Until there is a generic MIPS way of handing the DTB over from bootloader to
kernel we rely on a built in devicetrees. The OF code also remaps those register
ranges that we use global in our drivers.
Signed-off-by: John Crispin <blogic@openwrt.org>
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/4895/
The current code uses multiple if statements for
demultiplexing the different interrupt sources.
Additionally, the MISC interrupt controller has
32 interrupt sources and the current code does not
handles all of them.
Get rid of the if statements and process all interrupt
sources in a loop to fix these issues.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/4874/
Signed-off-by: John Crispin <blogic@openwrt.org>
GPIO function selection is not working on the AR934x
SoCs because the offset of the function selection
register is different on those.
Add a helper routine which returns the correct
register address based on the SoC type, and use
that in the 'ath79_gpio_function_*' routines.
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/4870/
Signed-off-by: John Crispin <blogic@openwrt.org>
Newer toolchains support the DSP and DSP Rev2 instructions. This patch
performs a check for that support and adds compiler and assembler
flags for only the files that need use those instructions.
Signed-off-by: Steven J. Hill <sjhill@mips.com>
Acked-by: Florian Fainelli <florian@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/4752/
Signed-off-by: John Crispin <blogic@openwrt.org>
On multi-chip boards, the first core on slave SoCs may take much
more time to wakeup. Add code to wait for the core to come up before
proceeding with the rest of the boot up.
Update xlp_wakeup_core to also skip the boot node and the boot CPU
initialization which is already complete.
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4783/
Signed-off-by: John Crispin <blogic@openwrt.org>
Doing calibrate delay on a hardware thread will be inaccurate since
it depends on the load on other threads in the core. It will also
slow down the boot process when done for 128 hardware threads. Switch
to a pre-computed loops per jiffy based on the core frequency. The
value is computed based on the core frequency and roughly matches the
value calculated by calibrate_delay().
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4791/
Signed-off-by: John Crispin <blogic@openwrt.org>
The XLR/XLS/XLP PIC has a 8 countdown timers which run at the PIC
frequencey. One of these can be used as a clocksource to provide
timestamps that is common across cores. This can be used in place
of the count/compare clocksource which is per-CPU.
On XLR/XLS PIC registers are 32-bit, so we just use the lower 32-bits
of the PIC counter. On XLP, the whole 64-bit can be used.
Provide common macros and functions for PIC timer registers on XLR/XLS
and XLP, and use them to register a PIC clocksource.
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4786/
Signed-off-by: John Crispin <blogic@openwrt.org>
Since we now use r4k cache code for Netlogic XLP, it is
better to split L1 icache among the active threads, so that
threads won't step on each other while flushing icache.
The L1 dcache is already split among the threads in the core.
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4787/
Signed-off-by: John Crispin <blogic@openwrt.org>
Rename function xlp_enable_pci_bswap() to xlp_config_pci_bswap(), which
is a better description for its functionality. When compiled in
big-endian mode, xlp_config_pci_bswap() will configure the PCIe links
to byteswap. In little-endian mode, no swap configuration is needed
for the PCIe controller, and the function is empty.
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4802/
Signed-off-by: John Crispin <blogic@openwrt.org>
Provide functions ack_c0_eirr(), set_c0_eimr(), clear_c0_eimr()
and read_c0_eirr_and_eimr() that do the EIMR and EIRR operations
and update the interrupt handling code to use these functions.
Also, use the EIMR register functions to mask interrupts in the
irq code.
The 64-bit interrupt request and mask registers (EIRR and EIMR) are
accessed when the interrupts are off, and the common operations are
to set or clear a bit in these registers. Using the 64-bit c0 access
functions for these operations is not optimal in 32-bit, because it
will disable/restore interrupts and split/join the 64-bit value during
each register access.
Signed-off-by: Jayachandran C <jchandra@broadcom.com>
Patchwork: http://patchwork.linux-mips.org/patch/4790/
Signed-off-by: John Crispin <blogic@openwrt.org>
This code makes the irqs used by the EIU loadable from the DT. Additionally we
add a helper that allows the pinctrl layer to map external irqs to real irq
numbers.
Signed-off-by: John Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/4818/
The mmid macro is meant to be used to get the mm->context.id data
from the mm structure, but it seems to have been missed in a cuple
of files.
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Since the new ASID code in b5466f8728
("ARM: mm: remove IPI broadcasting on ASID rollover") was changed to
use 64bit operations it has broken the BE operation due to an issue
with the MM code accessing sub-fields of mm->context.id.
When running in BE mode we see the values in mm->context.id are stored
with the highest value first, so the LDR in the arch/arm/mm/proc-macros.S
reads the wrong part of this field. To resolve this, change the LDR in
the mmid macro to load from +4.
Acked-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
In order to allow drivers to specify private data for each controller,
this commit adds a private_data field to the struct hw_pci. This field
is an array of nr_controllers pointers that will be used to initialize
the private_data field of the corresponding controller's pci_sys_data
structure.
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
When using deferred driver probing, PCI host controller drivers may
actually require this function after the init stage.
Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
869486d5f51 (ARM: 7646/1: mm: use static_vm for managing static mapped
areas) introduced new warnings:
arch/arm/mm/mmu.c: In function 'pci_reserve_io':
arch/arm/mm/mmu.c:888:16: warning: unused variable 'addr'
arch/arm/mm/mmu.c:887:20: warning: unused variable 'vm'
because it failed to delete the two local variables it no longer used.
Fix this.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
A static mapped area is ARM-specific, so it is better not to use
generic vmalloc data structure, that is, vmlist and vmlist_lock
for managing static mapped area. And it causes some needless overhead and
reducing this overhead is better idea.
Now, we have newly introduced static_vm infrastructure.
With it, we don't need to iterate all mapped areas. Instead, we just
iterate static mapped areas. It helps to reduce an overhead of finding
matched area. And architecture dependency on vmalloc layer is removed,
so it will help to maintainability for vmalloc layer.
Reviewed-by: Nicolas Pitre <nico@linaro.org>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
In current implementation, we used ARM-specific flag, that is,
VM_ARM_STATIC_MAPPING, for distinguishing ARM specific static mapped area.
The purpose of static mapped area is to re-use static mapped area when
entire physical address range of the ioremap request can be covered
by this area.
This implementation causes needless overhead for some cases.
For example, assume that there is only one static mapped area and
vmlist has 300 areas. Every time we call ioremap, we check 300 areas for
deciding whether it is matched or not. Moreover, even if there is
no static mapped area and vmlist has 300 areas, every time we call
ioremap, we check 300 areas in now.
If we construct a extra list for static mapped area, we can eliminate
above mentioned overhead.
With a extra list, if there is one static mapped area,
we just check only one area and proceed next operation quickly.
In fact, it is not a critical problem, because ioremap is not frequently
used. But reducing overhead is better idea.
Another reason for doing this work is for removing architecture dependency
on vmalloc layer. I think that vmlist and vmlist_lock is internal data
structure for vmalloc layer. Some codes for debugging and stat inevitably
use vmlist and vmlist_lock. But it is preferable that they are used
as least as possible in outside of vmalloc.c
Now, I introduce an ARM-specific infrastructure for static mapped area. In
the following patch, we will use this and resolve above mentioned problem.
Reviewed-by: Nicolas Pitre <nico@linaro.org>
Tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Parts of the virtual memory layout (mainly the modules area) are
described using open-coded immediate values.
Use the SZ_ definitions from linux/sizes.h instead to make the code
clearer.
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>
On AMD family 15h processors, there are 4 new performance
counters (in addition to 6 core performance counters) that can
be used for counting northbridge events (i.e. DRAM accesses).
Their bit fields are almost identical to the core performance
counters. However, unlike the core performance counters, these
MSRs are shared between multiple cores (that share the same
northbridge).
We will reuse the same code path as existing family 10h
northbridge event constraints handler logic to enforce
this sharing.
Signed-off-by: Jacob Shin <jacob.shin@amd.com>
Acked-by: Stephane Eranian <eranian@google.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Jacob Shin <jacob.shin@amd.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1360171589-6381-7-git-send-email-jacob.shin@amd.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
