So, I forgot I used divested, did firmware update with wrt build (not divested). Broke one partition. Then booted into another, fw_setenv boot_part 1, didn't reboot, uploaded another firmware, boom - fatality, both unbootable partitions.
Knowing im f*cked up, disassembled router, took some uart cables
flipper gpio TX RX GND as a bridge - got output for two times in 1000.
cp2103 red chinese adapter that helped me on bunch other fkups and some new leads - no connection.
flipper developer board esp32-s2 - no connection
bought some prolific knockoff usb adapter usb-ttl - not outputing a thing.
bought another prolific knockoff usb adapter usb-ttl - yay, output is there. Though, cant send any signal, and venom>>> appears only wiggling rx (believe so) connection on boot.
While I have been reading and wiggling all the ends for past 20 hours in two days, I am seeking for help.
My brick wont speak to me. Here below is a picture where I am able to get, but not when all 3 leads are connected to uart.
It appears to me, that something is leaking, bad shielding, or bad cables. I might go buy 5 more adapters/cables, to verify.
Any opinions, how can I make this console typable? Or maybe there is another point for JTAG, which pin?
I have measured voltage, starting from ground pin:
(i mistakenly posted this in the wrong thread earlier)
i htinnk there may be a problem with recent sysupgrade logic on the wrt32x.
on my wrt32x (dual partition) r27107 will only sysupgrade from 1->2.
going the other way makes partition 1 unbootable.
i upgrade periodically and have generaaly save old builds i tried going back a few builds and found a similar pattern back to r26608 r26300 works fine in both directions. i think this corresponds to the interval when auc was being replaced by owut and changes were sysupgrade were being implemented this probably does not affect all devices otherwise it would have been reported and addressed…
There was a short lived issue on master which yielded an image that could not be flashed away from (in GUI, CLI worked), or over. I was forced to tftp an image over the borked partition. Everything was fine except the inability to flash away from the image, or to flash over the image from the other partition. I compile my image, so no AUC or ... Version was r26967, and still is on a rango, until I find the time to pull and connect a serial.
yes, it might have been the inability to flash away from the image.
since i only had that on one partition so i used mtd write an openwrt .img from r26300.
do you know what the issue was?
i'm concerned that the issue persists in a way that onnly allows sysupgrades from 1->2.
How do i tftp, if i dont know the router address? I cant setenv ipaddr. Anyonr got an idea?
Tried kwboot command uploading new wrt32x boot loader, but "reboot router" never picks up reboot.
As the issue was resolved with a later version I never bothered trying to bisect to the problem commit(s). Looking at the versions in my drop, r26300 is much earlier than versions that were working OK for me, so not sure...
@xkuza , Re: tftp. Get a USB TTL serial adaptor, see the OpenWrt wiki and / or the McDebian wiki. If you are a MS person the drivers for knockoff devices can be challenging, any *nix device is usually no problem.
thanks - i know all the serialaccess things. i still have acceess to the 2nd partition, which is currnetly on r27110. i cant sysupgrade to the first partition.
as a work-around -
....
mtd erase kernel1
mtd write openwrt*.imp kernel1
fw_setenv boot_part 1
reboot
...
this will get me going on parttion 1. if i reboot to partion 2 and try sysupgrading to kernel1 - it failes to write.
if i go back to r26300 then i can sysupgrade to both partitions.
weird; i can always keep r26300 on partition 1 and then sysupgrade to partition 2, but this seems not right.!
@anomeome FWIW - i went back to r26300 on both partitions. sysupgrade also saves configs on upgrade to other partition on this build too. i'm thinking something is not quite right with sysupgrade more recently. i'll await some developments. thanks t
@akuza - i'm wondering if you shold avoid the development snapshot builds for now and flash 23.05 builds until this is sorted?
Managed to get the rango off of the above image, moved to r27119-c609c3bd6e w kernel 6.6.45 using mtd and a config restore. Then booted back into the previous image from which I did the mtd write and did a successful GUI upgrade over it again to be sure of success.
Not sure why it did not occur to me to try mtd write image.img, thanks for the reminder there is more than one way to skin that critter. IIRC there were some nand commits around the image version where things got borked.
Good point, right you be, still borked, cannot flash away from the most recent image. Will break out the mamba on which I have a serial to see if there is anything to be seen, but I think it is probably only going to be found with a bisect...
Edit:
nothing of interest spotted
most recent version on my drop that still upgrades is r26868
the u-boot defaults for the wrt32x will silence the console. you can enter commands but not see the response.
to change this behavior, from the venom prompt, enter these commands
....
setenv silent 0
saveenv
...
you wont see your typing so be careful to avoid errors.
you might have to power cycle to get all u-boot output unsilenced - i cant remember,
once you set 'silent 0' into the u-boot environemnt and save it, it will stick until you enter 'resetenv'. so i suspect posters with wrt32x and uboot access might have already done that.
so, yes, when i first started on my wrt32x, i blindly and carefully entered the commands above. it's scary at first - agreed.
i dont know what clang is.
images are build in formware-selector, with these selections:
...
-wpad-basic-mbedtls
-libustream-mbedtls
wpad-openssl
libustream-openssl
