Gargoyle Forum

Don't see any revision number anywhere to make this any more specific,
and I haven't tried doing anything with the USB functionality, but WLAN
bridging and routing are working fine with Gargoyle 1.3.10
gargoyle_1.3.10-brcm47xx-squashfs.trx

(In all honesty, i went and changed the bridge address and lan-side
DHCP config to a 10.x.x.x/24 network, as that's what i had on my prior
router and have all my IPsec SPDs configured up with, and i lost the
ability to connect to the https service on the gargoyle-d ASUS; http still
works fine. I see

Feb 21 21:00:42 Gargoyle daemon.crit httpd_gargoyle[3745]: socket :: - Address family not supported by protocol
Feb 21 21:00:42 Gargoyle daemon.err httpd_gargoyle[3745]: unknown sockaddr family on listen socket - 0

in the log upon restarting the httpd but don't know if that's significant.
I don't believe i've bollixed up hosts or ethers or protocols or anything,
but, yeah, it's probably just me, and i'm not too concerned.)

This is good stuff. Thanks

--buck, Maryland, USA

Linux version 2.6.32.27 (eric@Osiris) (gcc version 4.3.3 (GCC) ) #1 Mon Feb 7 17:44
:57 EST 2011
CPU revision is: 00029029 (Broadcom BCM3302)
ssb: Core 0 found: ChipCommon (cc 0x800, rev 0x14, vendor 0x4243)
ssb: Core 1 found: Fast Ethernet (cc 0x806, rev 0x09, vendor 0x4243)
ssb: Core 2 found: MIPS 3302 (cc 0x816, rev 0x08, vendor 0x4243)
ssb: Core 3 found: USB 2.0 Host (cc 0x819, rev 0x02, vendor 0x4243)
ssb: Core 4 found: MEMC SDRAM (cc 0x80F, rev 0x04, vendor 0x4243)
ssb: Core 5 found: IEEE 802.11 (cc 0x812, rev 0x0D, vendor 0x4243)
ssb: Core 6 found: Roboswitch (cc 0x81C, rev 0x02, vendor 0x4243)
ssb: chipcommon status is 0x0
ssb: Found rev 0 PMU (capabilities 0x04A63400)
ssb: Initializing MIPS core...
ssb: set_irq: core 0x0806, irq 4 => 4
ssb: set_irq: core 0x0816, irq 5 => 2
ssb: set_irq: core 0x0812, irq 2 => 5
ssb: after irq reconfiguration
ssb: core 0x0800, irq : 2(S) 3* 4 5 6 D I
ssb: core 0x0806, irq : 2(S) 3 4* 5 6 D I
ssb: core 0x0816, irq : 2(S)* 3 4 5 6 D I
ssb: core 0x0819, irq : 2(S) 3 4 5 6* D I
ssb: core 0x080f, irq : 2(S) 3 4 5 6 D I*
ssb: core 0x0812, irq : 2(S) 3 4 5* 6 D I
ssb: core 0x081c, irq : 2(S) 3 4 5 6 D I*
ssb: Sonics Silicon Backplane found at address 0x18000000

OK, slight problem

See

http://comments.gmane.org/gmane.linux.d ... evel/10674

This was causing the WLAN to croak intermittently (seemingly during
the morning lull when nobody at our house had an active client station
running, and machines couldn't reassociate later on in the day, but
there's no data to back that up. I also have this impression that every-
thing was hunky-dory for a few days until somebody started war-
driving the neighborhood when everybody was away at work or i
twiddled some config that didn't need it and made this trouble for my-
self I don't know)

Anyway, since i applied the patch mentioned in that mailing-list thread,
i get the ``DMA RX: SKB overrun'' messages (see below), but the b43
keeps going and i haven't had to reboot the router

I applied this to the 1.3.10 code (since that was what i was running
and didn't wanted to hold that constant) and i had to use the mtd
method to start running the patched .trx build; then reapplied my saved
config. If you want/need to try for yourself, you can apply the attached
to the 1.3.10 tagged revision of the gargoyle git source and run
./local.sh
or (if you need to redo from start)
./local.sh FULL_BUILD=1
and you'll end up with the .trx in the images subdirectory, if all goes
well. (On my debian machine, i needed to install libncurses5-dev
and zlib1g-dev and subversion to round out my build environment;
YMMV)

Thanks again to Eric B. for making this as easy as could be

Now if somebody can tell me why our stupid (wired) Brother printer is
no longer saying bonjour to my wife's Mac and seems to have dropped
off the net, . . . i'd like to go back to ignoring the computers and read
all the books i got from the local Border's liquidation

--buck, md, usa

b43-phy0: Loading firmware version 410.2160 (2007-05-26 15:32:10)
b43-phy0: Loading firmware version 410.2160 (2007-05-26 15:32:10)
device wlan0 entered promiscuous mode
br-lan: port 2(wlan0) entering forwarding state
device wlan0 left promiscuous mode
br-lan: port 2(wlan0) entering disabled state
device wlan0 entered promiscuous mode
br-lan: port 2(wlan0) entering forwarding state
b43-phy0 ERROR: DMA RX: SKB overflow (len=2348, frameoffset=30,skb free=2368
b43-phy0 ERROR: DMA RX buffer too small (len: 47905, buffer: 2352, nr-dropped: 21)
b43-phy0 ERROR: DMA RX buffer too small (len: 19294, buffer: 2352, nr-dropped: 9)

Mar 5 19:29:00 gargoyle user.err kernel: b43-phy0 ERROR: DMA RX buffer too small (len: 22869, buffer: 2352, nr-dropped: 10)
Mar 5 20:38:37 gargoyle user.err kernel: b43-phy0 ERROR: DMA RX: SKB overflow (len=2352, frameoffset=30,skb free=2368
Mar 5 20:38:37 gargoyle user.err kernel: b43-phy0 ERROR: DMA RX buffer too small (len: 52407, buffer: 2352, nr-dropped: 23)

OK. Wireless was still occasionally crapping out, but things have
been fine for about a week now after rebuilding based on git commit

c60a37905caa045e93df42e3efbcda12af93bc5a

with the attached diff. (It pulls a slightly more recent revision of
openwrt than the gargoyle git commit, and still includes the prior post's
overrun patch, which, yes, is, unfortunately, still needed)

--buck

I've just bumped openwrt to the latest backfire version (r26536) which has a major mac80211 update. The dma fix in your patch no longer applies cleanly, and it looks like there's a lot more error checking that has been added to the dma_rx function in the latest version.

Could you try building building based on commit 99c4c640ec48fb373db1 and see if the problem is fixed there? If not, would you consider updating your patch so that it applies cleanly to the latest version? It would be much appreciated.

By the way, if you want me to notice your patch sooner in the future, fork the gargoyle mirror on github (github.com/ericpaulbishop/gargoyle) commit your changes there and send a pull request. This is the best way to contribute.

i checked out 99c4c640ec48fb373db1656c75d0f99fe6ffee63, and it
svn-ed down 26536, which downloaded
compat-wireless-2011-04-06, but the b43/dma.c patch still applies
cleanly thereto, and, indeed, to the current

http://git.kernel.org/?p=linux/kernel/g ... .c;hb=HEAD

and i don't have the pain tolerance to build without the patch, flash
my router, and have to revert if it cra*s out

but i will if you say so