On 12/10/2016 07:57 AM, Jessica Hamilton wrote:
On 10 December 2016 at 12:18, Ingo Weinhold <ingo_weinhold@xxxxxx> wrote:
As Fredrik already pointed out, the different load addresses are due to
ASLR. It may or may not be useful to disable it to get reproducible results.
For debugging early userland crashes I'd recommend panic()ing in
vm_page_fault() (turn dprintf() into panic() [1]). This way you have at
least the kernel debugging facilities at your disposal to inspect the
crashed team.
[1]
http://cgit.haiku-os.org/haiku/tree/src/system/kernel/vm/vm.cpp?id=19d8d9fa3983e6687caeab8fe1193861a5202ab6#n4283
I had added a call to kernel_debugger in the function that attempts to
start the userland debugger, but this works too :)
Having access to the kernel debugger is one thing, knowing what to
look for is another...
I haven't tried that yet, but in theory it should also be possible to write
a core dump [2] at that point. You'll need Haiku's Debugger to analyze it --
the format is only marginally compatible with Linux/BSD core dumps.
[2]
http://cgit.haiku-os.org/haiku/tree/headers/private/kernel/core_dump.h?id=19d8d9fa3983e6687caeab8fe1193861a5202ab6#n14
This doesn't seem to be available from the kernel debugger...