The existing rt2x00-9999.ebuild became useless some months back when the rt2x00 team started requiring large and complex git downloads rather than making recent code available in CVS. Since the availability of kernel 2.6.22, rt2x00 is again available via CVS, and this is now the best way to install the driver. This ebuild is a minor update to support recent changes in the driver and in the kernel. It requires kernel 2.6.22 or above (CVS rt2x00 is not available for older kernels). See: http://rt2x00.serialmonkey.com/wiki/index.php/Downloads Reproducible: Always
Created attachment 124773 [details] rt2x00-9999.ebuild
*** Bug 185301 has been marked as a duplicate of this bug. ***
Linux kernel at least since 2.6.23-rc2 has got "CONFIG_EEPROM_93CX6" option. Maybe the ebuild has to be modified to use kernel eeprom_93cx6.ko module rather than compiling it once more with rt2x00?
Another thing that I've recently noticed is that ebuild is not going to install well without FEATURES="-sandbox" option. With FEATURES="-sandbox" emerge rt2x00 it installs perfect. Ebuild post-install message says that I shouldn't fill bugs in Gentoo bugzilla, but this is not rt2x00 module bug, but something concerning CVS ebuild. I'll attach my emerge log with error message.
Created attachment 128031 [details] emerge rt2x00 log with FEATURES="sandbox"
Comment on attachment 128031 [details] emerge rt2x00 log with FEATURES="sandbox" Unrelated, see Bug 135745
a diff between the ebuilds would be nice
Created attachment 128476 [details] emerge -v =net-wireless/rt2x00-9999 log (ebuild from overlay) This is a log from my attempt to compile this driver. It seems like an upstream problem to me.
latest CVS has a bug Add these lines between done and linux-mod_src_compile sed -i -e 's/ rt2x00dev->interface.id,//g' rt2x00mac.c # SO quick and dirty... echo '#define IEEE80211_TXCTL_LONG_RETRY_LIMIT (1<<10)' >> rt2x00.h
Created attachment 130259 [details] rt2x00-9999.ebuild I couldn't get this to build and fixed the ebuild for my rt61 card. It is obviously not a quality ebuild but I figured it does not make much sense with upstream changing a lot. I hope they deliver a release soon.
I can't ermerge this on kernel 2.6.22-gentoo-r5 I did the echo '#define IEEE80211_TXCTL_LONG_RETRY_LIMIT (1<<10)' >> rt2x00.h but now I get /var/tmp/portage/net-wireless/rt2x00-9999/work/rt2x00-9999/rt2x00mac.c:62: error: too many arguments to function and 'ieee80211_ctstoself_get'/var/tmp/portage/net-wireless/rt2x00-9999/work/rt2x00-9999/rt2x00mac.c:66: error: too many arguments to function 'ieee80211_rts_get' I understand that this is due to incompatible api changes in the mm kernel (found http://lkml.org/lkml/fancy/2007/8/22/362)
Line echo "CONFIG_${M}_BUTTON=${button}" >> config Should really be echo "CONFIG_${M}_RFKILL=${button}" >> config Example config file says it should be 'BUTTON' however Makefile works with 'RFKILL' variables.
Recent changes in rt2x00 CVS driver add new compilation problems: http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4294 It seems that we are stuck with old CVS revisions or we need to patch our kernel's mac80211 stack.
(In reply to comment #13) > Recent changes in rt2x00 CVS driver add new compilation problems: > http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4294 > > It seems that we are stuck with old CVS revisions or we need to patch our > kernel's mac80211 stack. > rt2x00 will be integrated in the next kernel. So maybe you want to try out some of the vanilla rc kernel (I've experienced, that they are quite stable after rc4 or rc5.). Plus, the more the integrated driver will be tested, the more stable it might be, when the kernel is finally released. Or just wait till in 10-12 weeks the new kernel hits the tree.
Thanks, I'll wait for 2.6.24-rc1. A good news that rt2x00 will be in vanilla-sources. (In reply to comment #14) > (In reply to comment #13) > > Recent changes in rt2x00 CVS driver add new compilation problems: > > http://rt2x00.serialmonkey.com/phpBB2/viewtopic.php?t=4294 > > > > It seems that we are stuck with old CVS revisions or we need to patch our > > kernel's mac80211 stack. > > > rt2x00 will be integrated in the next kernel. So maybe you want to try out some > of the vanilla rc kernel (I've experienced, that they are quite stable after > rc4 > or rc5.). Plus, the more the integrated driver will be tested, the more stable > it might be, when the kernel is finally released. > > Or just wait till in 10-12 weeks the new kernel hits the tree. >
By the way, is the bug still actual? rt2x00 driver is in kernel 2.6.24 since -rc1 and rt2x00 maintainers do not care about driver compilation out of vanilla kernel or wireless-dev sources. I mean, if you use 2.6.23 for example, you possibly won't compile recent rt2x00 revision and devs won't help you.
WONTFIX.
*** Bug 201964 has been marked as a duplicate of this bug. ***
