perf: arm_pmu_acpi: Fix armpmu_alloc call from invalid context
PATCH 1/1 perf: arm_pmu_acpi: Fix armpmu_alloc call from invalid context
When booting an arm64 debug kernel with ACPI, I see:
BUG: sleeping function called from invalid context at mm/slab.h:420
in_atomic(): 0, irqs_disabled(): 128, pid: 12, name: cpuhp/0
1 lock held by cpuhp/0/12:
#0: (cpuhp_state-up){+.+.}, at: [<0000000057aa0dae>] cpuhp_thread_fun+0x13c/0x258
irq event stamp: 28
hardirqs last enabled at (27): [<000000000b861658>] _raw_spin_unlock_irq+0x38/0x58
hardirqs last disabled at (28): [<000000006231cfb1>] cpuhp_thread_fun+0xd0/0x258
softirqs last enabled at (0): [<0000000054d9737a>] copy_process.isra.32.part.33+0x450/0x1480
softirqs last disabled at (0): [< (null)>] (null)
CPU: 0 PID: 12 Comm: cpuhp/0 Not tainted 4.15.0+ #18
Hardware name: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.25 Oct 17 2016
Call trace:
dump_backtrace+0x0/0x188
show_stack+0x24/0x2c
dump_stack+0xa4/0xe0
___might_sleep+0x208/0x234
__might_sleep+0x58/0x8c
kmem_cache_alloc_trace+0x248/0x3e0
armpmu_alloc+0x38/0x1a8
arm_pmu_acpi_cpu_starting+0x11c/0x15c
cpuhp_invoke_callback+0x120/0x100c
cpuhp_thread_fun+0xe8/0x258
smpboot_thread_fn+0x170/0x268
kthread+0x110/0x13c
ret_from_fork+0x10/0x18With commit 7d88eb695a1f ("arm/perf: Convert to hotplug state machine"), arm_pmu uses the cpuhotplug framework to initialize the PMU driver when using ACPI. However, the arm_pmu_acpi_cpu_starting() callback comes before CPUHP_AP_ONLINE is reached which means it runs with interrupts diabled and tries to allocate memory with GFP_KERNEL alloc which may sleep.
Move CPUHP_AP_PERF_ARM_ACPI_STARTING to come after CPUHP_AP_ONLINE so that the arm_pmu initialization runs with interrupts enabled as it does when booting with device tree.
--- include/linux/cpuhotplug.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/cpuhotplug.h +1 -1
@@ -114,7 +114,6 @@ enum cpuhp_state {CPUHP_AP_ARM_VFP_STARTING ,CPUHP_AP_ARM64_DEBUG_MONITORS_STARTING ,CPUHP_AP_PERF_ARM_HW_BREAKPOINT_STARTING , -CPUHP_AP_PERF_ARM_ACPI_STARTING ,CPUHP_AP_PERF_ARM_STARTING ,CPUHP_AP_ARM_L2X0_STARTING , CPUHP_AP_ARM_ARCH_TIMER_STARTING,
@@ -146,6 +145,7 @@ enum cpuhp_state {CPUHP_AP_SMPBOOT_THREADS ,CPUHP_AP_X86_VDSO_VMA_ONLINE ,CPUHP_AP_IRQ_AFFINITY_ONLINE , +CPUHP_AP_PERF_ARM_ACPI_STARTING ,CPUHP_AP_PERF_ONLINE ,CPUHP_AP_PERF_X86_ONLINE , CPUHP_AP_PERF_X86_UNCORE_ONLINE,
Signature · 1 line
1.8.3.1
Hi Mark,
Quoted 19 lines · Mark Salter
When booting an arm64 debug kernel with ACPI, I see:
BUG: sleeping function called from invalid context at mm/slab.h:420 in_atomic(): 0, irqs_disabled(): 128, pid: 12, name: cpuhp/0 1 lock held by cpuhp/0/12: #0: (cpuhp_state-up){+.+.}, at: [<0000000057aa0dae>] cpuhp_thread_fun+0x13c/0x258 irq event stamp: 28 hardirqs last enabled at (27): [<000000000b861658>] _raw_spin_unlock_irq+0x38/0x58 hardirqs last disabled at (28): [<000000006231cfb1>] cpuhp_thread_fun+0xd0/0x258 softirqs last enabled at (0): [<0000000054d9737a>] copy_process.isra.32.part.33+0x450/0x1480 softirqs last disabled at (0): [< (null)>] (null) CPU: 0 PID: 12 Comm: cpuhp/0 Not tainted 4.15.0+ #18 Hardware name: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.25 Oct 17 2016 Call trace: dump_backtrace+0x0/0x188 show_stack+0x24/0x2c dump_stack+0xa4/0xe0 ___might_sleep+0x208/0x234 __might_sleep+0x58/0x8c
Quoted 8 lines
kmem_cache_alloc_trace+0x248/0x3e0
armpmu_alloc+0x38/0x1a8
arm_pmu_acpi_cpu_starting+0x11c/0x15c
cpuhp_invoke_callback+0x120/0x100c
cpuhp_thread_fun+0xe8/0x258
smpboot_thread_fn+0x170/0x268
kthread+0x110/0x13c
ret_from_fork+0x10/0x18I have patches to address this:
http://lists.infradead.org/pipermail/linux-arm-kernel/2018-February/557838.html
https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/log/?h=arm64/acpi-pmu-lockdep
Quoted 30 lines
With commit
7d88eb695a1f("arm/perf: Convert to hotplug state machine"),arm_pmuuses the cpuhotplug framework to initialize the PMU driver when using ACPI. However, thearm_pmu_acpi_cpu_starting()callback comes beforeCPUHP_AP_ONLINEis reached which means it runs with interrupts diabled and tries to allocate memory withGFP_KERNELalloc which may sleep.Move
CPUHP_AP_PERF_ARM_ACPI_STARTINGto come afterCPUHP_AP_ONLINEso that thearm_pmuinitialization runs with interrupts enabled as it does when booting with device tree.Fixes 7d88eb695a1f ("arm/perf: Convert to hotplug state machine")Signed-off-by Mark Salter <[email protected]>--- include/linux/cpuhotplug.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/cpuhotplug.h +0
-1@@ -114,7 +114,6 @@ enum cpuhp_state {CPUHP_AP_ARM_VFP_STARTING ,CPUHP_AP_ARM64_DEBUG_MONITORS_STARTING ,CPUHP_AP_PERF_ARM_HW_BREAKPOINT_STARTING , -CPUHP_AP_PERF_ARM_ACPI_STARTING ,CPUHP_AP_PERF_ARM_STARTING ,CPUHP_AP_ARM_L2X0_STARTING , CPUHP_AP_ARM_ARCH_TIMER_STARTING,
include/linux/cpuhotplug.h enum cpuhp_state { · line 148
Quoted 4 lines
include/linux/cpuhotplug.h +1
-0@@ -146,6 +145,7 @@ enum cpuhp_state {CPUHP_AP_SMPBOOT_THREADS ,CPUHP_AP_X86_VDSO_VMA_ONLINE ,CPUHP_AP_IRQ_AFFINITY_ONLINE , + CPUHP_AP_PERF_ARM_ACPI_STARTING,
We need CPUHP_AP_PERF_ARM_ACPI_STARTING to happen before CPUHP_AP_PERF_ARM_STARTING, and I think this re-ordering prevents us from correctly resetting the PMU and enabling percpu interrupts, at least in heterogeneous configurations (e.g. big.LITTLE systems like Juno).
I'm not sure whether we could safely move both callbacks this late.
Thanks,
Mark.
Quoted 31 lines · Mark Rutland
Hi Mark,
Quoted 19 lines · Mark Salter
When booting an arm64 debug kernel with ACPI, I see:
BUG: sleeping function called from invalid context at mm/slab.h:420 in_atomic(): 0, irqs_disabled(): 128, pid: 12, name: cpuhp/0 1 lock held by cpuhp/0/12: #0: (cpuhp_state-up){+.+.}, at: [<0000000057aa0dae>] cpuhp_thread_fun+0x13c/0x258 irq event stamp: 28 hardirqs last enabled at (27): [<000000000b861658>] _raw_spin_unlock_irq+0x38/0x58 hardirqs last disabled at (28): [<000000006231cfb1>] cpuhp_thread_fun+0xd0/0x258 softirqs last enabled at (0): [<0000000054d9737a>] copy_process.isra.32.part.33+0x450/0x1480 softirqs last disabled at (0): [< (null)>] (null) CPU: 0 PID: 12 Comm: cpuhp/0 Not tainted 4.15.0+ #18 Hardware name: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.25 Oct 17 2016 Call trace: dump_backtrace+0x0/0x188 show_stack+0x24/0x2c dump_stack+0xa4/0xe0 ___might_sleep+0x208/0x234 __might_sleep+0x58/0x8cQuoted 8 lines
kmem_cache_alloc_trace+0x248/0x3e0 armpmu_alloc+0x38/0x1a8 arm_pmu_acpi_cpu_starting+0x11c/0x15c cpuhp_invoke_callback+0x120/0x100c cpuhp_thread_fun+0xe8/0x258 smpboot_thread_fn+0x170/0x268 kthread+0x110/0x13c ret_from_fork+0x10/0x18
Quoted 4 lines
I have patches to address this:
http://lists.infradead.org/pipermail/linux-arm-kernel/2018-February/557838.html
https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/log/?h=arm64/acpi-pmu-lockdep
Awesome, I completely missed that. Thanks.
Quoted 46 lines
Quoted 31 lines
With commit
7d88eb695a1f("arm/perf: Convert to hotplug state machine"),arm_pmuuses the cpuhotplug framework to initialize the PMU driver when using ACPI. However, thearm_pmu_acpi_cpu_starting()callback comes beforeCPUHP_AP_ONLINEis reached which means it runs with interrupts diabled and tries to allocate memory withGFP_KERNELalloc which may sleep.Move
CPUHP_AP_PERF_ARM_ACPI_STARTINGto come afterCPUHP_AP_ONLINEso that thearm_pmuinitialization runs with interrupts enabled as it does when booting with device tree.Fixes 7d88eb695a1f ("arm/perf: Convert to hotplug state machine")Signed-off-by Mark Salter <[email protected]>--- include/linux/cpuhotplug.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)include/linux/cpuhotplug.h +0
-1@@ -114,7 +114,6 @@ enum cpuhp_state {CPUHP_AP_ARM_VFP_STARTING ,CPUHP_AP_ARM64_DEBUG_MONITORS_STARTING ,CPUHP_AP_PERF_ARM_HW_BREAKPOINT_STARTING , -CPUHP_AP_PERF_ARM_ACPI_STARTING ,CPUHP_AP_PERF_ARM_STARTING ,CPUHP_AP_ARM_L2X0_STARTING , CPUHP_AP_ARM_ARCH_TIMER_STARTING,include/linux/cpuhotplug.h enum cpuhp_state { · line 148
Quoted 4 lines
include/linux/cpuhotplug.h +1
-0@@ -146,6 +145,7 @@ enum cpuhp_state {CPUHP_AP_SMPBOOT_THREADS ,CPUHP_AP_X86_VDSO_VMA_ONLINE ,CPUHP_AP_IRQ_AFFINITY_ONLINE , + CPUHP_AP_PERF_ARM_ACPI_STARTING,We need
CPUHP_AP_PERF_ARM_ACPI_STARTINGto happen beforeCPUHP_AP_PERF_ARM_STARTING, and I think this re-ordering prevents us from correctly resetting the PMU and enabling percpu interrupts, at least in heterogeneous configurations (e.g. big.LITTLE systems like Juno).I'm not sure whether we could safely move both callbacks this late.
Thanks,
Mark.
Quoted 32 lines · Mark Salter
Quoted 31 lines · Mark Rutland
Hi Mark,
Quoted 27 lines · Mark Salter
When booting an arm64 debug kernel with ACPI, I see:
BUG: sleeping function called from invalid context at mm/slab.h:420 in_atomic(): 0, irqs_disabled(): 128, pid: 12, name: cpuhp/0 1 lock held by cpuhp/0/12: #0: (cpuhp_state-up){+.+.}, at: [<0000000057aa0dae>] cpuhp_thread_fun+0x13c/0x258 irq event stamp: 28 hardirqs last enabled at (27): [<000000000b861658>] _raw_spin_unlock_irq+0x38/0x58 hardirqs last disabled at (28): [<000000006231cfb1>] cpuhp_thread_fun+0xd0/0x258 softirqs last enabled at (0): [<0000000054d9737a>] copy_process.isra.32.part.33+0x450/0x1480 softirqs last disabled at (0): [< (null)>] (null) CPU: 0 PID: 12 Comm: cpuhp/0 Not tainted 4.15.0+ #18 Hardware name: AppliedMicro X-Gene Mustang Board/X-Gene Mustang Board, BIOS 3.06.25 Oct 17 2016 Call trace: dump_backtrace+0x0/0x188 show_stack+0x24/0x2c dump_stack+0xa4/0xe0 ___might_sleep+0x208/0x234 __might_sleep+0x58/0x8c kmem_cache_alloc_trace+0x248/0x3e0 armpmu_alloc+0x38/0x1a8 arm_pmu_acpi_cpu_starting+0x11c/0x15c cpuhp_invoke_callback+0x120/0x100c cpuhp_thread_fun+0xe8/0x258 smpboot_thread_fn+0x170/0x268 kthread+0x110/0x13c ret_from_fork+0x10/0x18
Quoted 6 lines
Quoted 4 lines
I have patches to address this:
http://lists.infradead.org/pipermail/linux-arm-kernel/2018-February/557838.html
https://git.kernel.org/pub/scm/linux/kernel/git/mark/linux.git/log/?h=arm64/acpi-pmu-lockdepAwesome, I completely missed that. Thanks.
As a heads-up, these were merged in v4.16-rc3.
Mark.