Move shutdown and reboot related code to a separate file, out of
process.c. This helps to avoid polluting process.c with non-process
related code.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Normally, when a CPU wants to clear a cache line to zero in the external
L2 cache, it would generate bus cycles to write each word as it would do
with any other data access.
However, a Cortex A9 connected to a L2C-310 has a specific feature where
the CPU can detect this operation, and signal that it wants to zero an
entire cache line. This feature, known as Full Line of Zeros (FLZ),
involves a non-standard AXI signalling mechanism which only the L2C-310
can properly interpret.
There are separate enable bits in both the L2C-310 and the Cortex A9 -
the L2C-310 needs to be enabled and have the FLZ enable bit set in the
auxiliary control register before the Cortex A9 has this feature
enabled.
Unfortunately, the suspend code was not respecting this - it's not
obvious from the code:
swsusp_arch_suspend()
cpu_suspend() /* saves the Cortex A9 auxiliary control register */
arch_save_image()
soft_restart() /* turns off FLZ in Cortex A9, and disables L2C */
cpu_resume() /* restores the Cortex A9 registers, inc auxcr */
At this point, we end up with the L2C disabled, but the Cortex A9 with
FLZ enabled - which means any memset() or zeroing of a full cache line
will fail to take effect.
A similar issue exists in the resume path, but it's slightly more
complex:
swsusp_arch_suspend()
cpu_suspend() /* saves the Cortex A9 auxiliary control register */
arch_save_image() /* image with A9 auxcr saved */
...
swsusp_arch_resume()
call_with_stack()
arch_restore_image() /* restores image with A9 auxcr saved above */
soft_restart() /* turns off FLZ in Cortex A9, and disables L2C */
cpu_resume() /* restores the Cortex A9 registers, inc auxcr */
Again, here we end up with the L2C disabled, but Cortex A9 FLZ enabled.
There's no need to turn off the L2C in either of these two paths; there
are benefits from not doing so - for example, the page copies will be
faster with the L2C enabled.
Hence, fix this by providing a variant of soft_restart() which can be
used without turning the L2 cache controller off, and use it in both
of these paths to keep the L2C enabled across the respective resume
transitions.
Fixes: 8ef418c717 ("ARM: l2c: trial at enabling some Cortex-A9 optimisations")
Reported-by: Sean Cross <xobs@kosagi.com>
Tested-by: Sean Cross <xobs@kosagi.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Initialize the VDSO page list at boot, install the VDSO mapping at
exec time, and update the data page during timer ticks. This code is
not built if CONFIG_VDSO is not enabled.
Account for the VDSO length when randomizing the offset from the
stack. The [vdso] and [vvar] pages are placed immediately following
the sigpage with separate _install_special_mapping calls.
We want to "penalize" systems lacking the arch timer as little
as possible. Previous versions of this code installed the VDSO
unconditionally and unmodified, making it a measurably slower way for
glibc to invoke the real syscalls on such systems. E.g. calling
gettimeofday via glibc goes from ~560ns to ~630ns on i.MX6Q.
If we can indicate to glibc that the time-related APIs in the VDSO are
not accelerated, glibc can continue to invoke the syscalls directly
instead of dispatching through the VDSO only to fall back to the slow
path.
Thus, if the architected timer is unusable for whatever reason, patch
the VDSO at boot time so that symbol lookups for gettimeofday and
clock_gettime return NULL. (This is similar to what powerpc does and
borrows code from there.) This allows glibc to perform the syscall
directly instead of passing control to the VDSO, which minimizes the
penalty. In my measurements the time taken for a gettimeofday call
via glibc goes from ~560ns to ~580ns (again on i.MX6Q), and this is
solely due to adding a test and branch to glibc's gettimeofday syscall
wrapper.
An alternative to patching the VDSO at boot would be to not install
the VDSO at all when the arch timer isn't usable. Another alternative
is to include a separate "dummy" vdso.so without gettimeofday and
clock_gettime, which would be selected at boot time. Either of these
would get cumbersome if the VDSO were to gain support for an API such
as getcpu which is unrelated to arch timer support.
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
If the kernel is running in hypervisor mode or monitor mode we'll
print UK6_32 or UK10_32 if we call into __show_regs(). Let's
update these strings to indicate the new modes that didn't exist
when this code was written.
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Pull restart handler infrastructure from Guenter Roeck:
"This series was supposed to be pulled through various trees using it,
and I did not plan to send a separate pull request. As it turns out,
the pinctrl tree did not merge with it, is now upstream, and uses it,
meaning there are now build failures.
Please pull this series directly to fix those build failures"
* tag 'restart-handler-for-v3.18' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
arm/arm64: unexport restart handlers
watchdog: sunxi: register restart handler with kernel restart handler
watchdog: alim7101: register restart handler with kernel restart handler
watchdog: moxart: register restart handler with kernel restart handler
arm: support restart through restart handler call chain
arm64: support restart through restart handler call chain
power/restart: call machine_restart instead of arm_pm_restart
kernel: add support for kernel restart handler call chain
The sigpage is currently placed alongside shared libraries etc in the
address space. Similar to what x86_64 does for its VDSO, place the
sigpage at a randomized offset above the stack so that learning the
base address of the sigpage doesn't help expose where shared libraries
are loaded in the address space (and vice versa).
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
_install_special_mapping allows the VMA to be identifed in
/proc/pid/maps without the use of arch_vma_name, providing a
slight net reduction in object size:
text data bss dec hex filename
2996 96 144 3236 ca4 arch/arm/kernel/process.o (before)
2956 104 144 3204 c84 arch/arm/kernel/process.o (after)
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
The TPIDRURO and TPIDRURW registers need to be flushed during exec;
otherwise TLS information is potentially leaked. TPIDRURO in
particular needs careful treatment. Since flush_thread basically
needs the same code used to set the TLS in arm_syscall, pull that into
a common set_tls helper in tls.h and use it in both places.
Similarly, TEEHBR needs to be cleared during exec as well. Clearing
its save slot in thread_info isn't right as there is no guarantee
that a thread switch will occur before the new program runs. Just
setting the register directly is sufficient.
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Acked-by: Will Deacon <will.deacon@arm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Pull second set of ARM changes from Russell King:
"This is the remainder of the ARM changes for this merge window.
Included in this request are:
- fixes for kprobes for big-endian support
- fix tracing in soft_restart
- avoid phys address overflow in kdump code
- fix reporting of read-only pmd bits in kernel page table dump
- remove unnecessary (and possibly buggy) call to outer_flush_all()
- fix a three sparse warnings (missing header file for function
prototypes)
- fix pj4 crashing single zImage (thanks to arm-soc merging changes
which enables this with knowledge that the corresponding fix had
not even been submitted for my tree before the merge window opened)
- vfp macro cleanups
- dump register state on undefined instruction userspace faults when
debugging"
* 'for-linus' of git://ftp.arm.linux.org.uk/~rmk/linux-arm:
Dump the registers on undefined instruction userspace faults
ARM: 8018/1: Add {inc,dec}_preempt_count asm macros
ARM: 8017/1: Move asm macro get_thread_info to asm/assembler.h
ARM: 8016/1: Check cpu id in pj4_cp0_init.
ARM: 8015/1: Add cpu_is_pj4 to distinguish PJ4 because it has some differences with V7
ARM: add missing system_misc.h include to process.c
ARM: 8009/1: dcscb.c: remove call to outer_flush_all()
ARM: 8014/1: mm: fix reporting of read-only PMD bits
ARM: 8012/1: kdump: Avoid overflow when converting pfn to physaddr
ARM: 8010/1: avoid tracers in soft_restart
ARM: kprobes-test: Workaround GAS .align bug
ARM: kprobes-test: use <asm/opcodes.h> for Thumb instruction building
ARM: kprobes-test: use <asm/opcodes.h> for ARM instruction building
ARM: kprobes-test: use <asm/opcodes.h> for instruction accesses
ARM: probes: fix instruction fetch order with <asm/opcodes.h>
arm_pm_restart(), arm_pm_idle() and soft_restart() are all declared in
system_misc.h, but this file is not included in process.c. Add this
missing include. Found via sparse:
arch/arm/kernel/process.c:98:6: warning: symbol 'soft_restart' was not declared. Should it be static?
arch/arm/kernel/process.c:127:6: warning: symbol 'arm_pm_restart' was not declared. Should it be static?
arch/arm/kernel/process.c:134:6: warning: symbol 'arm_pm_idle' was not declared. Should it be static?
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Use of tracers in local_irq_disable is causes abort loops when called
with irqs disabled using a temporary stack. Replace local_irq_disable
with raw_local_irq_disable instead to avoid tracers.
Signed-off-by: Sebastian Capella <sebastian.capella@linaro.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Pull ARM changes from Russell King:
- Perf updates from Will Deacon:
- Support for Qualcomm Krait processors (run perf on your phone!)
- Support for Cortex-A12 (run perf stat on your FPGA!)
- Support for perf_sample_event_took, allowing us to automatically decrease
the sample rate if we can't handle the PMU interrupts quickly enough
(run perf record on your FPGA!).
- Basic uprobes support from David Long:
This patch series adds basic uprobes support to ARM. It is based on
patches developed earlier by Rabin Vincent. That approach of adding
hooks into the kprobes instruction parsing code was not well received.
This approach separates the ARM instruction parsing code in kprobes out
into a separate set of functions which can be used by both kprobes and
uprobes. Both kprobes and uprobes then provide their own semantic action
tables to process the results of the parsing.
- ARMv7M (microcontroller) updates from Uwe Kleine-König
- OMAP DMA updates (recently added Vinod's Ack even though they've been
sitting in linux-next for a few months) to reduce the reliance of
omap-dma on the code in arch/arm.
- SA11x0 changes from Dmitry Eremin-Solenikov and Alexander Shiyan
- Support for Cortex-A12 CPU
- Align support for ARMv6 with ARMv7 so they can cooperate better in a
single zImage.
- Addition of first AT_HWCAP2 feature bits for ARMv8 crypto support.
- Removal of IRQ_DISABLED from various ARM files
- Improved efficiency of virt_to_page() for single zImage
- Patch from Ulf Hansson to permit runtime PM callbacks to be available for
AMBA devices for suspend/resume as well.
- Finally kill asm/system.h on ARM.
* 'for-linus' of git://ftp.arm.linux.org.uk/~rmk/linux-arm: (89 commits)
dmaengine: omap-dma: more consolidation of CCR register setup
dmaengine: omap-dma: move IRQ handling to omap-dma
dmaengine: omap-dma: move register read/writes into omap-dma.c
ARM: omap: dma: get rid of 'p' allocation and clean up
ARM: omap: move dma channel allocation into plat-omap code
ARM: omap: dma: get rid of errata global
ARM: omap: clean up DMA register accesses
ARM: omap: remove almost-const variables
ARM: omap: remove references to disable_irq_lch
dmaengine: omap-dma: cleanup errata 3.3 handling
dmaengine: omap-dma: provide register read/write functions
dmaengine: omap-dma: use cached CCR value when enabling DMA
dmaengine: omap-dma: move barrier to omap_dma_start_desc()
dmaengine: omap-dma: move clnk_ctrl setting to preparation functions
dmaengine: omap-dma: improve efficiency loading C.SA/C.EI/C.FI registers
dmaengine: omap-dma: consolidate clearing channel status register
dmaengine: omap-dma: move CCR buffering disable errata out of the fast path
dmaengine: omap-dma: provide register definitions
dmaengine: omap-dma: consolidate setup of CCR
dmaengine: omap-dma: consolidate setup of CSDP
...
no indication about irqs in PSR and only a single ISA. So skip the whole
decoding and just print the xPSR on v7-M.
Also mark two static variables as __maybe_unused to prevent the compiler
from emitting:
arch/arm/kernel/process.c:51:20: warning: 'processor_modes' defined but not used [-Wunused-variable]
arch/arm/kernel/process.c:58:20: warning: 'isa_modes' defined but not used [-Wunused-variable]
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Unfortunately, I never committed the fix to a nasty oops which can
occur as a result of that commit:
------------[ cut here ]------------
kernel BUG at /home/olof/work/batch/include/linux/mm.h:414!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 0 PID: 490 Comm: killall5 Not tainted 3.11.0-rc3-00288-gabe0308 #53
task: e90acac0 ti: e9be8000 task.ti: e9be8000
PC is at special_mapping_fault+0xa4/0xc4
LR is at __do_fault+0x68/0x48c
This doesn't show up unless you do quite a bit of testing; a simple
boot test does not do this, so all my nightly tests were passing fine.
The reason for this is that install_special_mapping() expects the
page array to stick around, and as this was only inserting one page
which was stored on the kernel stack, that's why this was blowing up.
Reported-by: Olof Johansson <olof@lixom.net>
Tested-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
