iwi(4) has recently changed its behavior with regard to firmware loading, now doing so automatically when it's placed in /boot/firmware. The iwi-firmware port however places firmware unconditionally in /usr/local/share/iwi-firmware. It should be made aware of /boot/firmware so reinstalling the port resolves the firmware open errors that result from upgrading. Fix: Install contents of /usr/local/share/iwi-firmware/ to /boot/firmware depending on OS version, presumably also conditionally installing the rc script now the driver can load firmware on its own. Documenting /boot/firmware in the iwi manpage or so would also help :)
Responsible Changed From-To: freebsd-ports-bugs->flz Over to maintainer
Can you try the following patch ? http://people.freebsd.org/~flz/local/ipw.diff -- Florent Thoumie flz@FreeBSD.org FreeBSD committer
State Changed From-To: open->closed I've committed a fix, should be ok now. Thanks for your reminder.
* Florent Thoumie (flz@FreeBSD.org) wrote: > I've committed a fix, should be ok now. Thanks for your reminder. Thanks, not quite perfect though; I needed to pass -DWITHOUT_CONTROL just to get to the OPTIONS menu, then I had to de-select it again there. Probably best to swap the logic so it only does IGNORE if WITH_CONTROL is defined explicitly, or downgrade it to a warning, ala: --- ipw-firmware/Makefile.orig Mon Dec 12 06:34:45 2005 +++ ipw-firmware/Makefile Mon Dec 12 06:35:12 2005 @@ -67,7 +67,7 @@ .if ${OSVERSION} > ${MIN7OSVERSION} SUB_LIST+= DONT_NEED_CONTROL="@comment " -. if !defined(WITHOUT_CONTROL) +. if defined(WITH_CONTROL) IGNORE= is configured with ${DRIVERNAME}control(8) which you don't need . end Also I've heard of a patch that makes iwicontrol work with the new iwi changes, so maybe it's best to just humor the user if they want it. -- Thomas 'Freaky' Hurst http://hur.st/
Le Lundi 12 décembre 2005 à 06:39 +0000, Thomas Hurst a écrit : > * Florent Thoumie (flz@FreeBSD.org) wrote: > > > I've committed a fix, should be ok now. Thanks for your reminder. > > Thanks, not quite perfect though; I needed to pass -DWITHOUT_CONTROL > just to get to the OPTIONS menu, then I had to de-select it again there. That's the point, make config does it for you. > Probably best to swap the logic so it only does IGNORE if WITH_CONTROL > is defined explicitly, or downgrade it to a warning, ala: Nope, that's how OPTIONS work. If default is "on", check for WITHOUT_${foo}. > --- ipw-firmware/Makefile.orig Mon Dec 12 06:34:45 2005 > +++ ipw-firmware/Makefile Mon Dec 12 06:35:12 2005 > @@ -67,7 +67,7 @@ > > .if ${OSVERSION} > ${MIN7OSVERSION} > SUB_LIST+= DONT_NEED_CONTROL="@comment " > -. if !defined(WITHOUT_CONTROL) > +. if defined(WITH_CONTROL) > IGNORE= is configured with ${DRIVERNAME}control(8) which you don't need > . end IGNORE is fine, I'll keep it that way. I'll probably default CONTROL to off later, once latest changes have been MFC'ed and some months have passed. > Also I've heard of a patch that makes iwicontrol work with the new > iwi changes, so maybe it's best to just humor the user if they want it. What's the point ? iwicontrol was just a workaround to put the firmware in cache, now the driver does it automatically. -- Florent Thoumie flz@FreeBSD.org FreeBSD committer
* Florent Thoumie (flz@FreeBSD.org) wrote: > Le Lundi 12 décembre 2005 à 06:39 +0000, Thomas Hurst a écrit : > > * Florent Thoumie (flz@FreeBSD.org) wrote: > > > > > I've committed a fix, should be ok now. Thanks for your reminder. > > > > Thanks, not quite perfect though; I needed to pass -DWITHOUT_CONTROL > > just to get to the OPTIONS menu, then I had to de-select it again there. > > That's the point, make config does it for you. Right, but this means that a "portupgrade iwi-firmware", or just a plain make results in the IGNORE; you need to explicitly disable it or call config, even if there's no previously saved configuration, which isn't behavior I've encountered in a port before, and which doesn't seem especially intuitive, this being -CURRENT aside ;) > > Also I've heard of a patch that makes iwicontrol work with the new > > iwi changes, so maybe it's best to just humor the user if they want it. > > What's the point ? iwicontrol was just a workaround to put the > firmware in cache, now the driver does it automatically. iwicontrol also shows a bunch of stats about the device; number of frames sent at the various supported speeds, number of retransmits, failures etc. Are these reproduced elsewhere too? -- Thomas 'Freaky' Hurst http://hur.st/