Joined: 08 Oct 2006
Location: Chicago, IL
|Posted: Thu Jun 11, 2015 12:30 am Post subject: Acer AO722 & Broadcom BCM4313
I have an Acer Aspire One 722 with a Broadcom (YUCK!) 4313 wireless chip.
1) I'm having the hardest time repairing what was once a working Broadcom 4313 chip (PCI ID 14e4:4727) after some kernel upgrade and emerge -avtNDu @world updates. I used an older kernel version (3.16.5-gentoo) to update @world because that's the only kernel I could get wireless to work from, which I hope wasn't a mistake because everything else seems fine, which I will expound upon in a bit. If I boot back into kernel 3.16.5-gentoo the wireless works as it did before. I'd get kernel panics that seemed associated with cfg80211, but it was tolerable considering past "fixes" for this chip whose support is horrible.
However, no matter what combination of solutions I try (there are a lot of them out there), I cannot get it working again in kernel 3.18.12-gentoo. The fact that it will still work in kernel 3.16.5-gentoo leads me to believe it may simply be a kernel configuration problem with my 3.18.12-gentoo version, however since I updated @world from the old kernel I'm not too sure if that's "doing it wrong".
The 3.16.5-gentoo kernel has some option to include IPW2100=m (Intel Pro-Wireless 2100), I think it was so that other things like LIB80211=m, FW_LOADER=y, and WIRELESS_EXT=y get pulled in as well, but my memory fails me as to why I did that.
2) Another thing that I think may be of importance is the fact that the broadcom-sta package fails during an emerge. However I can't remember whether I even needed to emerge broadcom-sta to get it working; it's not installed now yet works in 3.16.5-gentoo kernel.
3) Lastly, here is some dmesg output that seems significant to me when I boot from the 3.16.5-gentoo kernel:
|[ 11.467942] wl: module license 'Mixed/Proprietary' taints kernel.
[ 11.467953] Disabling lock debugging due to kernel taint
[ 11.532634] wlan0: Broadcom BCM4727 802.11 Hybrid Wireless Controller 220.127.116.11 (r487574)
Please bear in mind that I've tried most of the plethora of combinations of solutions and fixes found through a simple Google search, and I've gotten this card working before in more ways than one, to varying degrees of success. However this time it's being particularly difficult.
I can pastebin my kernel .config from 3.16.5-gentoo and 3.18.12-gentoo if that would help. I've done my best using diff on the .config's for 3.16.5-gentoo and 3.18.12-gentoo to make sure they're identical.
The lspci -n for the Broadcom chip:
|07:00.0 0280: 14e4:4727 (rev 01) |
Although I've attempted the solutions that involve b43-fwcutter, it doesn't seem that matters because the 3.16.5-gentoo .config isn't loading any firmware at all so I think I abandoned that avenue.
Using the in-kernel b43 module seems to yield a bit different results according to dmesg, as it begins to ask for some allhwsupport=1 option, similar to this:
|[ 9.625943] Support for cores revisions 0x17 and 0x18 disabled by module param allhwsupport=0. Try b43.allhwsupport=1 |
Unfortunately playing around with this didn't get me very far as I couldn't, for the life of me, figure out how to pass the option allhwsupport=1 from modprobe.
Any help would be GREATLY appreciated.