(CC list trimmed since this is just for initial feedback)
SRCU is not necessary to be compiled by default in all cases. For tinification
efforts not compiling SRCU unless necessary is desirable.
The current patch tries to make compiling SRCU optional by introducing a new
Kconfig option CONFIG_SRCU which is selected when any of the components making
use of SRCU are selected.
The current tinyconfig on x86 enables SRCU by default since x86 enables
PERF_EVENTS which in-turn require SRCU support. If we disable
PERF_EVENTS, we can disable compiling SRCU for tiny configs on x86 systems.
I tested this using randconfig and building about 15 kernels. This is definitely
not complete and could use more testing. Sticking it in next might help in that
regards.
Comments and suggestions are welcome. Please let me know if I should split up
this patch.
Signed-off-by: Pranith Kumar <[email protected]>
CC: Paul McKenney <[email protected]>
CC: Josh Triplett <[email protected]>
---
v2:
- handle cpufreq, devfreq and notifier
- build tested tinyconfig on powerpc
arch/arm/kvm/Kconfig | 1 +
arch/arm64/kvm/Kconfig | 1 +
arch/ia64/kvm/Kconfig | 1 +
arch/mips/kvm/Kconfig | 1 +
arch/powerpc/kvm/Kconfig | 1 +
arch/s390/kvm/Kconfig | 1 +
arch/tile/kvm/Kconfig | 1 +
arch/x86/kvm/Kconfig | 1 +
drivers/clk/Kconfig | 1 +
drivers/cpufreq/Kconfig | 1 +
drivers/devfreq/Kconfig | 1 +
drivers/md/Kconfig | 1 +
drivers/net/Kconfig | 1 +
fs/btrfs/Kconfig | 1 +
fs/notify/Kconfig | 1 +
init/Kconfig | 10 ++++++++++
kernel/notifier.c | 3 +++
kernel/rcu/Makefile | 3 ++-
lib/Kconfig.debug | 1 +
mm/Kconfig | 1 +
security/tomoyo/Kconfig | 1 +
21 files changed, 33 insertions(+), 1 deletion(-)
diff --git a/arch/arm/kvm/Kconfig b/arch/arm/kvm/Kconfig
index 466bd29..3afee5f 100644
--- a/arch/arm/kvm/Kconfig
+++ b/arch/arm/kvm/Kconfig
@@ -23,6 +23,7 @@ config KVM
select HAVE_KVM_CPU_RELAX_INTERCEPT
select KVM_MMIO
select KVM_ARM_HOST
+ select SRCU
depends on ARM_VIRT_EXT && ARM_LPAE
---help---
Support hosting virtualized guest machines. You will also
diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
index 8ba85e9..b334084 100644
--- a/arch/arm64/kvm/Kconfig
+++ b/arch/arm64/kvm/Kconfig
@@ -26,6 +26,7 @@ config KVM
select KVM_ARM_HOST
select KVM_ARM_VGIC
select KVM_ARM_TIMER
+ select SRCU
---help---
Support hosting virtualized guest machines.
diff --git a/arch/ia64/kvm/Kconfig b/arch/ia64/kvm/Kconfig
index 3d50ea9..f7d380b 100644
--- a/arch/ia64/kvm/Kconfig
+++ b/arch/ia64/kvm/Kconfig
@@ -29,6 +29,7 @@ config KVM
select HAVE_KVM_IRQ_ROUTING
select KVM_APIC_ARCHITECTURE
select KVM_MMIO
+ select SRCU
---help---
Support hosting fully virtualized guest machines using hardware
virtualization extensions. You will need a fairly recent
diff --git a/arch/mips/kvm/Kconfig b/arch/mips/kvm/Kconfig
index 30e334e..2ae1282 100644
--- a/arch/mips/kvm/Kconfig
+++ b/arch/mips/kvm/Kconfig
@@ -20,6 +20,7 @@ config KVM
select PREEMPT_NOTIFIERS
select ANON_INODES
select KVM_MMIO
+ select SRCU
---help---
Support for hosting Guest kernels.
Currently supported on MIPS32 processors.
diff --git a/arch/powerpc/kvm/Kconfig b/arch/powerpc/kvm/Kconfig
index 602eb51..af18e0f 100644
--- a/arch/powerpc/kvm/Kconfig
+++ b/arch/powerpc/kvm/Kconfig
@@ -21,6 +21,7 @@ config KVM
select PREEMPT_NOTIFIERS
select ANON_INODES
select HAVE_KVM_EVENTFD
+ select SRCU
config KVM_BOOK3S_HANDLER
bool
diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
index 646db9c..5fce52c 100644
--- a/arch/s390/kvm/Kconfig
+++ b/arch/s390/kvm/Kconfig
@@ -28,6 +28,7 @@ config KVM
select HAVE_KVM_IRQCHIP
select HAVE_KVM_IRQFD
select HAVE_KVM_IRQ_ROUTING
+ select SRCU
---help---
Support hosting paravirtualized guest machines using the SIE
virtualization capability on the mainframe. This should work
diff --git a/arch/tile/kvm/Kconfig b/arch/tile/kvm/Kconfig
index 2298cb1..1e968f7 100644
--- a/arch/tile/kvm/Kconfig
+++ b/arch/tile/kvm/Kconfig
@@ -21,6 +21,7 @@ config KVM
depends on HAVE_KVM && MODULES
select PREEMPT_NOTIFIERS
select ANON_INODES
+ select SRCU
---help---
Support hosting paravirtualized guest machines.
diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
index f9d16ff..7dc7ba5 100644
--- a/arch/x86/kvm/Kconfig
+++ b/arch/x86/kvm/Kconfig
@@ -40,6 +40,7 @@ config KVM
select HAVE_KVM_MSI
select HAVE_KVM_CPU_RELAX_INTERCEPT
select KVM_VFIO
+ select SRCU
---help---
Support hosting fully virtualized guest machines using hardware
virtualization extensions. You will need a fairly recent
diff --git a/drivers/clk/Kconfig b/drivers/clk/Kconfig
index 455fd17..53e64f1 100644
--- a/drivers/clk/Kconfig
+++ b/drivers/clk/Kconfig
@@ -13,6 +13,7 @@ config COMMON_CLK
bool
select HAVE_CLK_PREPARE
select CLKDEV_LOOKUP
+ select SRCU
---help---
The common clock framework is a single definition of struct
clk, useful across many platforms, as well as an
diff --git a/drivers/cpufreq/Kconfig b/drivers/cpufreq/Kconfig
index 3489f8f..ebaa0cc 100644
--- a/drivers/cpufreq/Kconfig
+++ b/drivers/cpufreq/Kconfig
@@ -2,6 +2,7 @@ menu "CPU Frequency scaling"
config CPU_FREQ
bool "CPU Frequency scaling"
+ select SRCU
help
CPU Frequency scaling allows you to change the clock speed of
CPUs on the fly. This is a nice method to save power, because
diff --git a/drivers/devfreq/Kconfig b/drivers/devfreq/Kconfig
index faf4e70..3891f67 100644
--- a/drivers/devfreq/Kconfig
+++ b/drivers/devfreq/Kconfig
@@ -1,5 +1,6 @@
menuconfig PM_DEVFREQ
bool "Generic Dynamic Voltage and Frequency Scaling (DVFS) support"
+ select SRCU
help
A device may have a list of frequencies and voltages available.
devfreq, a generic DVFS framework can be registered for a device
diff --git a/drivers/md/Kconfig b/drivers/md/Kconfig
index 5bdedf6..c355a22 100644
--- a/drivers/md/Kconfig
+++ b/drivers/md/Kconfig
@@ -5,6 +5,7 @@
menuconfig MD
bool "Multiple devices driver support (RAID and LVM)"
depends on BLOCK
+ select SRCU
help
Support multiple physical spindles through a single logical device.
Required for RAID and logical volume management.
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 4706386..221e623 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -176,6 +176,7 @@ config NETCONSOLE_DYNAMIC
config NETPOLL
def_bool NETCONSOLE
+ select SRCU
config NET_POLL_CONTROLLER
def_bool NETPOLL
diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig
index a66768e..80e9c18 100644
--- a/fs/btrfs/Kconfig
+++ b/fs/btrfs/Kconfig
@@ -8,6 +8,7 @@ config BTRFS_FS
select LZO_DECOMPRESS
select RAID6_PQ
select XOR_BLOCKS
+ select SRCU
help
Btrfs is a general purpose copy-on-write filesystem with extents,
diff --git a/fs/notify/Kconfig b/fs/notify/Kconfig
index 22c629e..2a24249 100644
--- a/fs/notify/Kconfig
+++ b/fs/notify/Kconfig
@@ -1,5 +1,6 @@
config FSNOTIFY
def_bool n
+ select SRCU
source "fs/notify/dnotify/Kconfig"
source "fs/notify/inotify/Kconfig"
diff --git a/init/Kconfig b/init/Kconfig
index 3ee28ae..80ac0a4 100644
--- a/init/Kconfig
+++ b/init/Kconfig
@@ -501,6 +501,14 @@ config TINY_RCU
endchoice
+config SRCU
+ bool "Sleepable form of RCU"
+ def_bool n
+ help
+ This option selects the sleepable version of RCU. This version
+ permits arbitrary sleeping or blocking within RCU read-side critical
+ sections.
+
config PREEMPT_RCU
def_bool TREE_PREEMPT_RCU
help
@@ -510,6 +518,7 @@ config PREEMPT_RCU
config TASKS_RCU
bool "Task_based RCU implementation using voluntary context switch"
default n
+ select SRCU
help
This option enables a task-based RCU implementation that uses
only voluntary context switch (not preemption!), idle, and
@@ -1586,6 +1595,7 @@ config PERF_EVENTS
depends on HAVE_PERF_EVENTS
select ANON_INODES
select IRQ_WORK
+ select SRCU
help
Enable kernel support for various performance events provided
by software and hardware.
diff --git a/kernel/notifier.c b/kernel/notifier.c
index 4803da6..f52858a 100644
--- a/kernel/notifier.c
+++ b/kernel/notifier.c
@@ -402,6 +402,7 @@ int raw_notifier_call_chain(struct raw_notifier_head *nh,
}
EXPORT_SYMBOL_GPL(raw_notifier_call_chain);
+#if CONFIG_SRCU
/*
* SRCU notifier chain routines. Registration and unregistration
* use a mutex, and call_chain is synchronized by SRCU (no locks).
@@ -528,6 +529,8 @@ void srcu_init_notifier_head(struct srcu_notifier_head *nh)
}
EXPORT_SYMBOL_GPL(srcu_init_notifier_head);
+#endif /* CONFIG_SRCU */
+
static ATOMIC_NOTIFIER_HEAD(die_chain);
int notrace notify_die(enum die_val val, const char *str,
diff --git a/kernel/rcu/Makefile b/kernel/rcu/Makefile
index 807ccfb..a484080 100644
--- a/kernel/rcu/Makefile
+++ b/kernel/rcu/Makefile
@@ -1,4 +1,5 @@
-obj-y += update.o srcu.o
+obj-y += update.o
+obj-$(CONFIG_SRCU) += srcu.o
obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
obj-$(CONFIG_TREE_RCU) += tree.o
obj-$(CONFIG_TREE_PREEMPT_RCU) += tree.o
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 4e35a5d..797511a 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -1199,6 +1199,7 @@ config RCU_TORTURE_TEST
tristate "torture tests for RCU"
depends on DEBUG_KERNEL
select TORTURE_TEST
+ select SRCU
default n
help
This option provides a kernel module that runs torture tests
diff --git a/mm/Kconfig b/mm/Kconfig
index 1d1ae6b..4395b12 100644
--- a/mm/Kconfig
+++ b/mm/Kconfig
@@ -325,6 +325,7 @@ config VIRT_TO_BUS
config MMU_NOTIFIER
bool
+ select SRCU
config KSM
bool "Enable KSM for page merging"
diff --git a/security/tomoyo/Kconfig b/security/tomoyo/Kconfig
index 8eb779b9d..604e718 100644
--- a/security/tomoyo/Kconfig
+++ b/security/tomoyo/Kconfig
@@ -5,6 +5,7 @@ config SECURITY_TOMOYO
select SECURITYFS
select SECURITY_PATH
select SECURITY_NETWORK
+ select SRCU
default n
help
This selects TOMOYO Linux, pathname-based access control.
--
1.9.1
On Wed, Dec 03, 2014 at 11:03:52PM -0500, Pranith Kumar wrote:
> (CC list trimmed since this is just for initial feedback)
>
> SRCU is not necessary to be compiled by default in all cases. For tinification
> efforts not compiling SRCU unless necessary is desirable.
>
> The current patch tries to make compiling SRCU optional by introducing a new
> Kconfig option CONFIG_SRCU which is selected when any of the components making
> use of SRCU are selected.
>
> The current tinyconfig on x86 enables SRCU by default since x86 enables
> PERF_EVENTS which in-turn require SRCU support. If we disable
> PERF_EVENTS, we can disable compiling SRCU for tiny configs on x86 systems.
>
> I tested this using randconfig and building about 15 kernels. This is definitely
> not complete and could use more testing. Sticking it in next might help in that
> regards.
>
> Comments and suggestions are welcome. Please let me know if I should split up
> this patch.
For the moment, I applied this to -rcu against v3.18-rc6 to get some 0day
testing on it.
There are conflicts against the tip of my tree, but will worry about
that later. I don't have a ready answer on whether or not this needs
to be split up -- in the worst case, we would add the SRCU Kconfig
parameter, add all the selects, then once they all hit mainline add the
Makefile change.
We will need stats on how much memory was saved, and a "size kernel/rcu/srcu.o"
after a UP/tiny build should do it. Could you please add this to the
commit log?
Josh, and additional thoughts?
Thanx, Paul
> Signed-off-by: Pranith Kumar <[email protected]>
> CC: Paul McKenney <[email protected]>
> CC: Josh Triplett <[email protected]>
> ---
> v2:
> - handle cpufreq, devfreq and notifier
> - build tested tinyconfig on powerpc
>
> arch/arm/kvm/Kconfig | 1 +
> arch/arm64/kvm/Kconfig | 1 +
> arch/ia64/kvm/Kconfig | 1 +
> arch/mips/kvm/Kconfig | 1 +
> arch/powerpc/kvm/Kconfig | 1 +
> arch/s390/kvm/Kconfig | 1 +
> arch/tile/kvm/Kconfig | 1 +
> arch/x86/kvm/Kconfig | 1 +
> drivers/clk/Kconfig | 1 +
> drivers/cpufreq/Kconfig | 1 +
> drivers/devfreq/Kconfig | 1 +
> drivers/md/Kconfig | 1 +
> drivers/net/Kconfig | 1 +
> fs/btrfs/Kconfig | 1 +
> fs/notify/Kconfig | 1 +
> init/Kconfig | 10 ++++++++++
> kernel/notifier.c | 3 +++
> kernel/rcu/Makefile | 3 ++-
> lib/Kconfig.debug | 1 +
> mm/Kconfig | 1 +
> security/tomoyo/Kconfig | 1 +
> 21 files changed, 33 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm/kvm/Kconfig b/arch/arm/kvm/Kconfig
> index 466bd29..3afee5f 100644
> --- a/arch/arm/kvm/Kconfig
> +++ b/arch/arm/kvm/Kconfig
> @@ -23,6 +23,7 @@ config KVM
> select HAVE_KVM_CPU_RELAX_INTERCEPT
> select KVM_MMIO
> select KVM_ARM_HOST
> + select SRCU
> depends on ARM_VIRT_EXT && ARM_LPAE
> ---help---
> Support hosting virtualized guest machines. You will also
> diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
> index 8ba85e9..b334084 100644
> --- a/arch/arm64/kvm/Kconfig
> +++ b/arch/arm64/kvm/Kconfig
> @@ -26,6 +26,7 @@ config KVM
> select KVM_ARM_HOST
> select KVM_ARM_VGIC
> select KVM_ARM_TIMER
> + select SRCU
> ---help---
> Support hosting virtualized guest machines.
>
> diff --git a/arch/ia64/kvm/Kconfig b/arch/ia64/kvm/Kconfig
> index 3d50ea9..f7d380b 100644
> --- a/arch/ia64/kvm/Kconfig
> +++ b/arch/ia64/kvm/Kconfig
> @@ -29,6 +29,7 @@ config KVM
> select HAVE_KVM_IRQ_ROUTING
> select KVM_APIC_ARCHITECTURE
> select KVM_MMIO
> + select SRCU
> ---help---
> Support hosting fully virtualized guest machines using hardware
> virtualization extensions. You will need a fairly recent
> diff --git a/arch/mips/kvm/Kconfig b/arch/mips/kvm/Kconfig
> index 30e334e..2ae1282 100644
> --- a/arch/mips/kvm/Kconfig
> +++ b/arch/mips/kvm/Kconfig
> @@ -20,6 +20,7 @@ config KVM
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> select KVM_MMIO
> + select SRCU
> ---help---
> Support for hosting Guest kernels.
> Currently supported on MIPS32 processors.
> diff --git a/arch/powerpc/kvm/Kconfig b/arch/powerpc/kvm/Kconfig
> index 602eb51..af18e0f 100644
> --- a/arch/powerpc/kvm/Kconfig
> +++ b/arch/powerpc/kvm/Kconfig
> @@ -21,6 +21,7 @@ config KVM
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> select HAVE_KVM_EVENTFD
> + select SRCU
>
> config KVM_BOOK3S_HANDLER
> bool
> diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
> index 646db9c..5fce52c 100644
> --- a/arch/s390/kvm/Kconfig
> +++ b/arch/s390/kvm/Kconfig
> @@ -28,6 +28,7 @@ config KVM
> select HAVE_KVM_IRQCHIP
> select HAVE_KVM_IRQFD
> select HAVE_KVM_IRQ_ROUTING
> + select SRCU
> ---help---
> Support hosting paravirtualized guest machines using the SIE
> virtualization capability on the mainframe. This should work
> diff --git a/arch/tile/kvm/Kconfig b/arch/tile/kvm/Kconfig
> index 2298cb1..1e968f7 100644
> --- a/arch/tile/kvm/Kconfig
> +++ b/arch/tile/kvm/Kconfig
> @@ -21,6 +21,7 @@ config KVM
> depends on HAVE_KVM && MODULES
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> + select SRCU
> ---help---
> Support hosting paravirtualized guest machines.
>
> diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
> index f9d16ff..7dc7ba5 100644
> --- a/arch/x86/kvm/Kconfig
> +++ b/arch/x86/kvm/Kconfig
> @@ -40,6 +40,7 @@ config KVM
> select HAVE_KVM_MSI
> select HAVE_KVM_CPU_RELAX_INTERCEPT
> select KVM_VFIO
> + select SRCU
> ---help---
> Support hosting fully virtualized guest machines using hardware
> virtualization extensions. You will need a fairly recent
> diff --git a/drivers/clk/Kconfig b/drivers/clk/Kconfig
> index 455fd17..53e64f1 100644
> --- a/drivers/clk/Kconfig
> +++ b/drivers/clk/Kconfig
> @@ -13,6 +13,7 @@ config COMMON_CLK
> bool
> select HAVE_CLK_PREPARE
> select CLKDEV_LOOKUP
> + select SRCU
> ---help---
> The common clock framework is a single definition of struct
> clk, useful across many platforms, as well as an
> diff --git a/drivers/cpufreq/Kconfig b/drivers/cpufreq/Kconfig
> index 3489f8f..ebaa0cc 100644
> --- a/drivers/cpufreq/Kconfig
> +++ b/drivers/cpufreq/Kconfig
> @@ -2,6 +2,7 @@ menu "CPU Frequency scaling"
>
> config CPU_FREQ
> bool "CPU Frequency scaling"
> + select SRCU
> help
> CPU Frequency scaling allows you to change the clock speed of
> CPUs on the fly. This is a nice method to save power, because
> diff --git a/drivers/devfreq/Kconfig b/drivers/devfreq/Kconfig
> index faf4e70..3891f67 100644
> --- a/drivers/devfreq/Kconfig
> +++ b/drivers/devfreq/Kconfig
> @@ -1,5 +1,6 @@
> menuconfig PM_DEVFREQ
> bool "Generic Dynamic Voltage and Frequency Scaling (DVFS) support"
> + select SRCU
> help
> A device may have a list of frequencies and voltages available.
> devfreq, a generic DVFS framework can be registered for a device
> diff --git a/drivers/md/Kconfig b/drivers/md/Kconfig
> index 5bdedf6..c355a22 100644
> --- a/drivers/md/Kconfig
> +++ b/drivers/md/Kconfig
> @@ -5,6 +5,7 @@
> menuconfig MD
> bool "Multiple devices driver support (RAID and LVM)"
> depends on BLOCK
> + select SRCU
> help
> Support multiple physical spindles through a single logical device.
> Required for RAID and logical volume management.
> diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
> index 4706386..221e623 100644
> --- a/drivers/net/Kconfig
> +++ b/drivers/net/Kconfig
> @@ -176,6 +176,7 @@ config NETCONSOLE_DYNAMIC
>
> config NETPOLL
> def_bool NETCONSOLE
> + select SRCU
>
> config NET_POLL_CONTROLLER
> def_bool NETPOLL
> diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig
> index a66768e..80e9c18 100644
> --- a/fs/btrfs/Kconfig
> +++ b/fs/btrfs/Kconfig
> @@ -8,6 +8,7 @@ config BTRFS_FS
> select LZO_DECOMPRESS
> select RAID6_PQ
> select XOR_BLOCKS
> + select SRCU
>
> help
> Btrfs is a general purpose copy-on-write filesystem with extents,
> diff --git a/fs/notify/Kconfig b/fs/notify/Kconfig
> index 22c629e..2a24249 100644
> --- a/fs/notify/Kconfig
> +++ b/fs/notify/Kconfig
> @@ -1,5 +1,6 @@
> config FSNOTIFY
> def_bool n
> + select SRCU
>
> source "fs/notify/dnotify/Kconfig"
> source "fs/notify/inotify/Kconfig"
> diff --git a/init/Kconfig b/init/Kconfig
> index 3ee28ae..80ac0a4 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -501,6 +501,14 @@ config TINY_RCU
>
> endchoice
>
> +config SRCU
> + bool "Sleepable form of RCU"
> + def_bool n
> + help
> + This option selects the sleepable version of RCU. This version
> + permits arbitrary sleeping or blocking within RCU read-side critical
> + sections.
> +
> config PREEMPT_RCU
> def_bool TREE_PREEMPT_RCU
> help
> @@ -510,6 +518,7 @@ config PREEMPT_RCU
> config TASKS_RCU
> bool "Task_based RCU implementation using voluntary context switch"
> default n
> + select SRCU
> help
> This option enables a task-based RCU implementation that uses
> only voluntary context switch (not preemption!), idle, and
> @@ -1586,6 +1595,7 @@ config PERF_EVENTS
> depends on HAVE_PERF_EVENTS
> select ANON_INODES
> select IRQ_WORK
> + select SRCU
> help
> Enable kernel support for various performance events provided
> by software and hardware.
> diff --git a/kernel/notifier.c b/kernel/notifier.c
> index 4803da6..f52858a 100644
> --- a/kernel/notifier.c
> +++ b/kernel/notifier.c
> @@ -402,6 +402,7 @@ int raw_notifier_call_chain(struct raw_notifier_head *nh,
> }
> EXPORT_SYMBOL_GPL(raw_notifier_call_chain);
>
> +#if CONFIG_SRCU
> /*
> * SRCU notifier chain routines. Registration and unregistration
> * use a mutex, and call_chain is synchronized by SRCU (no locks).
> @@ -528,6 +529,8 @@ void srcu_init_notifier_head(struct srcu_notifier_head *nh)
> }
> EXPORT_SYMBOL_GPL(srcu_init_notifier_head);
>
> +#endif /* CONFIG_SRCU */
> +
> static ATOMIC_NOTIFIER_HEAD(die_chain);
>
> int notrace notify_die(enum die_val val, const char *str,
> diff --git a/kernel/rcu/Makefile b/kernel/rcu/Makefile
> index 807ccfb..a484080 100644
> --- a/kernel/rcu/Makefile
> +++ b/kernel/rcu/Makefile
> @@ -1,4 +1,5 @@
> -obj-y += update.o srcu.o
> +obj-y += update.o
> +obj-$(CONFIG_SRCU) += srcu.o
> obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
> obj-$(CONFIG_TREE_RCU) += tree.o
> obj-$(CONFIG_TREE_PREEMPT_RCU) += tree.o
> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> index 4e35a5d..797511a 100644
> --- a/lib/Kconfig.debug
> +++ b/lib/Kconfig.debug
> @@ -1199,6 +1199,7 @@ config RCU_TORTURE_TEST
> tristate "torture tests for RCU"
> depends on DEBUG_KERNEL
> select TORTURE_TEST
> + select SRCU
> default n
> help
> This option provides a kernel module that runs torture tests
> diff --git a/mm/Kconfig b/mm/Kconfig
> index 1d1ae6b..4395b12 100644
> --- a/mm/Kconfig
> +++ b/mm/Kconfig
> @@ -325,6 +325,7 @@ config VIRT_TO_BUS
>
> config MMU_NOTIFIER
> bool
> + select SRCU
>
> config KSM
> bool "Enable KSM for page merging"
> diff --git a/security/tomoyo/Kconfig b/security/tomoyo/Kconfig
> index 8eb779b9d..604e718 100644
> --- a/security/tomoyo/Kconfig
> +++ b/security/tomoyo/Kconfig
> @@ -5,6 +5,7 @@ config SECURITY_TOMOYO
> select SECURITYFS
> select SECURITY_PATH
> select SECURITY_NETWORK
> + select SRCU
> default n
> help
> This selects TOMOYO Linux, pathname-based access control.
> --
> 1.9.1
>
On 12/04/2014 01:27 AM, Paul E. McKenney wrote:
> For the moment, I applied this to -rcu against v3.18-rc6 to get some 0day
> testing on it.
>
> There are conflicts against the tip of my tree, but will worry about
> that later. I don't have a ready answer on whether or not this needs
> to be split up -- in the worst case, we would add the SRCU Kconfig
> parameter, add all the selects, then once they all hit mainline add the
> Makefile change.
>
> We will need stats on how much memory was saved, and a "size kernel/rcu/srcu.o"
> after a UP/tiny build should do it. Could you please add this to the
> commit log?
>
So if we do not select CONFIG_SRCU, srcu.o will not be compiled at all. So we should save about
text data bss dec hex filename
2007 0 0 2007 7d7 kernel/rcu/srcu.o
as in size of arch/powerpc/boot/zImage changes from
text data bss dec hex filename
831552 64180 23944 919676 e087c arch/powerpc/boot/zImage : before
829504 64180 23952 917636 e0084 arch/powerpc/boot/zImage : after
so about ~2000 bytes I guess.
--
Pranith
On Wed, Dec 03, 2014 at 11:03:52PM -0500, Pranith Kumar wrote:
> (CC list trimmed since this is just for initial feedback)
>
> SRCU is not necessary to be compiled by default in all cases. For tinification
> efforts not compiling SRCU unless necessary is desirable.
>
> The current patch tries to make compiling SRCU optional by introducing a new
> Kconfig option CONFIG_SRCU which is selected when any of the components making
> use of SRCU are selected.
>
> The current tinyconfig on x86 enables SRCU by default since x86 enables
> PERF_EVENTS which in-turn require SRCU support. If we disable
> PERF_EVENTS, we can disable compiling SRCU for tiny configs on x86 systems.
To the best of my knowledge, "select" still doesn't recursively enable
things that the selected symbol selects, so I think you need to "select
SRCU" from X86, right next to the "select PERF_EVENTS".
Also, can you try your patch on top of the current patch series to
compile without perf (tiny/disable-perf), and see if you can
successfully compile without SRCU on x86?
> I tested this using randconfig and building about 15 kernels. This is definitely
> not complete and could use more testing. Sticking it in next might help in that
> regards.
>
> Comments and suggestions are welcome. Please let me know if I should split up
> this patch.
I don't think this needs splitting; it's a very small and obvious patch
combined with a few selects.
The code looks good to me. Assuming it compiles on x86, with tinyconfig
and with allyesconfig minus SRCU (and whatever requires it), this seems
reasonable.
Thanks for working on this!
- Josh Triplett
> v2:
> - handle cpufreq, devfreq and notifier
> - build tested tinyconfig on powerpc
>
> arch/arm/kvm/Kconfig | 1 +
> arch/arm64/kvm/Kconfig | 1 +
> arch/ia64/kvm/Kconfig | 1 +
> arch/mips/kvm/Kconfig | 1 +
> arch/powerpc/kvm/Kconfig | 1 +
> arch/s390/kvm/Kconfig | 1 +
> arch/tile/kvm/Kconfig | 1 +
> arch/x86/kvm/Kconfig | 1 +
> drivers/clk/Kconfig | 1 +
> drivers/cpufreq/Kconfig | 1 +
> drivers/devfreq/Kconfig | 1 +
> drivers/md/Kconfig | 1 +
> drivers/net/Kconfig | 1 +
> fs/btrfs/Kconfig | 1 +
> fs/notify/Kconfig | 1 +
> init/Kconfig | 10 ++++++++++
> kernel/notifier.c | 3 +++
> kernel/rcu/Makefile | 3 ++-
> lib/Kconfig.debug | 1 +
> mm/Kconfig | 1 +
> security/tomoyo/Kconfig | 1 +
> 21 files changed, 33 insertions(+), 1 deletion(-)
>
> diff --git a/arch/arm/kvm/Kconfig b/arch/arm/kvm/Kconfig
> index 466bd29..3afee5f 100644
> --- a/arch/arm/kvm/Kconfig
> +++ b/arch/arm/kvm/Kconfig
> @@ -23,6 +23,7 @@ config KVM
> select HAVE_KVM_CPU_RELAX_INTERCEPT
> select KVM_MMIO
> select KVM_ARM_HOST
> + select SRCU
> depends on ARM_VIRT_EXT && ARM_LPAE
> ---help---
> Support hosting virtualized guest machines. You will also
> diff --git a/arch/arm64/kvm/Kconfig b/arch/arm64/kvm/Kconfig
> index 8ba85e9..b334084 100644
> --- a/arch/arm64/kvm/Kconfig
> +++ b/arch/arm64/kvm/Kconfig
> @@ -26,6 +26,7 @@ config KVM
> select KVM_ARM_HOST
> select KVM_ARM_VGIC
> select KVM_ARM_TIMER
> + select SRCU
> ---help---
> Support hosting virtualized guest machines.
>
> diff --git a/arch/ia64/kvm/Kconfig b/arch/ia64/kvm/Kconfig
> index 3d50ea9..f7d380b 100644
> --- a/arch/ia64/kvm/Kconfig
> +++ b/arch/ia64/kvm/Kconfig
> @@ -29,6 +29,7 @@ config KVM
> select HAVE_KVM_IRQ_ROUTING
> select KVM_APIC_ARCHITECTURE
> select KVM_MMIO
> + select SRCU
> ---help---
> Support hosting fully virtualized guest machines using hardware
> virtualization extensions. You will need a fairly recent
> diff --git a/arch/mips/kvm/Kconfig b/arch/mips/kvm/Kconfig
> index 30e334e..2ae1282 100644
> --- a/arch/mips/kvm/Kconfig
> +++ b/arch/mips/kvm/Kconfig
> @@ -20,6 +20,7 @@ config KVM
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> select KVM_MMIO
> + select SRCU
> ---help---
> Support for hosting Guest kernels.
> Currently supported on MIPS32 processors.
> diff --git a/arch/powerpc/kvm/Kconfig b/arch/powerpc/kvm/Kconfig
> index 602eb51..af18e0f 100644
> --- a/arch/powerpc/kvm/Kconfig
> +++ b/arch/powerpc/kvm/Kconfig
> @@ -21,6 +21,7 @@ config KVM
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> select HAVE_KVM_EVENTFD
> + select SRCU
>
> config KVM_BOOK3S_HANDLER
> bool
> diff --git a/arch/s390/kvm/Kconfig b/arch/s390/kvm/Kconfig
> index 646db9c..5fce52c 100644
> --- a/arch/s390/kvm/Kconfig
> +++ b/arch/s390/kvm/Kconfig
> @@ -28,6 +28,7 @@ config KVM
> select HAVE_KVM_IRQCHIP
> select HAVE_KVM_IRQFD
> select HAVE_KVM_IRQ_ROUTING
> + select SRCU
> ---help---
> Support hosting paravirtualized guest machines using the SIE
> virtualization capability on the mainframe. This should work
> diff --git a/arch/tile/kvm/Kconfig b/arch/tile/kvm/Kconfig
> index 2298cb1..1e968f7 100644
> --- a/arch/tile/kvm/Kconfig
> +++ b/arch/tile/kvm/Kconfig
> @@ -21,6 +21,7 @@ config KVM
> depends on HAVE_KVM && MODULES
> select PREEMPT_NOTIFIERS
> select ANON_INODES
> + select SRCU
> ---help---
> Support hosting paravirtualized guest machines.
>
> diff --git a/arch/x86/kvm/Kconfig b/arch/x86/kvm/Kconfig
> index f9d16ff..7dc7ba5 100644
> --- a/arch/x86/kvm/Kconfig
> +++ b/arch/x86/kvm/Kconfig
> @@ -40,6 +40,7 @@ config KVM
> select HAVE_KVM_MSI
> select HAVE_KVM_CPU_RELAX_INTERCEPT
> select KVM_VFIO
> + select SRCU
> ---help---
> Support hosting fully virtualized guest machines using hardware
> virtualization extensions. You will need a fairly recent
> diff --git a/drivers/clk/Kconfig b/drivers/clk/Kconfig
> index 455fd17..53e64f1 100644
> --- a/drivers/clk/Kconfig
> +++ b/drivers/clk/Kconfig
> @@ -13,6 +13,7 @@ config COMMON_CLK
> bool
> select HAVE_CLK_PREPARE
> select CLKDEV_LOOKUP
> + select SRCU
> ---help---
> The common clock framework is a single definition of struct
> clk, useful across many platforms, as well as an
> diff --git a/drivers/cpufreq/Kconfig b/drivers/cpufreq/Kconfig
> index 3489f8f..ebaa0cc 100644
> --- a/drivers/cpufreq/Kconfig
> +++ b/drivers/cpufreq/Kconfig
> @@ -2,6 +2,7 @@ menu "CPU Frequency scaling"
>
> config CPU_FREQ
> bool "CPU Frequency scaling"
> + select SRCU
> help
> CPU Frequency scaling allows you to change the clock speed of
> CPUs on the fly. This is a nice method to save power, because
> diff --git a/drivers/devfreq/Kconfig b/drivers/devfreq/Kconfig
> index faf4e70..3891f67 100644
> --- a/drivers/devfreq/Kconfig
> +++ b/drivers/devfreq/Kconfig
> @@ -1,5 +1,6 @@
> menuconfig PM_DEVFREQ
> bool "Generic Dynamic Voltage and Frequency Scaling (DVFS) support"
> + select SRCU
> help
> A device may have a list of frequencies and voltages available.
> devfreq, a generic DVFS framework can be registered for a device
> diff --git a/drivers/md/Kconfig b/drivers/md/Kconfig
> index 5bdedf6..c355a22 100644
> --- a/drivers/md/Kconfig
> +++ b/drivers/md/Kconfig
> @@ -5,6 +5,7 @@
> menuconfig MD
> bool "Multiple devices driver support (RAID and LVM)"
> depends on BLOCK
> + select SRCU
> help
> Support multiple physical spindles through a single logical device.
> Required for RAID and logical volume management.
> diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
> index 4706386..221e623 100644
> --- a/drivers/net/Kconfig
> +++ b/drivers/net/Kconfig
> @@ -176,6 +176,7 @@ config NETCONSOLE_DYNAMIC
>
> config NETPOLL
> def_bool NETCONSOLE
> + select SRCU
>
> config NET_POLL_CONTROLLER
> def_bool NETPOLL
> diff --git a/fs/btrfs/Kconfig b/fs/btrfs/Kconfig
> index a66768e..80e9c18 100644
> --- a/fs/btrfs/Kconfig
> +++ b/fs/btrfs/Kconfig
> @@ -8,6 +8,7 @@ config BTRFS_FS
> select LZO_DECOMPRESS
> select RAID6_PQ
> select XOR_BLOCKS
> + select SRCU
>
> help
> Btrfs is a general purpose copy-on-write filesystem with extents,
> diff --git a/fs/notify/Kconfig b/fs/notify/Kconfig
> index 22c629e..2a24249 100644
> --- a/fs/notify/Kconfig
> +++ b/fs/notify/Kconfig
> @@ -1,5 +1,6 @@
> config FSNOTIFY
> def_bool n
> + select SRCU
>
> source "fs/notify/dnotify/Kconfig"
> source "fs/notify/inotify/Kconfig"
> diff --git a/init/Kconfig b/init/Kconfig
> index 3ee28ae..80ac0a4 100644
> --- a/init/Kconfig
> +++ b/init/Kconfig
> @@ -501,6 +501,14 @@ config TINY_RCU
>
> endchoice
>
> +config SRCU
> + bool "Sleepable form of RCU"
> + def_bool n
> + help
> + This option selects the sleepable version of RCU. This version
> + permits arbitrary sleeping or blocking within RCU read-side critical
> + sections.
> +
> config PREEMPT_RCU
> def_bool TREE_PREEMPT_RCU
> help
> @@ -510,6 +518,7 @@ config PREEMPT_RCU
> config TASKS_RCU
> bool "Task_based RCU implementation using voluntary context switch"
> default n
> + select SRCU
> help
> This option enables a task-based RCU implementation that uses
> only voluntary context switch (not preemption!), idle, and
> @@ -1586,6 +1595,7 @@ config PERF_EVENTS
> depends on HAVE_PERF_EVENTS
> select ANON_INODES
> select IRQ_WORK
> + select SRCU
> help
> Enable kernel support for various performance events provided
> by software and hardware.
> diff --git a/kernel/notifier.c b/kernel/notifier.c
> index 4803da6..f52858a 100644
> --- a/kernel/notifier.c
> +++ b/kernel/notifier.c
> @@ -402,6 +402,7 @@ int raw_notifier_call_chain(struct raw_notifier_head *nh,
> }
> EXPORT_SYMBOL_GPL(raw_notifier_call_chain);
>
> +#if CONFIG_SRCU
> /*
> * SRCU notifier chain routines. Registration and unregistration
> * use a mutex, and call_chain is synchronized by SRCU (no locks).
> @@ -528,6 +529,8 @@ void srcu_init_notifier_head(struct srcu_notifier_head *nh)
> }
> EXPORT_SYMBOL_GPL(srcu_init_notifier_head);
>
> +#endif /* CONFIG_SRCU */
> +
> static ATOMIC_NOTIFIER_HEAD(die_chain);
>
> int notrace notify_die(enum die_val val, const char *str,
> diff --git a/kernel/rcu/Makefile b/kernel/rcu/Makefile
> index 807ccfb..a484080 100644
> --- a/kernel/rcu/Makefile
> +++ b/kernel/rcu/Makefile
> @@ -1,4 +1,5 @@
> -obj-y += update.o srcu.o
> +obj-y += update.o
> +obj-$(CONFIG_SRCU) += srcu.o
> obj-$(CONFIG_RCU_TORTURE_TEST) += rcutorture.o
> obj-$(CONFIG_TREE_RCU) += tree.o
> obj-$(CONFIG_TREE_PREEMPT_RCU) += tree.o
> diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> index 4e35a5d..797511a 100644
> --- a/lib/Kconfig.debug
> +++ b/lib/Kconfig.debug
> @@ -1199,6 +1199,7 @@ config RCU_TORTURE_TEST
> tristate "torture tests for RCU"
> depends on DEBUG_KERNEL
> select TORTURE_TEST
> + select SRCU
> default n
> help
> This option provides a kernel module that runs torture tests
> diff --git a/mm/Kconfig b/mm/Kconfig
> index 1d1ae6b..4395b12 100644
> --- a/mm/Kconfig
> +++ b/mm/Kconfig
> @@ -325,6 +325,7 @@ config VIRT_TO_BUS
>
> config MMU_NOTIFIER
> bool
> + select SRCU
>
> config KSM
> bool "Enable KSM for page merging"
> diff --git a/security/tomoyo/Kconfig b/security/tomoyo/Kconfig
> index 8eb779b9d..604e718 100644
> --- a/security/tomoyo/Kconfig
> +++ b/security/tomoyo/Kconfig
> @@ -5,6 +5,7 @@ config SECURITY_TOMOYO
> select SECURITYFS
> select SECURITY_PATH
> select SECURITY_NETWORK
> + select SRCU
> default n
> help
> This selects TOMOYO Linux, pathname-based access control.
> --
> 1.9.1
>
On Thu, Dec 04, 2014 at 10:37:39AM -0500, Pranith Kumar wrote:
> On 12/04/2014 01:27 AM, Paul E. McKenney wrote:
> > For the moment, I applied this to -rcu against v3.18-rc6 to get some 0day
> > testing on it.
> >
> > There are conflicts against the tip of my tree, but will worry about
> > that later. I don't have a ready answer on whether or not this needs
> > to be split up -- in the worst case, we would add the SRCU Kconfig
> > parameter, add all the selects, then once they all hit mainline add the
> > Makefile change.
> >
> > We will need stats on how much memory was saved, and a "size kernel/rcu/srcu.o"
> > after a UP/tiny build should do it. Could you please add this to the
> > commit log?
> >
>
> So if we do not select CONFIG_SRCU, srcu.o will not be compiled at all. So we should save about
>
> text data bss dec hex filename
> 2007 0 0 2007 7d7 kernel/rcu/srcu.o
>
> as in size of arch/powerpc/boot/zImage changes from
>
> text data bss dec hex filename
> 831552 64180 23944 919676 e087c arch/powerpc/boot/zImage : before
> 829504 64180 23952 917636 e0084 arch/powerpc/boot/zImage : after
>
> so about ~2000 bytes I guess.
Nice!
Josh, I recall you having a tinified .config somewhere, but don't see it
right off. If you do have one, could you please share it with Pranith?
Not a big deal, just don't want to overstate the savings too much. ;-)
Thanx, Paul
(resending, forgot CC's)
On Thu, Dec 4, 2014 at 11:13 AM, Josh Triplett <[email protected]> wrote:
>>
>> The current tinyconfig on x86 enables SRCU by default since x86 enables
>> PERF_EVENTS which in-turn require SRCU support. If we disable
>> PERF_EVENTS, we can disable compiling SRCU for tiny configs on x86 systems.
>
> To the best of my knowledge, "select" still doesn't recursively enable
> things that the selected symbol selects, so I think you need to "select
> SRCU" from X86, right next to the "select PERF_EVENTS".
Doesn't this prevent not selecting SRCU on x86 systems?
I think a warning will be generated during make if a dependent config
is not enabled.
>
> Also, can you try your patch on top of the current patch series to
> compile without perf (tiny/disable-perf), and see if you can
> successfully compile without SRCU on x86?
This branch had a small config error which was compiling perf related
code unconditionally.
warning: (HW_BREAKPOINTS) selects HAVE_HW_BREAKPOINT which has unmet
direct dependencies (PERF_EVENTS)
With the following patch applied on top of it I was able to compile my patch.
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 10c51ff..0ecc069 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -588,6 +588,7 @@ config HW_BREAKPOINTS
default y
select HAVE_HW_BREAKPOINT
select HAVE_MIXED_BREAKPOINTS_REGS
+ depends on PERF_EVENTS
---help---
Enable support for x86 hardware breakpoints for debuggers
and perf. This will implicitly enable perf-events.
>
>> I tested this using randconfig and building about 15 kernels. This is definitely
>> not complete and could use more testing. Sticking it in next might help in that
>> regards.
>>
>> Comments and suggestions are welcome. Please let me know if I should split up
>> this patch.
>
> I don't think this needs splitting; it's a very small and obvious patch
> combined with a few selects.
>
> The code looks good to me. Assuming it compiles on x86, with tinyconfig
> and with allyesconfig minus SRCU (and whatever requires it), this seems
For allyesconfig minus SRCU(using tiny/disable-perf) I needed the
following patch to get it to build
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 10c51ff..0ecc069 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -588,6 +588,7 @@ config HW_BREAKPOINTS
default y
select HAVE_HW_BREAKPOINT
select HAVE_MIXED_BREAKPOINTS_REGS
+ depends on PERF_EVENTS
---help---
Enable support for x86 hardware breakpoints for debuggers
and perf. This will implicitly enable perf-events.
diff --git a/fs/Kconfig.binfmt b/fs/Kconfig.binfmt
index 370b24c..2669ddf 100644
--- a/fs/Kconfig.binfmt
+++ b/fs/Kconfig.binfmt
@@ -182,6 +182,7 @@ config BINFMT_MISC
config COREDUMP
bool "Enable core dump support" if EXPERT
default y
+ depends on PERF_EVENTS
help
This option enables support for performing core dumps. You almost
certainly want to say Y here. Not necessary on systems that never
diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
index 8639819..b008c597 100644
--- a/kernel/trace/Kconfig
+++ b/kernel/trace/Kconfig
@@ -125,6 +125,7 @@ config TRACING_SUPPORT
# irqflags tracing for your architecture.
depends on TRACE_IRQFLAGS_SUPPORT || PPC32
depends on STACKTRACE_SUPPORT
+ depends on PERF_EVENTS
default y
if TRACING_SUPPORT
--
If the above two patches are good, then may be you should add them to
the tiny/disable-perf branch?
I will submit SRCU patch on top of rcu/dev so that Paul can pick it up
in his tree.
Thanks!
--
Pranith