I planed to do similar generic entry transaction as Guo Ren did[1], and
I had some commits in local. Since Guo has sent out the series, I
dropped my version and just provide those in my local repo but missing
in Guo's series. However, this doesn't mean this series depends on
Guo's series, in fact except the first one, the remaining three patches
are independent on generic entry.

[1]https://lore.kernel.org/linux-riscv/20220918155246.1203293-1-guoren@kernel.org/T/#t

Since v1:
  - consolidate gp saving/restoring in mcount-dyn.S as well
  - avoid non-trival memory waste if NR_CPUs is large
  - collect Reviewed-by tag.

Hi Guo,

I only added your reviewed-by tag to patch1, but dropped tag for patch3
and patch4 because the patch changed a bit.

Thanks

Jisheng Zhang (4):
  riscv: remove extra level wrappers of trace_hardirqs_{on,off}
  riscv: consolidate ret_from_kernel_thread into ret_from_fork
  riscv: fix race when vmap stack overflow and remove shadow_stack
  riscv: entry: consolidate general regs saving into save_gp

 arch/riscv/include/asm/asm-prototypes.h |   1 -
 arch/riscv/include/asm/asm.h            |  65 ++++++++++
 arch/riscv/include/asm/thread_info.h    |   4 +-
 arch/riscv/kernel/Makefile              |   2 -
 arch/riscv/kernel/asm-offsets.c         |   1 +
 arch/riscv/kernel/entry.S               | 154 +++---------------------
 arch/riscv/kernel/mcount-dyn.S          |  58 +--------
 arch/riscv/kernel/process.c             |   5 +-
 arch/riscv/kernel/trace_irq.c           |  27 -----
 arch/riscv/kernel/trace_irq.h           |  11 --
 arch/riscv/kernel/traps.c               |  31 +++--
 11 files changed, 104 insertions(+), 255 deletions(-)
 delete mode 100644 arch/riscv/kernel/trace_irq.c
 delete mode 100644 arch/riscv/kernel/trace_irq.h

-- 
2.34.1