Commit b1dc7a72 authored by Andy Lutomirski's avatar Andy Lutomirski Committed by Sasha Levin

x86/nmi: Enable nested do_nmi() handling for 64-bit kernels

[ Upstream commit 9d050416 ]

32-bit kernels handle nested NMIs in C.  Enable the exact same
handling on 64-bit kernels as well.  This isn't currently
necessary, but it will become necessary once the asm code starts
allowing limited nesting.
Signed-off-by: default avatarAndy Lutomirski <luto@kernel.org>
Reviewed-by: default avatarSteven Rostedt <rostedt@goodmis.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Signed-off-by: default avatarIngo Molnar <mingo@kernel.org>
Signed-off-by: default avatarSasha Levin <sasha.levin@oracle.com>
parent b8d1382c
...@@ -408,15 +408,15 @@ static void default_do_nmi(struct pt_regs *regs) ...@@ -408,15 +408,15 @@ static void default_do_nmi(struct pt_regs *regs)
NOKPROBE_SYMBOL(default_do_nmi); NOKPROBE_SYMBOL(default_do_nmi);
/* /*
* NMIs can hit breakpoints which will cause it to lose its * NMIs can hit breakpoints which will cause it to lose its NMI context
* NMI context with the CPU when the breakpoint does an iret. * with the CPU when the breakpoint or page fault does an IRET.
*/ *
#ifdef CONFIG_X86_32 * As a result, NMIs can nest if NMIs get unmasked due an IRET during
/* * NMI processing. On x86_64, the asm glue protects us from nested NMIs
* For i386, NMIs use the same stack as the kernel, and we can * if the outer NMI came from kernel mode, but we can still nest if the
* add a workaround to the iret problem in C (preventing nested * outer NMI came from user mode.
* NMIs if an NMI takes a trap). Simply have 3 states the NMI *
* can be in: * To handle these nested NMIs, we have three states:
* *
* 1) not running * 1) not running
* 2) executing * 2) executing
...@@ -430,15 +430,14 @@ NOKPROBE_SYMBOL(default_do_nmi); ...@@ -430,15 +430,14 @@ NOKPROBE_SYMBOL(default_do_nmi);
* (Note, the latch is binary, thus multiple NMIs triggering, * (Note, the latch is binary, thus multiple NMIs triggering,
* when one is running, are ignored. Only one NMI is restarted.) * when one is running, are ignored. Only one NMI is restarted.)
* *
* If an NMI hits a breakpoint that executes an iret, another * If an NMI executes an iret, another NMI can preempt it. We do not
* NMI can preempt it. We do not want to allow this new NMI * want to allow this new NMI to run, but we want to execute it when the
* to run, but we want to execute it when the first one finishes. * first one finishes. We set the state to "latched", and the exit of
* We set the state to "latched", and the exit of the first NMI will * the first NMI will perform a dec_return, if the result is zero
* perform a dec_return, if the result is zero (NOT_RUNNING), then * (NOT_RUNNING), then it will simply exit the NMI handler. If not, the
* it will simply exit the NMI handler. If not, the dec_return * dec_return would have set the state to NMI_EXECUTING (what we want it
* would have set the state to NMI_EXECUTING (what we want it to * to be when we are running). In this case, we simply jump back to
* be when we are running). In this case, we simply jump back * rerun the NMI handler again, and restart the 'latched' NMI.
* to rerun the NMI handler again, and restart the 'latched' NMI.
* *
* No trap (breakpoint or page fault) should be hit before nmi_restart, * No trap (breakpoint or page fault) should be hit before nmi_restart,
* thus there is no race between the first check of state for NOT_RUNNING * thus there is no race between the first check of state for NOT_RUNNING
...@@ -461,49 +460,36 @@ enum nmi_states { ...@@ -461,49 +460,36 @@ enum nmi_states {
static DEFINE_PER_CPU(enum nmi_states, nmi_state); static DEFINE_PER_CPU(enum nmi_states, nmi_state);
static DEFINE_PER_CPU(unsigned long, nmi_cr2); static DEFINE_PER_CPU(unsigned long, nmi_cr2);
#define nmi_nesting_preprocess(regs) \ #ifdef CONFIG_X86_64
do { \
if (this_cpu_read(nmi_state) != NMI_NOT_RUNNING) { \
this_cpu_write(nmi_state, NMI_LATCHED); \
return; \
} \
this_cpu_write(nmi_state, NMI_EXECUTING); \
this_cpu_write(nmi_cr2, read_cr2()); \
} while (0); \
nmi_restart:
#define nmi_nesting_postprocess() \
do { \
if (unlikely(this_cpu_read(nmi_cr2) != read_cr2())) \
write_cr2(this_cpu_read(nmi_cr2)); \
if (this_cpu_dec_return(nmi_state)) \
goto nmi_restart; \
} while (0)
#else /* x86_64 */
/* /*
* In x86_64 things are a bit more difficult. This has the same problem * In x86_64, we need to handle breakpoint -> NMI -> breakpoint. Without
* where an NMI hitting a breakpoint that calls iret will remove the * some care, the inner breakpoint will clobber the outer breakpoint's
* NMI context, allowing a nested NMI to enter. What makes this more * stack.
* difficult is that both NMIs and breakpoints have their own stack.
* When a new NMI or breakpoint is executed, the stack is set to a fixed
* point. If an NMI is nested, it will have its stack set at that same
* fixed address that the first NMI had, and will start corrupting the
* stack. This is handled in entry_64.S, but the same problem exists with
* the breakpoint stack.
* *
* If a breakpoint is being processed, and the debug stack is being used, * If a breakpoint is being processed, and the debug stack is being
* if an NMI comes in and also hits a breakpoint, the stack pointer * used, if an NMI comes in and also hits a breakpoint, the stack
* will be set to the same fixed address as the breakpoint that was * pointer will be set to the same fixed address as the breakpoint that
* interrupted, causing that stack to be corrupted. To handle this case, * was interrupted, causing that stack to be corrupted. To handle this
* check if the stack that was interrupted is the debug stack, and if * case, check if the stack that was interrupted is the debug stack, and
* so, change the IDT so that new breakpoints will use the current stack * if so, change the IDT so that new breakpoints will use the current
* and not switch to the fixed address. On return of the NMI, switch back * stack and not switch to the fixed address. On return of the NMI,
* to the original IDT. * switch back to the original IDT.
*/ */
static DEFINE_PER_CPU(int, update_debug_stack); static DEFINE_PER_CPU(int, update_debug_stack);
#endif
static inline void nmi_nesting_preprocess(struct pt_regs *regs) dotraplinkage notrace __kprobes void
do_nmi(struct pt_regs *regs, long error_code)
{ {
if (this_cpu_read(nmi_state) != NMI_NOT_RUNNING) {
this_cpu_write(nmi_state, NMI_LATCHED);
return;
}
this_cpu_write(nmi_state, NMI_EXECUTING);
this_cpu_write(nmi_cr2, read_cr2());
nmi_restart:
#ifdef CONFIG_X86_64
/* /*
* If we interrupted a breakpoint, it is possible that * If we interrupted a breakpoint, it is possible that
* the nmi handler will have breakpoints too. We need to * the nmi handler will have breakpoints too. We need to
...@@ -514,22 +500,8 @@ static inline void nmi_nesting_preprocess(struct pt_regs *regs) ...@@ -514,22 +500,8 @@ static inline void nmi_nesting_preprocess(struct pt_regs *regs)
debug_stack_set_zero(); debug_stack_set_zero();
this_cpu_write(update_debug_stack, 1); this_cpu_write(update_debug_stack, 1);
} }
}
static inline void nmi_nesting_postprocess(void)
{
if (unlikely(this_cpu_read(update_debug_stack))) {
debug_stack_reset();
this_cpu_write(update_debug_stack, 0);
}
}
#endif #endif
dotraplinkage notrace void
do_nmi(struct pt_regs *regs, long error_code)
{
nmi_nesting_preprocess(regs);
nmi_enter(); nmi_enter();
inc_irq_stat(__nmi_count); inc_irq_stat(__nmi_count);
...@@ -539,8 +511,17 @@ do_nmi(struct pt_regs *regs, long error_code) ...@@ -539,8 +511,17 @@ do_nmi(struct pt_regs *regs, long error_code)
nmi_exit(); nmi_exit();
/* On i386, may loop back to preprocess */ #ifdef CONFIG_X86_64
nmi_nesting_postprocess(); if (unlikely(this_cpu_read(update_debug_stack))) {
debug_stack_reset();
this_cpu_write(update_debug_stack, 0);
}
#endif
if (unlikely(this_cpu_read(nmi_cr2) != read_cr2()))
write_cr2(this_cpu_read(nmi_cr2));
if (this_cpu_dec_return(nmi_state))
goto nmi_restart;
} }
NOKPROBE_SYMBOL(do_nmi); NOKPROBE_SYMBOL(do_nmi);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment