Weird crashes since upgrading to Vivaldi 2.3



  • Since I've upgraded to 2.3 (2.3.1440.60 currently), Vivaldi often crashes. I'm not sure what causes it, but it leaves core dumps around:

    coredumpctl info
               PID: 10840 (vivaldi-bin)
               UID: 1000 (ferdl)
               GID: 1000 (ferdl)
            Signal: 11 (SEGV)
         Timestamp: Mon 2019-03-18 11:00:11 CET (5h 4min ago)
      Command Line: /opt/vivaldi/vivaldi-bin
        Executable: /opt/vivaldi/vivaldi-bin
     Control Group: /user.slice/user-1000.slice/session-2.scope
              Unit: session-2.scope
             Slice: user-1000.slice
           Session: 2
         Owner UID: 1000 (ferdl)
           Boot ID: 81d1e90217254943b29fe04bbf447d6d
        Machine ID: fa423859063e4fdbbd66dcc135370321
          Hostname: ferdi.local
           Storage: /var/lib/systemd/coredump/core.vivaldi-bin.1000.81d1e90217254943b29fe04bbf447d6d.10840.155290321>
           Message: Process 10840 (vivaldi-bin) of user 1000 dumped core.
    
                    Stack trace of thread 10840:
                    #0  0x0000560e120e2707 n/a (vivaldi-bin)
                    #1  0x0000560e0fcaedba n/a (vivaldi-bin)
                    #2  0x0000560e1218aee0 n/a (vivaldi-bin)
                    #3  0x0000560e12127e05 n/a (vivaldi-bin)
                    #4  0x0000560e0fe5dbcd n/a (vivaldi-bin)
                    #5  0x0000560e0fe7951e n/a (vivaldi-bin)
                    #6  0x0000560e0fe7a040 n/a (vivaldi-bin)
                    #7  0x0000560e0fe98646 n/a (vivaldi-bin)
                    #8  0x00007f30ad4cb7bf g_main_context_dispatch (libglib-2.0.so.0)
                    #9  0x00007f30ad4cd739 n/a (libglib-2.0.so.0)
    

    Tracing Vivaldi with strace reveals tons and tons of the following messages:

    recvmsg(31, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    recvmsg(31, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    recvmsg(35, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    recvmsg(35, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    poll([{fd=11, events=POLLIN}, {fd=12, events=POLLIN}, {fd=31, events=POLLIN}, {fd=35, events=POLLIN}], 4, 0) = 1 ([{fd=12, revents=POLLIN}])
    recvmsg(35, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    recvmsg(35, {msg_namelen=0}, 0)         = -1 EAGAIN (Resource temporarily unavailable)
    read(12, "!", 2)                        = 1
    

    Other symptoms are that it just won't repaint its window contents anymore until I manually kill off the vivaldi process.

    I'm on Arch Linux here. Anybody else experiences such crashes/hangs? Vivaldi was running very stable until recently. 😞



  • To follow up on my own post:

    After upgrading several times (first patches for 2.3, then upgrade to 2.4.1488.35-1) I am still stuck with this dreaded problem.

    Vivaldi freezes often several times a day. When in this condition, the browser will not re-paint the content of any of its windows.
    I am also unable to bring up the task manager. The only solution every time is to kill off all Vivaldi processes and just restart the thing.

    Any ideas? I'm rather close to going back to Google Chrome. Also, noone ever replied to my bug report which I submitted to bugs@vivaldi.com. Do they just ignore bug reports or what? I'd be glad to help out debugging this problem and provide further information, but I really don't know where to start.

    The only thing I can say so far that it may be connected to losses of network connectivity, i.e., when I switch from one LAN to another.
    All this basically never occured up until 2.3. The browser was running for weeks on end without any problem.



  • I get occasional freezes too, since about 2.3, but it's infrequent (once every week or two).

    But, also beginning with 2.3, Vivaldi shuts down about once a day. It's always when it's idle, a few seconds after typing or mouse activity. All processes are gone, and on restart it comes right back to where it was.

    I'm now running 2.4.1488.35 under Debian Stretch.


  • Moderator

    Black paeg and freeze? Perhaps freeze of graphic driver.
    Try to deactivate "Use of Hardware Acceleration" in Settings → Webpages

    Or a Out of Memory situation.


  • Moderator

    @ferdiG said in Weird crashes since upgrading to Vivaldi 2.3:

    noone ever replied to my bug report

    Bug number?



  • Summary: Frequent crashes upon upgrading to 2.3
    Key: VB-50555



  • @Gwen-Dragon I will try deactivating the hardware accel.

    Memory should not be a problem... computer has 32GB of RAM and my user account does not have any memory limits apart from stack size. When Vivaldi crashes hardly 10GB RAM of those 32 are even in use.

    ulimit output:
    -t: cpu time (seconds) unlimited
    -f: file size (blocks) unlimited
    -d: data seg size (kbytes) unlimited
    -s: stack size (kbytes) 8192
    -c: core file size (blocks) unlimited
    -m: resident set size (kbytes) unlimited
    -u: processes 127187
    -n: file descriptors 1024
    -l: locked-in-memory size (kbytes) 64
    -v: address space (kbytes) unlimited
    -x: file locks unlimited
    -i: pending signals 127187
    -q: bytes in POSIX msg queues 819200
    -e: max nice 0
    -r: max rt priority 0
    -N 15: unlimited



  • Turning off hardware acceleration mitigated part of the problem. However Vivaldi still occasionally crashes with signal 5, mostly during night time. Does this mean it hits some debug code? I really wonder why nobody else seems to have these sort of problems. Maybe no Vivaldi users on Linux? 🙂

               PID: 2008 (vivaldi-bin)
               UID: 1000 (ferdl)
               GID: 1000 (ferdl)
            Signal: 5 (TRAP)
         Timestamp: Tue 2019-04-09 03:51:08 CEST (6h ago)
      Command Line: /opt/vivaldi/vivaldi-bin
        Executable: /opt/vivaldi/vivaldi-bin
     Control Group: /user.slice/user-1000.slice/session-2.scope
              Unit: session-2.scope
             Slice: user-1000.slice
           Session: 2
         Owner UID: 1000 (ferdl)
           Boot ID: 704038f036e34c78981009cc27a23485
        Machine ID: fa423859063e4fdbbd66dcc135370321
          Hostname: ferdi.local
           Storage: /var/lib/systemd/coredump/core.vivaldi-bin.1000.704038f036e34c78981009cc27a23485.2008.1554774668000000.lz4 (truncated)
           Message: Process 2008 (vivaldi-bin) of user 1000 dumped core.
                    
                    Stack trace of thread 2034:
                    #0  0x0000557ecff130f1 n/a (/opt/vivaldi/vivaldi-bin)

  • Moderator

    May be a Out of Memory crash?


  • Vivaldi Team

    We have not yet implemented proper crashlogging support on Linux (yet). Nonetheless you can use GDB (GNU Project Debugger) to capture a backtrace when Vivaldi crashes on Linux. Sadly, it requires a special binary with symbols that exactly matches the version your are crashing. Upgrade to the latest stable version 2.4 (1488.36) if you have not already and reproduce the crash there.

    Here is how the process basically works.

    unxz vivaldi-2.4.1488.36-x64-debug.xz
    
    • Installing the debug binary
    sudo install -m755 vivaldi-2.4.1488.36-x64-debug /opt/vivaldi/vivaldi-2.4.1488.36-debug
    
    • Capturing a full crash

    To capture a crash in our UI (full browser crash) issue the following in a terminal window to start Vivaldi under GDB (Note: this requires gdb to be installed):

    gdb -ex run --args /opt/vivaldi/vivaldi-2.4.1488.36-debug
    

    Now repeat the procedure that you usually use to crash the browser. Once Vivaldi has crashed you can issue the following in the GDB console to obtain a backtrace of the crash

    bt
    

    Copy the entire backtrace to your clipboard and paste it here or send it along in a bug report.


  • Vivaldi Team

    @ferdiG said in Weird crashes since upgrading to Vivaldi 2.3:

    I really wonder why nobody else seems to have these sort of problems. Maybe no Vivaldi users on Linux?

    Actually we have hundreds of thousands of users on Linux. It probably just isn't a common crash but with a backtrace I am sure we will get to the bottom of it. Thanks for the reports thus far!



  • @ferdiG said in Weird crashes since upgrading to Vivaldi 2.3:

    I really wonder why nobody else seems to have these sort of problems. Maybe no Vivaldi users on Linux? 🙂

    Ever since about 2.3 I've had fairly frequent random crashes. I always use the snapshot channel, and I remember it being worse early on, to the point where I'd have multiple crashes in a day. For me it is usually when closing a bunch of tabs in succession, and it seems like it's only on Amazon tabs. But it's mostly better now crashes when closing a bunch of tabs is better (but not completely gone).

    Occasionally (like maybe once a week) Vivaldi will totally freeze up and become unresponsive (the page is still displayed, but I cannot scroll, close the tab, or switch to another tab). I have to kill the Vivaldi process and restart. I haven't found any pattern on these crashes.


  • Vivaldi Team

    @ugly want to try getting a backtrace with the steps I posted previously?



  • @Gwen-Dragon said in Weird crashes since upgrading to Vivaldi 2.3:

    May be a Out of Memory crash?

    I don't think so. There are no memory limitations on my user, and the machine has 32GB out of which hardly 10 are every used.


  • Moderator

    @ferdiG Wait until the Linux devs have analyzed your sent debug logs.



  • @ruario Thank you for the instructions. I installed the x64 binary and I am running it now under gdb. Will post results once I experience the next crash.


  • Moderator

    @ferdiG said in Weird crashes since upgrading to Vivaldi 2.3:

    Will post results

    Please, not do post here, send them to Vivaldi's bugreport tracker so devs can check.
    Please read how to Report a bug for Vivaldi carefully and then report the bug to Vivaldi bugtracker.



  • @Gwen-Dragon said in Weird crashes since upgrading to Vivaldi 2.3:

    Please read how to Report a bug for Vivaldi carefully and then report the bug to Vivaldi bugtracker.

    Just FYI, I did this several weeks ago without ever receiving an answer. I would have happily supplied them with debug data, but I never received a reply to my bug report. That's why I started posting on this forum in the first place!


  • Moderator

    @ferdiG Give me the bug numbers, please.



  • @Gwen-Dragon
    there you go:

             Summary: Frequent crashes upon upgrading to 2.3
                 Key: VB-50555
             Project: Vivaldi Browser
    
         Environment: Vivaldi version: 2.3.1440.60

 

Looks like your connection to Vivaldi Forum was lost, please wait while we try to reconnect.