<digitalcircuit>
ipq806x CPU L2 cache 1.4 GHz bug hunt update: USB power meter finally arrived! I'll play around with trying to catch if the USB 3.0 1 TB HDD power draw goes rogue on the build with L2 cache enabled.
<digitalcircuit>
I also still need to look into figuring out a "safe" raised CPU voltage level for the DTSI, and I may try formatting an old 256 GB SSD and connecting it to the router to see if I can recreate the issue with a non-spinning-rust device.
<Tusker>
ok, nice :) good luck, I am really only at "tweaking" level for DTS, have to really read through bunches of source code to find out what the different things do
<digitalcircuit>
Thanks! Pretty much the same here - diving in headfirst to this issue might have not been my wisest choice (though nobody seems too annoyed yet), but I'm certainly learning a lot along the way :)
<Tusker>
well, involvement is not usually looked upon badly... when people come in and complain without rolling up their sleeves is when people get annoyed
<aiyion>
blocktrron_: the build was a success; though I do not know what I've done in the kernel and kenrnel_initramfs line, yet. I pushed all local changes.
<aiyion>
0xf7a00c is not found in the repo though. So either, I've found a variant of the winbond chip, that's not yet registered in openwrt, or the number I found is some gibberish based on another problem.
<aiyion>
Or there's another source that registers the id outside of the openwrt repo.
<PaulFertser>
aiyion: do you have a logic analyser?
<aiyion>
No, sorry
<aiyion>
spi flasher and an ancient oscilloscope.
<PaulFertser>
aiyion: I'd still try varying the frequency. Set it to 20 MHz and see if the read number changes. Who knows, I'd say worth a try.
<PaulFertser>
aiyion: I wouldn't recommend probing for frequency, you might arrive at something that's out of spec, works for you but not for the others. I'd check what real frequency is used with a 'scope and how the signal looks.
<aiyion>
The way to go is, I take a stock device, hook up clock to an oscilloscope and extract the frequency?
<PaulFertser>
aiyion: what _I_ would try to do is to determine if the frequency as you set matches what's really used. You set 20 MHz so I would use a scope to check it really is almost exactly 20 MHz. If it matches I would check the datasheet for this model specifically (there're FV and BV winbond variants that differ IIRC) to see what maximum frequency is allowed for regular commands using the Vcc you have
<PaulFertser>
there. Then I would try setting that max frequency and if it doesn't work I'd use the scope to see if the signals are still good. If the edges are not sharp enough I would check if drive strength of the SPI pins of the SoC can be adjusted.
<PaulFertser>
Because higher frequencies require higher drive strength usually.
<PaulFertser>
aiyion: it's ok to choose a conservative number for the initial device submission. You can improve the performance later in a follow up patch.
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
<aiyion>
This will not much be of a blocker; a friend of mine just agreed to meet me after hours in the university; they've got sufficient hf-hardware.
<aiyion>
* be much
<karlp>
russell--: but the cheap ones are reallly pushing it to reach 24MHz.
<aiyion>
thats what I wanted to say :)
<karlp>
and if aiyion is looking at signal integrity at 100Mz.....
<karlp>
a LA isn't really the right tool anyway.
<aiyion>
I only find 10ish up to 24MHz
<aiyion>
find 50 bucks to 100Mhz
<aiyion>
but that will still not nealry be enough :/
<aiyion>
Either way, I really appreciate this chatroom and the community around it. It's amazing how much patience you put into beginners like me, thanks!
<PaulFertser>
karlp: I mentioned LA only to check what data is there when one uses lower clock but the device still gives odd JEDEC numbers.
<aiyion>
the jedec number does no make sense.
<aiyion>
*now
paper_ has joined #openwrt-devel
<paper_>
hi, is there a convenient way to list files in a built package? maybe a common script collection?
<PaulFertser>
aiyion: yes, so LA is no longer suggested :)
<karlp>
paper_: I normally just do it by hand, untar the data tar inside the package.
<karlp>
there's probably a better tool...
<karlp>
$ tar -xf bin/packages/mips_24kc/packages/lua-cjson_2.1.0-2_mips_24kc.ipk ./data.tar.gz -O | tar -tz
<karlp>
the control.tar.gz has the postinst/prerm scripts from the makefile and similar
<paper_>
karlp: I have something very similar, but not as pretty, thanks
dedeckeh has quit [Remote host closed the connection]
nitroshift has joined #openwrt-devel
rejoicetreat has joined #openwrt-devel
<aiyion>
There's a path on routers /sys/class/spi_master/spi0/spi0.0/of_node/spi-max-frequency that contains gibberish; which tool do i use to extract its value?
<aiyion>
is there a liteweight hexeditor we have available?
<fpsusername[m]>
<Habbie> "fpsusername, i suspect my soic16..." <- This morning I got a notification from the PostNL app, really thought it would be the stuff. Just some letters
<aiyion>
hexdump delivers "0000000 0131 2d00" and "0000004" on seperate lines
<aiyion>
1312d00 is the 20MHz I configured; what does the four mean?
<ldir>
anyone else getting recursive dependency issues with ethtool ?
aleksander has joined #openwrt-devel
<aiyion>
PaulFertser, russell--: thanks. Given this path does not exist on the original ar71xx images; is there another way to read the spi-speed from software?
<PaulFertser>
aiyion: don't you have sources for the original image?
<fpsusername[m]>
Habbie holy f sht. Fastest AliExpress delivery I've had so far. Ordered on the 28th, clip and jumpers arrived today. 8 days from China 😳
<aiyion>
I'll verify this later.
<aiyion>
Have a nice afternoon everyone.
dangole has quit [Remote host closed the connection]
<fpsusername[m]>
Totally missed it that the pitch of the SOIC 16 chip is 0.127mm, so I can't attach the jumper wires to it. Dang it. Bought another clip now lmao, not going to waste my time soldering it
<Habbie>
fpsusername[m], oh, wrong pitch on the clip?
<Habbie>
fpsusername[m], i trusted you on that :D
<fpsusername[m]>
Well, no
<fpsusername[m]>
The clip is fine
<fpsusername[m]>
I just bought the clip without soldered wires
Acinonyx_ has joined #openwrt-devel
<fpsusername[m]>
But I forgot to check the pitch
<fpsusername[m]>
The most common pitch for pins is 2.54mm and I stupidly assumed the same for the chip, even though it's 1.27mm
<Habbie>
also, funny story, this week it is -raining- complaints about international shipments being stuck in postnl
<fpsusername[m]>
I mean, I could cut the wires and solder them to it, but eh, I don't have heatshrink and at that pitch size I'll probably make a mess soldering without a third hand stand
<fpsusername[m]>
The pins where you'd connect wires to are 1.27mm away from each other
<Habbie>
at the 'back'
<Habbie>
just like the pitch for the pins at the 'front'
<fpsusername[m]>
But, normal jumper wires (which you'd connect to your Raspberry/Arduino/Dev board) are meant for pins that are 2.54mm apart from each other
<Habbie>
oh, and your duponts are too big to fit between the 1.27mm things?
<Habbie>
right
<fpsusername[m]>
Yes
<Habbie>
i had the same problem with a CC2531 pin header last week, so i had to order another plug for that
<fpsusername[m]>
So I could cut the heads of from the jumpers and solder the wires directly to the pins on the clip
<Habbie>
right
<fpsusername[m]>
Or buy another clip
<Habbie>
which is a way less risky deal than soldering to the SPI, of course
<Habbie>
right, because clips with wires exist
<Habbie>
i should order one just in case
<fpsusername[m]>
I don't feel like soldering on 1.27mm pitch with a mediocre soldering iron
<fpsusername[m]>
A 3rd hand stand would be very helpful though. I should get one tbh
<Habbie>
yep, those are great
<fpsusername[m]>
But then I'll be lost looking at tons of models finding out a great one and eventually spending more money than I would like
<fpsusername[m]>
That's why I still don't own a damn multi meter
<Habbie>
that's why i own two Action.nl multi meters
<Habbie>
they'll be fine until they're not
<fpsusername[m]>
The ones from AliExpress seem great, but the resolution sucks. You can only measure from 0.1mA or more
<fpsusername[m]>
EEVblog tested them and the resolution of the measurements sucks if you are testing out embedded electronics
<fpsusername[m]>
And I'm not ready to spend 100+ euro for a decent multi meter that I'd rarely use