Chameleon

Issue 393: Unknown Exception Vector 15 ... System haltet

Reported by Stephan Quernheim, Apr 19, 2015

Today I downloaded and compiled the recent version 2657 of 
chameleon. I copied the file boot to my OSX partition to replace the 
old one (ver. 2207). After the restart the following error occurred 
immediatly before opening the chameleon booter screen:

Unknown Exception Vector 15
EDI 0x0, ESI 0xd0228096, EBP 0x1fae4, ESP 0x1f9ac
EBX 0xf30123, EDX 0x40021, ECX 0x0, EAX 0xc0
Exception Code 0x0, EIP 0x4069f, CS 0x8, EFLAGS 0x246
System Haltet

OSX is on a separate SSD and BIOS is set to boot from this device 
first.

Replacing the file ver. 2657 with 2207 cures the "problem" 
immediately.

Am I missing something?

Comment 1 by Konstantinos Giannakas, Apr 21, 2015

I confirm too, this error.

I compiled latest version r2657 without issues (10.9.5) on my 
Gigabyte GA-P55M-UD4 (Core i7-860) and it works well upon each 
reboot.

But when running same binary/version r2657 on my newer Gigabyte 
GA-Z87N-WIFI (Core i7-4790K) and upon each boot (10.9.5 too) I have 
this error as well.

Reverting to older "boot" file fixed this. Kindly assist. 
Is this somewhat incompatible with newer boards/chipsets or UEFI 
BIOSes?

Comment 2 by Konstantinos Giannakas, Apr 21, 2015

Hello again, reverted to r2643 that had handy, still produces the 
error (on the newer motherboard). Another previous build r2530 that 
I had in a bootable USB stick, loads as expected.

In my setup, there are *no* modules present or loaded, just a 
working DSDT.aml, SSDT.aml, smbios.plist file and the 
org.Chameleon.boot.plist of course. And basic Theme folder.

Attached output error. If I find more versions working after r2530 
will report back.

Comment 3 by Konstantinos Giannakas, Apr 21, 2015

Update (had these pre-compiled and stored in backup):

r2530 works
r2542 works
r2570 works
r2587 works
r2593 does not work
r2643 does not work

Same vector 15 error, system halted. From the logs in Cornerstone 
(SVN) I cannot detect the issue on r2587 or later, only that fix in 
fake_efi.c?

2.592: Merge Interrupt Management changes from Zenith432 branch
2.591: Fix KernelPanic for x32 machine located in fake_efi.c
2.590: Typo fake_efi.c (/trunk/i386/libsaio)

Comment 4 by ErmaC , Apr 28, 2015

Pls try the revision 2662.

ErmaC
Status: AwaitingInformation

Comment 5 by Konstantinos Giannakas, Apr 29, 2015

Hi ErmaC, I can confirm build 2662 just compiled, works as expected 
(back to normal) on my same motherboard and setup (Gigabyte 
GA-Z87N-WIFI and Core i7-4790K).

Many thanks. Can I ask if you managed to detect the issue, and what 
it was? We don't want to see it repeated again... Thanks the whole 
team.

Comment 6 by ErmaC , Apr 29, 2015

Hi the issue was located in the commit "2.592" (Merge 
Interrupt Management) and was discovered by Bronya
http://www.insanelymac.com/forum/topic/231075-chameleon-23svn-officia
l-pkg-installer-binaries/?p=2131222
So the commit/feature still here but I disable his function with 
MACRO switch.

cordially

ErmaC
Status: Fixed

Comment 7 by Konstantinos Giannakas, Apr 29, 2015

Thanks ErmaC. Is this a temporary fix or a proper bug-fix? Just 
trying to understand what you mean by "disable his function 
with MACRO switch.". Hope anyway never to see this again. Many 
thanks. Will follow that InsanelyMac thread now.

Comment 8 by ErmaC , Apr 29, 2015

I mean this:
http://forge.voodooprojects.org/p/chameleon/source/tree/HEAD/trunk/i3
86/boot2/boot.c#L63

then here:
http://forge.voodooprojects.org/p/chameleon/source/tree/HEAD/trunk/i3
86/boot2/boot.c#L84

here:
http://forge.voodooprojects.org/p/chameleon/source/tree/HEAD/trunk/i3
86/boot2/boot.c#L455

and here:
http://forge.voodooprojects.org/p/chameleon/source/tree/HEAD/trunk/i3
86/boot2/boot.c#L867

a "simple" boolean "On/Off"

So the feature is still present in the source but disabled when it 
will be compiled...
If you wont try change the 0 to 1 and build the project.
-----------------------------------
#ifndef DEBUG_INTERRUPTS
	#define DEBUG_INTERRUPTS 1
#endif
-----------------------------------

ErmaC

Comment 9 by Konstantinos Giannakas, Apr 30, 2015

Thanks ErmaC, no need to tamper with anything :-)
For me #define DEBUG_INTERRUPTS 0 = ALWAYS, until the code is 
revisited and perhaps improved. Hopefully, this will be also 
reflected on the comments/notes for that release so I can avoid :-D  
Thanks again for your time!

Comment 10 by Stephan Quernheim, Apr 30, 2015

Thank you very much ErmaC!

Created: 8 years 11 months ago by Stephan Quernheim

Updated: 8 years 10 months ago

Status: Fixed

Followed by: 2 persons

Labels:
Priority:Medium
Type:Defect