I noticed the message from mbm that said experimental was now in CVS.  I built TESTED from CVS last night and flashed the squash fs into a WRT54GS (a v1 I guess, the sticker on the bottom doesn't have anything after WRT54GS).

When I put power to it here is what happens:

LAN 1-4 flash a few times while the power LED flashes
After a while the DMZ light will come (when it does the power LED goes solid) on and then go off
Next, it appears to reset as the 1-4 LEDs flash again
Power, DMZ both stay on for a while
DMZ will finally go off and then the WLAN LED will come on.

This acts way different than the last time I loaded experimental on a v1.1 GS (I have a version 1, 1.1 and a 2 GS).

Is it OK to put the experimental code in a V1 GS?

The next thing I noticed is running passwd doesn't work.  No matter what I try, it says "An error occurred updating the password file.

Can anyone school me on what is up or what I should try to fix it?  I'm an embedded software engineer but I'm new to the MIPS and openwrt.  I'm a little rusty using CVS too ... too many years using ClearCase.  Been using Debian since 1994.

This looks like good stuff!  I hope I can get educated and become usefull around here.

Regards,

Brian

Update to my last post regarding experimental built from CVS running on a v1 WRT54GS.

I was checking out dmesg and I noticed a line that says rootfs doesn't start on an erase block boundary -- force read-only.  Could this be why the passwd command isn't working?

I'll have to check out what an older experimental build dmesg output looks like on one of my v1.1 GS boxes.

Thanks in advance for info. to point me in the right direction.

Hello. I'm experimenting a couple of WR850G (v1 & v3) with openwrt-motorola-squashfs and openwrt-motorola-jffs2 and faced some trouble with experimental releases. I have flashed WRT54Gs before without problems

I renamed openwrt-motorola-squashfs.bin to trx and uploaded it through web firmware page (on R850G v1). Missed the boot_wait off on next reboot warning and the unit bricked (problably by the same cause of the second unit, plz read below)

decided to try the v3. this time enabled boot_wait and rebooted/tried to flash via tftp. the unit accepted it, answer pings but I cannot telnet to it. as boot_wait is on I can reflash with the other firmware file, but the behaviour is the same. boots up, I can ping it (192.168.10.1) but no telnet

as to debug this blindly is very difficult, I'd appreciate any hints on experimental with WR850G. I'll try to debrick the v1 as soon as I can find the 3.3v pin on the 9 pin header -- in, out & gnd are available on openwrt docs. I have a level shifter!!! :-)

the motorola default firmware turning boot_wait off on every reboot is a real pain. hopefully I'll have openwrt running on these beasts shortly to get rid of motorola's weird firmware.

Cheers
!3runo
from Brazil

Hi there

just to let you know, my problem was timing. it seems to took a long time to firstboot complete its run on WR850G (longer than on WRT54G). After some time the units simply start working! I was able to install openwrt experimental on both units, using

- openwrt-motorola-squashfs.bin on v1, renaming the file to trx and using motorola web interface to upgrade

- openwrt-generic-jffs2-4MB.trx and openwrt-generic-squashfs.trx on v3, by boot_wait and tftping the files to the unit

I still could not find a 3.3v source on the unit to be able to use a level shifter. Let's see about that.

Cheers
!3runo
from Brazil

P.S. Wish to thanks frop from irc.freenode.net #wrt54g for his online hints on this install.

you need to reboot after the flashing. then you can relogin and set a password for ssh.
after the next reboot you only can login via ssh.

On an asus wl500gx, replace eth0 with vlan1 for:

nvram set wan_ifnames=vlan1
nvram set wan_ifname=vlan1

Otherwise the WAN port doesn't work.

I was trying to 'make' the experimental release, however the build did not finish. The error is embedded below. May I know if there is a patch available to overcome this problem?

Regards,
L.

make[2]: Entering directory `/samples2/openwrt/target/utils'
make[2]: Nothing to be done for `prepare'.
make[2]: Leaving directory `/samples2/openwrt/target/utils'
make -C utils compile
make[2]: Entering directory `/samples2/openwrt/target/utils'
mkdir -p /samples2/openwrt/staging_dir_mipsel/bin
cp -a /samples2/openwrt/build_mipsel/target-utils/* /samples2/openwrt/staging_dir_mipsel/bin/
make[2]: Leaving directory `/samples2/openwrt/target/utils'
make[1]: Leaving directory `/samples2/openwrt/target'
make -C package compile
make[1]: Entering directory `/samples2/openwrt/package'
make[2]: Entering directory `/samples2/openwrt/package/openwrt'

make[3]: Entering directory `/samples2/openwrt/package/openwrt/libnvram'
make[3]: *** No rule to make target `/samples2/openwrt/build_mipsel/libnvram/nvram_linux.o', needed by `/samples2/openwrt/build_mipsel/libnvram/libnvram.so'.  Stop.
make
[3]: Leaving directory `/samples2/openwrt/package/openwrt/libnvram'

make[2]: *** [libnvram-compile] Error 2

make[2]: Leaving directory `/samples2/openwrt/package/openwrt'
make[1]: *** [openwrt-compile] Error 2

make[1]: Leaving directory `/samples2/openwrt/package'
make: *** [package_compile] Error 2

[root@ie10dtlx2458 openwrt]# ls

That's the same problems as this one.  Try GNU make 3.80.

Same problem here, is there any explanation why?

Thanks. Updating the version of make fixed the problem.

Regards,
L.

Loading ip_nat_proto_gre crashed the linksys.

root@OpenWrt:/lib/modules/2.4.30# insmod ip_nat_proto_gre
Using /lib/modules/2.4.30/ip_nat_proto_gre.o

A few seconds later, the linksys reboots.

Specs:
Linksys WRT54G v1.1
Linux OpenWrt 2.4.30 #1 Sat Apr 23 18:13:56 EDT 2005 mips unknown
root@OpenWrt:~# dmesg
CPU revision is: 00024000
Primary instruction cache 8kB, physically tagged, 2-way, linesize 16 bytes.
Primary data cache 4kB, 2-way, linesize 16 bytes.
Linux version 2.4.30 (root@reboot2.lan) (gcc version 3.3.5) #1 Sat Apr 23 18:13:56 EDT 2005

I installed the latest experimental build (dated: 23-apr-2005 )

Wouter

Another one:

root@OpenWrt:~# ipkg update
Downloading http://openwrt.org/downloads/experiment … s/Packages ...
Connecting to openwrt.org[208.185.247.160]:80
Packages             100% |**********************************************************************************************************| 34297       00:00 ETA
Done.
Updated list of available packages in /usr/lib/ipkg/lists/experimental
root@OpenWrt:~# ipkg upgrade
Package kmod-diag (2.4.30-1) installed in root is up to date
Package kmod-brcm-wl (2.4.30-1) installed in root is up to date
Package kmod-brcm-et (2.4.30-1) installed in root is up to date
Package kmod-ppp (2.4.30-1) installed in root is up to date
Package kmod-pppoe (2.4.30-1) installed in root is up to date
Package openwrt-utils (1) installed in root is up to date
Package kmod-wlcompat (2.4.30-1) installed in root is up to date
Package bridge (1.0.6-1) installed in root is up to date
Package busybox (1.00-2) installed in root is up to date
Package dropbear (0.45-2) installed in root is up to date
Package dnsmasq (2.22-1) installed in root is up to date
Package iptables (1.3.1-1) installed in root is up to date
Package ppp (2.4.3-4) installed in root is up to date
Package ppp-pppoe-plugin (2.4.3-4) installed in root is up to date
Package wireless-tools (28.pre6-1) installed in root is up to date
Package zlib (1.2.2-1) installed in root is up to date

And now reboots my linksys WRT54G v1.1

Wouter.

Latest experimental snapshot still contains a bug in the ethernet driver, which crashes older v1 Hardware (e.g Linksys WRT54Gv1 or v1.1).
We will post a new snapshot soon. In the mean time, please try my updated build at http://openwrt.inf.fh-brs.de/~nbd/gcc34/ and see if this fixes your crashes.

Hi,

I have tried your build:
root@OpenWrt:~# dmesg
CPU revision is: 00024000
Primary instruction cache 8kB, physically tagged, 2-way, linesize 16 bytes.
Primary data cache 4kB, 2-way, linesize 16 bytes.
Linux version 2.4.30 (nbd@ux-2y02) (gcc version 3.4.3) #1 Fri Apr 29 21:53:53 CEST 2005

I can now load the modules, but I have troubles obtaining an ip address. As soon as I connect the internet interface, the lights start flickering but it doesn't get an ip address.

I will further into it.

Wouter

Last build had a bug in the init scripts, so that IP forwarding (and other stuff) wouldn't get set. I have updated the build at the same URL.

Are you sure you have resolved the bug? My WRT54G reboots every 10 minutes.

Not entirely sure, but at least one user had it working stable... Maybe it's not fixed yet. We're working on an alternative (free) ethernet driver that will fix those issues.

What does you latest build says when you ask dmesg?

Looks like the "heavy traffic crash bug" is solved as stated at http://openwrt.org/forum/viewtopic.php?id=1384. Only had to:

nvram set clkfreq=216
nvram commit
reboot

And no more reboots/crashes on my WRT with heavy traffic. It's been working for 6 hours and had any reboot or crashes. Before doing this it used to reboot each 10 minuts, and crashed on heavy traffic.
The WRT has 2005.03.29 openwrt experimental firmware.

Now ksoftirqd_CPU0 doesn't takes 90% of CPU usage, it takes 15-17% CPU usage while transfering 500 Kbytes/sec and 30-37% while transfering 1000 Kbytes/sec. Haven't tested higher transfer bandwiths.

Update: 24 hours with this new setting and had 0 reboots/crashes :-)

(Last edited by acidbits on 3 May 2005, 12:41)

I have installed the build of 5th may of http://openwrt.inf.fh-brs.de/~nbd/gcc34/ and I still have problems obtaining an ip address on the wan interface.

If there is no cable connected to the wan interface, the lights of the lan interface keeps on burning and flash only when there is traffic.

When I connect the cable to the wan interface, all lights (lan +wan) keeps on flashing even if there is no traffic.

Any help?

(Last edited by WouterC on 6 May 2005, 09:00)

Of all of the miniscule bug fixes...

BusyBox httpd HTML encode string (-e) doesn't output properly formatted HTML entities.. It's missing the semi-colon.

i.e.
'httpd -e "<html>"' results in:
&#60html&#62

Unless, of course, numbered entities shouldn't have the ; terminator.

If this is all that's left, you're doing oK.

Hello everybody!
I'm running openwrt experimental (cvs, 06.05.2005) on an Asus wl500g and I'm experiencing random segmentation faults and other unexpected behaviour e.g.:

root@OpenWrt:~# ps -A
  PID  Uid     VmSize Stat Command
    1 root        384 S   init
(...)
  436 root        424 S < /usr/sbin/pppoa3 -c -vpi 8 -vci 48
Illegal instruction

root@OpenWrt:~# ps -A
  PID  Uid     VmSize Stat Command
Segmentation fault

that didn't show up with the firmware I were previously using (Oleg's custom firmware 1.7.5.9-5). I don't know if (or how) this is related to the "large file transfer reboot bug", but I trying different values from 108 to 128 for the clkfreq nvram variable didn't help (the wl500g is normally clocked at 125MHz). Any ideas on this issue?

Hi, I have another problem, this time with Quagga. When I compile custom firmware with quagga included (e.g. zebra, ripd), only ripd is installed. I made another tests and it seems that only the last "part" of the quagga is installed into firmware - e.g. if you select ripd, bgpd and ospfd, only ospfd is installed. In file lists in /usr/somewhere all parts are listed correctly though. I suspect some Makefile bug but I have no time to fix it myself now.

Please use the quagga packages that are available here :

http://openwrt.alphacore.net/experimental/

The "official" ones are completely buggy and there are duplicated librarires instead of symlinks for instance.

OK. Seems like you checked out the 'TESTED' revision, which doesn't have that fix yet. To fix this, you can either grab 'HEAD' from CVS and build it or try my updated build at http://openwrt.inf.fh-brs.de/~nbd/gcc34/ if you don't want to compile everything again.