Menu
▾
▴
Re: [E1000-devel] 6.2.15 changes affect IPMI
|
From: Leech, C. <chr...@in...> - 2006-10-18 20:58:15
|
I take that back, CHECKSUM_UNNECESSARY isn't affected only CHECKSUM_HW. Only the IP fragmented UDP checksum assist might be disabled in this case. - Chris > -----Original Message----- > From: e10...@li...=20 > [mailto:e10...@li...] On Behalf=20 > Of Leech, Christopher > Sent: Wednesday, October 18, 2006 1:54 PM > To: ch...@ll...; Brandeburg, Jesse > Cc: e10...@li... > Subject: Re: [E1000-devel] 6.2.15 changes affect IPMI >=20 >=20 > I think the length as reported by the Rx descriptors now needs to be > adjusted by -4 when filling out the skb, otherwise Rx checksum offload > will stop working. >=20 > If the CRC is reported as being a valid part of the frame=20 > data, then IP > will "trim" the frame to the end of the IP packet and disregard any > hardware checksum offload information. See the call to > pksb_trim_rcsum() in ip_rcv(). >=20 > - Chris >=20 > > -----Original Message----- > > From: e10...@li...=20 > > [mailto:e10...@li...] On=20 > Behalf Of Al Chu > > Sent: Wednesday, October 18, 2006 1:26 PM > > To: Brandeburg, Jesse > > Cc: e10...@li... > > Subject: Re: [E1000-devel] 6.2.15 changes affect IPMI > >=20 > >=20 > > Hi Jesse, > >=20 > > Is this all that needs to be done?? > >=20 > > --- e1000_main.c.orig 2006-10-18 13:09:07.859552000 -0700 > > +++ e1000_main.c 2006-10-18 13:05:32.786046000 -0700 > > @@ -1588,8 +1588,11 @@ e1000_setup_rctl(struct e1000_adapter *a > > E1000_RCTL_LBM_NO | E1000_RCTL_RDMTS_HALF | > > (adapter->hw.mc_filter_type << E1000_RCTL_MO_SHIFT); > >=20 > > - if (adapter->hw.mac_type > e1000_82543) > > - rctl |=3D E1000_RCTL_SECRC; > > + /* FIXME: disable the stripping of CRC because it breaks > > + * BMC firmware connected over SMBUS > > + if (adapter->hw.mac_type > e1000_82543) > > + rctl |=3D E1000_RCTL_SECRC; > > + */ > >=20 > > if (adapter->hw.tbi_compatibility_on =3D=3D 1) > > rctl |=3D E1000_RCTL_SBP; > > @@ -1650,7 +1653,9 @@ e1000_setup_rctl(struct e1000_adapter *a > > rfctl |=3D E1000_RFCTL_IPV6_DIS; > > E1000_WRITE_REG(&adapter->hw, RFCTL, rfctl); > >=20 > > - rctl |=3D E1000_RCTL_DTYP_PS | E1000_RCTL_SECRC; > > + /* FIXME: disable the stripping of CRC because it > > breaks > > + * BMC firmware connected over SMBUS */ > > + rctl |=3D E1000_RCTL_DTYP_PS /* |=20 > E1000_RCTL_SECRC */; > >=20 > > psrctl |=3D adapter->rx_ps_bsize0 >> > > E1000_PSRCTL_BSIZE0_SHIFT; > >=20 > > If so, no need for a patch. =20 > >=20 > > Thanks, > > Al > >=20 > > On Wed, 2006-10-18 at 10:53 -0700, Brandeburg, Jesse wrote: > > > Hi Al, sorry to cause you pain. > > >=20 > > > we have already found and fixed this issue. When the=20 > > driver turned on > > > the SECRC bit, which causes hardware to strip CRCs from=20 > > incoming packets > > > before transferring to SMBUS (ipmi) or to the host, it=20 > > broke some BMCs > > > that expect the ethernet CRC to be present. > > >=20 > > > e1000-7.0.41 was the first driver with the fix on our=20 > > sourceforge site. > > > the full patch really should include the later driver=20 > > changes to remove > > > the CRC in software too. I can make a patch for your=20 > > redhat kernel if > > > you would like, but simply trying the latest driver from > > > http://prdownloads.sf.net/e1000/e1000-7.2.9.tar.gz (takes you to a > > > download page, not actual file as url suggests) would be a=20 > > useful first > > > step. > > >=20 > > > Have you filed a bugzilla at redhat? > > >=20 > > > Jesse > > >=20 > > > -----Original Message----- > > > From: e10...@li... > > > [mailto:e10...@li...] On=20 > > Behalf Of Al Chu > > > Sent: Wednesday, October 18, 2006 10:28 AM > > > To: e10...@li... > > > Subject: [E1000-devel] 6.2.15 changes affect IPMI > > >=20 > > > Howdy everyone, > > >=20 > > > After upgrading our e1000 driver from 6.0.58 to 7.0.33 (due to an > > > errata update from Redhat) a number of our motherboards stopped > > > working with IPMI. =20 > > >=20 > > > After some grunt work, I found that the issue was caused by=20 > > the changes > > > between 6.1.16 and 6.2.15. The adapter is a 82546. At=20 > > this point in > > > time, nothing stands out in the changes that would make me=20 > > think IPMI > > > would stop working. > > >=20 > > > Is there anyway I could get an individual patch series for=20 > > the changes > > > between these two releases to try and narrow down the=20 > > cause. I'm more > > > than glad to hunt this down but not quite sure where the=20 > > CVS repository > > > is to grab the individual changes. > > >=20 > > > Thanks, > > > Al > > >=20 > > --=20 > > Albert Chu > > ch...@ll... > > 925-422-5311 > > Computer Scientist > > High Performance Systems Division > > Lawrence Livermore National Laboratory > >=20 > >=20 > > -------------------------------------------------------------- > > ----------- > > Using Tomcat but need to do more? Need to support web=20 > > services, security? > > Get stuff done quickly with pre-integrated technology to make=20 > > your job easier > > Download IBM WebSphere Application Server v.1.0.1 based on=20 > > Apache Geronimo > > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057& > dat=3D121642 > _______________________________________________ > E1000-devel mailing list > E10...@li... > https://lists.sourceforge.net/lists/listinfo/e1000-devel >=20 > -------------------------------------------------------------- > ----------- > Using Tomcat but need to do more? Need to support web=20 > services, security? > Get stuff done quickly with pre-integrated technology to make=20 > your job easier > Download IBM WebSphere Application Server v.1.0.1 based on=20 > Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057& dat=3D121642 _______________________________________________ E1000-devel mailing list E10...@li... https://lists.sourceforge.net/lists/listinfo/e1000-devel |
