https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=298560
--- Comment #3 from Chen Zhao <[email protected]> --- Created attachment 274796 --> https://bugs.freebsd.org/bugzilla/attachment.cgi?id=274796&action=edit Native FreeBSD TCP ISS regression reproducer (source, runner, and instructions) I also reproduced this issue on an amd64 FreeBSD VM running in Microsoft Azure. System: FreeBSD 15.1-RELEASE-p3, GENERIC amd64 Kernel revision: releng/15.1-n283611-88e7371d9dc2 Command: sudo sh run.sh long The test transferred 2 GiB + 128 KiB over a single TCP connection. All 1,470,969 advancing ACKs were processed through header prediction, with zero retransmissions. After a window change forced general TCP processing, the final valid ACK carrying one byte of data ('X') was incorrectly discarded as a ghost ACK. recv() returned -1 with errno 35 (EWOULDBLOCK), and the ghost ACK counter increased by one. The reproducer reported: REPRODUCED: valid ACK and X discarded as a ghost ACK Full terminal output follows: ``` azureuser@chenzhao-test:~/freebsd-iss-reproducer $ sudo sh run.sh long FreeBSD chenzhao-test 15.1-RELEASE-p3 FreeBSD 15.1-RELEASE-p3 releng/15.1-n283611-88e7371d9dc2 GENERIC amd64 15.1-RELEASE-p3 15.1-RELEASE-p3 mode=long initial_ghost_ack=rejected target=2147614720 progress: bytes=268435600 elapsed_ms=804 progress: bytes=536871200 elapsed_ms=1605 progress: bytes=805306800 elapsed_ms=2403 progress: bytes=1073742400 elapsed_ms=3207 progress: bytes=1342178000 elapsed_ms=4009 progress: bytes=1610613600 elapsed_ms=4816 progress: bytes=1879049200 elapsed_ms=5620 progress: bytes=2147484800 elapsed_ms=6421 transfer: bytes=2147614720 packets=1470969 advancing_acks=1470969 predicted_acks=1470969 snd_una_minus_iss=2147614721 retransmits=0 elapsed_ms=6422 request: received=-1 errno=35 ghost_ack_delta=1 REPRODUCED: valid ACK and X discarded as a ghost ACK ``` -- You are receiving this mail because: You are the assignee for the bug.
