FreeBSD 9.0-RELEASE amd64 freezes on a xfx i780 motherboard nforce chipset when connecting 2 computers together directly. There was no kernel panic screen shown. There is something wrong with the nfe0 driver when the link is connected and passing packets. This PR report that has been closed references this exact problem in a different way. http://www.freebsd.org/cgi/query-pr.cgi?pr=157132 I am attaching output of pciconf -lcbv, vmstat -i. Fix: NONE Patch attached with submission follows: How-To-Repeat: With a xfx i780 motherboard nforce chipset,connect 2 computers manually. assigning them ips and netmask manually.
Responsible Changed From-To: freebsd-bugs->freebsd-net Over to maintainer(s).
State Changed From-To: open->feedback Would you post dmesg output related with nfe(4) and its PHY driver? Your vmstat output indicates it didn't generate any interrupts from ethernet controller so how about disabling MSI-X or MSI?
Responsible Changed From-To: freebsd-net->yongari Grab.
On Thu, Mar 01, 2012 at 10:12:20AM -0800, Robert Z wrote: > Hi Yongari > > Found this in the FreeBSD email lists hope it helps. The list talks about? > other boards having issues with the same chipset. The quick hack? > there doing is blacklisting MSI support on motherboards in the nfe driver. > > I wish the MSI issue could be solved instead of doing it this way, witch is? I wonder how driver could fix silicon bug. MSI/MSI-X with NVIDIA chipset is horribly broken in various ways. Making it work on these chipset may require lots of hacks by accessing undocumented registers as it did in Linux(NVIDIA also did not open data sheet). However, this also does not guarantee working MSI/MSI-X. As you noted, FreeBSD lacks such hacks used in Linux so there could be cases that MSI/MSI-X does not work on FreeBSD even if Linux manages to make MSI/MSI-X work under certain conditions. The FreeBSD workaround you mentioned is not bad decision for a controller that is known to be broken in BIOS/chipset level. If MCP55 chipset was broken I would have it black-listed in PCI layer. For your case it's still not clear whether the issue came from broken BIOS or PCI bridges. Show me the output of "kenv | grep smbios". > just a workaround. > > > http://lists.freebsd.org/pipermail/svn-src-head/2010-November/022489.html? > > > Robert
For bugs matching the following criteria: Status: In Progress Changed: (is less than) 2014-06-01 Reset to default assignee and clear in-progress tags. Mail being skipped