Reported by Nikolaus Schaller, Oct 16, 2013
there has been a report that WiFi speed is low. This appears to be related to SDIO interrupt handling and was apparently improved in the 2.6.32 kernel by this patch: http://git.goldelico.com/?p=gta04-kernel.git;a=commit;h=f8c08bcf62b09 c928217f56dc599c628
Comment 1 by Nikolaus Schaller, Oct 19, 2013
There was also a report/patch for 3.5-rc2: http://lists.goldelico.com/pipermail/gta04-owner/2012-June/002483.htm l I was under the impression that it was included in 3.7 and therefore in our 3.12-rc5.
Comment 2 by Nikolaus Schaller, Dec 15, 2013
What is the influence (if at all) of the different WiFi firmware variants we can install through Debian nonfree? http://packages.debian.org/wheezy/firmware-libertas There are different variants: -rw-r--r-- 1 root root 122800 Jan 6 2013 /lib/firmware/libertas/sd8686_v8.bin -rw-r--r-- 1 root root 2464 Jan 6 2013 /lib/firmware/libertas/sd8686_v8_helper.bin -rw-r--r-- 1 root root 124188 Jan 6 2013 /lib/firmware/libertas/sd8686_v9.bin -rw-r--r-- 1 root root 2516 Jan 6 2013 /lib/firmware/libertas/sd8686_v9_helper.bin -rw-r--r-- 1 root root 122916 Feb 29 2008 /lib/firmware/sd8686.bin -rw-r--r-- 1 root root 2516 May 12 2009 /lib/firmware/sd8686_helper.bin Or maybe we are patching an old firmware over the latest Debian through http://git.goldelico.com/?p=gta04-rootfs.git;a=tree;f=debian/config/l ib/firmware;hb=HEAD
Comment 3 by Nikolaus Schaller, Dec 15, 2013
After reinstaling firmware-libertas we have root@gta04:~# ls -l /lib/firmware/sd8686* lrwxrwxrwx 1 root root 22 Jan 6 2013 /lib/firmware/sd8686.bin -> libertas/sd8686_v8.bin lrwxrwxrwx 1 root root 29 Jan 6 2013 /lib/firmware/sd8686_helper.bin -> libertas/sd8686_v8_helper.bin I.e. we indeed have a newer version than the patch. And it is v8 but still not yet v9.
Comment 4 by Lukas M, Jun 28, 2014
In the 3.12-replicant kernel, the following commit improved wifi speed by a factor of 4 (according to PaulK): http://git.goldelico.com/?p=gta04-kernel.git;a=commit;h=a32da16bc04ae d3271794f6356e3102a3377a979
Comment 5 by Nikolaus Schaller, Jun 28, 2014
Looks nice if he has made it 3.12 compatible. A problem might be that none of us understands it and can upstream it :( And we have to check if 3.16 and DT did not do something similar (SDIO with interrupts). I remember that there is some DT option to define if the SDIO interface should support IRQ or not. But I am not sure.
Comment 6 by Nikolaus Schaller, Jan 8, 2017
Is this still an issue?
Comment 7 by Andreas Kemnade, Jan 8, 2017
the mentioned stuff is in our feature branches, so it is solved in our tree, but cleaning it up for upstream still needs to be done