linux-arm-kernel.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: nicolas.pitre@linaro.org (Nicolas Pitre)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH 05/16] ARM: bL_head: vlock-based first man election
Date: Wed, 09 Jan 2013 19:20:40 -0500	[thread overview]
Message-ID: <1357777251-13541-6-git-send-email-nicolas.pitre@linaro.org> (raw)
In-Reply-To: <1357777251-13541-1-git-send-email-nicolas.pitre@linaro.org>

From: Dave Martin <dave.martin@linaro.org>

Instead of requiring the first man to be elected in advance (which
can be suboptimal in some situations), this patch uses a per-
cluster mutex to co-ordinate selection of the first man.

This should also make it more feasible to reuse this code path for
asynchronous cluster resume (as in CPUidle scenarios).

Signed-off-by: Dave Martin <dave.martin@linaro.org>
Signed-off-by: Nicolas Pitre <nicolas.pitre@linaro.org>
---
 arch/arm/common/Makefile  |  2 +-
 arch/arm/common/bL_head.S | 91 ++++++++++++++++++++++++++++++++++++++++-------
 2 files changed, 80 insertions(+), 13 deletions(-)

diff --git a/arch/arm/common/Makefile b/arch/arm/common/Makefile
index 50880c494f..894c2ddf9b 100644
--- a/arch/arm/common/Makefile
+++ b/arch/arm/common/Makefile
@@ -15,4 +15,4 @@ obj-$(CONFIG_PCI_HOST_ITE8152)  += it8152.o
 obj-$(CONFIG_ARM_TIMER_SP804)	+= timer-sp.o
 obj-$(CONFIG_FIQ_GLUE)		+= fiq_glue.o fiq_glue_setup.o
 obj-$(CONFIG_FIQ_DEBUGGER)	+= fiq_debugger.o
-obj-$(CONFIG_BIG_LITTLE)	+= bL_head.o bL_entry.o
+obj-$(CONFIG_BIG_LITTLE)	+= bL_head.o bL_entry.o vlock.o
diff --git a/arch/arm/common/bL_head.S b/arch/arm/common/bL_head.S
index f7a64ac127..e70dd432e8 100644
--- a/arch/arm/common/bL_head.S
+++ b/arch/arm/common/bL_head.S
@@ -16,6 +16,8 @@
 #include <linux/linkage.h>
 #include <asm/bL_entry.h>
 
+#include "vlock.h"
+
 .if BL_SYNC_CLUSTER_CPUS
 .error "cpus must be the first member of struct bL_cluster_sync_struct"
 .endif
@@ -64,10 +66,11 @@ ENTRY(bL_entry_point)
 	 * position independent way.
 	 */
 	adr	r5, 3f
-	ldmia	r5, {r6, r7, r8}
+	ldmia	r5, {r6, r7, r8, r11}
 	add	r6, r5, r6			@ r6 = bL_entry_vectors
 	ldr	r7, [r5, r7]			@ r7 = bL_power_up_setup_phys
 	add	r8, r5, r8			@ r8 = bL_sync
+	add	r11, r5, r11			@ r11 = first_man_locks
 
 	mov	r0, #BL_SYNC_CLUSTER_SIZE
 	mla	r8, r0, r10, r8			@ r8 = bL_sync cluster base
@@ -83,11 +86,25 @@ ENTRY(bL_entry_point)
 	@ At this point, the cluster cannot unexpectedly enter the GOING_DOWN
 	@ state, because there is at least one active CPU (this CPU).
 
-	@ Check if the cluster has been set up yet:
+	mov	r0, #.Lvlock_size
+	mla	r11, r0, r10, r11		@ r11 = cluster first man lock
+	mov	r0, r11
+	mov	r1, r9				@ cpu
+	bl	vlock_trylock
+
+	cmp	r0, #0				@ failed to get the lock?
+	bne	cluster_setup_wait		@ wait for cluster setup if so
+
 	ldrb	r0, [r8, #BL_SYNC_CLUSTER_CLUSTER]
-	cmp	r0, #CLUSTER_UP
-	beq	cluster_already_up
+	cmp	r0, #CLUSTER_UP			@ cluster already up?
+	bne	cluster_setup			@ if not, set up the cluster
+
+	@ Otherwise, release the first man lock and skip setup:
+	mov	r0, r11
+	bl	vlock_unlock
+	b	cluster_setup_complete
 
+cluster_setup:
 	@ Signal that the cluster is being brought up:
 	mov	r0, #INBOUND_COMING_UP
 	strb	r0, [r8, #BL_SYNC_CLUSTER_INBOUND]
@@ -102,26 +119,47 @@ ENTRY(bL_entry_point)
 cluster_teardown_wait:
 	ldrb	r0, [r8, #BL_SYNC_CLUSTER_CLUSTER]
 	cmp	r0, #CLUSTER_GOING_DOWN
-	wfeeq
-	beq	cluster_teardown_wait
+	bne	first_man_setup
+	wfe
+	b	cluster_teardown_wait
+
+first_man_setup:
+	@ If the outbound gave up before teardown started, skip cluster setup:
 
-	@ power_up_setup is responsible for setting up the cluster:
+	cmp	r0, #CLUSTER_UP
+	beq	cluster_setup_leave
+
+	@ power_up_setup is now responsible for setting up the cluster:
 
 	cmp	r7, #0
 	mov	r0, #1		@ second (cluster) affinity level
 	blxne	r7		@ Call power_up_setup if defined
 
+	dsb
+	mov	r0, #CLUSTER_UP
+	strb	r0, [r8, #BL_SYNC_CLUSTER_CLUSTER]
+
+cluster_setup_leave:
 	@ Leave the cluster setup critical section:
 
-	dsb
 	mov	r0, #INBOUND_NOT_COMING_UP
 	strb	r0, [r8, #BL_SYNC_CLUSTER_INBOUND]
-	mov	r0, #CLUSTER_UP
-	strb	r0, [r8, #BL_SYNC_CLUSTER_CLUSTER]
 	dsb
 	sev
 
-cluster_already_up:
+	mov	r0, r11
+	bl	vlock_unlock
+	b	cluster_setup_complete
+
+	@ In the contended case, non-first men wait here for cluster setup
+	@ to complete:
+cluster_setup_wait:
+	ldrb	r0, [r8, #BL_SYNC_CLUSTER_CLUSTER]
+	cmp	r0, #CLUSTER_UP
+	wfene
+	bne	cluster_setup_wait
+
+cluster_setup_complete:
 	@ If a platform-specific CPU setup hook is needed, it is
 	@ called from here.
 
@@ -150,11 +188,40 @@ bL_entry_gated:
 3:	.word	bL_entry_vectors - .
 	.word	bL_power_up_setup_phys - 3b
 	.word	bL_sync - 3b
+	.word	first_man_locks - 3b
 
 ENDPROC(bL_entry_point)
 
 	.bss
-	.align	5
+
+	@ Magic to size and align the first-man vlock structures
+	@ so that each does not cross a 1KB boundary.
+	@ We also must ensure that none of these shares a cacheline with
+	@ any data which might be accessed through the cache.
+
+	.equ	.Log2, 0
+	.rept	11
+		.if (1 << .Log2) < VLOCK_SIZE
+			.equ .Log2, .Log2 + 1
+		.endif
+	.endr
+	.if	.Log2 > 10
+		.error "vlock struct is too large for guaranteed barrierless access ordering"
+	.endif
+	.equ	.Lvlock_size, 1 << .Log2
+
+	@ The presence of two .align directives here is deliberate: we must
+	@ align to whichever of the two boundaries is larger:
+	.align	__CACHE_WRITEBACK_ORDER
+	.align	.Log2
+first_man_locks:
+	.rept	BL_NR_CLUSTERS
+	.space	.Lvlock_size
+	.endr
+	.size	first_man_locks, . - first_man_locks
+	.type	first_man_locks, #object
+
+	.align	__CACHE_WRITEBACK_ORDER
 
 	.type	bL_entry_vectors, #object
 ENTRY(bL_entry_vectors)
-- 
1.8.0

  parent reply	other threads:[~2013-01-10  0:20 UTC|newest]

Thread overview: 132+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-01-10  0:20 [PATCH 00/16] big.LITTLE low-level CPU and cluster power management Nicolas Pitre
2013-01-10  0:20 ` [PATCH 01/16] ARM: b.L: secondary kernel entry code Nicolas Pitre
2013-01-10  7:12   ` Stephen Boyd
2013-01-10 15:30     ` Nicolas Pitre
2013-01-10 15:34   ` Catalin Marinas
2013-01-10 16:47     ` Nicolas Pitre
2013-01-11 11:45       ` Catalin Marinas
2013-01-11 12:05         ` Lorenzo Pieralisi
2013-01-11 12:19         ` Dave Martin
2013-01-10 23:05   ` Will Deacon
2013-01-11  1:26     ` Nicolas Pitre
2013-01-11 10:55       ` Will Deacon
2013-01-11 11:35         ` Dave Martin
2013-01-11 17:16   ` Santosh Shilimkar
2013-01-11 18:10     ` Nicolas Pitre
2013-01-11 18:30       ` Santosh Shilimkar
2013-03-07  7:37   ` Pavel Machek
2013-03-07  8:57     ` Nicolas Pitre
2013-01-10  0:20 ` [PATCH 02/16] ARM: b.L: introduce the CPU/cluster power API Nicolas Pitre
2013-01-10 23:08   ` Will Deacon
2013-01-11  2:30     ` Nicolas Pitre
2013-01-11 10:58       ` Will Deacon
2013-01-11 11:29       ` Dave Martin
2013-01-11 17:26   ` Santosh Shilimkar
2013-01-11 18:33     ` Nicolas Pitre
2013-01-11 18:41       ` Santosh Shilimkar
2013-01-11 19:54         ` Nicolas Pitre
2013-01-10  0:20 ` [PATCH 03/16] ARM: b.L: introduce helpers for platform coherency exit/setup Nicolas Pitre
2013-01-10 12:01   ` Dave Martin
2013-01-10 19:04     ` Nicolas Pitre
2013-01-11 11:30       ` Dave Martin
2013-01-10 16:53   ` Catalin Marinas
2013-01-10 17:59     ` Nicolas Pitre
2013-01-10 21:50       ` Catalin Marinas
2013-01-10 22:31         ` Nicolas Pitre
2013-01-11 10:36           ` Dave Martin
2013-01-10 22:32     ` Nicolas Pitre
2013-01-10 23:13   ` Will Deacon
2013-01-11  1:50     ` Nicolas Pitre
2013-01-11 11:09       ` Dave Martin
2013-01-11 17:46   ` Santosh Shilimkar
2013-01-11 18:07     ` Dave Martin
2013-01-11 18:34       ` Santosh Shilimkar
2013-01-14 17:08   ` Dave Martin
2013-01-14 17:15     ` Catalin Marinas
2013-01-14 18:10       ` Dave Martin
2013-01-14 21:34         ` Catalin Marinas
2013-01-10  0:20 ` [PATCH 04/16] ARM: b.L: Add baremetal voting mutexes Nicolas Pitre
2013-01-10 23:18   ` Will Deacon
2013-01-11  3:15     ` Nicolas Pitre
2013-01-11 11:03       ` Will Deacon
2013-01-11 16:57       ` Dave Martin
2013-01-10  0:20 ` Nicolas Pitre [this message]
2013-01-10  0:20 ` [PATCH 06/16] ARM: b.L: generic SMP secondary bringup and hotplug support Nicolas Pitre
2013-01-11 18:02   ` Santosh Shilimkar
2013-01-14 18:05     ` Achin Gupta
2013-01-15  6:32       ` Santosh Shilimkar
2013-01-15 11:18         ` Achin Gupta
2013-01-15 11:26           ` Santosh Shilimkar
2013-01-15 18:53           ` Dave Martin
2013-01-14 16:35   ` Will Deacon
2013-01-14 16:51     ` Nicolas Pitre
2013-01-15 19:09       ` Dave Martin
2013-01-10  0:20 ` [PATCH 07/16] ARM: bL_platsmp.c: close the kernel entry gate before hot-unplugging a CPU Nicolas Pitre
2013-01-14 16:37   ` Will Deacon
2013-01-14 16:53     ` Nicolas Pitre
2013-01-14 17:00       ` Will Deacon
2013-01-14 17:11         ` Catalin Marinas
2013-01-14 17:15         ` Nicolas Pitre
2013-01-14 17:23           ` Will Deacon
2013-01-14 18:26           ` Russell King - ARM Linux
2013-01-14 18:49             ` Nicolas Pitre
2013-01-15 18:40             ` Dave Martin
2013-01-16 16:06               ` Catalin Marinas
2013-01-10  0:20 ` [PATCH 08/16] ARM: bL_platsmp.c: make sure the GIC interface of a dying CPU is disabled Nicolas Pitre
2013-01-11 18:07   ` Santosh Shilimkar
2013-01-11 19:07     ` Nicolas Pitre
2013-01-12  6:50       ` Santosh Shilimkar
2013-01-12 16:47         ` Nicolas Pitre
2013-01-13  4:37           ` Santosh Shilimkar
2013-01-14 17:53           ` Lorenzo Pieralisi
2013-01-14 16:39   ` Will Deacon
2013-01-14 16:54     ` Nicolas Pitre
2013-01-14 17:02       ` Will Deacon
2013-01-14 17:18         ` Nicolas Pitre
2013-01-14 17:24           ` Will Deacon
2013-01-14 17:56             ` Lorenzo Pieralisi
2013-01-10  0:20 ` [PATCH 09/16] ARM: vexpress: Select the correct SMP operations at run-time Nicolas Pitre
2013-01-10  0:20 ` [PATCH 10/16] ARM: vexpress: introduce DCSCB support Nicolas Pitre
2013-01-11 18:12   ` Santosh Shilimkar
2013-01-11 19:13     ` Nicolas Pitre
2013-01-12  6:52       ` Santosh Shilimkar
2013-01-10  0:20 ` [PATCH 11/16] ARM: vexpress/dcscb: add CPU use counts to the power up/down API implementation Nicolas Pitre
2013-01-10  0:20 ` [PATCH 12/16] ARM: vexpress/dcscb: do not hardcode number of CPUs per cluster Nicolas Pitre
2013-01-10  0:20 ` [PATCH 13/16] drivers: misc: add ARM CCI support Nicolas Pitre
2013-01-11 18:20   ` Santosh Shilimkar
2013-01-11 19:22     ` Nicolas Pitre
2013-01-12  6:53       ` Santosh Shilimkar
2013-01-15 18:34       ` Dave Martin
2013-01-10  0:20 ` [PATCH 14/16] ARM: TC2: ensure powerdown-time data is flushed from cache Nicolas Pitre
2013-01-10 18:50   ` Dave Martin
2013-01-10 19:13     ` Nicolas Pitre
2013-01-11 11:38       ` Dave Martin
2013-01-10  0:20 ` [PATCH 15/16] ARM: vexpress/dcscb: handle platform coherency exit/setup and CCI Nicolas Pitre
2013-01-10 12:05   ` Dave Martin
2013-01-11 18:27   ` Santosh Shilimkar
2013-01-11 19:28     ` Nicolas Pitre
2013-01-12  7:21       ` Santosh Shilimkar
2013-01-14 12:25         ` Lorenzo Pieralisi
2013-01-15  6:23           ` Santosh Shilimkar
2013-01-15 18:20             ` Dave Martin
2013-01-16  6:33               ` Santosh Shilimkar
2013-01-16 10:03                 ` Lorenzo Pieralisi
2013-01-16 10:12                   ` Santosh Shilimkar
2013-01-10  0:20 ` [PATCH 16/16] ARM: vexpress/dcscb: probe via device tree Nicolas Pitre
2013-01-10  0:46 ` [PATCH 00/16] big.LITTLE low-level CPU and cluster power management Rob Herring
2013-01-10  5:04   ` Nicolas Pitre
2013-01-10 23:01 ` Will Deacon
2013-01-14  9:56 ` Joseph Lo
2013-01-14 14:05   ` Nicolas Pitre
2013-01-15  2:44     ` Joseph Lo
2013-01-15 16:44       ` Nicolas Pitre
2013-01-16 16:02         ` Catalin Marinas
2013-01-16 21:18           ` Nicolas Pitre
2013-01-17 17:55             ` Catalin Marinas
2013-01-15 18:31     ` Dave Martin
2013-03-07  8:27 ` Pavel Machek
2013-03-07  9:12   ` Nicolas Pitre
2013-03-07  9:40     ` Pavel Machek
2013-03-07  9:56       ` Nicolas Pitre
2013-03-07 14:51         ` Pavel Machek
2013-03-07 15:42           ` Nicolas Pitre

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1357777251-13541-6-git-send-email-nicolas.pitre@linaro.org \
    --to=nicolas.pitre@linaro.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).