Re: [PATCH V2 0/3] riscv: Enable LOCKDEP

From: Guo Ren
Date: Thu Jul 09 2020 - 19:15:47 EST


Thank you, Palmer

On Fri, Jul 10, 2020 at 6:06 AM Palmer Dabbelt <palmerdabbelt@xxxxxxxxxx> wrote:
>
> On Sat, 27 Jun 2020 06:57:05 PDT (-0700), guoren@xxxxxxxxxx wrote:
> > From: Guo Ren <guoren@xxxxxxxxxxxxxxxxx>
> >
> > Lockdep is needed by proving the spinlocks and rwlocks. To support it,
> > we need to add TRACE_IRQFLAGS codes in kernel/entry.S. These patches
> > follow Documentation/irqflags-tracing.txt.
> >
> > Fixup 2 bugs that block the lockdep implementation.
> >
> > ---
> > Changes in v2
> > - Remove sX regs recovery codes which are unnecessary, because
> > callee will handle them. Thx Greentime :)
> >
> > - Move "restore a0 - a7" to handle_syscall, but if _TIF_SYSCALL_WORK
> > is set, "restore a1 - a7" is still duplicated. I prefer a C wrapper
> > for syscall.
> >
> > Guo Ren (2):
> > riscv: Fixup static_obj() fail
> > riscv: Enable LOCKDEP_SUPPORT & fixup TRACE_IRQFLAGS_SUPPORT
> >
> > Zong Li (1):
> > riscv: Fixup lockdep_assert_held with wrong param cpu_running
> >
> > arch/riscv/Kconfig | 3 +++
> > arch/riscv/kernel/entry.S | 33 ++++++++++++++++++++++++++++++++-
> > arch/riscv/kernel/smpboot.c | 1 -
> > arch/riscv/kernel/vmlinux.lds.S | 2 +-
> > 4 files changed, 36 insertions(+), 3 deletions(-)
>
> These are on for-next. As far as I can tell lockdep is working, but I'm just
> doing some simple boot tests.
>
> Thanks!



--
Best Regards
Guo Ren

ML: https://lore.kernel.org/linux-csky/