linux/arch/arc/include/asm/spinlock.h

382 lines
8.4 KiB

locking/spinlock, arch: Update and fix spin_unlock_wait() implementations This patch updates/fixes all spin_unlock_wait() implementations. The update is in semantics; where it previously was only a control dependency, we now upgrade to a full load-acquire to match the store-release from the spin_unlock() we waited on. This ensures that when spin_unlock_wait() returns, we're guaranteed to observe the full critical section we waited on. This fixes a number of spin_unlock_wait() users that (not unreasonably) rely on this. I also fixed a number of ticket lock versions to only wait on the current lock holder, instead of for a full unlock, as this is sufficient. Furthermore; again for ticket locks; I added an smp_rmb() in between the initial ticket load and the spin loop testing the current value because I could not convince myself the address dependency is sufficient, esp. if the loads are of different sizes. I'm more than happy to remove this smp_rmb() again if people are certain the address dependency does indeed work as expected. Note: PPC32 will be fixed independently Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: chris@zankel.net Cc: cmetcalf@mellanox.com Cc: davem@davemloft.net Cc: dhowells@redhat.com Cc: james.hogan@imgtec.com Cc: jejb@parisc-linux.org Cc: linux@armlinux.org.uk Cc: mpe@ellerman.id.au Cc: ralf@linux-mips.org Cc: realmz6@gmail.com Cc: rkuo@codeaurora.org Cc: rth@twiddle.net Cc: schwidefsky@de.ibm.com Cc: tony.luck@intel.com Cc: vgupta@synopsys.com Cc: ysato@users.sourceforge.jp Cc: linux-kernel@vger.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-05-26 10:35:03 +02:00
ARCv2: spinlock: remove the extra smp_mb before lock, after unlock - ARCv2 LLSC spinlocks have smp_mb() both before and after the LLSC instructions, which is not required per lkmm ACQ/REL semantics. smp_mb() is only needed _after_ lock and _before_ unlock. So remove the extra barriers. The reason they were there was mainly historical. At the time of initial SMP Linux bringup on HS38 cores, I was too conservative, given the fluidity of both hw and sw. The last attempt to ditch the extra barrier showed some hackbench regression which is apparently not the case now (atleast for LLSC case, read on...) - EX based spinlocks (!CONFIG_ARC_HAS_LLSC) still needs the extra smp_mb(), not due to lkmm, but due to some hardware shenanigans. W/o that, hackbench triggers RCU stall splat so extra DMB is retained !LLSC based systems are not realistic Linux sstem anyways so they can afford to be a nit suboptimal ;-) | [ARCLinux]# for i in (seq 1 1 5) ; do hackbench; done | Running with 10 groups 400 process | INFO: task hackbench:158 blocked for more than 10 seconds. | Not tainted 4.20.0-00005-g96b18288a88e-dirty #117 | "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. | hackbench D 0 158 135 0x00000000 | | Stack Trace: | watchdog: BUG: soft lockup - CPU#3 stuck for 59s! [hackbench:469] | Modules linked in: | Path: (null) | CPU: 3 PID: 469 Comm: hackbench Not tainted 4.20.0-00005-g96b18288a88e-dirty | | [ECR ]: 0x00000000 => Check Programmer's Manual | [EFA ]: 0x00000000 | [BLINK ]: do_exit+0x4a6/0x7d0 | [ERET ]: _raw_write_unlock_irq+0x44/0x5c - And while at it, remove the extar smp_mb() from EX based arch_read_trylock() since the spin lock there guarantees a full barrier anyways - For LLSC case, hackbench threads improves with this patch (HAPS @ 50MHz) ---- before ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 16.253 | Time: 16.445 | Time: 16.590 | Time: 16.721 | Time: 16.544 ---- after ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 15.638 | Time: 15.730 | Time: 15.870 | Time: 15.842 | Time: 15.729 Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2019-03-07 13:29:59 -08:00
ARCv2: spinlock: remove the extra smp_mb before lock, after unlock - ARCv2 LLSC spinlocks have smp_mb() both before and after the LLSC instructions, which is not required per lkmm ACQ/REL semantics. smp_mb() is only needed _after_ lock and _before_ unlock. So remove the extra barriers. The reason they were there was mainly historical. At the time of initial SMP Linux bringup on HS38 cores, I was too conservative, given the fluidity of both hw and sw. The last attempt to ditch the extra barrier showed some hackbench regression which is apparently not the case now (atleast for LLSC case, read on...) - EX based spinlocks (!CONFIG_ARC_HAS_LLSC) still needs the extra smp_mb(), not due to lkmm, but due to some hardware shenanigans. W/o that, hackbench triggers RCU stall splat so extra DMB is retained !LLSC based systems are not realistic Linux sstem anyways so they can afford to be a nit suboptimal ;-) | [ARCLinux]# for i in (seq 1 1 5) ; do hackbench; done | Running with 10 groups 400 process | INFO: task hackbench:158 blocked for more than 10 seconds. | Not tainted 4.20.0-00005-g96b18288a88e-dirty #117 | "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. | hackbench D 0 158 135 0x00000000 | | Stack Trace: | watchdog: BUG: soft lockup - CPU#3 stuck for 59s! [hackbench:469] | Modules linked in: | Path: (null) | CPU: 3 PID: 469 Comm: hackbench Not tainted 4.20.0-00005-g96b18288a88e-dirty | | [ECR ]: 0x00000000 => Check Programmer's Manual | [EFA ]: 0x00000000 | [BLINK ]: do_exit+0x4a6/0x7d0 | [ERET ]: _raw_write_unlock_irq+0x44/0x5c - And while at it, remove the extar smp_mb() from EX based arch_read_trylock() since the spin lock there guarantees a full barrier anyways - For LLSC case, hackbench threads improves with this patch (HAPS @ 50MHz) ---- before ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 16.253 | Time: 16.445 | Time: 16.590 | Time: 16.721 | Time: 16.544 ---- after ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 15.638 | Time: 15.730 | Time: 15.870 | Time: 15.842 | Time: 15.729 Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2019-03-07 13:29:59 -08:00
ARCv2: spinlock: remove the extra smp_mb before lock, after unlock - ARCv2 LLSC spinlocks have smp_mb() both before and after the LLSC instructions, which is not required per lkmm ACQ/REL semantics. smp_mb() is only needed _after_ lock and _before_ unlock. So remove the extra barriers. The reason they were there was mainly historical. At the time of initial SMP Linux bringup on HS38 cores, I was too conservative, given the fluidity of both hw and sw. The last attempt to ditch the extra barrier showed some hackbench regression which is apparently not the case now (atleast for LLSC case, read on...) - EX based spinlocks (!CONFIG_ARC_HAS_LLSC) still needs the extra smp_mb(), not due to lkmm, but due to some hardware shenanigans. W/o that, hackbench triggers RCU stall splat so extra DMB is retained !LLSC based systems are not realistic Linux sstem anyways so they can afford to be a nit suboptimal ;-) | [ARCLinux]# for i in (seq 1 1 5) ; do hackbench; done | Running with 10 groups 400 process | INFO: task hackbench:158 blocked for more than 10 seconds. | Not tainted 4.20.0-00005-g96b18288a88e-dirty #117 | "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. | hackbench D 0 158 135 0x00000000 | | Stack Trace: | watchdog: BUG: soft lockup - CPU#3 stuck for 59s! [hackbench:469] | Modules linked in: | Path: (null) | CPU: 3 PID: 469 Comm: hackbench Not tainted 4.20.0-00005-g96b18288a88e-dirty | | [ECR ]: 0x00000000 => Check Programmer's Manual | [EFA ]: 0x00000000 | [BLINK ]: do_exit+0x4a6/0x7d0 | [ERET ]: _raw_write_unlock_irq+0x44/0x5c - And while at it, remove the extar smp_mb() from EX based arch_read_trylock() since the spin lock there guarantees a full barrier anyways - For LLSC case, hackbench threads improves with this patch (HAPS @ 50MHz) ---- before ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 16.253 | Time: 16.445 | Time: 16.590 | Time: 16.721 | Time: 16.544 ---- after ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 15.638 | Time: 15.730 | Time: 15.870 | Time: 15.842 | Time: 15.729 Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2019-03-07 13:29:59 -08:00
ARCv2: spinlock: remove the extra smp_mb before lock, after unlock - ARCv2 LLSC spinlocks have smp_mb() both before and after the LLSC instructions, which is not required per lkmm ACQ/REL semantics. smp_mb() is only needed _after_ lock and _before_ unlock. So remove the extra barriers. The reason they were there was mainly historical. At the time of initial SMP Linux bringup on HS38 cores, I was too conservative, given the fluidity of both hw and sw. The last attempt to ditch the extra barrier showed some hackbench regression which is apparently not the case now (atleast for LLSC case, read on...) - EX based spinlocks (!CONFIG_ARC_HAS_LLSC) still needs the extra smp_mb(), not due to lkmm, but due to some hardware shenanigans. W/o that, hackbench triggers RCU stall splat so extra DMB is retained !LLSC based systems are not realistic Linux sstem anyways so they can afford to be a nit suboptimal ;-) | [ARCLinux]# for i in (seq 1 1 5) ; do hackbench; done | Running with 10 groups 400 process | INFO: task hackbench:158 blocked for more than 10 seconds. | Not tainted 4.20.0-00005-g96b18288a88e-dirty #117 | "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. | hackbench D 0 158 135 0x00000000 | | Stack Trace: | watchdog: BUG: soft lockup - CPU#3 stuck for 59s! [hackbench:469] | Modules linked in: | Path: (null) | CPU: 3 PID: 469 Comm: hackbench Not tainted 4.20.0-00005-g96b18288a88e-dirty | | [ECR ]: 0x00000000 => Check Programmer's Manual | [EFA ]: 0x00000000 | [BLINK ]: do_exit+0x4a6/0x7d0 | [ERET ]: _raw_write_unlock_irq+0x44/0x5c - And while at it, remove the extar smp_mb() from EX based arch_read_trylock() since the spin lock there guarantees a full barrier anyways - For LLSC case, hackbench threads improves with this patch (HAPS @ 50MHz) ---- before ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 16.253 | Time: 16.445 | Time: 16.590 | Time: 16.721 | Time: 16.544 ---- after ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 15.638 | Time: 15.730 | Time: 15.870 | Time: 15.842 | Time: 15.729 Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2019-03-07 13:29:59 -08:00
ARC: Workaround spinlock livelock in SMP SystemC simulation Some ARC SMP systems lack native atomic R-M-W (LLOCK/SCOND) insns and can only use atomic EX insn (reg with mem) to build higher level R-M-W primitives. This includes a SystemC based SMP simulation model. So rwlocks need to use a protecting spinlock for atomic cmp-n-exchange operation to update reader(s)/writer count. The spinlock operation itself looks as follows: mov reg, 1 ; 1=locked, 0=unlocked retry: EX reg, [lock] ; load existing, store 1, atomically BREQ reg, 1, rety ; if already locked, retry In single-threaded simulation, SystemC alternates between the 2 cores with "N" insn each based scheduling. Additionally for insn with global side effect, such as EX writing to shared mem, a core switch is enforced too. Given that, 2 cores doing a repeated EX on same location, Linux often got into a livelock e.g. when both cores were fiddling with tasklist lock (gdbserver / hackbench) for read/write respectively as the sequence diagram below shows: core1 core2 -------- -------- 1. spin lock [EX r=0, w=1] - LOCKED 2. rwlock(Read) - LOCKED 3. spin unlock [ST 0] - UNLOCKED spin lock [EX r=0,w=1] - LOCKED -- resched core 1---- 5. spin lock [EX r=1] - ALREADY-LOCKED -- resched core 2---- 6. rwlock(Write) - READER-LOCKED 7. spin unlock [ST 0] 8. rwlock failed, retry again 9. spin lock [EX r=0, w=1] -- resched core 1---- 10 spinlock locked in #9, retry #5 11. spin lock [EX gets 1] -- resched core 2---- ... ... The fix was to unlock using the EX insn too (step 7), to trigger another SystemC scheduling pass which would let core1 proceed, eliding the livelock. Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2013-09-25 16:53:32 +05:30
ARC: Workaround spinlock livelock in SMP SystemC simulation Some ARC SMP systems lack native atomic R-M-W (LLOCK/SCOND) insns and can only use atomic EX insn (reg with mem) to build higher level R-M-W primitives. This includes a SystemC based SMP simulation model. So rwlocks need to use a protecting spinlock for atomic cmp-n-exchange operation to update reader(s)/writer count. The spinlock operation itself looks as follows: mov reg, 1 ; 1=locked, 0=unlocked retry: EX reg, [lock] ; load existing, store 1, atomically BREQ reg, 1, rety ; if already locked, retry In single-threaded simulation, SystemC alternates between the 2 cores with "N" insn each based scheduling. Additionally for insn with global side effect, such as EX writing to shared mem, a core switch is enforced too. Given that, 2 cores doing a repeated EX on same location, Linux often got into a livelock e.g. when both cores were fiddling with tasklist lock (gdbserver / hackbench) for read/write respectively as the sequence diagram below shows: core1 core2 -------- -------- 1. spin lock [EX r=0, w=1] - LOCKED 2. rwlock(Read) - LOCKED 3. spin unlock [ST 0] - UNLOCKED spin lock [EX r=0,w=1] - LOCKED -- resched core 1---- 5. spin lock [EX r=1] - ALREADY-LOCKED -- resched core 2---- 6. rwlock(Write) - READER-LOCKED 7. spin unlock [ST 0] 8. rwlock failed, retry again 9. spin lock [EX r=0, w=1] -- resched core 1---- 10 spinlock locked in #9, retry #5 11. spin lock [EX gets 1] -- resched core 2---- ... ... The fix was to unlock using the EX insn too (step 7), to trigger another SystemC scheduling pass which would let core1 proceed, eliding the livelock. Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2013-09-25 16:53:32 +05:30
ARC: Workaround spinlock livelock in SMP SystemC simulation Some ARC SMP systems lack native atomic R-M-W (LLOCK/SCOND) insns and can only use atomic EX insn (reg with mem) to build higher level R-M-W primitives. This includes a SystemC based SMP simulation model. So rwlocks need to use a protecting spinlock for atomic cmp-n-exchange operation to update reader(s)/writer count. The spinlock operation itself looks as follows: mov reg, 1 ; 1=locked, 0=unlocked retry: EX reg, [lock] ; load existing, store 1, atomically BREQ reg, 1, rety ; if already locked, retry In single-threaded simulation, SystemC alternates between the 2 cores with "N" insn each based scheduling. Additionally for insn with global side effect, such as EX writing to shared mem, a core switch is enforced too. Given that, 2 cores doing a repeated EX on same location, Linux often got into a livelock e.g. when both cores were fiddling with tasklist lock (gdbserver / hackbench) for read/write respectively as the sequence diagram below shows: core1 core2 -------- -------- 1. spin lock [EX r=0, w=1] - LOCKED 2. rwlock(Read) - LOCKED 3. spin unlock [ST 0] - UNLOCKED spin lock [EX r=0,w=1] - LOCKED -- resched core 1---- 5. spin lock [EX r=1] - ALREADY-LOCKED -- resched core 2---- 6. rwlock(Write) - READER-LOCKED 7. spin unlock [ST 0] 8. rwlock failed, retry again 9. spin lock [EX r=0, w=1] -- resched core 1---- 10 spinlock locked in #9, retry #5 11. spin lock [EX gets 1] -- resched core 2---- ... ... The fix was to unlock using the EX insn too (step 7), to trigger another SystemC scheduling pass which would let core1 proceed, eliding the livelock. Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2013-09-25 16:53:32 +05:30
ARCv2: spinlock: remove the extra smp_mb before lock, after unlock - ARCv2 LLSC spinlocks have smp_mb() both before and after the LLSC instructions, which is not required per lkmm ACQ/REL semantics. smp_mb() is only needed _after_ lock and _before_ unlock. So remove the extra barriers. The reason they were there was mainly historical. At the time of initial SMP Linux bringup on HS38 cores, I was too conservative, given the fluidity of both hw and sw. The last attempt to ditch the extra barrier showed some hackbench regression which is apparently not the case now (atleast for LLSC case, read on...) - EX based spinlocks (!CONFIG_ARC_HAS_LLSC) still needs the extra smp_mb(), not due to lkmm, but due to some hardware shenanigans. W/o that, hackbench triggers RCU stall splat so extra DMB is retained !LLSC based systems are not realistic Linux sstem anyways so they can afford to be a nit suboptimal ;-) | [ARCLinux]# for i in (seq 1 1 5) ; do hackbench; done | Running with 10 groups 400 process | INFO: task hackbench:158 blocked for more than 10 seconds. | Not tainted 4.20.0-00005-g96b18288a88e-dirty #117 | "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. | hackbench D 0 158 135 0x00000000 | | Stack Trace: | watchdog: BUG: soft lockup - CPU#3 stuck for 59s! [hackbench:469] | Modules linked in: | Path: (null) | CPU: 3 PID: 469 Comm: hackbench Not tainted 4.20.0-00005-g96b18288a88e-dirty | | [ECR ]: 0x00000000 => Check Programmer's Manual | [EFA ]: 0x00000000 | [BLINK ]: do_exit+0x4a6/0x7d0 | [ERET ]: _raw_write_unlock_irq+0x44/0x5c - And while at it, remove the extar smp_mb() from EX based arch_read_trylock() since the spin lock there guarantees a full barrier anyways - For LLSC case, hackbench threads improves with this patch (HAPS @ 50MHz) ---- before ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 16.253 | Time: 16.445 | Time: 16.590 | Time: 16.721 | Time: 16.544 ---- after ---- | | [ARCLinux]# for i in 1 2 3 4 5; do hackbench 10 thread; done | Running with 10 groups 400 threads | Time: 15.638 | Time: 15.730 | Time: 15.870 | Time: 15.842 | Time: 15.729 Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
2019-03-07 13:29:59 -08:00