Re: [PATCH V3 4/7] riscv: convert to generic entry

From: Guo Ren <guoren@kernel.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: arnd@arndb.de, palmer@rivosinc.com, tglx@linutronix.de,
	luto@kernel.org,  conor.dooley@microchip.com, heiko@sntech.de,
	jszhang@kernel.org,  lazyparser@gmail.com, falcon@tinylab.org,
	chenhuacai@kernel.org,  apatel@ventanamicro.com,
	atishp@atishpatra.org, palmer@dabbelt.com,
	 paul.walmsley@sifive.com, bigeasy@linutronix.de,
	linux-arch@vger.kernel.org,  linux-kernel@vger.kernel.org,
	linux-riscv@lists.infradead.org,
	 Guo Ren <guoren@linux.alibaba.com>
Subject: Re: [PATCH V3 4/7] riscv: convert to generic entry
Date: Wed, 7 Sep 2022 08:59:20 +0800	[thread overview]
Message-ID: <CAJF2gTTqVc_CvxT+t4D2Z1Gy_r_nrbdgAGcELFm9tgjOaCyJYg@mail.gmail.com> (raw)
In-Reply-To: <YxcQ6NoPf3AH0EXe@hirez.programming.kicks-ass.net>

On Tue, Sep 6, 2022 at 5:20 PM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Mon, Sep 05, 2022 at 11:54:20PM -0400, guoren@kernel.org wrote:
>
> > +asmlinkage void noinstr do_riscv_irq(struct pt_regs *regs)
> > +{
> > +     struct pt_regs *old_regs;
> > +     irqentry_state_t state = irqentry_enter(regs);
> > +
> > +     irq_enter_rcu();
> > +     old_regs = set_irq_regs(regs);
> > +     handle_arch_irq(regs);
> > +     set_irq_regs(old_regs);
> > +     irq_exit_rcu();
> > +
> > +     irqentry_exit(regs, state);
> > +}
>
> The above is right in that everything that calls irqentry_enter() should
> be noinstr; however all the below instances get it wrong:
>
> >  #define DO_ERROR_INFO(name, signo, code, str)                                \
> >  asmlinkage __visible __trap_section void name(struct pt_regs *regs)  \
> >  {                                                                    \
> > +     irqentry_state_t state = irqentry_enter(regs);                  \
> >       do_trap_error(regs, signo, code, regs->epc, "Oops - " str);     \
> > +     irqentry_exit(regs, state);                                     \
> >  }
> >
> >  DO_ERROR_INFO(do_trap_unknown,
> > @@ -123,18 +126,22 @@ int handle_misaligned_store(struct pt_regs *regs);
> >
> >  asmlinkage void __trap_section do_trap_load_misaligned(struct pt_regs *regs)
> >  {
> > +     irqentry_state_t state = irqentry_enter(regs);
> >       if (!handle_misaligned_load(regs))
> >               return;
> >       do_trap_error(regs, SIGBUS, BUS_ADRALN, regs->epc,
> >                     "Oops - load address misaligned");
> > +     irqentry_exit(regs, state);
> >  }
> >
> >  asmlinkage void __trap_section do_trap_store_misaligned(struct pt_regs *regs)
> >  {
> > +     irqentry_state_t state = irqentry_enter(regs);
> >       if (!handle_misaligned_store(regs))
> >               return;
> >       do_trap_error(regs, SIGBUS, BUS_ADRALN, regs->epc,
> >                     "Oops - store (or AMO) address misaligned");
> > +     irqentry_exit(regs, state);
> >  }
> >  #endif
> >  DO_ERROR_INFO(do_trap_store_fault,
> > @@ -158,6 +165,8 @@ static inline unsigned long get_break_insn_length(unsigned long pc)
> >
> >  asmlinkage __visible __trap_section void do_trap_break(struct pt_regs *regs)
> >  {
> > +     irqentry_state_t state = irqentry_enter(regs);
> > +
> >  #ifdef CONFIG_KPROBES
> >       if (kprobe_single_step_handler(regs))
> >               return;
> > @@ -185,6 +194,8 @@ asmlinkage __visible __trap_section void do_trap_break(struct pt_regs *regs)
> >               regs->epc += get_break_insn_length(regs->epc);
> >       else
> >               die(regs, "Kernel BUG");
> > +
> > +     irqentry_exit(regs, state);
> >  }
> >  NOKPROBE_SYMBOL(do_trap_break);
>
> > +asmlinkage void do_page_fault(struct pt_regs *regs)
> > +{
> > +     irqentry_state_t state = irqentry_enter(regs);
> > +
> > +     __do_page_fault(regs);
> > +
> > +     irqentry_exit(regs, state);
> > +}
> >  NOKPROBE_SYMBOL(do_page_fault);
>
> Without noinstr the compiler is free to insert instrumentation (think
> all the k*SAN, KCov, GCov, ftrace etc..) which can call code we're not
> yet ready to run this early in the entry path, for instance it could
> rely on RCU which isn't on yet, or expect lockdep state.
I'll add a patch to fix it in the next version. Thx for pointing it out.

>
>


-- 
Best Regards
 Guo Ren

_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv