create_branch(), create_cond_branch() and translate_branch() return the
instruction that they create, or return 0 to signal an error. Separate
these concerns in preparation for an instruction type that is not just
an unsigned int. Fill the created instruction to a pointer passed as
the first parameter to the function and use a non-zero return value to
signify an error.
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Link: https://lore.kernel.org/r/20200506034050.24806-6-jniethe5@gmail.com
A modulo operation is used for calculating the current offset from a
breakpoint within the breakpoint table. As instruction lengths are
always a power of 2, this can be replaced with a bitwise 'and'. The
current check for word alignment can be replaced with checking that the
lower 2 bits are not set.
Suggested-by: Christophe Leroy <christophe.leroy@c-s.fr>
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Link: https://lore.kernel.org/r/20200506034050.24806-5-jniethe5@gmail.com
The instructions for xmon's breakpoint are stored bpt_table[] which is in
the data section. This is problematic as the data section may be marked
as no execute. Move bpt_table[] to the text section.
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200506034050.24806-4-jniethe5@gmail.com
To execute an instruction out of line after a breakpoint, the NIP is set
to the address of struct bpt::instr. Here a copy of the instruction that
was replaced with a breakpoint is kept, along with a trap so normal flow
can be resumed after XOLing. The struct bpt's are located within the
data section. This is problematic as the data section may be marked as
no execute.
Instead of each struct bpt holding the instructions to be XOL'd, make a
new array, bpt_table[], with enough space to hold instructions for the
number of supported breakpoints. A later patch will move this to the
text section.
Make struct bpt::instr a pointer to the instructions in bpt_table[]
associated with that breakpoint. This association is a simple mapping:
bpts[n] -> bpt_table[n * words per breakpoint]. Currently we only need
the copied instruction followed by a trap, so 2 words per breakpoint.
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Alistair Popple <alistair@popple.id.au>
Link: https://lore.kernel.org/r/20200506034050.24806-3-jniethe5@gmail.com
For modifying instructions in xmon, patch_instruction() can serve the
same role that store_inst() is performing with the advantage of not
being specific to xmon. In some places patch_instruction() is already
being using followed by store_inst(). In these cases just remove the
store_inst(). Otherwise replace store_inst() with patch_instruction().
Signed-off-by: Jordan Niethe <jniethe5@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Nicholas Piggin <npiggin@gmail.com>
Link: https://lore.kernel.org/r/20200506034050.24806-2-jniethe5@gmail.com
The ps3_mm_region_destroy() and ps3_mm_vas_destroy() routines
are called very late in the shutdown via kexec's mmu_cleanup_all
routine. By the time mmu_cleanup_all runs it is too late to use
udbg_printf, and calling it will cause PS3 systems to hang.
Remove all debugging statements from ps3_mm_region_destroy() and
ps3_mm_vas_destroy() and replace any error reporting with calls
to lv1_panic.
With this change builds with 'DEBUG' defined will not cause kexec
reboots to hang, and builds with 'DEBUG' defined or not will end
in lv1_panic if an error is encountered.
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/7325c4af2b4c989c19d6a26b90b1fec9c0615ddf.1589049250.git.geoff@infradead.org
System Reset and Machine Check interrupts that are not recoverable due
to being nested or interrupting when RI=0 currently panic. This is not
necessary, and can often just kill the current context and recover.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@c-s.fr>
Link: https://lore.kernel.org/r/20200508043408.886394-16-npiggin@gmail.com
Similarly to the previous patch, do not trace system reset. This code
is used when there is a crash or hang, and tracing disturbs the system
more and has been known to crash in the crash handling path.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@c-s.fr>
Acked-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Link: https://lore.kernel.org/r/20200508043408.886394-15-npiggin@gmail.com
Rather than notrace annotations throughout a significant part of the
machine check code across kernel/ pseries/ and powernv/ which can
easily be broken and is infrequently tested, use paca->ftrace_enabled
to blanket-disable tracing of the real-mode non-maskable handler.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@c-s.fr>
Acked-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Link: https://lore.kernel.org/r/20200508043408.886394-14-npiggin@gmail.com
machine_check_early() is taken as an NMI, so nmi_enter() is used
there. machine_check_exception() is no longer taken as an NMI (it's
invoked via irq_work in the case a machine check hits in kernel mode),
so remove the nmi_enter() from that case.
In NMI context, hash faults don't try to refill the hash table, which
can lead to crashes accessing non-pinned kernel pages. System reset
still has this potential problem.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Drop change in show_regs() which breaks Book3E]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-12-npiggin@gmail.com
With the previous patch, machine checks can use rtas_call_unlocked()
which avoids the RTAS spinlock which would deadlock if a machine
check hits while making an RTAS call.
This also avoids the complex RTAS error logging which has more RTAS
calls and includes kmalloc (which can return memory beyond RMA, which
would also crash).
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-11-npiggin@gmail.com
This allows rtas_args to be put on the machine check stack, which
avoids a lot of complications with re-entrancy deadlocks.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@c-s.fr>
Reviewed-by: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Link: https://lore.kernel.org/r/20200508043408.886394-10-npiggin@gmail.com
PAPR does not specify that fwnmi sreset should be interlocked, and
PowerVM (and therefore now QEMU) do not require it.
These "ibm,nmi-interlock" calls are ignored by firmware, but there
is a possibility that the sreset could have interrupted a machine
check and release the machine check's interlock too early, corrupting
it if another machine check came in.
This is an extremely rare case, but it should be fixed for clarity
and reducing the code executed in the sreset path. Firmware also
does not provide error information for the sreset case to look at, so
remove that comment.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Use __be64 to silence some sparse warnings]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-9-npiggin@gmail.com
If there is some error with the fwnmi save area, r3 has already been
modified which doesn't help with debugging.
Only update r3 when to restore the saved value.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-8-npiggin@gmail.com
This was discovered developing qemu fwnmi sreset support. This
off-by-one bug means the last 16 bytes of the rtas area can not
be used for a 16 byte save area.
It's not a serious bug, and QEMU implementation has to retain a
workaround for old kernels, but it's good to tighten it.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Acked-by: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Link: https://lore.kernel.org/r/20200508043408.886394-7-npiggin@gmail.com
In the interest of reducing code and possible failures in the
machine check and system reset paths, grab the "ibm,nmi-interlock"
token at init time.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@c-s.fr>
Reviewed-by: Mahesh Salgaonkar <mahesh@linux.ibm.com>
Link: https://lore.kernel.org/r/20200508043408.886394-6-npiggin@gmail.com
A spare interrupt stack slot is needed to save irq state when
reconciling NMIs (sreset and decrementer soft-nmi). _DAR is used
for this, but we want to reconcile machine checks as well, which
do use _DAR. Switch to using RESULT instead, as it's used by
system calls.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-4-npiggin@gmail.com
The architecture allows for machine check exceptions to cause idle
wakeups which resume at the 0x200 address which has to return via
the idle wakeup code, but the early machine check handler is run
first.
The case of a no state-loss sleep is broken because the early
handler uses non-volatile register r1 , which is needed for the wakeup
protocol, but it is not restored.
Fix this by loading r1 from the MCE exception frame before returning
to the idle wakeup code. Also update the comment which has become
stale since the idle rewrite in C.
This crash was found and fix confirmed with a machine check injection
test in qemu powernv model (which is not upstream in qemu yet).
Fixes: 10d91611f4 ("powerpc/64s: Reimplement book3s idle code in C")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200508043408.886394-2-npiggin@gmail.com
EEH device state is currently removed (by eeh_remove_device()) during
the device release handler, which is invoked as the device's reference
count drops to zero. This may take some time, or forever, as other
threads may hold references.
However, the PCI device state is released synchronously by
pci_stop_and_remove_bus_device(). This mismatch causes problems, for
example the device may be re-discovered as a new device before the
release handler has been called, leaving the PCI and EEH state
mismatched.
So instead, call eeh_remove_device() from the bus device removal
handlers, which are called synchronously in the removal path.
Signed-off-by: Sam Bobroff <sbobroff@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/0a1f5105d3a33b1c090bba31de63eb0cdd25de7b.1588045502.git.sbobroff@linux.ibm.com
If a device is hot unplgged during EEH recovery, it's possible for the
RTAS call to ibm,configure-pe in pseries_eeh_configure() to return
parameter error (-3), however negative return values are not checked
for and this leads to an infinite loop.
Fix this by correctly bailing out on negative values.
Signed-off-by: Sam Bobroff <sbobroff@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Nathan Lynch <nathanl@linux.ibm.com>
Link: https://lore.kernel.org/r/1b0a6010a647dc915816e44845b64d72066676a7.1588045502.git.sbobroff@linux.ibm.com
Currently we don't report anything useful in /proc/<pid>/status:
$ grep Speculation_Store_Bypass /proc/self/status
Speculation_Store_Bypass: unknown
Our mitigation is currently always a barrier instruction, which
doesn't map that well onto the existing possibilities for the PR_SPEC
values.
However even if we added a "barrier" type PR_SPEC value, userspace
would still need to consult some other source to work out which type
of barrier to use. So reporting "vulnerable" seems sufficient, as
userspace can see that and then consult its source to determine what
barrier to use.
Signed-off-by: Gustavo Walbon <gwalbon@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200402124929.3574166-1-mpe@ellerman.id.au
There's no need to cast in task_pt_regs() as tsk->thread.regs should
already be a struct pt_regs. If someone's using task_pt_regs() on
something that's not a task but happens to have a thread.regs then
we'll deal with them later.
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200428123152.73566-1-mpe@ellerman.id.au
Aneesh increased the size of struct pt_regs by 16 bytes and started
seeing this WARN_ON:
smp: Bringing up secondary CPUs ...
------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at arch/powerpc/kernel/process.c:455 giveup_all+0xb4/0x110
Modules linked in:
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 5.7.0-rc2-gcc-8.2.0-1.g8f6a41f-default+ #318
NIP: c00000000001a2b4 LR: c00000000001a29c CTR: c0000000031d0000
REGS: c0000000026d3980 TRAP: 0700 Not tainted (5.7.0-rc2-gcc-8.2.0-1.g8f6a41f-default+)
MSR: 800000000282b033 <SF,VEC,VSX,EE,FP,ME,IR,DR,RI,LE> CR: 48048224 XER: 00000000
CFAR: c000000000019cc8 IRQMASK: 1
GPR00: c00000000001a264 c0000000026d3c20 c0000000026d7200 800000000280b033
GPR04: 0000000000000001 0000000000000000 0000000000000077 30206d7372203164
GPR08: 0000000000002000 0000000002002000 800000000280b033 3230303030303030
GPR12: 0000000000008800 c0000000031d0000 0000000000800050 0000000002000066
GPR16: 000000000309a1a0 000000000309a4b0 000000000309a2d8 000000000309a890
GPR20: 00000000030d0098 c00000000264da40 00000000fd620000 c0000000ff798080
GPR24: c00000000264edf0 c0000001007469f0 00000000fd620000 c0000000020e5e90
GPR28: c00000000264edf0 c00000000264d200 000000001db60000 c00000000264d200
NIP [c00000000001a2b4] giveup_all+0xb4/0x110
LR [c00000000001a29c] giveup_all+0x9c/0x110
Call Trace:
[c0000000026d3c20] [c00000000001a264] giveup_all+0x64/0x110 (unreliable)
[c0000000026d3c90] [c00000000001ae34] __switch_to+0x104/0x480
[c0000000026d3cf0] [c000000000e0b8a0] __schedule+0x320/0x970
[c0000000026d3dd0] [c000000000e0c518] schedule_idle+0x38/0x70
[c0000000026d3df0] [c00000000019c7c8] do_idle+0x248/0x3f0
[c0000000026d3e70] [c00000000019cbb8] cpu_startup_entry+0x38/0x40
[c0000000026d3ea0] [c000000000011bb0] rest_init+0xe0/0xf8
[c0000000026d3ed0] [c000000002004820] start_kernel+0x990/0x9e0
[c0000000026d3f90] [c00000000000c49c] start_here_common+0x1c/0x400
Which was unexpected. The warning is checking the thread.regs->msr
value of the task we are switching from:
usermsr = tsk->thread.regs->msr;
...
WARN_ON((usermsr & MSR_VSX) && !((usermsr & MSR_FP) && (usermsr & MSR_VEC)));
ie. if MSR_VSX is set then both of MSR_FP and MSR_VEC are also set.
Dumping tsk->thread.regs->msr we see that it's: 0x1db60000
Which is not a normal looking MSR, in fact the only valid bit is
MSR_VSX, all the other bits are reserved in the current definition of
the MSR.
We can see from the oops that it was swapper/0 that we were switching
from when we hit the warning, ie. init_task. So its thread.regs points
to the base (high addresses) in init_stack.
Dumping the content of init_task->thread.regs, with the members of
pt_regs annotated (the 16 bytes larger version), we see:
0000000000000000 c000000002780080 gpr[0] gpr[1]
0000000000000000 c000000002666008 gpr[2] gpr[3]
c0000000026d3ed0 0000000000000078 gpr[4] gpr[5]
c000000000011b68 c000000002780080 gpr[6] gpr[7]
0000000000000000 0000000000000000 gpr[8] gpr[9]
c0000000026d3f90 0000800000002200 gpr[10] gpr[11]
c000000002004820 c0000000026d7200 gpr[12] gpr[13]
000000001db60000 c0000000010aabe8 gpr[14] gpr[15]
c0000000010aabe8 c0000000010aabe8 gpr[16] gpr[17]
c00000000294d598 0000000000000000 gpr[18] gpr[19]
0000000000000000 0000000000001ff8 gpr[20] gpr[21]
0000000000000000 c00000000206d608 gpr[22] gpr[23]
c00000000278e0cc 0000000000000000 gpr[24] gpr[25]
000000002fff0000 c000000000000000 gpr[26] gpr[27]
0000000002000000 0000000000000028 gpr[28] gpr[29]
000000001db60000 0000000004750000 gpr[30] gpr[31]
0000000002000000 000000001db60000 nip msr
0000000000000000 0000000000000000 orig_r3 ctr
c00000000000c49c 0000000000000000 link xer
0000000000000000 0000000000000000 ccr softe
0000000000000000 0000000000000000 trap dar
0000000000000000 0000000000000000 dsisr result
0000000000000000 0000000000000000 ppr kuap
0000000000000000 0000000000000000 pad[2] pad[3]
This looks suspiciously like stack frames, not a pt_regs. If we look
closely we can see return addresses from the stack trace above,
c000000002004820 (start_kernel) and c00000000000c49c (start_here_common).
init_task->thread.regs is setup at build time in processor.h:
#define INIT_THREAD { \
.ksp = INIT_SP, \
.regs = (struct pt_regs *)INIT_SP - 1, /* XXX bogus, I think */ \
The early boot code where we setup the initial stack is:
LOAD_REG_ADDR(r3,init_thread_union)
/* set up a stack pointer */
LOAD_REG_IMMEDIATE(r1,THREAD_SIZE)
add r1,r3,r1
li r0,0
stdu r0,-STACK_FRAME_OVERHEAD(r1)
Which creates a stack frame of size 112 bytes (STACK_FRAME_OVERHEAD).
Which is far too small to contain a pt_regs.
So the result is init_task->thread.regs is pointing at some stack
frames on the init stack, not at a pt_regs.
We have gotten away with this for so long because with pt_regs at its
current size the MSR happens to point into the first frame, at a
location that is not written to by the early asm. With the 16 byte
expansion the MSR falls into the second frame, which is used by the
compiler, and collides with a saved register that tends to be
non-zero.
As far as I can see this has been wrong since the original merge of
64-bit ppc support, back in 2002.
Conceptually swapper should have no regs, it never entered from
userspace, and in fact that's what we do on 32-bit. It's also
presumably what the "bogus" comment is referring to.
So I think the right fix is to just not-initialise regs at all. I'm
slightly worried this will break some code that isn't prepared for a
NULL regs, but we'll have to see.
Remove the comment in head_64.S which refers to us setting up the
regs (even though we never did), and is otherwise not really accurate
any more.
Reported-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200428123130.73078-1-mpe@ellerman.id.au
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
sizeof(flexible-array-member) triggers a warning because flexible array
members have incomplete type[1]. There are some instances of code in
which the sizeof operator is being incorrectly/erroneously applied to
zero-length arrays and the result is zero. Such instances may be hiding
some bugs. So, this work (flexible-array member conversions) will also
help to get completely rid of those sorts of issues.
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200507185755.GA15014@embeddedor
The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:
struct foo {
int stuff;
struct boo array[];
};
By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.
Also, notice that, dynamic memory allocations won't be affected by
this change:
"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]
sizeof(flexible-array-member) triggers a warning because flexible array
members have incomplete type[1]. There are some instances of code in
which the sizeof operator is being incorrectly/erroneously applied to
zero-length arrays and the result is zero. Such instances may be hiding
some bugs. So, this work (flexible-array member conversions) will also
help to get completely rid of those sorts of issues.
This issue was found with the help of Coccinelle.
[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 7649773293 ("cxgb3/l2t: Fix undefined behaviour")
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200507185749.GA14994@embeddedor
It's not very nice to zero trap for this, because then system calls no
longer have trap_is_syscall(regs) invariant, and we can't distinguish
between sc and scv system calls (in a later patch).
Take one last unused bit from the low bits of the pt_regs.trap word
for this instead. There is not a really good reason why it should be
in trap as opposed to another field, but trap has some concept of
flags and it exists. Ideally I think we would move trap to 2-byte
field and have 2 more bytes available independently.
Add a selftests case for this, which can be seen to fail if
trap_norestart() is changed to return false.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Make them static inlines]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200507121332.2233629-4-mpe@ellerman.id.au
A new system call interrupt will be added with a new trap number.
Hide the explicit 0xc00 test behind an accessor to reduce churn
in callers.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Make it a static inline]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200507121332.2233629-3-mpe@ellerman.id.au
The pt_regs.trap field keeps 4 low bits for some metadata about the
trap or how it was handled, which is masked off in order to test the
architectural trap number.
Add a set_trap() accessor to set this, equivalent to TRAP() for
returning it. This is actually not quite the equivalent of TRAP()
because it always clears the low bits, which may be harmless if
it can only be updated via ptrace syscall, but it seems dangerous.
In fact settting TRAP from ptrace doesn't seem like a great idea
so maybe it's better deleted.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Make it a static inline rather than a shouty macro]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200507121332.2233629-2-mpe@ellerman.id.au
My 'pengutronix' address is defunct for years. Merge the entries and use
the proper contact address.
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200502142642.18979-1-wsa@kernel.org
Replace relaswap with built-in one, because relaswap
does a simple byte to byte swap.
Since Spectre mitigations have made indirect function calls more
expensive, and the default simple byte copies swap is implemented
without them, an "optimized" custom swap function is now
a waste of time as well as code.
Signed-off-by: Andrey Abramov <st5pub@yandex.ru>
Reviewed-by: George Spelvin <lkml@sdf.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/994931554238042@iva8-b333b7f98ab0.qloud-c.yandex.net
Fix a cut'n'paste error in a warning message. This should be
'cpu-idle-state-residency-ns' to match the property searched in the
previous 'of_property_read_u32_array()'
Fixes: 9c7b185ab2 ("powernv/cpuidle: Parse dt idle properties into global structure")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200502115949.139000-1-christophe.jaillet@wanadoo.fr
This merges the lockless page table walk rework series from Aneesh.
Because it touches powerpc KVM code we are sharing it with the kvm-ppc
tree in our topic/ppc-kvm branch.
This is the cover letter from Aneesh:
Avoid IPI while updating page table entries.
Problem Summary:
Slow termination of KVM guest with large guest RAM config due to a
large number of IPIs that were caused by clearing level 1 PTE
entries (THP) entries. This is shown in the stack trace below.
- qemu-system-ppc [kernel.vmlinux] [k] smp_call_function_many
- smp_call_function_many
- 36.09% smp_call_function_many
serialize_against_pte_lookup
radix__pmdp_huge_get_and_clear
zap_huge_pmd
unmap_page_range
unmap_vmas
unmap_region
__do_munmap
__vm_munmap
sys_munmap
system_call
__munmap
qemu_ram_munmap
qemu_anon_ram_free
reclaim_ramblock
call_rcu_thread
qemu_thread_start
start_thread
__clone
Why we need to do IPI when clearing PMD entries:
This was added as part of commit: 13bd817bb8 ("powerpc/thp: Serialize pmd clear against a linux page table walk")
serialize_against_pte_lookup makes sure that all parallel lockless
page table walk completes before we convert a PMD pte entry to regular
pmd entry. We end up doing that conversion in the below scenarios
1) __split_huge_zero_page_pmd
2) do_huge_pmd_wp_page_fallback
3) MADV_DONTNEED running parallel to page faults.
local_irq_disable and lockless page table walk:
The lockless page table walk work with the assumption that we can
dereference the page table contents without holding a lock. For this
to work, we need to make sure we read the page table contents
atomically and page table pages are not going to be freed/released
while we are walking the table pages. We can achieve by using a rcu
based freeing for page table pages or if the architecture implements
broadcast tlbie, we can block the IPI as we walk the page table pages.
To support both the above framework, lockless page table walk is done
with irq disabled instead of rcu_read_lock()
We do have two interface for lockless page table walk, gup fast and
__find_linux_pte. This patch series makes __find_linux_pte table walk
safe against the conversion of PMD PTE to regular PMD.
gup fast:
gup fast is already safe against THP split because kernel now
differentiate between a pmd split and a compound page split. gup fast
can run parallel to a pmd split and we prevent a parallel gup fast to
a hugepage split, by freezing the page refcount and failing the
speculative page ref increment.
Similar to how gup is safe against parallel pmd split, this patch
series updates the __find_linux_pte callers to be safe against a
parallel pmd split. We do that by enforcing the following rules.
1) Don't reload the pte value, because that can be updated in
parallel.
2) Code should be able to work with a stale PTE value and not the
recent one. ie, the pte value that we are looking at may not be the
latest value in the page table.
3) Before looking at pte value check for _PAGE_PTE bit. We now do this
as part of pte_present() check.
Performance:
This speeds up Qemu guest RAM del/unplug time as below
128 core, 496GB guest:
Without patch:
munmap start: timer = 13162 ms, PID=7684
munmap finish: timer = 95312 ms, PID=7684 - delta = 82150 ms
With patch (upto removing IPI)
munmap start: timer = 196449 ms, PID=6681
munmap finish: timer = 196488 ms, PID=6681 - delta = 39ms
With patch (with adding the tlb invalidate in pmdp_huge_get_and_clear_full)
munmap start: timer = 196345 ms, PID=6879
munmap finish: timer = 196714 ms, PID=6879 - delta = 369ms
Link: https://lore.kernel.org/r/20200505071729.54912-1-aneesh.kumar@linux.ibm.com
MADV_DONTNEED holds mmap_sem in read mode and that implies a
parallel page fault is possible and the kernel can end up with a level 1 PTE
entry (THP entry) converted to a level 0 PTE entry without flushing
the THP TLB entry.
Most architectures including POWER have issues with kernel instantiating a level
0 PTE entry while holding level 1 TLB entries.
The code sequence I am looking at is
down_read(mmap_sem) down_read(mmap_sem)
zap_pmd_range()
zap_huge_pmd()
pmd lock held
pmd_cleared
table details added to mmu_gather
pmd_unlock()
insert a level 0 PTE entry()
tlb_finish_mmu().
Fix this by forcing a tlb flush before releasing pmd lock if this is
not a fullmm invalidate. We can safely skip this invalidate for
task exit case (fullmm invalidate) because in that case we are sure
there can be no parallel fault handlers.
This do change the Qemu guest RAM del/unplug time as below
128 core, 496GB guest:
Without patch:
munmap start: timer = 196449 ms, PID=6681
munmap finish: timer = 196488 ms, PID=6681 - delta = 39ms
With patch:
munmap start: timer = 196345 ms, PID=6879
munmap finish: timer = 196714 ms, PID=6879 - delta = 369ms
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200505071729.54912-23-aneesh.kumar@linux.ibm.com
Now that all the lockless page table walk is careful w.r.t the PTE
address returned, we can now revert
commit: 13bd817bb8 ("powerpc/thp: Serialize pmd clear against a linux page table walk.")
We also drop the equivalent IPI from other pte updates routines. We still keep
IPI in hash pmdp collapse and that is to take care of parallel hash page table
insert. The radix pmdp collapse flush can possibly be removed once I am sure
generic code doesn't have the any expectations around parallel gup walk.
This speeds up Qemu guest RAM del/unplug time as below
128 core, 496GB guest:
Without patch:
munmap start: timer = 13162 ms, PID=7684
munmap finish: timer = 95312 ms, PID=7684 - delta = 82150 ms
With patch:
munmap start: timer = 196449 ms, PID=6681
munmap finish: timer = 196488 ms, PID=6681 - delta = 39ms
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200505071729.54912-21-aneesh.kumar@linux.ibm.com
This adds _PAGE_PTE check and makes sure we validate the pte value returned via
find_kvm_host_pte.
NOTE: this also considers _PAGE_INVALID to the software valid bit.
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200505071729.54912-20-aneesh.kumar@linux.ibm.com
Jordan Niethe