CLOCKSOURCE: mips-gic: Allow GIC clock to be specified in device-tree
As an alternative to the "clock-frequency" property, allow the GIC
timer operating clock to be specified in the device-tree instead.
This is useful on systems which use common clock or where the GIC
is not fixed to a particular frequency and is instead, for example,
derived from the CPU clock.
Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Cc: James Hogan <james.hogan@imgtec.com> Cc: Rob Herring <robh+dt@kernel.org> Cc: Pawel Moll <pawel.moll@arm.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Ian Campbell <ijc+devicetree@hellion.org.uk> Cc: Kumar Gala <galak@codeaurora.org> Cc: Daniel Lezcano <daniel.lezcano@linaro.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: devicetree@vger.kernel.org Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9309/
Computing sum introduces true data dependency. This patch removes some
true data depdendencies, hence increases instruction level parallelism.
This patch brings up to 50% csum performance gain on Loongson 3a.
One example about how this patch works is in CSUM_BIGCHUNK1:
// ** original ** vs ** patch applied **
ADDC(sum, t0) ADDC(t0, t1)
ADDC(sum, t1) ADDC(t2, t3)
ADDC(sum, t2) ADDC(sum, t0)
ADDC(sum, t3) ADDC(sum, t2)
In the original implementation, each ADDC(sum, ...) depends on the sum
value updated by previous ADDC(as source operand).
With this patch applied, the first two ADDC operations are independent,
hence can be executed simultaneously if possible.
Another example is in the "copy and sum calculating chunk":
// ** original ** vs ** patch applied **
STORE(t0, UNIT(0) ... STORE(t0, UNIT(0) ...
ADDC(sum, t0) ADDC(t0, t1)
STORE(t1, UNIT(1) ... STORE(t1, UNIT(1) ...
ADDC(sum, t1) ADDC(sum, t0)
STORE(t2, UNIT(2) ... STORE(t2, UNIT(2) ...
ADDC(sum, t2) ADDC(t2, t3)
STORE(t3, UNIT(3) ... STORE(t3, UNIT(3) ...
ADDC(sum, t3) ADDC(sum, t2)
With this patch applied, ADDC and the **next next** ADDC are independent.
Huacai Chen [Sun, 29 Mar 2015 02:54:05 +0000 (10:54 +0800)]
MIPS: Hibernate: flush TLB entries earlier
We found that TLB mismatch not only happens after kernel resume, but
also happens during snapshot restore. So move it to the beginning of
swsusp_arch_suspend().
Signed-off-by: Huacai Chen <chenhc@lemote.com> Cc: <stable@vger.kernel.org> Cc: Steven J. Hill <Steven.Hill@imgtec.com> Cc: linux-mips@linux-mips.org Cc: Fuxin Zhang <zhangfx@lemote.com> Cc: Zhangjin Wu <wuzhangjin@gmail.com> Cc: stable@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9621/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Florian Fainelli [Thu, 26 Mar 2015 04:55:15 +0000 (21:55 -0700)]
MIPS: BMIPS: restrict DTB selection to BMIPS_GENERIC
Since we are always sourcing arch/mips/bmips/Kconfig and there is no
dependency on BMIPS_GENERIC, we will offer building BMIPS-related DTBs
while this is not relevant for the other MIPS platforms.
Ralf Baechle [Fri, 27 Mar 2015 14:17:31 +0000 (15:17 +0100)]
MIPS: BMIPS: Flush the readahead cache after DMA.
BMIPS 3300/435x/438x CPUs have a readahead cache that is separate from
the L1/L2. During a DMA operation, accesses adjacent to a DMA buffer
may cause parts of the DMA buffer to be prefetched into the RAC. To
avoid possible coherency problems, flush the RAC upon DMA completion.
Derived from Kevin Cernekee's https://patchwork.linux-mips.org/patch/9602/.
Ralf Baechle [Fri, 27 Mar 2015 12:11:51 +0000 (13:11 +0100)]
MIPS: ath25: Remove unused DMA helper functions.
These got merged with the ath25 support after 4e7f72660c39 (MIPS: Remove
unnecessary platform dma helper functions) had already removed them for
all other platforms.
Rafał Miłecki [Mon, 1 Dec 2014 06:58:18 +0000 (07:58 +0100)]
MIPS: BCM47xx: Move NVRAM header to the include/linux/.
There are two reasons for having this header in the common place:
1) Simplifying drivers that read NVRAM entries. We will be able to
safely call bcm47xx_nvram_* functions without #ifdef-s.
2) Getting NVRAM driver out of MIPS arch code. This is needed to support
BCM5301X arch which also requires this NVRAM driver. Patch for that
will follow once we get is reviewed.
David Daney [Tue, 24 Feb 2015 23:35:34 +0000 (15:35 -0800)]
MIPS: Expand __swp_offset() to carry 40 significant bits for 64-bit kernel.
With CONFIG_MIGRATION, the PFN of the migrating pages is stored in
__swp_offset(), so we must have enough bits to store the largest
possible PFN. OCTEON NUMA systems have 41 bits of physical address
space, so with 4K pages (12-bits), we need at least 29 bits to store
the PFN.
The current width of 24-bits is too narrow, so expand it all the way
out to 40-bits. This leaves the low order 16 bits as zero which does
not interfere with any of the PTE bits.
MIPS: Provide fallback reboot/poweroff/halt implementations
If a machine-specific hook is not implemented for restart, poweroff,
or halt, fall back to halting secondary CPUs, disabling interrupts,
and spinning. In the case of restart, attempt to restart the system
via do_kernel_restart() (which will call any registered restart
handlers) before halting.
Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Cc: James Hogan <james.hogan@imgtec.com> Cc: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9600/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
MIPS: smp: Make stop_this_cpu() actually stop the CPU
Since cpu_wait() enables interrupts upon return, CPUs which have
entered stop_this_cpu() may still end up handling interrupts.
This can lead to the softlockup detector firing on a panic or
restart/poweroff/halt. Just disable interrupts and spin to ensure
nothing else runs on the CPU once it has entered stop_this_cpu().
Signed-off-by: Andrew Bresticker <abrestic@chromium.org> Cc: James Hogan <james.hogan@imgtec.com> Cc: Maciej W. Rozycki <macro@linux-mips.org> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9601/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
David Daney [Wed, 3 Dec 2014 19:12:23 +0000 (11:12 -0800)]
MIPS: Add FPU emulator counter for emulated delay slots.
Delay slot emulation in the FPU emulator is the only kernel user of an
executable stack, it is also very slow. Add a counter so we can see
how many of these emulations are done.
Jayachandran C [Wed, 7 Jan 2015 11:28:35 +0000 (16:58 +0530)]
MIPS: Netlogic: Do not enable SUE for core
Enabling the SUE bit for core can can result in rare cache errors
which are difficult to track down, so do not enable it. This can
cause a minor performance loss in some tests.
Jayachandran C [Wed, 7 Jan 2015 11:28:33 +0000 (16:58 +0530)]
MIPS: Netlogic: Update function to read DRAM BARs
Change name of xlp_get_dram_map to nlm_get_dram_map to be consistent
with the rest of the functions in the file. Pass the the size of the
array 'dram_map' to the function, and ensure that it does not write
past the end of the array.
Jayachandran C [Wed, 7 Jan 2015 11:28:29 +0000 (16:58 +0530)]
MIPS: Netlogic: Use MIPS topology.h
commit bda4584cd943 ("MIPS: Support CPU topology files in sysfs")
added topology related macros for all MIPS platforms and commit bbbf6d8768f5 ("MIPS: NL: Fix nlm_xlp_defconfig build error")
removed most of the contents from mach-netlogic/topology.h.
The netlogic specific topology is not needed anymore, we just need
to setup the package field in current_cpu_data.
MIPS: Netlogic: Fix frequency calculation register
Change the PIC frequency calculation to use the register that has the
current configuration. The existing code used the register that is
written to change frequency, which can have an invalid value if the
firmware did not set it up correctly.
Qingmin Liu [Wed, 7 Jan 2015 11:28:23 +0000 (16:58 +0530)]
MIPS: Netlogic: Fix nlm_xlp2_get_pic_frequency to use ref_div
The variable ref_div is initialized to the correct divisor but not
used in the frequency calculation. This caused incorrect frequency
to be reported when the clock divisor is not 3.
Signed-off-by: Qingmin Liu <qingmin@broadcom.com> Signed-off-by: Jayachandran C <jchandra@broadcom.com> Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8884/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Shanghui Liu [Wed, 7 Jan 2015 11:28:22 +0000 (16:58 +0530)]
MIPS: Netlogic: Fix wait for slave CPUs
For core 0, the condition of "cpu == bootcpu" is always true, so it
does not wait for other three threads to become ready. Fix this by
using correct check.
Signed-off-by: Shanghui Liu <shliu@broadcom.com> Signed-off-by: Jayachandran C <jchandra@broadcom.com> Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/8881/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Remove variable self-assignments.
This silences a bunch of -Wself-assign warnings reported by clang.
The changed code can be compiled without warnings by both gcc and clang.
Daniel Sanders [Tue, 24 Feb 2015 15:25:10 +0000 (15:25 +0000)]
MIPS: LLVMLinux: Fix an 'inline asm input/output type mismatch' error.
Replace incorrect matching constraint that caused the error with an alternative
that still has the required constraints on the inline assembly.
This is the error message reported by clang:
arch/mips/include/asm/checksum.h:285:27: error: unsupported inline asm: input with type '__be32' (aka 'unsigned int') matching output with type 'unsigned short'
"0" (htonl(len)), "1" (htonl(proto)), "r" (sum));
^~~~~~~~~~~~
The changed code can be compiled successfully by both gcc and clang.
Signed-off-by: Daniel Sanders <daniel.sanders@imgtec.com> Signed-off-by: Toma Tabacu <toma.tabacu@imgtec.com> Suggested-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: Leonid Yegoshin <Leonid.Yegoshin@imgtec.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9313/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Toma Tabacu [Tue, 24 Feb 2015 15:25:09 +0000 (15:25 +0000)]
MIPS: LLVMLinux: Fix a 'cast to type not present in union' error.
Remove a cast to the 'mips16e_instruction' union inside an if
condition and instead do an assignment to a local
'union mips16e_instruction' variable's 'full' member before the if
statement and use this variable in the if condition.
This is the error message reported by clang:
arch/mips/kernel/branch.c:38:8: error: cast to union type from type 'unsigned short' not present in union
if (((union mips16e_instruction)inst).ri.opcode
^ ~~~~
The changed code can be compiled successfully by both gcc and clang.
Signed-off-by: Toma Tabacu <toma.tabacu@imgtec.com> Signed-off-by: Daniel Sanders <daniel.sanders@imgtec.com> Cc: Andreas Herrmann <andreas.herrmann@caviumnetworks.com> Cc: David Daney <david.daney@cavium.com> Cc: Manuel Lauss <manuel.lauss@gmail.com> Cc: linux-mips@linux-mips.org Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/9312/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Daniel Sanders [Tue, 24 Feb 2015 15:02:57 +0000 (15:02 +0000)]
MIPS: Changed current_thread_info() to an equivalent supported by both clang and GCC
Without this, a 'break' instruction is executed very early in the boot and
the boot hangs.
The problem is that clang doesn't honour named registers on local variables
and silently treats them as normal uninitialized variables. However, it
does honour them on global variables.
Signed-off-by: Daniel Sanders <daniel.sanders@imgtec.com> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Paul Burton <paul.burton@imgtec.com> Cc: Markos Chandras <markos.chandras@imgtec.com> Cc: James Hogan <james.hogan@imgtec.com> Cc: linux-mips@linux-mips.org Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Cc: David Daney <ddaney.cavm@gmail.com> Acked-by: Behan Webster <behanw@converseincode.com>
Patchwork: https://patchwork.linux-mips.org/patch/9311/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Steven J. Hill [Fri, 27 Feb 2015 00:16:39 +0000 (18:16 -0600)]
MIPS: XPA: Add new configuration file.
Add in new config files for enabling a XPA platform.
Signed-off-by: Steven J. Hill <Steven.Hill@imgtec.com> Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/9354/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
David Daney [Thu, 5 Mar 2015 14:31:29 +0000 (17:31 +0300)]
MIPS: OCTEON: Add semaphore to serialize bootbus accesses.
Some hardware blocks attached to the OCTEON bootbus run asynchronously
to accesses from the CPUs. These include MMC/SD host, CF(when using
DMA), and NAND controller. A bus error, or corrupt data may occur if
a CPU is trying to access a bootbus connected device at the same time
the bus is running asynchronous operations.
To work around these problems we add this semaphore that must be
acquired before initiating bootbus activity. Subsequent patches will
add users for this.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:20 +0000 (09:49 -0800)]
MIPS: BMIPS: Add DTS files for several platforms
Most of the supported chips use legacy (non-DT) bootloaders, so they will
need to select an appropriate builtin DTB at compile time until the
bootloader is updated. Provide suitable DTS files, and a means to compile
one of them into the kernel image.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:18 +0000 (09:49 -0800)]
MIPS: BMIPS: Refresh BCM3384 DTS files
The DT bindings for this platform have changed as the bootloader and
product requirements evolved. In particular, there are both
Linux-on-Zephyr and Linux-on-Viper configurations.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:15 +0000 (09:49 -0800)]
MIPS: BMIPS: Delete the irqchip driver from irq.c
BCM3384/BCM63xx can use the common drivers/irqchip/irq-bcm7120-l2.c for
this purpose; BCM7xxx will use drivers/irqchip/irq-bcm7038-l1.c. We no
longer need this code under arch/mips.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:06 +0000 (09:49 -0800)]
IRQCHIP: Add new driver for BCM7038-style level 1 interrupt controllers
This is the main peripheral IRQ controller on the BCM7xxx MIPS chips;
it has the following characteristics:
- 64 to 160+ level IRQs
- Atomic set/clear registers
- Reasonably predictable register layout (N status words, then N
mask status words, then N mask set words, then N mask clear words)
- SMP affinity supported on most systems
- Typically connected to MIPS IRQ 2,3,2,3 on CPUs 0,1,2,3
This driver registers one IRQ domain and one IRQ chip to cover all
instances of the block. Up to 4 instances of the block may appear, as
it supports 4-way IRQ affinity on BCM7435.
The same block exists on the ARM BCM7xxx chips, but typically the ARM GIC
is used instead. So this driver is primarily intended for MIPS STB chips.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:05 +0000 (09:49 -0800)]
IRQCHIP: bcm7120-l2: Add support for BCM3380-style controllers
These controllers support multiple enable/status pairs (64+ IRQs),
can put the enable/status words at different offsets, and do not
support multiple parent IRQs.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:04 +0000 (09:49 -0800)]
IRQCHIP: bcm7120-l2: Split STB-specific logic into its own function
The BCM7xxx instances of this block (listed in the register manual as
simply "IRQ0") all have the following items in common:
- brcm,int-map-mask: for routing different bits in the L2 to different
parent IRQs
- brcm,int-fwd-mask: for hardwiring certain IRQs to bypass the L2 and
use dedicated L1 lines
- one enable/status pair (32 bits only)
Much of the driver code can be shared with BCM3380-style controllers, but
in order to do this cleanly, let's split out the BCM7xxx-specific logic
first.
Kevin Cernekee [Thu, 25 Dec 2014 17:49:03 +0000 (09:49 -0800)]
IRQCHIP: bcm7120-l2: Refactor driver for arbitrary IRQEN/IRQSTAT offsets
Currently the driver assumes that REG_BASE+0x00 is the IRQ enable mask,
and REG_BASE+0x04 is the IRQ status mask. This is true on BCM3384 and
BCM7xxx, but it is not true for some of the controllers found on BCM63xx
chips. So we will change a couple of key assumptions:
- Don't assume that both the IRQEN and IRQSTAT registers will be
covered by a single ioremap() operation.
- Don't assume any particular ordering (IRQSTAT might show up before
IRQEN on some chips).
- For an L2 controller with >=64 IRQs, don't assume that every
IRQEN/IRQSTAT pair will use the same register spacing.
This patch changes the "plumbing" but doesn't yet provide a way for users
to instantiate a controller with arbitrary IRQEN/IRQSTAT offsets.
Brian Norris [Thu, 25 Dec 2014 17:49:02 +0000 (09:49 -0800)]
IRQCHIP: brcmstb-l2: don't clear wakeable interrupts at init time
Wakeable interrupts might be pending at boot/init time, because wakeup
interrupts might have triggered a resume from S5. So don't clear such
wakeups.
This means that any driver which requests a wakeable interrupt bit
should be prepared to handle an interrupt as soon as they call
request_irq(). (This is technically already the correct development
practice, but some drivers probably expect not to receive interrupts
until they have performed some I/O.)
Kevin Cernekee [Thu, 25 Dec 2014 17:49:00 +0000 (09:49 -0800)]
MIPS: bcm3384: Rename "bcm3384" target to "bmips"
This platform is configured primarily through device tree, and we can
reuse the same code to support a bunch of other chips. Change the name
to reflect this.
[ralf@linux-mips.org: Fix conflicts with other patches.]
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:35 +0000 (10:30 -0800)]
MIPS: Add support for fine granularity task level IRQ time accounting
With sched_clock being ready, it makes sense to add the option of IRQ time
accounting -- When we have a fast enough sched_clock, IRQ time accounting
will be enabled (see sched_clock_register).
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:33 +0000 (10:30 -0800)]
MIPS: csrc-sb1250: Remove FSF mail address from GPL notice
This is to make checkpatch.pl happy for the next patch. It would otherwise
say --
ERROR: Do not include the paragraph about writing to the Free Software
Foundation's mailing address from the sample GPL notice. The FSF has
changed addresses in the past, and may do so again. Linux already includes
a copy of the GPL.
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:27 +0000 (10:30 -0800)]
MIPS: csrc-ioasic: Remove FSF mail address from GPL notice
This is to make checkpatch.pl happy for the next patch. It would otherwise
say --
ERROR: Do not include the paragraph about writing to the Free Software
Foundation's mailing address from the sample GPL notice. The FSF has
changed addresses in the past, and may do so again. Linux already includes
a copy of the GPL.
Acked-by: Maciej W. Rozycki <macro@linux-mips.org> Signed-off-by: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com> Cc: linux-mips@linux-mips.org Cc: macro@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/9481/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:25 +0000 (10:30 -0800)]
MIPS: csrc-bcm1480: Remove FSF mail address from GPL notice
This is to make checkpatch.pl happy for the next patch. It would otherwise
say --
ERROR: Do not include the paragraph about writing to the Free Software
Foundation's mailing address from the sample GPL notice. The FSF has
changed addresses in the past, and may do so again. Linux already includes
a copy of the GPL.
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:23 +0000 (10:30 -0800)]
MIPS: Add sched_clock support
This will provide sched_clock interface to implement individual
read_sched_clock(). Not for CAVIUM_OCTEON_SOC as it defines its own
sched_clock() directly (not using the sched_clock_register interface).
GENERIC_SCHED_CLOCK can be selected by architectures other than ARM. The
current dependencies of CLKSRC_VERSATILE make it possible that other
architectures will have CLKSRC_VERSATILE available in configuration once
they select GENERIC_SCHED_CLOCK, whereas this clock source should be solely
available to ARM in reality.
This patch adds one more dependency to CLKSRC_VERSATILE to fix the issue.
Signed-off-by: Deng-Cheng Zhu <dengcheng.zhu@imgtec.com> Reported-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Russell King <linux@arm.linux.org.uk> Cc: LKML <linux-kernel@vger.kernel.org> Cc: linux-arm-kernel@lists.infradead.org Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/9476/ Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:20 +0000 (10:30 -0800)]
MIPS: Fall back to generic implementation of cmpxchg64 on 32-bit platforms
This is in preparation of adding HAVE_VIRT_CPU_ACCOUNTING_GEN support in
the next patch.
Without having cmpxchg64 to use the generic implementation, kernel linking
will complain:
kernel/built-in.o: In function `cputime_adjust':
cputime.c:(.text+0x33748): undefined reference to `__cmpxchg_called_with_bad_pointer'
cputime.c:(.text+0x33810): undefined reference to `__cmpxchg_called_with_bad_pointer'
Deng-Cheng Zhu [Sat, 7 Mar 2015 18:30:19 +0000 (10:30 -0800)]
MIPS: Add SCHED_HRTICK support
We have HIGH_RES_TIMERS to support SCHED_HRTICK. But SCHED_HRTICK is in
kernel/Kconfig.hz where HZ values unsuitable for MIPS are defined. So we
simply add this config in arch/mips/Kconfig as opposed to including the
whole kernel/Kconfig.hz.
projects / linux-beck.git / log