[Bug 297383] axge(4): RX aggregation timer far too short, uninitialised reads on control transfer failure, TX padding bit never set
| Newsgroups | gmane.os.freebsd.devel.net |
|---|---|
| Message-ID | <[email protected]/bugzilla/> |
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=297383 --- Comment #7 from Oleg <[email protected]> --- Thank you for performing the A/B/A test and verifying that the IMOD register actually changed. Your results show that reducing the interval to 16 microseconds does not improve throughput on your AMD FCH controller. You previously mentioned a local TX queue-depth change from 1 to 4, but I do not see that change in either attached patch. Was it enabled during your latest 711–716 Mbit/s TX tests? If it was not, could you repeat the test with that change and consider attaching the TX patch? If it was enabled, the remaining limitation may be elsewhere in the axge TX path rather than in xHCI interrupt moderation. -- You are receiving this mail because: You are the assignee for the bug.