Kernel Panics in 1.9?

I've had two complete machine shutdowns since I started using 1.9.

Both times I was just sitting on the ground in IM. Nothing fancy.

Same here. Last night was the first time I was on SL for any length of time since the latest upgrade. My system crashed hard, twice, during the evening. 10.4.5 with all updates, here, on an iMac G5.

SecondLife ran great until 1.9.0. The update to 1.9.0 (17) doesn't seem any different than the first 1.9.0 client. They're crashing my Mac hard (dim screen and frozen with a message that I must force reboot), frequently. I've had about 9 or 10 crashes in the last three days. This system has only crashed that way once in the prior 18 months I've owned it. It's making SL very frustrating to try to use.

This happens sporaidcally. I reinstalled OS X, but continued to crash anyway.

Also running 10.4.5 on an early iMac5:

GeForce FX 5200:

Chipset Model: GeForce FX 5200
Type: Display
Bus: AGP
Slot: AGP
VRAM (Total): 64 MB
Vendor: nVIDIA (0x10de)
Device ID: 0x0329
Revision ID: 0x00b1
ROM Revision: 2103
Displays:
iMac:
Display Type: LCD
Resolution: 1680 x 1050
Depth: 32-bit Color
Built-In: Yes
Core Image: Supported
Main Display: Yes
Mirror: Off
Online: Yes
Quartz Extreme: Supported
Display:
Status: No display connected

You all probably already know this but I'll repost it anyway...

/164/0d/73940/1.html#post799365

For a bit of balance, I haven't had a panic for many months and the cause was a Logitech mouse driver.

Try looking in /Library/Logs and ~/Library/Logs for the panic.log it might help you identify a problem.

Just because a panic suddenly starts happening when you use an app, doesn't always mean the app is at fault. Often it's because the app is using hardware or driver resources that are not used in your normal day to day use.

I'm not saying SL isn't the cause, just that you should suspect other causes too.

I've run temperature monitors while running SL, and while they rise initially, once the fans rev up they drop a bit, and more or less hold that for hours. There are no peaks either at the CPU or the sensor in the upper part of the enclosure correlated with the crashes.

I've run the hardware tests from both the hardware test CD and Techtools Deluxe, as soon after a crash as possible, with no hardware problems found.

The panic logs are all pretty similar. In particular, the PC is always 0xB754 from the start of com.Apple.GeForce, implying that when things go bad, the system is noticing it at the same point of the video driver code. Here's the latest from the log.

Wed Mar 29 00:09:02 2006


Unresolved kernel trap(cpu 0): 0x300 - Data access DAR=0x000000000000000E PC=0x0000000000918754
Latest crash info for cpu 0:
Exception state (sv=0x27A95000)
PC=0x00918754; MSR=0x00009030; DAR=0x0000000E; DSISR=0x40000000; LR=0x009369B8; R1=0x179C3940; XCP=0x0000000C (0x300 - Data access)
Backtrace:
0x00927A98 0x009369B8 0x00918CDC 0x002E7EE0 0x002E9DAC 0x0008C4B0
0x000291C0 0x000233AC 0x000AC02C 0x00000000
backtrace terminated - frame not mapped or invalid: 0xBFFFEAA0

Kernel loadable modules in backtrace (with dependencies):
com.apple.GeForce(4.1.@0x90d000
dependency: com.apple.iokit.IOPCIFamily(1.7)@0x458000
dependency: com.apple.iokit.IOGraphicsFamily(1.4.1)@0x5b9000
dependency: com.apple.iokit.IONDRVSupport(1.4.1)@0x5dd000
dependency: com.apple.NVDAResman(4.1.@0x5f3000
Proceeding back via exception chain:
Exception state (sv=0x27A95000)
previously dumped as "Latest" state. skipping...
Exception state (sv=0x472E1C80)
PC=0x9000B1E8; MSR=0x0000F930; DAR=0x002A5000; DSISR=0x42000000; LR=0x9000B13C; R1=0xBFFFEAA0; XCP=0x00000030 (0xC00 - System call)

Kernel version:
Darwin Kernel Version 8.5.0: Sun Jan 22 10:38:46 PST 2006; root:xnu-792.6.61.obj~1/RELEASE_PPC
panic(cpu 0 caller 0xFFFF0003): 0x300 - Data access
Latest stack backtrace for cpu 0:
Backtrace:
0x00095718 0x00095C30 0x0002683C 0x000A8384 0x000ABD00
Proceeding back via exception chain:
Exception state (sv=0x27A95000)
PC=0x00918754; MSR=0x00009030; DAR=0x0000000E; DSISR=0x40000000; LR=0x009369B8; R1=0x179C3940; XCP=0x0000000C (0x300 - Data access)
Backtrace:
0x00927A98 0x009369B8 0x00918CDC 0x002E7EE0 0x002E9DAC 0x0008C4B0
0x000291C0 0x000233AC 0x000AC02C 0x00000000
backtrace terminated - frame not mapped or invalid: 0xBFFFEAA0

Kernel loadable modules in backtrace (with dependencies):
com.apple.GeForce(4.1.@0x90d000
dependency: com.apple.iokit.IOPCIFamily(1.7)@0x458000
dependency: com.apple.iokit.IOGraphicsFamily(1.4.1)@0x5b9000
dependency: com.apple.iokit.IONDRVSupport(1.4.1)@0x5dd000
dependency: com.apple.NVDAResman(4.1.@0x5f3000
Exception state (sv=0x472E1C80)
PC=0x9000B1E8; MSR=0x0000
*********

Same here. My panic.log looks pretty much like Keirsten Pierterson's.


Tue Mar 28 21:04:28 2006


Unresolved kernel trap(cpu 0): 0x300 - Data access DAR=0x000000000000000E PC=0x000000003609C754
Latest crash info for cpu 0:
Exception state (sv=0x3039F280)
PC=0x3609C754; MSR=0x00009030; DAR=0x0000000E; DSISR=0x40000000; LR=0x360BA9B8; R1=0x17EEB940; XCP=0x0000000C (0x300 - Data access)
Backtrace:
0x360ABA98 0x360BA9B8 0x3609CCDC 0x002E7EE0 0x002E9DAC 0x0008C4B0
0x000291C0 0x000233AC 0x000AC02C 0x00000000
backtrace terminated - frame not mapped or invalid: 0xBFFFE0F0

Kernel loadable modules in backtrace (with dependencies):
com.apple.GeForce(4.1.@0x36091000
dependency: com.apple.iokit.IOPCIFamily(1.7)@0x2803a000
dependency: com.apple.iokit.IOGraphicsFamily(1.4.1)@0x35cfa000
dependency: com.apple.iokit.IONDRVSupport(1.4.1)@0x35d1e000
dependency: com.apple.NVDAResman(4.1.@0x35d34000
Proceeding back via exception chain:
Exception state (sv=0x3039F280)
previously dumped as "Latest" state. skipping...
Exception state (sv=0x304D9500)
PC=0x9000B1E8; MSR=0x0000F930; DAR=0x222B93E4; DSISR=0x42000000; LR=0x9000B13C; R1=0xBFFFE0F0; XCP=0x00000030 (0xC00 - System call)

And, if you hadn't done the math, your crash is at the exact same offset (0xb754) in the same driver. I have an earlyish iMac G5 (it's hard to say how early, since both the midplane and the video system were replaced after separate failures).

No. I didn't do the math, but it *looked* right.

I ordered my iMac G5 the day they were announced -- so it's pretty old as iMac G5s go. And I had the capacitor-of-doom problems, so my midplane has been replaced.

have you tried the Alpha version of the Universal build? just to see if it is any better.... I've been running it, but on an intel Mac, and its been great, no crashes of any kind.

Yes, I tried the UB build, too. It didn't behave any differently than the PowerPC only build for me. The Intel Macs have different video cards, too, which might explain why they're not crashing.

I've asked two of our Mac developers to investigate. I will post to the known issues forum if/when they find something useful!

I noticed one of the systems has only 64 MB of VRAM. That's really, really low. Insufficient RAM or VRAM is probably one of the biggest unrecognized sources of problems with using resource-intensive applications like SL. When I increased the RAM in my G4 PowerBook from 768 MB to 1.25 GB, SL ran a lot better. I use an Intel iMac with 1 GB RAM and 256 MB VRAM now, using the 1.9 alpha Universal Binary, and it runs better and faster than on my dad's high-end HP Windows machine (3.2 MHz, 1 GB RAM).

Thank you, Karen Linden. This problem wasn't happening before 1.9.

Kat, we don't have much choice in the VRAM department. Those of us with iMacs are stuck with whatever is built into our machines. In the case of the early iMac G5s, that was 64 MiB of video RAM. I have 1 GiB of normal RAM.

Seems to be alot of iMac users having this problem. Running a maxed out Dual G5 desktop with 10.4.5 here. Can't remember the last time I experienced the dreaded panic attack - even with SL.

Hope the Mac gurus over there on the Left Coast find a fix soon.

For what it's worth, Keirsten Pierterson and I seem to have identical configurations. (No, I am not Keirsten's alt!) And this was not a problem before 1.9. In fact, in terms of graphics, 1.8 was pretty stable -- the texture flashing problem was very much decreased.

And it still happens in the same spot with OS 10.4.6 applied, too. But that didn't change the GeForce driver.

I'm having the same problem, this really sucks a lot and the problem has been around for ages now. Isn't anybody going to fix it?

I'm basically not going to start SL at all, I really don't like having to force-restart my iMac every so often because of kernel panics.

For a while, I was hopeful that 1.9.21 had gotten rid of this, since I ran a couple times for a few hours with no problem. Then, tonight, I had two kernel panics with the same footprint as previously.

Nothing takes down my Mac like 1.9 does. Hard, hard crashes. I keep sending in those crash reports.

OSX 10.4.6
Dual 2.5 GHz Dual Core PowerPC G5
2.5 GB RAM
256 MB GeForce 7800GT

Has something changed in the last few days with the data the grid is sending to the client? I realized I'd spent more time in SL that usual, lately, and it was because I didn't take the crashes as a sign that I'd had enough now. When I checked my uptime, I see I haven't crashed in over 6 days. (That this is a positive milestone is sad, really, given that I never crashed before running SL.) Nothing has changed on the client side of this, so I'm wondering if maybe the grid is no longer sending whatever caused my client to crash?

I really should have kept my mouth shut. Same old crash happened tonight.