Rage Inducing Bugs
I have run into a lot of bugs lately. Maybe it’s actually a normal amount, but these bugs, especially taken together, have caused me quite a bit of rage. Writing is an outlet for me and at the very least you can all enjoy the show, so here goes!
X11 Text Thing
I tweeted about this one a few days ago. The gist is that, sometimes, when going back and forth from suspend, font data in video memory gets corrupted. I have a theory that it has to do with switching from X11 to DRI (the old school TTYs), but it is not the most reproducable thing in the world, so this is where I’ve had to leave it.
I reported a bug against Firefox recently about overlay windows not getting shown. There is a workaround for this, and the Firefox team (or at least a Firefox individual, Karl Tomlinson) has been willing to look at my errors and have a dialog with me. I have a sinking feeling that this could be a kernel driver bug or maybe a GTK3 bug, but I have no idea how to verify that.
Vim has been crashing on my computer for a while now. I turned on coredumps for
gvim only so that I could repro it about three weeks ago, and I finally got the
core yesterday. I dutifully inspected the core dump with
Program terminated with signal SIGBUS, Bus error. #0 0x00007fa650515757 in ?? ()
Worthless. I knew I needed debugging symbols, but it turns out there is no
vim-dbg. A while ago (like, eight years ago) Debian (and thus Ubuntu)
started storing debugging symbols in a completely separate repository.
Thankfully a Debian developer, Niels Thykier, was kind enough to point this out
to me that I was able to install the debugging symbols. If you want to do that
yourself you can follow instructions
here, but I have to warn you, you
will get errors, because I don’t think Ubuntu has really put much effort into
this working well.
After installing the debugging symbols I got this much more useful backtrace:
#0 0x00007fa650515757 in kill () at ../sysdeps/unix/syscall-template.S:84 #1 0x0000555fad98c273 in may_core_dump () at os_unix.c:3297 #2 0x0000555fad98dd20 in may_core_dump () at os_unix.c:3266 #3 mch_exit (r=1) at os_unix.c:3263 #4 <signal handler called> #5 in_id_list (cur_si=<optimized out>, [email protected]=0x555fb0591700, list=0x6578655f3931313e, [email protected]=0x555faf7497a0, contained=0) at syntax.c:6193 #6 0x0000555fad9fb902 in syn_current_attr ([email protected]=0, [email protected]=0, [email protected]=0x0, [email protected]=0) at syntax.c:2090 #7 0x0000555fad9fc1b4 in syn_finish_line ([email protected]=0) at syntax.c:1781 #8 0x0000555fad9fcd3f in syn_finish_line (syncing=0) at syntax.c:758 #9 syntax_start (wp=0x555faf633720, lnum=3250) at syntax.c:536 #10 0x0000555fad9fcf45 in syn_get_foldlevel (wp=0x555faf633720, [email protected]=3250) at syntax.c:6546 #11 0x0000555fad9167e9 in foldlevelSyntax (flp=0x7ffe2b90beb0) at fold.c:3222 #12 0x0000555fad917fe8 in foldUpdateIEMSRecurse ([email protected]=0x555faf633828, [email protected]=1, [email protected]=1, [email protected]=0x7ffe2b90beb0, [email protected]=0x555fad9167a0 <foldlevelSyntax>, [email protected]=7532, topflags=2) at fold.c:2652 #13 0x0000555fad918dbf in foldUpdateIEMS (bot=7532, top=1, wp=0x555faf633720) at fold.c:2292 #14 foldUpdate ([email protected]=0x555faf633720, [email protected]=1, [email protected]=2147483647) at fold.c:835 #15 0x0000555fad919123 in checkupdate ([email protected]=0x555faf633720) at fold.c:1187 #16 0x0000555fad91936a in checkupdate (wp=0x555faf633720) at fold.c:217 #17 hasFoldingWin (win=0x555faf633720, lnum=5591, firstp=0x555faf633798, [email protected]=0x0, [email protected]=1, [email protected]=0x0) at fold.c:158 #18 0x0000555fad91942e in hasFolding (lnum=<optimized out>, firstp=<optimized out>, [email protected]=0x0) at fold.c:133 #19 0x0000555fad959c3e in update_topline () at move.c:291 #20 0x0000555fad9118ee in buf_reload ([email protected]=0x555faf25e210, [email protected]=33204) at fileio.c:7155 #21 0x0000555fad911d0c in buf_check_timestamp ([email protected]=0x555faf25e210, [email protected]=1) at fileio.c:6997 #22 0x0000555fad912422 in check_timestamps (focus=1) at fileio.c:6664 #23 0x0000555fada1091b in ui_focus_change (in_focus=<optimized out>) at ui.c:3203 #24 0x0000555fad91fd96 in vgetc () at getchar.c:1670 #25 0x0000555fad920019 in safe_vgetc () at getchar.c:1801 #26 0x0000555fad96e775 in normal_cmd (oap=0x7ffe2b90c440, toplevel=1) at normal.c:627 #27 0x0000555fada5d665 in main_loop (cmdwin=0, noexmode=0) at main.c:1359 #28 0x0000555fad88d21d in main (argc=<optimized out>, argv=<optimized out>) at main.c:1051
Linux Kernel Bugs
I found a bug in the Linux Kernel, probably related to the nvidia drivers, but I’m not totally sure. I’d love for this to get resolved, though reporting kernel bugs to Ubuntu has not gone well for me in the past.
The kernel bug above causes the computer to crash during xrandr events; this means that I end up with vim writing a fresh new session file during the event (thanks to the excellent Obsession by Tim Pope) and the session file getting hopelessly corrupted, because it fails midwrite.
I foolishly mentioned this on the #vim channel on freenode and was reminded how often IRC channels are actually unrelated to aptitude. The people in the channel seemed to think that if the kernel crashes, there is nothing that can be done by a program to avoid losing data. I will argue that while it is hard, it is not impossible. The most basic thing that can and should be done is:
- Write to a tempfile
- Rename the tempfile to the final file
I will likely submit the above as a proper bug to the Vim team tomorrow. In the meantime this must also be done in Obsession, and I have submitted a small patch to do what I outlined above. I’m battle testing it now and will know soon if it resolves the problem.
I feel better. At the very least I’ve submitted bugs, and in one of the most annoying cases, been able to submit a patch. When you run into a bug, why not do the maintainer a solid and report it? And if you can, fix it!Posted Tue, May 10, 2016