CPU interrupts need to be disabled on a cpu being taken down.
When a cpu is hot-plugged out of the system the following sequence occurs.
On the CPU where the hotplug sequence was initiated:
cpu_down
_cpu_down {
__cpu_notify(CPU_DOWN_PREPARE
__stop_machine(take_cpu_down
wait for cpu to run disable code.
__cpu_die
}
On the CPU being disabled:
take_cpu_down
__cpu_disable {
mp_ops->cpu_disable
bmips_cpu_disable
clear_c0_status(IE_IRQ5) (added)
cpu_notify(CPU_DYING...
}
Before the cpu_notifier is called with CPU_DYING, all interrupts on the
dying cpu must be disabled. This guarantees that before tick_notify is
called with the CPU_DYING event and sets the clock device pointer to
NULL, there can not be any more clock interrupts.
When this wasn't done, an unfortunately-timed timer interrupt sometimes
caused hangs immediately prior to system suspend:
Debug PM is not enabled. To enable partial suspend, rebuild kernel with CONFIG_PM_DEBUG
Pass 1 out of 1,PM: Syncing filesystems ... mode=none, tp1=done.
1, flags=5, cycle_tp=, sleep=
Freezing user space processes ... (elapsed 0.01 seconds) done.
Freezing remaining freezable tasks ... (elapsed 0.01 seconds) done.
PM: suspend of devices complete after 54.199 msecs
PM: late suspend of devices complete after 0.172 msecs
Disabling non-boot CPUs ...
SMP: CPU1 is offline
INFO: rcu_sched detected stalls on CPUs/tasks: { 3} (detected by 0, t=62537 jiffies)
Call Trace:
[<804baa78>] dump_stack+0x8/0x34
[<8008a2d8>] __rcu_pending+0x4b8/0x55c
[<8008adf4>] rcu_check_callbacks+0x78/0x180
[<80037830>] update_process_times+0x40/0x6c
[<80072fe4>] tick_sched_timer+0x74/0xe4
[<80050180>] __run_hrtimer.clone.30+0x64/0x140
[<80051150>] hrtimer_interrupt+0x19c/0x4bc
[<8000cdb8>] c0_compare_interrupt+0x50/0x88
[<80081b18>] handle_irq_event_percpu+0x5c/0x2f4
[<80086490>] handle_percpu_irq+0x8c/0xc0
[<800811b4>] generic_handle_irq+0x34/0x54
[<800067dc>] do_IRQ+0x18/0x2c
[<8000375c>] plat_irq_dispatch+0xd0/0x128
[<80004a04>] ret_from_irq+0x0/0x4
[<80004c40>] r4k_wait+0x20/0x40
[<80006b6c>] cpu_idle+0x98/0xf0
[<805d3988>] start_kernel+0x424/0x440
Signed-off-by: Jon Fraser <jfraser@broadcom.com>
Signed-off-by: Kevin Cernekee <cernekee@gmail.com>
Cc: f.fainelli@gmail.com
Cc: mbizon@freebox.fr
Cc: jogo@openwrt.org
Cc: linux-mips@linux-mips.org
Cc: devicetree@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/8160/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The hybrid FPR scheme exists to allow for compatibility between existing
FP32 code and newly compiled FP64A code. Such code should hopefully be
rare in the real world, and for the moment is difficult to come across.
All code except that built for the FP64 ABI can correctly execute using
the hybrid FPR scheme, so debugging the hybrid FPR implementation can
be eased by forcing all such code to use it. This is undesirable in
general due to the trap & emulate overhead of the hybrid FPR
implementation, but is a very useful option to have for debugging.
Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: linux-fsdevel@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/7680/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Hybrid FPRs is a scheme where scalar FP registers are 64b wide, but
accesses to odd indexed single registers use bits 63:32 of the
preceeding even indexed 64b register. In this mode all FP code
except that built for the plain FP64 ABI can execute correctly. Most
notably a combination of FP64A & FP32 code can execute correctly,
allowing for existing FP32 binaries to be linked with new FP64A binaries
that can make use of 64 bit FP & MSA.
Hybrid FPRs are implemented by setting both the FR & FRE bits, trapping
& emulating single precision FP instructions (via Reserved Instruction
exceptions) whilst allowing others to execute natively. It therefore has
a penalty in terms of execution speed, and should only be used when no
fully native mode can be. As more binaries are recompiled to use either
the FPXX or FP64(A) ABIs, the need for hybrid FPRs should diminish.
However in the short to mid term it allows for a gradual transition
towards that world, rather than a complete ABI break which is not
feasible for some users & not desirable for many.
A task will be executed using the hybrid FPR scheme when its
TIF_HYBRID_FPREGS flag is set & TIF_32BIT_FPREGS is clear. A further
patch will set the flags as necessary, this patch simply adds the
infrastructure necessary for the hybrid FPR mode to work.
Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: linux-fsdevel@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/7683/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The MIPS GIC supports 7 local interrupts, 2 of which are the GIC
local watchdog and count/compare timer. The remainder are CPU
interrupts which may optionally be re-routed through the GIC.
GIC hardware IRQs 0-6 are now used for local interrupts while
hardware IRQs 7+ are used for external (shared) interrupts.
Note that the 5 CPU interrupts may not be re-routable through
the GIC. In that case mapping will fail and the vectors reported
in C0_IntCtl should be used instead. gic_get_c0_compare_int() and
gic_get_c0_perfcount_int() will return the correct IRQ number to
use for the C0 timer and perfcounter interrupts based on the
routability of those interrupts through the GIC.
A separate irq_chip, with callbacks that mask/unmask the local
interrupt on all CPUs, is used for the C0 timer and performance
counter interrupts since all other platforms do not use the percpu
IRQ API for those interrupts.
Malta, SEAD-3, and the GIC clockevent driver have been updated
to use local interrupts and the R4K clockevent driver has been
updated to poll for C0 timer interrupts through the GIC when
the GIC is present.
Signed-off-by: Andrew Bresticker <abrestic@chromium.org>
Acked-by: Jason Cooper <jason@lakedaemon.net>
Reviewed-by: Qais Yousef <qais.yousef@imgtec.com>
Tested-by: Qais Yousef <qais.yousef@imgtec.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Jeffrey Deans <jeffrey.deans@imgtec.com>
Cc: Markos Chandras <markos.chandras@imgtec.com>
Cc: Paul Burton <paul.burton@imgtec.com>
Cc: Jonas Gorski <jogo@openwrt.org>
Cc: John Crispin <blogic@openwrt.org>
Cc: David Daney <ddaney.cavm@gmail.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/7819/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Add new 'noftlb' kernel command line option to disable the FTLB.
Since the kernel command line is not available when probing and
enabling the CPU features in cpu_probe(), we let the kernel configure
the FTLB during the config4 decode operation and we disable the FTLB later
on, once the command line has become available to us. This should have
no negative effects since FTLB isn't used so early in the boot process.
FTLB increases the effective TLB size leading to less TLB misses. However,
sometimes it's useful to be able to disable it when debugging memory related
core features or other hardware components.
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: http://patchwork.linux-mips.org/patch/7586/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Currently, arch_trigger_all_cpu_backtrace() is defined in only x86 and
sparc which have an NMI. But in case of softlockup, it could be possible
to dump backtrace of all cpus. and this could be helpful for debugging.
for example, if system has 2 cpus.
CPU 0 CPU 1
acquire read_lock()
try to do write_lock()
,,,
missing read_unlock()
In this case, softlockup will occur becasuse CPU 0 does not call
read_unlock(). And dump_stack() print only backtrace for "CPU 0". If
CPU1's backtrace is printed it's very helpful.
[ralf@linux-mips.org: Fixed whitespace and formatting issues.]
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/8200/
Based on the spatch
@@
expression e;
@@
- return (e);
+ return e;
with heavy hand editing because some of the changes are either whitespace
or identation only or result in excessivly long lines.
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
The save_fp_context & restore_fp_context pointers were being assigned
to the wrong variables if either:
- The kernel is configured for UP & runs on a system without an FPU,
since b2ead52828 "MIPS: Move & rename
fpu_emulator_{save,restore}_context".
- The kernel is configured for EVA, since ca750649e0 "MIPS: kernel:
signal: Prevent save/restore FPU context in user memory".
This would lead to FP context being clobbered incorrectly when setting
up a sigcontext, then the garbage values being saved uselessly when
returning from the signal.
Fix by swapping the pointer assignments appropriately.
Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: stable@vger.kernel.org # v3.15+
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8230/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Make use of the Config6/FLTBP bit to set the probability of a TLBWR
instruction to hit the FTLB or the VTLB. A value of 0 (which may be
the default value on certain cores, such as proAptiv or P5600)
means that a TLBWR instruction will never hit the VTLB which
leads to performance limitations since it effectively decreases
the number of available TLB slots.
Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
Reviewed-by: James Hogan <james.hogan@imgtec.com>
Cc: <stable@vger.kernel.org> # v3.15+
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8368/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Implement the microMIPS encoding of the J instruction for the purpose of
the static keys feature, fixing a crash early on in bootstrap as the
kernel is unhappy seeing the ISA bit set in jump table entries. Make
sure the ISA bit correctly reflects the instruction encoding chosen for
the kernel, 0 for the standard MIPS and 1 for the microMIPS encoding.
Also make sure the instruction to patch is a 32-bit NOP in the microMIPS
mode as by default the 16-bit short encoding is assumed
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8516/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Correct the check for the span of the 256MB segment addressable by the J
instruction according to this instruction's semantics. The calculation
of the jump target is applied to the address of the delay-slot
instruction that immediately follows. Adjust the check accordingly by
adding 4 to `e->code' that holds the address of the J instruction
itself.
Signed-off-by: Maciej W. Rozycki <macro@codesourcery.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8515/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Starting with version 2.24.51.20140728 MIPS binutils complain loudly
about mixing soft-float and hard-float object files, leading to this
build failure since GCC is invoked with "-msoft-float" on MIPS:
{standard input}: Warning: .gnu_attribute 4,3 requires `softfloat'
LD arch/mips/alchemy/common/built-in.o
mipsel-softfloat-linux-gnu-ld: Warning: arch/mips/alchemy/common/built-in.o
uses -msoft-float (set by arch/mips/alchemy/common/prom.o),
arch/mips/alchemy/common/sleeper.o uses -mhard-float
To fix this, we detect if GAS is new enough to support "-msoft-float" command
option, and if it does, we can let GCC pass it to GAS; but then we also need
to sprinkle the files which make use of floating point registers with the
necessary ".set hardfloat" directives.
Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Cc: Linux-MIPS <linux-mips@linux-mips.org>
Cc: Matthew Fortune <Matthew.Fortune@imgtec.com>
Cc: Markos Chandras <Markos.Chandras@imgtec.com>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Patchwork: https://patchwork.linux-mips.org/patch/8355/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Even if CMA is disabled, the for_each_memblock macro expands
to run reserve_bootmem once. Hence, reserve_bootmem attempts to
reserve location 0 of size 0.
Add a check to avoid that.
Issue was highlighted during testing with EVA enabled.
resrve_bootmem used to exit gracefully when passed arguments to
reserve 0 size location at 0 without EVA.
But with EVA enabled, macros would point to different addresses
and the code would trigger a BUG.
Signed-off-by: Zubair Lutfullah Kakakhel <Zubair.Kakakhel@imgtec.com>
Tested-by: Markos Chandras <markos.chandras@imgtec.com>
Tested-by: Huacai Chen <chenhc@lemote.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/8231/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
