lol

I'm reading this chapter because it began with this master class of snark that made me say "gosh maybe I don't know all the basics, I'd better not skip it"

@brenns10 I need to put this in mine

@ljs you don't want your readers to get too cocky

@brenns10 lol the second they hit the anon VMA stuff all such feeling will disappear and be replaced with despair

@ljs @brenns10 did you see that anon_vma ->degree confusion bug that was fixed in 2022, where an assumption in vma mergability checks is broken and you get an anon page mapped in a VMA which is not connected to the page's anon_vma, and that leads to anon_vma UAF?
https://project-zero.issues.chromium.org/issues/42451486

@brenns10 which book, btw?

@jann @ljs @brenns10 my colleague managed to create a livepatch that somehow avoided adding the new fields, and I've reviewed it as that it should be working, but forgot all the details since ;)

@ljs @brenns10 one of those bug reports that probably involved me drawing a bunch of possible scenarios on paper...
the reproducer requires five fork() calls, plus some VMA splitting and merging

@vbabka @brenns10 @ljs oh yes, this patch was how I learned that security fixes that change struct layouts can be very annoying for people

@jann @ljs @brenns10 ha, I believe I was discarding these kind of my own drawings among other stuff just week ago

@jann @brenns10 fucking hell Jann why are you like this

Oh wait this is like cotton candy for exploit writers sigh.

Well at least I didn't take on maintainership responsibility recently... Oh wait FUCK

@jann @brenns10 merging much nicer now due to some genius

@vbabka It's the early access PDF for "Building a Debugger". So far so good! I've only just gotten to chapter 3 but clearly I enjoy the author's sense of humor which makes it easier reading. I'm excited to get to some of the DWARF handling chapters especially.

https://nostarch.com/building-a-debugger

@brenns10 @vbabka According to the Table of Contents, the kernel support for debuggers (gdbserver I guess) isn't part of the content, right? I've always been interested in learning more, but I never took a look.

@brenns10 @vbabka How is the chapter about DWARF?

@andersonc0d3
@vbabka Yeah it's definitely not a book focusing on debuggers for kernels. There's a small section in chapter 22 about remote debugging but it's just a "FYI this exists" sort of thing.

If you did want to see an example of gdbserver support in a kernel debugger you could check this new pull request for gdbserver support in drgn, hot off the presses from Daniel Thompson, kgdb co-maintainer :)

https://github.com/osandov/drgn/pull/444

@brenns10 @vbabka Thanks, Stephen. I am going to check it out.

@jarkko @vbabka I'll let you know when I get there!

@brenns10 @vbabka I have to give kudos to anyone who dares to make any sense of DWARF :-)

@jarkko @brenns10 my impression is that these days people rather invent new stuff to avoid using dwarf where possible ;)

@vbabka @jarkko @brenns10 honestly I prefer being called a 'short king' guys

@ljs @brenns10 @jarkko @vbabka it's 'short and thin king' now

@ljs @brenns10 @jarkko that means "short but needs a king size bed"?

@vbabka @brenns10 @jarkko oi!!!

Super king fyi

@ljs @brenns10 @jarkko do you use it orthogonally

@vbabka @brenns10 @jarkko you're the worst human being alive

@ljs @brenns10 @jarkko @vbabka worst human, best ape!

@vbabka @brenns10 ya it's sick, something like ACPI feels cleanest thing ever engineered compared to DWARF