[PATCH v3 11/11] powerpc/64: use interrupt restart table to speed up return from interrupt

[Nicholas Piggin]

Excerpts from Michael Ellerman's message of June 15, 2021 11:44 pm:
> Nicholas Piggin <npiggin at gmail.com> writes:
>> diff --git a/arch/powerpc/lib/feature-fixups.c b/arch/powerpc/lib/feature-fixups.c
>> index fe26f2fa0f3f..fbe94e2d5011 100644
>> --- a/arch/powerpc/lib/feature-fixups.c
>> +++ b/arch/powerpc/lib/feature-fixups.c
>> @@ -412,12 +430,19 @@ void do_entry_flush_fixups(enum l1d_flush_type types)
>>  	stop_machine(__do_entry_flush_fixups, &types, NULL);
>>  }
>>  
>> -void do_rfi_flush_fixups(enum l1d_flush_type types)
>> +static int __do_rfi_flush_fixups(void *data)
>>  {
>> +	enum l1d_flush_type types = *(enum l1d_flush_type *)data;
>>  	unsigned int instrs[3], *dest;
>>  	long *start, *end;
>>  	int i;
>>  
>> +	if (types & L1D_FLUSH_FALLBACK)
>> +		rfi_exit_not_reentrant = true;
>> +	else
>> +		rfi_exit_not_reentrant = false;
>> +	update_interrupt_exit();
> 
> This is not happy:

Ah, needs a static_key_enable_cpuslocked. I'll fix.

Thanks,
Nick

> 
> [    0.000000][    T0] ============================================
> [    0.000000][    T0] WARNING: possible recursive locking detected
> [    0.000000][    T0] 5.13.0-rc2-00118-gca433a3a44e3 #1 Not tainted
> [    0.000000][    T0] --------------------------------------------
> [    0.000000][    T0] swapper/0 is trying to acquire lock:
> [    0.000000][    T0] c00000000252fa10 (cpu_hotplug_lock){....}-{0:0}, at: static_key_enable+0x24/0x50
> [    0.000000][    T0]
> [    0.000000][    T0] but task is already holding lock:
> [    0.000000][    T0] c00000000252fa10 (cpu_hotplug_lock){....}-{0:0}, at: stop_machine+0x2c/0x60
> [    0.000000][    T0]
> [    0.000000][    T0] other info that might help us debug this:
> [    0.000000][    T0]  Possible unsafe locking scenario:
> [    0.000000][    T0]
> [    0.000000][    T0]        CPU0
> [    0.000000][    T0]        ----
> [    0.000000][    T0]   lock(cpu_hotplug_lock);
> [    0.000000][    T0]   lock(cpu_hotplug_lock);
> [    0.000000][    T0]
> [    0.000000][    T0]  *** DEADLOCK ***
> [    0.000000][    T0]
> [    0.000000][    T0]  May be due to missing lock nesting notation
> [    0.000000][    T0]
> [    0.000000][    T0] 1 lock held by swapper/0:
> [    0.000000][    T0]  #0: c00000000252fa10 (cpu_hotplug_lock){....}-{0:0}, at: stop_machine+0x2c/0x60
> [    0.000000][    T0]
> [    0.000000][    T0] stack backtrace:
> [    0.000000][    T0] CPU: 0 PID: 0 Comm: swapper Not tainted 5.13.0-rc2-00118-gca433a3a44e3 #1
> [    0.000000][    T0] Call Trace:
> [    0.000000][    T0] [c0000000027db8f0] [c00000000093dd28] dump_stack+0xec/0x144 (unreliable)
> [    0.000000][    T0] [c0000000027db940] [c0000000001ed5b4] __lock_acquire+0x1744/0x28b0
> [    0.000000][    T0] [c0000000027dba70] [c0000000001ef338] lock_acquire+0x128/0x600
> [    0.000000][    T0] [c0000000027dbb70] [c00000000015035c] cpus_read_lock+0x4c/0x170
> [    0.000000][    T0] [c0000000027dbba0] [c0000000003c2594] static_key_enable+0x24/0x50
> [    0.000000][    T0] [c0000000027dbbd0] [c0000000000ae87c] __do_rfi_flush_fixups+0x7c/0x300
> [    0.000000][    T0] [c0000000027dbc80] [c0000000002ab7e4] stop_machine_cpuslocked+0xe4/0x200
> [    0.000000][    T0] [c0000000027dbcf0] [c0000000002ab940] stop_machine+0x40/0x60
> [    0.000000][    T0] [c0000000027dbd30] [c0000000000aef30] do_rfi_flush_fixups+0x30/0x50
> [    0.000000][    T0] [c0000000027dbd60] [c000000000040890] setup_rfi_flush+0xa0/0x140
> [    0.000000][    T0] [c0000000027dbdd0] [c00000000201c6c4] pnv_setup_arch+0x304/0x4ac
> [    0.000000][    T0] [c0000000027dbe60] [c00000000200a31c] setup_arch+0x374/0x3c4
> [    0.000000][    T0] [c0000000027dbee0] [c000000002003d08] start_kernel+0xb0/0x790
> [    0.000000][    T0] [c0000000027dbf90] [c00000000000d79c] start_here_common+0x1c/0x600
> [    0.000000][    T0] rfi-flush: patched 12 locations (fallback displacement flush)
> 
> 
> cheers
>