Message ID | 1418221700-45052-1-git-send-email-seth.forshee@canonical.com |
---|---|
State | New |
Headers | show |
Seth, You let me know you tested it via IRC using a dummy firmware driver. Seems like reasonable to do in Vivid. --chris On 12/10/2014 08:28 AM, Seth Forshee wrote: > udev no longer supports firmware loading by default and ignores > the uevent, and the kernel blocks until loading times out 60 > seconds later. For some specific Intel wireless hardware revs > this happens twice as the driver tries to load two firmware > versions which don't exist for those cards. > > Since udev is out of the firmware loading game it no longer > makes sense to have the userspace fallback enabled at all. It's > still possible for a driver to explicitly request a userspace > helper, as is done by dell_rbu. > > BugLink: http://bugs.launchpad.net/bugs/1398458 > Signed-off-by: Seth Forshee <seth.forshee@canonical.com> > --- > debian.master/config/config.common.ubuntu | 2 +- > debian.master/config/enforce | 6 ++++++ > 2 files changed, 7 insertions(+), 1 deletion(-) > > diff --git a/debian.master/config/config.common.ubuntu b/debian.master/config/config.common.ubuntu > index 0f4fd6e..ddb92a7 100644 > --- a/debian.master/config/config.common.ubuntu > +++ b/debian.master/config/config.common.ubuntu > @@ -2369,7 +2369,7 @@ CONFIG_FWTTY_MAX_CARD_PORTS=32 > CONFIG_FWTTY_MAX_TOTAL_PORTS=64 > CONFIG_FW_LOADER=y > CONFIG_FW_LOADER_USER_HELPER=y > -CONFIG_FW_LOADER_USER_HELPER_FALLBACK=y > +# CONFIG_FW_LOADER_USER_HELPER_FALLBACK is not set > CONFIG_GACT_PROB=y > CONFIG_GADGET_UAC1=y > CONFIG_GAMEPORT=m > diff --git a/debian.master/config/enforce b/debian.master/config/enforce > index 70929ad..bcff5c8 100644 > --- a/debian.master/config/enforce > +++ b/debian.master/config/enforce > @@ -140,3 +140,9 @@ value CONFIG_I2C_DESIGNWARE_PLATFORM m | \ > > # Required if /init is a shell script. > value CONFIG_BINFMT_SCRIPT y > + > +# Newer udevs don't handle firmware loading, and having the userspace > +# fallback enabled in the kernel just results in big delays if we do > +# fall back. > +# See LP:1398458 > +value CONFIG_FW_LOADER_USER_HELPER_FALLBACK n >
diff --git a/debian.master/config/config.common.ubuntu b/debian.master/config/config.common.ubuntu index 0f4fd6e..ddb92a7 100644 --- a/debian.master/config/config.common.ubuntu +++ b/debian.master/config/config.common.ubuntu @@ -2369,7 +2369,7 @@ CONFIG_FWTTY_MAX_CARD_PORTS=32 CONFIG_FWTTY_MAX_TOTAL_PORTS=64 CONFIG_FW_LOADER=y CONFIG_FW_LOADER_USER_HELPER=y -CONFIG_FW_LOADER_USER_HELPER_FALLBACK=y +# CONFIG_FW_LOADER_USER_HELPER_FALLBACK is not set CONFIG_GACT_PROB=y CONFIG_GADGET_UAC1=y CONFIG_GAMEPORT=m diff --git a/debian.master/config/enforce b/debian.master/config/enforce index 70929ad..bcff5c8 100644 --- a/debian.master/config/enforce +++ b/debian.master/config/enforce @@ -140,3 +140,9 @@ value CONFIG_I2C_DESIGNWARE_PLATFORM m | \ # Required if /init is a shell script. value CONFIG_BINFMT_SCRIPT y + +# Newer udevs don't handle firmware loading, and having the userspace +# fallback enabled in the kernel just results in big delays if we do +# fall back. +# See LP:1398458 +value CONFIG_FW_LOADER_USER_HELPER_FALLBACK n
udev no longer supports firmware loading by default and ignores the uevent, and the kernel blocks until loading times out 60 seconds later. For some specific Intel wireless hardware revs this happens twice as the driver tries to load two firmware versions which don't exist for those cards. Since udev is out of the firmware loading game it no longer makes sense to have the userspace fallback enabled at all. It's still possible for a driver to explicitly request a userspace helper, as is done by dell_rbu. BugLink: http://bugs.launchpad.net/bugs/1398458 Signed-off-by: Seth Forshee <seth.forshee@canonical.com> --- debian.master/config/config.common.ubuntu | 2 +- debian.master/config/enforce | 6 ++++++ 2 files changed, 7 insertions(+), 1 deletion(-)