<div dir="ltr"><div><br><div class="gmail_extra"><br><div class="gmail_quote">On 30 July 2017 at 21:54, dequis <span dir="ltr"><<a href="mailto:dx@dxzone.com.ar" target="_blank">dx@dxzone.com.ar</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr">[resending this without the attachment, sorry if it ends up being posted twice]<br><div><div class="gmail_extra"><br><div class="gmail_quote"><span class="gmail-">On 9 March 2017 at 17:46, dequis <span dir="ltr"><<a href="mailto:dx@dxzone.com.ar" target="_blank">dx@dxzone.com.ar</a>></span> wrote:<br></span><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div><span class="gmail-">Hi.<br>
<br>
I got a crash with anki (git version with pyqt5 instead of 4). I'm not<br></span>
sure how to reproduce it, but [snip]<br></div></blockquote><span class="gmail-"><div><br></div><div>Hey there, it's me again with the anki
crashes. Still happening with pyqt 5.9. I took a break from using the
anki desktop app for a few months but now it's time to deal with it
again.</div></span></div><span class="gmail-"><div><br></div><div>I still don't know how to intentionally
reproduce it other than "just use the app normally for a while", but
using the app normally for a while works (Which I'd totally recommend
since anki is an excellent app, but that's not a very reliable way to
reproduce it). Takes hours to reproduce but looks like it happens
roughly once every 12-48 hours.<br></div><div><br></div><div>So I prepared a bit better this time:</div><div><br></div><div>- got debug symbols for everything (thanks the-compiler for the repo!)</div><div>- patched the sip build scripts to not strip on 'make install'<br></div><div>- installed the excellent python gdb extensions</div><div>- replaced malloc with a tcmalloc_debug to make it crash more reliably</div><div>- set PYTHONMALLOC=malloc</div><div>- and ran the whole thing under rr</div><div><br></div><div>Way
better than valgrind, since I have time travel now, and I can replay
this as many times as I want. I gave it a shot to try to extract as much
info as I could.<br></div><div><br></div></span><div>Here's the annotated gdb/rr session:</div><div><br></div><div><a href="http://dump.dequis.org/G21sm.txt" target="_blank">http://dump.dequis.org/G21sm.<wbr>txt</a><br></div><div><br></div><div> And here's what I learnt:</div><span class="gmail-"><div><br></div><div>-
The object being freed is EditCurrent, a subclass of QDialog (I think
it's the dialog opened from the edit button during a review)<br></div>- The free happens during garbage collection because it needs to break a reference cycle between EditCurrent and Editor<div class="gmail_quote">- Some interesting interactions with the code that calls javascript to do "saveNow"</div><div class="gmail_quote"><br></div><div class="gmail_quote">It
got hairy at some point and I didn't reach the initial allocation of
the object - lots of incref/decref in code related to saveNow. I'll
continue later.<br></div><div class="gmail_quote"><br></div><div class="gmail_quote">Also worth noting that i'm using a slightly old git revision of anki,
43a662a installed april 15. Didn't want to upgrade just in case the bug
stopped happening.</div><div class="gmail_quote"><br></div><div class="gmail_quote">One recent anki commit caught my attention, "fix duplicate
constructor call in editcurrent", three days ago, removes a duplicate call to "<span class="gmail-m_7728960927058748368gmail-m_7386095962584855335gmail-blob-code-inner">QDialog.<span class="gmail-m_7728960927058748368gmail-m_7386095962584855335gmail-pl-c1">__init__</span></span>". Who knows if it's relevant. It takes forever to find out so I'd rather stay with what I have.</div><div class="gmail_quote"><br></div>Any
suggestions on how to continue debugging this would be appreciated. I
got some new ideas on how to reproduce it, but nothing seems to work so
far.</span></div></div></div>
</blockquote></div></div><br>Okay, so, the good news is that I got decent repro steps now.<br><br>The bad (but also good) news is that this is anki's fault and it was indeed fixed by removing that duplicate constructor call.<br><br><a href="https://github.com/dae/anki/commit/5ef1692c781356c23d7f3c1fecbb61cb57f684cb">https://github.com/dae/anki/commit/5ef1692c781356c23d7f3c1fecbb61cb57f684cb</a><br><br>I'm not sure where you normally set the boundary of where to blame the application, but I hope this is some kind of misbehavior pyqt can detect and fail early instead of messing up reference counts.<br><br>Requirements:<br><br>- Python 3.6<br>- Anki 2.1.0beta4 or current git with the commit above reverted.<br>- Have at least one deck with one card ("add" in the main window)<br>- Optional: gperftools or an equivalent package containing tcmalloc (other mallocs are less reliable for debugging)<br><br>Procedure:<br><br>1. Start anki with:<br><br>$ LD_PRELOAD=/usr/lib/libtcmalloc_debug.so PYTHONMALLOC=malloc python3 /usr/bin/anki<br><br>Or using glibc's malloc, which is less reliable:<br><br>$ MALLOC_PERTURB_=255 PYTHONMALLOC=malloc python3 /usr/bin/anki<br><br>2. Click the deck name<br>3. Click study now<br>4. Click edit (bottom left)<br>5. Close the edit window<br>6. Press ctrl+: (colon)<br>7. Enter "gc.collect()"<br>8. Press ctrl+enter<br>9. Click edit again<br>10. If using MALLOC_PERTURB_ and it doesn't crash, repeat from step 5 a couple of times.</div></div>