> > I'm not sure how the issue got fixed with the non-android devices because > they were doing the same thing on the 2.4ghz radio before a recent patch > but I was hoping that this could have a similar fix. > > > If I may add a tidbit (and just as camden, I'm no expert either)...
That being said, I have a few android devices myself and while most of them play nice, some of them are notorious for indicating a problem where none exists (at least not with OpenWrt). However, given that trunk is always changing, it might be a combination of OpenWrt changes and a badly written kernel WiFi driver for the android device in question. For example, I own a Google Nexus 4 and over the course of time, starting from the 10th installment of android (Jelly Bean) through to the 11th (Kit-Kat), they broke the 5 GHz WiFi on the 3-digit channels (149+). Not necessarily Google's fault (but still Google's fault technically), because the related WiFi code for the qualcomm msm itself if borked in that stock kernel. They did fix the problem with delayed notification where any notification after the screen went to sleep was delayed until the device was manually awakened (caused by bad ARP translation offloading code). However, they introduced horrible throughput on the 3-digit wifi channels (which is what I use in my house). What am I talking about? The Phone's 3-digit 5GHz wifi on the stock kernel doesn't reach a download throughput anywhere higher than 12mbps (even though it connects near at 65mbps). It definitely worked with stock 4.2.2 but not with 4.4.2. All my other Android devices are able saturate my WAN connection throughput during speed tests, just not the nexus 4. How do I know this? After trying out several other ROMs i figured out the CyanogenMod (CM) uses a different code base (Code Aurora) and has a much more functional WiFi at 5GHz than the one used by Google (Android Open Source Project a.k.a AOSP). Problem is that qualcomm is done committing any changes to the kernel used by AOSP (as far as I know), and their most recent and functional WiFi code is available in Code Aurora (go figure why...). To make matters worse, this isn't the case for all the devices. My Dad's Nexus 5 doesn't have any issues with the stock AOSP/Google ROM. So the bottom line is, android devices' WiFi performance is like a "baux auf chaucklets" (read: box of chocolates), and you may be trying to look for a solution in the wrong place. That being said, I'm sure debug mode/TCPdump may be able to help you narrow down the issue. All I want to point out is that it "might" turn out to be a dead-end (not saying it "is"). What I want to know is whether this problem exists with the router running STOCK firmware from Western Digital. I have a Western Digital MyNet N750 (similar hardware maybe), and I have not had this issue. And even if such an issue occurred, I'm more likely to blame Android's WiFi driver and not OpenWrt, but hey that's just me. Also, do you have any other OpenWrt compatible router where this problem doesn't occur? Thanks, Chirag
_______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
