https://github.com/DavidSpickett commented:

You keep saying security but if you're debugging a running kernel surely most 
bets are off at that point?

I think "safety" might be a better term. Safety covers breaking the running 
kernel, and kinda implies security, but not so much that we'd have to prove 
that debugging a running kernel is "secure" by whatever definition.

Our error handling is not very principled so I suspect there are paths that 
would hide this error. I don't think it's worth auditing all that though, as 
long as something simple like assigning to a variable, or `memory write ...` 
surfaces it. That's the sort of thing I'd try if something more complex failed 
without explanation.

In other words: please confirm there's at least 1 top level lldb command that 
can produce the error message.

> Implement ProcessFreeBSDKernelCore::DoWriteMemory() to write data on kernel 
> dump or /dev/mem.

`/dev/mem` is a window into the running kernel, so I see the utility of writing 
to that.

For kernel dumps, what's the use case? Aren't these offline files akin to elf 
cores?

https://github.com/llvm/llvm-project/pull/183237
_______________________________________________
lldb-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits

Reply via email to