Skip to contentSkip to conversation

Fix: xtensa: add missing sync_core

PATCH In review Mathieu Desnoyers 2 participants 5 messages Aug 27, 2017 – Sep 12, 2017 Tensilica Xtensa port (xtensa)

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix: xtensa: add missing sync_core

The membarrier system call now requires all architectures to implement sync_core(). On Xtensa, it is provided by the EXTW instruction.

[ Completely untested! Can someone on the xtensa side confirm whether
EXTW is the right way to serialize core execution and try it out ? ]

CC Peter Zijlstra <[email protected]>
CC Paul E. McKenney <[email protected]>
CC Chris Zankel <[email protected]>
CC Max Filippov <[email protected]>
---
 arch/xtensa/include/asm/processor.h | 9 +++++++++
 1 file changed, 9 insertions(+)
arch/xtensa/include/asm/processor.h +9 -0
index 30ee8c608853..b435bd6adbd6 100644
@@ -248,5 +248,14 @@ static inline unsigned long get_er(unsigned long addr)
 
 #endif /* XCHAL_HAVE_EXTERN_REGS */
 
+static inline void sync_core(void)
+{
+	/*
+	 * Synchronize the core execution pipeline. Acts as a compiler
+	 * barrier.
+	 */
+	asm volatile ("extw" : : : "memory");
+}
+
 #endif	/* __ASSEMBLY__ */
 #endif	/* _XTENSA_PROCESSOR_H */
Signature · 1 line
2.11.0
maintainer
#1 lore raw

Hi Mathieu,

Quoted 5 lines · Mathieu Desnoyers

The membarrier system call now requires all architectures to implement sync_core(). On Xtensa, it is provided by the EXTW instruction.

[ Completely untested! Can someone on the xtensa side confirm whether
EXTW is the right way to serialize core execution and try it out ? ]

Thanks for the patch. I'm currently travelling, I'll give it a try next week once I'm back at work.

Signature · 2 lines
Thanks.
-- Max
#2 lore raw
Quoted 4 lines · ----- On Aug 28, 2017, at 1:12 PM, Max Filippov [email protected] wrote:

Hi Mathieu,

On Mon, Aug 28, 2017 at 12:36 AM, Mathieu Desnoyers <[email protected]> wrote:

Quoted 8 lines
Quoted 5 lines

The membarrier system call now requires all architectures to implement sync_core(). On Xtensa, it is provided by the EXTW instruction.

[ Completely untested! Can someone on the xtensa side confirm whether
EXTW is the right way to serialize core execution and try it out ? ]

Thanks for the patch. I'm currently travelling, I'll give it a try next week once I'm back at work.

Hi Max,

I think we may need to flush the icache to make it consistent with the dcache too on xtensa, in addition to the EXTW. The goal here is to allow JIT engines to reclaim and re-use memory after they discard dynamically generated code. This is similar to what we'd need to do on arm32, where they have inconsistent d/i-caches.

Thoughts ?

Thanks,

Mathieu

Quoted 4 lines

--
Thanks.
-- Max

Signature · 3 lines
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
maintainer
#3 lore raw

Hi Mathieu,

Quoted 9 lines · Mathieu Desnoyers
Quoted 8 lines · ----- On Aug 28, 2017, at 1:12 PM, Max Filippov [email protected] wrote:
Quoted 5 lines · Mathieu Desnoyers

The membarrier system call now requires all architectures to implement sync_core(). On Xtensa, it is provided by the EXTW instruction.

[ Completely untested! Can someone on the xtensa side confirm whether
EXTW is the right way to serialize core execution and try it out ? ]

Quoted 8 lines
Quoted 2 lines

Thanks for the patch. I'm currently travelling, I'll give it a try next week once I'm back at work.

I think we may need to flush the icache to make it consistent with the dcache too on xtensa, in addition to the EXTW. The goal here is to allow JIT engines to reclaim and re-use memory after they discard dynamically generated code. This is similar to what we'd need to do on arm32, where they have inconsistent d/i-caches.

my understanding is that to support JIT engines on xtensa we need to do icache/dcache synchronization, this procedure is covered in the ISYNC instruction description in the ISA book, it involves MEMW and ISYNC, but not EXTW. EXTW is meant to work as a CPU barrier that orders all externally visible CPU signals, which seems unnecessary.

Interestingly, currently we don't have MEMW between dcache flush and icache invalidation, so I need to add it to be consistent with the documented procedure. Then I believe that sync_core implementation should invoke flush_dcache_all followed by MEMW followed by invalidate_icache_all. Does that sound right?

Signature · 2 lines
Thanks.
-- Max
#4 lore raw
Quoted 22 lines · ----- On Sep 12, 2017, at 12:37 AM, Max Filippov [email protected] wrote:

Hi Mathieu,

Quoted 9 lines · Mathieu Desnoyers
Quoted 8 lines · ----- On Aug 28, 2017, at 1:12 PM, Max Filippov [email protected] wrote:
Quoted 5 lines · Mathieu Desnoyers

The membarrier system call now requires all architectures to implement sync_core(). On Xtensa, it is provided by the EXTW instruction.

[ Completely untested! Can someone on the xtensa side confirm whether
EXTW is the right way to serialize core execution and try it out ? ]

Quoted 8 lines
Quoted 2 lines

Thanks for the patch. I'm currently travelling, I'll give it a try next week once I'm back at work.

I think we may need to flush the icache to make it consistent with the dcache too on xtensa, in addition to the EXTW. The goal here is to allow JIT engines to reclaim and re-use memory after they discard dynamically generated code. This is similar to what we'd need to do on arm32, where they have inconsistent d/i-caches.

Quoted 5 lines

my understanding is that to support JIT engines on xtensa we need to do icache/dcache synchronization, this procedure is covered in the ISYNC instruction description in the ISA book, it involves MEMW and ISYNC, but not EXTW. EXTW is meant to work as a CPU barrier that orders all externally visible CPU signals, which seems unnecessary.

Good point! Yes, combining MEMW and ISYNC seems to be what we need here.

My upcoming proposal for core-serializing membarrier command issues smp_mb() (memw) and sync_core(). I would then think that just using the "isync" instruction is a good candidate for the sync_core() implementation, given that it is similar to the effect of the core serializing "cpuid" instruction on Intel.

Quoted 6 lines

Interestingly, currently we don't have MEMW between dcache flush and icache invalidation, so I need to add it to be consistent with the documented procedure. Then I believe that sync_core implementation should invoke flush_dcache_all followed by MEMW followed by invalidate_icache_all. Does that sound right?

In a different leg of the thread targeting arm64, we discussed that we should probably only make that membarrier command serialize the core, without taking care of flushing the caches. Cache flushing instructions can be executed from user-space on some architectures, or through another arch-specific system call that targets the address range that needs to be flushed.

For SMP variants of xtensa, that cache flushing system call should consider whether icache invalidation needs to be performed on all CPUs when reclaiming JIT code.

Thoughts ?

Thanks,

Mathieu

Quoted 4 lines

--
Thanks.
-- Max

Signature · 3 lines
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com