On Wed, Dec 15, 2021 at 11:46:42AM +0800, Landon wrote:
> The error appear sometimes,  It doesn't always come up.
> Do you know what the problem is ?
> 
> [   28.829750] proc: Bad value for 'hidepid'
> [   29.253033] SQUASHFS error: xz decompression failed, data probably corrupt
> [   29.260761] SQUASHFS error: squashfs_read_data failed to read block 0x1b3f8cc

The data in the SquashFS volume is corrupted.  By "the error appears sometimes"
do you mean that:

1. Sometimes you build an image and it is garbage but other times you build an
   image and it works fine?

2. Some machines can be flashed with an image and they work, but other machines
   can be flashed with an image and they fail?

3. The same machine can be flashed with an image and sometimes it works and
   sometimes it doesn't?

4. A machine can be flashed once and then some reboots work fine and some
   reboots do not?

#1 would likely be a problem with your flash layout where the squashfs is
getting corrupted by an earlier part of the image being too big.

#2-4 would likely be some hardware and/or SPI configuration issue.

I would likely try slowing down the SPI bus speeds and/or turning of QSPI if
you have that enabled to get something stable and then work from there.  If this
is a new board it is possible you have some SI issues on the SPI line
(reflections, bad capacitance, incorrect level shifter IC for SPI speeds, etc.).

-- 
Patrick Williams