linux.git/arch/mips/kernel, branch v4.11

MIPS: KGDB: Use kernel context for sleeping threads

2017-04-12T20:29:22+00:00

KGDB is a kernel debug stub and it can't be used to debug userland as it
can only safely access kernel memory.

On MIPS however KGDB has always got the register state of sleeping
processes from the userland register context at the beginning of the
kernel stack. This is meaningless for kernel threads (which never enter
userland), and for user threads it prevents the user seeing what it is
doing while in the kernel:

(gdb) info threads
  Id   Target Id         Frame
  ...
  3    Thread 2 (kthreadd) 0x0000000000000000 in ?? ()
  2    Thread 1 (init)   0x000000007705c4b4 in ?? ()
  1    Thread -2 (shadowCPU0) 0xffffffff8012524c in arch_kgdb_breakpoint () at arch/mips/kernel/kgdb.c:201

Get the register state instead from the (partial) kernel register
context stored in the task's thread_struct for resume() to restore. All
threads now correctly appear to be in context_switch():

(gdb) info threads
  Id   Target Id         Frame
  ...
  3    Thread 2 (kthreadd) context_switch (rq=, cookie=..., next=, prev=0x0) at kernel/sched/core.c:2903
  2    Thread 1 (init)   context_switch (rq=, cookie=..., next=, prev=0x0) at kernel/sched/core.c:2903
  1    Thread -2 (shadowCPU0) 0xffffffff8012524c in arch_kgdb_breakpoint () at arch/mips/kernel/kgdb.c:201

Call clobbered registers which aren't saved and exception registers
(BadVAddr & Cause) which can't be easily determined without stack
unwinding are reported as 0. The PC is taken from the return address,
such that the state presented matches that found immediately after
returning from resume().

Fixes: 8854700115ec ("[MIPS] kgdb: add arch support for the kernel's kgdb core")
Signed-off-by: James Hogan 
Cc: Jason Wessel 
Cc: linux-mips@linux-mips.org
Cc: stable@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/15829/
Signed-off-by: Ralf Baechle

MIPS: smp-cps: Fix potentially uninitialised value of core

2017-04-12T19:34:30+00:00

Turning on DEBUG in smp-cps.c, or compiling the kernel with
CONFIG_DYNAMIC_DEBUG enabled results the build error:

arch/mips/kernel/smp-cps.c: In function 'play_dead':
./include/linux/dynamic_debug.h:126:3: error: 'core' may be used
uninitialized in this function [-Werror=maybe-uninitialized]

Fix this by always initialising the variable.

Fixes: 0d2808f338c7 ("MIPS: smp-cps: Add support for CPU hotplug of MIPSr6 processors")
Signed-off-by: Matt Redfearn 
Cc: James Hogan 
Cc: Masahiro Yamada 
Cc: Paul Burton 
Cc: Ingo Molnar 
Cc: Andrew Morton 
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/15848/
Signed-off-by: Ralf Baechle

MIPS: KASLR: Add missing header files

2017-04-12T15:02:04+00:00

After the split of linux/sched.h, KASLR stopped building.

Fix this by including the correct header file for init_thread_union

Signed-off-by: Matt Redfearn 
Cc: James Hogan 
Cc: Marcin Nowakowski 
Cc: Steven J. Hill 
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/15849/
Signed-off-by: Ralf Baechle

MIPS: Avoid BUG warning in arch_check_elf

2017-04-12T13:22:13+00:00

arch_check_elf contains a usage of current_cpu_data that will call
smp_processor_id() with preemption enabled and therefore triggers a
"BUG: using smp_processor_id() in preemptible" warning when an fpxx
executable is loaded.

As a follow-up to commit b244614a60ab ("MIPS: Avoid a BUG warning during
prctl(PR_SET_FP_MODE, ...)"), apply the same fix to arch_check_elf by
using raw_current_cpu_data instead. The rationale quoted from the previous
commit:

"It is assumed throughout the kernel that if any CPU has an FPU, then
all CPUs would have an FPU as well, so it is safe to perform the check
with preemption enabled - change the code to use raw_ variant of the
check to avoid the warning."

Fixes: 46490b572544 ("MIPS: kernel: elf: Improve the overall ABI and FPU mode checks")
Signed-off-by: James Cowgill 
CC:  # 4.0+
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/15951/
Signed-off-by: Ralf Baechle

MIPS: cevt-r4k: Fix out-of-bounds array access

2017-04-10T11:31:12+00:00

calculate_min_delta() may incorrectly access a 4th element of buf2[]
which only has 3 elements. This may trigger undefined behaviour and has
been reported to cause strange crashes in start_kernel() sometime after
timer initialization when built with GCC 5.3, possibly due to
register/stack corruption:

sched_clock: 32 bits at 200MHz, resolution 5ns, wraps every 10737418237ns
CPU 0 Unable to handle kernel paging request at virtual address ffffb0aa, epc == 8067daa8, ra == 8067da84
Oops[#1]:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.9.18 #51
task: 8065e3e0 task.stack: 80644000
$ 0   : 00000000 00000001 00000000 00000000
$ 4   : 8065b4d0 00000000 805d0000 00000010
$ 8   : 00000010 80321400 fffff000 812de408
$12   : 00000000 00000000 00000000 ffffffff
$16   : 00000002 ffffffff 80660000 806a666c
$20   : 806c0000 00000000 00000000 00000000
$24   : 00000000 00000010
$28   : 80644000 80645ed0 00000000 8067da84
Hi    : 00000000
Lo    : 00000000
epc   : 8067daa8 start_kernel+0x33c/0x500
ra    : 8067da84 start_kernel+0x318/0x500
Status: 11000402 KERNEL EXL
Cause : 4080040c (ExcCode 03)
BadVA : ffffb0aa
PrId  : 0501992c (MIPS 1004Kc)
Modules linked in:
Process swapper/0 (pid: 0, threadinfo=80644000, task=8065e3e0, tls=00000000)
Call Trace:
[<8067daa8>] start_kernel+0x33c/0x500
Code: 24050240  0c0131f9  24849c64  41606020  000000c0  0c1a45e6 00000000  0c1a5f44

UBSAN also detects the same issue:

================================================================
UBSAN: Undefined behaviour in arch/mips/kernel/cevt-r4k.c:85:41
load of address 80647e4c with insufficient space
for an object of type 'unsigned int'
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 4.9.18 #47
Call Trace:
[<80028f70>] show_stack+0x88/0xa4
[<80312654>] dump_stack+0x84/0xc0
[<8034163c>] ubsan_epilogue+0x14/0x50
[<803417d8>] __ubsan_handle_type_mismatch+0x160/0x168
[<8002dab0>] r4k_clockevent_init+0x544/0x764
[<80684d34>] time_init+0x18/0x90
[<8067fa5c>] start_kernel+0x2f0/0x500
=================================================================

buf2[] is intentionally only 3 elements so that the last element is the
median once 5 samples have been inserted, so explicitly prevent the
possibility of comparing against the 4th element rather than extending
the array.

Fixes: 1fa405552e33f2 ("MIPS: cevt-r4k: Dynamically calculate min_delta_ns")
Reported-by: Rabin Vincent 
Signed-off-by: James Hogan 
Tested-by: Rabin Vincent 
Cc: linux-mips@linux-mips.org
Cc:  # 4.7.x-
Patchwork: https://patchwork.linux-mips.org/patch/15892/
Signed-off-by: Ralf Baechle

MIPS: perf: fix deadlock

2017-04-10T10:25:07+00:00

mipsxx_pmu_handle_shared_irq() calls irq_work_run() while holding the
pmuint_rwlock for read.  irq_work_run() can, via perf_pending_event(),
call try_to_wake_up() which can try to take rq->lock.

However, perf can also call perf_pmu_enable() (and thus take the
pmuint_rwlock for write) while holding the rq->lock, from
finish_task_switch() via perf_event_context_sched_in().

This leads to an ABBA deadlock:

 PID: 3855   TASK: 8f7ce288  CPU: 2   COMMAND: "process"
  #0 [89c39ac8] __delay at 803b5be4
  #1 [89c39ac8] do_raw_spin_lock at 8008fdcc
  #2 [89c39af8] try_to_wake_up at 8006e47c
  #3 [89c39b38] pollwake at 8018eab0
  #4 [89c39b68] __wake_up_common at 800879f4
  #5 [89c39b98] __wake_up at 800880e4
  #6 [89c39bc8] perf_event_wakeup at 8012109c
  #7 [89c39be8] perf_pending_event at 80121184
  #8 [89c39c08] irq_work_run_list at 801151f0
  #9 [89c39c38] irq_work_run at 80115274
 #10 [89c39c50] mipsxx_pmu_handle_shared_irq at 8002cc7c

 PID: 1481   TASK: 8eaac6a8  CPU: 3   COMMAND: "process"
  #0 [8de7f900] do_raw_write_lock at 800900e0
  #1 [8de7f918] perf_event_context_sched_in at 80122310
  #2 [8de7f938] __perf_event_task_sched_in at 80122608
  #3 [8de7f958] finish_task_switch at 8006b8a4
  #4 [8de7f998] __schedule at 805e4dc4
  #5 [8de7f9f8] schedule at 805e5558
  #6 [8de7fa10] schedule_hrtimeout_range_clock at 805e9984
  #7 [8de7fa70] poll_schedule_timeout at 8018e8f8
  #8 [8de7fa88] do_select at 8018f338
  #9 [8de7fd88] core_sys_select at 8018f5cc
 #10 [8de7fee0] sys_select at 8018f854
 #11 [8de7ff28] syscall_common at 80028fc8

The lock seems to be there to protect the hardware counters so there is
no need to hold it across irq_work_run().

Signed-off-by: Rabin Vincent 
Signed-off-by: Ralf Baechle

Merge branch 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus

2017-04-06T20:16:34+00:00

Pull MIPS fixes from Ralf Baechle:
 "Lantiq:
    - Fix adding xbar resoures causing a panic

  Loongson3:
    - Some Loongson 3A don't identify themselves as having an FTLB so
      hardwire that knowledge into CPU probing.
    - Handle Loongson 3 TLB peculiarities in the fast path of the RDHWR
      emulation.
    - Fix invalid FTLB entries with huge page on VTLB+FTLB platforms
    - Add missing calculation of S-cache and V-cache cache-way size

  Ralink:
    - Fix typos in rt3883 pinctrl data

  Generic:
    - Force o32 fp64 support on 32bit MIPS64r6 kernels
    - Yet another build fix after the linux/sched.h changes
    - Wire up statx system call
    - Fix stack unwinding after introduction of IRQ stack
    - Fix spinlock code to build even for microMIPS with recent binutils

  SMP-CPS:
    - Fix retrieval of VPE mask on big endian CPUs"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
  MIPS: IRQ Stack: Unwind IRQ stack onto task stack
  MIPS: c-r4k: Fix Loongson-3's vcache/scache waysize calculation
  MIPS: Flush wrong invalid FTLB entry for huge page
  MIPS: Check TLB before handle_ri_rdhwr() for Loongson-3
  MIPS: Add MIPS_CPU_FTLB for Loongson-3A R2
  MIPS: Lantiq: fix missing xbar kernel panic
  MIPS: smp-cps: Fix retrieval of VPE mask on big endian CPUs
  MIPS: Wire up statx system call
  MIPS: Include asm/ptrace.h now linux/sched.h doesn't
  MIPS: ralink: Fix typos in rt3883 pinctrl
  MIPS: End spinlocks with .insn
  MIPS: Force o32 fp64 support on 32bit MIPS64r6 kernels

Merge branch 'regset' (PTRACE_SETREGSET data leakage)

2017-03-29T15:55:25+00:00

Merge PTRACE_SETREGSET leakage fixes from Dave Martin:
 "This series is the collection of fixes I proposed on this topic, that
  have not yet appeared upstream or in the stable branches,

  The issue can leak kernel stack, but doesn't appear to allow userspace
  to attack the kernel directly.  The affected architectures are c6x,
  h8300, metag, mips and sparc.

  [ Mark Salter points out that c6x has no MMU or other mechanism to
    prevent userspace access to kernel code or data on c6x, but it
    doesn't hurt to clean that case up too. ]

  The bugs arise from use of user_regset_copyin(). Users of
  user_regset_copyin() can work in one of two ways:

   1) Copy directly to thread_struct or equivalent. (This seems to be
      the design assumption of the regset API, and is the most common
      approach.)

   2) Copy to a local variable and then transfer to thread_struct. (A
      significant minority of cases.)

  Buggy code typically involves approach 2"

* emailed patches from Dave Martin :
  sparc/ptrace: Preserve previous registers for short regset write
  mips/ptrace: Preserve previous registers for short regset write
  metag/ptrace: Reject partial NT_METAG_RPIPE writes
  metag/ptrace: Provide default TXSTATUS for short NT_PRSTATUS
  metag/ptrace: Preserve previous registers for short regset write
  h8300/ptrace: Fix incorrect register transfer count
  c6x/ptrace: Remove useless PTRACE_SETREGSET implementation

mips/ptrace: Preserve previous registers for short regset write

2017-03-29T15:54:17+00:00

Ensure that if userspace supplies insufficient data to PTRACE_SETREGSET
to fill all the registers, the thread's old registers are preserved.

Signed-off-by: Dave Martin 
Signed-off-by: Linus Torvalds

MIPS: IRQ Stack: Unwind IRQ stack onto task stack

2017-03-22T10:53:57+00:00

When the separate IRQ stack was introduced, stack unwinding only
proceeded as far as the top of the IRQ stack, leading to kernel
backtraces being less useful, lacking the trace of what was interrupted.

Fix this by providing a means for the kernel to unwind the IRQ stack
onto the interrupted task stack. The processor state is saved to the
kernel task stack on interrupt. The IRQ_STACK_START macro reserves an
unsigned long at the top of the IRQ stack where the interrupted task
stack pointer can be saved. After the active stack is switched to the
IRQ stack, save the interrupted tasks stack pointer to the reserved
location.

Fix the stack unwinding code to look for the frame being the top of the
IRQ stack and if so get the next frame from the saved location. The
existing test does not work with the separate stack since the ra is no
longer pointed at ret_from_{irq,exception}.

The test to stop unwinding the stack 32 bytes from the top of a stack
must be modified to allow unwinding to continue up to the location of
the saved task stack pointer when on the IRQ stack. The low / high marks
of the stack are set depending on whether the sp is on an irq stack or
not.

Signed-off-by: Matt Redfearn 
Cc: Paolo Bonzini 
Cc: Marcin Nowakowski 
Cc: Masanari Iida 
Cc: Chris Metcalf 
Cc: James Hogan 
Cc: Paul Burton 
Cc: Ingo Molnar 
Cc: Jason A. Donenfeld 
Cc: Andrew Morton 
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/15788/
Signed-off-by: Ralf Baechle