Properly determine libusb read size for large reports (Fixes #274) by sudobash1 · Pull Request #728 · libusb/hidapi

sudobash1 · 2025-03-25T20:27:04Z

Currently the libusb version of hidapi simply reads up to wMaxPacketSize bytes as the report. This is problematic when reports are longer than wMaxPacketSize. The current behavior will split that report up.

The proper solution is to review the report descriptor to find the longest input report and use that as the length of the libusb_fill_interrupt_transfer buffer. (Note: there is no need to manually get multiple USB packets and concatenate them together to fit the report length. USB already handles that for us.)

This will still work for HID devices when some input reports are shorter than others. The HID device will just send a short packet terminator and libusb will give us the shorter buffer.

The substance of these changes is in the get_max_input_size method. It uses the same basic report descriptor parsing as get_usage. I considered changing the code so that there could be shared parsing code, but I decided that was overkill for this.

Fixes #274

sudobash1 · 2025-03-25T22:08:47Z

PR updated to remove different signedness comparison compiler warning when compiling with -Wall.

Be-ing · 2025-03-26T01:07:51Z

Thank you for finally taking care of this longstanding bug.

Youw

Thanks for this, I'll perform a thorrow review a bit later.
@Be-ing can you confirm it fixes the issue with long reports too?

JoergAtGithub · 2025-03-26T18:21:25Z

+   Requires an opened device with *claimed interface*.
+
+   The return value is the size on success and -1 on failure. */
+static size_t get_max_input_size(libusb_device_handle *handle, int interface_num, uint16_t expected_report_descriptor_size)


This function seems to be the functional the same as InputReportByteLength from https://learn.microsoft.com/en-us/windows-hardware/drivers/ddi/hidpi/ns-hidpi-_hidp_caps on Windows.

We have a lot of test data for the Windows unit test, which stores the result of this function:

hidapi/windows/test/data/045E_02FF_0005_0001.pp_data

Line 21 in 0ab6c14

pp_data->caps_info[0]->ReportByteLength = 16

together with the real ReportDescriptor https://github.com/libusb/hidapi/blob/master/windows/test/data/045E_02FF_0005_0001_real.rpt_desc

What do you think about creating a unit test for the new get_max_input_size function using the _real.rpt_desc files as input and compare the result with the value stored in the .pp_data.

Nice! That does look like a great dataset. I'll see if I can figure a clean way to make a unit test, and see how it fares with that data.

Currently the get_max_input_size is a static function, but I need a way to use it externally in the unit test. It seems that the windows library has some extensions prefixed with winapi. Should I rename the get max function to hid_libusbapi_get_max_input_report_size and mark it HID_API_EXPORT_CALL?

I will also need to change how it gets the report descriptor, but that won't be a problem.

Making it with HID_API_EXPORT_CALL would make it a public (at least on binary level) function, which is undesirable, if it is required for tests only. Maybe export it only when building unit-tests etc. or try to avoid it by using static linking or smth else (or have it in an internal header file?)

Have a look at the Windows unit test ( https://github.com/libusb/hidapi/tree/master/windows/test ) for the report descriptor reconstructor. There we had exactly the same challenges, hid_winapi_descriptor_reconstruct_pp_data was the internal function to test there.

@JoergAtGithub I did actually look at hid_winapi_descriptor_reconstruct_pp_data, and it is marked as HID_API_EXPORT_CALL in the header file:

hidapi/windows/hidapi_winapi.h

Line 68 in 0ab6c14

int HID_API_EXPORT_CALL hid_winapi_descriptor_reconstruct_pp_data(void *hidp_preparsed_data, unsigned char *buf, size_t buf_size);

Although, now that I am looking, it is not marked in the C file:

hidapi/windows/hidapi_descriptor_reconstruct.c

Line 181 in 0ab6c14

int hid_winapi_descriptor_reconstruct_pp_data(void *preparsed_data, unsigned char *buf, size_t buf_size)

WRT @Youw's suggestion. I could simplify the function a bit so that it doesn't depend on anything in hid.c and then move it to hid_max_input_report.h. If that is acceptable, it is probably the easiest solution.

hid_winapi_descriptor_reconstruct_pp_data, and it is marked as HID_API_EXPORT_CALL in the header file

Right - that is a public API function. That is also the reason why there is a winapi prefix in the name.

it is not marked in the C file

Not required, if it is marked in the header.

I could simplify the function a bit so that it doesn't depend on anything in hid.c and then move it to hid_max_input_report.h. If that is acceptable, it is probably the easiest solution.

Yes, sounds like simples solution for now. Go for it.

I have a test, but it isn't passing. It could be that I am misunderstanding the Windows HID structures. I have only a vague understanding of them.

For example, in the 046A_0011_0006_0001 test data, it gives a max input size of 9 bytes:

hidapi/windows/test/data/046A_0011_0006_0001.pp_data

Lines 16 to 19 in 0ab6c14

pp_data->caps_info[0]->FirstCap = 0

pp_data->caps_info[0]->LastCap = 2

pp_data->caps_info[0]->NumberOfCaps = 2

pp_data->caps_info[0]->ReportByteLength = 9

But two input caps (cap[0] and cap[1]) are:

hidapi/windows/test/data/046A_0011_0006_0001.pp_data

Lines 37 to 38 in 0ab6c14

pp_data->cap[0]->BitSize = 1

pp_data->cap[0]->ReportCount = 8

hidapi/windows/test/data/046A_0011_0006_0001.pp_data

Lines 83 to 84 in 0ab6c14

pp_data->cap[1]->BitSize = 8

pp_data->cap[1]->ReportCount = 6

According to the documentation I read, the report size is just BitSize * ReportCount (plus a byte for the report number I assume).

This gives report sizes of (1*8)/8 + 1 = 2 and (8*6)/8 + 1 = 7. So I don't know where the ReportByteLength = 9 value is coming from.

Do either of you know what I am missing here?

mcuee · 2025-03-27T05:12:38Z

BTW, there is a typo here.
@Youw and @JoergAtGithub.

hidapi/windows/test/data/046A_0011_0006_0001.pp_data

Line 4 in 0ab6c14

    
           dev->manufacturer_string = "dev->product_string      = "dev->release_number      = 0x0100

dev->manufacturer_string = "dev->product_string      = "dev->release_number      = 0x0100

JoergAtGithub · 2025-03-27T06:17:16Z

I put 046A_0011_0006_0001_real.rpt_desc into https://eleccelerator.com/usbdescreqparser/ and got:

0x05, 0x01,        // Usage Page (Generic Desktop Ctrls)
0x09, 0x06,        // Usage (Keyboard)
0xA1, 0x01,        // Collection (Application)
0x05, 0x07,        //   Usage Page (Kbrd/Keypad)
0x19, 0xE0,        //   Usage Minimum (0xE0)
0x29, 0xE7,        //   Usage Maximum (0xE7)
0x15, 0x00,        //   Logical Minimum (0)
0x25, 0x01,        //   Logical Maximum (1)
0x75, 0x01,        //   Report Size (1)
0x95, 0x08,        //   Report Count (8)
0x81, 0x02,        //   Input (Data,Var,Abs,No Wrap,Linear,Preferred State,No Null Position)
0x75, 0x08,        //   Report Size (8)
0x95, 0x01,        //   Report Count (1)
0x81, 0x03,        //   Input (Const,Var,Abs,No Wrap,Linear,Preferred State,No Null Position)
0x19, 0x00,        //   Usage Minimum (0x00)
0x29, 0xDD,        //   Usage Maximum (0xDD)
0x15, 0x00,        //   Logical Minimum (0)
0x26, 0xDD, 0x00,  //   Logical Maximum (221)
0x75, 0x08,        //   Report Size (8)
0x95, 0x06,        //   Report Count (6)
0x81, 0x00,        //   Input (Data,Array,Abs,No Wrap,Linear,Preferred State,No Null Position)
0x05, 0x08,        //   Usage Page (LEDs)
0x19, 0x01,        //   Usage Minimum (Num Lock)
0x29, 0x03,        //   Usage Maximum (Scroll Lock)
0x15, 0x00,        //   Logical Minimum (0)
0x25, 0x01,        //   Logical Maximum (1)
0x75, 0x01,        //   Report Size (1)
0x95, 0x03,        //   Report Count (3)
0x91, 0x02,        //   Output (Data,Var,Abs,No Wrap,Linear,Preferred State,No Null Position,Non-volatile)
0x75, 0x05,        //   Report Size (5)
0x95, 0x01,        //   Report Count (1)
0x91, 0x03,        //   Output (Const,Var,Abs,No Wrap,Linear,Preferred State,No Null Position,Non-volatile)
0xC0,              // End Collection

The InputReport contains:

8x1bit data
1x8bit constant padding
6x8bit data
This sums to 8byte, but InputReportByteLength is defined as follows:
Specifies the maximum size, in bytes, of all the input reports. Includes the report ID, which is prepended to the report data. If report ID is not used, the ID value is zero.
So we've to add a byte for the ReportID and are at 9 bytes.

JoergAtGithub · 2025-03-27T06:20:44Z

BTW, there is a typo here. @Youw and @JoergAtGithub.

hidapi/windows/test/data/046A_0011_0006_0001.pp_data

Line 4 in 0ab6c14

dev->manufacturer_string = "dev->product_string = "dev->release_number = 0x0100
dev->manufacturer_string = "dev->product_string      = "dev->release_number      = 0x0100

There is something missing, as this is autogenerated by pp_data_dump, I guess there was something in the manufacturer_string , that pp_data_dump couldn't handle. Could you please open a dedicated issue for this, as this is unrelated to this PR.

mcuee · 2025-03-27T07:34:55Z

There is something missing, as this is autogenerated by pp_data_dump, I guess there was something in the manufacturer_string , that pp_data_dump couldn't handle. Could you please open a dedicated issue for this, as this is unrelated to this PR.

pp_data_dump issue for 046A_0011_0006_0001 -- Lower Priority #729

JoergAtGithub · 2025-03-27T19:48:22Z

+     if(NOT EXISTS "${TEST_PP_DATA}")
+          message(FATAL_ERROR "Missing '${TEST_PP_DATA}' file for '${TEST_CASE}' test case")
+     endif()
+     set(TEST_EXPECTED_DESCRIPTOR "${CMAKE_CURRENT_LIST_DIR}/../../windows/test/data/${TEST_CASE}_expected.rpt_desc")


You need to parse the _real.rpt_desc, not the _expected.rpt_desc. The real is what is dumped from the device, and the expected is what the Windows ReportDescriptor-Reconstructor generates out of the .pp_data.

I see. I assumed that the real and expected should contain equivalent reports.

The _real.rpt_desc files do not follow a consistent format, so should I parse them manually (should be pretty doable with a bit of regex) into new files and add them to the repo?

Maybe you can put them into https://eleccelerator.com/usbdescreqparser/ to unify the format. That makes them also human readable.

JoergAtGithub · 2025-03-27T19:49:30Z

+			if (report_count < 0 || report_size < 0) {
+				/* We are missing size or count. That isn't good. */
+				return 0;


Suggested change

if (report_count < 0 || report_size < 0) {

/* We are missing size or count. That isn't good. */

return 0;

if (report_count < 0 || report_size < 0) {

/* We are missing size or count. That isn't good. */

return -1;

This would mean a corrupt ReportDescriptor

I wasn't consistent with whether I was using size_t or ssize_t. In my last commit, I settled on size_t, but perhaps I should change it back to ssize_t so that we can more clearly differentiate between errors and a zero value (if there are no feature reports, it will return 0, which is not an error).

Makes sense!

sudobash1 · 2025-03-27T20:00:14Z

Thanks @JoergAtGithub for walking me through that. I was misreading the descriptor.

I have added tests for libusb using the same data as the windows tests. And I extended the functionality of the libusb method to be able to calculate the maximum output and feature report sizes as well. This has no functional use currently, but it lets us run three times as many tests since the pp_data files have all three max sizes available.

JoergAtGithub · 2025-03-27T20:10:00Z

And I extended the functionality of the libusb method to be able to calculate the maximum output and feature report sizes as well. This has no functional use currently, but it lets us run three times as many tests since the pp_data files have all three max sizes available.

Independent of this PR, I think it would generally make sense to store these 3 values in the device structure. On Windows we would have to use the values InputReportByteLength, OutputReportByteLength and FeatureReportByteLength from https://learn.microsoft.com/en-us/windows-hardware/drivers/ddi/hidpi/ns-hidpi-_hidp_caps . On all other backends we could use your parser to determine them - and with the testcase you proved that they are the same as on Windows.
In this way, we were able to programmatically determine the required buffer sizes for read/write operations on all platforms. @Youw What do you think about this?

Youw · 2025-03-28T10:31:54Z


-if(WIN32)
-    # so far only Windows has tests
+if(WIN32 OR HIDAPI_WITH_LIBUSB)


also need to update builds.yml

Youw · 2025-03-28T10:35:37Z

+set(CMAKE_VERSION_SUPPORTS_ENVIRONMENT_MODIFICATION "3.22")
+
+foreach(TEST_CASE ${HID_DESCRIPTOR_RECONSTRUCT_TEST_CASES})
+     set(TEST_PP_DATA "${CMAKE_CURRENT_LIST_DIR}/../../windows/test/data/${TEST_CASE}.pp_data")


maybe move the test data to <root>/test_data ?

Youw

I don't have a strong opinion on test implementation (I trust @JoergAtGithub on this one).
libusb implementation seem fine.

Lets make sure it runs with CI on Github Actions and we're good to go here.

Youw · 2025-03-28T10:39:49Z

And I extended the functionality of the libusb method to be able to calculate the maximum output and feature report sizes as well. This has no functional use currently, but it lets us run three times as many tests since the pp_data files have all three max sizes available.

Independent of this PR, I think it would generally make sense to store these 3 values in the device structure. On Windows we would have to use the values InputReportByteLength, OutputReportByteLength and FeatureReportByteLength from https://learn.microsoft.com/en-us/windows-hardware/drivers/ddi/hidpi/ns-hidpi-_hidp_caps . On all other backends we could use your parser to determine them - and with the testcase you proved that they are the same as on Windows. In this way, we were able to programmatically determine the required buffer sizes for read/write operations on all platforms. @Youw What do you think about this?

Lets continue here: #731

mcuee · 2025-03-28T14:46:09Z

This PR does not seem to work.

Test device is the same as the one used in Issue #274. The FW is a mod of Jan Axelson's FX2HID example and codes are included in the following #274 comment.

HID reports of more than 64 bytes and libusb backend #274 (comment)

I can reproduce the issue reported in #274 with hidapi git libusb backend, hidraw backend is okay.

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_hidraw_git --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading 129-byte input report 0, 250 msec timeout...read 128 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_libusb_git --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading 129-byte input report 0, 250 msec timeout...read 64 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00
Closing device

With this PR, no change in hidraw backend behavior, which is good.

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_hidraw_pr728 --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading 129-byte input report 0, 250 msec timeout...read 128 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

But the libusb backend fix is not working.

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ make -f Makefile_pr728.libusb 
cc -I/usr/local/include/libusb-1.0 -I ../hidapi_pr728/hidapi -c ../hidapi_pr728/libusb/hid.c -o ../hidapi_pr728/libusb/hid.o
cc -I/usr/local/include/libusb-1.0 -I ../hidapi_pr728/hidapi -c hidapitester.c -o hidapitester.o
cc -I/usr/local/include/libusb-1.0 -I ../hidapi_pr728/hidapi ../hidapi_pr728/libusb/hid.o hidapitester.o -o hidapitester -L/usr/local/lib -lusb-1.0 -pthread

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ mv hidapitester hidapitester_libusb_pr728 

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_libusb_pr728 --vidpid 0925:1234 --open --buflen 256 -l 128 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 128-bytes...wrote 128 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
Reading 128-byte input report 0, 250 msec timeout...read 0 bytes:
Closing device

mcuee · 2025-03-28T14:50:44Z

HID Report Descriptor.

ReportDscr:

	db 06h, 0A0h, 0FFh ;    Usage Page (FFA0h, vendor defined)
	db 09h, 01h     ;       Usage (vendor defined)
	db 0A1h, 01h    ;       Collection (Application)
	db 09h, 02h     ;       Usage (vendor defined)
	db 0A1h, 00h    ;       Collection (Physical)
	db 06h, 0A1h, 0FFh ;    Usage Page (vendor defined)

;; The Input report
	db 09h, 03h     ;       Usage (vendor defined)
	db 09h, 04h     ;       Usage (vendor defined)
	db 15h, 80h	;	Logical minimum (80h or -128)
	db 25h, 7Fh	;	Logical maximum (7Fh or 127)
	db 35h, 00h	;	Physical minimum (0)
	db 45h, 0FFh	;	Physical maximum (255)
	db 75h, 08h	;	Report size (8 bits)
	db 95h, 80h	;	Report count (128 fields)
	db 81h, 02h	;	Input (data, variable, absolute)

;; The Output report
	db 09h, 05h     ;       Usage (vendor defined)
	db 09h, 06h     ;       Usage (vendor defined)
	db 15h, 80h	;	Logical minimum (80h or -128)
	db 25h, 7Fh	;	Logical maximum (7Fh or 127)
	db 35h, 00h	;	Physical minimum (0)
	db 45h, 0FFh	;	Physical maximum (255)
	db 75h, 08h	;	Report size (8 bits)
	db 95h, 80h	;	Report count (128 fields)
	db 91h, 02h	;	Output (data, variable, absolute)

	db 0C0h         ;       End Collection (Physical)
	db 0C0h         ;       End Collection (Application)

ReportDscrEnd:

hidtest is okay.

mcuee@UbuntuSwift3 ~/build/hid/hidapi (master)$ sudo ./hidtest/hidtest-libusb 
hidapi test/example tool. Compiled with hidapi version 0.15.0, runtime version 0.15.0.
Compile-time version matches runtime version of hidapi.

Device Found
  type: 0925 1234
  path: 3-2.1:1.0
  serial_number: (null)
  Manufacturer: CYPRESS
  Product:      EZ-USB FX2 HID USBHIDIO
  Release:      0
  Interface:    0
  Usage (page): 0x0 (0x0)
  Bus type: 1 (USB)

  Report Descriptor: (52 bytes)
0x06, 0xa0, 0xff, 0x09, 0x01, 0xa1, 0x01, 0x09, 0x02, 0xa1, 
0x00, 0x06, 0xa1, 0xff, 0x09, 0x03, 0x09, 0x04, 0x15, 0x80, 
0x25, 0x7f, 0x35, 0x00, 0x45, 0xff, 0x75, 0x08, 0x95, 0x80, 
0x81, 0x02, 0x09, 0x05, 0x09, 0x06, 0x15, 0x80, 0x25, 0x7f, 
0x35, 0x00, 0x45, 0xff, 0x75, 0x08, 0x95, 0x80, 0x91, 0x02, 
0xc0, 0xc0, 
unable to open device

sudobash1 · 2025-03-29T18:20:37Z

If I have time, I'll try to address @JoergAtGithub and @Youw's comments on Monday.

@mcuee I tried feeding the report descriptor that you sent into get_max_report_size and it returned 129, which seems correct. I'll try to look into this on Monday as well.

I did notice a typo in your test. When you ran ./hidapitester_libusb_pr728, you specified -l 128 which should have been -l 129, but I think it should have returned data anyway.

mcuee · 2025-03-30T00:19:38Z

Let me try again.

mcuee@UbuntuSwift3 ~/build/cyusb/fxload_libusb (master)$ sudo ./fxload -v -t fx2lp -I ./fx2hid.hex -D /dev/bus/usb/003/019 
microcontroller type: fx2lp
single stage:  load on-chip memory
open RAM hexfile image ./fx2hid.hex
stop CPU
write on-chip, addr 0x0600 len  234 (0x00ea)
write on-chip, addr 0x09e5 len   10 (0x000a)
write on-chip, addr 0x0380 len  428 (0x01ac)
write on-chip, addr 0x0080 len  768 (0x0300)
write on-chip, addr 0x0033 len    3 (0x0003)
write on-chip, addr 0x09ff len    7 (0x0007)
write on-chip, addr 0x07b8 len   92 (0x005c)
write on-chip, addr 0x0851 len   59 (0x003b)
write on-chip, addr 0x05fe len    2 (0x0002)
write on-chip, addr 0x0a07 len    4 (0x0004)
write on-chip, addr 0x09ef len    8 (0x0008)
write on-chip, addr 0x09b0 len   18 (0x0012)
write on-chip, addr 0x09f7 len    8 (0x0008)
write on-chip, addr 0x09c2 len   18 (0x0012)
write on-chip, addr 0x0a0b len    6 (0x0006)
write on-chip, addr 0x091d len   40 (0x0028)
write on-chip, addr 0x096a len   24 (0x0018)
write on-chip, addr 0x06ea len   22 (0x0016)
write on-chip, addr 0x099a len   22 (0x0016)
write on-chip, addr 0x088c len   54 (0x0036)
write on-chip, addr 0x0982 len   24 (0x0018)
write on-chip, addr 0x0814 len   61 (0x003d)
write on-chip, addr 0x0a11 len   36 (0x0024)
write on-chip, addr 0x08f1 len   44 (0x002c)
write on-chip, addr 0x08c2 len   47 (0x002f)
write on-chip, addr 0x0945 len   20 (0x0014)
write on-chip, addr 0x05b8 len   70 (0x0046)
write on-chip, addr 0x0959 len   17 (0x0011)
write on-chip, addr 0x0043 len    3 (0x0003)
write on-chip, addr 0x0053 len    3 (0x0003)
write on-chip, addr 0x0700 len  184 (0x00b8)
write on-chip, addr 0x0000 len    3 (0x0003)
write on-chip, addr 0x052c len   12 (0x000c)
write on-chip, addr 0x09d4 len   17 (0x0011)
write on-chip, addr 0x0538 len  128 (0x0080)
write on-chip, addr 0x0a06 len    1 (0x0001)
... WROTE: 2497 bytes, 36 segments, avg 69
reset CPU

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_libusb_pr728 --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input 
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading 129-byte input report 0, 250 msec timeout...read 0 bytes:
Closing device

mcuee@UbuntuSwift3 ~/build/hid/hidapitester (master)$ sudo ./hidapitester_hidraw_pr728 --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input 
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading 129-byte input report 0, 250 msec timeout...read 128 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

mcuee · 2025-03-30T03:11:55Z

Same problem under FreeBSD 14.1 Release.

This is with a physical machine, Chuwi mini PC, Intel J4125 CPU, 8GB/256GB configuration.

Without this PR, hidapi git has the problem mentioned in #274.

But then somehow this PR does not work.

mcuee@freebsd14:~/build/hidapi_pr730 $ uname -a
FreeBSD freebsd14 14.1-RELEASE FreeBSD 14.1-RELEASE releng/14.1-n267679-10e31f0946d8 GENERIC amd64

mcuee@freebsd14:~ $ sudo lsusb
Bus /dev/usb Device /dev/ugen0.5: ID 04b4:8613 Cypress Semiconductor Corp. CY7C68013 EZ-USB FX2 USB 2.0 Development Kit
Bus /dev/usb Device /dev/ugen0.6: ID 04d8:003f Microchip Technology, Inc. 
Bus /dev/usb Device /dev/ugen0.4: ID 13d3:3503 IMC Networks 
Bus /dev/usb Device /dev/ugen0.3: ID 1ea7:0064 SHARKOON Technologies GmbH 2.4GHz Wireless rechargeable vertical mouse [More&Better]
Bus /dev/usb Device /dev/ugen0.2: ID 04f2:0760 Chicony Electronics Co., Ltd Acer KU-0760 Keyboard
Bus /dev/usb Device /dev/ugen0.1: ID 0000:0000  

mcuee@freebsd14:~/build/hidapi_pr730 $ fxload 
no device specified!
usage: fxload [-vV] [-B backend] [-l] [-t type] [-D devpath]
		[-I firmware_hexfile] [-s loader] [-c config_byte]
		[-L link] [-m mode]
... [-D devpath] overrides DEVICE= in env
... device types:  one of an21, fx, fx2, fx2lp, fx3
... at least one of -I, -L, -m is required

mcuee@freebsd14:~/build/hidapi_pr730 $ fxload -t fx2lp -I ./fx2hid.hex -D vid=0x04b4,pid=0x8613

mcuee@freebsd14:~/build/hidapi_pr730 $ lsusb
Bus /dev/usb Device /dev/ugen0.6: ID 04d8:003f Microchip Technology, Inc. 

mcuee@freebsd14:~/build/hidapi_pr730 $ sudo chmod 666 /dev/ugen0.5

mcuee@freebsd14:~/build/hidapi_pr730 $ lsusb
Bus /dev/usb Device /dev/ugen0.5: ID 0925:1234 Lakeview Research 
Bus /dev/usb Device /dev/ugen0.6: ID 04d8:003f Microchip Technology, Inc. 

mcuee@freebsd14:~/build/hidapitester $ ./hidapitester_libusb_git --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 64 bytes:
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00
Closing device

mcuee@freebsd14:~/build/hidapitester $ ./hidapitester_libusb_pr728 --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 0 bytes:
Closing device

mcuee · 2025-04-01T00:01:36Z

If I use the original fx2hid example (loop back of two bytes output report and input report), it seems to me this PR works fine. But that just means this PR has no regression.
http://janaxelson.com/hidpage.htm
http://janaxelson.com/files/fx2hid.zip

mcuee@freebsd14:~/build/hidapitester $ sudo ./hidapitester_libusb_pr728 --vidpid 0925:1234 --open --buflen 3 -l 3 --send-output 0,1,2 --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 3-bytes...wrote 3 bytes:
 00 01 02
Reading up to 3-byte input report, 250 msec timeout...read 2 bytes:
 01 02 00
Closing device

mcuee@freebsd14:~/build/hidapitester $ sudo ./hidapitester_libusb_git --vidpid 0925:1234 --open --buflen 3 -l 3 --send-output 0,1,2 --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 3-bytes...wrote 3 bytes:
 00 01 02
Reading up to 3-byte input report, 250 msec timeout...read 2 bytes:
 01 02 00
Closing device

cederom · 2025-06-19T00:55:55Z

@sudobash1 great news! I will be able to provide soonest feedback around Monday sorry if you are ready before and want to go ahead then go ahead :-) I am waiting for the board and will have to see how to set things up but then I will be able to help you in other cases :-)

Is there any sort of bad-usb device out there that could provide us various test cases from hardware point of view (i.e. simulate bad descriptors, strange transfers, etc)?

mcuee · 2025-06-19T06:55:31Z

@sudobash1

Other than testing, @Youw will need to review the codes and then approve this PR.

But now the first thing is to fix the CI build -- two builds failed because of the same issue.

[ 37%] Building C object src/libusb/CMakeFiles/hidapi_libusb.dir/hid.c.o
/__w/hidapi/hidapi/hidapisrc/libusb/hid.c: In function ‘get_max_report_size’:
/__w/hidapi/hidapi/hidapisrc/libusb/hid.c:333:29: error: comparison of integer expressions of different signedness: ‘int’ and ‘enum report_descr_type’ [-Werror=sign-compare]
  333 |                 if (key_cmd == report_type) { /* Input / Output / Feature */
      |                             ^~
cc1: all warnings being treated as errors
make[2]: *** [src/libusb/CMakeFiles/hidapi_libusb.dir/build.make:79: src/libusb/CMakeFiles/hidapi_libusb.dir/hid.c.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:206: src/libusb/CMakeFiles/hidapi_libusb.dir/all] Error 2
make: *** [Makefile:[13](https://github.com/libusb/hidapi/actions/runs/15746712859/job/44397056740?pr=728#step:7:14)6: all] Error 2

sudobash1 · 2025-06-19T20:24:51Z

But now the first thing is to fix the CI build -- two builds failed because of the same issue.

Oops. I thought I had fixed that. I guess not. I have pushed e2391b6 to address compiler warnings. It compiles cleanly with -Wall -Wextra on gcc-15 and clang-20 on my Fedora box now. Hopefully all the CI compile now.

mcuee · 2025-06-20T02:08:08Z

Oops. I thought I had fixed that. I guess not. I have pushed e2391b6 to address compiler warnings. It compiles cleanly with -Wall -Wextra on gcc-15 and clang-20 on my Fedora box now. Hopefully all the CI compile now.

Yes, CI builds are okay now.

sudobash1 · 2025-06-20T06:04:21Z

Ah, it looks like there a few comments from @Youw that I need to address too. Most significant is probably updating builds.yml (to make sure the tests are enabled).

I'll look through this on Friday.

JoergAtGithub · 2025-06-20T06:30:04Z

It would be good if it could clarify in the comments what exactly the “report size” returned by get_max_report_size comprises.

Here it is only about the InputReports that are sent unrequested by the HID device via interupt transfers (hid_read / hid_read_timeout).
However, InputReports can also be actively retrieved with hid_get_input_report via ControlTransfers, where the ReportId byte is always included, even if the device does not use ReportIDs (in this case the ReportID-byte contains the value 0).
And get_max_report_size also supports feature and output reports, where it's different too.

In other code of mine, I introduced the unofficial term payload-size for the report size without the ReportID-byte. There I add 1 depending on the context where it is used. There I've also a boolean that indicates, if the device uses ReportIDs.

mcuee · 2025-06-23T00:06:08Z

@sudobash1

First test under Windows -- it seems the reported write/read length is not consistent with native API, one byte higher.

Native Windows HID backend-- wrote 129 bytes, read 128 bytes
libusb backend -- wrote 130 bytes, read 129 bytes

But this is getting close. I will test under Linux and FreeBSD later.

 UCRT64 /c/work/libusb/hidapitester
$ ./hidapitester_git_winapi.exe --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 128 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

$ ./hidapitester_pr728_libusb_new.exe --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 130 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 129 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

$ ./hidapitester_pr728_winapi_new.exe --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37
,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,10
3,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 128 bytes:
 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20
 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F 40
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00
Closing device

Dirty fix to be able to build under Windows.

PS C:\work\libusb\hidapi_pr728_new> git diff
diff --git a/libusb/hid.c b/libusb/hid.c
index 0f06389..c830479 100644
--- a/libusb/hid.c
+++ b/libusb/hid.c
@@ -34,8 +34,8 @@
 #include <unistd.h>
 #include <sys/types.h>
 #include <sys/stat.h>
-#include <sys/ioctl.h>
-#include <sys/utsname.h>
+///#include <sys/ioctl.h>
+//#include <sys/utsname.h>
 #include <fcntl.h>
 #include <wchar.h>

hidapitester Makefile hack and binaries (against hidapi git and this PR, Windows native backend and libusb backend).

hidapitester_win.zip

cederom · 2025-10-01T23:33:27Z

Sorry for the delay, extremely overloaded, I have the CY7C680 board, working on FreeBSD.

# uname -a
FreeBSD hexagon 14.2-RELEASE-p1 FreeBSD 14.2-RELEASE-p1 GENERIC amd64

# usbconfig
(..)
ugen1.12: <CY7C68013 EZ-USB FX2 USB 2.0 Development Kit Cypress Semiconductor Corp.> at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (100mA)

# usbconfig -d 1.12 dump_all_desc
ugen1.12: <CY7C68013 EZ-USB FX2 USB 2.0 Development Kit Cypress Semiconductor Corp.> at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (100mA)

  bLength = 0x0012
  bDescriptorType = 0x0001
  bcdUSB = 0x0200
  bDeviceClass = 0x00ff  <Vendor specific>
  bDeviceSubClass = 0x00ff
  bDeviceProtocol = 0x00ff
  bMaxPacketSize0 = 0x0040
  idVendor = 0x04b4
  idProduct = 0x8613
  bcdDevice = 0xa001
  iManufacturer = 0x0000  <no string>
  iProduct = 0x0000  <no string>
  iSerialNumber = 0x0000  <no string>
  bNumConfigurations = 0x0001

 Configuration index 0

    bLength = 0x0009
    bDescriptorType = 0x0002
    wTotalLength = 0x00ab
    bNumInterfaces = 0x0001
    bConfigurationValue = 0x0001
    iConfiguration = 0x0000  <no string>
    bmAttributes = 0x0080
    bMaxPower = 0x0032

    Interface 0
      bLength = 0x0009
      bDescriptorType = 0x0004
      bInterfaceNumber = 0x0000
      bAlternateSetting = 0x0000
      bNumEndpoints = 0x0000
      bInterfaceClass = 0x00ff  <Vendor specific>
      bInterfaceSubClass = 0x00ff
      bInterfaceProtocol = 0x00ff
      iInterface = 0x0000  <no string>


    Interface 0 Alt 1
      bLength = 0x0009
      bDescriptorType = 0x0004
      bInterfaceNumber = 0x0000
      bAlternateSetting = 0x0001
      bNumEndpoints = 0x0006
      bInterfaceClass = 0x00ff  <Vendor specific>
      bInterfaceSubClass = 0x00ff
      bInterfaceProtocol = 0x00ff
      iInterface = 0x0000  <no string>

     Endpoint 0
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0001  <OUT>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 1
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0081  <IN>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 2
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0002  <OUT>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 3
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0004  <OUT>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 4
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0086  <IN>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 5
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0088  <IN>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000


    Interface 0 Alt 2
      bLength = 0x0009
      bDescriptorType = 0x0004
      bInterfaceNumber = 0x0000
      bAlternateSetting = 0x0002
      bNumEndpoints = 0x0006
      bInterfaceClass = 0x00ff  <Vendor specific>
      bInterfaceSubClass = 0x00ff
      bInterfaceProtocol = 0x00ff
      iInterface = 0x0000  <no string>

     Endpoint 0
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0001  <OUT>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 1
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0081  <IN>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 2
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0002  <OUT>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0200
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 3
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0004  <OUT>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 4
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0086  <IN>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0200
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 5
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0088  <IN>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000


    Interface 0 Alt 3
      bLength = 0x0009
      bDescriptorType = 0x0004
      bInterfaceNumber = 0x0000
      bAlternateSetting = 0x0003
      bNumEndpoints = 0x0006
      bInterfaceClass = 0x00ff  <Vendor specific>
      bInterfaceSubClass = 0x00ff
      bInterfaceProtocol = 0x00ff
      iInterface = 0x0000  <no string>

     Endpoint 0
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0001  <OUT>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 1
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0081  <IN>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 2
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0002  <OUT>
        bmAttributes = 0x0001  <ISOCHRONOUS>
        wMaxPacketSize = 0x0200
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 3
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0004  <OUT>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 4
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0086  <IN>
        bmAttributes = 0x0001  <ISOCHRONOUS>
        wMaxPacketSize = 0x0200
        bInterval = 0x0001
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 5
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0088  <IN>
        bmAttributes = 0x0002  <BULK>
        wMaxPacketSize = 0x0200
        bInterval = 0x0000
        bRefresh = 0x0000
        bSynchAddress = 0x0000

% ./hidapitester --version
hidapitester version: v0.5
hidapi version: 0.16.0

% ./hidapitester --list
0B05/19AF: AsusTek Computer Inc. - AURA LED Controller
534D/2109: MACROSILICON - USB3. 0 capture
0B0E/24CA:  - Jabra Link 380
09DA/6406: A4tech - A4tech KB
09DA/6406: A4tech - A4tech KB
046D/C099: Logitech - G502 X
046D/C099: Logitech - G502 X
044F/B106: Thrustmaster - T.Flight Stick X

Had to power on board with button pressed to be able to flash firmware:

# fxload -t fx2 -I fx2hid_128bytes/fx2hid.hex -D vid=04b4,pid=8613 -v
microcontroller type: fx2
single stage:  load on-chip memory
open RAM hexfile image fx2hid_128bytes/fx2hid.hex
stop CPU
write on-chip, addr 0x0600 len  234 (0x00ea)
write on-chip, addr 0x09e5 len   10 (0x000a)
write on-chip, addr 0x0380 len  428 (0x01ac)
write on-chip, addr 0x0080 len  768 (0x0300)
write on-chip, addr 0x0033 len    3 (0x0003)
write on-chip, addr 0x09ff len    7 (0x0007)
write on-chip, addr 0x07b8 len   92 (0x005c)
write on-chip, addr 0x0851 len   59 (0x003b)
write on-chip, addr 0x05fe len    2 (0x0002)
write on-chip, addr 0x0a07 len    4 (0x0004)
write on-chip, addr 0x09ef len    8 (0x0008)
write on-chip, addr 0x09b0 len   18 (0x0012)
write on-chip, addr 0x09f7 len    8 (0x0008)
write on-chip, addr 0x09c2 len   18 (0x0012)
write on-chip, addr 0x0a0b len    6 (0x0006)
write on-chip, addr 0x091d len   40 (0x0028)
write on-chip, addr 0x096a len   24 (0x0018)
write on-chip, addr 0x06ea len   22 (0x0016)
write on-chip, addr 0x099a len   22 (0x0016)
write on-chip, addr 0x088c len   54 (0x0036)
write on-chip, addr 0x0982 len   24 (0x0018)
write on-chip, addr 0x0814 len   61 (0x003d)
write on-chip, addr 0x0a11 len   36 (0x0024)
write on-chip, addr 0x08f1 len   44 (0x002c)
write on-chip, addr 0x08c2 len   47 (0x002f)
write on-chip, addr 0x0945 len   20 (0x0014)
write on-chip, addr 0x05b8 len   70 (0x0046)
write on-chip, addr 0x0959 len   17 (0x0011)
write on-chip, addr 0x0043 len    3 (0x0003)
write on-chip, addr 0x0053 len    3 (0x0003)
write on-chip, addr 0x0700 len  184 (0x00b8)
write on-chip, addr 0x0000 len    3 (0x0003)
write on-chip, addr 0x052c len   12 (0x000c)
write on-chip, addr 0x09d4 len   17 (0x0011)
write on-chip, addr 0x0538 len  128 (0x0080)
write on-chip, addr 0x0a06 len    1 (0x0001)
... WROTE: 2497 bytes, 36 segments, avg 69
reset CPU

Now it shows as:

ugen1.12: <CYPRESS EZ-USB FX2 HID USBHIDIO> at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (80mA)

% usbconfig -d 1.12 dump_all_desc
ugen1.12: <CYPRESS EZ-USB FX2 HID USBHIDIO> at usbus1, cfg=0 md=HOST spd=HIGH (480Mbps) pwr=ON (80mA)

  bLength = 0x0012
  bDescriptorType = 0x0001
  bcdUSB = 0x0200
  bDeviceClass = 0x0000  <Probed by interface class>
  bDeviceSubClass = 0x0000
  bDeviceProtocol = 0x0000
  bMaxPacketSize0 = 0x0040
  idVendor = 0x0925
  idProduct = 0x1234
  bcdDevice = 0x0000
  iManufacturer = 0x0001  <CYPRESS>
  iProduct = 0x0002  <EZ-USB FX2 HID USBHIDIO>
  iSerialNumber = 0x0000  <no string>
  bNumConfigurations = 0x0001

 Configuration index 0

    bLength = 0x0009
    bDescriptorType = 0x0002
    wTotalLength = 0x0029
    bNumInterfaces = 0x0001
    bConfigurationValue = 0x0001
    iConfiguration = 0x0003  <retrieving string failed>
    bmAttributes = 0x0080
    bMaxPower = 0x0028

    Interface 0
      bLength = 0x0009
      bDescriptorType = 0x0004
      bInterfaceNumber = 0x0000
      bAlternateSetting = 0x0000
      bNumEndpoints = 0x0002
      bInterfaceClass = 0x0003  <HID device>
      bInterfaceSubClass = 0x0000
      bInterfaceProtocol = 0x0000
      iInterface = 0x0000  <no string>

      Additional Descriptor

      bLength = 0x09
      bDescriptorType = 0x21
      bDescriptorSubType = 0x10
       RAW dump:
       0x00 | 0x09, 0x21, 0x10, 0x01, 0x00, 0x01, 0x22, 0x34,
       0x08 | 0x00

     Endpoint 0
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0081  <IN>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0005
        bRefresh = 0x0000
        bSynchAddress = 0x0000

     Endpoint 1
        bLength = 0x0007
        bDescriptorType = 0x0005
        bEndpointAddress = 0x0001  <OUT>
        bmAttributes = 0x0003  <INTERRUPT>
        wMaxPacketSize = 0x0040
        bInterval = 0x0005
        bRefresh = 0x0000
        bSynchAddress = 0x0000


% ./hidapitester --list
0B05/19AF: AsusTek Computer Inc. - AURA LED Controller
534D/2109: MACROSILICON - USB3. 0 capture
0B0E/24CA:  - Jabra Link 380
09DA/6406: A4tech - A4tech KB
09DA/6406: A4tech - A4tech KB
044F/B106: Thrustmaster - T.Flight Stick X
046D/C099: Logitech - G502 X
046D/C099: Logitech - G502 X
0925/1234: CYPRESS - EZ-USB FX2 HID USBHIDIO


% ./hidapitester --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 64 bytes:
 68 89 37 FE 79 58 09 E7 88 CE 3F 33 9D CF B7 7D 95 C7 8F AD 2E 84 97 EF FF 71 FC 3F F5 D7 70 6E
 9E 93 5E FE F8 DF BA 74 DE B7 AB 6F 9D 3F AF BB 5B FE BF CF BE FB EE 03 B7 EF 8A 0C B3 BB E7 6A
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00
Closing device


% ./hidapitester --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128  --read-input
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F
 40 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F
 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F
 80
Reading up to 129-byte input report, 250 msec timeout...read 64 bytes:
 41 42 43 44 45 46 47 48 49 4A 4B 4C 4D 4E 4F 50 51 52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60
 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F 80
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00
Closing device

% ./hidapitester --vidpid 0925:1234 --open --buflen 256 -l 129 --send-output 0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37
Opening device, vid/pid: 0x0925/0x1234
Writing output report of 129-bytes...wrote 129 bytes:
 00 01 02 03 04 05 06 07 08 09 0A 0B 0C 0D 0E 0F 10 11 12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F
 20 21 22 23 24 25 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
 00
Closing device

acolombier · 2025-10-31T10:47:25Z

I realised I forgot to update my previously flawed comment - the issue was obviously on my end.
Tested this patch on Android and work like a charm! (ref)

Fixes libusb#274

Co-authored-by: Ihor Dutchak <ihor.youw@gmail.com>

mcuee · 2025-12-30T02:03:37Z

@sudobash1

Just wondering if you have any updates on this one. I think it is pretty close now.

mcuee · 2026-03-20T01:46:06Z

Just update the branch to trigger a new build.

It will be good to move this PR forward. May need more testing and reviews.

mcuee · 2026-04-19T07:57:27Z

@Youw

Just wondering if it is worth asking Github Copilot to review this PR. It seems to be pretty close to be okay.

Youw

Coming back to this after dropping it for a while — did a spec/kernel cross-check pass and a code re-read. Leaving hardware testing to those with the devices.

Algorithm validation

get_max_report_size matches HID 1.11 (§6.2.2.7, §8 for Report ID semantics) and is structurally identical to Linux's report-descriptor parsing in drivers/hid/hid-core.c:

hid_add_field(): report->size += parser->global.report_size * parser->global.report_count;
hid_register_report(): if (id != 0) report_enum->numbered = 1;
On receive, one byte is consumed as the ID prefix iff numbered — which is exactly the + report_id_used semantic here.

The FX2 128-byte fix is the subtle one: the hang wasn't the size math, it was the unconditional +1 making the buffer a non-multiple of wMaxPacketSize and suppressing the libusb short-packet terminator. Good catch on making it conditional on the Report ID tag actually being used.

Reusing windows/test/data/*.pp_data for a cross-backend parity test is a nice consistency proof.

Suggestions before merge

Scope of "Fixes #274" — this resolves the long-report truncation aspect, but #274 also asks for Full-Speed misconfiguration detection / reporting to the app. Consider softening to "Partially fixes" or adding a note.
Platform length asymmetry that @mcuee observed on Windows (libusb wrote 130 / read 129 vs native HID wrote 129 / read 128) — worth a sentence in the PR body pointing to #731 so users don't re-file it as a regression.
Inline nits on specific lines below.

Algorithm + spec alignment LGTM on the core change.

Extended version:

Comparison with the spec and other implementations
HID 1.11 spec (§6.2.2.7 & §5.6 / §8): "If a Report ID tag is used anywhere in Report descriptor, all data reports for the device are preceded by a single byte ID field." Report ID = 0 is a placeholder meaning "no report IDs". Reports are sized as Σ(report_count × report_size) bits, rounded up to bytes. The PR's algorithm and the conditional +report_id_used byte match the spec exactly.

Windows (HIDP_CAPS): InputReportByteLength/OutputReportByteLength/FeatureReportByteLength — these always include the Report-ID byte (value 0 when unused). That's the source of the read/write-length platform asymmetry and is exactly what follow-up issue #731 is about. The PR's test case subtracts 1 byte from these values when ReportID == 0x00 to reconcile with the libusb convention (see parse_max_input_report_size in max_input_report_size_test.c).

Linux kernel (drivers/hid/hid-core.c): Uses the same algorithm. hid_add_field() accumulates report->size += parser->global.report_size * parser->global.report_count. In hid_register_report(), if (id != 0) report_enum->numbered = 1;. In hid_report_raw_event, when numbered is true the first data byte is consumed as the ID (cdata++; csize--;). Size is bounded by max_buffer_size - 1 bits when numbered. This is structurally identical to what the PR does — so the PR's approach is directly validated by the reference kernel implementation.

hidapi/linux (hidraw backend): Avoids parsing entirely — delegates to the kernel: read(dev->device_handle, data, length) with user-supplied length, because the kernel has already sized the per-report buffer. That's why hidraw "just works" for the 128-byte FX2 firmware and why the libusb backend is the outlier.

Strengths
Correct algorithm. Matches HID 1.11 and Linux kernel's parsing exactly, including the subtle conditional Report-ID byte.
Good test coverage via reuse of 25 real-world Windows pp_data fixtures — this is a nice cross-backend consistency proof.
Safe fallback to input_ep_max_packet_size when the descriptor is corrupt/unparseable — no regression risk for devices already working.
Fixes the classic "exact multiple of max packet size hangs" libusb gotcha — the conditional +1 was the critical insight; without it devices with payloads that happen to be multiples of 64 would hang forever waiting for a short packet.
Remaining concerns worth flagging on review
Re-read of report descriptor. hidapi_initialize_device now calls hid_get_report_descriptor_libusb during init (new control transfer on every open). This is a second descriptor fetch in addition to any caller-driven hid_get_report_descriptor(). Consider caching.
HID_API_MAX_REPORT_DESCRIPTOR_SIZE buffer on stack (4096 bytes typically). libusb/hid.c:1314 in the diff allocates this on the stack — acceptable, but worth noting for embedded targets.
cur_size reset timing. In the parser, cur_size is reset on each Report ID encounter. For descriptors that interleave INPUT/OUTPUT items between Report ID changes this is correct; just note that the max_size comparison happens only at Report-ID boundaries and at the end — which matches the spec semantics where each Report ID opens a fresh report namespace.
Platform-length asymmetry (mcuee's Windows observation) is intentionally out of scope — tracked in #731. Worth linking from the PR description.
PR description still says "Fixes #274" but #274 has broader asks (detection of USB Full-Speed misconfiguration, speed reporting to app). This PR fixes the "long reports get truncated" aspect; the speed/error-reporting aspect remains open.
Typo in test comment (.dpt_desc → .rpt_desc) in libusb/test/CMakeLists.txt.
Bottom line
The PR is technically sound, aligned with HID 1.11 and the Linux kernel reference, and has been hardware-validated on Linux, FreeBSD and Android. The earlier "doesn't work" reports were a spec-compliance bug in the Report-ID handling that sudobash1 fixed. Primary blockers now are administrative: final Youw approval + merge, plus perhaps addressing the minor concerns above. The broader cross-platform length-consistency question is correctly deferred to #731.

Youw · 2026-04-19T19:58:52Z

+# <name>_expected.rpt_desc - reconstructed HID Report Descriptor out of .pp_data file;
+#
+# (Non-required by test):
+# <name>_real.dpt_desc - the original report rescriptor used to create a test case;


Two typos: dpt_desc → rpt_desc, and rescriptor → descriptor.

Youw · 2026-04-19T19:58:52Z


 	dev->report_descriptor_size = get_report_descriptor_size_from_interface_descriptors(intf_desc);

+	unsigned char report_descriptor[HID_API_MAX_REPORT_DESCRIPTOR_SIZE];


Two minor things in this block:

4 KiB on the stack (HID_API_MAX_REPORT_DESCRIPTOR_SIZE) — fine for desktop, may be tight on some embedded targets. Worth heap-allocating to match the rest of the backend.

This is now a second descriptor fetch per hid_open (one here, and another whenever a caller later invokes hid_get_report_descriptor()). Consider caching the bytes on hid_device so both consumers share them.

1 - totally dismiss, 4KB is fine on any platform where HIDAPI shall be considered to be used

Youw · 2026-04-19T20:01:01Z

This is the result of the review by Claude Opus 4.7 (locally).

JoergAtGithub · 2026-04-19T21:48:37Z

PR description still says "Fixes #274" but #274 has broader asks (detection of USB Full-Speed misconfiguration, speed reporting to app). This PR fixes the "long reports get truncated" aspect; the speed/error-reporting aspect remains open.

I think this statement is not correct. The USB standard specifies the max read size per speed category. With this PR every read size should work and therefore every USB speed.

sudobash1 mentioned this pull request Mar 25, 2025

HID reports of more than 64 bytes and libusb backend #274

Open

sudobash1 force-pushed the libusb_input_size branch from 068fd62 to 52c6cd9 Compare March 25, 2025 22:06

mcuee added bug Something isn't working libusb Related to libusb backend labels Mar 26, 2025

Youw reviewed Mar 26, 2025

View reviewed changes

Comment thread libusb/hid.c Outdated

JoergAtGithub reviewed Mar 26, 2025

View reviewed changes

mcuee mentioned this pull request Mar 27, 2025

pp_data_dump issue for 046A_0011_0006_0001 -- Lower Priority #729

Closed

sudobash1 force-pushed the libusb_input_size branch 2 times, most recently from 2708264 to 3710895 Compare March 27, 2025 19:42

JoergAtGithub suggested changes Mar 27, 2025

View reviewed changes

Youw reviewed Mar 28, 2025

View reviewed changes

Comment thread libusb/test/CMakeLists.txt Outdated

Youw reviewed Mar 28, 2025

View reviewed changes

Youw mentioned this pull request Mar 28, 2025

Store Input/Output/Feature reports sizes for all platforms #731

Open

mcuee requested a review from JoergAtGithub June 19, 2025 06:44

sudobash1 force-pushed the libusb_input_size branch 2 times, most recently from 44328f8 to e2391b6 Compare June 19, 2025 20:16

acolombier mentioned this pull request Oct 1, 2025

[libusb] hid_read doesn't include the report ID #759

Closed

This comment was marked as outdated.

Sign in to view

Stephen and others added 7 commits December 30, 2025 10:03

Properly determine libusb read size for large reports

76b72c6

Fixes libusb#274

Fixes to report size calculation and added tests

09017e5

Allow computation of max report size for other types

415db44

Don't assign -1 to size_t. That's silly

8b07bcf

Do not increase max_report_size by one if not using report numbers

3b3c70a

Fix compiler warnings

c358d33

Remove comment

e19ba07

Co-authored-by: Ihor Dutchak <ihor.youw@gmail.com>

mcuee force-pushed the libusb_input_size branch from a4c4da9 to e19ba07 Compare December 30, 2025 02:03

Merge branch 'master' into libusb_input_size

6eedce1

Merge branch 'master' into libusb_input_size

cda57eb

Youw reviewed Apr 19, 2026

View reviewed changes

	pp_data->caps_info[0]->FirstCap = 0
	pp_data->caps_info[0]->LastCap = 2
	pp_data->caps_info[0]->NumberOfCaps = 2
	pp_data->caps_info[0]->ReportByteLength = 9

	pp_data->cap[0]->BitSize = 1
	pp_data->cap[0]->ReportCount = 8

	pp_data->cap[1]->BitSize = 8
	pp_data->cap[1]->ReportCount = 6


		dev->report_descriptor_size = get_report_descriptor_size_from_interface_descriptors(intf_desc);

		unsigned char report_descriptor[HID_API_MAX_REPORT_DESCRIPTOR_SIZE];

Conversation

sudobash1 commented Mar 25, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sudobash1 commented Mar 25, 2025

Uh oh!

Be-ing commented Mar 26, 2025

Uh oh!

Youw left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

mcuee commented Mar 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

JoergAtGithub commented Mar 27, 2025

Uh oh!

JoergAtGithub commented Mar 27, 2025

Uh oh!

mcuee commented Mar 27, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

JoergAtGithub Mar 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sudobash1 commented Mar 27, 2025

Uh oh!

JoergAtGithub commented Mar 27, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Youw left a comment

Choose a reason for hiding this comment

Uh oh!

Youw commented Mar 28, 2025

Uh oh!

mcuee commented Mar 28, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mcuee commented Mar 28, 2025

Uh oh!

sudobash1 commented Mar 29, 2025

Uh oh!

mcuee commented Mar 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mcuee commented Mar 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mcuee commented Apr 1, 2025

Uh oh!

sudobash1 commented Mar 25, 2025 •

edited

Loading

mcuee commented Mar 27, 2025 •

edited

Loading

JoergAtGithub Mar 27, 2025 •

edited

Loading

mcuee commented Mar 28, 2025 •

edited

Loading

mcuee commented Mar 30, 2025 •

edited

Loading

mcuee commented Mar 30, 2025 •

edited

Loading

cederom commented Jun 19, 2025 •

edited

Loading

mcuee commented Jun 23, 2025 •

edited

Loading

cederom commented Oct 1, 2025 •

edited

Loading

Youw left a comment •

edited

Loading