heap grower PANIC

[mvfranz]

While building gcc 4.3.3 on R1 Alpha the following kernel panic occurred.

PANIC: ASSERT FAILED(/BOM/haiku/haiku/src/system/kernel/vm/vm_page_.cpp:1842): sReservedPages >= count

I looked at the backtrace but it looked like it was in the debugger and did not tell me more information than the like of the kernel code that caused the problem.

[mmlr]

Likely another instance of the address space overflow. If this is the case it should be fixed in the trunk. A backtrace would be helpful to determine where exactly it failed to be able to confirm the likelihood of it being the overflow. Just to make sure btw, you realize that GCC4 is already installed in the alpha release and you don't have to build it yourself?

[mvfranz]

I wrote down some of the backtrace, but it was very tedious. I did a quick search to see if there was way to get the bt off my system, but did not find a way. Next time I will take a picture.

I know gcc4 is on the system. I am testing some stuff and want to ensure that I can build gcc 4.3.3. The build seems to create a good stress test for the system.

[anevilyak]

You can either take a picture or get it across serial if you have the requisite cable, ports and a second machine.

[mvfranz]

Kernel Panic backtrace -second time

[mvfranz]

Since I have been using Haiku pretty much every day for development I have found that this happens every day. If I let my pc run over night there is a 100% chance that I will have this PANIC in the morning. This happens on both a VMware image and my real hardware.

What is interesting as nothing is running at the time. My builds have finished and Haiku is just sitting idle.

[mmlr]

Does this still happen? If it was caused by address space overflows it should be fixed for quite some time now.

[mmlr]

For lack of feedback I'm going to close this under the assumption that it was an address space overflow which was fixed in the meantime, please reopen or create a new ticket if seen again.