00:01 <kerneltoast> didja fixit yet

00:02 <fche> patience pedewan

00:02 <kerneltoast> SLACKER

00:02 <fche> well I never!!!

00:04 <fche> ... why on earth does that probe body even have lock code emitted into it ... I see no globals being used

00:04 <fche> just a print

00:05 <fche> and anyway in that .c file

00:06 <kerneltoast> yeah i've seen quite a few empty locks[]

00:06 <fche> ok, wild, that should be an easy optimization to add (back)

00:06 <kerneltoast> i got a more interesting backtrace for ya

00:06 <fche> but doesn't explain the problem here

00:06 <kerneltoast> static inline void _stp_runtime_entryfn_put_context(struct context *c)

00:06 <kerneltoast> if (c) {

00:06 <kerneltoast> {

00:06 <kerneltoast> + BUG_ON(!preempt_count());

00:06 <kerneltoast> if (c == _stp_runtime_get_context())

00:06 <kerneltoast> atomic_dec(&c->busy);

00:06 <kerneltoast> /* else, warn about bad state? */

00:07 <kerneltoast> https://gist.github.com/kerneltoast/9f1d409da81908f694b8f7f6df89f160

00:07 <fche> ummm wait is preempt_count() the right metric?

00:08 <fche> the stapiu_probe_handler calls at the least local_irq_save

00:08 <kerneltoast> yeah it's the right metric

00:08 <kerneltoast> stapiu_probe_handler doesn't cover the panic in the backtrace with the irq save

00:09 <kerneltoast> that occurs at the stapiu_probe_handler()

00:09 <fche> begin/end probes do not normally bother with local-irq

00:10 <fche> just preempt_disable in entryfn_get_context

00:10 <fche> in the first case, where you had put the BUGON on the get_context

00:11 <fche> isn't that block just a few lines below a preempt_disable(); call?

00:11 <kerneltoast> !preempt_count() means that preempt is *enabled*

00:11 <fche> how can that be?

00:11 <kerneltoast> the problem lies elsewhere

00:11 <kerneltoast> there's an unbalanced preempt enable

00:12 <kerneltoast> the BUG_ON(!preempt_count()); in _stp_runtime_entryfn_put_context means that preempt is about to be re-enabled despite already being enabled

00:12 <fche> right

00:13 <fche> ok yes that would explain things, but I thought the kernel warns if the preempt count goes negative

00:13 <kerneltoast> this is a 3.10 kernel

00:13 <kerneltoast> it only warns you after it panics

00:14 <kerneltoast> i bet it doesn't have the machinery to warn on preempt count going negative

00:15 <kerneltoast> fche, i added the .c and .ko artifacts to this gist: https://gist.github.com/kerneltoast/9f1d409da81908f694b8f7f6df89f160

00:28 <kerneltoast> fche, how can i make the testsuite modules get built with debug symbols?

00:37 <fche> -BCONFIG_DEBUG_INFO=y I believe

00:39 <kerneltoast> make installcheck RUNTESTFLAGS=' -BCONFIG_DEBUG_INFO=y' ?

00:39 <fche> nah, that's an stap flag

00:39 <kerneltoast> what do

01:33 <irker378> systemtap: fche systemtap.git:master * release-4.3-137-gafe0acfac / runtime/transport/relay_v2.c: transport relay_v2: drop "dropped" facility

01:33 <irker378> systemtap: fche systemtap.git:master * release-4.3-138-g7615cae79 / buildrun.cxx runtime/linux/autoconf-lockdown-debugfs.c runtime/procfs.c runtime/transport/control.c runtime/transport/debugfs.c runtime/transport/procfs.c runtime/transport/relay_v2.c runtime/transport/transport.c runtime/transport/transport.h staprun/ctl.c staprun/relay.c testsuite/buildok/trans-debugfs.stp testsuite/buildok/trans-procfs.stp: PR26665: relayfs-on-pr

01:33 <irker378> systemtap: fche systemtap.git:master * release-4.3-139-g471111339 / testsuite/buildok/floatingpoint.stp: testsuite tweak: buildok/floatingpoint.stp chmod a+x

01:46 <irker378> systemtap: fche systemtap.git:master * release-4.3-140-g1a4c75501 / : pre-release: update-docs

01:50 <kerneltoast> guess you're going forth with the release eh

01:51 <kerneltoast> with some luck maybe I'll have this fixed by the end of the week

03:06 <irker378> systemtap: fche systemtap.git:master * release-4.3-141-gf3cc67f98 / testsuite/systemtap.examples/index.html testsuite/systemtap.examples/index.txt testsuite/systemtap.examples/keyword-index.html testsuite/systemtap.examples/keyword-index.txt testsuite/systemtap.examples/metadatabase.db: pre-release: regenerate example index

03:06 <irker378> systemtap: fche systemtap.git:master * release-4.3-142-g988f439af / NEWS session.cxx systemtap.spec: pre-release: version timestamping, NEWS tweaks

03:08 <irker378> systemtap: fche systemtap.git:release-4.4 * release-4.4 / NEWS session.cxx systemtap.spec: pre-release: version timestamping, NEWS tweaks

18:15 <irker300> systemtap: alizhang systemtap.git:azhang/pr13838-f32 * release-4.3-137-gddd0f5ef6 / main.cxx runtime/softfloat.c runtime/softfloat.h runtime/softfloat/internals.h runtime/softfloat/primitives.h runtime/softfloat/specialize.h tapset/floatingpoint.stp testsuite/buildok/floatingpoint.stp: PR13838: Add float32 support and corresponding test cases

21:53 <irker300> systemtap: alizhang systemtap.git:azhang/pr13838-f32 * release-4.3-138-g59e694673 / runtime/softfloat.c tapset/floatingpoint.stp: PR13838: get rid of unused functions