250 replies to this topic

[JoeMNSF]

The DSDT file gets stuck at the PCI Configuration Begin again, although I can't get ed working to apply the patches to any .dsl files, it just either gives me a ? or 357599 as the output. Same results when entering the ed commands manually from the patch files.

[bcc9]

JoeMNSF, on 03 December 2011 - 06:31 PM, said:

The DSDT file gets stuck at the PCI Configuration Begin again, although I can't get ed working to apply the patches to any .dsl files, it just either gives me a ? or 357599 as the output. Same results when entering the ed commands manually from the patch files.

I just got access to my old 1340 again, and the new DSDT.aml is working on it just fine for me.  So I'm still at a loss as to why folks are having problems.
As for ed, looks like the OSX version is not parsing the comments as expected (as it works in the gnu version used by linux distributions).  But this avoids that issue:

% tcsh
% cp orig.dsl fixed.dsl
% foreach i ( ed* )
foreach? echo $i
foreach? grep -v '^#' $i | ed - fixed.dsl
foreach? end
 

[bcc9]

Ok, for your debugging pleasure, here is a set of incremental aml files to try.
orig.aml is the original DSDT from a14 bios
ed1.aml is the DSDT with edit #1
ed2.aml is the DSDT with edit#1+edit#2
...
ed9.aml is the DSDT with all 9 edits.
I renumbered the edits a bit so that they would all compile incrementally (well except ed7).

You can install all the .aml files in /Extra and boot with DSDT=filename, for example DSDT=ed1.aml
This should make troubleshooting easy for those who are having hanging problems.

Attached Files

•  dsdt_debug.zip   436.15K   26 downloads

[bcc9]

I've also updated the dsdt bundle in post #1 to v3, so that it matches the above debug edits, fixes the patch script for OSX, and address a problem with patch #9.

[bfbruin]

Sorry, bcc9.

I tried DSDT.aml from v3, and then attempted to troubleshoot via orig.aml and ed1-6, 8 and 9. (ed7.aml was not in the zip files, though it was there as a dsl.)

None of them worked to get past [PCI Configuration Begin]. Only DSDT=no got it to the usual booting state of bridges 8 devices 22, followed by bridge 4 device 1 cardbus 0 and so forth.

Maybe I should roll back to A14 like Dual_Shock?

[bcc9]

bfbruin, on 04 December 2011 - 12:59 AM, said:

Sorry, bcc9.

I tried DSDT.aml from v3, and then attempted to troubleshoot via orig.aml and ed1-6, 8 and 9. (ed7.aml was not in the zip files, though it was there as a dsl.)

None of them worked to get past [PCI Configuration Begin]. Only DSDT=no got it to the usual booting state of bridges 8 devices 22, followed by bridge 4 device 1 cardbus 0 and so forth.

Maybe I should roll back to A14 like Dual_Shock?

If you really can't even boot with DSDT=orig.aml, then something is pretty broken.  You're using chameleon r1649?
Are you trying to boot the 10.7 install partition (the first boot), or the partition you made during the install (the second boot)?

Dual shock's post didn't indicate whether or not he was able to boot with a vanilla .aml under a14 or a15 so his limited data seems to only show that downgrading to a14 didn't fix the use of a modified .aml for those who are having trouble.

[bfbruin]

I've been using 1493, so I'll upgrade to the 1649 now. UPDATE: Cannot sem to find Chameleon 1649, as builds.voodooprojects appears to be down.

I'm booting my Lion partition, not my Install Partition.

[bcc9]

bfbruin, on 04 December 2011 - 01:18 AM, said:

I've been using 1493, so I'll upgrade to the 1649 now. UPDATE: Cannot sem to find Chameleon 1649, as builds.voodooprojects appears to be down.

I wouldn't sweat it; either version should be fine (both work fine for me).

I'm assuming you can't boot into single user mode or safe mode with orig.aml?  Does the system really seem to have hung or just fails to display anything after "pci configuration begin" (with disk activity continuing)?

[bcc9]

If those who are having problems with hanging at "pci configuration begin" can't get anywhere with dsdt troubleshooting, then I recommend testing the old workaround of using the 10.6.8  IOPCIFamily.  If that works, then it's clear what has been going wrong and what needs a better fix.

[metamorphoise]

I had a go with your debug DSDTs and like bfbruin, even DSDT=orig.aml gives me a hang on [PCI config begin]
i tried to boot to safe mode with -x DSDT=orig.aml npci=0x2000 and it hung at a message slightly before [PCI config begin], something like "PFC64 0x5100000000" or similar.
I would like to try the 10.6.8 IOPCIFamily.kext workaround but I have no 10.6 installation to get the file and have not been able to find a downloadable 10.6.8 kext


Here is some information that may or may not be pertinent:
-I'm using A14 BIOS
-I'm using Chameleon SVN 1650 as the 1649 link seems to be down
-I needed to remove the SMmemtype from smbios.plist as my system seemed to have trouble booting with that, I think I changed the ram in my 1340 at some stage, still 4Gb.
-I am installing off a clean lion install done with Tonymac's Unibeast tool, could this be causing problems?
-Maybe 50% of the time, it seems to boot fine but no graphics are displayed (monitor turns off) and i have to boot with GraphicsEnalber=No
-The kexts in /E/E are 9400m_gstate_inject, AppleACPIPS2Nub, fakesmc, VoodooBattery & VoodooPS2Controller

PS, the links to the OpenHaltRestart and your HCI bluetooth pkg seem to be broken, also some links in the older 10.6 guide. they send me to the main insanelymac page. seems like something changed with this board's handling of links?

Thanks,
Tom

[ticktick]

thanks for detailed tutorial. After upgrade to 10.7.2 I am stuck at still waiting for root device and it never goes further.

[bcc9]

metamorphoise, on 04 December 2011 - 01:34 PM, said:

I had a go with your debug DSDTs and like bfbruin, even DSDT=orig.aml gives me a hang on [PCI config begin]
i tried to boot to safe mode with -x DSDT=orig.aml npci=0x2000 and it hung at a message slightly before [PCI config begin], something like "PFC64 0x5100000000" or similar.
I would like to try the 10.6.8 IOPCIFamily.kext workaround but I have no 10.6 installation to get the file and have not been able to find a downloadable 10.6.8 kext

The 10.6 iopcifamily kexts are floating all over the place.  I've never tried them, so I don't know which one works with various flavors of 10.7.  Perhaps oleg's: http://cl.ly/3R460x3f092m0F0F393q ?  Recommend you check back in the 10.6 thread for details.

metamorphoise, on 04 December 2011 - 01:34 PM, said:

-I needed to remove the SMmemtype from smbios.plist as my system seemed to have trouble booting with that, I think I changed the ram in my 1340 at some stage, still 4Gb.

Makes no sense, the SMmemtype should just be cosmetic (allows the memory to show up as ddr3 in system profiler).  Any memory you added would have been ddr3 so the SMmemtype would still be right...

metamorphoise, on 04 December 2011 - 01:34 PM, said:

-I am installing off a clean lion install done with Tonymac's Unibeast tool, could this be causing problems?

You're having problems, but you're using a completely different install method, not following directions?  I'm not sure how I can help you with that.  Presumably tonymac's installers would give you different .plists in /Extra.  
There's plenty more debugging that you could do such as booting with Wait=yes, and comparing the no dsdt case vs the DSDT=orig.aml case.

metamorphoise, on 04 December 2011 - 01:34 PM, said:

-Maybe 50% of the time, it seems to boot fine but no graphics are displayed (monitor turns off) and i have to boot with GraphicsEnalber=No
-The kexts in /E/E are 9400m_gstate_inject, AppleACPIPS2Nub, fakesmc, VoodooBattery & VoodooPS2Controller

PS, the links to the OpenHaltRestart and your HCI bluetooth pkg seem to be broken, also some links in the older 10.6 guide. they send me to the main insanelymac page. seems like something changed with this board's handling of links?

That kext list seems about right, for starters, of course they can't just be in /E/E if you're using UseKernelCache=Yes
Those other links used to be fine, looks like the board corrupted them and also made them no longer work.  
Update: Fixed.  Definintely a forum bug.  These same links are busted in the 10.6 1340 thread too.

[LatinMcG]

try changing

Device (PCI0)
		{
			Name (_ADR, 0x00)
			Name (_HID, EisaId ("PNP0A08"))
			Name (_CID, EisaId ("PNP0A03"))
			Name (_UID, 0x01)

to

Device (PCI0)
		{
			Name (_HID, EisaId ("PNP0A08"))
			Name (_CID, EisaId ("PNP0A03"))
			Name (_ADR, 0x00)
			Name (_UID, 0x00)

root uid 1 for AMW0 as its on same _SB both UID 0 might conflict.. so change AMW0 _UID to 0x01

[bcc9]

LatinMcG, on 05 December 2011 - 06:21 PM, said:

try changing

Why, to obviate the need for PCIRootUID being set in org.chameleon.Boot.plist?  (This can also be achieved by simply moving the _UID assignment to before the _HID&_CID assignments.)  This avoids the problem with chameleon's buggy .aml parsing so that it is able to find the _UID value without a PCIRootUID assignment.

Or is the intent here to somehow help the hang at "PCI Configuration Begin" case?

LatinMcG, on 05 December 2011 - 06:21 PM, said:

root uid 1 for AMW0 as its on same _SB both UID 0 might conflict.. so change AMW0 _UID to 0x01

Seems this shouldn't conflict with AMW0 either way as the two don't share _HID or _CID values.  "The
_UID must be unique across all devices with either a common _HID or _CID." in the spec...

[LatinMcG]

when 10.6.8 cme out my inspiron would not boot. till i read somewhere the tequila fix worked due to UID being 0 not 1.
yes it can be done in chameleon..but the real mac way is 0x00.
my inspiron 1520 didnt have _UID at all . i try to add under _ADR and no boot.
so i replaced _ADR with _UID and it works. no more pci config hang.

the amw0  i wasnt sure if the UID had to be non conflicting or not.. i guess both can say Zero

[bcc9]

I'm not seeing anything in the two "tequila fix" threads I can find about the UID change having any effect on hangs at "pci configuration begin".  At least I don't see how changing the UID number for systems that already have both ADR & UID defined would have that effect.  But folks who are having problems are free to experiment of course.

Yes, patching the dsdt so as to not need to configure PCIRootUID seems slightly preferable.  Seems some people are having trouble booting consistently with PCIRootUID set (metamorphoise's 50% failure with graphics sounds like this situation).

[metamorphoise]

I seem to have tracked down what was causing my [PCI config hang], my extracted DSDT was different from the original DSDT this guide is based on.

I have attached the relevant files for others with the same problem to try, but I imagine that your DSDT may be different again so here is the process I followed:
I extracted my DSDT using DSDTESE and ran it through filemerge (nice diff program, installed when you install Xcode) with the orig.dsl from dsdt.a14.lion.v3.zip. filemerge showed 9 differences, 4 being insignificant differences in comments at the top and 5 being significant differences like:

metamorphoise1340orig.dsl				OperationRegion (ASLD, SystemMemory, 0x8FECF5CA, 0x00000100)
orig.dsl				OperationRegion (ASLD, SystemMemory, 0xAFECF5CA, 0x00000100)

and

metamorphoise1340orig.dsl				OperationRegion (ROM1, SystemMemory, 0x8FECF6CA, 0x00010000)
orig.dsl				OperationRegion (ROM1, SystemMemory, 0xAFECF6CA, 0x00010000)

I then renamed my extracted dsl to orig.dsl and copied it to the dsdt.a14.lion.zip directory, overwriting the existing file. Then ran the cmd file to patch my dsl, and we have a new fixed.dsl

Drop that fixed.dsl into iaslMe (http://www.insanelym...howtopic=189272), move the resulting aml to /Extra/DSDT.aml, and boot! (i also did chmod 775&&chown root:wheel on it but don't think that is necessary)




Now my system seems to be almost fully functional, but there are still some issues with the graphics that I will look into now but the symptoms are below and any help would be greatly appreciated:
Often the system will boot seemingly normally but nothing is displayed when it gets to login (screen backlight is on but screen is black, disk and keyboard lights functioning) and I have to reboot one or more times till it decides to work, or boot with GraphicsEnabler=No

Attached Files

•  metamorphoiseDSDT.zip   80.43K   12 downloads

[bcc9]

metamorphoise, on 07 December 2011 - 01:45 AM, said:

I extracted my DSDT using DSDTESE and ran it through filemerge (nice diff program, installed when you install Xcode) with the orig.dsl from dsdt.a14.lion.v3.zip. filemerge showed 9 differences, 4 being insignificant differences in comments at the top and 5 being significant differences like:

Interesting, that's the first time anyone has reported any dsdt differences (and we've been sharing dsdts for this laptop for almost 3 years now).
This is with a14 bios, right?  And you have a 9400m?

[metamorphoise]

bcc9, on 07 December 2011 - 02:20 AM, said:

Interesting, that's the first time anyone has reported any dsdt differences (and we've been sharing dsdts for this laptop for almost 3 years now).
This is with a14 bios, right?  And you have a 9400m?

Yes it is A14 BIOS

You may be onto something with the graphics card though, I bought this laptop 2nd hand and thought it was a 9400m.
When the GraphicsEnabler works, About This Mac identifies it as a "GeForce 9400M G 256 MB"
But if i put my service tag (BQ6SWK1) into Dell website, it says my original configuration was/is "NVIDIA® GeForce® 210M - 512MB"

I have to scoot to work now so can't really look into it until tonight, but seems like this is some form of Hybrid SLI model. From what i can see here, '9500m hybrid SLI' consists of a 9400m & 9200m GS, but if mine is a 9400m & 210m maybe the original purchaser upgraded the graphics.

On another note, I tried applying LatinMcG's patch to my DSDT and removing the PCIRootUID string&key from org.chameleon.Boot.plist, no change. still boots only sometimes.

[LatinMcG]

at third step in tequila fix.(this is from 2009 no lion or pci config issues then but relevant)

the differences seem to be in ram amount changes: SystemMemory, 0x8FECF5CA

btw the 9400 is similar chip to 210. (rebranded with a lil tweaking in the rom and hdmi i think i read somewhere)