Date: Sun, 10 Feb 2013 02:29:39 +0400 From: Andrey Zonov <zont@FreeBSD.org> To: Matt Burke <mattblists@icritical.com> Cc: freebsd-current@FreeBSD.org Subject: Re: [patch] Userland DTrace Message-ID: <5116CDD3.4000503@FreeBSD.org> In-Reply-To: <51152216.9080905@icritical.com> References: <51152216.9080905@icritical.com>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] On 2/8/13 8:04 PM, Matt Burke wrote: > I've been spending some time trying to get the fasttrap provider to work > on FreeBSD without panicing. I believe I have succeeded, at least to the > point where it's no longer panicing. > > There were two panic causes. The first was > http://www.freebsd.org/cgi/query-pr.cgi?pr=165541 - the FreeBSD port of > fasttrap.c caused ftp_rcount to be left >0. To fix this I've got rid of > the early return and reverted to the opensolaris way. > > A second panic then showed up intermittently when fasttrap_pid_cleanup_cb > was run while something in userland had locks. Using sx_try_xlock calls > has stopped the panics and shouldn't affect operation AFAICT. > > This is against r246454. > > > Although this has fixed the panics for me, I'm finding a lot of stuff just > isn't actually working, with dtrace and the traced process just chewing > CPU. Truss on the dtrace shows a heck of a lot of ptrace() calls and I > have no idea what the target is doing... CPU time is split 2:1 > dtrace:target > Great! This fixes panics for me too, but I still cannot get something useful tracing and after detaching by ctrl+c my programs still segfaults. Please look at one style comment below. > > Also noteworthy is the LOR on the first time you try to use the fasttrap > provider: http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/165479 > > The lock order there seems right, so I'm guessing "something else" must > have done it wrong first? How can I find out what the "something else" > is? > > > Thanks > > > --- a/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c > +++ b/sys/cddl/contrib/opensolaris/uts/common/dtrace/dtrace.c > @@ -7536,9 +7536,23 @@ dtrace_unregister(dtrace_provider_id_t id) > return (EBUSY); > } > } else { > +#if defined(sun) > mutex_enter(&dtrace_provider_lock); > mutex_enter(&mod_lock); > mutex_enter(&dtrace_lock); > +#else > + if (sx_try_xlock(&dtrace_provider_lock) == 0) s/sx_try_xlock/mutex_tryenter/ Look at sys/cddl/compat/opensolaris/sys/mutex.h > + return (EBUSY); > + if (sx_try_xlock(&mod_lock) == 0) { > + mutex_exit(&dtrace_provider_lock); > + return (EBUSY); > + } > + if (sx_try_xlock(&dtrace_lock) == 0) { > + mutex_exit(&mod_lock); > + mutex_exit(&dtrace_provider_lock); > + return (EBUSY); > + } > +#endif > } > > /* > --- a/sys/cddl/contrib/opensolaris/uts/common/dtrace/fasttrap.c > +++ b/sys/cddl/contrib/opensolaris/uts/common/dtrace/fasttrap.c > @@ -1116,23 +1116,28 @@ fasttrap_pid_disable(void *arg, dtrace_id_t id, void *parg) > > ASSERT(id == probe->ftp_id); > > - mutex_enter(&provider->ftp_mtx); > - > /* > * We won't be able to acquire a /proc-esque lock on the process > * iff the process is dead and gone. In this case, we rely on the > * provider lock as a point of mutual exclusion to prevent other > * DTrace consumers from disabling this probe. > */ > - if ((p = pfind(probe->ftp_pid)) == NULL) { > - mutex_exit(&provider->ftp_mtx); > - return; > + > +#if defined(sun) > + if ((p = sprlock(probe->ftp_pid)) != NULL) { > + ASSERT(!(p->p_flag & SVFORK)); > + mutex_exit(&p->p_lock); > + } > +#else > + if ((p = pfind(probe->ftp_pid)) != NULL) { > + _PHOLD(p); > + PROC_UNLOCK(p); > } > -#ifdef __FreeBSD__ > - _PHOLD(p); > - PROC_UNLOCK(p); > #endif > > + mutex_enter(&provider->ftp_mtx); > + > + > /* > * Disable all the associated tracepoints (for fully enabled probes). > */ > @@ -1154,6 +1159,13 @@ fasttrap_pid_disable(void *arg, dtrace_id_t id, void *parg) > if (provider->ftp_retired && !provider->ftp_marked) > whack = provider->ftp_marked = 1; > mutex_exit(&provider->ftp_mtx); > + > +#if defined(sun) > + mutex_enter(&p->p_lock); > + sprunlock(p); > +#else > + PRELE(p); > +#endif > } else { > /* > * If the process is dead, we're just waiting for the > @@ -1167,9 +1179,6 @@ fasttrap_pid_disable(void *arg, dtrace_id_t id, void *parg) > if (whack) > fasttrap_pid_cleanup(); > > -#ifdef __FreeBSD__ > - PRELE(p); > -#endif > if (!probe->ftp_enabled) > return; > > -- Andrey Zonov [-- Attachment #2 --] -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.18 (Darwin) Comment: GPGTools - http://gpgtools.org iQEcBAEBAgAGBQJRFs3WAAoJEBWLemxX/CvT8jMH/0gNWX0R9dTzGVy20Ii0uRLT JoH4AuS8AaGZ+K4TP4BPPZ38ARnUfZMeM7jwzqCiXBuERgVa6LWC0k/uFkPIboCG Uwr0TMq9pCSGl+HqNVBG0gX2rHagh2UbCfl+xCAAnHpdWgO8sdB9tVfj+UEPGJDp eSsVwTQtSegqpR+t7VabBeBFhBs9TbmT8mvhLWd25JK4bSI+gGNrkhgRqvKZOi4b 5i41wLKXnBQhoPQYh6hazLFwXnYJU+ZK3HW35Ezix4kxPwrnzib++QlPOKGCpsWr 4rWuv94xzdAE2KVEG8a2ZGVznoJ1nrTBl0CSeJdazBnLard036E7X/XcR8mpoDE= =FQRp -----END PGP SIGNATURE-----
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5116CDD3.4000503>