search

flozz / rivalcfg

CLI tool and Python library to configure SteelSeries gaming mice
https://flozz.github.io/rivalcfg/
Do What The F*ck You Want To Public License
761 stars 61 forks source link

Requested device or endpoint not found: 1038:183a:03 on MacOS Ventura #200

Closed OptimusKoala closed 1 year ago

OptimusKoala commented 1 year ago

Hi :)

I installed rivalcfg 4.8 with pip3 and I'm running Python 3.9.9 on MacOS Ventura 13.3 When my mouse is plugged I have this error when I try to print help

Traceback (most recent call last):
  File "/Users/optimuskoala/.pyenv/versions/3.9.9/bin/rivalcfg", line 8, in <module>
    sys.exit(main())
  File "/Users/optimuskoala/.pyenv/versions/3.9.9/lib/python3.9/site-packages/rivalcfg/__main__.py", line 78, in main
    mouse = get_first_mouse()
  File "/Users/optimuskoala/.pyenv/versions/3.9.9/lib/python3.9/site-packages/rivalcfg/__init__.py", line 29, in get_first_mouse
    return mouse.get_mouse(
  File "/Users/optimuskoala/.pyenv/versions/3.9.9/lib/python3.9/site-packages/rivalcfg/mouse.py", line 36, in get_mouse
    hid_device = usbhid.open_device(vendor_id, product_id, profile["endpoint"])
  File "/Users/optimuskoala/.pyenv/versions/3.9.9/lib/python3.9/site-packages/rivalcfg/usbhid.py", line 96, in open_device
    raise DeviceNotFound(
rivalcfg.usbhid.DeviceNotFound: Requested device or endpoint not found: 1038:183a:03

I can see my mousse with the print-debug option

RIVALCFG
========
Version: 4.8.0
Installation path: /Users/optimuskoala/.pyenv/versions/3.9.9/lib/python3.9/site-packages/rivalcfg

OPERATING SYSTEM
================
OS: Darwin
PYTHON
======
Python version: 3.9.9
HIDAPI version: 0.13.1

PLUGGED STEELSERIES DEVICES ENDPOINTS
=====================================
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)
1038:183a | 00 | SteelSeries Aerox 3 Wireless (firmware v0)

Any idea to makes it work ? :)

flozz commented 1 year ago

I do not know why the mouse is listed bu is not found when trying to open it... I do not have macOS to help debuging this :(

OptimusKoala commented 1 year ago

Hi Fabien :) Do you want some debuging trace or something ? Maybe rivalcfg is not supported under macOS Ventura ? (I think this is not the issue here)

I tried to put my mousse on usb, then bluetooth and then wifi but it's every time the same error.

flozz commented 1 year ago

I think I do not know macOS enough to be really helpful here... :(

What is strange here is that you have 6 endpoints for the device but all are displayed as being the "endpoint 0"... Maybe an other software or a kernel diver is already attached to the device and makes some interference?

OptimusKoala commented 1 year ago

I don't have steelseries driver installed.

I could try to make it work myself by cloning the repo and try in local. I'll keep you in touch :)

glyph commented 1 year ago

I have this issue as well.

here's what a full dump of my enumerated devices looks like:

{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4294982071',
 'product_id': 5926,
 'product_string': 'SteelSeries Rival 650 Wireless',
 'release_number': 294,
 'serial_number': '000000000000',
 'usage': 1,
 'usage_page': 65473,
 'vendor_id': 4152}
{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4294982077',
 'product_id': 5926,
 'product_string': 'SteelSeries Rival 650 Wireless',
 'release_number': 294,
 'serial_number': '000000000000',
 'usage': 2,
 'usage_page': 1,
 'vendor_id': 4152}
{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4294982077',
 'product_id': 5926,
 'product_string': 'SteelSeries Rival 650 Wireless',
 'release_number': 294,
 'serial_number': '000000000000',
 'usage': 1,
 'usage_page': 1,
 'vendor_id': 4152}
{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4294982075',
 'product_id': 5926,
 'product_string': 'SteelSeries Rival 650 Wireless',
 'release_number': 294,
 'serial_number': '000000000000',
 'usage': 6,
 'usage_page': 1,
 'vendor_id': 4152}
{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4294982075',
 'product_id': 5926,
 'product_string': 'SteelSeries Rival 650 Wireless',
 'release_number': 294,
 'serial_number': '000000000000',
 'usage': 1,
 'usage_page': 12,
 'vendor_id': 4152}

As you can see, interface_number is not unique here, so filtering on it selects a device at random. After some trial and error, I discovered that filtering on interface["usage_page"] == 65472 allowed me to find the one that would report battery status reliably. I don't know if this is the same device used for other stuff; I have only the vaguest idea what "usage pages" are or how HID devices work.

glyph commented 1 year ago

I put up a PR for this, #202, since I was having this issue with my mouse too. I don't know when it started occurring, but "when upgrading to Ventura" seems plausible.

flozz commented 1 year ago

on my rival 100 it seems to be 0:

 {'path': b'5-2:1.0',
  'vendor_id': 4152,
  'product_id': 5890,
  'serial_number': '',
  'release_number': 101,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 100 Gaming Mouse',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 0},
 {'path': b'5-2:1.1',
  'vendor_id': 4152,
  'product_id': 5890,
  'serial_number': '',
  'release_number': 101,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 100 Gaming Mouse',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 1},

I will check with all my other SteelSeries mice to see if it is specific to Aerox and Rival 650 / or if it may be specific to macOS Ventura...

flozz commented 1 year ago

Ok I made some test on Linux with my Rival 650.

When I enumerate interfaces I have:

 {'path': b'2-7:1.0',
  'vendor_id': 4152,
  'product_id': 5926,
  'serial_number': '000000000000',
  'release_number': 294,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 650 Wireless',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 0},
 {'path': b'2-7:1.1',
  'vendor_id': 4152,
  'product_id': 5926,
  'serial_number': '000000000000',
  'release_number': 294,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 650 Wireless',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 1},
 {'path': b'2-7:1.2',
  'vendor_id': 4152,
  'product_id': 5926,
  'serial_number': '000000000000',
  'release_number': 294,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 650 Wireless',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 2},
 {'path': b'2-7:1.3',
  'vendor_id': 4152,
  'product_id': 5926,
  'serial_number': '000000000000',
  'release_number': 294,
  'manufacturer_string': 'SteelSeries',
  'product_string': 'SteelSeries Rival 650 Wireless',
  'usage_page': 0,
  'usage': 0,
  'interface_number': 3},

With Wireshark I can capture this:

image

So the problem is:

Page ID Page Name hidusage.h constant
0x01 Generic Desktop Controls HID_USAGE_PAGE_GENERIC
0x05 Game Controls HID_USAGE_PAGE_GAME
0x08 LEDs HID_USAGE_PAGE_LED
0x09 Button HID_USAGE_PAGE_BUTTON

I do not know what is going on on macOS Ventura, but there is clearly an issue somewhere. We will have to find a way to know which interface is the right one without using the usage page and the interface_number :/

flozz commented 1 year ago

I had an idea to fix the issue on macOS Ventura (without breaking Linux). I pushed it on the 200-macos-ventura-endpoint branch. Can someone test if it works? :)

git clone https://github.com/flozz/rivalcfg.git
cd rivalcfg
git checkout 200-macos-ventura-endpoint
python3 -m venv __env__
source __env__/bin/activate
pip install -e .

python -m rivalcfg [...]

See changes: https://github.com/flozz/rivalcfg/commit/57756e92ca335efdff6e9ebaf2b95d4db2f20018

flozz commented 1 year ago

UP

I really need someone on macOS Ventura to test it work :)

matthewvinton commented 1 year ago

I am on Ventura, experiencing the issue. I will test by end of week. Thanks!

johnelliott commented 1 year ago

I just checked out the 200-macos-ventura-endpoint branch and it worked, it seems fixed to me 👍. Way to go @flozz :)

glyph commented 1 year ago

Sadly 200-macos-ventura-endpoint doesn't fix it for me. As before, the connection is still unreliable, and I freaquently get.

Traceback (most recent call last):
  File "/Users/glyph/.local/bin/rivalcfg", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/Users/glyph/Projects/rivalcfg/rivalcfg/__main__.py", line 81, in main
    raise error
  File "/Users/glyph/Projects/rivalcfg/rivalcfg/__main__.py", line 78, in main
    mouse = get_first_mouse()
            ^^^^^^^^^^^^^^^^^
  File "/Users/glyph/Projects/rivalcfg/rivalcfg/__init__.py", line 29, in get_first_mouse
    return mouse.get_mouse(
           ^^^^^^^^^^^^^^^^
  File "/Users/glyph/Projects/rivalcfg/rivalcfg/mouse.py", line 36, in get_mouse
    hid_device = usbhid.open_device(vendor_id, product_id, profile["endpoint"])
                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/glyph/Projects/rivalcfg/rivalcfg/usbhid.py", line 101, in open_device
    device.open_path(path)
  File "hid.pyx", line 154, in hid.device.open_path
OSError: open failed
glyph commented 1 year ago

I also still get Unable to get the battery level sometimes, so the behavior seems identical.

glyph commented 1 year ago

I also double checked to make sure it wasn't a wired/wireless issue; I tried while wired, while on wireless, and wired with the wireless entirely disconnected.

flozz commented 1 year ago

@johnelliott Thank you for the feedback!

@glyph Your issue seems de be different from the previous one:

rivalcfg.usbhid.DeviceNotFound: Requested device or endpoint not found

— VS —

OSError: open failed

The endpoint seems to be found but hidapi is not able to open it (an other software or driver is attached to the device? or a permission issue?). We will have to find why it cannot be opened ^^"

flozz commented 1 year ago

@glyph I turned your comment as a new issue as it is not the same problem.

I close this issue as the initial problem is solved. The fix will be released soon :)

glyph commented 1 year ago

@glyph I turned your comment as a new issue as it is not the same problem.

I close this issue as the initial problem is solved. The fix will be released soon :)

Sounds good. Looking forward to a fix for that one :)

glyph commented 1 year ago

The endpoint seems to be found but hidapi is not able to open it (an other software or driver is attached to the device? or a permission issue?). We will have to find why it cannot be opened ^^"

I should ask, is there anything diagnostic you'd like me to do? I'm sure it's not permissions, but I don't know what else it might be.

johnelliott commented 1 year ago

I do also notice some intermittent issues with Ventura. Before the 200-macos-ventura-endpoint branch it would never work.

Now when I wake my mac from sleep and the mouse and usb adapter seem to have disassociated I need to run the script again, and it seems to work within 30 seconds or so after a few tries. Perhaps I'm seeing the same issues as @glyph.

flozz commented 1 year ago

I reopened the issue because I think my fix work for some devices (the ones that use an endpoint != 0, like Aerox) but not for devices that use endpoint 0... For those one, the first endpoint listed will be used and it can be the wrong one. I have to tweak the code a bit more... :)

flozz commented 1 year ago

Can someone on macOS Ventura give me the result of the following command?

python3 -c "import platform ; print(platform.system()) ; print(platform.version()) ; print(platform.mac_ver())"

→ Someone on Twitter gave me the answer:

Darwin
Darwin Kernel Version 22.5.0: Mon Apr 24 20:53:44 PDT 2023; root:xnu-8796.121.2~5/RELEASE_ARM64_T8103
('13.4', ('', '', ''), 'arm64')
flozz commented 1 year ago

I pushed a new fix on the 200-macos-ventura-endpoint branch: https://github.com/flozz/rivalcfg/commit/0ed42261dda14d7e1990bf744bafb09942e02647

→ Can you test if it work? → Please remind me what mouse model you are using (to allow me to know if it works for both devices that use endpoint 3 and 0).

:)

glyph commented 1 year ago

('13.4', ('', '', ''), 'arm64')

Same here.

glyph commented 1 year ago

→ Can you test if it work?

Kind of. It's better now, but still not completely reliable:

  ★ for each in $(seq 20); do rivalcfg --battery-level; done
Unable to get the battery level
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Unable to get the battery level
Unable to get the battery level
Charging [========  ] 89 %
Charging [========  ] 89 %
Unable to get the battery level
Charging [========  ] 89 %
Charging [========  ] 89 %
Unable to get the battery level
Unable to get the battery level
Charging [========  ] 89 %
Unable to get the battery level
Unable to get the battery level
Charging [========  ] 89 %
Unable to get the battery level
Charging [========  ] 89 %
Unable to get the battery level
glyph commented 1 year ago

Contrast my usage-page branch:

  ★ for each in $(seq 20); do rivalcfg --battery-level; done
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
Charging [========  ] 89 %
glyph commented 1 year ago

As far as what hardware I'm using: "1038:172b | 00 | SteelSeries Rival 650 Wireless (firmware v0)"

flozz commented 1 year ago

Ok... I need more info to figure what is happening... I pushed a new version of the code in a new branch : 200-macos-ventura-endpoint-test that prints some debug information.

Can you run it multiple times like you did just before and then share results here? :)

johnelliott commented 1 year ago

200-macos-ventura-endpoint-test ... run it multiple times like you did just before and then share results here? :)

Ran my shell script invoking the program with my flags, this time I didn't see the lights on the mouse change:

$ ./aerox.sh
================================================================================
_IS_MACOS_VENTURA: 1
--------------------------------------------------------------------------------
All interfaces:
[{'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010661',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 6,
  'usage_page': 1,
  'vendor_id': 4152},
 {'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010657',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 1,
  'usage_page': 12,
  'vendor_id': 4152},
 {'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010659',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 1,
  'usage_page': 65473,
  'vendor_id': 4152},
 {'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010663',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 1,
  'usage_page': 65472,
  'vendor_id': 4152},
 {'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010665',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 2,
  'usage_page': 1,
  'vendor_id': 4152},
 {'interface_number': 0,
  'manufacturer_string': 'SteelSeries',
  'path': b'DevSrvsID:4295010665',
  'product_id': 6200,
  'product_string': 'SteelSeries Aerox 3 Wireless',
  'release_number': 299,
  'serial_number': '',
  'usage': 1,
  'usage_page': 1,
  'vendor_id': 4152}]
--------------------------------------------------------------------------------
macOS Ventura found interface:
{'interface_number': 0,
 'manufacturer_string': 'SteelSeries',
 'path': b'DevSrvsID:4295010659',
 'product_id': 6200,
 'product_string': 'SteelSeries Aerox 3 Wireless',
 'release_number': 299,
 'serial_number': '',
 'usage': 1,
 'usage_page': 65473,
 'vendor_id': 4152}
--------------------------------------------------------------------------------
glyph commented 1 year ago

Here's a log of me running it 20 times:

https://gist.github.com/glyph/09d9b3db000e04634bc66dced662a3db

flozz commented 1 year ago

Thank you for the feedbacks.

It found the following issue on the HIDAPI lib that is used by Rivalcfg to access USB devices: https://github.com/libusb/hidapi/issues/531

They released a fix 3 days ago!

https://github.com/libusb/hidapi/releases/tag/hidapi-0.14.0

Once the Python binding of HIDAPI released we will be able to test if it fixes the issue for us, else we will hack in Rivalcfg. :)

I will revert the change on master waiting for the update.

johnelliott commented 1 year ago

I just logged back on to say I'd found a way to more reliably reproduce the bug: after the system goes to sleep and the usb mouse RF receiver is off, the bug happens until I switch the mouse off and back on again.

Now as I read the above, it looks like the problem is upstream.

glyph commented 1 year ago

They released a fix 3 days ago!

This makes me feel a little less crazy — as I was hacking around the shifting USB endpoint attributes, I was thinking "this… can't be how it's supposed to work, right?" Glad that this is getting fixed in the right place!

glyph commented 1 year ago

Went to go check whether cython-hidapi had an issue to update to the next release, and I saw @flozz was already on it :). For those following along at home, here's the link: https://github.com/trezor/cython-hidapi/issues/151

johnelliott commented 1 year ago

Thanks so much all, I wish I was more of a help. :)

flozz commented 1 year ago

hidapi was updated to v0.14! If someone can test on macOS with the master branch and this version (pip install hidapi==0.14) :)

flozz commented 1 year ago

I published the v4.9.0 where hidapi is updated to v0.14, so it should work on macOS Ventura. Please let me know if it is not the case. :)

glyph commented 1 year ago

Looking good: https://gist.github.com/glyph/e858a95d4008f905ae6e4516aaa6563d