Segmentation Fault (SIGSEGV) in bool Unit::Killed()

vegastrike / Vega-Strike-Engine-Source

Vega Strike Engine

Other

258 stars 43 forks source link

Segmentation Fault (SIGSEGV) in bool Unit::Killed() #11

Open stephengtuggy opened 5 years ago

stephengtuggy commented 5 years ago

Just now, I got a segfault in the inline bool method Unit::Killed(). Judging by the comment just above, indicating that the void Kill() method deletes an object, it seems to me that trying to read the value of Killed is a bad idea in general. It would never return true, only false or... segfault. Why? Because in the latter case, the object has been deleted, and it no longer exists in memory!

We should probably do away with the whole concept of Killed, and use smart pointers instead, to avoid trying to access memory that has been reclaimed.

stephengtuggy commented 5 years ago

After further experimentation and examination of the code, it looks like we might need the Killed concept after all. However, to prevent other threads from trying to read that property on an object that has already been destroyed, I suggest using some form of thread-safe smart pointers. (See #8 ).

stephengtuggy commented 5 years ago

std::atomic_flag or similar might be worth using for this property, also, in combination with smart pointers for the Unit object (and descendants) as a whole.

LifWirser commented 4 years ago

While a "segfault" report is serious, this report is lacking enough information not to be

The minimum need for a proper BUG report is

The type of os on which the error occured
short description of machine error occured on (64 bit ibm w nvidia-blah)
error message displayed (segfault at blah.cpp ...)
breif description of actions that caused problem
is problem repeatable

Based on lack of needed informatiom I recommend this issue be closed

stephengtuggy commented 4 years ago

My apologies. I thought I had filled in the environment information for this bug months ago.

Okay, here goes; The OS was Ubuntu 18.04 LTS, I believe. 64-bit CPU: 8th-generation Intel Core i5-8500, 6-core, 3.00 GHz Main system RAM: 16 GB of DDR4 Graphics: GeForce GTX 1050 Ti, with 4 GB of dedicated GDDR5 Motherboard: Gigabyte H370M-DS3H Main system storage: 500GB M.2 NVMe SSD

I think there were two error messages displayed, one after the other. I no longer remember the first one, only the second one, which was the segfault message, in the terminal that I was running vegastrike from.

I do not have a specific series of actions to reproduce the problem, but I am guaranteed to see it on my hardware within about 20 minutes of gameplay or less, one way or another.

@LifWirser do you see anything else I'm missing?

stephengtuggy commented 3 years ago

This might still be an issue; I'm not sure. If it is, the problem may only show up when DESTRUCTDEBUG is defined.