Add USBHost v2 #4032

i404788 · 2025-04-01T08:00:18Z

Follow up to #3307 (see for discussion)

This add UsbHostDriver & UsbChannel traits allowing HAL's to implement Usb Host support.

Example handlers are in embassy-usb/src/handlers/, and there is additionally a UAC handler made by @AlexCharlton available.

Changes since #3307:

UsbChannels now have a set_timeout to enable configurable timeouts
Descriptor parsing is refactored to use iterators, this is more intuitive since parsing needs to be done sequentially
Configuration of Usb Devices now happens after enumeration, and can be done either by the handler (usually deferring to the active config, example handlers) or the application.

Pending tasks:

Update embassy-rp to use the latest change
Update stm32g0 to add set_timeout
Update esp-hal (out-of-tree) to add set_timeout
Documentation & configs clean up

…isochronuous channels

Use stat_tx and stat_rx instead of nak() flag

Should only be included for usb_v4 peripherals.

Move retarget_channel from UsbHostDriver to UsbChannel trait implementation.

It's already implemented in mod.rs

Compiles and runs ok on a rp2040 board.

lynxD · 2025-04-10T12:17:49Z

examples/rp/src/bin/usb_host_keyboard.rs

+    println!("Found device with speed = {:?}", speed);
+
+    let enum_info = usbhost.enumerate_root_bare(speed, 1).await.unwrap();
+    let mut kbd = KbdHandler::try_register(&usbhost, &enum_info)


How would I go about supporting the keyboard being unplugged and plugged in again?
Might be worth it to add that to the example.

EDIT:
It seems to me that KbdHandler::wait_for_event should return when the device is disconnected so the driver can be cleaned up. Right now it awaiting on that just hangs forever.

There is HandlerEvent::HandlerDisconnected but it's not triggered on disconnect. I don't know if it's a bug or it's for something else.

I've managed to get re-plugging devices to partially work. Seems channels get exhausted after some time (probably due to lack of drop_channel). All in all it's reliable only the first time the device is plugged in.

#![no_std] #![no_main] use defmt::*; use embassy_executor::Spawner; use embassy_rp::{bind_interrupts, peripherals, usb}; use embassy_usb_driver::host::{UsbHostDriver, DeviceEvent}; use embassy_usb::host::UsbHostBusExt; use embassy_usb::handlers::{UsbHostHandler, kbd}; use embassy_futures::select::{select, Either}; use embassy_usb_driver::Speed; use {defmt_rtt as _, panic_probe as _}; bind_interrupts!(struct Irqs { USBCTRL_IRQ => usb::host::InterruptHandler<peripherals::USB>; }); #[embassy_executor::main] async fn main(_spawner: Spawner) -> ! { let p = embassy_rp::init(Default::default()); let mut usbhost = usb::host::Driver::new(p.USB, Irqs); loop { match usbhost.wait_for_device_event().await { DeviceEvent::Connected(speed) => { handle_device(&mut usbhost, speed).await; } DeviceEvent::Disconnected => { warn!("Disconnected when there's no device?"); } } }; } async fn handle_device( usbhost: &mut embassy_rp::usb::host::Driver<'_, peripherals::USB>, speed: Speed) { let enum_info = usbhost.enumerate_root_bare(speed, 1).await.unwrap(); info!("enum_info.device_desc = {}", enum_info.device_desc); let mut kbd = kbd::KbdHandler::try_register(usbhost, &enum_info).await.expect("Couldn't register keyboard"); handle_kbd_events(usbhost, &mut kbd).await; } async fn handle_kbd_events( usbhost: &mut embassy_rp::usb::host::Driver<'_, peripherals::USB>, kbd: &mut kbd::KbdHandler<embassy_rp::usb::host::Driver<'_, peripherals::USB>>){ loop { let result = select(usbhost.wait_for_device_event(), kbd.wait_for_event()).await; match result { Either::First(dev_event) => match dev_event { DeviceEvent::Connected(_) => { defmt::panic!("More than one device?"); } DeviceEvent::Disconnected => { info!("Disconnected!"); break; } } Either::Second(kbd_event) => { info!("keyboard event = {}", kbd_event); } } }; }

lynxD · 2025-04-24T17:47:04Z

embassy-usb/src/host/mod.rs

+    /// Request the underlying bytes for a descriptor of a specific type.
+    /// bytes.len() determines how many bytes are read at maximum.
+    /// This can be used for descriptors of varying length, which are parsed by the caller.
+    async fn request_descriptor_bytes<T: USBDescriptor>(


This can't use "T: USBDescriptor" as USBDescriptor trait requires the descriptor to be of a fixed size.
I'll try preparing a patch that addresses takes DESC_TYPE as another another argument.
I'm not sure if it would be better to have USBDescriptor not to enforce fixed size (I don't know how/if this can be accomplished). Maybe rename it to something like FixedSizeUSBDescriptor.
From what I understand(correct me if I'm wrong) HID Report descriptors are also USB descriptors but they aren't fixed size.

Yeah, we're kind of deprecating the USBDescriptor trait (as it was originally used) as it's only useful for fetching the headers of descriptors. Since we are restricted on memory buffer the typical flow of a descriptor request is fetch header -> check if static buffer can support descriptor -> fetch buffer as size of descriptor. For HID specifically you fetch the entire ConfigurationDescriptor and then use the RawDescriptorIterator as shown here 39c9d16

Could definitely rename it, maybe USBDescriptorHeader or something similar would be more apt?

Report descriptors don't even contain their own length(it's in corresponding HID descriptor). It's more akin to a string.
So I'm not sure it would be useful for that.
I don't have better idea for the name.

Hmm it seems like the length of the report should be in the wItemLength in HIDDescriptor (EndpointDescriptor eqv.). So yeah USBDescriptorHeader wouldn't apply for ReportDescriptors

lynxD · 2025-04-24T17:56:46Z

embassy-usb/src/handlers/mod.rs

+use crate::host::descriptor::{ConfigurationDescriptor, DeviceDescriptor, USBDescriptor};
+use crate::host::ControlChannelExt;
+
+pub mod hub;


From the name alone this could mean two things. Either host hub driver or implementation of a device hub.
I can imagine both being useful. One could make a device that shows as a hub with two devices connected. Heck pio on rp2040 can be used as USB host so it would be possible to make a hub based on that eg a rp2040 based USB keyboard with a port for mice. Or split keyboard that is connected together via USB (and one that actually works as a true USB port albeit slow).

I think it would be great to make it immediately obvious that this is host hub driver.
What do You folks think?

The handlers module is current for host drivers but we can move it to host. Also I think it would be interesting to have an emulated hub for supporting multiple virtual devices, but that seems better for another PR.

lynxD · 2025-04-24T18:59:45Z

embassy-usb/src/handlers/kbd.rs

+        const SET_PROTOCOL: u8 = 0x0B;
+        const BOOT_PROTOCOL: u16 = 0x0000;
+        if let Err(err) = control_channel
+            .class_request_out(SET_PROTOCOL, BOOT_PROTOCOL, iface.interface_number as u16, &[])


This is a bit unfortunate as according to the HID spec the simplified, boot protocol is optional and the full report protocol is mandatory. So we should probably just use the report protocol for everything. I'd think we want to support something like game pads/joysticks anyway and the boot protocol is defined for keyboard and mice only.
I'll trying to write report protocol support atm.

Yeah this was just made from an example of a previous keyboard driver to get something that works for a few devices we tested, not necessarily fully spec-compliant.
If report protocol can help increase compatibility then we should definitely incorporate that.

I've been tinkering with hid last few days. I've came to conclusion it would deserve a crate of it's own. From what I see Bluetooth reuses the same report descriptors and parsing those is not insignificant amount of code.
I'd think we would be better served by a different example than a keyboard. I haven't checked but it seems to me that mass storage class would should be simple, useful and short. I might get around to writing it later.

I think it would be good to have another example that relies on vendor_id and product_id. So that we get the driver choice right. I don't have idea of a device that would not have a class, Not be a device that should have been a class device. Would have a simple and short driver. Would be useful to embassy users. Anyone has an idea? If we don't get any we can make a dummy driver.

0.000431 DEBUG Detecting device └─ usb_host_keyboard::____embassy_main_task::{async_fn#0} @ src/bin/usb_host_keyboard.rs:25 Found device with speed = Low 0.161293 ERROR panicked at src/bin/usb_host_keyboard.rs:36:65: called `Result::unwrap()` on an `Err` value: ChannelError(Stall)

USBDescriptor trait requires the descriptor to be of a fixed size and this function is meant for getting descriptors of variable size.

lynxD · 2025-04-25T11:41:35Z

I've fixed some bugs:
https://github.com/lynxD/embassy/tree/usb-asynch-v2

Qix- · 2025-04-25T12:04:18Z

@lynxD you can PR into the author's branch on their fork by the way. If merged, the changes will show up here automatically.

Fixes for usb-asynch

lynxD · 2025-04-30T18:39:26Z

embassy-usb/src/host/descriptor.rs

+            // FIXME: Fallible, propegate errors?
+            let iface = InterfaceDescriptor::try_from_bytes(remaining_buf).ok()?;
+            self.offset += iface.len as usize + iface.buffer.len();
+            self.index += 1;


There is a bug here. It assumes that there is one interface descriptor per interface.

A DS4 gamepad gives me following configuration descriptor decoded with (http://eleccelerator.com/usbdescreqparser/):

0x09, // bLength 0x02, // bDescriptorType (Configuration) 0xE1, 0x00, // wTotalLength 225 0x04, // bNumInterfaces 4 0x01, // bConfigurationValue 0x00, // iConfiguration (String Index) 0xC0, // bmAttributes Self Powered 0xFA, // bMaxPower 500mA 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x00, // bInterfaceNumber 0 0x00, // bAlternateSetting 0x00, // bNumEndpoints 0 0x01, // bInterfaceClass (Audio) 0x01, // bInterfaceSubClass (Audio Control) 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x0A, // bLength 0x24, // bDescriptorType (See Next Line) 0x01, // bDescriptorSubtype (CS_INTERFACE -> HEADER) 0x00, 0x01, // bcdADC 1.00 0x47, 0x00, // wTotalLength 71 0x02, // binCollection 0x02 0x01, // baInterfaceNr 1 0x02, // baInterfaceNr 2 0x0C, // bLength 0x24, // bDescriptorType (See Next Line) 0x02, // bDescriptorSubtype (CS_INTERFACE -> INPUT_TERMINAL) 0x01, // bTerminalID 0x01, 0x01, // wTerminalType (USB Streaming) 0x06, // bAssocTerminal 0x02, // bNrChannels 2 0x03, 0x00, // wChannelConfig (Left and Right Front) 0x00, // iChannelNames 0x00, // iTerminal 0x0A, // bLength 0x24, // bDescriptorType (See Next Line) 0x06, // bDescriptorSubtype (CS_INTERFACE -> FEATURE_UNIT) 0x02, // bUnitID 0x01, // bSourceID 0x01, // bControlSize 1 0x03, 0x00, // bmaControls[0] (Mute,Volume) 0x00, 0x00, // bmaControls[1] (None) 0x09, // bLength 0x24, // bDescriptorType (See Next Line) 0x03, // bDescriptorSubtype (CS_INTERFACE -> OUTPUT_TERMINAL) 0x03, // bTerminalID 0x02, 0x04, // wTerminalType (Headset) 0x04, // bAssocTerminal 0x02, // bSourceID 0x00, // iTerminal 0x0C, // bLength 0x24, // bDescriptorType (See Next Line) 0x02, // bDescriptorSubtype (CS_INTERFACE -> INPUT_TERMINAL) 0x04, // bTerminalID 0x02, 0x04, // wTerminalType (Headset) 0x03, // bAssocTerminal 0x01, // bNrChannels 1 0x00, 0x00, // wChannelConfig 0x00, // iChannelNames 0x00, // iTerminal 0x09, // bLength 0x24, // bDescriptorType (See Next Line) 0x06, // bDescriptorSubtype (CS_INTERFACE -> FEATURE_UNIT) 0x05, // bUnitID 0x04, // bSourceID 0x01, // bControlSize 1 0x03, 0x00, // bmaControls[0] (Mute,Volume) 0x00, // iFeature 0x09, // bLength 0x24, // bDescriptorType (See Next Line) 0x03, // bDescriptorSubtype (CS_INTERFACE -> OUTPUT_TERMINAL) 0x06, // bTerminalID 0x01, 0x01, // wTerminalType (USB Streaming) 0x01, // bAssocTerminal 0x05, // bSourceID 0x00, // iTerminal 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x01, // bInterfaceNumber 1 0x00, // bAlternateSetting 0x00, // bNumEndpoints 0 0x01, // bInterfaceClass (Audio) 0x02, // bInterfaceSubClass (Audio Streaming) 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x01, // bInterfaceNumber 1 0x01, // bAlternateSetting 0x01, // bNumEndpoints 1 0x01, // bInterfaceClass (Audio) 0x02, // bInterfaceSubClass (Audio Streaming) 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x07, // bLength 0x24, // bDescriptorType (See Next Line) 0x01, // bDescriptorSubtype (CS_INTERFACE -> AS_GENERAL) 0x01, // bTerminalLink 0x01, // bDelay 1 0x01, 0x00, // wFormatTag (PCM) 0x0B, // bLength 0x24, // bDescriptorType (See Next Line) 0x02, // bDescriptorSubtype (CS_INTERFACE -> FORMAT_TYPE) 0x01, // bFormatType 1 0x02, // bNrChannels (Stereo) 0x02, // bSubFrameSize 2 0x10, // bBitResolution 16 0x01, // bSamFreqType 1 0x00, 0x7D, 0x00, // tSamFreq[1] 32000 Hz 0x09, // bLength 0x05, // bDescriptorType (See Next Line) 0x01, // bEndpointAddress (OUT/H2D) 0x09, // bmAttributes (Isochronous, Adaptive, Data EP) 0x84, 0x00, // wMaxPacketSize 132 0x01, // bInterval 1 (unit depends on device speed) 0x00, // bRefresh 0x00, // bSyncAddress 0x07, // bLength 0x25, // bDescriptorType (See Next Line) 0x01, // bDescriptorSubtype (CS_ENDPOINT -> EP_GENERAL) 0x00, // bmAttributes (None) 0x00, // bLockDelayUnits 0x00, 0x00, // wLockDelay 0 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x02, // bInterfaceNumber 2 0x00, // bAlternateSetting 0x00, // bNumEndpoints 0 0x01, // bInterfaceClass (Audio) 0x02, // bInterfaceSubClass (Audio Streaming) 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x02, // bInterfaceNumber 2 0x01, // bAlternateSetting 0x01, // bNumEndpoints 1 0x01, // bInterfaceClass (Audio) 0x02, // bInterfaceSubClass (Audio Streaming) 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x07, // bLength 0x24, // bDescriptorType (See Next Line) 0x01, // bDescriptorSubtype (CS_INTERFACE -> AS_GENERAL) 0x06, // bTerminalLink 0x01, // bDelay 1 0x01, 0x00, // wFormatTag (PCM) 0x0B, // bLength 0x24, // bDescriptorType (See Next Line) 0x02, // bDescriptorSubtype (CS_INTERFACE -> FORMAT_TYPE) 0x01, // bFormatType 1 0x01, // bNrChannels (Mono) 0x02, // bSubFrameSize 2 0x10, // bBitResolution 16 0x01, // bSamFreqType 1 0x80, 0x3E, 0x00, // tSamFreq[1] 16000 Hz 0x09, // bLength 0x05, // bDescriptorType (See Next Line) 0x82, // bEndpointAddress (IN/D2H) 0x05, // bmAttributes (Isochronous, Async, Data EP) 0x22, 0x00, // wMaxPacketSize 34 0x01, // bInterval 1 (unit depends on device speed) 0x00, // bRefresh 0x00, // bSyncAddress 0x07, // bLength 0x25, // bDescriptorType (See Next Line) 0x01, // bDescriptorSubtype (CS_ENDPOINT -> EP_GENERAL) 0x00, // bmAttributes (None) 0x00, // bLockDelayUnits 0x00, 0x00, // wLockDelay 0 0x09, // bLength 0x04, // bDescriptorType (Interface) 0x03, // bInterfaceNumber 3 0x00, // bAlternateSetting 0x02, // bNumEndpoints 2 0x03, // bInterfaceClass 0x00, // bInterfaceSubClass 0x00, // bInterfaceProtocol 0x00, // iInterface (String Index) 0x09, // bLength 0x21, // bDescriptorType (HID) 0x11, 0x01, // bcdHID 1.11 0x00, // bCountryCode 0x01, // bNumDescriptors 0x22, // bDescriptorType[0] (HID) 0xFB, 0x01, // wDescriptorLength[0] 507 0x07, // bLength 0x05, // bDescriptorType (Endpoint) 0x84, // bEndpointAddress (IN/D2H) 0x03, // bmAttributes (Interrupt) 0x40, 0x00, // wMaxPacketSize 64 0x05, // bInterval 5 (unit depends on device speed) 0x07, // bLength 0x05, // bDescriptorType (Endpoint) 0x03, // bEndpointAddress (OUT/H2D) 0x03, // bmAttributes (Interrupt) 0x40, 0x00, // wMaxPacketSize 64 0x05, // bInterval 5 (unit depends on device speed) // 225 bytes

Notice how there are multiple descriptors with bInterfaceNumber = 1.
This iterator would never get to interface with bInterfaceNumber = 3 (the hid class one).

I'll fix it so that the user will get all of them. At minimum this needs to be mentioned in doc string so the user knows to expect more then num_interfaces members.

Hmm yeah seems like we're not accounting for bAlternateSetting variations, we could try to parse all descriptors (should account for most edge-cases) but it would be a bit slower. Would definitely be good to mention in docs as reminder of this behaviour.

I've made a fix to this and some other problems:
i404788#2

@i404788 kind reminder, Could you take a look?

@i404788 Ping

lynxD · 2025-05-06T18:19:55Z

embassy-usb-driver/src/host.rs

+/// Errors returned by [`ChannelOut::write`] and [`ChannelIn::read`]
+#[derive(Copy, Clone, Eq, PartialEq, Debug)]
+#[cfg_attr(feature = "defmt", derive(defmt::Format))]
+pub enum ChannelError {


Am I wrong or what we call channel is called pipe in USB spec?
If so I'd think we should align ourselves with the spec.

Qix- · 2025-05-25T13:58:55Z

From what I gather this doesn't cover ULPI-based hosts in e.g. stm32, does it?

lynxD · 2025-06-19T15:30:20Z

It's been more than a month since last activity of @i404788 (this PR's author) here. I've created new PR to push this forward.
#4321

i404788 · 2025-06-20T20:06:47Z

@lynxD Thanks for picking it up, I currently (and probably won't anytime soon) have time to work on it, so I'll close my PR in favor of #4321.

If I do end up with some spare time somehow, I'll help review/resolve pending issues.

ferris and others added 30 commits March 22, 2025 14:08

Add interrupt/isochron scheduling registers

f96c7a4

Add helper for framelistlen

2f40d35

Add doping register

30f4ff1

Describe Hcdma reg, add HcdmaB & split hctsiz->xfersiz for interrupt/…

b42fb9c

…isochronuous channels

Fix hcmda reg definition

d22bb83

Add more regs

1b3b832

Add descriptor DMA setting & fix perschedena

e96a29f

Add frame interval dynamic loading ctl

f81ac75

Add frame list addr register

5793a9f

Fix perschena mutability & fifo no-op

143c963

Restore product_id W without no-op

705d29e

Improve docs & remove unnecessary unsafe

e8459cf

Create initial USB Host driver

5743c97

Create usb_host HAL driver module

62d27c6

Make set_address async and wait for completion

e43dc4a

Implement basic request_out api

8a696e0

Implement basic out_request for requesting descriptor

fa07fbe

Use proper status flaxs for detecting finisheed transactions

95433e8

Use stat_tx and stat_rx instead of nak() flag

Implement multi packet receiving

8781600

Implement Endpoint0 change max_oacket_size and realloc

d3dd928

Add USBHost Trait to embassy-usb-driver crate

b467053

Implement Read for non-control endpoints

24a3c63

Remove debug prints

cb03291

Reuse channel write/read for control transfers

612e898

Disable channels on error

c198d3e

Make usb_host available for usb_v4 peripherals

60fe570

Add USB Host implementation and example for stm32g0

156bc53

Add docstrings

8a41d00

Add comment about NUCLE0-G0B1RE together with X-NUCLEO-DRP1M1

a1ebc19

Try fix conditional compilation for usb_host.rs

da8fe18

Should only be included for usb_v4 peripherals.

lynxD and others added 13 commits April 9, 2025 12:48

embassy_rp: rm duplcated code

3cf826a

embassy-rp: Allign to changed traits

6dbfacc

Move retarget_channel from UsbHostDriver to UsbChannel trait implementation.

embassy-rp: Remove duplicate code

28210df

It's already implemented in mod.rs

Allign to DeviceEvent::Connected being parametrized by Speed

9777225

Remove unused use-s

d5a3c8a

Use uppercase names for globals

cdf6c31

Use EndpointInfo instead of EndpointDescriptor

fc24fad

Correct type name in doc

2546f79

Placeholder implementation of set_timeout to satisfy trait

793fb4d

Fix examples/rp/.../usb_host_keyboard.rs

d69c13d

Compiles and runs ok on a rp2040 board.

Remove drop_channel and move to unimplemented Drop for Channel

cbd8538

Remove usize from OUT direction requests

21a66b0

Add missing Synopsys registers

16b39fa

lynxD reviewed Apr 10, 2025

View reviewed changes

lynxD reviewed Apr 24, 2025

View reviewed changes

lynxD added 3 commits April 24, 2025 22:36

Fix compilation

6c12419

Take desc_type from argument in request_descriptor_bytes

af5f2ef

USBDescriptor trait requires the descriptor to be of a fixed size and this function is meant for getting descriptors of variable size.

Merge pull request #1 from lynxD/usb-asynch-v2

853f770

Fixes for usb-asynch

lynxD reviewed Apr 30, 2025

View reviewed changes

lynxD reviewed May 6, 2025

View reviewed changes

lynxD mentioned this pull request Jun 19, 2025

Add USBHost #4321

Draft

i404788 closed this Jun 20, 2025

Add USBHost v2 #4032

Add USBHost v2 #4032

Conversation

i404788 commented Apr 1, 2025

Uh oh!

lynxD Apr 10, 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

i404788 Apr 25, 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!

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!

lynxD Apr 24, 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!

Choose a reason for hiding this comment

Uh oh!

lynxD Apr 29, 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!

lynxD commented Apr 25, 2025

Uh oh!

Qix- commented Apr 25, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lynxD May 5, 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!

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!

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

Uh oh!

Uh oh!

lynxD commented Jun 19, 2025

Uh oh!

i404788 commented Jun 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

lynxD Apr 10, 2025 •

edited

Loading

i404788 Apr 25, 2025 •

edited

Loading

lynxD Apr 24, 2025 •

edited

Loading

lynxD Apr 29, 2025 •

edited

Loading

lynxD May 5, 2025 •

edited

Loading

Qix- commented May 25, 2025 •

edited

Loading

i404788 commented Jun 20, 2025 •

edited

Loading