On Wed, Aug 30, 2023 at 09:48:48PM -0600, Simon Glass wrote:
Hi AKASHI,
On Tue, 22 Aug 2023 at 19:50, AKASHI Takahiro takahiro.akashi@linaro.org wrote:
The output from "dm tree" or "dm uclass" is a bit annoying if the number of devices available on the system is huge. (This is especially true on sandbox when I debug some DM code.)
With this patch, we can specify the uclass name or the device name that we are interested in in order to limit the output.
For instance,
=> dm uclass usb uclass 121: usb 0 usb@1 @ 0bcff8b0, seq 1
uclass 124: usb
=> dm tree usb:usb@1 Class Index Probed Driver Name
usb 0 [ ] usb_sandbox usb@1 usb_hub 0 [ ] usb_hub `-- hub usb_emul 0 [ ] usb_sandbox_hub `-- hub-emul usb_emul 1 [ ] usb_sandbox_flash |-- flash-stick@0 usb_emul 2 [ ] usb_sandbox_flash |-- flash-stick@1 usb_emul 3 [ ] usb_sandbox_flash |-- flash-stick@2 usb_emul 4 [ ] usb_sandbox_keyb `-- keyb@3
If you want forward-matching against a uclass or udevice name, you can specify "-e" option.
I don't really know what 'forward matching' means.
Really? I believed that forward-matching was a common word. I meant that it searches for any string starting with a specific sub-string. In other words, it would be "^<sub-string>" in a regular expression. So, for example, "usb" should match with "usbABC", "usb-DEF", "usb_GHI" and so on, but not match with "ABCusb".
Please use forward instead of forword in the code
=> dm uclass -e usb uclass 15: usb_emul 0 hub-emul @ 0bcffb00, seq 0 1 flash-stick@0 @ 0bcffc30, seq 1 2 flash-stick@1 @ 0bcffdc0, seq 2 3 flash-stick@2 @ 0bcfff50, seq 3 4 keyb@3 @ 0bd000e0, seq 4
uclass 64: usb_mass_storage
uclass 121: usb 0 usb@1 @ 0bcff8b0, seq 1
uclass 122: usb_dev_generic
uclass 123: usb_hub 0 hub @ 0bcff9b0, seq 0
uclass 124: usb
=> dm tree -e usb Class Index Probed Driver Name
usb 0 [ ] usb_sandbox usb@1 usb_hub 0 [ ] usb_hub `-- hub usb_emul 0 [ ] usb_sandbox_hub `-- hub-emul usb_emul 1 [ ] usb_sandbox_flash |-- flash-stick@0 usb_emul 2 [ ] usb_sandbox_flash |-- flash-stick@1 usb_emul 3 [ ] usb_sandbox_flash |-- flash-stick@2 usb_emul 4 [ ] usb_sandbox_keyb `-- keyb@3
Signed-off-by: AKASHI Takahiro takahiro.akashi@linaro.org
v2
- allow for forward-matching against the name
- update command doc
cmd/dm.c | 48 ++++++++++++++---- doc/usage/cmd/dm.rst | 30 ++++++++++- drivers/core/dump.c | 116 ++++++++++++++++++++++++++++++++----------- include/dm/util.h | 15 ++++-- 4 files changed, 165 insertions(+), 44 deletions(-)
Reviewed-by: Simon Glass sjg@chromium.org
Thanks for doing this. It will be a big time-saver. I also wonder if it would be better if the default were to do a substring search and
Initially, I implemented so, but felt it is annoying to see (sometimes many) unexpected matched devices, especially when you know the exact name of device. See my example "dm uclass -e usb".
you have to add a flag to search for a single device?
Does 'single' mean the first matched word or exactly-same one?
Either way, I don't have a strong opinion here, though.
Thanks, -Takahiro Akashi
See below
diff --git a/cmd/dm.c b/cmd/dm.c index 3263547cbec6..1aa86aab9c1c 100644 --- a/cmd/dm.c +++ b/cmd/dm.c @@ -59,11 +59,26 @@ static int do_dm_dump_static_driver_info(struct cmd_tbl *cmdtp, int flag, static int do_dm_dump_tree(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[]) {
bool sort;sort = argc > 1 && !strcmp(argv[1], "-s");dm_dump_tree(sort);
bool extended = false, sort = false;char *device = NULL;for (; argc > 1; argc--, argv++) {if (argv[1][0] != '-')break;if (!strcmp(argv[1], "-e")) {extended = true;} else if (!strcmp(argv[1], "-s")) {sort = true;} else {printf("Unknown parameter: %s\n", argv[1]);return 0;}}if (argc > 1)device = argv[1];dm_dump_tree(device, extended, sort); return 0;} @@ -71,7 +86,20 @@ static int do_dm_dump_tree(struct cmd_tbl *cmdtp, int flag, int argc, static int do_dm_dump_uclass(struct cmd_tbl *cmdtp, int flag, int argc, char *const argv[]) {
dm_dump_uclass();
bool extended = false;char *uclass = NULL;if (argc > 1) {if (!strcmp(argv[1], "-e")) {extended = true;argc--;argv++;}if (argc > 1)uclass = argv[1];}dm_dump_uclass(uclass, extended); return 0;} @@ -91,8 +119,8 @@ static char dm_help_text[] = "dm drivers Dump list of drivers with uclass and instances\n" DM_MEM_HELP "dm static Dump list of drivers with static platform data\n"
"dm tree [-s] Dump tree of driver model devices (-s=sort)\n""dm uclass Dump list of instances for each uclass"
"dm tree [-s][-e][name] Dump tree of driver model devices (-s=sort)\n""dm uclass [-e][name] Dump list of instances for each uclass" ;#endif
@@ -102,5 +130,5 @@ U_BOOT_CMD_WITH_SUBCMDS(dm, "Driver model low level access", dm_help_text, U_BOOT_SUBCMD_MKENT(drivers, 1, 1, do_dm_dump_drivers), DM_MEM U_BOOT_SUBCMD_MKENT(static, 1, 1, do_dm_dump_static_driver_info),
U_BOOT_SUBCMD_MKENT(tree, 2, 1, do_dm_dump_tree),U_BOOT_SUBCMD_MKENT(uclass, 1, 1, do_dm_dump_uclass));
U_BOOT_SUBCMD_MKENT(tree, 4, 1, do_dm_dump_tree),U_BOOT_SUBCMD_MKENT(uclass, 3, 1, do_dm_dump_uclass));diff --git a/doc/usage/cmd/dm.rst b/doc/usage/cmd/dm.rst index 74c6b01e3619..12b7edeed685 100644 --- a/doc/usage/cmd/dm.rst +++ b/doc/usage/cmd/dm.rst @@ -12,8 +12,8 @@ Synopis dm devres dm drivers dm static
- dm tree [-s]
- dm uclass
- dm tree [-s][-e] [uclass name]
- dm uclass [-e] [udevice name]
Description
@@ -127,6 +127,12 @@ If -s is given, the top-level devices (those which are children of the root device) are shown sorted in order of uclass ID, so it is easier to find a particular device type.
+If -e is given, forward-matching against existing devices is +made and only the matched devices are shown.
I don't understand this...can you expand it a bit so it is clear what -e does and what forward-matching is?
+If a device name is given, forward-matching against existing devices is +made and only the matched devices are shown.
dm uclass
Regards, Simon