fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged
<sscox>
kerneltoast fche and I were chatting about it. (good find!)
khaled has quit [Quit: Konversation terminated!]
hpt has joined #systemtap
khaled has joined #systemtap
<kerneltoast>
sscox, this commit has been a thorn in our sides... the first commit broke atvar functionality for non-PIE binaries and i fixed it with a subsequent commit
<kerneltoast>
but the subsequent commit still left TLS broken
<kerneltoast>
the original commit was made to fix vma functionality for recentish php binaries
<kerneltoast>
because of a compiler optimization that php began using
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #systemtap
fdalleau_away is now known as fdalleau
orivej has quit [Ping timeout: 252 seconds]
LW has joined #systemtap
hpt has quit [Ping timeout: 252 seconds]
hpt has joined #systemtap
hpt has quit [Ping timeout: 252 seconds]
orivej has joined #systemtap
orivej has quit [Ping timeout: 265 seconds]
tromey has joined #systemtap
amerey has joined #systemtap
orivej has joined #systemtap
irker751 has joined #systemtap
<irker751>
systemtap: smakarov systemtap.git:master * release-4.5-1-g77abac91d / NEWS configure.ac doc/SystemTap_Beginners_Guide/en-US/Book_Info.xml systemtap.spec testsuite/configure.ac: version bumps for the next release-4.6
<irker751>
systemtap: smakarov systemtap.git:master * release-4.5-2-ge8f5c058f / configure testsuite/configure: autoconf: version bumps for the next release-4.6
dxu has quit [Ping timeout: 248 seconds]
fdalleau is now known as fdalleau_away
tromey has quit [Quit: ERC (IRC client for Emacs 27.1)]
dxu has joined #systemtap
<sscox>
kerneltoast this extra condition? Yep seems to fix it; running the testsuite with it: if (res == -ESRCH /* || vm_start + offset == addr */)
<kerneltoast>
sscox, yeah
<kerneltoast>
but that's needed to fix vma tracking with php
<kerneltoast>
it's a conundrum
<fche>
maybe need a different solution for php.
<sscox>
oh urgh, gotcha
<kerneltoast>
if (!strcmp(current->comm, "php"))
<kerneltoast>
sscox, i can send you the nasty reproducers for the various cases we need to consider for this game of whack-a-mole
<fche>
pretty sure scox would run screaming from this part of the code, I don't think he'd had occasion to touch it before
<kerneltoast>
ah, just like how i ran screaming from tapset.cxx
<fche>
yes but scox runs really well
<kerneltoast>
unlike this part of the code
<fche>
hey now
<kerneltoast>
;_;
<kerneltoast>
fche, btw how big of an issue is the TLS breakage since 4.5 just came out
<fche>
it's not great, pretty sure we'll end up backporting fixes for it
<fche>
into our distro builds
<kerneltoast>
foiled by php
LW has quit [Read error: Connection reset by peer]
LW has joined #systemtap
orivej has quit [Ping timeout: 260 seconds]
<sscox>
can any reproducers be molded into testcases for the testsuite?
<fche>
for the problematic php binary? yeah I think we figured out what it was -- a funky linker option to make page sizes very large
<fche>
so we can syntesize other binaries to trigger the problems
<fche>
but
<fche>
still not sure we really Have to
<sscox>
aha, gotcha
<sscox>
glanced at vma.c when I added user space tracking which was thankfully way easier