Cool VL Viewer forum

Greetings!

Last Sunday I updated from 1.32.0.20 to 22 and also installed 1.32.1.5. (Also, I had my Gentoo system do its weekly updates.) Ever since I crash continuously, mostly when I am in busy sims but on empty sims too. The crashes happen with the previous versions of the viewers too and even with other viewers (so it's not CoolVLViewer and it's not the current versions), but I need help debugging. I'm attaching the logs. Any idea will help.

Thanks!

I see a lot of abnormal warnings about texture caching failures ("Failed to cache texture"), which would hint for a file system issue. Partition full ?... Failing disk ?

Also, the "Packet out of order" warnings are many and unusual, which could be the sign of a bad network or overloaded system.

There are also many "Received animation state for unknown avatar": one or two such warnings may happen, but so many of them is strange...

As for the stack trace, it points to a probable issue in the graphics driver, causing a return address corruption:
There is no way a call to an OpenGL function in radeonsi_dri.so could lead to HBFloaterAreaSearch::onClickInspect() to be called on return, so this is clearly a corrupted return address and the corruption happened in radeonsi_dri.so... Since you also get crashes with other viewers, and unless the issue is indeed with your disk being full or failing, I'd recommend you try with a different graphics driver version.

To go further, you'd need to recompile the viewer yourself from sources (so to have the viewer binary symbols file at hand; no need for a debug build: even a release build will produce that symbols file) and start the resulting viewer from the command line with ./viewer-linux-x86_64-release/newview/CoolVLViewer-x86_64-1.32.1.5/cool_vl_viewer --debug (for gdb, or --lldb for clang's lldb if you got it, since it is actually more competent than gdb), and then type 'bt' in the debugger when the viewer crashes...

Thank you! I don't think my disk is failing, because I have no issues anywhere else than second life. (I have not tried running steam, but I will.) I have tried with a self-compiled viewer (your patch about the HUDs under RLV thing) and binary downloaded by the website. I will try the --debug option and report back. I can paste you a full list of packages that Gentoo updated last Sunday onwards if you want, but there is no mesa in it, there is a kernel though (6.6.30 from 6.6.28) and cmake 3.28.5.

Verify your drive health and make a short self-test using smartctl (e.g. smartctl -a /dev/sda, smartctl -t short /dev/sda, etc) or using GSmartControl, for a more user-friendly interface...

A failing drive would also let traces in the system logs (usually held in /var/log/messages for systemd-less Linux distros, or using journalctl --system for systemd ones).

Well, you did not mention that update...

You should rather revert all the changes and verify if it fixes your issue. If yes, then apply updates to packages, one (or one correlated group) after the other, until you find the culprit...

Note that DRM is also partly implemented in the kernel, so a change to the latter could have introduced a graphics rendering related bug (or Mesa might need to be recompiled for that new kernel !): start with reverting the kernel itself (or boot under the old kernel if the update did not remove it) and see how it fares...

Both the short and the long self-tests came back good, so that's something. I followed your advice and booted with the previous kernel version and I had no crashes at all. I am attaching the last log, if you want to take a look at it. I saw no "Failed to cache texture" or "Received animation state for unknown avatar" warnings, but I saw a lot of "Packet out of order" warnings.

You got your reply... The new kernel is either buggy, or some other package in your distro needs to be recompiled for it (Mesa would be the first suspect).

The absence of the first warning (texture caching failure) could be the sign that your new kernel also had issues with file systems (in excess of the OpenGL crash bug it caused)...

This is usually the sign of a flacky or congested network, but this is not always something you can solve at your side (a bad ISP, or a congested route to the host sims could explain it as well); still, verify your network device/connection.

E.g. for an Ethernet connection, check for dropped packets or errors on the interface: ifconfig, or its equivalent for your distro, would tell you so in the interface stats numbers (and this would be the sign of a bad Ethernet connector or cable).

For a WiFi connection, it could be caused by a bad radio transmission between your PC and the WiFi router: try and move the antennas or the router itself; a "direct view" from your PC is best, of course, or as direct as it can be, with the minimum number of obstacles (walls). WiFi radio waves also got the bad habit to reflect on surfaces (e.g. doors) and inhabitants, and may cause fading as a result when the reflection interferes with the waves which followed another path (multi-path propagation interferences)... Due to the short wave length of WiFi radio signals, moving the antennas by only a few decimeters may suffice to improve the communications quality.

Thank you very much for your help!

This seems fixed with kernel 6.6.32.

It is always wise to check the kernel change log, when you get crashes or bugs just after updating the kernel (especially with AMD and Intel (i)GPUs, since their Mesa driver makes heavy use of their DRM counterparts in the Linux kernel).

Quite a few fixes went in drm/amd with kernel v6.6.31 and kernel v6.6.32...