[GH-ISSUE #4888] Pursuit Force - Extreme Justice (bezier/spline with < 4 uv counts) #2031

Closed
opened 2026-03-18 06:38:21 +03:00 by kerem · 20 comments
Owner

Originally created by @fastrizwaan on GitHub (Dec 20, 2013).
Original GitHub issue: https://github.com/hrydgard/ppsspp/issues/4888

As soon as we enter the Tunnel, the ppsspp crashes. Crash happens in Windows 32bit, Linux SDL and Qt versions.

screenshot from 2013-12-20 23 27 30

As per this post, version 0.9.1 did not crash but crashes in later versions.
http://forums.ppsspp.org/showthread.php?tid=371&pid=57578#pid57578

Kindly look into the issue, it's a fun action game.

Actual PSP gameplay:
http://www.youtube.com/watch?v=ysDMU-NFpoY&t=1m23s

Originally created by @fastrizwaan on GitHub (Dec 20, 2013). Original GitHub issue: https://github.com/hrydgard/ppsspp/issues/4888 As soon as we enter the Tunnel, the ppsspp crashes. Crash happens in Windows 32bit, Linux SDL and Qt versions. ![screenshot from 2013-12-20 23 27 30](https://f.cloud.github.com/assets/1161682/1792484/42228022-69a0-11e3-9a25-1466e8585276.png) As per this post, version 0.9.1 did not crash but crashes in later versions. http://forums.ppsspp.org/showthread.php?tid=371&pid=57578#pid57578 Kindly look into the issue, it's a fun action game. Actual PSP gameplay: http://www.youtube.com/watch?v=ysDMU-NFpoY&t=1m23s
kerem 2026-03-18 06:38:21 +03:00
Author
Owner

@hrydgard commented on GitHub (Dec 20, 2013):

Does it still crash if you turn on "Low quality spline/bezier curves"? (yes, I know that the graphics will be semi-broken).

If it doesn't crash with that setting on, the crash is obviously in the spline/bezier code.

<!-- gh-comment-id:31029118 --> @hrydgard commented on GitHub (Dec 20, 2013): Does it still crash if you turn on "Low quality spline/bezier curves"? (yes, I know that the graphics will be semi-broken). If it doesn't crash with that setting on, the crash is obviously in the spline/bezier code.
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

Yes, it still crashes with Low quality spline/bezier curves enabled. Tested with Qt version.

<!-- gh-comment-id:31029404 --> @fastrizwaan commented on GitHub (Dec 20, 2013): Yes, it still crashes with Low quality spline/bezier curves enabled. Tested with Qt version.
Author
Owner

@unknownbrackets commented on GitHub (Dec 20, 2013):

Can you bisect the crash? See the wiki for details.

-[Unknown]

<!-- gh-comment-id:31037156 --> @unknownbrackets commented on GitHub (Dec 20, 2013): Can you bisect the crash? See the wiki for details. -[Unknown]
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

v0.9.1-134-g554ca84 does not crash, as in the git biset wiki,

should I go 100 commits up, till i get a crash?

ok i'm going backwards from the current 0.9.6

<!-- gh-comment-id:31044515 --> @fastrizwaan commented on GitHub (Dec 20, 2013): v0.9.1-134-g554ca84 does not crash, as in the git biset wiki, should I go 100 commits up, till i get a crash? ok i'm going backwards from the current 0.9.6
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

How do I go up in commits?

v0.9.1-1517-gc84a9da doesn't crash

This take me back not forward :(
git checkout HEAD~100

<!-- gh-comment-id:31046474 --> @fastrizwaan commented on GitHub (Dec 20, 2013): How do I go up in commits? v0.9.1-1517-gc84a9da doesn't crash This take me back not forward :( `git checkout HEAD~100`
Author
Owner

@hrydgard commented on GitHub (Dec 20, 2013):

git checkout master

should take you back to the start. Or you can find the commit hash in the commit list here and just do

git checkout fa83042

or whatever.

Or use "git bisect", read up on it, it's pretty easy.

<!-- gh-comment-id:31046674 --> @hrydgard commented on GitHub (Dec 20, 2013): git checkout master should take you back to the start. Or you can find the commit hash in the commit list here and just do git checkout fa83042 or whatever. Or use "git bisect", read up on it, it's pretty easy.
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

Bisecting: 0 revisions left to test after this (roughly 0 steps)
[f701f0c347] Windows buildfix.

<!-- gh-comment-id:31048738 --> @fastrizwaan commented on GitHub (Dec 20, 2013): Bisecting: 0 revisions left to test after this (roughly 0 steps) [f701f0c34708c687fee79a14a3774badce8ef4e4] Windows buildfix.
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

v0.9.1-1669-gf701f0c is the last version which does not crash! Yay :)

<!-- gh-comment-id:31048804 --> @fastrizwaan commented on GitHub (Dec 20, 2013): v0.9.1-1669-gf701f0c is the last version which does not crash! Yay :)
Author
Owner

@fastrizwaan commented on GitHub (Dec 20, 2013):

bash-4.2$ git bisect good
d696e5b02f is the first bad commit
commit d696e5b02f
Author: Henrik Rydgård hrydgard@gmail.com
Date: Thu Oct 10 17:36:42 2013 +0200

Try a different heuristic for the bezier subdivision factor

:040000 040000 a8ac54cdbb 5d77f9c613 M GPU

<!-- gh-comment-id:31048926 --> @fastrizwaan commented on GitHub (Dec 20, 2013): bash-4.2$ git bisect good d696e5b02fd54a7d6200b5b8e9507641a505c007 is the first bad commit commit d696e5b02fd54a7d6200b5b8e9507641a505c007 Author: Henrik Rydgård hrydgard@gmail.com Date: Thu Oct 10 17:36:42 2013 +0200 ``` Try a different heuristic for the bezier subdivision factor ``` :040000 040000 a8ac54cdbbeb20c219a264eec262a21a68a72358 5d77f9c6131f6cf0cc445cea86a284a9749b42ed M GPU
Author
Owner

@hrydgard commented on GitHub (Dec 20, 2013):

Alright, then it is beziers after all. Thanks for your investigation so far :)

<!-- gh-comment-id:31049750 --> @hrydgard commented on GitHub (Dec 20, 2013): Alright, then it is beziers after all. Thanks for your investigation so far :)
Author
Owner

@unknownbrackets commented on GitHub (Jun 27, 2014):

Does this still happen? There have been some improvements in the bezier code.

-[Unknown]

<!-- gh-comment-id:47296459 --> @unknownbrackets commented on GitHub (Jun 27, 2014): Does this still happen? There have been some improvements in the bezier code. -[Unknown]
Author
Owner

@fastrizwaan commented on GitHub (Jun 27, 2014):

yes, it still crashes. Tested with ppsspp-v0.9.8_1425. Do i need to change some settings, anywhere in ppsspp?

<!-- gh-comment-id:47300686 --> @fastrizwaan commented on GitHub (Jun 27, 2014): yes, it still crashes. Tested with ppsspp-v0.9.8_1425. Do i need to change some settings, anywhere in ppsspp?
Author
Owner

@unknownbrackets commented on GitHub (Jun 27, 2014):

What if you try the "medium" or "low" bezier quality? Atlhough that's not a solution if it still crashes.

-[Unknown]

<!-- gh-comment-id:47303587 --> @unknownbrackets commented on GitHub (Jun 27, 2014): What if you try the "medium" or "low" bezier quality? Atlhough that's not a solution if it still crashes. -[Unknown]
Author
Owner

@fastrizwaan commented on GitHub (Jun 27, 2014):

I tested with Low, Medium and High bezier quality, all the 3 crashes the game at entering the tunnel.

<!-- gh-comment-id:47304274 --> @fastrizwaan commented on GitHub (Jun 27, 2014): I tested with Low, Medium and High bezier quality, all the 3 crashes the game at entering the tunnel.
Author
Owner

@daniel229 commented on GitHub (Dec 3, 2014):

Still crash with v0.9.9.1-1046-gca8ba95 no matter what quality is.
01

savestate
http://1drv.ms/1vMO0S4

<!-- gh-comment-id:65379947 --> @daniel229 commented on GitHub (Dec 3, 2014): Still crash with v0.9.9.1-1046-gca8ba95 no matter what quality is. ![01](https://cloud.githubusercontent.com/assets/3481559/5278353/5f9632ea-7b13-11e4-8a8e-b94ee9537678.png) savestate http://1drv.ms/1vMO0S4
Author
Owner

@xebra commented on GitHub (Dec 5, 2014):

I checked that this issue crash on this address(0x08AFBDF8).
Using @daniel229's savestate.
It seems related to bezier, but I think it's not bezier issue. I guess it's broken opcode issue.
As I thought, this issue seems very similar to this.
Tantei Jinguji Saburo will crash at open the inventory (I'll report this issue later)
http://forums.ppsspp.org/showthread.php?tid=6869
pursuit ge 1
We need more research about these issues.

<!-- gh-comment-id:65795716 --> @xebra commented on GitHub (Dec 5, 2014): I checked that this issue crash on this address(0x08AFBDF8). Using @daniel229's savestate. It seems related to bezier, but I think it's not bezier issue. I guess it's broken opcode issue. As I thought, this issue seems very similar to this. Tantei Jinguji Saburo will crash at open the inventory (I'll report this issue later) http://forums.ppsspp.org/showthread.php?tid=6869 ![pursuit ge 1](https://cloud.githubusercontent.com/assets/1607422/5316879/34209dfe-7cd4-11e4-80f8-01741096e57b.png) We need more research about these issues.
Author
Owner

@daniel229 commented on GitHub (Dec 5, 2014):

For the Tantei Jinguji Saburo,it won't crash if replace the jpn0.pgf with other font file like the one from PSP.

<!-- gh-comment-id:65797965 --> @daniel229 commented on GitHub (Dec 5, 2014): For the Tantei Jinguji Saburo,it won't crash if replace the jpn0.pgf with other font file like the one from PSP.
Author
Owner

@xebra commented on GitHub (Dec 5, 2014):

Really? Thanks for useful information. I'll try to replace font file.

Well, I tested this game issue avoid bezier 1x7 like this

void TransformDrawEngine::SubmitBezier(void* control_points, void* indices, int count_u, int count_v, GEPatchPrimType prim_type, u32 vertType) {
    Flush();

    if (count_u == 1 || count_v == 1)
        return;
....

It's no problem about play the game...
So, probably bezier issue...
But 1x7 bezier is what? curved line? not curved surface...
Does PSP support curved line? If so we need support it.

<!-- gh-comment-id:65800907 --> @xebra commented on GitHub (Dec 5, 2014): Really? Thanks for useful information. I'll try to replace font file. Well, I tested this game issue avoid bezier 1x7 like this ``` void TransformDrawEngine::SubmitBezier(void* control_points, void* indices, int count_u, int count_v, GEPatchPrimType prim_type, u32 vertType) { Flush(); if (count_u == 1 || count_v == 1) return; .... ``` It's no problem about play the game... So, probably bezier issue... But 1x7 bezier is what? curved line? not curved surface... Does PSP support curved line? If so we need support it.
Author
Owner

@unknownbrackets commented on GitHub (Apr 10, 2016):

It seems from my testing that 1x7 and similar doesn't draw anything.

Does this game still work fine without crashing with the latest code? Any visible graphical issues? If it's working, I think we're safe to close this.

-[Unknown]

<!-- gh-comment-id:208064448 --> @unknownbrackets commented on GitHub (Apr 10, 2016): It seems from my testing that 1x7 and similar doesn't draw anything. Does this game still work fine without crashing with the latest code? Any visible graphical issues? If it's working, I think we're safe to close this. -[Unknown]
Author
Owner

@daniel229 commented on GitHub (Apr 11, 2016):

Everything still looks fine with master build.

<!-- gh-comment-id:208123028 --> @daniel229 commented on GitHub (Apr 11, 2016): Everything still looks fine with master build.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
starred/ppsspp#2031
No description provided.