[Bug Report] Playground version 463 vm

[Butz_san]

# Playground version
svn 463

# Cpu options selected
all on

# Plugins used
GSDX 0.1.12
Linuzappz Iso CDVD 0.7.0
ZeroSPU 0.45
LilyPad 0.9.4

# Description of the bug and of how to replicate it
I was trying to change the frame limit from 0(default) to 70, suddenly when I hit the accept button, I got a Crash, windows said that Pcsx2 PP 463 crashed. Besides I got a message that it's updating without doing anything , I attach a picture of the message that I got in the output window, is it normal?

[kabooz]

# Description of the bug and of how to replicate it
I was trying to change the frame limit from 0(default) to 70, suddenly when I hit the accept button, I got a Crash, windows said that Pcsx2 PP 463 crashed. Besides I got a message that it's updating without doing anything , I attach a picture of the message that I got in the output window, is it normal?
[/quote]

tried to replicate the bug but it didn't happen to me... neither on the vm nor on the vtlb build... but i did have a random crash this this night... (when i changed the VU settings) so i think it was before the EE rec reset was displayed in the console (oh god we thank you for the console)
I used the VM build

do you have the problem every time?


EDIT: ah almost forgot^^ yes i have the profiles too ^^ seems to be for easier debugging

[Zeydlitz]

Fist of all, console messages was fixed in 468, it's simply profiler,it does not meant to be in a release. It has no relation for crush. And crush when user change something in UI sometimes happened, yes. We could only hope that they would be fixed someday.

[Butz_san]

Thanks for the answers

[Air]

The crash in the CpuDlg box is a known issue as of this morning, when I ran into it myself a couple times. It's caused by some messy code in the VUrecs that does a remarkably poor job of trying to allocate a virtual memory block for it's recompiled code buffer.

All of the recompilers (EE, IOP, VU0 and VU1) have recompiled code buffers. Each buffer is allocated using virtual memory, regardless of the Pcsx2 build, because Virtual Memory is the only way to create memory allocations with Code Execution Rights. The EE and IOP allocate smartly -- they try lots of blocks until one works. The VU tries one block, and dies (without error no less!) if the block is unavailable.

I'll take a look at fixing it now.