Architectures should fully validate whether kexec is possible as part of
machine_kexec_prepare(), so that user-space's kexec_load() operation can
report any problems. Performing validation in machine_kexec() itself is
too late, since it is not allowed to return.
Prior to this patch, ARM's machine_kexec() was testing after-the-fact
whether machine_kexec_prepare() was able to disable all but one CPU.
Instead, modify machine_kexec_prepare() to validate all conditions
necessary for machine_kexec_prepare()'s to succeed. BUG if the validation
succeeded, yet disabling the CPUs didn't actually work.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The __cpuinit type of throwaway sections might have made sense
some time ago when RAM was more constrained, but now the savings
do not offset the cost and complications. For example, the fix in
commit 5e427ec2d0 ("x86: Fix bit corruption at CPU resume time")
is a good example of the nasty type of bugs that can be created
with improper use of the various __init prefixes.
After a discussion on LKML[1] it was decided that cpuinit should go
the way of devinit and be phased out. Once all the users are gone,
we can then finally remove the macros themselves from linux/init.h.
Note that some harmless section mismatch warnings may result, since
notify_cpu_starting() and cpu_up() are arch independent (kernel/cpu.c)
and are flagged as __cpuinit -- so if we remove the __cpuinit from
the arch specific callers, we will also get section mismatch warnings.
As an intermediate step, we intend to turn the linux/init.h cpuinit
related content into no-ops as early as possible, since that will get
rid of these warnings. In any case, they are temporary and harmless.
This removes all the ARM uses of the __cpuinit macros from C code,
and all __CPUINIT from assembly code. It also had two ".previous"
section statements that were paired off against __CPUINIT
(aka .section ".cpuinit.text") that also get removed here.
[1] https://lkml.org/lkml/2013/5/20/589
Cc: Russell King <linux@arm.linux.org.uk>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Conflicts:
arch/arm/kernel/smp.c
Please pull these miscellaneous LPAE fixes I've been collecting for a while
now for 3.11. They've been tested and reviewed by quite a few people, and most
of the patches are pretty trivial. -- Will Deacon.
Add comments to machine_shutdown()/halt()/power_off()/restart() that
describe their purpose and/or requirements re: CPUs being active/not.
In machine_shutdown(), replace the call to smp_send_stop() with a call to
disable_nonboot_cpus(). This completely disables all but one CPU, thus
satisfying the requirement that only a single CPU be active for kexec.
Adjust Kconfig dependencies for this change.
In machine_halt()/power_off()/restart(), call smp_send_stop() directly,
rather than via machine_shutdown(); these functions don't need to
completely de-activate all CPUs using hotplug, but rather just quiesce
them.
Remove smp_kill_cpus(), and its call from smp_send_stop().
smp_kill_cpus() was indirectly calling smp_ops.cpu_kill() without calling
smp_ops.cpu_die() on the target CPUs first. At least some implementations
of smp_ops had issues with this; it caused cpu_kill() to hang on Tegra,
for example. Since smp_send_stop() is only used for shutdown, halt, and
power-off, there is no need to attempt any kind of CPU hotplug here.
Adjust Kconfig to reflect that machine_shutdown() (and hence kexec)
relies upon disable_nonboot_cpus(). However, this alone doesn't guarantee
that hotplug will work, or even that hotplug is implemented for a
particular piece of HW that a multi-platform zImage runs on. Hence, add
error-checking to machine_kexec() to determine whether it did work.
Suggested-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Tested-by: Zhangfei Gao <zhangfei.gao@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The MPU initialisation on the primary core is performed in two stages, one
minimal stage to ensure the CPU can boot and a second one after
sanity_check_meminfo. As the memory configuration is known by the time we
boot secondary cores only a single step is necessary, provided the values
for DRSR are passed to secondaries.
This patch implements this arrangement. The configuration generated for the
MPU regions is made available to the secondary core, which can then use the
asm MPU intialisation code to program a complete region configuration.
This is necessary for SMP configurations without an MMU, as the MPU
initialisation is the only way to ensure that memory is specified as
'shared'.
Signed-off-by: Jonathan Austin <jonathan.austin@arm.com>
Reviewed-by: Will Deacon <will.deacon@arm.com>
CC: Nicolas Pitre <nico@linaro.org>
nommu systems do not require any page tables, so don't try to initialise
them when bringing up secondary cores.
Signed-off-by: Will Deacon <will.deacon@arm.com>
This patch redefines the early boot time use of the R4 register to steal a few
low order bits (ARCH_PGD_SHIFT bits) on LPAE systems. This allows for up to
38-bit physical addresses.
Signed-off-by: Cyril Chemparathy <cyril@ti.com>
Signed-off-by: Vitaly Andrianov <vitalya@ti.com>
Acked-by: Nicolas Pitre <nico@linaro.org>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: Subash Patel <subash.rp@samsung.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Before f7b861b7a6 ("arm: Use generic idle loop") ARM would kill the
CPU within the rcu idle section. Now that the rcu_idle_enter()/exit()
pair have been pushed lower down in the idle loop this is no longer true
and so using RCU_NONIDLE here is no longer necessary and also harmful
because RCU is not actually idle at this point.
Cc: Russell King <linux@arm.linux.org.uk>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull ARM updates from Russell King:
"The major items included in here are:
- MCPM, multi-cluster power management, part of the infrastructure
required for ARMs big.LITTLE support.
- A rework of the ARM KVM code to allow re-use by ARM64.
- Error handling cleanups of the IS_ERR_OR_NULL() madness and fixes
of that stuff for arch/arm
- Preparatory patches for Cortex-M3 support from Uwe Kleine-König.
There is also a set of three patches in here from Hugh/Catalin to
address freeing of inappropriate page tables on LPAE. You already
have these from akpm, but they were already part of my tree at the
time he sent them, so unfortunately they'll end up with duplicate
commits"
* 'for-linus' of git://git.linaro.org/people/rmk/linux-arm: (77 commits)
ARM: EXYNOS: remove unnecessary use of IS_ERR_VALUE()
ARM: IMX: remove unnecessary use of IS_ERR_VALUE()
ARM: OMAP: use consistent error checking
ARM: cleanup: OMAP hwmod error checking
ARM: 7709/1: mcpm: Add explicit AFLAGS to support v6/v7 multiplatform kernels
ARM: 7700/2: Make cpu_init() notrace
ARM: 7702/1: Set the page table freeing ceiling to TASK_SIZE
ARM: 7701/1: mm: Allow arch code to control the user page table ceiling
ARM: 7703/1: Disable preemption in broadcast_tlb*_a15_erratum()
ARM: mcpm: provide an interface to set the SMP ops at run time
ARM: mcpm: generic SMP secondary bringup and hotplug support
ARM: mcpm_head.S: vlock-based first man election
ARM: mcpm: Add baremetal voting mutexes
ARM: mcpm: introduce helpers for platform coherency exit/setup
ARM: mcpm: introduce the CPU/cluster power API
ARM: multi-cluster PM: secondary kernel entry code
ARM: cacheflush: add synchronization helpers for mixed cache state accesses
ARM: cpu hotplug: remove majority of cache flushing from platforms
ARM: smp: flush L1 cache in cpu_die()
ARM: tegra: remove tegra specific cpu_disable()
...
Flush the L1 cache for the CPU which is going down in cpu_die() so
that we don't end up with all platforms doing this. This ensures
that any cache lines we own are pushed out before the cache becomes
inaccessible.
We may end up subsequently creating some dirty cache lines - for
example, with the complete() call, but this update must become
visible to other CPUs before __cpu_die() can proceed. Subsequent
accesses from the platforms cpu_die() function should _not_ matter.
Also place a mb() after the complete() call to ensure that this is
visible to other CPUs.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Commit 70264367a2 ("ARM: 7653/2: do not scale loops_per_jiffy when
using a constant delay clock") fixed a problem with our timer-based
delay loop, where loops_per_jiffy is scaled by cpufreq yet used directly
by the timer delay ops.
This patch fixes the problem in a more elegant way by keeping a private
ticks_per_jiffy field in the delay ops, independent of loops_per_jiffy
and therefore not subject to scaling. The loop-based delay continues to
use loops_per_jiffy directly, as it should.
Acked-by: Nicolas Pitre <nico@linaro.org>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
With recent arm broadcast time clean-up from Mark Rutland, the dummy
broadcast device is always registered with timer subsystem. And since
the rating of the dummy clock event is very high, it may be preferred
over a real clock event.
This is a change in behavior from past and not an intended one. So
reduce the rating of the dummy clock-event so that real clock-event
device is selected when available.
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The ARM ARM requires branch predictor maintenance if, for a given ASID,
the instructions at a specific virtual address appear to change.
From the kernel's point of view, that means:
- Changing the kernel's view of memory (e.g. switching to the
identity map)
- ASID rollover (since ASIDs will be re-allocated to new tasks)
This patch adds explicit branch predictor maintenance when either of the
two conditions above are met.
Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Pull late ARM updates from Russell King:
"Here is the late set of ARM updates for this merge window; in here is:
- The ARM parts of the broadcast timer support, core parts merged
through tglx's tree. This was left over from the previous merge to
allow the dependency on tglx's tree to be resolved.
- A fix to the VFP code which shows up on Raspberry Pi's, as well as
fixing the fallout from a previous commit in this area.
- A number of smaller fixes scattered throughout the ARM tree"
* 'for-linus' of git://git.linaro.org/people/rmk/linux-arm:
ARM: Fix broken commit 0cc41e4a21 corrupting kernel messages
ARM: fix scheduling while atomic warning in alignment handling code
ARM: VFP: fix emulation of second VFP instruction
ARM: 7656/1: uImage: Error out on build of multiplatform without LOADADDR
ARM: 7640/1: memory: tegra_ahb_enable_smmu() depends on TEGRA_IOMMU_SMMU
ARM: 7654/1: Preserve L_PTE_VALID in pte_modify()
ARM: 7653/2: do not scale loops_per_jiffy when using a constant delay clock
ARM: 7651/1: remove unused smp_timer_broadcast #define
Pull ARM virtualization changes:
"This contains parts of the ARM KVM support that have dependencies on
other patches merged through the arm-soc tree. In combination with
patches coming through Russell's tree, this will finally add full
support for the kernel based virtual machine on ARM, which has been
awaited for some time now.
Further, we now have a separate platform for virtual machines and qemu
booting that is used by both Xen and KVM, separating these from the
Versatile Express reference implementation. Obviously, this new
platform is multiplatform capable so it can be combined with existing
machines in the same kernel."
* tag 'virt' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (38 commits)
ARM: arch_timer: include linux/errno.h
arm: arch_timer: add missing inline in stub function
ARM: KVM: arch_timers: Wire the init code and config option
ARM: KVM: arch_timers: Add timer world switch
ARM: KVM: arch_timers: Add guest timer core support
ARM: KVM: Add VGIC configuration option
ARM: KVM: VGIC initialisation code
ARM: KVM: VGIC control interface world switch
ARM: KVM: VGIC interrupt injection
ARM: KVM: vgic: retire queued, disabled interrupts
ARM: KVM: VGIC virtual CPU interface management
ARM: KVM: VGIC distributor handling
ARM: KVM: VGIC accept vcpu and dist base addresses from user space
ARM: KVM: Initial VGIC infrastructure code
ARM: KVM: Keep track of currently running vcpus
KVM: ARM: Introduce KVM_ARM_SET_DEVICE_ADDR ioctl
ARM: gic: add __ASSEMBLY__ guard to C definitions
ARM: gic: define GICH offsets for VGIC support
ARM: gic: add missing distributor defintions
ARM: mach-virt: fixup machine descriptor after removal of sys_timer
...
Pull ARM SoC cleanups from Arnd Bergmann:
"A large number of cleanups, all over the platforms. This is dominated
largely by the Samsung platforms (s3c, s5p, exynos) and a few of the
others moving code out of arch/arm into more appropriate subsystems.
The clocksource and irqchip drivers are now abstracted to the point
where platforms that are already cleaned up do not need to even
specify the driver they use, it can all get configured from the device
tree as we do for normal device drivers. The clocksource changes
basically touch every single platform in the process.
We further clean up the use of platform specific header files here,
with the goal of turning more of the platforms over to being
"multiplatform" enabled, which implies that they cannot expose their
headers to architecture independent code any more.
It is expected that no functional changes are part of the cleanup.
The overall reduction in total code lines is mostly the result of
removing broken and obsolete code."
* tag 'cleanup' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (133 commits)
ARM: mvebu: correct gated clock documentation
ARM: kirkwood: add missing include for nsa310
ARM: exynos: move exynos4210-combiner to drivers/irqchip
mfd: db8500-prcmu: update resource passing
drivers/db8500-cpufreq: delete dangling include
ARM: at91: remove NEOCORE 926 board
sunxi: Cleanup the reset code and add meaningful registers defines
ARM: S3C24XX: header mach/regs-mem.h local
ARM: S3C24XX: header mach/regs-power.h local
ARM: S3C24XX: header mach/regs-s3c2412-mem.h local
ARM: S3C24XX: Remove plat-s3c24xx directory in arch/arm/
ARM: S3C24XX: transform s3c2443 subirqs into new structure
ARM: S3C24XX: modify s3c2443 irq init to initialize all irqs
ARM: S3C24XX: move s3c2443 irq code to irq.c
ARM: S3C24XX: transform s3c2416 irqs into new structure
ARM: S3C24XX: modify s3c2416 irq init to initialize all irqs
ARM: S3C24XX: move s3c2416 irq init to common irq code
ARM: S3C24XX: Modify s3c_irq_wake to use the hwirq property
ARM: S3C24XX: Move irq syscore-ops to irq-pm
clocksource: always define CLOCKSOURCE_OF_DECLARE
...
When udelay() is implemented using an architected timer, it is wrong
to scale loops_per_jiffy when changing the CPU clock frequency since
the timer clock remains constant.
The lpj should probably become an implementation detail relevant to
the CPU loop based delay routine only and more confined to it. In the
mean time this is the minimal fix needed to have expected delays with
the timer based implementation when cpufreq is also in use.
Reported-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Nicolas Pitre <nico@linaro.org>
Tested-by: Viresh Kumar <viresh.kumar@linaro.org>
Acked-by: Liviu Dudau <Liviu.Dudau@arm.com>
Cc: stable@vger.kernel.org
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The assignment of clock_event_device::broadcast can be done by timer
core as of 12ad100046: "clockevents: Add generic timer broadcast
function", and the arm code moved over to this as of 3d06770eef: "arm:
Add generic timer broadcast support", but left a dangling #define when
!CONFIG_GENERIC_TIMER_BROADCAST.
This patch removes the now unused #define.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Implement timer_broadcast for the arm architecture, allowing for the use
of clock_event_device_drivers decoupled from the timer tick broadcast
mechanism.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Currently, the ARM backend must maintain a redundant list of timers for
the purpose of centralising timer broadcast functionality. This prevents
sharing timer drivers across architectures.
This patch moves the pain of dealing with timer broadcasts to the core
clockevents tick broadcast code, which already maintains its own list
of timers.
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Reviewed-by: Stephen Boyd <sboyd@codeaurora.org>
Remove some silly wrapper functions which aren't really required:
platform_smp_prepare_cpus
platform_secondary_init
platform_cpu_die
This simplifies the code.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Stephen Warren
Paul Gortmaker
Russell King
Jonathan Austin
Will Deacon
Cyril Chemparathy
Stephen Boyd
Linus Torvalds
Thomas Gleixner
Santosh Shilimkar
Nicolas Pitre
Mark Rutland
Olof Johansson