Date: Sun, 31 Jul 2022 22:42:23 -0700 From: Cy Schubert <Cy.Schubert@cschubert.com> To: Mark Johnston <markj@freebsd.org>, freebsd-current@freebsd.org, chuck@freebsd.org, mhorne@FreeBSD.org, allanjude@FreeBSD.org Subject: Re: DTrace Error Message-ID: <20220801054223.9624010C@slippy.cwsent.com> In-Reply-To: <20220725153706.A2BB3A6@slippy.cwsent.com> References: <20220723035223.57CDBD7@slippy.cwsent.com> <20220723141444.85620189@slippy.cwsent.com> <YtxAbplpPtLy2Ll1@nuc> <20220723185533.9EA7D11E@slippy.cwsent.com> <20220724030857.B57FAFC@slippy.cwsent.com> <20220724170719.139BB40F@slippy.cwsent.com> <Yt62QYRLicwanWxC@nuc> <20220725153706.A2BB3A6@slippy.cwsent.com>
next in thread | previous in thread | raw e-mail | index | archive | help
In message <20220725153706.A2BB3A6@slippy.cwsent.com>, Cy Schubert writes:
> In message <Yt62QYRLicwanWxC@nuc>, Mark Johnston writes:
> > On Sun, Jul 24, 2022 at 10:07:19AM -0700, Cy Schubert wrote:
> > > In message <20220724030857.B57FAFC@slippy.cwsent.com>, Cy Schubert writes
> :
> > > > In message <20220723185533.9EA7D11E@slippy.cwsent.com>, Cy Schubert wri
> te
> > s:
> > > > > In message <YtxAbplpPtLy2Ll1@nuc>, Mark Johnston writes:
> > > > > > On Sat, Jul 23, 2022 at 07:14:44AM -0700, Cy Schubert wrote:
> > > > > > > In message <20220723035223.57CDBD7@slippy.cwsent.com>, Cy Schuber
> t
> > writ
> > > > es
> > > > > :
> > > > > > > > I'm not sure if this is because my obj tree needs a fresh rebui
> ld
> > and
> > > >
> > > > > > > > reinstall or if this is a legitimate problem. Regardless of the
> d
> > trac
> > > > e
> > > > > > > > command entered, whether it be a fbt or sdt, the following erro
> r
> > occu
> > > > rs
> > > > > :
> > > > > > > >
> > > > > > > > slippy# dtrace -n 'fbt::ieee80211_vap_setup:entry { printf("ent
> er
> > ing
> > > > > > > > ieee80211_vap_setup\n"); }'
> > > > > > > > dtrace: invalid probe specifier fbt::ieee80211_vap_setup:entry
> {
> > > > > > > > printf("entering ieee80211_vap_setup\n"); }: "/usr/lib/dtrace/p
> si
> > nfo.
> > > > d"
> > > > > ,
> > > > > > > > line 1: failed to copy type of 'pr_gid': Conflicting type is al
> re
> > ady
> > > > de
> > > > > fi
> > > > > > ned
> > > > > > > > slippy#
> > > > > > > >
> > > > > > > > Old DTrace scripts I've used months or even years ago also fail
> w
> > ith
> > > > th
> > > > > e
> > > > > > > > same error. It's not this one probe. All probes result in the p
> r_
> > gid
> > > > er
> > > > > ro
> > > > > > r.
> > > > > > > >
> > > > > > > > I'm currently rebuilding my "prod" tree from scratch with the h
> op
> > e th
> > > > at
> > > > >
> > > > > > > > it's simply something out of sync. But, should it not be, has a
> ny
> > one
> > > > el
> > > > > se
> > > > > >
> > > > > > > > encountered this lately?
> > > > > > >
> > > > > > > A full clean rebuild and installworld/kernel did not change the r
> es
> > ult.
> > > >
> > > > > > > This is a new problem.
> > > > > >
> > > > > > I don't see any such problem on a system built from commit 151abc80
> cd
> > e,
> > > > > > using GENERIC. Are you using a custom kernel config? Which kernel
> > > > > > modules do you have loaded?
> > > > >
> > > > > [...]
> > > >
> > > > chuck@ emailed me privately suggesting a roll back to cb2ae6163174. The
>
> > > > problem is fixed. I'm creating a special branch that reverts only the l
> lv
> > m
> > > > commits since then.
> > >
> > > llvm 14 is not the problem. There must be something else after cb2ae61631
> 74
> >
> > > that is causing the regression.
> >
> > Are you able to bisect? I spent a bit of time trying to replicate the
> > problem based on your kernel config, without any luck yet.
>
> How fortuitous is this email. I just rebooted my sandbox again and
> discovered this is related to non-INVARIANT kernels. Enabling INVARIANTS
> "fixes" dtrace. There must be some commit since cb2ae6163174 that affected
> non-INVARIANT kernels. As to which one, I'm not sure yet.
The commit that introduced the regression to non-INVARIANT kernels is
2449b9e5fe565be757a4b29093fd1c9c6ffcf3c9. Looking at the diff I don't
see how it caused the problem but reverting it locally addresses the
regression. (Of course one needs to disable building the mac_ddb module
in order to have the build succeed.)
Without looking at it closer, I suspect that dtrace could be sensitive to
one of the struct changes.
--
Cheers,
Cy Schubert <Cy.Schubert@cschubert.com>
FreeBSD UNIX: <cy@FreeBSD.org> Web: http://www.FreeBSD.org
NTP: <cy@nwtime.org> Web: https://nwtime.org
e**(i*pi)+1=0
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20220801054223.9624010C>