Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

The default 256 bytes sometimes is just not enough.
We usually provide earlycon=... and console=... and ip=...
All this and more may need more room.
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Acked-by: NVineet Gupta <vgupta@synopsys.com>

Since the CTOP is SMT hardware multi-threaded, we need to hint
the HW that now will be a very good time to do a hardware
thread context switching. This is done by issuing the schd.rw
instruction (binary coded here so as to not require specific
revision of GCC to build the kernel).
sched.rw means that Thread becomes eligible for execution by
the threads scheduler after all pending read/write
transactions were completed.

Implementing cpu_relax_lowlatency() with barrier()
Since with current semantics of cpu_relax() it may take a
while till yielded CPU will get back.
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Acked-by: NVineet Gupta <vgupta@synopsys.com>

With generic "identity" num of CPUs is limited to 256 (8 bit).
We use our alternative AUX register GLOBAL_ID (12 bit).
Now we can support up to 4096 CPUs.
Signed-off-by: NNoam Camus <noamc@ezchip.com>

NPS device got 256 cores and each got 16 HW threads (SMT).
We use EZchip dedicated ISA to trigger HW scheduler of the
core that current HW thread belongs to.
This scheduling makes sure that data beyond barrier is available
to all HW threads in core and by that to all in device (4K).
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Cc: Peter Zijlstra <peterz@infradead.org>

We need our own implementaions since we lack LLSC support.
Our extended ISA provided with optimized solution for all 32bit
operations we see in these three headers.
Signed-off-by: NNoam Camus <noamc@ezchip.com>

NPS use special mapping right below TASK_SIZE.
Hence we need to lower STACK_TOP so that user stack won't
overlap NPS special mapping.
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Acked-by: NVineet Gupta <vgupta@synopsys.com>

If we hold rwlock and interrupt occures we may
end up spinning on it for ever during softirq.
Note that this lock is an internal lock
and since the lock is free to be used from any context,
the lock needs to be IRQ-safe.

Below you may see an example for interrupt we get while
nl_table_lock is holding its rw->lock_mutex and we spinned
on it for ever.

The concept for the fix was taken from SPARC.

[2015-05-12 19:16:12] Stack Trace:
[2015-05-12 19:16:12]   arc_unwind_core+0xb8/0x11c
[2015-05-12 19:16:12]   dump_stack+0x68/0xac
[2015-05-12 19:16:12]   _raw_read_lock+0xa8/0xac
[2015-05-12 19:16:12]   netlink_broadcast_filtered+0x56/0x35c
[2015-05-12 19:16:12]   nlmsg_notify+0x42/0xa4
[2015-05-12 19:16:13]   neigh_update+0x1fe/0x44c
[2015-05-12 19:16:13]   neigh_event_ns+0x40/0xa4
[2015-05-12 19:16:13]   arp_process+0x46e/0x5a8
[2015-05-12 19:16:13]   __netif_receive_skb_core+0x358/0x500
[2015-05-12 19:16:13]   process_backlog+0x92/0x154
[2015-05-12 19:16:13]   net_rx_action+0xb8/0x188
[2015-05-12 19:16:13]   __do_softirq+0xda/0x1d8
[2015-05-12 19:16:14]   irq_exit+0x8a/0x8c
[2015-05-12 19:16:14]   arch_do_IRQ+0x6c/0xa8
[2015-05-12 19:16:14]   handle_interrupt_level1+0xe4/0xf0
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Acked-by: NPeter Zijlstra <peterz@infradead.org>

On ARC, lower 2G of address space is translated and used for
 - user vaddr space (region 0 to 5)
 - unused kernel-user gutter (region 6)
 - kernel vaddr space (region 7)

where each region simply represents 256MB of address space.

The kernel vaddr space of 256MB is used to implement vmalloc, modules
So far this was enough, but not on EZChip system with 4K CPUs (given
that per cpu mechanism uses vmalloc for allocating chunks)

So allow VMALLOC_SIZE to be configurable by expanding down into the unused
kernel-user gutter region which at default 256M was excessive anyways.

Also use _BITUL() to fix a build error since PGDIR_SIZE cannot use "1UL"
as called from assembly code in mm/tlbex.S
Signed-off-by: NNoam Camus <noamc@ezchip.com>
[vgupta: rewrote changelog, debugged bootup crash due to int vs. hex]
Acked-by: NVineet Gupta <vgupta@synopsys.com>

UAPI header should not use Kconfig items

Use __BIG_ENDIAN__ defined as a compiler intrinsic
Signed-off-by: NNoam Camus <noamc@ezchip.com>
[vgupta: fix changelog]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

There are no more users of this - so RIP!
Signed-off-by: NAlexey Brodkin <abrodkin@synopsys.com>
[vgupta: update changelog]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

This will be needed for switching to linear irq domain as
irq_create_mapping() called by intr code needs the IRQ numbers
in addition to existing usage in mcip.c for requesting the irq
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

 - timer frequency is derived from DT (no longer rely on top level
   DT "clock-frequency" probed early and exported by asm/clk.h)

 - TIMER0_IRQ need not be exported across arch code, confined to intc as
   it is property of same

 - Any failures in clockevent setup are considered pedantic and system
   panic()'s as there is no generic fallback (unlike clocksource where
   a jiffies based soft clocksource always exists)
Acked-by: NDaniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

ARC Timers so far have been handled as "legacy" w/o explicit description
in DT. This poses challenge for newer platforms wanting to use them.
This series will eventually help move timers over to DT.

This patch does a small change of using a CPU notifier to set clockevent
on non-boot CPUs. So explicit setup is done only on boot CPU (which will
later be done by DT)
Signed-off-by: NNoam Camus <noamc@ezchip.com>
[vgupta: broken off from a bigger patch]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

- The idea is to remove the API usage since it has a subltle
  design flaw - relies on being called on cpu0 first. This is true for
  some early per cpu irqs such as TIMER/IPI, but not for late probed
  per cpu peripherals such a perf. And it's usage in perf has already
  bitten us once: see c6317bc7
  ("ARCv2: perf: Ensure perf intr gets enabled on all cores") where we
  ended up open coding it anyways

- The seeming duplication will go away once we start using cpu notifier
  for timer setup
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

Initial HIGHMEM support on ARC was introduced for PAE40 where the low
memory (0x8000_0000 based) and high memory (0x1_0000_0000) were
physically contiguous. So CONFIG_FLATMEM sufficed (despite a peipheral
hole in the middle, which wasted a bit of struct page memory, but things
worked).

However w/o PAE, highmem was not possible and we could only reach
~1.75GB of DDR. Now there is a use case to access ~4GB of DDR w/o PAE40
The idea is to have low memory at canonical 0x8000_0000 and highmem
at 0 so enire 4GB address space is available for physical addressing
This needs additional platform/interconnect mapping to convert
the non contiguous physical addresses into linear bus adresses.

From Linux point of view, non contiguous divide means FLATMEM no
longer works and DISCONTIGMEM is needed to track the pfns in the 2
regions.

This scheme would also work for PAE40, only better in that we don't
waste struct page memory for the peripheral hole.

The DT description will be something like

    memory {
        ...
        reg = <0x80000000 0x200000000   /* 512MB: lowmem */
               0x00000000 0x10000000>;  /* 256MB: highmem */
   }
Signed-off-by: NNoam Camus <noamc@ezchip.com>
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

So a benign looking cleanup which macro'ized PAGE_SHIFT shifts turned
out to be bad (since it was done non-sensically across the board).

It caused boot failures with PAE40 as forced cast to (unsigned long)
from newly introduced virt_to_pfn() was causing truncatiion of the
(long long) pte/paddr values.

It is OK to use this in accessors dealing with kernel virtual address,
pointers etc, but not for PTE values themelves.

Fixes: cJ2ff5cf2735c ("ARC: mm: Use virt_to_pfn() for addr >> PAGE_SHIFT pattern)
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

While reviewing a different change to asm-generic/io.h Arnd spotted that
ARC ioread32 and ioread32be both of which come from asm-generic versions
are not symmetrical in terms of calling the io barriers.

generic ioread32   -> ARC readl()                  [ has barriers]
generic ioread32be -> __be32_to_cpu(__raw_readl()) [ lacks barriers]

While generic ioread32be is being remediated to call readl(), that involves
a swab32(), causing double swaps on ioread32be() on Big Endian systems.

So provide our versions of big endian IO accessors to ensure io barrier
calls while also keeping them optimal
Suggested-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NArnd Bergmann <arnd@arndb.de>
Cc: stable@vger.kernel.org  [4.2+]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

- The asm helpers for calling into irq tracer were missing

- Add calls to above helpers in low level assembly entry code for ARCv2

- irq_save() uses CLRI to disable interrupts and returns the prev interrupt
  state (in STATUS32) in a specific encoding (and not the raw value of
  STATUS32). This is usable with SETI in irq_restore(). However
  save_flags() reads the raw value of STATUS32 which doesn't pair with
  irq_save/restore() and thus needs fixing.
Signed-off-by: NEvgeny Voevodin <evgeny.voevodin@intel.com>
[vgupta: updated changelog and also added some comments]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

During mmaping of frame-buffer pages to user-space
fb_protect() is called to set proper page settings.

In case of ARC we need to mark pages that are mmaped to
user as uncached because of 2 reasons:
 * Huge amount of data if passing through data cache will
   thrash cache a lot making cache almost useless for other
   less traffic hungry processes.
 * Data written by user in FB will be immediately available for
   hardware (such as PGU etc) without requirements to flush data
   cache regularly.
Signed-off-by: NAlexey Brodkin <abrodkin@synopsys.com>
Cc: linux-snps-arc@lists.infradead.org
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

The peripheral address space is architectural address window which is
uncached and typically used to wire up peripherals.

For ARC700 cores (ARCompact ISA based) this was fixed to 1GB region
0xC000_0000 - 0xFFFF_FFFF.

For ARCv2 based HS38 cores the start address is flexible and can be
0xC, 0xD, 0xE, 0xF 000_000 by programming AUX_NON_VOLATILE_LIMIT reg
(typically done in bootloader)

Further in cas of PAE, the physical address can extend beyond 4GB so
need to confine this check, otherwise all pages beyond 4GB will be
treated as uncached
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

To support dma in physical memory beyond 4GB with PAE40
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

With THP refcounting work, no need to mark PMDs splitting.

(ARC got missed under the sweeping arch change as THP support was likely
not present in orig baseline)
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

linux-next for 4.6-rc1 timeline reported ARC build failures !THP

| arch/arc/include/asm/tlbflush.h:29:0: warning: "flush_pmd_tlb_range" redefined [enabled by default]
| arch/arc/include/asm/tlbflush.h:29:0: warning: "flush_pmd_tlb_range" redefined [enabled by default]
| arch/arc/include/asm/tlbflush.h:29:0: warning: "flush_pmd_tlb_range" redefined [enabled by default]

Turns out that commit ("mm/thp/migration: switch from flush_tlb_range
to flush_pmd_tlb_range") triggered the issue while the problem was in
ARC code where THP specific helpers were not guarded with #ifdef.
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

This patch updates all instances of csum_tcpudp_magic and
csum_tcpudp_nofold to reflect the types that are usually used as the source
inputs.  For example the protocol field is populated based on nexthdr which
is actually an unsigned 8 bit value.  The length is usually populated based
on skb->len which is an unsigned integer.

This addresses an issue in which the IPv6 function csum_ipv6_magic was
generating a checksum using the full 32b of skb->len while
csum_tcpudp_magic was only using the lower 16 bits.  As a result we could
run into issues when attempting to adjust the checksum as there was no
protocol agnostic way to update it.

With this change the value is still truncated as many architectures use
"(len + proto) << 8", however this truncation only occurs for values
greater than 16776960 in length and as such is unlikely to occur as we stop
the inner headers at ~64K in size.

I did have to make a few minor changes in the arm, mn10300, nios2, and
score versions of the function in order to support these changes as they
were either using things such as an OR to combine the protocol and length,
or were using ntohs to convert the length which would have truncated the
value.

I also updated a few spots in terms of whitespace and type differences for
the addresses.  Most of this was just to make sure all of the definitions
were in sync going forward.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

ARC architecture has 2 instruction sets: ARCompact/ARCv2.
While same gcc supports compiling for either (using appropriate toggles),
we can't use the same toolchain to build kernel because libgcc needs
to be unique and the toolchian (uClibc based) is not multilibed.

uClibc toolchain is convenient since it allows all userspace and
kernel to be built with a single install for an ISA.

This however means 2 gnu installs (with same triplet prefix) are needed
for building for 2 ISA and need to be in PATH.
As developers we keep switching the builds, but would occassionally fail
to update the PATH leading to usage of wrong tools. And this would only
show up at the end of kernel build when linking incompatible libgcc.

So the initial solution was to have gcc define a special preprocessor macro
DEFAULT_CPU_xxx which is unique for default toolchain configuration.
Claudiu proposed using grep for an existing preprocessor macro which is
again uniquely defined per ISA.

Cc: Michal Marek <mmarek@suse.cz>
Suggested-by: NClaudiu Zissulescu <claziss@synopsys.com>
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

read{l,w}() write{l,w}() primitives should use le{16,32}_to_cpu() and
cpu_to_le{16,32}() respectively to ensure device registers are read
correctly in Big Endian CPU configuration.

Per Arnd Bergmann
| Most drivers using readl() or readl_relaxed() expect those to perform byte
| swaps on big-endian architectures, as the registers tend to be fixed endian

This was needed for getting UART to work correctly on a Big Endian ARC.

The ARC accessors originally were fine, and the bug got introduced
inadventently by commit b8a03302 ("ARCv2: barriers")

Fixes: b8a03302 ("ARCv2: barriers")
Link: http://lkml.kernel.org/r/201603100845.30602.arnd@arndb.de
Cc: Alexey Brodkin <abrodkin@synopsys.com>
Cc: stable@vger.kernel.org  [4.2+]
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: NLada Trimasova <ltrimas@synopsys.com>
[vgupta: beefed up changelog, added Fixes/stable tags]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

commit 80f42084 removed the ARC bitops microoptimization but failed
to prune the comments to same effect

Fixes: 80f42084 ("ARC: Make ARC bitops "safer" (add anti-optimization)")
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

Signed-off-by: NAdam Buchbinder <adam.buchbinder@gmail.com>
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

Add PCI support to ARC and update drivers/pci Makefile enabling the ARC
arch to use the generic PCI setup functions.

[bhelgaas: fold in Joao's pci-dma-compat.h & pci-bridge.h build fix (I
should have caught this myself, sorry]
Signed-off-by: NJoao Pinto <jpinto@synopsys.com>
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Acked-by: NVineet Gupta <vgupta@synopsys.com>

They are not symmetric with each other, neither are used in real world
(can not be found by grep command in source code root directory), so
remove them.
Signed-off-by: NChen Gang <gang.chen.5i5j@gmail.com>
Acked-by: NRussell King <rmk+kernel@arm.linux.org.uk>
Acked-by: NGreg Ungerer <gerg@uclinux.org>
Signed-off-by: NArnd Bergmann <arnd@arndb.de>

Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

There is no real ARC700 based SMP SoC so remove IPI definition.
EZChip's SMP ARC700 is going to use a different intc and IPI provider
anyways.
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

ARConnect/MCIP Inter-Core-Interrupt module can't send interrupt to
local core. So use core intc capability to trigger software
interrupt to self, using an unsued IRQ #21.

This showed up as csd deadlock with LTP trace_sched on a dual core
system. This test acts as scheduler fuzzer, triggering all sorts of
schedulting activity. Trouble starts with IPI to self, which doesn't get
delivered (effectively lost due to H/w capability), but the msg intended
to be sent remain enqueued in per-cpu @ipi_data.

All subsequent IPIs to this core from other cores get elided due to the
IPI coalescing optimization in ipi_send_msg_one() where a pending msg
implies an IPI already sent and assumes other core is yet to ack it.
After the elided IPI, other core simply goes into csd_lock_wait()
but never comes out as this core never sees the interrupt.

Fixes STAR 9001008624

Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@vger.kernel.org>        [4.2]
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

- ARCv2 uses a seperate BCR for {I,D}CCM base address:
  ARCompact encoded both base/size in same BCR

- Size encoding in common BCR is different for ARCompact/ARCv2
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

MMUv4 supports 2 concurrent page sizes: Normal and Super [4K to 16M]

So far Linux supported a single super page size for a given Normal page,
depending on the software page walking address split.
e.g. we had 11:8:13 address split for 8K page, which meant super page
was 2 ^(8+13) = 2M (given that THP size has to be PMD_SHIFT)

Now we turn this around, by allowing multiple Super Pages in Kconfig
(currently 2M and 16M only) and forcing page walker address split to
PGDIR_SHIFT and PAGE_SHIFT

For configs without Super page, things are same as before and
PGDIR_SHIFT can be hacked to get non default address split

The motivation for this change is a customer who needs 16M super page
and a 8K Normal page combo.
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

ARC HS Cores support configurable multiple interrupt priorities of upto
16 levels.

There is processor "interrupt preemption threshhold" in STATUS32.E[4:1]
And several places need to set this up:
1. seed value as kernel is booting
2. seed value for user space programs
3. Arg to SLEEP instruction in idle task (what interrupt prio can wake)
4. Per-IRQ line prioirty (i.e. what is the priority of interrupt
   raised by a peripheral or timer or perf counter...

Currently above sites use the highest priority 0. This can be potential
problem when multiple priorities are supported. e.g. user space could
only be interrupted by P0 interrupt, not others...
So turn this over and instead make default interruption level to be
the lowest priority possible 15. This should be fine even if there are
fewer priority levels configured (say two: P0 HIGH, P1 LOW)

This feature also effectively disables FIRQ feature if present in
hardware config. With old code, a P0 interrupt would be FIRQ, needing
special handling (ISR or Register Banks) which is NOT supported yet.
Now it not be P0 (P15 or whatever is lowest prio) so FIRQ is not
triggered.
Signed-off-by: NVineet Gupta <vgupta@synopsys.com>

Signed-off-by: NVineet Gupta <vgupta@synopsys.com>