IGCIT / Intel-GPU-Community-Issue-Tracker-IGCIT

IGCIT is a Community-driven issue tracker for Intel GPUs.
GNU General Public License v3.0
115 stars 4 forks source link

CS2 random crash at executable startup (no window displayed) #589

Closed freak2fast4u closed 9 months ago

freak2fast4u commented 10 months ago

Checklist [README]

Game / Application [Required]

Counter-Strike 2

Game Platform [Required]

Other game platform

No response

Processor / Processor Number [Required]

Ryzen 7800X3D

Graphic Card [Required]

Intel Arc A770 16GB LE

GPU Driver Version [Required]

31.0.101.4952

Rendering API [Required]

Windows Build Number [Required]

Other Windows build number

19045.3636

Intel System Support Utility report

igcit_ssu_20231108_cs2startcrashrandom_anon.txt

Description and steps to reproduce [Required]

Start the game, notice no window opened and Steam reports the game can be started. No popup, no crash, nothing, the process just silently exits. Random.

Game graphic quality [Required]

Game resolution [Required]

2560x1440

Game VSync [Required]

Off

Game display mode [Required]

Detailed game settings [Required]

N/A

Device / Platform name

ASUS TUF GAMING B650-PLUS WIFI

Crash dumps [Required, if applicable]

cs2_2023_1108_154646_0_accessviolation.zip

Save game

No response

freak2fast4u commented 10 months ago

Excerpt from WinDbg : 2023-11-08 16_03_30-E__SteamLibrary_steamapps_common_Counter-Strike Global Offensive_game_bin_win64_

As a side note, I was trying to use the latest PresentMon 0.6 beta along side, but this worked multiple times before and after, so I'm not sure it has anything to do with it.

freak2fast4u commented 10 months ago

Another occurence today (happened twice in fact, but here's the latest one) : cs2_2023_1116_232030_0_accessviolation.zip

image

crash_dump_synthesis_windbg_cs2.exe_20231116.txt

Joseph-Intel commented 10 months ago

Hey @freak2fast4u thanks for bringing up your concern. I was able to test this out and i wasn't getting any crashes and was able to play the game. I know this game updated a couple of times since this post. I am also on the latest graphics driver V1014953. Would you be able to test this out and let me know? I recommend checking the files or reinstalling the game as well if you haven't done so.

freak2fast4u commented 10 months ago

Hi @Joseph-Intel, thank you for your interest. Sadly, the problem appears randomly, and also rarely. Maybe on average once every fifty runs ? Reproducing it will be tedious. Re-running the game a second time after a crash usually works fine.

It also just so turns out that it happened to me twice yesterday alone (I only uploaded the second occurence in my last comment).

Valve has indeed rolled out a massive patch to CS2 on 16/11 (a few hours after my previous comment) which I recieved today. The patch notes don't mention any crash-related fixes, but I'll stay on the lookout for any spurious launch crashes that might happen going forward, if any.

I sure hope both crash dumps I provided give actual pointers to where exactly in the driver code the crash occurs, wether it's induced by bad input from CS2's game engine or an actual oopsie in the Intel Arc driver.

The game crashed a total of 13 times in the past 2 months, and I've been starting/restarting it quite a lot.

image

I'll keep you updated anyways :)

Joseph-Intel commented 10 months ago

@freak2fast4u thank you for updating! we will continue testing and let you know if we find anything

freak2fast4u commented 10 months ago

It happened yet again today with latest beta drivers (4972 currently) on latest CS2 build (20/11/2023). cs2_2023_1125_162007_0_accessviolation.zip

crash_dump_synthesis_windbg_cs2.exe_20231125.txt

I played the game earlier without issues, and just relaunched it without any issues either. Completely random, and still happening with up-to-date software on all sides.

image

Take note that one of the first things CS2 does when launching is playing an intro video, so it isn't necessarily 3D-related. Could it be related to some hw-accelerated decoding of the launch video that randomly fails ? DXVA/DXVA2 come to my mind as a plausible area of exploration, but who knows ... See attached dump for a more detailed view of what's going on :)

As with anything "random" in software, there has to be a race condition somewhere leading to that reading of a null pointer variable before it's actually initialized ^^"

Joseph-Intel commented 9 months ago

Hi @freak2fast4u After more testing i still have not been able to reproduce. Since i can not reproduce the issue i cannot pursue or file an internal report. @IGCIT Please help close this case. This case can be reopened at anytime that you can help find a way to reproduce it more reliably or if others are having this issue. Thanks for your time

freak2fast4u commented 9 months ago

Seems fair by me, if noone else is complaining then it has to be something specific with my setup. I agree your time would be better spent elsewhere. Thanks for your time also. See you around :)