Modules that die...

I was playing around with some module code the other day and I messed up the write-back into R12.
Which meant that attempts to use the module caused an abort when trying to LDR from the bogus address, and attempts to *RMKill the module was met with “Not a heap block”. Essentially, the module could not be killed.

I figured there might be a way to fake out a safe exit by poking some location in memory (modaddr!-4 ?) but frankly – not only was a reboot quicker, but the OS should not fail to kill off a module in this manner.

If the module itself refuses to die, that is one specific API-documented case. However in the case of workspace pointer corruption, the OS should say something like “Workspace corrupt – could not be freed” and then remove the module anyway as was instructed.

Can this be added? It would be a heck of a lot simpler for debugging stuff to have a more resiliant *RMKill.

Have you tried DebugTools from the ROOL site?

No, I’ll go look for that. Thanks.

Doesn’t change the fact, however, that an error in releasing the module should not render it otherwise unkillable…

Rick said

Waaaah

As it happens, my !Modules utility you’ve already asked me to release has precisely the functionality you want.

And for exactly the same reason. ;-)