View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000544 | Spring engine | General | public | 2007-05-31 01:58 | 2013-10-02 16:24 |
| Reporter | redstar | Assigned To | abma | ||
| Priority | normal | Severity | feature | Reproducibility | always |
| Status | closed | Resolution | won't fix | ||
| Summary | 0000544: Vista TDR mishandling | ||||
| Description | When a gpu operation takes too long it times out and the Operating system interprets it as a driver/gpu hang, and reloads the gpu driver in order to recover rather than showing a blue screen of death. This happens while spring is running, and each time it happens spring does not recover and instead crashes with c++ runtime exception message window, it then continues to run in the taskbar unplayable and connected to the host untill closed by the user via task manager. | ||||
| Additional Information | More detail including links to articles can be found here: http://spring.clan-sy.com/phpbb/viewtopic.php?t=10117 The device remove event occurs when a TDR crash is in progress. There are checks for this that can be used provided by the Direct X 9+10 APIs. Here is such a check from Direct3D 10: http://msdn2.microsoft.com/en-us/library/bb173571.aspx | ||||
| Tags | No tags attached. | ||||
| Checked infolog.txt for Errors | |||||
|
|
Perhaps try run in compatibility mode for xp? |
|
|
A graphics driver is outside the scope of compatability abstraction layer and communicates directly with the windows subsystems. As such running in XP compatability mode will generate exactly the same OpenGL instructions leading to exactly the same issue. Also the expected behaviour under Vista of a graphics driver crash is a baloon popup and a handling of a device remove event by directx. The expected behaviour under XP of a graphics driver crash is a blue screen of death. So aside from driver side fixes to prevent the issue to begin with, the problem must still be handled so that spring can recover, or a possible spring side cause must be found, aka graphics operations that last too long causing the driver subsystem to timeout and declare a driver crash has occurred. |
|
|
Ah, now I understand :-D, sorry bout that :-P |
|
|
as this requires directx i see that as feature-request. maybe directx will be never supported by the spring-engine. |
|
|
either SDL has this feature out of the box or we won't implement it. patches for it are welcome, but i guess it will be ugly as it seems to be a DirectX only feature. also, graphics drivers shouldn't crash! if they do, this functions won't really help as they could do it repeatedtly. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2007-05-31 01:58 | redstar | New Issue | |
| 2007-10-31 08:42 | ShadowOfLegends | Note Added: 0001387 | |
| 2007-10-31 14:22 | redstar | Note Added: 0001391 | |
| 2007-11-01 03:48 | ShadowOfLegends | Note Added: 0001394 | |
| 2010-11-24 18:59 | abma | Note Added: 0005973 | |
| 2010-11-24 18:59 | abma | Severity | crash => feature |
| 2013-10-02 16:24 | abma | Note Added: 0011656 | |
| 2013-10-02 16:24 | abma | Status | new => closed |
| 2013-10-02 16:24 | abma | Assigned To | => abma |
| 2013-10-02 16:24 | abma | Resolution | open => won't fix |
