This patch extends sigcontext in order to hold the most significant 64
bits of each vector register in addition to the MSA control & status
register. The least significant 64 bits are already saved as the scalar
FP context. This makes things a little awkward since the least & most
significant 64 bits of each vector register are not contiguous in
memory. Thus the copy_u & insert instructions are used to transfer the
values of the most significant 64 bits via GP registers.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6533/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

No current systems implementing MSA include support for vector register
partitioning which makes it somewhat difficult to implement support for
it in the kernel. Thus for the moment the kernel includes no such
support. However if the kernel were to be run on a system which
implemented register partitioning then it would not function correctly,
mishandling MSA disabled exceptions. Print a warning if run on a system
with vector register partitioning implemented to indicate this problem
should it occur.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6494/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch adds a simple handler for MSA FP exceptions which delivers a
SIGFPE to the running task. In the future it should probably be extended
to re-execute the instruction with the MSACSR.NX bit set in order to
generate results for any elements which did not cause an exception
before delivering the SIGFPE signal.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6432/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch adds support for context switching the MSA vector registers.
These 128 bit vector registers are aliased with the FP registers - an
FP register accesses the least significant bits of the vector register
with which it is aliased (ie. the register with the same index). Due to
both this & the requirement that the scalar FPU must be 64-bit (FR=1) if
enabled at the same time as MSA the kernel will enable MSA & scalar FP
at the same time for tasks which use MSA. If we restore the MSA vector
context then we might as well enable the scalar FPU since the reason it
was left disabled was to allow for lazy FP context restoring - but we
just restored the FP context as it's a subset of the vector context. If
we restore the FP context and have previously used MSA then we have to
restore the whole vector context anyway (see comment in
enable_restore_fp_context for details) so similarly we might as well
enable MSA.

Thus if a task does not use MSA then it will continue to behave as
without this patch - the scalar FP context will be saved & restored as
usual. But if a task executes an MSA instruction then it will save &
restore the vector context forever more.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6431/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch adds support for probing the MSAP bit within the Config3
register in order to detect the presence of the MSA ASE. Presence of the
ASE will be indicated in /proc/cpuinfo. The value of the MSA
implementation register will be displayed at boot to aid debugging and
verification of a correct setup, as is done for the FPU.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6430/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

When saving or restoring scalar FP context we want to access the least
significant 64 bits of each FP register. When the FP registers are 64
bits wide that is trivially the start of the registers value in memory.
However when the FP registers are wider this equivalence will no longer
be true for big endian systems. Define a new set of offset macros for
the least significant 64 bits of each saved FP register within thread
context, and make use of them when saving and restoring scalar FP
context.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6428/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

When we want to access 64-bit FP register values we can only treat
consecutive registers as being consecutive in memory when the width of
an FP register equals 64 bits. This assumption will not remain true once
MSA support is introduced, so provide a code path which copies each 64
bit FP register value in turn when the width of an FP register differs
from 64 bits.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6427/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This code assumed that saved FP registers are 64 bits wide, an
assumption which will no longer be true once MSA is introduced. This
patch modifies the code to copy the lower 64 bits of each register in
turn, which is safe for any FP register width >= 64 bits.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6425/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

NUM_FPU_REGS just makes it clearer what's going on, rather than the
magic hard coded 32.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6424/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

When a task which has used the FPU at some point in its past takes a
signal the kernel would previously always require the task to take
ownership of the FPU whilst setting up or restoring from the sigcontext.
That means that if the task has not used the FPU within this timeslice
then the kernel would enable the FPU, restore the task's FP context into
FPU registers and then save them into the sigcontext. This seems
inefficient, and if the signal handler doesn't use FP then enabling the
FPU & the extra memory accesses are entirely wasted work.

This patch modifies the sigcontext setup & restore code to copy directly
between the tasks saved FP context & the sigcontext for any tasks which
have used FP in the past but are not currently the FPU owner (ie. have
not used FP in this timeslice).
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NQais Yousef <qais.yousef@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6423/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

These functions aren't directly related to the FPU emulator at all, they
simply copy between a thread's saved context & a sigcontext. Thus move
them to the appropriate signal files & rename them accordingly. This
makes it clearer that the functions don't require the FPU emulator in
any way.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NQais Yousef <qais.yousef@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6422/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch replaces the fpureg_t typedef with a "union fpureg" enabling
easier access to 32 & 64 bit values. This allows the access macros used
in cp1emu.c to be simplified somewhat. It will also make it easier to
expand the width of the FP registers as will be done in a future
patch in order to support the 128 bit registers introduced with MSA.

No behavioural change is intended by this patch.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NQais Yousef <qais.yousef@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6532/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

When userland uses syscall() to perform an indirect system call
the actually system call that needs to be checked by the filter
is on the first argument. The kernel code needs to handle this case
by looking at the original syscall number in v0 and if it's
NR_syscall, then it needs to examine the first argument to
identify the real system call that will be executed.
Similarly, we need to 'virtually' shift the syscall() arguments
so the syscall_get_arguments() function can fetch the correct
arguments for the indirect system call.
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6404/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Reviewed-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6399/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

MIPS now has the infrastructure for dynamic seccomp-bpf
filtering
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6400/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This effectively renames __syscall_get_arch to syscall_get_arch
and implements a compatible interface for the seccomp API.
The seccomp code (kernel/seccomp.c) expects a syscall_get_arch
function to be defined for every architecture, so we drop
the leading underscores from the existing function.

This also makes use of the 'task' argument to determine the type
the process instead of assuming the process has the same
characteristics as the kernel it's running on.
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Reviewed-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6398/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This header was used only by Malta but is used no longer. Remove it. It
was also included unnecessarily in irq-gic.c, so that include is also
removed.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6366/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch introduces a new SMP implementation for systems implementing
the MIPS Coherent Processing System architecture. The kernel will make
use of the Coherence Manager, Cluster Power Controller & Global
Interrupt Controller in order to detect, bring up & make use of other
cores in the system. SMTC is not supported, so only a single TC per VPE
in the system is used. That is, this option enables an SMVP style setup
but across multiple cores.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6362/
Patchwork: https://patchwork.linux-mips.org/patch/6611/
Patchwork: https://patchwork.linux-mips.org/patch/6651/
Patchwork: https://patchwork.linux-mips.org/patch/6652/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

This patch introduces code to probe for a MIPS Cluster Power Controller
& accessor functions to allow for easy register access. This support
code will be used by a subsequent patch.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6361/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

The kernel currently only probes for a MIPS Coherence Manager in the
Malta interrupt code in order to detect & enable the GIC. However CM is
not Malta-specific, so this should really be more generic. This patch
introduces some non-Malta-specific code which probes for a CM and
performs some basic initialisation.

A new header, with temporarily duplicated register definitions, is
introduced in order to:

  1) Allow the new definitions to be correct with regards to the
     CM documentation, as many of those in gcmpregs.h aren't.

  2) Allow switching away from the REG() macro used via a few layers of
     nested macros in order to access registers in gcmpregs.h. This
     patch instead introduced accessor functions akin to the
     {read,write}_c0_* functions used for cop0 registers.

  3) Allow users of the CM to be migrated one by one.

  4) Switch from the name 'GCMP' to 'CM' since the Coherence Manager is
     what this code is actually dealing with.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6360/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

The GIC IPI functions aren't necessarily specific to the "CMP
framework" SMP implementation, and will be used elsewhere in a
subsequent commit. This patch adds cleaned up GIC IPI functions to a
separate file which is compiled when a new CONFIG_MIPS_GIC_IPI Kconfig
symbol is selected, and selects that symbol for CONFIG_MIPS_CMP.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6359/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

The 1074K is a multiprocessing coherent processing system (CPS) based
on modified 74K cores. This patch makes the 1074K an actual unique
CPU type, instead of a 74K derivative, which it is not.
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6389/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

Wire up for MIPS the new sched_setattr and sched_getattr system calls
added in commit d50dde5a (sched: Add new scheduler syscalls to
support an extended scheduling parameters ABI) merged in v3.14-rc1.
Signed-off-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NMarkos Chandras <markos.chandras@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6502/Signed-off-by: NRalf Baechle <ralf@linux-mips.org>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.  Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6320/

Adds processor ID of XLP 9XX to asm/cpu.h.  Update netlogic/xlp-hal/xlp.h
to add cpu_is_xlp9xx() and to update cpu_is_xlpii() to support XLP 9XX.
Signed-off-by: NJayachandran C <jchandra@broadcom.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6274/

Currently the supported ISA is only printed on the latest architectures.
Print it also on legacy platforms.
Signed-off-by: NAaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6295/

There is already an init_dsp function which checks cpu_has_dsp & calls
__init_dsp if it does. Make use of it instead of duplicating the same
code.
Signed-off-by: NPaul Burton <paul.burton@imgtec.com>
Reviewed-by: NQais Yousef <qais.yousef@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6148/

The linux build-bot recently reported a build error in arch/mips/kernel/vpe.c

     tree:   git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-linus
     head:   261000a5
     commit: 261000a5 [4/4] userns:  userns: Remove UIDGID_STRICT_TYPE_CHECKS
     config: make ARCH=mips maltaaprp_defconfig

     All error/warnings:

        arch/mips/kernel/vpe.c: In function 'vpe_open':
     >> arch/mips/kernel/vpe.c:1086:9: error: incompatible types when assigning to type 'unsigned int' from type 'kuid_t'
     >> arch/mips/kernel/vpe.c:1087:9: error: incompatible types when assigning to type 'unsigned int' from type 'kgid_t'

     vim +1086 arch/mips/kernel/vpe.c

     863abad4 Jesper Juhl   2010-10-30  1080			return -ENOMEM;
     863abad4 Jesper Juhl   2010-10-30  1081  		}
     e01402b1 Ralf Baechle  2005-07-14  1082  		v->plen = P_SIZE;
     e01402b1 Ralf Baechle  2005-07-14  1083  		v->load_addr = NULL;
     e01402b1 Ralf Baechle  2005-07-14  1084  		v->len = 0;
     e01402b1 Ralf Baechle  2005-07-14  1085
     d76b0d9b David Howells 2008-11-14 @1086		v->uid = filp->f_cred->fsuid;
     d76b0d9b David Howells 2008-11-14 @1087  		v->gid = filp->f_cred->fsgid;
     2600990e Ralf Baechle  2006-04-05  1088
     2600990e Ralf Baechle  2006-04-05  1089		v->cwd[0] = 0;
     2600990e Ralf Baechle  2006-04-05  1090 	 	ret = getcwd(v->cwd, VPE_PATH_MAX);

When examining the code to see what v->uid and v->gid were used for I
discovered that the only users in the kernel are vpe_getuid and
vpe_getgid, and that vpe_getuid and vpe_getgid are never called.

So instead of proposing a conversion to use kuid_t and kgid_t instead
of unsigned int/int as I normally would let's just kill this dead code
so no one has to worry about it further.

Deng-Cheng Zhu said:

This is a good catch. vpe_get[u|g]id was originally used by KSPD which
has been removed.

Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Reviewed-by: NDeng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>

Clean-up code according to the 'checkpatch.pl' script.
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NQais Yousef <Qais.Yousef@imgtec.com>
Patchwork: http://patchwork.linux-mips.org/patch/6097/Reviewed-by: NJohn Crispin <blogic@openwrt.org>

This patch adds RTLX API support for platforms having a CMP.
Signed-off-by: NDeng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NQais Yousef <Qais.Yousef@imgtec.com>
Patchwork: http://patchwork.linux-mips.org/patch/6095/Reviewed-by: NJohn Crispin <blogic@openwrt.org>

Split the RTLX functionality in preparation for adding support for CMP
platforms. Common functions remain in the original file and a new file
contains code specific to platforms that do not have a CMP.
Signed-off-by: NDeng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NQais Yousef <Qais.Yousef@imgtec.com>
Patchwork: http://patchwork.linux-mips.org/patch/6093/Reviewed-by: NJohn Crispin <blogic@openwrt.org>

This patch adds VPE loader support for platforms having a CMP.
Signed-off-by: NDeng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NQais Yousef <Qais.Yousef@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6092/

Split the VPE functionality in preparation for adding support
for CMP platforms. Common functions remain in the original file
and a new file contains code specific to platforms that do not
have a CMP present.
Signed-off-by: NDeng-Cheng Zhu <dengcheng.zhu@imgtec.com>
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Reviewed-by: NQais Yousef <Qais.Yousef@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6094/

Signed-off-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6152/

The interAptiv is a power-efficient multi-core microprocessor
for use in system-on-chip (SoC) applications. The interAptiv combines
a multi-threading pipeline with a coherence manager to deliver improved
computational throughput and power efficiency. The interAptiv can
contain one to four MIPS32R3 interAptiv cores, system level
coherence manager with L2 cache, optional coherent I/O port,
and optional floating point unit.
Signed-off-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6163/

Add a new mips/segments debugfs file to print the 6 segmentation
control registers for supported cores. A sample from a proAptiv core
is given below:
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6137/
Segment   Virtual    Size   Access Mode   Physical   Caching   EU
-------   -------    ----   -----------   --------   -------   --
   0      e0000000   512M      MK           UND         U       0
   1      c0000000   512M      MSK          UND         U       0
   2      a0000000   512M      UK           000         2       0
   3      80000000   512M      UK           000         3       0
   4      40000000    1G       MUSK         UND         U       1
   5      00000000    1G       MUSK         UND         U       1
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>

The Fixed Page Size TLB (FTLB) is a set-associative dual entry TLB. Its
purpose is to reduce the number of TLB misses by increasing the effective
TLB size and keep the implementation complexity to minimum levels.
A supported core can have both VTLB and FTLB.
Reviewed-by: NJames Hogan <james.hogan@imgtec.com>
Reviewed-by: NPaul Burton <paul.burton@imgtec.com>
Signed-off-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6139/

Signed-off-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6135/

The proAptiv Multiprocessing System is a power efficient multi-core
microprocessor for use in system-on-chip (SoC) applications.
The proAptiv Multiprocessing System combines a deep pipeline
with multi-issue out of order execution for improved computational
throughput. The proAptiv Multiprocessing System can contain one to
six MIPS32r3 proAptiv cores, system level coherence
manager with L2 cache, optional coherent I/O port, and optional
floating point unit.
Signed-off-by: NLeonid Yegoshin <Leonid.Yegoshin@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6134/

MIPS32R3 introduced a new set of Segmentation Control registers which
increase the flexibility of the segmented-based memory scheme.
Signed-off-by: NSteven J. Hill <Steven.Hill@imgtec.com>
Signed-off-by: NMarkos Chandras <markos.chandras@imgtec.com>
Signed-off-by: NJohn Crispin <blogic@openwrt.org>
Patchwork: http://patchwork.linux-mips.org/patch/6131/