25.12.2025 19:07, Theo de Raadt пишет:
> Hmm.
> 
> I think we chose dmesg.boot for a reason before, but I'm not sure I
> remember the reason.
> 
> Maybe we thought $(dmesg) could have rolled over with noisy debug messages,
> and thus hidden names of devices we want to identify, and therefore
> dmesg.boot was more reliable?
> 
> Here's a thought:  Should we look at both $(dmesg) and dmesg.boot ??

I certainly have machines where dmesg gets flooded with log spam,
so any $(dmesg) with a few hours of uptime won't have a single device.

If fw_update stops looking at dmesg.boot, there won't be any chance
for it to update existing firmware, right?