chrisf

Members
  • Content count

    132
  • Joined

  • Last visited

About chrisf

  • Rank
    Elite member

Recent Profile Visitors

633 profile views
  1. chrisf

    gotop

    You've bunged up the link somehow, there's a unicode byte order marker between i and / in cjbassi/gotop If you copy/paste into a url bar you'll get a 404 and scratch your head because it looks identical to a url with %EF%BB%BF in it
  2. My guess is PCB bad layout between the memory and SoC, where the production tolerances of the DRAM/SoC mean some work fine, other don't.
  3. chrisf

    [Solved] SDRAM parameters

    I'm not sure if I understood this correctly. You want clear and complete documentation from Allwinner?
  4. Personally, I solved my Espressobin problems by throwing it in the bin.
  5. chrisf

    Enabling 1wire support rock64

    It then gives you the freedom of using a w1 kernel driver or doing it all in userspace. Being i2c, it's a master clocked protocol, so there are no timing issues even if you don't have a hardware i2c port available. OWFS supports the Maxim i2c chips too
  6. chrisf

    What about Google Home

    What about porting Google Home to an ARM linux distro?
  7. chrisf

    Enabling 1wire support rock64

    If you want reliable, go for a 1wire to i2c bridge like the ds2482 or something. Bit banging timing sensitive protocols in a multi user OS is never a good idea. It wastes CPU time with spin waits and if the process gets pre-empted it will probably fail.
  8. You could start by trying to figure out what the touch screen is. lsusb should give you the USB PID/VID when it's plugged in. You could then have a look on http://www.linux-usb.org/usb.ids to see what it is. Alternatively find the part number for the touch controller chip on the screen. Armed with the knowledge of what drivers you're looking for, you can start looking for drivers. You may be able to pull them out of one of the RPi distros you mentioned.
  9. chrisf

    Daily (tech related) news diet

    Don't worry, Google is only a gold member of the Linux Foundation. Microsoft is a Platinum member...
  10. I decided not to waste any more time and money in it. I'm not going to pay to ship back a faulty $49 board, considering how much international postage costs in my part of the world. There's also the risk I experience the same faults with a replacement. My time is also not limitless, nor do I consider it worthless. I never got a chance to try mainline kernel. It wasn't stable enough to complete the process to change it. I was using the latest firmware from https://dl.armbian.com/espressobin/u-boot/
  11. Yesterday I tried getting my old espressobin going again. Flashed new firmware, new sd image. Back to the random kernel panics I used to get. Instead of trying to find the specific version of firmware I had that fixed the crashing. The boards are defective or DDR timing in the firmware is untested/too aggressive for the board layout. If you're lucky, the tolerance of the soc/ram chips on your board will work fine with the firmwares available. All I wanted was a board that is stable and runs at the speed advertised (1.2G) with the features advertised (advanced crypto acceleration). Last nights attempt was to use the latest mainline kernel that includes the crypto drivers. I never got a chance as it kept crashing during the apt processing trying to switch kernels. In the end I used a hammer and vacuumed up what was left to make sure I don't suffer this frustration again.
  12. chrisf

    New forum UI!!

    Looks like the logo doesn't work well with mobile. It's cutting off some text
  13. chrisf

    Ethernet flow control

    Apparently flow control on full duplex gigabit links can cause the link to operate at 100M if it's talking with both gigabit and 100M nodes at the same time. https://www.smallnetbuilder.com/lanwan/lanwan-features/30212-when-flow-control-is-not-a-good-thing That could be a reason why some gigabit switches don't support it. when running in a mixed fast/gigabit network it can impacts transfers between two gigabit nodes if one of them is also talking to a fast node. as a manufacturer you don't want to field complaints about poor network performance that is "fixed" by swapping your switch with another brand.
  14. chrisf

    Ethernet flow control

    Perhaps your 1000Mb/s switch does not support flow control? Don't both sides of the link need to support it for it to work?
  15. chrisf

    /sys/class/pwm as non-root user

    Google udev rules