Date: Tue, 20 Jan 2015 20:47:17 -0800 From: Adrian Chadd <adrian@freebsd.org> To: Alexey Dokuchaev <danfe@nsu.ru> Cc: "freebsd-wireless@freebsd.org" <wireless@freebsd.org> Subject: Re: Dual-band AR5414 card test-run on stable/8 and head Message-ID: <CAJ-VmomkzN39Vpugo2oNNiawUvfW9aSsmRyBdQ1X%2Bg6fhYH72Q@mail.gmail.com> In-Reply-To: <20150121040308.GA49520@regency.nsu.ru> References: <20150121040308.GA49520@regency.nsu.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
On 20 January 2015 at 20:03, Alexey Dokuchaev <danfe@nsu.ru> wrote: > Hi there, > > Recently I've purchased few assorted ARxxxx-based wireless cards. Original > plan was to replace 2200BG Intel of mine (it's not really stable and driver > does not support frame injection), but this goal had failed: my new awesome > (and quite expensive) industry-grade 9220-based dual-band card was too tall > for my laptop's miniPCI bay and had MMCX antenna plugs rather than U.FL. :( > > So I was only left to play with AR5414 2.4GHz/5GHz (per specs from the shop) > card; looks like it is Askey WLL4070-D50. > > After doing "ifconfig ath0" (on stable/8, this loads ath.ko): > > ath0: <Atheros 5413> mem 0xb0110000-0xb011ffff irq 17 at device 8.0 on pci6 > ath0: [ITHREAD] > ath0: AR5413 mac 10.5 RF5413 phy 6.1 > > .. and "kldload ath_pci" on head (r277422, ifconfig(8) alone does not DTRT): > > ath0: <Atheros 5413> mem 0xb0110000-0xb011ffff irq 17 at device 8.0 on pci6 > ath0: AR5413 mac 10.5 RF5413 phy 6.1 > ath0: 2GHz radio: 0x0000; 5GHz radio: 0x0063 > > Per `pciconf -lv`, it's class=0x020000 card=0x132910cf chip=0x001b168c > rev=0x01 hdr=0x00 network ethernet, by Atheros Communications Inc. Device > string differs between stable/8 (AR5006 family 802.11abg Wireless NIC) and > head (AR5413/AR5414 Wireless Network Adapter [AR5006X(S) 802.11abg]). > > Now, here's my quest. I first tried to use it under stable/8 (which is my > every-day working system). Apparently it could connect to the access point > and obtained DHCP lease, but I could not ping even the gateway (and thus > anything outside). I've also noticed two things: No idea why it doesn't work on stable/8. That's odd. It should work just fine. > 1). WiFi LED is always on, immediately once I press the power button (before > FreeBSD gets to boot). With 2200BG, it was actually correctly reflecting > network activity. Any chance it can be fixed to work with Atheros card(s)? > This applies to both stable/8 and head. The LED goes off only when laptop > is switched off or put to suspend (S3). It's likely hooked up to a GPIO pin that's hooked up to the mini-PCI slot. You can check the, say, 6 GPIOs to see if any do anything with the LED: sysctl dev.ath.0.ledpin=<x> (0, 1, 2, 3 .. 5) sysctl dev.ath.0.ledon=<1 | 0 - ie, the LED polarity> sysctl dev.ath.0.softled=1 Toggle softled 0 -> 1 each time you change ledpin, just to make sure the programming takes. > 2). I have "WiFi radio on/off" button on the keyboard (Fn-F2) which seems > to work with both cards, Intel and Askey. There is a difference, however: > with iwi(4), there are "radio turned off/on" messages in dmesg, but nothing > alike with ath(4). Since LED stays always on, it is impossible to easily > tell at which state is the card at any given moment. It's also likely a GPIO pin. rfkill is something I haven't played with all that much and I don't know if we have the ath(4) rfkill stuff plumbed into the userland in some useful fashion for the driver to tell you what's going on. > On -CURRENT, modulo that I had to "kldload ath_pci" instead of just saying > "ifconfig ath0", it works quite well. I could connect to the office WiFi > (at 2.4GHz, probably because the access point doesn't support or offer 5GHz > radio), and finally I could ssh wirelessly and has the same responsiveness > as on copper (previously, with 2200BG, it was pretty laggy, bad enough for > typing anything become a real PITA). Too bad I cannot stay on -CURRENT for > too long due to broken suspend/resume cycle. What's broken with suspend/resume? That should be fixed. > TL;DR: the card works on head, but not on stable/8; any revisions I should > take a look at, test, and ask for MFC if all goes well? What about always- > on LED, can it be controlled? Does ath(4) expects anything from e.g. ACPI > to be able to do so? I've no idea about what or why it broke on stable/8 but isn't on -HEAD. I've done some prettty significant surgery to ath and net80211 since stable/8. You can try narrowing it down by trying stable/9 and stable/10 to see if it started working during one of those releases. That'll narrow down the revisions to bisect. > I will probably continue to buy more of ath(4) cards to see how well they > behave. :-) Woo! -a
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-VmomkzN39Vpugo2oNNiawUvfW9aSsmRyBdQ1X%2Bg6fhYH72Q>