On 11/5/26 19:22, Greg Troxel wrote:
This could well be related. I have seen stalls on a per-device basis on
USB where other things (on the same controller and certainly on
different controllers) are ok.
I think the main question would be how NetBSD handles these events. If
just freezing is the expected response, at least when there's no disk
swap present, that would surprise me.
Regards,
Ellie
Tangent to not bore everyone, but here is how Linux handles these:
On Linux without a disk swap file, I've seen full USB brownouts (with
USB SSDs on a Raspberry Pi 5, which doesn't have the USB power of a
normal desktop), and usually it doesn't freeze at all:
The graphical desktop will happily continue to run while trying to
launch any command "dmesg" gives "I/O error" instead of hanging.
Meanwhile, you can watch the graphical apps die one by one as they try
to load something from the disk and fail, and the system typically
remains perfectly responsive.
If the "poweroff" binary happens to still be cached, which you can
ensure by looping while true; do sleep 5; poweroff --help; done in a
shell, you can even typically shut down a Linux system just fine.
