/17/23 00:27, Simon Glass wrote:
> > > This code is normally compiled for sifive, but sandbox can also compile
> > > it. We should not use UNIT_TEST as a synonym for SANDBOX, since it is
> > > possible to disable UNIT_TEST for sandbox.
> > > 
> > > Drop the condition since it isn't needed.
> > 
> > This is not what the patch does. It introduces another condition.
> > 
> > > 
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > Suggested-by: Sean Anderson <seanga2@gmail.com>
> > > ---
> > > 
> > > Changes in v3:
> > > - Just drop the condition instead
> > > 
> > >   include/k210/pll.h | 2 +-
> > >   1 file changed, 1 insertion(+), 1 deletion(-)
> > > 
> > > diff --git a/include/k210/pll.h b/include/k210/pll.h
> > > index fd16a89cb203..6dd60b2eb4fc 100644
> > > --- a/include/k210/pll.h
> > > +++ b/include/k210/pll.h
> > > @@ -13,7 +13,7 @@ struct k210_pll_config {
> > >       u8 od;
> > >   };
> > > 
> > > -#ifdef CONFIG_UNIT_TEST
> > > +#ifdef CONFIG_SANDBOX
> > 
> > We should be able to compile with and without unit tests on the MAIX
> > boards. Why should CONFIG_SANDBOX have any relevance here?
> > 
> > Why don't we make k210_pll_calc_config() non-static in all cases?
> 
> U-Boot is smaller when it is static. But the forward declaration can be made
> unconditionally, as long as it is unused.

I think part of the problem with this patch is confusing what it's
doing. The issue here is that we can (and this is good!) and do compile
drivers/clk/clk_k210.c for sandbox, so it gets coverity scans and so
forth.  However, if we build sandbox with UNIT_TEST=n then we get an
undefined reference to 'nop'.  Because sandbox doesn't define nop().
This header then provides nop for sandbox.  But, we can do something
clearer, now that this is spelled out.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 663 bytes --]

On Mon, Oct 16, 2023 at 04:28:23PM -0600, Simon Glass wrote:

> Now that everything is working, add a test to make sure that this
> builds correctly.
> 
> Signed-off-by: Simon Glass <sjg@chromium.org>
> Reviewed-by: Tom Rini <trini@konsulko.com>
> ---
> 
> Changes in v3:
> - Rebase on Tom's LONGHELP series
> - Correct 'of' typo
> 
>  test/py/tests/test_sandbox_opts.py | 20 ++++++++++++++++++++
>  1 file changed, 20 insertions(+)
>  create mode 100644 test/py/tests/test_sandbox_opts.py

This is not doing the equivalent of:
make sandbox_config
sed -i -e 's/CONFIG_CMDLINE=y/CONFIG_CMDLINE=n/' .config
make oldconfig
make all

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Peter Robinson]

On Wed, Oct 18, 2023 at 4:29 PM Simon Glass <sjg@chromium.org> wrote:
>
> Hi Tom,
>
> On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> >
> > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > available, add it as an explicit dependency.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v2)
> > >
> > > Changes in v2:
> > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > >
> > >  lib/efi_loader/Kconfig | 1 +
> > >  1 file changed, 1 insertion(+)
> > >
> > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > index 13cad6342c36..fca4b3eef270 100644
> > > --- a/lib/efi_loader/Kconfig
> > > +++ b/lib/efi_loader/Kconfig
> > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > >       depends on BLK
> > > +     depends on DM_ETH
> > >       depends on !EFI_APP
> > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > >       select CHARSET
> >
> > I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> > to disable networking entirely, we still are able to build with
> > EFI_LOADER enabled, and no warning / link errors.
>
> Fair enough.
>
> As of this patch the following errors are generated with -a ~CMDLINE :
>
> +lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
> reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> undefined reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> undefined reference to `eth_get_dev'
> +/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> +lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
> reference to `efi_get_image_parameters'
>
> which is why I added this patch. A later patch disables networking
> entirely with ~CMDLINE so that is why this problem goes away.
>
> This series was basically built up by disabling CMDLINE and then
> fixing the errors one by one. I didn't go back and check whether (at
> the end) all the patches were needed.

Was this just tested on sandbox? How do we know it's not going to
regress a vast amount of actual devices?

> I could do that and send a v4, if your more general concerns can be sorted out.
>
> Regards,
> Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 2742 bytes --]

On Thu, Oct 19, 2023 at 11:39:49AM +0100, Peter Robinson wrote:
> On Wed, Oct 18, 2023 at 4:29 PM Simon Glass <sjg@chromium.org> wrote:
> >
> > Hi Tom,
> >
> > On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > >
> > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > available, add it as an explicit dependency.
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > ---
> > > >
> > > > (no changes since v2)
> > > >
> > > > Changes in v2:
> > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > >
> > > >  lib/efi_loader/Kconfig | 1 +
> > > >  1 file changed, 1 insertion(+)
> > > >
> > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > index 13cad6342c36..fca4b3eef270 100644
> > > > --- a/lib/efi_loader/Kconfig
> > > > +++ b/lib/efi_loader/Kconfig
> > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > >       depends on BLK
> > > > +     depends on DM_ETH
> > > >       depends on !EFI_APP
> > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > >       select CHARSET
> > >
> > > I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> > > to disable networking entirely, we still are able to build with
> > > EFI_LOADER enabled, and no warning / link errors.
> >
> > Fair enough.
> >
> > As of this patch the following errors are generated with -a ~CMDLINE :
> >
> > +lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
> > reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> > +lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
> > reference to `efi_get_image_parameters'
> >
> > which is why I added this patch. A later patch disables networking
> > entirely with ~CMDLINE so that is why this problem goes away.
> >
> > This series was basically built up by disabling CMDLINE and then
> > fixing the errors one by one. I didn't go back and check whether (at
> > the end) all the patches were needed.
> 
> Was this just tested on sandbox? How do we know it's not going to
> regress a vast amount of actual devices?

It both shouldn't (DM_ETH is required), and this patch isn't moving
forward either.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Simon Glass]

Hi Tom,

On Wed, 18 Oct 2023 at 06:39, Tom Rini <trini@konsulko.com> wrote:
>
> On Tue, Oct 17, 2023 at 09:31:04PM -0600, Simon Glass wrote:
> > Hi Tom,
> >
> > On Tue, 17 Oct 2023 at 08:02, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 16, 2023 at 04:28:05PM -0600, Simon Glass wrote:
> > >
> > > > There are two cleanup functions needed during boot which depend on
> > > > CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()
> > > >
> > > > Provide static inline versions of these for when commands are not
> > > > enabled.
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > >
> > > NAK, these functions need to be available to boot the OS, regardless of
> > > command line existing or not.  Unwind things in the other direction
> > > please.
> >
> > That won't be a successful strategy. See my other reply on this. I
> > have had this idea since 2016 and have looked at it every now and
> > then. With bootstd it makes more sense since we actually have a
> > framework to boot without the command line. This series enables such
> > work, but cannot include it.
>
> I don't understand.  The "cleanup" functions here are to ensure that any
> devices that we've touched are in the state the OS needs them to be.
> It's not "we ran a command", it's "we loaded something off of USB (or
> network or MMC or ..)" and so we must have those called.

Right, but they are in with the bootm command code. We could pull them
out, but it is only a partialsolution. Really we need another series
to allow 'bootm' booting without CMDLINE. In other words, we cannot
boot most things without cmdline today. That has been the case before
this series and unfortunately this series doesn't change that.

At present we cannot even build without ~CMDLINE - that has regressed.
I am trying to patch things up and make it build and add a test.

Regards,
Simon

Re: [PATCH v3 04/32] cmd: Add a few more dependencies on CMDLINE

[Simon Glass]

Hi Heinrich,

On Wed, 18 Oct 2023 at 07:14, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
>
> On 10/17/23 00:27, Simon Glass wrote:
> > Add this to some more commands to avoid build errors with sandbox.
> >
> > Note that this is a temporary solution to expose more problems. A later
> > patch puts these behind an 'if CMDLINE'
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > ---
> >
> > Changes in v3:
> > - Add an explation as to why this patch is here
> >
> >   cmd/Kconfig | 6 ++++++
> >   1 file changed, 6 insertions(+)
> >
> > diff --git a/cmd/Kconfig b/cmd/Kconfig
> > index 5bc0a92d57ad..00a7f84bce99 100644
> > --- a/cmd/Kconfig
> > +++ b/cmd/Kconfig
> > @@ -231,6 +231,7 @@ menu "Boot commands"
> >
> >   config CMD_BOOTD
> >       bool "bootd"
> > +     depends on CMDLINE
>
> Why don't we simply use a single "if" for all commands?
> Adding the same dependency to 200+ commands does not feel right.
>
> If there is any CONFIG_CMD* that is needed without CMDLINE, then that
> code should move to lib/ or drivers/.

This is here to fix build failures at this point in the series.

One of the later patches does what you suggest (27 I think).

Regards,
Simon

Re: [PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Simon Glass]

Hi Tom,

On Wed, 18 Oct 2023 at 20:32, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 16, 2023 at 04:28:23PM -0600, Simon Glass wrote:
>
> > Now that everything is working, add a test to make sure that this
> > builds correctly.
> >
> > Signed-off-by: Simon Glass <sjg@chromium.org>
> > Reviewed-by: Tom Rini <trini@konsulko.com>
> > ---
> >
> > Changes in v3:
> > - Rebase on Tom's LONGHELP series
> > - Correct 'of' typo
> >
> >  test/py/tests/test_sandbox_opts.py | 20 ++++++++++++++++++++
> >  1 file changed, 20 insertions(+)
> >  create mode 100644 test/py/tests/test_sandbox_opts.py
>
> This is not doing the equivalent of:
> make sandbox_config
> sed -i -e 's/CONFIG_CMDLINE=y/CONFIG_CMDLINE=n/' .config
> make oldconfig
> make all

Buildman should do that automatically...is something failing?

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Peter,

On Thu, 19 Oct 2023 at 04:40, Peter Robinson <pbrobinson@gmail.com> wrote:
>
> On Wed, Oct 18, 2023 at 4:29 PM Simon Glass <sjg@chromium.org> wrote:
> >
> > Hi Tom,
> >
> > On Wed, 18 Oct 2023 at 07:37, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > >
> > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > available, add it as an explicit dependency.
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > ---
> > > >
> > > > (no changes since v2)
> > > >
> > > > Changes in v2:
> > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > >
> > > >  lib/efi_loader/Kconfig | 1 +
> > > >  1 file changed, 1 insertion(+)
> > > >
> > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > index 13cad6342c36..fca4b3eef270 100644
> > > > --- a/lib/efi_loader/Kconfig
> > > > +++ b/lib/efi_loader/Kconfig
> > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > >       depends on BLK
> > > > +     depends on DM_ETH
> > > >       depends on !EFI_APP
> > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > >       select CHARSET
> > >
> > > I don't think this is needed.  After reconfiguring qemu_arm64 to be able
> > > to disable networking entirely, we still are able to build with
> > > EFI_LOADER enabled, and no warning / link errors.
> >
> > Fair enough.
> >
> > As of this patch the following errors are generated with -a ~CMDLINE :
> >
> > +lib/efi_loader/efi_device_path.c:985:(.text+0xca4): undefined
> > reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> > undefined reference to `eth_get_dev'
> > +/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> > +lib/efi_loader/efi_device_path.c:1095:(.text+0xe00): undefined
> > reference to `efi_get_image_parameters'
> >
> > which is why I added this patch. A later patch disables networking
> > entirely with ~CMDLINE so that is why this problem goes away.
> >
> > This series was basically built up by disabling CMDLINE and then
> > fixing the errors one by one. I didn't go back and check whether (at
> > the end) all the patches were needed.
>
> Was this just tested on sandbox? How do we know it's not going to
> regress a vast amount of actual devices?

This was tested with buildman to check for config changes. The purpose
of this series.is to get back to what worked a few years ago (CMDLINE
could be disabled). If you don't disable CMDLINE, then it doesn't
really change anything.

Perhaps we should add a build test for non-network EFI_LOADER if
people are going to use it? I don't see any boards that do:

$ ./tools/qconfig.py -f EFI_LOADER ~NET
0 matches

That is a great tool for checking what is used in the code base.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Heinrich,

On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
>
> On 10/17/23 16:09, Tom Rini wrote:
> > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> >
> >> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> >> available, add it as an explicit dependency.
> >>
> >> Signed-off-by: Simon Glass <sjg@chromium.org>
> >> ---
> >>
> >> (no changes since v2)
> >>
> >> Changes in v2:
> >> - Add new patch to update EFI_LOADER to depend on DM_ETH
> >>
> >>   lib/efi_loader/Kconfig | 1 +
> >>   1 file changed, 1 insertion(+)
> >>
> >> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> >> index 13cad6342c36..fca4b3eef270 100644
> >> --- a/lib/efi_loader/Kconfig
> >> +++ b/lib/efi_loader/Kconfig
> >> @@ -11,6 +11,7 @@ config EFI_LOADER
> >>      # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> >>      depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> >>      depends on BLK
> >> +    depends on DM_ETH
> >>      depends on !EFI_APP
> >>      default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> >>      select CHARSET
> >
> > Does this work for you Heinrich, or do you want to clarify the
> > dependencies (and re-organize the code as needed) around networking?
> >
>
> We should be able to boot via EFI on devices without U-Boot network support.
>
> We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> CONFIG_DM_ETH.
>
> Why is this not sufficient?
> Is there a configuration that does not build?

The point of this series is to disable CMDLINE and fix up what breaks.

In this case we have some sort of breakage...perhaps Tom has already
found it, but otherwise could you take a look?

We should be able to disable NET and LTO in sandbox and still build.
But this fails at present[1]. You can try it on -master

Regards,
Simon

[1] sjg@sjg1:~/u$ crosfw sandbox -L
cmd: make -j4 'CROSS_COMPILE=' --no-print-directory 'HOSTSTRIP=true'
'QEMU_ARCH=' 'KCONFIG_NOSILENTUPDATE=1' 'O=/tmp/b/sandbox' 'NO_LTO=1'
-s 'BUILD_ROM=1' all
/usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_eth':
/home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:985:(.text+0xca4):
undefined reference to `eth_get_dev'
/usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
undefined reference to `eth_get_dev'
/usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
undefined reference to `eth_get_dev'
collect2: error: ld returned 1 exit status
make[1]: *** [/home/sjg/c/src/third_party/u-boot/files/Makefile:1765:
u-boot] Error 1
make: *** [Makefile:177: sub-make] Error 2

sjg@sjg1:~/u$ crosfw sandbox
sjg@sjg1:~/u$ (passes)

'crosfw xx' is just 'buildman --bo xxx'

diff --git a/boot/pxe_utils.c b/boot/pxe_utils.c
index a92bb896c63e..4e9996a92342 100644
--- a/boot/pxe_utils.c
+++ b/boot/pxe_utils.c
@@ -48,7 +48,7 @@ int pxe_get_file_size(ulong *sizep)

  return 0;
 }
-
+#if 0
 /**
  * format_mac_pxe() - obtain a MAC address in the PXE format
  *
@@ -82,7 +82,7 @@ int format_mac_pxe(char *outbuf, size_t outbuf_len)

  return 1;
 }
-
+#endif
 /**
  * get_relfile() - read a file relative to the PXE file
  *
diff --git a/configs/sandbox_defconfig b/configs/sandbox_defconfig
index 47417cb0391d..57bb6202dbb5 100644
--- a/configs/sandbox_defconfig
+++ b/configs/sandbox_defconfig
@@ -349,3 +349,4 @@ CONFIG_TEST_FDTDEC=y
 CONFIG_UNIT_TEST=y
 CONFIG_UT_TIME=y
 CONFIG_UT_DM=y
+# CONFIG_NET is not set

Regards,
Simon

Re: [PATCH v3 14/32] bootm: Allow building when cleanup functions are missing

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 2244 bytes --]

On Thu, Oct 19, 2023 at 07:59:29AM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Wed, 18 Oct 2023 at 06:39, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Tue, Oct 17, 2023 at 09:31:04PM -0600, Simon Glass wrote:
> > > Hi Tom,
> > >
> > > On Tue, 17 Oct 2023 at 08:02, Tom Rini <trini@konsulko.com> wrote:
> > > >
> > > > On Mon, Oct 16, 2023 at 04:28:05PM -0600, Simon Glass wrote:
> > > >
> > > > > There are two cleanup functions needed during boot which depend on
> > > > > CMD_BOOTM: bootm_disable_interrupts() and board_quiesce_devices()
> > > > >
> > > > > Provide static inline versions of these for when commands are not
> > > > > enabled.
> > > > >
> > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > >
> > > > NAK, these functions need to be available to boot the OS, regardless of
> > > > command line existing or not.  Unwind things in the other direction
> > > > please.
> > >
> > > That won't be a successful strategy. See my other reply on this. I
> > > have had this idea since 2016 and have looked at it every now and
> > > then. With bootstd it makes more sense since we actually have a
> > > framework to boot without the command line. This series enables such
> > > work, but cannot include it.
> >
> > I don't understand.  The "cleanup" functions here are to ensure that any
> > devices that we've touched are in the state the OS needs them to be.
> > It's not "we ran a command", it's "we loaded something off of USB (or
> > network or MMC or ..)" and so we must have those called.
> 
> Right, but they are in with the bootm command code. We could pull them

I don't think that's right.  For bootm, things are split correctly.  I'm
not quite sure about booti/bootz, and I can see what bootefi needs to be
split more.  But bootm is fine.

[snip]
> At present we cannot even build without ~CMDLINE - that has regressed.
> I am trying to patch things up and make it build and add a test.

To be clear here, I prefer build failure in that case over having broken
functionality. That way leads to developers expecting something to work
that will not. Those prep functions are critical. It's also a moot point
because we can have them today with CMDLINE=n.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 1118 bytes --]

On Thu, Oct 19, 2023 at 08:00:00AM -0600, Simon Glass wrote:
> Hi Tom,
> 
> On Wed, 18 Oct 2023 at 20:32, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 16, 2023 at 04:28:23PM -0600, Simon Glass wrote:
> >
> > > Now that everything is working, add a test to make sure that this
> > > builds correctly.
> > >
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > Reviewed-by: Tom Rini <trini@konsulko.com>
> > > ---
> > >
> > > Changes in v3:
> > > - Rebase on Tom's LONGHELP series
> > > - Correct 'of' typo
> > >
> > >  test/py/tests/test_sandbox_opts.py | 20 ++++++++++++++++++++
> > >  1 file changed, 20 insertions(+)
> > >  create mode 100644 test/py/tests/test_sandbox_opts.py
> >
> > This is not doing the equivalent of:
> > make sandbox_config
> > sed -i -e 's/CONFIG_CMDLINE=y/CONFIG_CMDLINE=n/' .config
> > make oldconfig
> > make all
> 
> Buildman should do that automatically...is something failing?

Yes. https://source.denx.de/u-boot/u-boot/-/jobs/716843 fails, but
shouldn't, and it's only the no LTO job, so maybe there's some conflict
there?

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[AKASHI Takahiro]

On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> Hi Heinrich,
> 
> On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> >
> > On 10/17/23 16:09, Tom Rini wrote:
> > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > >
> > >> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > >> available, add it as an explicit dependency.
> > >>
> > >> Signed-off-by: Simon Glass <sjg@chromium.org>
> > >> ---
> > >>
> > >> (no changes since v2)
> > >>
> > >> Changes in v2:
> > >> - Add new patch to update EFI_LOADER to depend on DM_ETH
> > >>
> > >>   lib/efi_loader/Kconfig | 1 +
> > >>   1 file changed, 1 insertion(+)
> > >>
> > >> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > >> index 13cad6342c36..fca4b3eef270 100644
> > >> --- a/lib/efi_loader/Kconfig
> > >> +++ b/lib/efi_loader/Kconfig
> > >> @@ -11,6 +11,7 @@ config EFI_LOADER
> > >>      # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > >>      depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > >>      depends on BLK
> > >> +    depends on DM_ETH
> > >>      depends on !EFI_APP
> > >>      default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > >>      select CHARSET
> > >
> > > Does this work for you Heinrich, or do you want to clarify the
> > > dependencies (and re-organize the code as needed) around networking?
> > >
> >
> > We should be able to boot via EFI on devices without U-Boot network support.
> >
> > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > CONFIG_DM_ETH.
> >
> > Why is this not sufficient?
> > Is there a configuration that does not build?
> 
> The point of this series is to disable CMDLINE and fix up what breaks.
> 
> In this case we have some sort of breakage...perhaps Tom has already
> found it, but otherwise could you take a look?
> 
> We should be able to disable NET and LTO in sandbox and still build.
> But this fails at present[1]. You can try it on -master

Obviously, it would be necessary to enclose efi_dp_from_eth()
with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
Then, we could drop "depends on DM_ETH".

Another possible place for completeness is "case UCLASS_ETH" clause
in dp_size(), but it seems to be harmless.

-Takahiro Akashi


> Regards,
> Simon
> 
> [1] sjg@sjg1:~/u$ crosfw sandbox -L
> cmd: make -j4 'CROSS_COMPILE=' --no-print-directory 'HOSTSTRIP=true'
> 'QEMU_ARCH=' 'KCONFIG_NOSILENTUPDATE=1' 'O=/tmp/b/sandbox' 'NO_LTO=1'
> -s 'BUILD_ROM=1' all
> /usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_eth':
> /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:985:(.text+0xca4):
> undefined reference to `eth_get_dev'
> /usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> undefined reference to `eth_get_dev'
> /usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> undefined reference to `eth_get_dev'
> collect2: error: ld returned 1 exit status
> make[1]: *** [/home/sjg/c/src/third_party/u-boot/files/Makefile:1765:
> u-boot] Error 1
> make: *** [Makefile:177: sub-make] Error 2
> 
> sjg@sjg1:~/u$ crosfw sandbox
> sjg@sjg1:~/u$ (passes)
> 
> 'crosfw xx' is just 'buildman --bo xxx'
> 
> diff --git a/boot/pxe_utils.c b/boot/pxe_utils.c
> index a92bb896c63e..4e9996a92342 100644
> --- a/boot/pxe_utils.c
> +++ b/boot/pxe_utils.c
> @@ -48,7 +48,7 @@ int pxe_get_file_size(ulong *sizep)
> 
>   return 0;
>  }
> -
> +#if 0
>  /**
>   * format_mac_pxe() - obtain a MAC address in the PXE format
>   *
> @@ -82,7 +82,7 @@ int format_mac_pxe(char *outbuf, size_t outbuf_len)
> 
>   return 1;
>  }
> -
> +#endif
>  /**
>   * get_relfile() - read a file relative to the PXE file
>   *
> diff --git a/configs/sandbox_defconfig b/configs/sandbox_defconfig
> index 47417cb0391d..57bb6202dbb5 100644
> --- a/configs/sandbox_defconfig
> +++ b/configs/sandbox_defconfig
> @@ -349,3 +349,4 @@ CONFIG_TEST_FDTDEC=y
>  CONFIG_UNIT_TEST=y
>  CONFIG_UT_TIME=y
>  CONFIG_UT_DM=y
> +# CONFIG_NET is not set
> 
> Regards,
> Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi,

On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
<takahiro.akashi@linaro.org> wrote:
>
> On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > Hi Heinrich,
> >
> > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > >
> > > On 10/17/23 16:09, Tom Rini wrote:
> > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > >
> > > >> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > >> available, add it as an explicit dependency.
> > > >>
> > > >> Signed-off-by: Simon Glass <sjg@chromium.org>
> > > >> ---
> > > >>
> > > >> (no changes since v2)
> > > >>
> > > >> Changes in v2:
> > > >> - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > >>
> > > >>   lib/efi_loader/Kconfig | 1 +
> > > >>   1 file changed, 1 insertion(+)
> > > >>
> > > >> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > >> index 13cad6342c36..fca4b3eef270 100644
> > > >> --- a/lib/efi_loader/Kconfig
> > > >> +++ b/lib/efi_loader/Kconfig
> > > >> @@ -11,6 +11,7 @@ config EFI_LOADER
> > > >>      # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > >>      depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > >>      depends on BLK
> > > >> +    depends on DM_ETH
> > > >>      depends on !EFI_APP
> > > >>      default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > >>      select CHARSET
> > > >
> > > > Does this work for you Heinrich, or do you want to clarify the
> > > > dependencies (and re-organize the code as needed) around networking?
> > > >
> > >
> > > We should be able to boot via EFI on devices without U-Boot network support.
> > >
> > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > CONFIG_DM_ETH.
> > >
> > > Why is this not sufficient?
> > > Is there a configuration that does not build?
> >
> > The point of this series is to disable CMDLINE and fix up what breaks.
> >
> > In this case we have some sort of breakage...perhaps Tom has already
> > found it, but otherwise could you take a look?
> >
> > We should be able to disable NET and LTO in sandbox and still build.
> > But this fails at present[1]. You can try it on -master
>
> Obviously, it would be necessary to enclose efi_dp_from_eth()
> with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> Then, we could drop "depends on DM_ETH".

Strange that it only happens on the non-LTO board, though?

>
> Another possible place for completeness is "case UCLASS_ETH" clause
> in dp_size(), but it seems to be harmless.

OK

Regards,
Simon


>
> -Takahiro Akashi
>
>
> > Regards,
> > Simon
> >
> > [1] sjg@sjg1:~/u$ crosfw sandbox -L
> > cmd: make -j4 'CROSS_COMPILE=' --no-print-directory 'HOSTSTRIP=true'
> > 'QEMU_ARCH=' 'KCONFIG_NOSILENTUPDATE=1' 'O=/tmp/b/sandbox' 'NO_LTO=1'
> > -s 'BUILD_ROM=1' all
> > /usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_eth':
> > /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:985:(.text+0xca4):
> > undefined reference to `eth_get_dev'
> > /usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:987:(.text+0xca9):
> > undefined reference to `eth_get_dev'
> > /usr/bin/ld: /home/sjg/c/src/third_party/u-boot/files/lib/efi_loader/efi_device_path.c:993:(.text+0xcc9):
> > undefined reference to `eth_get_dev'
> > collect2: error: ld returned 1 exit status
> > make[1]: *** [/home/sjg/c/src/third_party/u-boot/files/Makefile:1765:
> > u-boot] Error 1
> > make: *** [Makefile:177: sub-make] Error 2
> >
> > sjg@sjg1:~/u$ crosfw sandbox
> > sjg@sjg1:~/u$ (passes)
> >
> > 'crosfw xx' is just 'buildman --bo xxx'
> >
> > diff --git a/boot/pxe_utils.c b/boot/pxe_utils.c
> > index a92bb896c63e..4e9996a92342 100644
> > --- a/boot/pxe_utils.c
> > +++ b/boot/pxe_utils.c
> > @@ -48,7 +48,7 @@ int pxe_get_file_size(ulong *sizep)
> >
> >   return 0;
> >  }
> > -
> > +#if 0
> >  /**
> >   * format_mac_pxe() - obtain a MAC address in the PXE format
> >   *
> > @@ -82,7 +82,7 @@ int format_mac_pxe(char *outbuf, size_t outbuf_len)
> >
> >   return 1;
> >  }
> > -
> > +#endif
> >  /**
> >   * get_relfile() - read a file relative to the PXE file
> >   *
> > diff --git a/configs/sandbox_defconfig b/configs/sandbox_defconfig
> > index 47417cb0391d..57bb6202dbb5 100644
> > --- a/configs/sandbox_defconfig
> > +++ b/configs/sandbox_defconfig
> > @@ -349,3 +349,4 @@ CONFIG_TEST_FDTDEC=y
> >  CONFIG_UNIT_TEST=y
> >  CONFIG_UT_TIME=y
> >  CONFIG_UT_DM=y
> > +# CONFIG_NET is not set
> >
> > Regards,
> > Simon

Re: [PATCH v3 32/32] sandbox: Add a test for disabling CONFIG_CMDLINE

[Simon Glass]

Hi Tom,

On Thu, 19 Oct 2023 at 07:57, Tom Rini <trini@konsulko.com> wrote:
>
> On Thu, Oct 19, 2023 at 08:00:00AM -0600, Simon Glass wrote:
> > Hi Tom,
> >
> > On Wed, 18 Oct 2023 at 20:32, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 16, 2023 at 04:28:23PM -0600, Simon Glass wrote:
> > >
> > > > Now that everything is working, add a test to make sure that this
> > > > builds correctly.
> > > >
> > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > Reviewed-by: Tom Rini <trini@konsulko.com>
> > > > ---
> > > >
> > > > Changes in v3:
> > > > - Rebase on Tom's LONGHELP series
> > > > - Correct 'of' typo
> > > >
> > > >  test/py/tests/test_sandbox_opts.py | 20 ++++++++++++++++++++
> > > >  1 file changed, 20 insertions(+)
> > > >  create mode 100644 test/py/tests/test_sandbox_opts.py
> > >
> > > This is not doing the equivalent of:
> > > make sandbox_config
> > > sed -i -e 's/CONFIG_CMDLINE=y/CONFIG_CMDLINE=n/' .config
> > > make oldconfig
> > > make all
> >
> > Buildman should do that automatically...is something failing?
>
> Yes. https://source.denx.de/u-boot/u-boot/-/jobs/716843 fails, but
> shouldn't, and it's only the no LTO job, so maybe there's some conflict
> there?

For me this was the EFI patch to make it depend on net. Perhaps you
hit a different one?

I am seeing build errors with ~CMDLINE ~LTO with your series.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 3369 bytes --]

On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> Hi,
> 
> On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> <takahiro.akashi@linaro.org> wrote:
> >
> > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > Hi Heinrich,
> > >
> > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > >
> > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > >
> > > > >> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > >> available, add it as an explicit dependency.
> > > > >>
> > > > >> Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > >> ---
> > > > >>
> > > > >> (no changes since v2)
> > > > >>
> > > > >> Changes in v2:
> > > > >> - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > >>
> > > > >>   lib/efi_loader/Kconfig | 1 +
> > > > >>   1 file changed, 1 insertion(+)
> > > > >>
> > > > >> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > >> index 13cad6342c36..fca4b3eef270 100644
> > > > >> --- a/lib/efi_loader/Kconfig
> > > > >> +++ b/lib/efi_loader/Kconfig
> > > > >> @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > >>      # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > >>      depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > >>      depends on BLK
> > > > >> +    depends on DM_ETH
> > > > >>      depends on !EFI_APP
> > > > >>      default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > >>      select CHARSET
> > > > >
> > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > dependencies (and re-organize the code as needed) around networking?
> > > > >
> > > >
> > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > >
> > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > CONFIG_DM_ETH.
> > > >
> > > > Why is this not sufficient?
> > > > Is there a configuration that does not build?
> > >
> > > The point of this series is to disable CMDLINE and fix up what breaks.
> > >
> > > In this case we have some sort of breakage...perhaps Tom has already
> > > found it, but otherwise could you take a look?
> > >
> > > We should be able to disable NET and LTO in sandbox and still build.
> > > But this fails at present[1]. You can try it on -master
> >
> > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > Then, we could drop "depends on DM_ETH".
> 
> Strange that it only happens on the non-LTO board, though?

There's two issues. The first of which is that I think you need to
re-check your error exactly? With my series, and LTO also disabled the
problem is a call to efi_get_image_parameters() as that's defined in
cmd/bootefi.c, but also only used with cmdline invocations. So we can
fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
around that, and then discard efi_dp_from_name() entirely.

The second issue is that with LTO we more completely find the cases
where if x() calls y() and y() is undefined but nothing calls x() we can
just discard x() and not care that y() is undefined.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Heinrich Schuchardt]

On 10/21/23 20:26, Tom Rini wrote:
> On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
>> Hi,
>>
>> On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
>> <takahiro.akashi@linaro.org> wrote:
>>>
>>> On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
>>>> Hi Heinrich,
>>>>
>>>> On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
>>>>>
>>>>> On 10/17/23 16:09, Tom Rini wrote:
>>>>>> On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
>>>>>>
>>>>>>> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
>>>>>>> available, add it as an explicit dependency.
>>>>>>>
>>>>>>> Signed-off-by: Simon Glass <sjg@chromium.org>
>>>>>>> ---
>>>>>>>
>>>>>>> (no changes since v2)
>>>>>>>
>>>>>>> Changes in v2:
>>>>>>> - Add new patch to update EFI_LOADER to depend on DM_ETH
>>>>>>>
>>>>>>>    lib/efi_loader/Kconfig | 1 +
>>>>>>>    1 file changed, 1 insertion(+)
>>>>>>>
>>>>>>> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
>>>>>>> index 13cad6342c36..fca4b3eef270 100644
>>>>>>> --- a/lib/efi_loader/Kconfig
>>>>>>> +++ b/lib/efi_loader/Kconfig
>>>>>>> @@ -11,6 +11,7 @@ config EFI_LOADER
>>>>>>>       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
>>>>>>>       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
>>>>>>>       depends on BLK
>>>>>>> +    depends on DM_ETH
>>>>>>>       depends on !EFI_APP
>>>>>>>       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
>>>>>>>       select CHARSET
>>>>>>
>>>>>> Does this work for you Heinrich, or do you want to clarify the
>>>>>> dependencies (and re-organize the code as needed) around networking?
>>>>>>
>>>>>
>>>>> We should be able to boot via EFI on devices without U-Boot network support.
>>>>>
>>>>> We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
>>>>> eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
>>>>> CONFIG_DM_ETH.
>>>>>
>>>>> Why is this not sufficient?
>>>>> Is there a configuration that does not build?
>>>>
>>>> The point of this series is to disable CMDLINE and fix up what breaks.
>>>>
>>>> In this case we have some sort of breakage...perhaps Tom has already
>>>> found it, but otherwise could you take a look?
>>>>
>>>> We should be able to disable NET and LTO in sandbox and still build.
>>>> But this fails at present[1]. You can try it on -master
>>>
>>> Obviously, it would be necessary to enclose efi_dp_from_eth()
>>> with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
>>> Then, we could drop "depends on DM_ETH".
>>
>> Strange that it only happens on the non-LTO board, though?
>
> There's two issues. The first of which is that I think you need to
> re-check your error exactly? With my series, and LTO also disabled the
> problem is a call to efi_get_image_parameters() as that's defined in
> cmd/bootefi.c, but also only used with cmdline invocations. So we can
> fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> around that, and then discard efi_dp_from_name() entirely.
>
> The second issue is that with LTO we more completely find the cases
> where if x() calls y() and y() is undefined but nothing calls x() we can
> just discard x() and not care that y() is undefined.
>

I will send a patch for function efi_dp_from_eth().

@Simon

One thing that I don't understand is why we don't let the linker
eliminate the unused functions on the sandbox.

On other architectures we put each function into a separate text section
and let the linker eliminate the unused text sections:

arch/riscv/config.mk:29:
PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
LDFLAGS_u-boot          += --gc-sections -static -pie

Shouldn't we keep the sandbox close to what other architectures do?

Best regards

Heinrich

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 4719 bytes --]

On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> On 10/21/23 20:26, Tom Rini wrote:
> > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > Hi,
> > > 
> > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > <takahiro.akashi@linaro.org> wrote:
> > > > 
> > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > Hi Heinrich,
> > > > > 
> > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > 
> > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > 
> > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > available, add it as an explicit dependency.
> > > > > > > > 
> > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > ---
> > > > > > > > 
> > > > > > > > (no changes since v2)
> > > > > > > > 
> > > > > > > > Changes in v2:
> > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > 
> > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > 
> > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > >       depends on BLK
> > > > > > > > +    depends on DM_ETH
> > > > > > > >       depends on !EFI_APP
> > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > >       select CHARSET
> > > > > > > 
> > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > 
> > > > > > 
> > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > 
> > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > CONFIG_DM_ETH.
> > > > > > 
> > > > > > Why is this not sufficient?
> > > > > > Is there a configuration that does not build?
> > > > > 
> > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > 
> > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > found it, but otherwise could you take a look?
> > > > > 
> > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > But this fails at present[1]. You can try it on -master
> > > > 
> > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > Then, we could drop "depends on DM_ETH".
> > > 
> > > Strange that it only happens on the non-LTO board, though?
> > 
> > There's two issues. The first of which is that I think you need to
> > re-check your error exactly? With my series, and LTO also disabled the
> > problem is a call to efi_get_image_parameters() as that's defined in
> > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > around that, and then discard efi_dp_from_name() entirely.
> > 
> > The second issue is that with LTO we more completely find the cases
> > where if x() calls y() and y() is undefined but nothing calls x() we can
> > just discard x() and not care that y() is undefined.
> > 
> 
> I will send a patch for function efi_dp_from_eth().

There's no problem with efi_dp_from_eth as far as I can tell.

> @Simon
> 
> One thing that I don't understand is why we don't let the linker
> eliminate the unused functions on the sandbox.
> 
> On other architectures we put each function into a separate text section
> and let the linker eliminate the unused text sections:
> 
> arch/riscv/config.mk:29:
> PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> LDFLAGS_u-boot          += --gc-sections -static -pie
> 
> Shouldn't we keep the sandbox close to what other architectures do?

Oh my, I didn't realize that sandbox was missing the garbage collection
stuff.  Yes, that needs to be fixed first, then we can see what's next
to change, as there are some issues (my series first fixed CMDLINE=n on
qemu_arm64).

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 5184 bytes --]

On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > On 10/21/23 20:26, Tom Rini wrote:
> > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > Hi,
> > > > 
> > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > <takahiro.akashi@linaro.org> wrote:
> > > > > 
> > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > Hi Heinrich,
> > > > > > 
> > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > 
> > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > 
> > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > 
> > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > ---
> > > > > > > > > 
> > > > > > > > > (no changes since v2)
> > > > > > > > > 
> > > > > > > > > Changes in v2:
> > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > 
> > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > 
> > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > >       depends on BLK
> > > > > > > > > +    depends on DM_ETH
> > > > > > > > >       depends on !EFI_APP
> > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > >       select CHARSET
> > > > > > > > 
> > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > 
> > > > > > > 
> > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > 
> > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > CONFIG_DM_ETH.
> > > > > > > 
> > > > > > > Why is this not sufficient?
> > > > > > > Is there a configuration that does not build?
> > > > > > 
> > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > 
> > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > found it, but otherwise could you take a look?
> > > > > > 
> > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > But this fails at present[1]. You can try it on -master
> > > > > 
> > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > Then, we could drop "depends on DM_ETH".
> > > > 
> > > > Strange that it only happens on the non-LTO board, though?
> > > 
> > > There's two issues. The first of which is that I think you need to
> > > re-check your error exactly? With my series, and LTO also disabled the
> > > problem is a call to efi_get_image_parameters() as that's defined in
> > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > around that, and then discard efi_dp_from_name() entirely.
> > > 
> > > The second issue is that with LTO we more completely find the cases
> > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > just discard x() and not care that y() is undefined.
> > > 
> > 
> > I will send a patch for function efi_dp_from_eth().
> 
> There's no problem with efi_dp_from_eth as far as I can tell.
> 
> > @Simon
> > 
> > One thing that I don't understand is why we don't let the linker
> > eliminate the unused functions on the sandbox.
> > 
> > On other architectures we put each function into a separate text section
> > and let the linker eliminate the unused text sections:
> > 
> > arch/riscv/config.mk:29:
> > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > LDFLAGS_u-boot          += --gc-sections -static -pie
> > 
> > Shouldn't we keep the sandbox close to what other architectures do?
> 
> Oh my, I didn't realize that sandbox was missing the garbage collection
> stuff.  Yes, that needs to be fixed first, then we can see what's next
> to change, as there are some issues (my series first fixed CMDLINE=n on
> qemu_arm64).

My super quick pass at enabling
function-sections/data-sections/gc-sections shows there's nothing
further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
was looking at before.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi,

On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
>
> On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > On 10/21/23 20:26, Tom Rini wrote:
> > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > Hi,
> > > > >
> > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > >
> > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > Hi Heinrich,
> > > > > > >
> > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > >
> > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > >
> > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > >
> > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > ---
> > > > > > > > > >
> > > > > > > > > > (no changes since v2)
> > > > > > > > > >
> > > > > > > > > > Changes in v2:
> > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > >
> > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > >
> > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > >       depends on BLK
> > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > >       select CHARSET
> > > > > > > > >
> > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > >
> > > > > > > >
> > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > >
> > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > CONFIG_DM_ETH.
> > > > > > > >
> > > > > > > > Why is this not sufficient?
> > > > > > > > Is there a configuration that does not build?
> > > > > > >
> > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > >
> > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > found it, but otherwise could you take a look?
> > > > > > >
> > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > But this fails at present[1]. You can try it on -master
> > > > > >
> > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > Then, we could drop "depends on DM_ETH".
> > > > >
> > > > > Strange that it only happens on the non-LTO board, though?
> > > >
> > > > There's two issues. The first of which is that I think you need to
> > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > around that, and then discard efi_dp_from_name() entirely.
> > > >
> > > > The second issue is that with LTO we more completely find the cases
> > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > just discard x() and not care that y() is undefined.
> > > >
> > >
> > > I will send a patch for function efi_dp_from_eth().
> >
> > There's no problem with efi_dp_from_eth as far as I can tell.
> >
> > > @Simon
> > >
> > > One thing that I don't understand is why we don't let the linker
> > > eliminate the unused functions on the sandbox.
> > >
> > > On other architectures we put each function into a separate text section
> > > and let the linker eliminate the unused text sections:
> > >
> > > arch/riscv/config.mk:29:
> > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > >
> > > Shouldn't we keep the sandbox close to what other architectures do?
> >
> > Oh my, I didn't realize that sandbox was missing the garbage collection
> > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > to change, as there are some issues (my series first fixed CMDLINE=n on
> > qemu_arm64).
>
> My super quick pass at enabling
> function-sections/data-sections/gc-sections shows there's nothing
> further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> was looking at before.

I am not sure about the original reason, but sandbox probably
pre-dates wholesale enabling of gc-sections (I haven't looked). There
is also the issue of whether the host toolchain supports it. Sandbox
is supposed to run on a variety of OS types.

I am also not sure of the point ot it...since it is normally pretty
easy to use Kconfig to control features.

At least for me 'u-boot -D' doesn't work with Heinrich's patch. Do we
really need to make this change? It will rapidly devolve to the point
that coming back would be extremely painful.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Heinrich Schuchardt]

On 10/22/23 23:55, Simon Glass wrote:
> Hi,
>
> On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
>>
>> On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
>>> On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
>>>> On 10/21/23 20:26, Tom Rini wrote:
>>>>> On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
>>>>>> Hi,
>>>>>>
>>>>>> On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
>>>>>> <takahiro.akashi@linaro.org> wrote:
>>>>>>>
>>>>>>> On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
>>>>>>>> Hi Heinrich,
>>>>>>>>
>>>>>>>> On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
>>>>>>>>>
>>>>>>>>> On 10/17/23 16:09, Tom Rini wrote:
>>>>>>>>>> On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
>>>>>>>>>>
>>>>>>>>>>> Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
>>>>>>>>>>> available, add it as an explicit dependency.
>>>>>>>>>>>
>>>>>>>>>>> Signed-off-by: Simon Glass <sjg@chromium.org>
>>>>>>>>>>> ---
>>>>>>>>>>>
>>>>>>>>>>> (no changes since v2)
>>>>>>>>>>>
>>>>>>>>>>> Changes in v2:
>>>>>>>>>>> - Add new patch to update EFI_LOADER to depend on DM_ETH
>>>>>>>>>>>
>>>>>>>>>>>     lib/efi_loader/Kconfig | 1 +
>>>>>>>>>>>     1 file changed, 1 insertion(+)
>>>>>>>>>>>
>>>>>>>>>>> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
>>>>>>>>>>> index 13cad6342c36..fca4b3eef270 100644
>>>>>>>>>>> --- a/lib/efi_loader/Kconfig
>>>>>>>>>>> +++ b/lib/efi_loader/Kconfig
>>>>>>>>>>> @@ -11,6 +11,7 @@ config EFI_LOADER
>>>>>>>>>>>        # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
>>>>>>>>>>>        depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
>>>>>>>>>>>        depends on BLK
>>>>>>>>>>> +    depends on DM_ETH
>>>>>>>>>>>        depends on !EFI_APP
>>>>>>>>>>>        default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
>>>>>>>>>>>        select CHARSET
>>>>>>>>>>
>>>>>>>>>> Does this work for you Heinrich, or do you want to clarify the
>>>>>>>>>> dependencies (and re-organize the code as needed) around networking?
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> We should be able to boot via EFI on devices without U-Boot network support.
>>>>>>>>>
>>>>>>>>> We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
>>>>>>>>> eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
>>>>>>>>> CONFIG_DM_ETH.
>>>>>>>>>
>>>>>>>>> Why is this not sufficient?
>>>>>>>>> Is there a configuration that does not build?
>>>>>>>>
>>>>>>>> The point of this series is to disable CMDLINE and fix up what breaks.
>>>>>>>>
>>>>>>>> In this case we have some sort of breakage...perhaps Tom has already
>>>>>>>> found it, but otherwise could you take a look?
>>>>>>>>
>>>>>>>> We should be able to disable NET and LTO in sandbox and still build.
>>>>>>>> But this fails at present[1]. You can try it on -master
>>>>>>>
>>>>>>> Obviously, it would be necessary to enclose efi_dp_from_eth()
>>>>>>> with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
>>>>>>> Then, we could drop "depends on DM_ETH".
>>>>>>
>>>>>> Strange that it only happens on the non-LTO board, though?
>>>>>
>>>>> There's two issues. The first of which is that I think you need to
>>>>> re-check your error exactly? With my series, and LTO also disabled the
>>>>> problem is a call to efi_get_image_parameters() as that's defined in
>>>>> cmd/bootefi.c, but also only used with cmdline invocations. So we can
>>>>> fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
>>>>> around that, and then discard efi_dp_from_name() entirely.
>>>>>
>>>>> The second issue is that with LTO we more completely find the cases
>>>>> where if x() calls y() and y() is undefined but nothing calls x() we can
>>>>> just discard x() and not care that y() is undefined.
>>>>>
>>>>
>>>> I will send a patch for function efi_dp_from_eth().
>>>
>>> There's no problem with efi_dp_from_eth as far as I can tell.
>>>
>>>> @Simon
>>>>
>>>> One thing that I don't understand is why we don't let the linker
>>>> eliminate the unused functions on the sandbox.
>>>>
>>>> On other architectures we put each function into a separate text section
>>>> and let the linker eliminate the unused text sections:
>>>>
>>>> arch/riscv/config.mk:29:
>>>> PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
>>>> LDFLAGS_u-boot          += --gc-sections -static -pie
>>>>
>>>> Shouldn't we keep the sandbox close to what other architectures do?
>>>
>>> Oh my, I didn't realize that sandbox was missing the garbage collection
>>> stuff.  Yes, that needs to be fixed first, then we can see what's next
>>> to change, as there are some issues (my series first fixed CMDLINE=n on
>>> qemu_arm64).
>>
>> My super quick pass at enabling
>> function-sections/data-sections/gc-sections shows there's nothing
>> further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
>> was looking at before.
>
> I am not sure about the original reason, but sandbox probably
> pre-dates wholesale enabling of gc-sections (I haven't looked). There
> is also the issue of whether the host toolchain supports it. Sandbox
> is supposed to run on a variety of OS types.
>
> I am also not sure of the point ot it...since it is normally pretty
> easy to use Kconfig to control features.
>
> At least for me 'u-boot -D' doesn't work with Heinrich's patch. Do we
> really need to make this change? It will rapidly devolve to the point
> that coming back would be extremely painful.

It just shows where the code is not clean.

Linker generated lists referenced by the corresponding macros are
compiled in (as shows up in u-boot.map). But you have decided not to use
those macros to access the sandbox command line options. Now they are
removed as unreferenced garbage.

Can't we use the linker generated list macros instead of duplicating code?

* define SANDBOX_CMDLINE_OPT via ll_entry_declare().
* replace ll_entry_count() instead of __u_boot_sandbox_option_count()
* use ll_start(), ll_end() for iterating

Best regards

Heinrich

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 6831 bytes --]

On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> Hi,
> 
> On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > Hi,
> > > > > >
> > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > >
> > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > Hi Heinrich,
> > > > > > > >
> > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > >
> > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > >
> > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > >
> > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > ---
> > > > > > > > > > >
> > > > > > > > > > > (no changes since v2)
> > > > > > > > > > >
> > > > > > > > > > > Changes in v2:
> > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > >
> > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > >
> > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > >       depends on BLK
> > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > >       select CHARSET
> > > > > > > > > >
> > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > >
> > > > > > > > >
> > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > >
> > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > >
> > > > > > > > > Why is this not sufficient?
> > > > > > > > > Is there a configuration that does not build?
> > > > > > > >
> > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > >
> > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > found it, but otherwise could you take a look?
> > > > > > > >
> > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > >
> > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > >
> > > > > > Strange that it only happens on the non-LTO board, though?
> > > > >
> > > > > There's two issues. The first of which is that I think you need to
> > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > >
> > > > > The second issue is that with LTO we more completely find the cases
> > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > just discard x() and not care that y() is undefined.
> > > > >
> > > >
> > > > I will send a patch for function efi_dp_from_eth().
> > >
> > > There's no problem with efi_dp_from_eth as far as I can tell.
> > >
> > > > @Simon
> > > >
> > > > One thing that I don't understand is why we don't let the linker
> > > > eliminate the unused functions on the sandbox.
> > > >
> > > > On other architectures we put each function into a separate text section
> > > > and let the linker eliminate the unused text sections:
> > > >
> > > > arch/riscv/config.mk:29:
> > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > >
> > > > Shouldn't we keep the sandbox close to what other architectures do?
> > >
> > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > qemu_arm64).
> >
> > My super quick pass at enabling
> > function-sections/data-sections/gc-sections shows there's nothing
> > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > was looking at before.
> 
> I am not sure about the original reason, but sandbox probably
> pre-dates wholesale enabling of gc-sections (I haven't looked). There
> is also the issue of whether the host toolchain supports it. Sandbox
> is supposed to run on a variety of OS types.

It doesn't predate gc-sections, we've just never moved it to the
top-level Makefiles (and disabled on LTO) like we should have a long
time ago.  It's supported by all the compilers we support, and it's a
required feature (or, LTO) for U-Boot.

> I am also not sure of the point ot it...since it is normally pretty
> easy to use Kconfig to control features.

That's not the point of the feature.  The point is that we can discard
unused code without having to rely on #ifdef and __maybe_unused.

> At least for me 'u-boot -D' doesn't work with Heinrich's patch. Do we
> really need to make this change? It will rapidly devolve to the point
> that coming back would be extremely painful.

Do we absolutely need which, gc-sections or LTO? Yes, we do.  And I
don't see immediately how that changes --default_fdt ?

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 7465 bytes --]

On Mon, Oct 23, 2023 at 12:55:34AM +0200, Heinrich Schuchardt wrote:
> On 10/22/23 23:55, Simon Glass wrote:
> > Hi,
> > 
> > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > > 
> > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > Hi,
> > > > > > > 
> > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > > 
> > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > Hi Heinrich,
> > > > > > > > > 
> > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > > 
> > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > > 
> > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > > 
> > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > ---
> > > > > > > > > > > > 
> > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > > 
> > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > > 
> > > > > > > > > > > >     lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > >     1 file changed, 1 insertion(+)
> > > > > > > > > > > > 
> > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > >        # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > >        depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > >        depends on BLK
> > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > >        depends on !EFI_APP
> > > > > > > > > > > >        default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > >        select CHARSET
> > > > > > > > > > > 
> > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > > 
> > > > > > > > > > 
> > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > > 
> > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > > 
> > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > > 
> > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > > 
> > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > > 
> > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > > 
> > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > > 
> > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > > 
> > > > > > There's two issues. The first of which is that I think you need to
> > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > > 
> > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > just discard x() and not care that y() is undefined.
> > > > > > 
> > > > > 
> > > > > I will send a patch for function efi_dp_from_eth().
> > > > 
> > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > > 
> > > > > @Simon
> > > > > 
> > > > > One thing that I don't understand is why we don't let the linker
> > > > > eliminate the unused functions on the sandbox.
> > > > > 
> > > > > On other architectures we put each function into a separate text section
> > > > > and let the linker eliminate the unused text sections:
> > > > > 
> > > > > arch/riscv/config.mk:29:
> > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > > 
> > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > > 
> > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > qemu_arm64).
> > > 
> > > My super quick pass at enabling
> > > function-sections/data-sections/gc-sections shows there's nothing
> > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > was looking at before.
> > 
> > I am not sure about the original reason, but sandbox probably
> > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > is also the issue of whether the host toolchain supports it. Sandbox
> > is supposed to run on a variety of OS types.
> > 
> > I am also not sure of the point ot it...since it is normally pretty
> > easy to use Kconfig to control features.
> > 
> > At least for me 'u-boot -D' doesn't work with Heinrich's patch. Do we
> > really need to make this change? It will rapidly devolve to the point
> > that coming back would be extremely painful.
> 
> It just shows where the code is not clean.
> 
> Linker generated lists referenced by the corresponding macros are
> compiled in (as shows up in u-boot.map). But you have decided not to use
> those macros to access the sandbox command line options. Now they are
> removed as unreferenced garbage.
> 
> Can't we use the linker generated list macros instead of duplicating code?
> 
> * define SANDBOX_CMDLINE_OPT via ll_entry_declare().
> * replace ll_entry_count() instead of __u_boot_sandbox_option_count()
> * use ll_start(), ll_end() for iterating

Oh, it comes down to something like this?  I don't recall if we had
__used the first time we put this on sandbox, but we do now and we use
it for cases where LTO thinks it knows best.  We should be able to do
that too for sandbox.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Tom,

On Sun, 22 Oct 2023 at 16:45, Tom Rini <trini@konsulko.com> wrote:
>
> On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> > Hi,
> >
> > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > Hi,
> > > > > > >
> > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > >
> > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > Hi Heinrich,
> > > > > > > > >
> > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > >
> > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > >
> > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > >
> > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > ---
> > > > > > > > > > > >
> > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > >
> > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > >
> > > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > > >
> > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > >       depends on BLK
> > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > >       select CHARSET
> > > > > > > > > > >
> > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > >
> > > > > > > > > >
> > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > >
> > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > >
> > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > >
> > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > >
> > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > >
> > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > >
> > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > >
> > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > >
> > > > > > There's two issues. The first of which is that I think you need to
> > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > >
> > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > just discard x() and not care that y() is undefined.
> > > > > >
> > > > >
> > > > > I will send a patch for function efi_dp_from_eth().
> > > >
> > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > >
> > > > > @Simon
> > > > >
> > > > > One thing that I don't understand is why we don't let the linker
> > > > > eliminate the unused functions on the sandbox.
> > > > >
> > > > > On other architectures we put each function into a separate text section
> > > > > and let the linker eliminate the unused text sections:
> > > > >
> > > > > arch/riscv/config.mk:29:
> > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > >
> > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > >
> > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > qemu_arm64).
> > >
> > > My super quick pass at enabling
> > > function-sections/data-sections/gc-sections shows there's nothing
> > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > was looking at before.
> >
> > I am not sure about the original reason, but sandbox probably
> > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > is also the issue of whether the host toolchain supports it. Sandbox
> > is supposed to run on a variety of OS types.
>
> It doesn't predate gc-sections, we've just never moved it to the
> top-level Makefiles (and disabled on LTO) like we should have a long
> time ago.  It's supported by all the compilers we support, and it's a
> required feature (or, LTO) for U-Boot.

Remember that the sandbox linker script is special in that it adds to
the existing default one provided by the toolchain

>
> > I am also not sure of the point ot it...since it is normally pretty
> > easy to use Kconfig to control features.
>
> That's not the point of the feature.  The point is that we can discard
> unused code without having to rely on #ifdef and __maybe_unused.

That doesn't match with my understanding. An unused static will still
produce a warning if it is unused.

Re #ifdef, you will still see code discarded with sandbox if  'if
IS_ENABLED()' is used. It's just that the link-time

Sandbox doesn't really have a need to discard unused code. It doesn't
make much difference to the binary size, from what I can see, partly
because we enable as much as we can.

No LTO:
before: 3652159 220840 126576 3999575 3d0757 /tmp/b/sandbox/u-boot
after: 3515872 204016 124528 3844416 3aa940 /tmp/b/sandbox/u-boot
Saving 3.7%

With LTO:
before: 3391148 212224 124704 3728076 38e2cc /tmp/b/sandbox/u-boot
after: 3356324 200664 124704 3681692 382d9c /tmp/b/sandbox/u-boot
Saving 1%

That reflects the fact that LTO is quite a bit more powerful than linker hacks.

The other thing I have noticed over the years is that linker
discarding can change behaviour when unrelated changes are made. For
example, if I refactor a function the linker decides that a function
it calls is now needed, whereas before it was not. That sort of thing
is annoying with sandbox.

>
> > At least for me 'u-boot -D' doesn't work with Heinrich's patch. Do we
> > really need to make this change? It will rapidly devolve to the point
> > that coming back would be extremely painful.
>
> Do we absolutely need which, gc-sections or LTO? Yes, we do.  And I
> don't see immediately how that changes --default_fdt ?

I hope that with all the clang effort we might be closer to having
sandbox behave well with this change. But I still don't see the point.

Regards,
SImon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 9718 bytes --]

On Mon, Oct 23, 2023 at 12:05:28AM -0700, Simon Glass wrote:
> Hi Tom,
> 
> On Sun, 22 Oct 2023 at 16:45, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> > > Hi,
> > >
> > > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > > >
> > > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > > Hi,
> > > > > > > >
> > > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > > >
> > > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > > Hi Heinrich,
> > > > > > > > > >
> > > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > > >
> > > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > > >
> > > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > > >
> > > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > > ---
> > > > > > > > > > > > >
> > > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > > >
> > > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > > >
> > > > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > > > >
> > > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > > >       depends on BLK
> > > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > > >       select CHARSET
> > > > > > > > > > > >
> > > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > > >
> > > > > > > > > > >
> > > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > > >
> > > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > > >
> > > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > > >
> > > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > > >
> > > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > > >
> > > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > > >
> > > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > > >
> > > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > > >
> > > > > > > There's two issues. The first of which is that I think you need to
> > > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > > >
> > > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > > just discard x() and not care that y() is undefined.
> > > > > > >
> > > > > >
> > > > > > I will send a patch for function efi_dp_from_eth().
> > > > >
> > > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > > >
> > > > > > @Simon
> > > > > >
> > > > > > One thing that I don't understand is why we don't let the linker
> > > > > > eliminate the unused functions on the sandbox.
> > > > > >
> > > > > > On other architectures we put each function into a separate text section
> > > > > > and let the linker eliminate the unused text sections:
> > > > > >
> > > > > > arch/riscv/config.mk:29:
> > > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > > >
> > > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > > >
> > > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > > qemu_arm64).
> > > >
> > > > My super quick pass at enabling
> > > > function-sections/data-sections/gc-sections shows there's nothing
> > > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > > was looking at before.
> > >
> > > I am not sure about the original reason, but sandbox probably
> > > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > > is also the issue of whether the host toolchain supports it. Sandbox
> > > is supposed to run on a variety of OS types.
> >
> > It doesn't predate gc-sections, we've just never moved it to the
> > top-level Makefiles (and disabled on LTO) like we should have a long
> > time ago.  It's supported by all the compilers we support, and it's a
> > required feature (or, LTO) for U-Boot.
> 
> Remember that the sandbox linker script is special in that it adds to
> the existing default one provided by the toolchain

Yes, and we've also been using gc-sections since 2008, from a quick look
at the logs.

> > > I am also not sure of the point ot it...since it is normally pretty
> > > easy to use Kconfig to control features.
> >
> > That's not the point of the feature.  The point is that we can discard
> > unused code without having to rely on #ifdef and __maybe_unused.
> 
> That doesn't match with my understanding. An unused static will still
> produce a warning if it is unused.

Yes, you're misunderstanding how this feature works and why it's
required.  With v4 of my series, on sandbox and CMDLINE=n and LTO=n:
/usr/bin/ld: boot/scene_textline.o: in function `scene_textline_send_key':
/home/trini/work/u-boot/u-boot/boot/scene_textline.c:157: undefined reference to `cread_line_process_ch'
/usr/bin/ld: boot/scene_textline.o: in function `scene_textline_open':
/home/trini/work/u-boot/u-boot/boot/scene_textline.c:222: undefined reference to `cli_cread_init'
/usr/bin/ld: common/bootstage.o: in function `bootstage_fdt_add_report':
/home/trini/work/u-boot/u-boot/common/bootstage.c:323: undefined reference to `working_fdt'
/usr/bin/ld: common/cli.o: in function `run_commandf':
/home/trini/work/u-boot/u-boot/common/cli.c:154: undefined reference to `run_command'
/usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
/home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
/usr/bin/ld: drivers/fastboot/fb_command.o: in function `fastboot_acmd_complete':
/home/trini/work/u-boot/u-boot/drivers/fastboot/fb_command.c:350: undefined reference to `run_command'
/usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
/home/trini/work/u-boot/u-boot/lib/efi_loader/efi_device_path.c:1095: undefined reference to `efi_get_image_parameters'
/usr/bin/ld: net/net.o: in function `net_auto_load':
/home/trini/work/u-boot/u-boot/net/net.c:349: undefined reference to `tftp_start'
collect2: error: ld returned 1 exit status

And with re-enabling gc-sections:
/usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
/home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
collect2: error: ld returned 1 exit status

Because all of those other link failures are in unreachable code and the
rest of the code base is already setup to know that x() can call y() and
y() can just be undefined at link time if x() is never called and ends
up discarded.

I honestly didn't realize sandbox hadn't had this critical feature
enabled and I hope we haven't been doing too many odd shuffles of the
codebase because of it.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Tom,

On Mon, 23 Oct 2023 at 06:16, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 23, 2023 at 12:05:28AM -0700, Simon Glass wrote:
> > Hi Tom,
> >
> > On Sun, 22 Oct 2023 at 16:45, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> > > > Hi,
> > > >
> > > > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > > > >
> > > > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > > > Hi,
> > > > > > > > >
> > > > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > > > >
> > > > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > > > Hi Heinrich,
> > > > > > > > > > >
> > > > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > > > >
> > > > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > > > >
> > > > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > > > ---
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > > > >
> > > > > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > > > >       depends on BLK
> > > > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > > > >       select CHARSET
> > > > > > > > > > > > >
> > > > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > > > >
> > > > > > > > > > > >
> > > > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > > > >
> > > > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > > > >
> > > > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > > > >
> > > > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > > > >
> > > > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > > > >
> > > > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > > > >
> > > > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > > > >
> > > > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > > > >
> > > > > > > > There's two issues. The first of which is that I think you need to
> > > > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > > > >
> > > > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > > > just discard x() and not care that y() is undefined.
> > > > > > > >
> > > > > > >
> > > > > > > I will send a patch for function efi_dp_from_eth().
> > > > > >
> > > > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > > > >
> > > > > > > @Simon
> > > > > > >
> > > > > > > One thing that I don't understand is why we don't let the linker
> > > > > > > eliminate the unused functions on the sandbox.
> > > > > > >
> > > > > > > On other architectures we put each function into a separate text section
> > > > > > > and let the linker eliminate the unused text sections:
> > > > > > >
> > > > > > > arch/riscv/config.mk:29:
> > > > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > > > >
> > > > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > > > >
> > > > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > > > qemu_arm64).
> > > > >
> > > > > My super quick pass at enabling
> > > > > function-sections/data-sections/gc-sections shows there's nothing
> > > > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > > > was looking at before.
> > > >
> > > > I am not sure about the original reason, but sandbox probably
> > > > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > > > is also the issue of whether the host toolchain supports it. Sandbox
> > > > is supposed to run on a variety of OS types.
> > >
> > > It doesn't predate gc-sections, we've just never moved it to the
> > > top-level Makefiles (and disabled on LTO) like we should have a long
> > > time ago.  It's supported by all the compilers we support, and it's a
> > > required feature (or, LTO) for U-Boot.
> >
> > Remember that the sandbox linker script is special in that it adds to
> > the existing default one provided by the toolchain
>
> Yes, and we've also been using gc-sections since 2008, from a quick look
> at the logs.

Right, I just mean that sandbox does not use a standalone linker
script. It is not common to garbage-collect sections in executables,
so far as I know.

>
> > > > I am also not sure of the point ot it...since it is normally pretty
> > > > easy to use Kconfig to control features.
> > >
> > > That's not the point of the feature.  The point is that we can discard
> > > unused code without having to rely on #ifdef and __maybe_unused.
> >
> > That doesn't match with my understanding. An unused static will still
> > produce a warning if it is unused.
>
> Yes, you're misunderstanding how this feature works and why it's
> required.  With v4 of my series, on sandbox and CMDLINE=n and LTO=n:
> /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_send_key':
> /home/trini/work/u-boot/u-boot/boot/scene_textline.c:157: undefined reference to `cread_line_process_ch'
> /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_open':
> /home/trini/work/u-boot/u-boot/boot/scene_textline.c:222: undefined reference to `cli_cread_init'
> /usr/bin/ld: common/bootstage.o: in function `bootstage_fdt_add_report':
> /home/trini/work/u-boot/u-boot/common/bootstage.c:323: undefined reference to `working_fdt'
> /usr/bin/ld: common/cli.o: in function `run_commandf':
> /home/trini/work/u-boot/u-boot/common/cli.c:154: undefined reference to `run_command'
> /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> /usr/bin/ld: drivers/fastboot/fb_command.o: in function `fastboot_acmd_complete':
> /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_command.c:350: undefined reference to `run_command'
> /usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> /home/trini/work/u-boot/u-boot/lib/efi_loader/efi_device_path.c:1095: undefined reference to `efi_get_image_parameters'
> /usr/bin/ld: net/net.o: in function `net_auto_load':
> /home/trini/work/u-boot/u-boot/net/net.c:349: undefined reference to `tftp_start'
> collect2: error: ld returned 1 exit status
>
> And with re-enabling gc-sections:
> /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> collect2: error: ld returned 1 exit status
>
> Because all of those other link failures are in unreachable code and the
> rest of the code base is already setup to know that x() can call y() and
> y() can just be undefined at link time if x() is never called and ends
> up discarded.

All of those problems were corrected by my series. Perhaps that
accounts for the difference between them?

Just to take one example, working_fdt, while it might seem convenient
to leave it in cmd/fdt.c it doesn't actually make sense, since the var
is set when booting. Without that var DT fixups can't happen. So why
not pick up my patch that moves it to boot/fdt_support.c where it
belongs?

Looking at the list of errors, we should just fix them. They all
indicate real problems with the structure of the code. Another example
is fastboot which needs a condition for CMDLINE.

>
> I honestly didn't realize sandbox hadn't had this critical feature
> enabled and I hope we haven't been doing too many odd shuffles of the
> codebase because of it.

I'm a bit unsure where you are heading with this. Ultimate we don't
want to disable all the functionality when CMDLINE is disabled. I
think with the addition of 5-6 patches in your series you can fix it.

Regards,
Simon

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Tom Rini]

[-- Attachment #1: Type: text/plain, Size: 13226 bytes --]

On Mon, Oct 23, 2023 at 10:13:54AM -0700, Simon Glass wrote:
> Hi Tom,
> 
> On Mon, 23 Oct 2023 at 06:16, Tom Rini <trini@konsulko.com> wrote:
> >
> > On Mon, Oct 23, 2023 at 12:05:28AM -0700, Simon Glass wrote:
> > > Hi Tom,
> > >
> > > On Sun, 22 Oct 2023 at 16:45, Tom Rini <trini@konsulko.com> wrote:
> > > >
> > > > On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> > > > > Hi,
> > > > >
> > > > > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > > > > >
> > > > > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > > > > Hi,
> > > > > > > > > >
> > > > > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > > > > >
> > > > > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > > > > Hi Heinrich,
> > > > > > > > > > > >
> > > > > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > > > > >
> > > > > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > > > > ---
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > > > > >       depends on BLK
> > > > > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > > > > >       select CHARSET
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > > > > >
> > > > > > > > > > > > >
> > > > > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > > > > >
> > > > > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > > > > >
> > > > > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > > > > >
> > > > > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > > > > >
> > > > > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > > > > >
> > > > > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > > > > >
> > > > > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > > > > >
> > > > > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > > > > >
> > > > > > > > > There's two issues. The first of which is that I think you need to
> > > > > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > > > > >
> > > > > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > > > > just discard x() and not care that y() is undefined.
> > > > > > > > >
> > > > > > > >
> > > > > > > > I will send a patch for function efi_dp_from_eth().
> > > > > > >
> > > > > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > > > > >
> > > > > > > > @Simon
> > > > > > > >
> > > > > > > > One thing that I don't understand is why we don't let the linker
> > > > > > > > eliminate the unused functions on the sandbox.
> > > > > > > >
> > > > > > > > On other architectures we put each function into a separate text section
> > > > > > > > and let the linker eliminate the unused text sections:
> > > > > > > >
> > > > > > > > arch/riscv/config.mk:29:
> > > > > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > > > > >
> > > > > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > > > > >
> > > > > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > > > > qemu_arm64).
> > > > > >
> > > > > > My super quick pass at enabling
> > > > > > function-sections/data-sections/gc-sections shows there's nothing
> > > > > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > > > > was looking at before.
> > > > >
> > > > > I am not sure about the original reason, but sandbox probably
> > > > > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > > > > is also the issue of whether the host toolchain supports it. Sandbox
> > > > > is supposed to run on a variety of OS types.
> > > >
> > > > It doesn't predate gc-sections, we've just never moved it to the
> > > > top-level Makefiles (and disabled on LTO) like we should have a long
> > > > time ago.  It's supported by all the compilers we support, and it's a
> > > > required feature (or, LTO) for U-Boot.
> > >
> > > Remember that the sandbox linker script is special in that it adds to
> > > the existing default one provided by the toolchain
> >
> > Yes, and we've also been using gc-sections since 2008, from a quick look
> > at the logs.
> 
> Right, I just mean that sandbox does not use a standalone linker
> script.

Yes it does, arch/sandbox/cpu/u-boot.lds.

> It is not common to garbage-collect sections in executables,
> so far as I know.

I'm not sure how that's relevant here, honestly. And we didn't ask the
toolchain community to invent this for us, it's a standard feature that
we've found helpful for a number of reasons.

> > > > > I am also not sure of the point ot it...since it is normally pretty
> > > > > easy to use Kconfig to control features.
> > > >
> > > > That's not the point of the feature.  The point is that we can discard
> > > > unused code without having to rely on #ifdef and __maybe_unused.
> > >
> > > That doesn't match with my understanding. An unused static will still
> > > produce a warning if it is unused.
> >
> > Yes, you're misunderstanding how this feature works and why it's
> > required.  With v4 of my series, on sandbox and CMDLINE=n and LTO=n:
> > /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_send_key':
> > /home/trini/work/u-boot/u-boot/boot/scene_textline.c:157: undefined reference to `cread_line_process_ch'
> > /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_open':
> > /home/trini/work/u-boot/u-boot/boot/scene_textline.c:222: undefined reference to `cli_cread_init'
> > /usr/bin/ld: common/bootstage.o: in function `bootstage_fdt_add_report':
> > /home/trini/work/u-boot/u-boot/common/bootstage.c:323: undefined reference to `working_fdt'
> > /usr/bin/ld: common/cli.o: in function `run_commandf':
> > /home/trini/work/u-boot/u-boot/common/cli.c:154: undefined reference to `run_command'
> > /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> > /usr/bin/ld: drivers/fastboot/fb_command.o: in function `fastboot_acmd_complete':
> > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_command.c:350: undefined reference to `run_command'
> > /usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> > /home/trini/work/u-boot/u-boot/lib/efi_loader/efi_device_path.c:1095: undefined reference to `efi_get_image_parameters'
> > /usr/bin/ld: net/net.o: in function `net_auto_load':
> > /home/trini/work/u-boot/u-boot/net/net.c:349: undefined reference to `tftp_start'
> > collect2: error: ld returned 1 exit status
> >
> > And with re-enabling gc-sections:
> > /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> > collect2: error: ld returned 1 exit status
> >
> > Because all of those other link failures are in unreachable code and the
> > rest of the code base is already setup to know that x() can call y() and
> > y() can just be undefined at link time if x() is never called and ends
> > up discarded.
> 
> All of those problems were corrected by my series. Perhaps that
> accounts for the difference between them?

All of these are not real problems at this point in time is my point.

> Just to take one example, working_fdt, while it might seem convenient
> to leave it in cmd/fdt.c it doesn't actually make sense, since the var
> is set when booting. Without that var DT fixups can't happen. So why
> not pick up my patch that moves it to boot/fdt_support.c where it
> belongs?

And the issue is that when the command line is disabled NOTHING uses
"working_fdt". We don't need it. It would be discarded by real
platforms, if we moved it.

To rephrase things another way, skimming the code, that
bootm_find_images() does:
	if (IS_ENABLED(CONFIG_CMD_FDT))
		set_working_fdt_addr(map_to_sysmem(images.ft_addr));

And so we never set "working_fdt" when CMDLINE is off (since CMD_FDT is
off) IS the bug, and that needs to be fixed. That's when you should
figure out where "working_fdt" should be (and also maybe re-name it or
the local variable named "working_fdt" that the pxe_utils.c code uses)
and not before then.

> Looking at the list of errors, we should just fix them. They all
> indicate real problems with the structure of the code.

No, they do not. The codebase assume that we discard unreachable
functions. This allows us to NOT have to use #ifdefs and also not have
to make otherwise unreasonable code splits.

> Another example
> is fastboot which needs a condition for CMDLINE.

This is an example where yes, the functionality requires CMDLINE.
Because reading the code shows that fastboot will literally run
"bootcmd", and so requires that to be available.

> > I honestly didn't realize sandbox hadn't had this critical feature
> > enabled and I hope we haven't been doing too many odd shuffles of the
> > codebase because of it.
> 
> I'm a bit unsure where you are heading with this. Ultimate we don't
> want to disable all the functionality when CMDLINE is disabled. I
> think with the addition of 5-6 patches in your series you can fix it.

Where I'm heading with this is to say that sandbox must work like a real
platform does and discard unreachable code. You don't know what code
is or is not really needed for functionality X and needs to be moved for
it to not rely on what's in the command file vs elsewhere until you act
like other platforms do.

-- 
Tom

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 659 bytes --]

Re: [PATCH v3 22/32] efi: Update EFI_LOADER to depend on DM_ETH

[Simon Glass]

Hi Tom,

On Mon, 23 Oct 2023 at 10:45, Tom Rini <trini@konsulko.com> wrote:
>
> On Mon, Oct 23, 2023 at 10:13:54AM -0700, Simon Glass wrote:
> > Hi Tom,
> >
> > On Mon, 23 Oct 2023 at 06:16, Tom Rini <trini@konsulko.com> wrote:
> > >
> > > On Mon, Oct 23, 2023 at 12:05:28AM -0700, Simon Glass wrote:
> > > > Hi Tom,
> > > >
> > > > On Sun, 22 Oct 2023 at 16:45, Tom Rini <trini@konsulko.com> wrote:
> > > > >
> > > > > On Sun, Oct 22, 2023 at 02:55:32PM -0700, Simon Glass wrote:
> > > > > > Hi,
> > > > > >
> > > > > > On Sun, 22 Oct 2023 at 07:59, Tom Rini <trini@konsulko.com> wrote:
> > > > > > >
> > > > > > > On Sun, Oct 22, 2023 at 10:29:22AM -0400, Tom Rini wrote:
> > > > > > > > On Sun, Oct 22, 2023 at 08:08:11AM +0200, Heinrich Schuchardt wrote:
> > > > > > > > > On 10/21/23 20:26, Tom Rini wrote:
> > > > > > > > > > On Sat, Oct 21, 2023 at 08:43:08AM -0700, Simon Glass wrote:
> > > > > > > > > > > Hi,
> > > > > > > > > > >
> > > > > > > > > > > On Thu, 19 Oct 2023 at 17:30, AKASHI Takahiro
> > > > > > > > > > > <takahiro.akashi@linaro.org> wrote:
> > > > > > > > > > > >
> > > > > > > > > > > > On Thu, Oct 19, 2023 at 08:01:11AM -0600, Simon Glass wrote:
> > > > > > > > > > > > > Hi Heinrich,
> > > > > > > > > > > > >
> > > > > > > > > > > > > On Wed, 18 Oct 2023 at 06:55, Heinrich Schuchardt <xypron.glpk@gmx.de> wrote:
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > On 10/17/23 16:09, Tom Rini wrote:
> > > > > > > > > > > > > > > On Mon, Oct 16, 2023 at 04:28:13PM -0600, Simon Glass wrote:
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > > Since efi_device_path.c calls eth_get_dev() and assumes that Ethernet is
> > > > > > > > > > > > > > > > available, add it as an explicit dependency.
> > > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > > > > > > > > > > > > > > ---
> > > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > > (no changes since v2)
> > > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > > Changes in v2:
> > > > > > > > > > > > > > > > - Add new patch to update EFI_LOADER to depend on DM_ETH
> > > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > >    lib/efi_loader/Kconfig | 1 +
> > > > > > > > > > > > > > > >    1 file changed, 1 insertion(+)
> > > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > > diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > > index 13cad6342c36..fca4b3eef270 100644
> > > > > > > > > > > > > > > > --- a/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > > +++ b/lib/efi_loader/Kconfig
> > > > > > > > > > > > > > > > @@ -11,6 +11,7 @@ config EFI_LOADER
> > > > > > > > > > > > > > > >       # We need EFI_STUB_32BIT to be set on x86_32 with EFI_STUB
> > > > > > > > > > > > > > > >       depends on !EFI_STUB || !X86 || X86_64 || EFI_STUB_32BIT
> > > > > > > > > > > > > > > >       depends on BLK
> > > > > > > > > > > > > > > > +    depends on DM_ETH
> > > > > > > > > > > > > > > >       depends on !EFI_APP
> > > > > > > > > > > > > > > >       default y if !ARM || SYS_CPU = armv7 || SYS_CPU = armv8
> > > > > > > > > > > > > > > >       select CHARSET
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > > > Does this work for you Heinrich, or do you want to clarify the
> > > > > > > > > > > > > > > dependencies (and re-organize the code as needed) around networking?
> > > > > > > > > > > > > > >
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > We should be able to boot via EFI on devices without U-Boot network support.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > We already use IS_ENABLED(CONFIG_NETDEVICES) to avoid invoking
> > > > > > > > > > > > > > eth_get_dev() if there is no network. CONFIG_NETDEVICES=y selects
> > > > > > > > > > > > > > CONFIG_DM_ETH.
> > > > > > > > > > > > > >
> > > > > > > > > > > > > > Why is this not sufficient?
> > > > > > > > > > > > > > Is there a configuration that does not build?
> > > > > > > > > > > > >
> > > > > > > > > > > > > The point of this series is to disable CMDLINE and fix up what breaks.
> > > > > > > > > > > > >
> > > > > > > > > > > > > In this case we have some sort of breakage...perhaps Tom has already
> > > > > > > > > > > > > found it, but otherwise could you take a look?
> > > > > > > > > > > > >
> > > > > > > > > > > > > We should be able to disable NET and LTO in sandbox and still build.
> > > > > > > > > > > > > But this fails at present[1]. You can try it on -master
> > > > > > > > > > > >
> > > > > > > > > > > > Obviously, it would be necessary to enclose efi_dp_from_eth()
> > > > > > > > > > > > with "if defined(CONFIG_NETDEVICES)" (or DM_ETH).
> > > > > > > > > > > > Then, we could drop "depends on DM_ETH".
> > > > > > > > > > >
> > > > > > > > > > > Strange that it only happens on the non-LTO board, though?
> > > > > > > > > >
> > > > > > > > > > There's two issues. The first of which is that I think you need to
> > > > > > > > > > re-check your error exactly? With my series, and LTO also disabled the
> > > > > > > > > > problem is a call to efi_get_image_parameters() as that's defined in
> > > > > > > > > > cmd/bootefi.c, but also only used with cmdline invocations. So we can
> > > > > > > > > > fix that CMDLINE=n && LTO=n case with a IS_ENABLED(CONFIG_CMDLINE)
> > > > > > > > > > around that, and then discard efi_dp_from_name() entirely.
> > > > > > > > > >
> > > > > > > > > > The second issue is that with LTO we more completely find the cases
> > > > > > > > > > where if x() calls y() and y() is undefined but nothing calls x() we can
> > > > > > > > > > just discard x() and not care that y() is undefined.
> > > > > > > > > >
> > > > > > > > >
> > > > > > > > > I will send a patch for function efi_dp_from_eth().
> > > > > > > >
> > > > > > > > There's no problem with efi_dp_from_eth as far as I can tell.
> > > > > > > >
> > > > > > > > > @Simon
> > > > > > > > >
> > > > > > > > > One thing that I don't understand is why we don't let the linker
> > > > > > > > > eliminate the unused functions on the sandbox.
> > > > > > > > >
> > > > > > > > > On other architectures we put each function into a separate text section
> > > > > > > > > and let the linker eliminate the unused text sections:
> > > > > > > > >
> > > > > > > > > arch/riscv/config.mk:29:
> > > > > > > > > PLATFORM_RELFLAGS       += -fno-common -ffunction-sections -fdata-sections
> > > > > > > > > LDFLAGS_u-boot          += --gc-sections -static -pie
> > > > > > > > >
> > > > > > > > > Shouldn't we keep the sandbox close to what other architectures do?
> > > > > > > >
> > > > > > > > Oh my, I didn't realize that sandbox was missing the garbage collection
> > > > > > > > stuff.  Yes, that needs to be fixed first, then we can see what's next
> > > > > > > > to change, as there are some issues (my series first fixed CMDLINE=n on
> > > > > > > > qemu_arm64).
> > > > > > >
> > > > > > > My super quick pass at enabling
> > > > > > > function-sections/data-sections/gc-sections shows there's nothing
> > > > > > > further needed for CMDLINE=n and LTO=n on sandbox, not even the part I
> > > > > > > was looking at before.
> > > > > >
> > > > > > I am not sure about the original reason, but sandbox probably
> > > > > > pre-dates wholesale enabling of gc-sections (I haven't looked). There
> > > > > > is also the issue of whether the host toolchain supports it. Sandbox
> > > > > > is supposed to run on a variety of OS types.
> > > > >
> > > > > It doesn't predate gc-sections, we've just never moved it to the
> > > > > top-level Makefiles (and disabled on LTO) like we should have a long
> > > > > time ago.  It's supported by all the compilers we support, and it's a
> > > > > required feature (or, LTO) for U-Boot.
> > > >
> > > > Remember that the sandbox linker script is special in that it adds to
> > > > the existing default one provided by the toolchain
> > >
> > > Yes, and we've also been using gc-sections since 2008, from a quick look
> > > at the logs.
> >
> > Right, I just mean that sandbox does not use a standalone linker
> > script.
>
> Yes it does, arch/sandbox/cpu/u-boot.lds.
>
> > It is not common to garbage-collect sections in executables,
> > so far as I know.
>
> I'm not sure how that's relevant here, honestly. And we didn't ask the
> toolchain community to invent this for us, it's a standard feature that
> we've found helpful for a number of reasons.
>
> > > > > > I am also not sure of the point ot it...since it is normally pretty
> > > > > > easy to use Kconfig to control features.
> > > > >
> > > > > That's not the point of the feature.  The point is that we can discard
> > > > > unused code without having to rely on #ifdef and __maybe_unused.
> > > >
> > > > That doesn't match with my understanding. An unused static will still
> > > > produce a warning if it is unused.
> > >
> > > Yes, you're misunderstanding how this feature works and why it's
> > > required.  With v4 of my series, on sandbox and CMDLINE=n and LTO=n:
> > > /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_send_key':
> > > /home/trini/work/u-boot/u-boot/boot/scene_textline.c:157: undefined reference to `cread_line_process_ch'
> > > /usr/bin/ld: boot/scene_textline.o: in function `scene_textline_open':
> > > /home/trini/work/u-boot/u-boot/boot/scene_textline.c:222: undefined reference to `cli_cread_init'
> > > /usr/bin/ld: common/bootstage.o: in function `bootstage_fdt_add_report':
> > > /home/trini/work/u-boot/u-boot/common/bootstage.c:323: undefined reference to `working_fdt'
> > > /usr/bin/ld: common/cli.o: in function `run_commandf':
> > > /home/trini/work/u-boot/u-boot/common/cli.c:154: undefined reference to `run_command'
> > > /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> > > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> > > /usr/bin/ld: drivers/fastboot/fb_command.o: in function `fastboot_acmd_complete':
> > > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_command.c:350: undefined reference to `run_command'
> > > /usr/bin/ld: lib/efi_loader/efi_device_path.o: in function `efi_dp_from_name':
> > > /home/trini/work/u-boot/u-boot/lib/efi_loader/efi_device_path.c:1095: undefined reference to `efi_get_image_parameters'
> > > /usr/bin/ld: net/net.o: in function `net_auto_load':
> > > /home/trini/work/u-boot/u-boot/net/net.c:349: undefined reference to `tftp_start'
> > > collect2: error: ld returned 1 exit status
> > >
> > > And with re-enabling gc-sections:
> > > /usr/bin/ld: drivers/fastboot/fb_common.o: in function `fastboot_boot':
> > > /home/trini/work/u-boot/u-boot/drivers/fastboot/fb_common.c:137: undefined reference to `run_command'
> > > collect2: error: ld returned 1 exit status
> > >
> > > Because all of those other link failures are in unreachable code and the
> > > rest of the code base is already setup to know that x() can call y() and
> > > y() can just be undefined at link time if x() is never called and ends
> > > up discarded.
> >
> > All of those problems were corrected by my series. Perhaps that
> > accounts for the difference between them?
>
> All of these are not real problems at this point in time is my point.
>
> > Just to take one example, working_fdt, while it might seem convenient
> > to leave it in cmd/fdt.c it doesn't actually make sense, since the var
> > is set when booting. Without that var DT fixups can't happen. So why
> > not pick up my patch that moves it to boot/fdt_support.c where it
> > belongs?
>
> And the issue is that when the command line is disabled NOTHING uses
> "working_fdt". We don't need it. It would be discarded by real
> platforms, if we moved it.
>
> To rephrase things another way, skimming the code, that
> bootm_find_images() does:
>         if (IS_ENABLED(CONFIG_CMD_FDT))
>                 set_working_fdt_addr(map_to_sysmem(images.ft_addr));
>
> And so we never set "working_fdt" when CMDLINE is off (since CMD_FDT is
> off) IS the bug, and that needs to be fixed. That's when you should
> figure out where "working_fdt" should be (and also maybe re-name it or
> the local variable named "working_fdt" that the pxe_utils.c code uses)
> and not before then.
>
> > Looking at the list of errors, we should just fix them. They all
> > indicate real problems with the structure of the code.
>
> No, they do not. The codebase assume that we discard unreachable
> functions. This allows us to NOT have to use #ifdefs and also not have
> to make otherwise unreasonable code splits.
>
> > Another example
> > is fastboot which needs a condition for CMDLINE.
>
> This is an example where yes, the functionality requires CMDLINE.
> Because reading the code shows that fastboot will literally run
> "bootcmd", and so requires that to be available.
>
> > > I honestly didn't realize sandbox hadn't had this critical feature
> > > enabled and I hope we haven't been doing too many odd shuffles of the
> > > codebase because of it.
> >
> > I'm a bit unsure where you are heading with this. Ultimate we don't
> > want to disable all the functionality when CMDLINE is disabled. I
> > think with the addition of 5-6 patches in your series you can fix it.
>
> Where I'm heading with this is to say that sandbox must work like a real
> platform does and discard unreachable code. You don't know what code
> is or is not really needed for functionality X and needs to be moved for
> it to not rely on what's in the command file vs elsewhere until you act
> like other platforms do.

It is remarkable that this sandbox 'feature' remained undiscovered for so long.

I'm going to let this thread die. I think I have said everything worth
saying at this point.

Regards,
Simon