Mike Tancsa
2016-Jan-30 16:47 UTC
possible em regression (was Re: svn commit: r294958 - in stable/10: share/man/man4 sys/dev/e1000 sys/dev/ixgb sys/dev/netmap)
On 1/29/2016 8:23 PM, Marius Strobl wrote:> On Fri, Jan 29, 2016 at 03:41:57PM -0500, Mike Tancsa wrote: >> >> No multi queue. Stock GENERIC kernel with a couple of things removed. >> hw.em are just the defaults. I will try without TSO >> >> % ifconfig em0 >> em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 >> >> options=4209b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,WOL_MAGIC,VLAN_HWTSO> >> > > Hrm, that's strange, TSO4 should be enabled by default so apparently > you are already disabling it; what is the behavior if you turn it on? > Do you use a < Gigabit link?Hi Marius, Thanks for looking. The ifconfig output was after I turned off tso as Harry suggested to try. Its been 24hrs and I have not seen any resets. I will wait another 36hrs or so and then turn it back on to see if the problem comes back. this link is 100Mb. The other em nic onboard is a different chip. Its at gigabit and seems to work fine with tso enabled # pciconf -lvcb em1 em1 at pci0:14:0:0: class=0x020000 card=0x109a15d9 chip=0x109a8086 rev=0x00 hdr=0x00 vendor = 'Intel Corporation' device = '82573L Gigabit Ethernet Controller' class = network subclass = ethernet bar [10] = type Memory, range 32, base 0xe8b00000, size 131072, enabled bar [18] = type I/O Port, range 32, base 0x6000, size 32, enabled cap 01[c8] = powerspec 2 supports D0 D3 current D0 cap 05[d0] = MSI supports 1 message, 64 bit enabled with 1 message cap 10[e0] = PCI-Express 1 endpoint max data 128(256) RO NS link x1(x1) speed 2.5(2.5) ecap 0001[100] = AER 1 0 fatal 1 non-fatal 0 corrected ecap 0003[140] = Serial 1 003048ffff9c59f1 # ifconfig em1 em1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 options=4219b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,WOL_MAGIC,VLAN_HWTSO> ether 00:30:48:9c:59:f1 inet6 fe80::230:48ff:fe9c:59f1%em1 prefixlen 64 scopeid 0x2 nd6 options=23<PERFORMNUD,ACCEPT_RTADV,AUTO_LINKLOCAL> media: Ethernet autoselect (1000baseT <full-duplex>) status: active The original problem nic is em0 pciconf -lvcb em0 em0 at pci0:13:0:0: class=0x020000 card=0x108c15d9 chip=0x108c8086 rev=0x03 hdr=0x00 vendor = 'Intel Corporation' device = '82573E Gigabit Ethernet Controller (Copper)' class = network subclass = ethernet bar [10] = type Memory, range 32, base 0xe8a00000, size 131072, enabled bar [18] = type I/O Port, range 32, base 0x5000, size 32, enabled cap 01[c8] = powerspec 2 supports D0 D3 current D0 cap 05[d0] = MSI supports 1 message, 64 bit enabled with 1 message cap 10[e0] = PCI-Express 1 endpoint max data 128(256) RO NS link x1(x1) speed 2.5(2.5) ecap 0001[100] = AER 1 0 fatal 1 non-fatal 0 corrected ecap 0003[140] = Serial 1 003048ffff9c59f0 -- ------------------- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, mike at sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/
Marius Strobl
2016-Jan-30 17:26 UTC
possible em regression (was Re: svn commit: r294958 - in stable/10: share/man/man4 sys/dev/e1000 sys/dev/ixgb sys/dev/netmap)
On Sat, Jan 30, 2016 at 11:47:19AM -0500, Mike Tancsa wrote:> On 1/29/2016 8:23 PM, Marius Strobl wrote: > > On Fri, Jan 29, 2016 at 03:41:57PM -0500, Mike Tancsa wrote: > >> > >> No multi queue. Stock GENERIC kernel with a couple of things removed. > >> hw.em are just the defaults. I will try without TSO > >> > >> % ifconfig em0 > >> em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 > >> > >> options=4209b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,WOL_MAGIC,VLAN_HWTSO> > >> > > > > Hrm, that's strange, TSO4 should be enabled by default so apparently > > you are already disabling it; what is the behavior if you turn it on? > > Do you use a < Gigabit link? > > Hi Marius, > Thanks for looking. The ifconfig output was after I turned off tso as > Harry suggested to try. Its been 24hrs and I have not seen any resets. > I will wait another 36hrs or so and then turn it back on to see if the > problem comes back. > > this link is 100Mb.Ah, okay, that at least makes sense. Can you please verify that with the attached patch applied, you have a setup that works out of the box? Marius -------------- next part -------------- A non-text attachment was scrubbed... Name: em_tso_gig_only_10.diff Type: text/x-diff Size: 780 bytes Desc: not available URL: <http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20160130/d6e9d6ef/attachment.diff>
Mike Tancsa
2016-Feb-01 22:11 UTC
possible em regression (was Re: svn commit: r294958 - in stable/10: share/man/man4 sys/dev/e1000 sys/dev/ixgb sys/dev/netmap)
On 1/30/2016 12:26 PM, Marius Strobl wrote:> > Ah, okay, that at least makes sense. Can you please verify that with > the attached patch applied, you have a setup that works out of the > box? >Hi, The patch does not apply cleanly # patch < em_tso_gig_only_10.diff Hmm... Looks like a unified diff to me... The text leading up to this was: -------------------------- |Index: sys/dev/e1000/if_em.c |==================================================================|--- sys/dev/e1000/if_em.c (revision 294962) |+++ sys/dev/e1000/if_em.c (working copy) -------------------------- Patching file sys/dev/e1000/if_em.c using Plan A... Hunk #1 failed at 1377. 1 out of 1 hunks failed--saving rejects to sys/dev/e1000/if_em.c.rej done # cat sys/dev/e1000/if_em.c.rej @@ -1377,8 +1377,15 @@ ifp->if_hwassist = 0; if (ifp->if_capenable & IFCAP_TXCSUM) ifp->if_hwassist |= (CSUM_TCP | CSUM_UDP); - if (ifp->if_capenable & IFCAP_TSO4) - ifp->if_hwassist |= CSUM_TSO; + /* + ** There have proven to be problems with TSO when not + ** at full gigabit speed, so disable the assist automatically + ** when at lower speeds. -jfv + */ + if (ifp->if_capenable & IFCAP_TSO4) { + if (adapter->link_speed == SPEED_1000) + ifp->if_hwassist |= CSUM_TSO; + } /* Configure for OS presence */ em_init_manageability(adapter); -- ------------------- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, mike at sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/
Mike Tancsa
2016-Feb-02 14:49 UTC
possible em regression (was Re: svn commit: r294958 - in stable/10: share/man/man4 sys/dev/e1000 sys/dev/ixgb sys/dev/netmap)
On 1/30/2016 12:26 PM, Marius Strobl wrote:> On Sat, Jan 30, 2016 at 11:47:19AM -0500, Mike Tancsa wrote: >> On 1/29/2016 8:23 PM, Marius Strobl wrote: >>> On Fri, Jan 29, 2016 at 03:41:57PM -0500, Mike Tancsa wrote: >>>> >>>> No multi queue. Stock GENERIC kernel with a couple of things removed. >>>> hw.em are just the defaults. I will try without TSO >>>> >>>> % ifconfig em0 >>>> em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 >>>> >>>> options=4209b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,WOL_MAGIC,VLAN_HWTSO> >>>> >>> >>> Hrm, that's strange, TSO4 should be enabled by default so apparently >>> you are already disabling it; what is the behavior if you turn it on? >>> Do you use a < Gigabit link? >> >> Hi Marius, >> Thanks for looking. The ifconfig output was after I turned off tso as >> Harry suggested to try. Its been 24hrs and I have not seen any resets. >> I will wait another 36hrs or so and then turn it back on to see if the >> problem comes back. >> >> this link is 100Mb. > > Ah, okay, that at least makes sense. Can you please verify that with > the attached patch applied, you have a setup that works out of the > box?Hi, Should the nic come up with TSO disabled by default ? After reboot, I see # ifconfig em0 em0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500 options=4219b<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,VLAN_HWCSUM,TSO4,WOL_MAGIC,VLAN_HWTSO> ether 00:30:48:9c:59:f0 ---Mike -- ------------------- Mike Tancsa, tel +1 519 651 3400 Sentex Communications, mike at sentex.net Providing Internet services since 1994 www.sentex.net Cambridge, Ontario Canada http://www.tancsa.com/