fix(dcd_ch32_usbfs): fix dropped bulk OUT packets by properly re-arming EP_RX_CTRL state#3622
Merged
Conversation
…ng EP_RX_CTRL state
|
| target | .text | .rodata | .data | .bss | total | % diff |
|---|---|---|---|---|---|---|
| ch32v307v_r1_1v0/net_lwip_webserver | 52,120 → 56,304 (+4,184) | — | — | 18,276 → 26,492 (+8,216) | 73,388 → 85,792 (+12,404) | +16.9% |
| ch32v307v_r1_1v0/dfu_runtime | 8,908 → 9,360 (+452) | — | — | 1,056 → 1,092 (+36) | 12,556 → 13,048 (+492) | +3.9% |
| ch32v307v_r1_1v0/hid_generic_inout | 10,032 → 10,500 (+468) | — | — | 1,260 → 1,296 (+36) | 13,884 → 14,388 (+504) | +3.6% |
| ch32v307v_r1_1v0/hid_multiple_interface | 10,852 → 11,316 (+464) | — | — | 1,136 → 1,172 (+36) | 14,580 → 15,084 (+504) | +3.5% |
| ch32v307v_r1_1v0/hid_boot_interface | 10,876 → 11,340 (+464) | — | — | 1,136 → 1,172 (+36) | 14,604 → 15,108 (+504) | +3.5% |
| ch32v307v_r1_1v0/hid_composite | 11,224 → 11,692 (+468) | — | — | 1,160 → 1,196 (+36) | 14,976 → 15,484 (+508) | +3.4% |
| ch32f205r-r0/dfu_runtime | 7,580 → 7,928 (+348) | — | — | 1,416 → 1,452 (+36) | 11,440 → 11,824 (+384) | +3.4% |
| ch32f205r-r0/hid_generic_inout | 8,588 → 8,944 (+356) | — | — | 1,620 → 1,656 (+36) | 12,480 → 12,872 (+392) | +3.1% |
| ch32v307v_r1_1v0/audio_test | 12,508 → 12,984 (+476) | — | — | 1,712 → 1,748 (+36) | 16,844 → 17,360 (+516) | +3.1% |
| ch32f205r-r0/hid_multiple_interface | 9,200 → 9,556 (+356) | — | — | 1,496 → 1,532 (+36) | 13,140 → 13,532 (+392) | +3.0% |
Contributor
There was a problem hiding this comment.
Pull request overview
This PR targets reliability of bulk OUT transfers on WCH CH32 USBFS device controller by ensuring OUT endpoints are properly re-armed via EP_RX_CTRL, preventing missed receive callbacks and dropped packets during sustained high-rate transfers.
Changes:
- Update
update_out()to programEP_RX_CTRL(ACK/NAK) for non-EP0 OUT endpoints based on whether an OUT transfer is still active. - Update
dcd_edpt_xfer()to explicitly re-enable OUT endpoint RX ACK when a new OUT transfer is queued.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
118
to
124
|
|
||
| if (ep == 0) { | ||
| EP_RX_CTRL(0) = USBFS_EP_R_RES_ACK; | ||
| } else { | ||
| EP_RX_CTRL(ep) = (EP_RX_CTRL(ep) & ~USBFS_EP_R_RES_MASK) | | ||
| (xfer->valid ? USBFS_EP_R_RES_ACK : USBFS_EP_R_RES_NAK); | ||
| } |
| if (dir == TUSB_DIR_IN) { | ||
| update_in(rhport, ep, true); | ||
| } else { | ||
| EP_RX_CTRL(ep) = (EP_RX_CTRL(ep) & ~USBFS_EP_R_RES_MASK) | USBFS_EP_R_RES_ACK; |
Collaborator
|
Thank you, I think we should also remove ACK enablement in I don't have CH30V20x at hand but surprised by their USB IP implementation, if CPU load is high setting NAK after packet received can still missing data. |
Collaborator
|
I bought a nanoch32v203 (and v305), I do observe OUT EP send ACK blindly without checking if data is actually read 😱 😱 Captured with CDC example: I think to fix it correctly we need to disable |
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
OUT is still buggy Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
1 task
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
It would casue race condition, SETUP packet is always acked. Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Refactor the driver to follow USBFS style for easier maintenance. Replace the old packet/response helpers with explicit queue and update paths for IN and OUT transfers. Introduce transfer validity tracking and per-endpoint data toggle state. Reset toggle state on init, close-all, endpoint close, clear-stall, and bus reset. Initialize endpoint controls consistently in NAK + TOG_0 mode. Tighten EP0 setup/status handling and route transfer IRQ processing through the transfer-flag path. Stop enabling ISO_ACT in INT_EN and clear unhandled interrupt flags explicitly. Signed-off-by: HiFiPhile <admin@hifiphile.com>
Signed-off-by: HiFiPhile <admin@hifiphile.com>
Collaborator
|
I've included a large transfer reliability patch for both USBFS/USBHS driver.
Tested on CH32V203/CH32V305 with Concurrent ISO transfer is not working yet, eg |
HiFiPhile
approved these changes
May 23, 2026
This was referenced May 23, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
When stress-testing a custom Vendor interface with lots of fast bulk transfers on CH32V30x (which use the USBFS controller), I noticed that a lot of data was dropped because
tud_vendor_rx_cb()would not always fire.After investigating, it seemed like
dcd_ch32_usbfswasn't updating theEP_RX_CTRLregister after a transfer completed or when a new one was queued.update_out()now properly setsEP_RX_CTRLto ACK or NAK for all endpoints based on transfer state, anddcd_edpt_xfer()explicitly re-enables ACK when re-queuing an OUT transfer.Bulk transfers now complete reliably without data loss or missed callbacks.