powerpc/64s/interrupt: Perf NMI should not take normal exit path

NMI interrupts should exit with EXCEPTION_RESTORE_REGS not with
interrupt_return_srr, which is what the perf NMI handler currently does.
This breaks if a PMI hits after interrupt_exit_user_prepare_main() has
switched the context tracking to user mode, then the CT_WARN_ON() in
interrupt_exit_kernel_prepare() fires because it returns to kernel with
context set to user.

This could possibly be solved by soft-disabling PMIs in the exit path,
but that reduces our ability to profile that code. The warning could be
removed, but it's potentially useful.

All other NMIs and soft-NMIs return using EXCEPTION_RESTORE_REGS, so
this makes perf interrupts consistent with that and seems like the best
fix.

Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Squash in fixups from Nick]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20221006140413.126443-3-npiggin@gmail.com
This commit is contained in:
Nicholas Piggin 2022-10-07 00:04:11 +10:00 committed by Michael Ellerman
parent a073672eb0
commit dc398a084d
2 changed files with 21 additions and 7 deletions

View File

@ -2357,9 +2357,21 @@ EXC_VIRT_END(performance_monitor, 0x4f00, 0x20)
EXC_COMMON_BEGIN(performance_monitor_common)
GEN_COMMON performance_monitor
addi r3,r1,STACK_FRAME_OVERHEAD
bl performance_monitor_exception
lbz r4,PACAIRQSOFTMASK(r13)
cmpdi r4,IRQS_ENABLED
bne 1f
bl performance_monitor_exception_async
b interrupt_return_srr
1:
bl performance_monitor_exception_nmi
/* Clear MSR_RI before setting SRR0 and SRR1. */
li r9,0
mtmsrd r9,1
kuap_kernel_restore r9, r10
EXCEPTION_RESTORE_REGS hsrr=0
RFI_TO_KERNEL
/**
* Interrupt 0xf20 - Vector Unavailable Interrupt.

View File

@ -377,13 +377,15 @@ notrace unsigned long interrupt_exit_kernel_prepare(struct pt_regs *regs)
* CT_WARN_ON comes here via program_check_exception, so avoid
* recursion.
*
* Skip the assertion on PMIs to work around a problem caused by NMI
* PMIs incorrectly taking this interrupt return path, it's possible
* for this to hit after interrupt exit to user switches context to
* user. See also the comment in the performance monitor handler in
* exceptions-64e/s.S
* Skip the assertion on PMIs on 64e to work around a problem caused
* by NMI PMIs incorrectly taking this interrupt return path, it's
* possible for this to hit after interrupt exit to user switches
* context to user. See also the comment in the performance monitor
* handler in exceptions-64e.S
*/
if (TRAP(regs) != INTERRUPT_PROGRAM && TRAP(regs) != INTERRUPT_PERFMON)
if (!IS_ENABLED(CONFIG_PPC_BOOK3E_64) &&
TRAP(regs) != INTERRUPT_PROGRAM &&
TRAP(regs) != INTERRUPT_PERFMON)
CT_WARN_ON(ct_state() == CONTEXT_USER);
kuap = kuap_get_and_assert_locked();