arm64: replace read_lock to rcu lock in call_step_hook (2024)

Yang Shi Feb. 8, 2016, 10:49 p.m. UTC

BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:917in_atomic(): 1, irqs_disabled(): 128, pid: 383, name: shPreemption disabled at:[<ffff800000124c18>] kgdb_cpu_enter+0x158/0x6b8CPU: 3 PID: 383 Comm: sh Tainted: G W 4.1.13-rt13 #2Hardware name: Freescale Layerscape 2085a RDB Board (DT)Call trace:[<ffff8000000885e8>] dump_backtrace+0x0/0x128[<ffff800000088734>] show_stack+0x24/0x30[<ffff80000079a7c4>] dump_stack+0x80/0xa0[<ffff8000000bd324>] ___might_sleep+0x18c/0x1a0[<ffff8000007a20ac>] __rt_spin_lock+0x2c/0x40[<ffff8000007a2268>] rt_read_lock+0x40/0x58[<ffff800000085328>] single_step_handler+0x38/0xd8[<ffff800000082368>] do_debug_exception+0x58/0xb8Exception stack(0xffff80834a1e7c80 to 0xffff80834a1e7da0)7c80: ffffff9c ffffffff 92c23ba0 0000ffff 4a1e7e40 ffff8083 001bfcc4 ffff80007ca0: f2000400 00000000 00000000 00000000 4a1e7d80 ffff8083 0049501c ffff80007cc0: 00005402 00000000 00aaa210 ffff8000 4a1e7ea0 ffff8083 000833f4 ffff80007ce0: ffffff9c ffffffff 92c23ba0 0000ffff 4a1e7ea0 ffff8083 001bfcc0 ffff80007d00: 4a0fc400 ffff8083 00005402 00000000 4a1e7d40 ffff8083 00490324 ffff80007d20: ffffff9c 00000000 92c23ba0 0000ffff 000a0000 00000000 00000000 000000007d40: 00000008 00000000 00080000 00000000 92c23b8b 0000ffff 92c23b8e 0000ffff7d60: 00000038 00000000 00001cb2 00000000 00000005 00000000 92d7b498 0000ffff7d80: 01010101 01010101 92be9000 0000ffff 00000000 00000000 00000030 00000000[<ffff8000000833f4>] el1_dbg+0x18/0x6cThis issue is similar with 62c6c61("arm64: replace read_lock to rcu lock incall_break_hook"), but comes to single_step_handler.This also solves kgdbts boot test silent hang issue on 4.4 -rt kernel.Signed-off-by: Yang Shi <yang.shi@linaro.org>--- arch/arm64/kernel/debug-monitors.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-)

* Yang Shi | 2016-02-08 14:49:24 [-0800]:>BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:917…>This issue is similar with 62c6c61("arm64: replace read_lock to rcu lock in>call_break_hook"), but comes to single_step_handler.>>This also solves kgdbts boot test silent hang issue on 4.4 -rt kernel.I am going to apply this for -RT.Sebastian

On Mon, Feb 08, 2016 at 02:49:24PM -0800, Yang Shi wrote:> BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:917> in_atomic(): 1, irqs_disabled(): 128, pid: 383, name: sh> Preemption disabled at:[<ffff800000124c18>] kgdb_cpu_enter+0x158/0x6b8> > CPU: 3 PID: 383 Comm: sh Tainted: G W 4.1.13-rt13 #2> Hardware name: Freescale Layerscape 2085a RDB Board (DT)> Call trace:> [<ffff8000000885e8>] dump_backtrace+0x0/0x128> [<ffff800000088734>] show_stack+0x24/0x30> [<ffff80000079a7c4>] dump_stack+0x80/0xa0> [<ffff8000000bd324>] ___might_sleep+0x18c/0x1a0> [<ffff8000007a20ac>] __rt_spin_lock+0x2c/0x40> [<ffff8000007a2268>] rt_read_lock+0x40/0x58> [<ffff800000085328>] single_step_handler+0x38/0xd8> [<ffff800000082368>] do_debug_exception+0x58/0xb8> Exception stack(0xffff80834a1e7c80 to 0xffff80834a1e7da0)> 7c80: ffffff9c ffffffff 92c23ba0 0000ffff 4a1e7e40 ffff8083 001bfcc4 ffff8000> 7ca0: f2000400 00000000 00000000 00000000 4a1e7d80 ffff8083 0049501c ffff8000> 7cc0: 00005402 00000000 00aaa210 ffff8000 4a1e7ea0 ffff8083 000833f4 ffff8000> 7ce0: ffffff9c ffffffff 92c23ba0 0000ffff 4a1e7ea0 ffff8083 001bfcc0 ffff8000> 7d00: 4a0fc400 ffff8083 00005402 00000000 4a1e7d40 ffff8083 00490324 ffff8000> 7d20: ffffff9c 00000000 92c23ba0 0000ffff 000a0000 00000000 00000000 00000000> 7d40: 00000008 00000000 00080000 00000000 92c23b8b 0000ffff 92c23b8e 0000ffff> 7d60: 00000038 00000000 00001cb2 00000000 00000005 00000000 92d7b498 0000ffff> 7d80: 01010101 01010101 92be9000 0000ffff 00000000 00000000 00000030 00000000> [<ffff8000000833f4>] el1_dbg+0x18/0x6c> > This issue is similar with 62c6c61("arm64: replace read_lock to rcu lock in> call_break_hook"), but comes to single_step_handler.> > This also solves kgdbts boot test silent hang issue on 4.4 -rt kernel.> > Signed-off-by: Yang Shi <yang.shi@linaro.org>Acked-by: Will Deacon <will.deacon@arm.com>Will

On Tue, 9 Feb 2016 14:54:26 +0000Will Deacon <will.deacon@arm.com> wrote:> Acked-by: Will Deacon <will.deacon@arm.com>Will,The patch looks good to me. Do you want to take it through your tree?It benefits mainline too as a rcu_read_lock() is more efficient thanrwlocks. Although I will say this is a slow path anyway.-- Steve

On Tue, Feb 09, 2016 at 10:07:58AM -0500, Steven Rostedt wrote:> On Tue, 9 Feb 2016 14:54:26 +0000> Will Deacon <will.deacon@arm.com> wrote:> > > > Acked-by: Will Deacon <will.deacon@arm.com>> > Will,Hi Steve,> The patch looks good to me. Do you want to take it through your tree?> > It benefits mainline too as a rcu_read_lock() is more efficient than> rwlocks. Although I will say this is a slow path anyway.I was thinking that Catalin would queue it for 4.6 in the arm64 tree,since that's probably easiest in case any unlikely conflicts crop up.Will

On Tue, 9 Feb 2016 16:04:42 +0000Will Deacon <will.deacon@arm.com> wrote: > > The patch looks good to me. Do you want to take it through your tree?> > > > It benefits mainline too as a rcu_read_lock() is more efficient than> > rwlocks. Although I will say this is a slow path anyway. > > I was thinking that Catalin would queue it for 4.6 in the arm64 tree,> since that's probably easiest in case any unlikely conflicts crop up.Oh, OK.That's fine. I just didn't want you to think that the RT folks weregoing to pull it in for mainline.-- Steve

On Tue, Feb 09, 2016 at 04:04:42PM +0000, Will Deacon wrote:> On Tue, Feb 09, 2016 at 10:07:58AM -0500, Steven Rostedt wrote:> > On Tue, 9 Feb 2016 14:54:26 +0000> > Will Deacon <will.deacon@arm.com> wrote:> > > > > > > Acked-by: Will Deacon <will.deacon@arm.com>> > > The patch looks good to me. Do you want to take it through your tree?> > > > It benefits mainline too as a rcu_read_lock() is more efficient than> > rwlocks. Although I will say this is a slow path anyway.> > I was thinking that Catalin would queue it for 4.6 in the arm64 tree,> since that's probably easiest in case any unlikely conflicts crop up.Queued.

Patch

diff --git a/arch/arm64/kernel/debug-monitors.c b/arch/arm64/kernel/debug-monitors.cindex 8aee3ae..c1492ba 100644--- a/arch/arm64/kernel/debug-monitors.c+++ b/arch/arm64/kernel/debug-monitors.c@@ -186,20 +186,21 @@  static void clear_regs_spsr_ss(struct pt_regs *regs) /* EL1 Single Step Handler hooks */ static LIST_HEAD(step_hook);-static DEFINE_RWLOCK(step_hook_lock);+static DEFINE_SPINLOCK(step_hook_lock); void register_step_hook(struct step_hook *hook) {-write_lock(&step_hook_lock);-list_add(&hook->node, &step_hook);-write_unlock(&step_hook_lock);+spin_lock(&step_hook_lock);+list_add_rcu(&hook->node, &step_hook);+spin_unlock(&step_hook_lock); } void unregister_step_hook(struct step_hook *hook) {-write_lock(&step_hook_lock);-list_del(&hook->node);-write_unlock(&step_hook_lock);+spin_lock(&step_hook_lock);+list_del_rcu(&hook->node);+spin_unlock(&step_hook_lock);+synchronize_rcu(); } /*@@ -213,15 +214,15 @@  static int call_step_hook(struct pt_regs *regs, unsigned int esr) struct step_hook *hook; int retval = DBG_HOOK_ERROR; -read_lock(&step_hook_lock);+rcu_read_lock(); -list_for_each_entry(hook, &step_hook, node){+list_for_each_entry_rcu(hook, &step_hook, node){ retval = hook->fn(regs, esr); if (retval == DBG_HOOK_HANDLED) break; } -read_unlock(&step_hook_lock);+rcu_read_unlock(); return retval; }