[prev in list] [next in list] [prev in thread] [next in thread] 

List:       freebsd-arm
Subject:    Re: Oddities on Pi4 under -current
From:       bob prohaska <fbsd () www ! zefox ! net>
Date:       2021-10-12 18:33:50
Message-ID: 20211012183350.GA2046 () www ! zefox ! net
[Download RAW message or body]

On Tue, Oct 12, 2021 at 04:08:31PM +0100, tech-lists wrote:
> 
> On Mon, Oct 11, 2021 at 05:56:41PM -0700, bob prohaska wrote:
> > A power conservation feature seems a likely culprit, if one has been
> > newly added or turned on.
> > 
> > The delayed response appeared after a world/kernel update, I think about
> > two weeks ago. I didn't make any manual config changes. Dropped ssh sessions
> > have been a problem for a while now, but this seems noticeably worse.
> 
> ** this rpi4 has aluminium case heatsink and recommended-by-rpi PSU.

Mine has a plastic case with a ~1" fan, no heatsinks installed, and a
Canakit power supply.

> 
> My rpi4/8GB has the following in /etc/rc.conf
> [...]
> powerd_enable="YES"

Same here.

> powerd_flags="-r 1"

No flags for powerd

> [...]
> 
> means it'd only shift down if use was 0.00 IIUC. So in practice it never
> shifts down.
> 
> in /boot/msdos/config.txt :
> [...]
> over_voltage=8
> arm_freq=2147
> sdram_freq_min=3200

No such changes

> [...]
> 
> results in:
> [...]
> % sysctl dev.cpu.0
> dev.cpu.0.temperature: 72.0C
> dev.cpu.0.freq_levels: 2100/-1 600/-1
> dev.cpu.0.freq: 2100

I get bob@nemesis:~ % sysctl dev.cpu.0
dev.cpu.0.temperature: 43.1C
dev.cpu.0.freq_levels: 1500/-1 600/-1
dev.cpu.0.freq: 1500

> 
> at the time of writing:
> % uptime
>  4:01p.m.  up 16 days,  3:10, 5 users, load averages: 4.49, 6.47, 4.68
> 
> because it's running a poudriere build.
> 

Mine just started a poudriere build of www/firefox. It'll heat up
some, but probably stay below 60 C. Uptime hasn't been a problem.
 
> ssh never times out. It's running stable/13-n245223-af8d48ac4b8 built
> 15th Sept.

I'm dealing with -current built October 10th. Most ssh connections to
the machine remain connected just fine. It's just those communicating
with the usb-serial adapter that tend to break off. 

Thanks for writing,

bob prohaska



[prev in list] [next in list] [prev in thread] [next in thread]