yangsong8-a1 opened a new pull request, #21993:
URL: https://github.com/apache/nuttx/pull/21993

   ## Summary
   
   Two concurrency fixes in the CDC/ACM class driver, both visible with
   CONFIG_CDCACM_DISABLE_TXBUF=y:
   
   - cdcacm_sndpacket() overwrote priv->wrcontainer with NULL after an
     interrupt-time re-entry (wrcomplete -> inner sndpacket) had already
     assigned a new container, leaking the container and losing data.
     A local snapshot with a conditional write fixes the race.
   - When txfree ran empty after EP_SUBMIT, xmit.buffer still pointed at
     the submitted req->buf, so the serial layer kept writing into the
     buffer the controller owns.  xmit.buffer is now set to NULL so the
     serial layer treats the FIFO as full.
   - g_init_lock moves into cdcacm_classobject() to protect both call
     paths, and cdcuart_txempty() sets priv->ispolling under priv->lock.
   - cdcacm.c now includes <nuttx/mutex.h> directly instead of relying on
     another header to pull it in.
   
   ## Impact
   
   drivers/usbdev/cdcacm.c only.  No API or Kconfig change.  Behavior
   changes only with CONFIG_CDCACM_DISABLE_TXBUF=y (default n).
   
   ## Testing
   
   Ubuntu 22.04, arm-none-eabi-gcc 10.3.  Builds with no new warnings:
   
   - raspberrypi-pico-2:usbnsh (with and without
     CONFIG_CDCACM_DISABLE_TXBUF=y)
   
   sim:usbdev with SIM_USB_RAW_GADGET, dummy_hcd on the host:
   
   - conn 1 (CDCACM+CDCECM), conn 0 (RNDIS) and conn 2 (CDCNCM)
     composite gadgets all enumerate (cdc_acm -> ttyACM0, cdc_ether /
     rndis_host / cdc_ncm -> usb0).
   - With CONFIG_CDCACM_DISABLE_TXBUF=y, `echo USB-TX-OK-123456789 >
     /dev/ttyACM0` in nsh is received unchanged on the host's
     /dev/ttyACM0 (20 bytes).  The write goes through the fixed
     sndpacket() path.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to