diff options
| author | Kevin Bowling <kbowling@FreeBSD.org> | 2025-11-21 05:02:00 +0000 |
|---|---|---|
| committer | Kevin Bowling <kbowling@FreeBSD.org> | 2025-11-21 06:44:25 +0000 |
| commit | aa30bab9a92e1be230b9708bff9f33aae7d384e5 (patch) | |
| tree | df14c5277c70a2f2e98ca18ff66686eed7a69946 | |
| parent | 5818b6ee552b302f5300934f9b8cb94881867a5f (diff) | |
e1000: Bump 82574/82583 PBA to 32K
The reporter contacted me with packet loss and throughput fluctuations
on a low power machine (Intel J1900) that got worse with the recent AIM
algorithm in FreeBSD 14.2+.
32K RX PBA matches Linux default. Add a conditional path since we don't
otherwise do a fixup for jumbo frames to retain space for two frames in
Tx.
With this change and an additional errata change, the throughput meets
line rate for the reporter.
Reported by: Codin <codin@nagi.ftp.sh>
Tested by: Codin <codin@nagi.ftp.sh>
MFC after: 2 weeks
| -rw-r--r-- | sys/dev/e1000/if_em.c | 6 |
1 files changed, 5 insertions, 1 deletions
diff --git a/sys/dev/e1000/if_em.c b/sys/dev/e1000/if_em.c index b8ea3168330b..02f4c431badd 100644 --- a/sys/dev/e1000/if_em.c +++ b/sys/dev/e1000/if_em.c @@ -3080,9 +3080,13 @@ em_reset(if_ctx_t ctx) case e1000_82573: pba = E1000_PBA_12K; /* 12K for Rx, 20K for Tx */ break; + /* 82574/82583: Total Packet Buffer is 40K */ case e1000_82574: case e1000_82583: - pba = E1000_PBA_20K; /* 20K for Rx, 20K for Tx */ + if (hw->mac.max_frame_size > 8192) + pba = E1000_PBA_22K; /* 22K for Rx, 18K for Tx */ + else + pba = E1000_PBA_32K; /* 32K for RX, 8K for Tx */ break; case e1000_ich8lan: pba = E1000_PBA_8K; |