Re: June 2025 stabilization week

From: Herbert J. Skuhra <herbert_at_gojira.at>
Date: Wed, 25 Jun 2025 14:00:37 UTC 
On Tue, 24 Jun 2025 22:52:48 +0200, Michael Tuexen wrote:
> 
> > On 24. Jun 2025, at 22:08, Michael Tuexen <tuexen@freebsd.org> wrote:
> > 
> >> On 24. Jun 2025, at 21:41, Gleb Smirnoff <glebius@FreeBSD.org> wrote:
> >> 
> >> On Tue, Jun 24, 2025 at 12:28:25PM +0200, Michael Tuexen wrote:
> >> M> In /etc/devfs.conf I have
> >> M> 
> >> M> own bpf* root:network
> >> M> perm bpf* 0660
> >> M> 
> >> M> and this results in
> >> M> tuexen@epyc:~ % ls -l /dev/bpf*
> >> M> crw-rw----  1 root network 0x27 Apr 26 20:13 /dev/bpf
> >> M> lrwxr-xr-x  1 root wheel      3 Apr 26 20:13 /dev/bpf0 -> bpf
> >> M> 
> >> M> This allows members of the network group to capture traffic.
> >> M> 
> >> M> However, with main-n278173-81e6c0168d46 I get
> >> M> tuexen@epyc:~ % ls -l /dev/bpf*
> >> M> crw-------  1 root wheel 0x27 Jun 24 12:25 /dev/bpf
> >> M> lrwxr-xr-x  1 root wheel    3 Jun 24 12:25 /dev/bpf0 -> bpf
> >> M> 
> >> M> It looks like the devd did not what it did in the past.
> >> 
> >> I failed to reproduce that on fe55780e5f10, which is very
> >> close to 81e6c0168d46. I have what is expected:
> >> 
> >> crw-rw----  1 root network 0x1d 24 ÉÀÎÑ  12:37 /dev/bpf
> >> lrwxr-xr-x  1 root wheel      3 24 ÉÀÎÑ  12:37 /dev/bpf0 -> bpf
> >> 
> > I just tried the updated version of May 2025 stabweek, and that is
> > working fine. Let me test fe55780e5f10...
> > I observe this problem on an arm64 and on an amd64 system...
> fe55780e5f10 show the problem for me:
> tuexen@epyc:~ % ls -l /dev/bpf*
> crw-------  1 root wheel 0x27 Jun 24 22:48 /dev/bpf
> lrwxr-xr-x  1 root wheel    3 Jun 24 22:48 /dev/bpf0 -> bpf

4deb9760a9d8?

Reverting 7faddeb395b7 and 4deb9760a9d8 and running etcupdate seem to
resolve the issue.