Debugging inconsistent Inno deinitialize error

Charles Pigott

unread,

Jul 10, 2021, 2:42:06 AM7/10/21

to inno...@googlegroups.com

Hi,

We have a fairly large installer (in compressed size and amount of pascal) that when running occasionally hangs with a msgbox containing an error similar to:

Access violation at address <addr> in module 'setup.tmp'. Read of address <addr>.

The installer is compiled with a basic ISCC.exe foo.iss command, running Inno version 6.0.5. The installer is tested/used via a CI system using flags /SILENT /VERYSILENT /CURRENTUSER and others.

Because the installer is so large, and the error happens so inconsistently, trying to cut the installer down to reproduce the issue is basically impossible.

I did find https://stackoverflow.com/questions/27089898/how-to-debug-inno-setup-exceptions

but I've checked every single defined function within the iss files and none of them have any issues with an undefined Result variable.

After enabling logging, the last line of the log is always "Deinitializing Setup.", which leads me to believe that the issue is somewhere in Inno's own DeinitSetup function -

https://github.com/jrsoftware/issrc/blob/is-6_0_5/Projects/Main.pas#L3543-L3621

We don't have any DeinitializeSetup/GetCustomSetupExitCode functions, which I suppose narrows it down slightly.

Any further ideas for debugging this?

Thanks,

Charles

--

Register for our upcoming webinar with NXP & Curtiss-Wright: Verifying Avionics Multicore Hardware & Software (CAST-32A & A(M)C 20-193) .

José Gonçalves

unread,

Jul 11, 2021, 2:06:32 AM7/11/21

to innosetup

You could try and add more log info to this DeinitSetup procedure and recompile your version of IS + your installer.

Message has been deleted

Charles Pigott

unread,

Jul 26, 2021, 6:19:09 AM7/26/21

to inno...@googlegroups.com

I could! But I have no experience with compiling IS (just using binary packages) and was rather hoping I could avoid it...

Charles

Eivind Bakkestuen

unread,

Jul 26, 2021, 6:32:51 PM7/26/21

to inno...@googlegroups.com

You are saying that the exact same executable may crash once, and when run again in the same testing environment, it may run many times and not crash?

What is the actual message (with actual <...> addresses)?

Are you using any external modules (COM/DLL)?

On Mon, Jul 26, 2021 at 8:13 PM 'Charles Pigott' via innosetup <inno...@googlegroups.com> wrote:

We could! But compiling IS at all is something I'd like to avoid, having not done it before...

Charles

--
You received this message because you are subscribed to the Google Groups "innosetup" group.
To unsubscribe from this group and stop receiving emails from it, send an email to innosetup+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/innosetup/21920606-fad4-44eb-ba57-e915f0245f68n%40googlegroups.com.

Charles Pigott

unread,

Jul 27, 2021, 6:45:34 AM7/27/21

to inno...@googlegroups.com

I don't have complete records, and it hasn't happened for a while now, but these are all the ones I can find in various chat logs:

Access violation at address 00000008 in module 'setup-rvs_vtesteditor_07_01-WF20210714_x64.tmp'. Execution of address 00000008.
Access violation at address 00000008 in module 'setup-rvs_v3.14BF20210209_x64.tmp'. Execution of address 00000008.
Access violation at address 6F707845 in module 'setup-rvs_vWF20200604_x64.tmp'. Execution of address 6F707845.

No external modules.

From: eivind.b...@gmail.com <Unknown>

Date: Monday, 26 July 2021 at 23:32:51 UTC+1

Subject: Re: Debugging inconsistent Inno deinitialize error

To: inno...@googlegroups.com <Unknown>

Reply all

Reply to author

Forward