Lair Of The Multimedia Guru

For a long time i used cheap wired logitech or microsoft mice and that was ok-ish. But from time to time one needs to buy a new mouse for a new computer, because the old is so dirty its disgusting or the really rare that it stopped working 100% reliable. So for some forgotten reason i bought a new one and my mistake here is that i trusted the amazon ratings (never do that). So i got a “noname == tecknet” mouse, it was cheap had more buttons than the cheap logitech one. It even was the “Amazon choice” for wired mouse. How it got that rating, i can not comprehend. The problem with this mouse is it simply does not work reliable. You turn the scroll wheel one way and it sometimes registers that and sometimes registers the opposite direction.
So as that thing was just annoying, i picked a new mouse again and using my brain slightly more this time, i concluded that any gaming mouse must be reliable as gamers dont like loosing from lost or misread buttons. So i choose based on specs and price a corsair gaming katar pro xt. Nice mouse nice hardware. linux support well, kind of. It works fine out of the box if you like your computer looking like a Christmas tree.

Switching off that LED

My first thought was of course that there must be a tool on linux to turn this off. I failed to find it though.
So the next was screwdriver / soldering iron but that was really not feeling right.
So i spend several hours implementing enough support in ckb_next to adjust the LED for this mouse. This would have been quite trivial with a windows box running the official driver and listening to it but i didnt had a windows machine, so randomly reading and writing things to the mouse was the way to go. This was especially fun as the mouse remembers its settings when disconnected. So if you successfully set something you have an incentive to undo that if you want it working as before. I didnt destroy the mouse no, basically blindfolded stabbing it with random writes allowed me to identify a few more fields of what seems called the bragi protocol. So the protocol header now lists the fields for the hardware DPI values for the 3 modes and also the corresponding colors. the mask of enabled DPI modes, a DPI index (which the hardware does not seem to bound check at least not the way id expect). And the LED brightness for both software and hardware modes. I very interestingly did not find how the LED color itself is set or how the hardware mode animation is controlled. So there are unknown things left to be found. But for my purpose what i wanted (turing the LED off) i found all i wanted.
Code is here, pull request is also send, so it may or may not end in the official ckb-next

A few days ago i wanted to print a page, nothing special here thats a common thing. But this time was different, my Samsung CLP-365 color laser printer did not print, its main LED red, the 4 toner leds all the same and lighting. On the linux side no error message or anything. Paper was in it, no paper stuck anywhere, power cycling did not help. None of the button combinations i could find that where intended to print debug information did print anything. Searching the wide net lead to a service manual and the tip to look at the display which my printer did not had and some windows software which required a windows machine. A youtube video pointed to the waste toner container but that was not the reason for the failure though it was messy.
What finally helped was connecting the printer to a mac, its printer drivers finally provided a useful error message, namely that the yellow toner cartridge was incompatible. It of course was not a original samsung one because the originals cost as much as a new printer. The other 3 cartridges where still the originals, the yellow one was bought from amazon a bit less than a year ago and it worked fine for a few months. (and this seemed not related to a firmware update)
The final confirmation that this was the issue came today as i received a replacement yellow toner cartridge again non original of course. And putting that in the printer, it came back to life, for now at least.
Iam not sure what i should think about this, but this is uncool. First samsung WTF, why if theres a fatal issue with a toner cartridge why do the LEDs for the toners all light up equally that gives 0 hope to one finding the issue. And iam not commenting about the whole toner original vs compatibility thing except id like the CEOs of all th e printer manufacteurs that make it hard to produce compatible toner cartridges to cleanup all the avoidable waste this creates. Go with a row boat please to the Great Pacific Garbage Patch and pick your incompatible toners and still good but misdiagnosed printers out and recycle them properly.

Thanks

FATE strikes back

The Cubox no longer boots, the Panda boards power supply no longer works, it actually literally fell apparent when i unplugged it

The USB stick in my Panda board failed, huge amounts of bad sectors. Ive had it running with the power supply from the dead cubox and the sdcard as storage device.
That lead to out of disk space issues as fate is not just evil but its also big. So today the panda board has a new usb stick (SanDisk Extreme Go 64GB USB 3.1), which was the cheaptest brand name + fast one i found quickly.
Some benchmarks of the SD card vs the USB stick

sd card configure (-O1 gcc-4.4)
real 0m50.546s
user 0m24.523s
sys 0m11.164s

sd card build -j2
real 24m46.453s
user 38m38.586s
sys 3m14.938s

sd card fate -j2 -k
real 75m42.462s
user 104m30.688s
sys 8m33.727s

usb configure (-O1 gcc-4.4)
real 0m47.655s
user 0m24.719s
sys 0m11.297s

usb build -j2
real 21m25.833s
user 38m32.438s
sys 3m5.820s

usb fate -j2 -k
real 57m29.724s
user 103m6.180s
sys 7m30.867s

My i7-5820K box, which runs several fate client VMs also was showing signs of age, its main disk (Samsung SSD 850 PRO 512GB) showed an accelerating increase in number of reallocated sectors. That being after over 5 years of continuous 24/7 use by multiple FATE VMs, reading and writing. Ive a few days ago replaced it by a Samsung SSD 970 PRO 1TB. That also means theres spare space now for new fate clients if i find the time to setup something …

2 days ago before going to bed i started some tests on my box. When i woke up yesterday, i couldnt login remotely to the box, looking more careful the box was waiting for keyboard input during boot. It had issues previously and i intended to replace it for a long time. That was on hold because the ryzen 3950X isnt available yet which i wanted to use for a new box.
So i helped it boot, restarted the task and a bit later the box fans and disks spun down it rebooted again. My first thought was power supply or mainboard issue. The box after all was used alot and was somewhat old. I had a power supply laying around which was intended for my the future 3950X. So a little unhappy i moved the other machine that blocked access to the failing box away. Next with some dust mask and vacuum cleaner i removed many years of dust. Having removed all dust that failed to hold on. I wanted to check if it still booted, what temperature it reached and if maybe it worked now. The moment i switched it on i was glad i had the dust mask still on, for some reason more dust was blown out by the noctua fans. And it was interesting to see the dust coming out and some of it being sucked into my still running vacuum cleaner. The machine booted and i restarted the task, it didnt fail. So i tried again and then run fate in a loop. It reached 86°C cpu temp IIRC but it didnt fail or reboot again. Now a day later, the machine showed no further anomalies, it seems happy again.
This was the first time I ve had a system apparently failing from nothing but dust.

A few days ago when i wanted to run some benchmarks on my cubox, i run into some odd errors, investigation led to the micro sd card. There where several bad sectors that could not be read anymore.
Thats the first time ive myself seen a sandisk micro sd fail. ive seen transcend (really quickly) and kingston (not so quick) fails but no sandisk or samsung till now.
Another point for FATE, another disk it won against…
But how badly damaged is the sd card? ive run ddrescue on it for a few days:

# ddrescue -r 5 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7947 MB, errsize: 493 kB, errors: 217
Current status
rescued: 7947 MB, errsize: 254 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 48, average rate: 122 B/s
opos: 7389 MB, time since last successful read: 1.5 m
Finished
# ddrescue -r 20 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7947 MB, errsize: 254 kB, errors: 48
Current status
rescued: 7948 MB, errsize: 193 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 37, average rate: 10 B/s
opos: 7389 MB, time since last successful read: 7.2 m
Finished
# ddrescue -r 40 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7948 MB, errsize: 193 kB, errors: 37
Current status
rescued: 7948 MB, errsize: 176 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 36, average rate: 1 B/s
opos: 7389 MB, time since last successful read: 41.4 m
Finished
# ddrescue -r 80 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7948 MB, errsize: 176 kB, errors: 36
Current status
rescued: 7948 MB, errsize: 172 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 35, average rate: 0 B/s
opos: 7389 MB, time since last successful read: 4.7 h
Finished
# ddrescue -r 160 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7948 MB, errsize: 172 kB, errors: 35
Current status
rescued: 7948 MB, errsize: 163 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 33, average rate: 0 B/s
opos: 7389 MB, time since last successful read: 41.7 m
Finished
# time ddrescue -r 320 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7948 MB, errsize: 163 kB, errors: 33
Current status
rescued: 7948 MB, errsize: 151 kB, current rate: 0 B/s
ipos: 7389 MB, errors: 30, average rate: 0 B/s
opos: 7389 MB, time since last successful read: 4.7 h
Finished

real 1119m16.116s
user 0m1.274s
sys 0m3.877s
# time ddrescue -r 640 /dev/sdd backups/cubox-2019-01-07 backups/cubox-2019-01-07.log

GNU ddrescue 1.17
Press Ctrl-C to interrupt
Initial status (read from logfile)
rescued: 7948 MB, errsize: 151 kB, errors: 30
Current status
rescued: 7948 MB, errsize: 147 kB, current rate: 0 B/s
ipos: 4393 MB, errors: 30, average rate: 0 B/s
opos: 4393 MB, time since last successful read: 15.1 h
Retrying bad sectors… Retry 375^C
Interrupted by user

real 1236m21.926s
user 0m1.383s
sys 0m4.204s

Its interesting to note that even after quite some time there are still damaged sectors which can be recovered. It was just myself loosing patience that made me stop it. Also i have a backup somewhere IIRC so this recovery was just out of curiosity. And maybe ill play more with this card in the future to see if theres anything that can be done to improve recovery.

Hopefully ill have the cubox with a new sd card up and running fate soon again … (should be quite trivial)

A few days ago i noticed that my panda-board died. This is one of the 2 ARM systems on fate.ffmpeg.org.
Resetting or unplugging and re-plugging did not help. Replacing the power supply with a random 5v supply made it come back to life.
The failed power supply is a HNP-24-050 “HN Power Germany”. This is now probably the 3rd or 4th failed power supply for my panda board. The one its running on currently is a random one from ali-express. I must say iam really disappointed by the trash that some german electronic shops sell. All the failed power supplies where bought from germany, not china. The failed supply still produces 5v but when subjected to the slightest load its output collapses to 0, even 200mA is too much.
Ive read on the net that some digikey supply is recommended but both part numbers i found are marked obsolete.
Anyone has any recommendations ? I am not too positive that the 2€ +free shipping supply its running on currently will last very long.
I know i can just run it off a brand name ATX supply or a bench supply but that is a bit inconvenient.
Also if anyone wants to add more ARM hardware to fate.ffmpeg.org, so we have a bit more redundancy, thats certainly welcome

After running into the row-hammer issue with new DDR3 SO-DIMMS for my notebook, ive tested all my used computers ram. I think i had tested at least some of it previously. These where DDR4 Micron Technology 16ATF1G64AZ-2G1A1 and CML16GX3M4A1600C9 from corsair. Both showed no issues on multiple passes of the row hammer test. Ive also now obtained new SO-DIMMS for my notebook, M471B1G73EB0-YK0 from Samsung and HMT41GS6AFR8A-PB from SK Hynix. Both as well have shown 0 issues in the row hammer test on multiple passes. So it seems that the issue is not as widespread in 2017-11 as looking at a set of 2 modules would have suggested. Ill return the faulty modules …
Iam a bit curious who the manufacturer of the chips on the faulty modules is but there are stickers over them and as i want to return them i dont want to peel these off. And in dmidecode they are just listed as “Corsair”. I guess if Corsair wants to make it hard to find out who made them, they will have to take the blame here.

My notebook has 8gb ram and without swap occasionally that was not enough. So i thought hey trivial put more memory in. And while online resources disagreed on the maximum supported for my Acer Aspire V3-571G. It seems fine with 2 8gb sticks. But …
When running memtest (MemTest86 V7.4 actually, which was the first google pointed me to). It displayed after a while a note, “[Note] RAM may be vulnerable to high frequency row hammer bit flips”. After a short wtf moment, and a bit of research, this is a hardware bug that can be used to flip bits one normally has no access to and exploit an affected machine. I think i did actually read about it years ago but did not immediately connect the note message to it.
But what i find really disturbing is that we have november 2017 many years since its known. And these are newly bought sodimm modules (brand name in fact from corsair, CMSX16GX3M2B1600C9. Both of the 2 modules alone also are affected.
Un-voluntarily that also leads myself to the question, does anyone know of similar sodimm modules (can be slower i dont care) which do not have this issue? Or a way to increase the ram refresh rate on a notebook where this seems not available in BIOS? (I guess i could go read the chipset datasheets but maybe some tool already exists …)

About 3 months ago my cheap/small color laser printer stopped eating. It tried but most of the time it just failed and very soon it completely stopped working, giving a generic paper stuck kind of error so i had to take it apart and fix it as i needed it. I am writing this 3 month after fixing it, as i didnt had time before, but maybe its still useful for someone.
I have the suspicion all printers of this design will fail this way.

Disassembly

The printer is held together with screws and little plastic hooks, below are some pictures of my disassembly of it. The part we need to get to, is a small steel part which is moved by a solenoid.





Below is the part that causes the problem, the sticky foam thing on it has become slightly sticky on the wrong side and is also squished. The device still works if the solenoid is pulsed with sufficient power.

Repair

This is very trivial, remove the decaying black foam thingy and replace it with something that is non sticky on one side and
has dimensions somewhat similar to the original. I used very strong double sided tape to get the thickness and covered in chinese “blue” painters tape from my 3d printer to make it non sticky on one side. If you can, test that the solenoid works before reassembly.

Reassembly

Just undo the disassembly in reverse order, weave all the wires back in as they where before.


3 Months since the repair above, my printer still works and had 0 paper feed failures (unless it run out of paper).

A few days ago i replaced my acer notebooks keyboard, several of its keys had become unreliable months ago I first attempted to clean it but that just broke some little plastic part off with no improvement. The process was a bit boing, 24 pages of the (well written) service manual 36 screws some epoxy and 2 hours later my note book was working again with new keyboard.
This notebook is definitely not built to be serviced, having to disassemble it partly from the bottom to get to the top. Some screws screw into plastic, many of them have the plastic cracked after screw removial making reassembly tricky. Some screws screwed into metal inserts, 2 of which broke off without me even touching them, requiring the epoxy to reattach the metal inserts onto the brittle plastic. There are also plastic latches before you disassemble things, the ones of the main door at the bottom are indestructable it seems but the ones deeper inside seem not intended to be reused judging from the smaller number of clicking sounds when reassembling …
My previous acer notebook i once dropped and it still works. It also still has its original keyboard and the hp omnibook i had before i could use until many of the keys on its keyboard had no readable symbols on them anymore. Though today that old hp notebooks plastic parts have also become extreemly brittle but its keyboard could be exchanged by just removing one or 2 screws IIRC.
In the distant past electronics seemed to last forever becoming obsolete long before failing. Now it seems most stuff fails long before becoming obsolete.

Both keyboards after the replacement:

And most disappointingly no parts where left unused after reassembly nor where any missing. I guess my workspace was too tidy.