250 replies to this topic

[wormx]

Ah, so it turns out I am no longer getting the header read size 200/Incorrect image signature  errors/warnings. However, the computer is still unresponsive after going to "sleep." The fan is still spinning, and the power light is still on solid, but the caps lock key doesn't change. Any other ideas on where I should look?
Thanks again!

[bcc9]

wormx, on 13 April 2013 - 04:01 PM, said:

Any other ideas on where I should look?
Thanks again!

So you could either figure out how your setup now differs from what is recommended in post #1, or just start troubleshooting which driver might be failing to suspend correctly.  Most likely it's the display driver or usb failing to suspend correctly.  If you don't want to figure out what's different in your setup vs the recommendations, I'd start by temporarily taking out the nvidia drivers and not using usb, and see how that changes things.

[bcc9]

Anyone with a 9400m 1340 upgrade from 10.8.2 to 10.8.3 successfully?
I finally had a chance to try it, and I cannot get the system to come up multi-user.  Without intervention, the system console shows everything starting, gets to the point where the graphics driver is starting, and the screen goes to black and the system locks up (no ping response).  Same thing if I boot with safe mode.

I can remove some of the nvidia graphics drivers and get the system to lock up right before the screen goes black instead, or I can get the system to panic with IOPPlatformPluginLegacy in the traceback if I remove all the nvidia graphics drivers.

[theconnactic]

I don't know if it's related, but i had 10.8.3 (now i'm on 10.8.4) installed in my ION system (The Asus netbook of my sig). Since i think ION and 9400m are basically the same silicon, i don't get why the XPS would have graphics issues with the updates.

All the best!

[bcc9]

theconnactic, on 30 April 2013 - 07:45 AM, said:

I don't know if it's related, but i had 10.8.3 (now i'm on 10.8.4) installed in my ION system (The Asus netbook of my sig). Since i think ION and 9400m are basically the same silicon, i don't get why the XPS would have graphics issues with the updates.

All the best!

Google shows a bunch of users (even genuine mac users) having trouble with their nvidia graphics going from 10.8.2->10.8.3, but I don't see much in common.  You're probably using slightly different injection strings than me; and maybe the cuda drivers are now somehow required to keep the nvidia driver working on these old systems?

[theconnactic]

I don't think ION/9400m has any CUDA cores, so CUDA drivers should neither work nor make a difference. I do inject it via DSDT - yet i need GraphicsEnabler=Yes anyway: without both GE=Y and the DSDT, the system won't start.

You can try to examine the DSDTs available on the 1201n topic to find if any suits your hardware: good luck!

All the best!

[bcc9]

theconnactic, on 30 April 2013 - 05:21 PM, said:

I don't think ION/9400m has any CUDA cores, so CUDA drivers should neither work nor make a difference. I do inject it via DSDT - yet i need GraphicsEnabler=Yes anyway: without both GE=Y and the DSDT, the system won't start.

You can try to examine the DSDTs available on the 1201n topic to find if any suits your hardware: good luck!

All the best!

Cuda drivers install on the 9400m and nvidia's web page show the 9400m as a supported platform.
Some users have claimed installing the cuda drivers to be a necessary step to get working video, but I've never experienced it.
I just tried the 10.8.3 upgrade again and cuda made no difference for my 10.8.3 failure case.

I don't need help on what the injection strings are, I've been supporting 9400m&9500m users here for years with regards to that.  Just wondering if anyone knew what changed with 10.8.3 that would cause this failure.

[theconnactic]

bcc9, on 30 April 2013 - 06:12 PM, said:

I don't need help on what the injection strings are, I've been supporting 9400m&9500m users here for years with regards to that.  Just wondering if anyone knew what changed with 10.8.3 that would cause this failure.

Sorry, bbc9, but how could i know? To me, you were just someone asking help, and i tried my best. But i just realized you're the one that wrote the HDA script, so obviously any DSDT injection would be a piece of cake for you. My bad. (OFF: by the way, do you know directions for an AppleHDA patching guide from where i can learn to patch it manually, completely from scratch, for my VT 1705, since it's not supported by any current patch?)

And you are right about CUDA and the 9400m, despite the nVidia site stating 16 "stream cores" (not CUDA cores) for the 9400m: just downloaded the latest CUDA drivers, which i found were missing in my system, and CUDA-Z now displays the CUDA cores correctly (picture below). But then again, i've been running OSX on my 1201n all the way from 10.7.0 to 10.8.4 beta without a single graphics issue, so the issue must lie somewhere else. I'd say, risking to be a little futile, to test the injections of the 1201n topic.

All the best!

 Screen Shot 2013-04-30 at 5.02.53 PM.png   90.83K   3 downloads

[bcc9]

bcc9, on 30 April 2013 - 06:44 AM, said:

I can remove some of the nvidia graphics drivers and get the system to lock up right before the screen goes black instead, or I can get the system to panic with IOPPlatformPluginLegacy in the traceback if I remove all the nvidia graphics drivers.

So this panic message was the clue.  Prior to 10.8.3, I could boot with nv_disable=1 and that would disable the nvidia graphics and the system would come up with vga graphics.  With 10.8.3, I'd get the panic with IOPPlatformPluginLegacy and ACPI_SMC_PlatformPlugin in the traceback.

So, I moved ACPI_SMC_PlatformPlugin out of /S/L/E and viola no more panic, and also I can boot with the 10.8.3 nvidia drivers.  No more hang.

[bcc9]

theconnactic, on 30 April 2013 - 08:02 PM, said:

Sorry, bbc9, but how could i know? To me, you were just someone asking help, and i tried my best. But i just realized you're the one that wrote the HDA script, so obviously any DSDT injection would be a piece of cake for you. My bad. (OFF: by the way, do you know directions for an AppleHDA patching guide from where i can learn to patch it manually, completely from scratch, for my VT 1705, since it's not supported by any current patch?)

No worries.  The HDA script was somewhat simple actually compared to some of the other reverse engineering I've had to do (like ATI graphics, intel hd graphics, nvidia HDMI audio, etc.).
In any case, AppleHDA worked for the VT2021 I tried, so I suspect VT1705 would work too.  The projectosx thread I referenced in my applehda thread,  http://www.projectos...owtopic=465��is the most complete/current instructions I know of.  There are other older somewhat obsolete pieces such as the pincap under linux thread: http://www.insanelym...ded&pid=1055645
These days you may be able to crib your pin configuration from voodoohda instead of using linux.

[bcc9]

bcc9, on 30 April 2013 - 11:36 PM, said:

So, I moved ACPI_SMC_PlatformPlugin out of /S/L/E and viola no more panic, and also I can boot with the 10.8.3 nvidia drivers.  No more hang.

Digging further, it was 9400m_gstate_inject that was triggering the problems.  Looks like mapping this system's AGPM g-state to control-id 16 no longer works.  Everything is fine without 9400m_gstate_inject (and I no longer need to break  ACPI_SMC_PlatformPlugin).