Date: Thu, 17 Nov 2011 22:05:29 +0100 From: Attilio Rao <attilio@freebsd.org> To: mdf@freebsd.org Cc: Kostik Belousov <kostikbel@gmail.com>, Alexander Motin <mav@freebsd.org>, freebsd-current@freebsd.org, Andriy Gapon <avg@freebsd.org> Subject: Re: Stop scheduler on panic Message-ID: <CAJ-FndDLTpognq6VTZS514XY%2Beh12LrW=XSoS5=6eF3T7KApUg@mail.gmail.com> In-Reply-To: <CAMBSHm-jne0qFb5A9ua1N_KOuc8O9e=pjX7_iwMG2dODzpy%2B_A@mail.gmail.com> References: <20111113083215.GV50300@deviant.kiev.zoral.com.ua> <201111171137.18663.jhb@freebsd.org> <4EC53D1B.4000308@FreeBSD.org> <201111171409.37629.jhb@freebsd.org> <4EC563BB.60209@FreeBSD.org> <CAJ-FndB7-UeFVQhqe0sTnpJ2PhWO5ijaCLaE-6bzMU%2B8=gYXeg@mail.gmail.com> <CAMBSHm-jne0qFb5A9ua1N_KOuc8O9e=pjX7_iwMG2dODzpy%2B_A@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
2011/11/17 <mdf@freebsd.org>: > On Thu, Nov 17, 2011 at 12:54 PM, Attilio Rao <attilio@freebsd.org> wrote= : >> 2011/11/17 Andriy Gapon <avg@freebsd.org>: >>> BTW, it is my opinion that we really should not let the debugger code c= all >>> mi_switch for any reason. >> >> Yes, I agree with this, this is why the sched_bind() in boot() is >> broken (immagine calling things like doadump from KDB. KDB right now >> can be thought as a first cut of this patch because it does disable >> the CPUs when entering the context, thus, the bug here is that if you >> stop all CPUs including CPU0 and later on you want bind on it you are >> death). > > Another patch related to this area we have at $WORK: > > =C2=A0#if defined(SMP) > - =C2=A0 =C2=A0 =C2=A0 /* > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* Bind us to CPU 0 so that all shutdown code= runs there. =C2=A0Some > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* systems don't shutdown properly (i.e., ACP= I power off) if we > - =C2=A0 =C2=A0 =C2=A0 =C2=A0* run on another processor. > - =C2=A0 =C2=A0 =C2=A0 =C2=A0*/ > - =C2=A0 =C2=A0 =C2=A0 thread_lock(curthread); > - =C2=A0 =C2=A0 =C2=A0 sched_bind(curthread, 0); > - =C2=A0 =C2=A0 =C2=A0 thread_unlock(curthread); > - =C2=A0 =C2=A0 =C2=A0 KASSERT(PCPU_GET(cpuid) =3D=3D 0, ("%s: not runnin= g on cpu 0", __func__)); > + =C2=A0 =C2=A0 =C2=A0 /* > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* sched_bind can't be done reliably inside o= f panic. =C2=A0cpu_reset() will > + =C2=A0 =C2=A0 =C2=A0 =C2=A0* rebind us in any case, more reliably. > + =C2=A0 =C2=A0 =C2=A0 =C2=A0*/ > + =C2=A0 =C2=A0 =C2=A0 if (panicstr =3D=3D NULL) { > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 /* > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* Bind us to CPU= 0 so that all shutdown code runs there. =C2=A0Some > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* systems don't = shutdown properly (i.e., ACPI power off) if we > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0* run on another= processor. > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0*/ > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 thread_lock(curthread)= ; > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 sched_bind(curthread, = 0); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 thread_unlock(curthrea= d); > + =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 KASSERT(PCPU_GET(cpuid= ) =3D=3D 0, ("boot: not running on cpu 0")); > + =C2=A0 =C2=A0 =C2=A0 } > =C2=A0#endif > =C2=A0 =C2=A0 =C2=A0 =C2=A0/* We're in the process of rebooting. */ > =C2=A0 =C2=A0 =C2=A0 =C2=A0rebooting =3D 1; This doesn't cover the KDB case which is the most broken here. (I'm a bit unsure about the name of functions and I cannot check now, but in short): - you enter KDB via debug.kdb.enter=3D1 (for example) - kdb_enter() stop CPUs and if it is on CPU1 it stops CPU0 - you call functions entering boot() from KDB prompt (IIRC "call doadump" should do it) - boot() wants to bind on CPU0 which is turned off This case only take care of panic, which is not enough. Attilio --=20 Peace can only be achieved by understanding - A. Einstein
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-FndDLTpognq6VTZS514XY%2Beh12LrW=XSoS5=6eF3T7KApUg>