08:27 <narmstrong> steev: yes

08:28 <narmstrong> steev: kernelci tests BE kernels, I’ll have a look

08:52 <sheepman> hi all, I'm still having issues with ethernet on my S912 box, no matter what I try I end up with ethernet working for a while then it suddenly starts dropping packets and/or experiencing extremely high ping latency. A reboot brings it back for a while but then it happens again. Also not 100% sure but seems to get triggered faster when the box is under load

08:53 <sheepman> currently running 4.19.0-rc7 (one of balbes builds) on ARMBIAN 5.62

08:53 <sheepman> oh it's gigabit ethernet btw

10:02 <yangxuan> narmstrong: have you ever tried enable CC_STACKPROTECTOR ? I make menuconfig then search it, but nothing found

10:53 <narmstrong> yangxuan: no but it should work, nothing related to Amlogic SoCs specificities

10:54 <narmstrong> sheepman: try to disable eee with ethtool

10:55 <yangxuan> narmstrong: yes, this should not related to SoC, but I can't find it, and can't enable it with kernel scripts, so it's kind of weird

10:56 <yangxuan> and seems it not depends on other config

10:56 <sheepman> narmstrong: it is already, it stops it dying under load but this is definitely a different issue

11:00 <sheepman> think eee is disabled by the dts i'm using

11:00 <sheepman> (just fyi)

13:41 <narmstrong> sheepman: a similar issue was fixed a hopefully backported to 4.18, can you use the latest stable tree ?

13:42 <sheepman> i tend to just use some of the prebuilt images available but i'll certainly have a look

13:48 <sheepman> RTL8211F chipset btw

16:28 <steev> narmstrong: i thought the kernelci tests tested with a initramfs, not on emmc/sdcard since writing those would possibly cause hardware issues

16:57 <xdarklight> steev: it does run from initramfs - but it's the kernel's job to detect the SD card (if there's one inserted) and the eMMC flash (if available on the board). so if you're seeing an error during SD card initialization then kernelci should have the same problem (at least that's the idea :))

16:58 <steev> xdarklight: si, I just meant, if the hardware doesn’t exist it can’t error

16:58 <yangxuan> narmstrong: when connect my s905d box with ssh, some time it will stuck for a while, have to wait some seconds to continue typing, if this is device specific or happen on most amlogic device with mainline kernel?

16:59 <xdarklight> steev: kernelci's Odroid-C2 with CONFIG_CPU_BIG_ENDIAN=y: https://storage.kernelci.org/mainline/master/v4.19-4404-g44adbac8f721/arm64/defconfig+CONFIG_CPU_BIG_ENDIAN=y/lab-baylibre-seattle/boot-meson-gxbb-odroidc2.html

17:00 <probono> hi all, i am still looking for a way to make a _full_ nand image of my box. any hints? (there seems to be no official restore img available online for my type of box, and actually many people are looking for one)

17:01 <steev> xdarklight: yep, that’s pretty much what I see here

17:02 <xdarklight> probono: unfortunately it's not possible with the mainline kernel (and I never tried with the Amlogic one) because Amlogic's kernel uses a custom flash format (they're storing the block status inside the main area, which mainline can't parse)

17:31 <probono> hm. if I save the content of all the partitions mentioned in the dtd, can I then be sure that I can fully recover an entirely bricked machine?

17:32 <probono> is there a way to make an Amlogic factory-like .image from those dd'd partitions?

17:32 <probono> err, "dts"

17:36 <xdarklight> not on NAND, IIRC the first NAND page is not referenced in the .dtb/.dts but hardcoded somewhere in u-boot / chip firmware and it's used to store some of the NAND settings (which then allows the chip firmware to load u-boot)

17:37 <probono> this is what i was fearing

17:38 <probono> thanks for confirming

17:50 <xdarklight> probono: you can always put u-boot on an SD card - just make sure to properly break your NAND then ;)

17:51 <xdarklight> if firmware can't find a valid bootloader on NAND it tries to load u-boot from SD card

17:52 <xdarklight> however, that only works if firmware detects the bootloader on NAND as invalid (I managed to overwrite my bootloader on NAND with 0x00 or 0xff once - firmware loaded u-boot from SD card just fine)

17:52 <vagrantc> defaulting to removable media sure makes troubleshooting easier

18:03 <probono> i figured so much, yes

18:41 <steev> xdarklight: wait... i just noticed at the top... it says pass - shouldn't it fail?

20:01 <xdarklight> steev: it also doesn't see any errors or warnings - not sure about that. maybe khilman knows why (as https://storage.kernelci.org/mainline/master/v4.19-4404-g44adbac8f721/arm64/defconfig+CONFIG_CPU_BIG_ENDIAN=y/lab-baylibre-seattle/boot-meson-gxbb-odroidc2.html shows some errors at the end of the log)