fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged
<fche>
surprised that in this case, out_fd is not simply initialized to avail_cpus[0]'s value
<fche>
ie., fix this flow at the data level rather than control flow changes here
<kerneltoast>
yea...
<kerneltoast>
we could
<kerneltoast>
but then our friends with a zillion cpus will have that big array initialized
<fche>
bet all this is MAX_NR_CPU based in any case
<fche>
s/based/bounded/
<fche>
eek I see this out_fd[avail_cpus[...]] thing all over, yuck
<fche>
well anyway, thanks, looks plausible, if it works for you, ship it
<fche>
did one of the tests catch this?
<kerneltoast>
i didn't notice it from the testsuite, but our lean testsuite vomits without this
<fche>
barf
<kerneltoast>
ERROR: Couldn't write to output 0 for cpu 14, exiting.: Bad file descriptor
<fche>
we'd welcome little testsuite extension patches too
<fche>
though I recall agentzh had a whole other framework
<kerneltoast>
ain't nobody got time fo dat
<kerneltoast>
yeah ours is perl based
<kerneltoast>
it is one speedy mcqueen
<irker714>
systemtap: sultan systemtap.git:master * release-4.4-38-gb4b5a29b5 / staprun/relay.c: staprun: use the correct out_fd when bulkmode and fsize_max aren't used
<fche>
otherwise things stable for y'all now?
<kerneltoast>
i think so, i'm currently cleaning up our testsuite gore
<kerneltoast>
i suspect the rest of our testsuite gore is just tests needing to be updated
<kerneltoast>
or me being bad at merging
<kerneltoast>
that patch i just pushed cleaned up most of it
derek0883 has quit [Remote host closed the connection]
derek0883 has joined #systemtap
derek0883 has quit [Remote host closed the connection]
tromey has quit [Quit: ERC (IRC client for Emacs 27.1)]
<kerneltoast>
fche, i got a question about one of your comments
<fche>
no comment
<kerneltoast>
k
<kerneltoast>
thanks
<kerneltoast>
if (s.verbose > 3) // suppress in PR25841 context
<kerneltoast>
what's this mean in elaborate.cxx
<kerneltoast>
could you please elaborate
<kerneltoast>
.cxx
<fche>
I expect this warning message in that context is not useful to see, probably redundant with another run of of a similar optimization pass
derek0883 has joined #systemtap
<fche>
turn on stap --vp 04000 and see the extra messages that are printed
<kerneltoast>
ah ok
<kerneltoast>
i saw that comment and had flashbacks to the print driver
<kerneltoast>
"what do you mean you can't print in this context?? is stap running inside an NMI????? what is even real anymore"
<fche>
hahaha
<fche>
userspace and print_warning() is quite mild in comparison
<fche>
FEAR NOT
<fche>
fear is the mind killer
<kerneltoast>
"We have nothing to fear but death and taxes." -Jenbamin Franksevelt