Re: [RFC][PATCH 4/7] x86/debug: Move historical SYSENTER junk into exc_debug_kernel()

[Daniel Thompson]

On Fri, Aug 21, 2020 at 12:19:03PM +0200, peterz@xxxxxxxxxxxxx wrote:
> On Thu, Aug 20, 2020 at 04:28:28PM +0100, Daniel Thompson wrote:
> 
> > Specifically I've entered the kdb in pretty much the simplest way
> > possible: a direct call to kgdb_breakpoint() from a task context. I
> > generate a backtrace to illustrate this, just to give you a better
> > understanding of what might be happening) and then ran the single step.
> 
> 
> > + make -C .. O=$PWD x86_64_defconfig
> > + ../scripts/config --enable RUNTIME_TESTING_MENU
> > + ../scripts/config --enable DEBUG_INFO --enable DEBUG_FS --enable KALLSYMS_ALL --enable MAGIC_SYSRQ --enable KGDB --enable KGDB_TESTS --enable KGDB_KDB --enable KDB_KEYBOARD --enable LKDTM
> > + ../scripts/config --enable PROVE_LOCKING --enable DEBUG_ATOMIC_SLEEP
> > + make olddefconfig
> 
> 
> That asked me about a kgdb boottime selftest, which I figured was a good
> idea, but even without my patches that seems to fail, and fail so that
> it doesn't boot :/

If you are running to fork test (the F100 in the default test string of
kgdbts=V1F100 kgdbwait) then you will need rodata=off to avoid problems
setting or clearing breakpoints (you also need to have at last 100 calls
to fork for the test suite to complete; a modern kernel without the
excessive forking for hotplug will usually make it well into userspace
before racking up this many forks).

This will prevent the test suite from hanging the boot although I think
you will still see the still see the stacktrace from the hardware
watchpoint test (this is unrelated to current patchset).


Daniel.