Install of 14.04 fails with 3.13.0-20-generic kernel

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1300416

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

The installation fails and apport-collect is unavailable in the installation environment. Therefore, I'm unable to run this command.

Roderick - does it make any difference if you use an ethernet cable during the install ? It appears you're having some issues with a Broadcom wireless driver.

No, an Ethernet cable doesn't seem to have any effect. My initial report was with Ethernet connected. I've since tried with it disconnected, and the symptoms look the same. I'm attaching /var/log/syslog from the cable-disconnected case, though, in case that gives you any more clues.

Roderick - a possible workaround might be to disable wifi from the BIOS, do the install, then restore wifi.

I've got a build which I think will fix this problem. I'm trying to figure out how to get an iso for you to test (the instructions I've found so far seem to be out of date), but in case you have some other way to test I went ahead and posted the build to:

http://people.canonical.com/~sforshee/lp1300416/linux-3.13.0-21.43+lp1300416/

This won't get you working wireless since the firmware will still be missing, but it shouldn't hang. I'll try to get an iso that you can try soon.

Tim, disabling the Wi-Fi radio in the firmware did not resolve the problem. (I didn't see any other options related to Wi-Fi.) Of course, that's moot if Seth's fix works, so I look forward to testing it.

I'm still not having any luck in generating an is which will boot successfully on more than one of my systems. Seems like it shouldn't be this hard ...

Anyway, maybe we'll try another avenue. When booting to the installer, if you add "brcmsmac.blacklist=1" to the kernel boot options it should prevent the module from loading at all. This should be enough to get you through the install. On UEFI systems I get a normal grub menu, so pressing 'e' on the install option and adding that text to the end of the "linux ..." line before booting does the trick.

After installation has finished you can verify that you still have the bug. If you have a /lib/firmware/brcm directory, temporarily rename it to something else, e.g. /lib/firmware/brcm-bak, then run 'sudo modprobe -r brcmsmac, sudo modprobe brcmsmac' as root (if you don't have /lib/firmware/brcm you don't need to do any of that). You should immediately see the "brcmsmac: fail to load firmware brcm/bcm43xx-0.fw" message appear in dmesg, and a couple of minutes later you should see the stack traces. After this you'll like have to hard power-off your system to reboot.

Now install the test build and reboot to that kernel. Try the same sequence of steps as above. You should still see the "fail to load firmware" message, but you shouldn't get the stack traces in dmesg. If you get these results then the problem should be gone in the installer as well.

Thanks!

I found out there was a patch pending upstream nearly the same as mine, so we went ahead and pulled it in.

This bug was fixed in the package linux - 3.13.0-23.45

---------------
linux (3.13.0-23.45) trusty; urgency=low

  [ Tim Gardner ]

  * Release Tracking Bug
    - LP: #1302083

  [ Emil Goode ]

  * SAUCE: (no-up) brcmsmac: fix deadlock on missing firmware
    - LP: #1300416

  [ Moni Shoua ]

  * SAUCE: (no-up) IB/core: Don't resolve passive side RoCE L2 address in cma req handler
    Merged at the request of Mellanox engineers. It shold be making its way upstream.
    http://www.spinics.net/lists/linux-rdma/msg19417.html

  [ Or Gerlitz ]

  * [Config] CONFIG_INFINIBAND_USNIC=m

  [ Tim Gardner ]

  * [Config] CONFIG_MLX4_EN_VXLAN=y

  [ Upstream Kernel Changes ]

  * This set of patches essentially updates the infiniband subsystem to 3.14 plus some
    bug fixes from 3.15. The initial set of commits were determined using:

    git log --pretty=oneline --reverse v3.13..cd6362befe4cc7bf589a5236d2a780af2d47bcc9 -- drivers/infiniband drivers/net/ethernet/mellanox include/linux/mlx* include/rdma

    A few tree-wide patches were dropped since they were not essential.
    This is the list of commits that were actually applied. As you can see there were
    a few back ports, but all were simple context collisions and easily resolved.

    (cherry picked from commit 7c6d74d23a33a946bcf08ba2d3e52d31943b7342)
    (cherry picked from commit 7b25d81b7ff03bb1893a9c7f97797b891a772deb)
    (cherry picked from commit 73e74ab4e0d030f28d640507998156a22d4211f8)
    (cherry picked from commit eb17711bc1d6611e934af5b6dabc225936084128)
    (cherry picked from commit 8e1a28e8e6797449dfdfa4739002d1e5939355a8)
    (cherry picked from commit 84c864038d6d991be81344fc3168ec2c2f7a8d06)
    (cherry picked from commit d03a68f8217ea03492e4f7928db222dc6544792a)
    (cherry picked from commit be902ab122fcc59ba6a8588e820c31861d75a5a4)
    (cherry picked from commit e4b59a1cb6f8feb03f356b0abfd20451f05d7470)
    (cherry picked from commit 0276a330617a0cf380f09e5065299078d3d45886)
    (cherry picked from commit 982290a7fe36e528af292d3e3b61939b1900bfc6)
    (cherry picked from commit c5266d40b0a26546d0ebedb44dd4145088b85cb8)
    (cherry picked from commit 7ffdf726cfe0d188907bdbb0e7729fb35a69c219)
    (cherry picked from commit 837052d0ccc5a789a578f8b628ba154b63bd51ea)
    (cherry picked from commit c0623e587d869b4b18e077d64a8524ea364b5b77)
    (cherry picked from commit ad7d4eaed995d76fb24a18e202fdf5072197ff0a)
    (cherry picked from commit 2156d9a8ac0202f0158d407063cb850afffd3f56)
    (cherry picked from commit b912b2f8fc71df4c3ffa7a9fe2c2227e8bcdaa07)
    (cherry picked from commit 9ba75fb0c4b92416b94640b5a043c323a457f14a)
    (cherry picked from commit 74b9c3ea847f060c784e86453f1ad77dd05a7a8f)
    (cherry picked from commit e6a767582942d6fd9da0ddea673f5a7017a365c7)
    (cherry picked from commit fe5e8a1acc7fd877b6706053cf88c418c33fe7a3)
    (cherry picked from commit be8348df6efac6b602f2ad3210139bccf0dbe3d7)
    (cherry picked from commit e3cf00d0a87f025db5855a43a67c67a41fa79fef)
    (cherry picked from commit 8192d4acb5c5376c0f6756f2106ab243036c8c7d)
    (cherry picked from commit 301a0dd68e5ddd22d992a58f466b621987d9df3b)
    (cherry picked from commit 2183b990b67b761f81c...

I've verified that the 3.13-23-gneric kernel does indeed fix the problem on the systems I'd been using.