d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

The TCP checksum offload handling in the 8111B/8168B and 8101E PCIe can

Browse Source 
apparently be confused by short TCP segments that have been manually
padded to the minimum ethernet frame size. The driver does short frame
padding in software as a workaround for a bug in the 8169 PCI devices
that causes short IP fragments to be corrupted due to an apparent
conflict between the hardware autopadding and hardware IP checksumming.

To fix this, we avoid software padding for short TCP segments, since
the hardware seems to autopad and checksum these correctly (even the
older 8169 NICs get these right). Short UDP packets appear to be
handled correctly in all cases. This should work around the IP header
checksum bug in the 8169 while not tripping the TCP checksum bug in
the 8111B/8168B and 8101E.
This commit is contained in:
Bill Paul 2007-01-25 17:30:30 +00:00
parent 63cb891e8b
commit e2bcb489ef
1 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
sys/dev/re/if_re.c
View File

@ -2071,8 +2071,13 @@ re_encap(sc, m_head, idx)
* the mbuf chain has too many fragments so the coalescing code
* below can assemble the packet into a single buffer that's
* padded out to the mininum frame size.
*
* Note: this appears unnecessary for TCP, and doing it for TCP
* with PCIe adapters seems to result in bad checksums.
*/
if (arg.rl_flags && (*m_head)->m_pkthdr.len < RL_MIN_FRAMELEN)
if (arg.rl_flags && !(arg.rl_flags & RL_TDESC_CMD_TCPCSUM) &&
(*m_head)->m_pkthdr.len < RL_MIN_FRAMELEN)
error = EFBIG;
else
error = bus_dmamap_load_mbuf(sc->rl_ldata.rl_mtag, map,