Default viewer refusing to load on Windows 10

[nogaku]

Just did a clean install of the latest version of the default viewer downloaded from the official site. When I double click the second life viewer icon on the desktop it will pop up a "Loading Second Life" window that very quickly disappears. then the entire program exits without producing any error. I have tried restarting my computer, reinstalling second life, and updating my drivers, and none of this helped. I am running Windows 10.

here are what the last lines of the second life log look like:

Quote

2019-04-23T12:03:01Z INFO #InitInfo# newview/llappviewer.cpp(799) LLAppViewer::init : Configuration initialized.
2019-04-23T12:03:01Z WARNING # llxml/llcontrol.h(260) LLControlGroup::get : Control MemProfiling not found.
2019-04-23T12:03:01Z INFO #Init# newview/llappcorehttp.cpp(389) LLAppCoreHttp::refreshSettings : HTTP Pipelining enabled!
2019-04-23T12:03:01Z INFO #InitInfo# newview/llappviewer.cpp(827) LLAppViewer::init : LLCore::Http initialized.
2019-04-23T12:03:01Z INFO #AppInit# newview/llmachineid.cpp(220) LLMachineID::init :  Serial Number : 0000000000000000

Any ideas would be greatly appreciated.

[Whirly Fizzle]

2019-04-23T12:03:01Z INFO #AppInit# newview/llmachineid.cpp(220) LLMachineID::init :  Serial Number : 0000000000000000 

Did you zero out the serial number or is that what actually printed in the log?
Did this problem only start when you did the clean viewer install or did you do that to try & fix the problem?

If I were you I'd file a JIRA issue for this so the Lindens can get the full logs & look up your crash reports etc.
https://community.secondlife.com/knowledgebase/english/how-to-report-a-bug-r224/

[nogaku]

35 minutes ago, Whirly Fizzle said:

2019-04-23T12:03:01Z INFO #AppInit# newview/llmachineid.cpp(220) LLMachineID::init :  Serial Number : 0000000000000000 

Did you zero out the serial number or is that what actually printed in the log?
Did this problem only start when you did the clean viewer install or did you do that to try & fix the problem?

If I were you I'd file a JIRA issue for this so the Lindens can get the full logs & look up your crash reports etc.
https://community.secondlife.com/knowledgebase/english/how-to-report-a-bug-r224/

no, the serial number is like that in the logs

and this is my first install on this machine and it's always done this.

[Whirly Fizzle]

I have to ask - is your copy of Windows legit? The serial number shouldn't be all zeros.

Are you able to get another viewer launched, for example Firestorm?  Does Firestorm show a different Machine ID/Serial in the log?
If Firestorm also crashes to desktop, in the logs folder you should see a dump folder that's kept for any saved session. Inside the dump folder, is there a dmp file?
You won't get this on the LL viewer because they use Bugsplat now for crash reporting.

If there's a dmp file generated for the crash I can get a crash stack for you.

Edited April 23, 2019 by Whirly Fizzle

[nogaku]

I am running a legitimate install of Windows 10 upgraded from Windows 7 ages ago.

Here are the last few lines of Firestorm.log

Quote

2019-04-23T15:55:35Z INFO #InitInfo#  newview/llappviewer.cpp(953) LLAppViewer::init : Configuration initialized.
2019-04-23T15:55:35Z INFO #Init#  newview/llappcorehttp.cpp(414) LLAppCoreHttp::refreshSettings : HTTP Pipelining enabled!
2019-04-23T15:55:35Z INFO #Init#  newview/llappcorehttp.cpp(537) LLAppCoreHttp::refreshSettings : Application settings overriding default mesh fetch concurrency.  New value:  16
2019-04-23T15:55:35Z INFO #InitInfo#  newview/llappviewer.cpp(989) LLAppViewer::init : LLCore::Http initialized.
2019-04-23T15:55:35Z INFO #AppInit#  newview/llmachineid.cpp(220) LLMachineID::init :  Serial Number : 0000000000000000
2019-04-23T15:55:35Z INFO #CRASHREPORT#  llcommon/llapp.cpp(1179) windows_post_minidump_callback : generated minidump: C:\Users\[REDACTED]\AppData\Roaming\Firestorm_x64\logs\dump-7ca05eda-fa3c-bae2-4798-1e9e2cbc05c8\12e40559-5caf-448c-a1e1-ba54bdf36ba6.dmp
2019-04-23T15:55:35Z INFO #  llcommon/llapp.cpp(1184) windows_post_minidump_callback : Entering Windows Exception Handler...
2019-04-23T15:55:35Z INFO #CRASHREPORT#  newview/llappviewer.cpp(4246) LLAppViewer::handleViewerCrash : Handle viewer crash entry.
2019-04-23T15:55:35Z INFO #CRASHREPORT#  newview/llappviewer.cpp(4248) LLAppViewer::handleViewerCrash : Last render pool type: 0
2019-04-23T15:55:35Z INFO #  llcommon/llmemory.cpp(155) LLMemory::logMemoryInfo : Current allocated physical memory(KB): 41876KB
2019-04-23T15:55:35Z INFO #  llcommon/llmemory.cpp(156) LLMemory::logMemoryInfo : Current allocated page size (KB): 26228KB
2019-04-23T15:55:35Z INFO #  llcommon/llmemory.cpp(157) LLMemory::logMemoryInfo : Current available physical memory(KB): 1635845KB
2019-04-23T15:55:35Z INFO #  llcommon/llmemory.cpp(158) LLMemory::logMemoryInfo : Current max usable memory(KB): 1677721KB
2019-04-23T15:55:35Z INFO #MarkerFile#  newview/llappviewer.cpp(4337) LLAppViewer::handleViewerCrash : Created crash marker file C:\Users\[REDACTED]\AppData\Roaming\Firestorm_x64\logs\Firestorm.error_marker
2019-04-23T15:55:35Z INFO #  newview/llappviewer.cpp(3629) LLAppViewer::writeDebugInfo : Opening debug file C:\Users\[REDACTED]\AppData\Roaming\Firestorm_x64\logs\dump-7ca05eda-fa3c-bae2-4798-1e9e2cbc05c8\\dynamic_debug_info.log

here is the dump file from firestorm, which I certainly hope contains no sensitive information

Edited April 23, 2019 by nogaku
screwed up quote tags

[Whirly Fizzle]

Got the dmp, you can remove the file now.

[Whirly Fizzle]

Here is the full crash stack: https://pastebin.com/2h1C3LCa

Important bit:

Operating system: Windows NT
                  10.0.17134
CPU: amd64
     family 6 model 58 stepping 9
     4 CPUs
 
Crash reason:  EXCEPTION_ACCESS_VIOLATION_READ
Crash address: 0x0
Process uptime: not available
 
Thread 0 (crashed)
 0  Firestorm-Releasex64.exe!LLMachineID::init() [llmachineid.cpp : 224 + 0x4]
 1  Firestorm-Releasex64.exe!LLAppViewer::init() [llappviewer.cpp : 991 + 0x5]
 2  Firestorm-Releasex64.exe!WinMain [llappviewerwin32.cpp : 354 + 0x9]
 3  Firestorm-Releasex64.exe!__tmainCRTStartup [crtexe.c : 618 + 0x15]
 4  kernel32.dll + 0x14034
 5  ntdll.dll + 0x73691
 6  KERNELBASE.dll + 0xebc40

 

So it very much looks like the crash is caused because the viewer can't get the machine ID/serial.

I'm going to ask the devs for advice about this & I'll get back to you here.
I would  still file a LL JIRA for this crash though.

[Whirly Fizzle]

2 hours ago, nogaku said:

and this is my first install on this machine

Is this a new computer?

[Whirly Fizzle]

wmic path Win32_VideoController get NAME

If you run this command in a cmd prompt, does it list your video cards or do you get an error message eg) "wmic is not recognised as an internal or external command, operable program or batch file" ?
If you get an error, please post it.
You should not get an error.  I'm curious if you have broken WMI on your system which means the viewer can't get the machine ID.

One of the Firestorm devs has offered to make a patch to the viewer to try & find out more detail about why the viewer crashes.
You would need to install & run a custom build of Firestorm with the patch to get the needed info.
Are you willing to do this?  It's fine if you are not, I totally understand.
 

 

[Whirly Fizzle]

For those interested, here is the patch:  it should also hopefully fix the crash.
The dev said it was a null pointer.

I:\phoenix-firestorm-lgpl>hg diff
diff -r 0814d9f99101 indra/newview/llmachineid.cpp
--- a/indra/newview/llmachineid.cpp     Mon Apr 22 16:32:15 2019 +0300
+++ b/indra/newview/llmachineid.cpp     Tue Apr 23 18:43:14 2019 +0100
@@ -217,6 +217,16 @@

             // Get the value of the Name property
             hr = pclsObj->Get(L"SerialNumber", 0, &vtProp, 0, 0);
+                       if( FAILED(hr) || !vtProp.bstrVal || vtProp.vt != VT_BSTR )
+                       {
+                               std::stringstream strHr;
+                               strHr << std::hex << hr;
+                               LL_WARNS() << "pclsObj->Get(L'SerialNumber') failed with hr" << strHr.str().c_str() << " and vtProp.vt = " << (U32)vtProp.vt << LL_EN
DL;
+                               pclsObj->Release();
+                               continue;
+                       }
+
+
             LL_INFOS("AppInit") << " Serial Number : " << vtProp.bstrVal << LL_ENDL;
             // use characters in the returned Serial Number to create a byte array of size len
             BSTR serialNumber ( vtProp.bstrVal);

 

[nogaku]

it is not a new computer. it is in fact a very old install of windows which is probably why it's so glitchy.

The command correctly lists my current video card.

I can run a custom build of firestorm if you'd like. I just hope you're not doing it only for my sake.

It occurs to me I have working linux dualboot on this machine and I could probably run firestorm from there if all else fails.

[Whirly Fizzle]

1 minute ago, nogaku said:

I can run a custom build of firestorm if you'd like. I just hope you're not doing it only for my sake.

Great.  It's no trouble. You are probably not the only one affected by this crash so being willing to test a fix is very helpful & for the good of everyone 
I have meetings for a while but I'll fire off the patched Firestorm build as soon as I can & I'll send it to you in a forum PM.
If this fixes the crash then the fix can be contributed to Linden Lab too.

 

[nogaku]

I think I figured it out. It might not have a serial number because I built this computer myself from a new motherboard.

here is the output from wmic bios get serialnumber

SerialNumber
To be filled by O.E.M.

[Whirly Fizzle]

I sent you the patched build of Firestorm in a forum PM.

Keeping fingers crossed 

[nogaku]

I have installed the patched version of Firestorm and it seems to have resolved everything. Was able to log in and get to the game. Thank you so much!

[Whirly Fizzle]

Fix pushed: https://hg.phoenixviewer.com/phoenix-firestorm-lgpl/rev/513868ae7af6

[Whirly Fizzle]

Bug report filed against the LL viewer: BUG-226885 - Viewer crashes on launch if serial is either unset or does not return a string