While computing slice mask for the free area we need make sure we only
search in the addr limit applicable for this mmap. We update the
slb_addr_limit after we request for a mmap above 128TB. But the
following mmap request with hint addr below 128TB should still limit
its search to below 128TB. ie. we should not use slb_addr_limit to
compute slice mask in this case. Instead, we should derive high addr
limit based on the mmap hint addr value.
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Recently we added a CPU feature for Power9 DD2.0, to capture the fact
that some workarounds are required only on Power9 DD1 and DD2.0 but
not DD2.1 or later.
Then in commit 9d2f510a66 ("powerpc/64s/idle: avoid POWER9 DD1 and
DD2.0 ERAT workaround on DD2.1") and commit e3646330cf
"powerpc/64s/idle: avoid POWER9 DD1 and DD2.0 PMU workaround on
DD2.1") we changed CPU_FTR_SECTIONs to check for DD1 or DD20, eg:
BEGIN_FTR_SECTION
PPC_INVALIDATE_ERAT
END_FTR_SECTION_IFSET(CPU_FTR_POWER9_DD1 | CPU_FTR_POWER9_DD20)
Unfortunately although this reads as "if set DD1 or DD2.0", the or is
a bitwise or and actually generates a mask of both bits. The code that
does the feature patching then checks that the value of the CPU
features masked with that mask are equal to the mask.
So the end result is we're checking for DD1 and DD20 being set, which
never happens. Yes the API is terrible.
Removing the ERAT workaround on DD2.0 results in random SEGVs, the
system tends to boot, but things randomly die including sometimes
dhclient, udev etc.
To fix the problem and hopefully avoid it in future, we remove the
DD2.0 CPU feature and instead add a DD2.1 (or later) feature. This
allows us to easily express that the workarounds are required if DD2.1
is not set.
At some point we will drop the DD1 workarounds entirely and some of
this can be cleaned up.
Fixes: 9d2f510a66 ("powerpc/64s/idle: avoid POWER9 DD1 and DD2.0 ERAT workaround on DD2.1")
Fixes: e3646330cf ("powerpc/64s/idle: avoid POWER9 DD1 and DD2.0 PMU workaround on DD2.1")
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On 64-bit Book3s, when we take an instruction fault the reason for the
fault may be reported in SRR1. For data faults the reason is reported
in DSISR (Data Storage Instruction Status Register).
The reasons reported in each do not necessarily correspond, so we mask
the SRR1 bits before copying them to the DSISR, which is then used by
the page fault code.
Prior to commit b4c001dc44 ("powerpc/mm: Use symbolic constants for
filtering SRR1 bits on ISIs") we used a hard-coded mask of 0x58200000,
which corresponds to:
DSISR_NOHPTE 0x40000000 /* no translation found */
DSISR_NOEXEC_OR_G 0x10000000 /* exec of no-exec or guarded */
DSISR_PROTFAULT 0x08000000 /* protection fault */
DSISR_KEYFAULT 0x00200000 /* Storage Key fault */
That commit added a #define for the mask, DSISR_SRR1_MATCH_64S, but
incorrectly used a different similarly named DSISR_BAD_FAULT_64S.
This had the effect of changing the mask to 0xa43a0000, which omits
everything but DSISR_KEYFAULT.
Luckily this had no visible effect, because in practice we hardly use
the DSISR bits. The lack of DSISR_NOHPTE means a TLB flush
optimisation was missed in the native HPTE code, and DSISR_NOEXEC_OR_G
and DSISR_PROTFAULT are both only used to trigger rare warnings.
So we got lucky, but let's fix it. The new value only has bits between
17 and 30 set, so we can continue to use andis.
Fixes: b4c001dc44 ("powerpc/mm: Use symbolic constants for filtering SRR1 bits on ISIs")
Cc: stable@vger.kernel.org # v4.14+
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Radix keeps no meaningful state in addr_limit, so remove it from radix
code and rename to slb_addr_limit to make it clear it applies to hash
only.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Radix VA space allocations test addresses against mm->task_size which
is 512TB, even in cases where the intention is to limit allocation to
below 128TB.
This results in mmap with a hint address below 128TB but address +
length above 128TB succeeding when it should fail (as hash does after
the previous patch).
Set the high address limit to be considered up front, and base
subsequent allocation checks on that consistently.
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
While mapping hints with a length that cross 128TB are disallowed,
MAP_FIXED allocations that cross 128TB are allowed. These are failing
on hash (on radix they succeed). Add an additional case for fixed
mappings to expand the addr_limit when crossing 128TB.
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Hash unconditionally resets the addr_limit to default (128TB) when the
mm context is initialised. If a process has > 128TB mappings when it
forks, the child will not get the 512TB addr_limit, so accesses to
valid > 128TB mappings will fail in the child.
Fix this by only resetting the addr_limit to default if it was 0. Non
zero indicates it was duplicated from the parent (0 means exec()).
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When allocating VA space with a hint that crosses 128TB, the SLB
addr_limit variable is not expanded if addr is not > 128TB, but the
slice allocation looks at task_size, which is 512TB. This results in
slice_check_fit() incorrectly succeeding because the slice_count
truncates off bit 128 of the requested mask, so the comparison to the
available mask succeeds.
Fix this by using mm->context.addr_limit instead of mm->task_size for
testing allocation limits. This causes such allocations to fail.
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Reported-by: Florian Weimer <fweimer@redhat.com>
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently userspace is able to request mmap() search between 128T-512T
by specifying a hint address that is greater than 128T. But that means
a hint of 128T exactly will return an address below 128T, which is
confusing and wrong.
So fix the logic to check the hint is greater than *or equal* to 128T.
Fixes: f4ea6dcb08 ("powerpc/mm: Enable mappings above 128TB")
Cc: stable@vger.kernel.org # v4.12+
Suggested-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Suggested-by: Nicholas Piggin <npiggin@gmail.com>
[mpe: Split out of Nick's bigger patch]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Commit 398a719d34 ("powerpc/mm: Update bits used to skip hash_page")
mistakenly dropped the DSISR_DABRMATCH bit from the mask of bit tested
to skip trying to hash a page.
As a result, the DABR matches would no longer be detected.
This adds it back. We open code it in the 2 places where it matters
rather than fold it into DSISR_BAD_FAULT_32S/64S because this isn't
technically a bad fault and while we would never hit it with the
current code, I prefer if page_fault_is_bad() didn't trigger on these.
Fixes: 398a719d34 ("powerpc/mm: Update bits used to skip hash_page")
Cc: stable@vger.kernel.org # v4.14
Tested-by: Pedro Miraglia Franco de Carvalho <pedromfc@br.ibm.com>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
When a uprobe is installed on an instruction that we currently do not
emulate, we copy the instruction into a xol buffer and single step
that instruction. If that instruction generates a fault, we abort the
single stepping before invoking the signal handler. Once the signal
handler is done, the uprobe trap is hit again since the instruction is
retried and the process repeats.
We use uprobe_deny_signal() to detect if the xol instruction triggered
a signal. If so, we clear TIF_SIGPENDING and set TIF_UPROBE so that the
signal is not handled until after the single stepping is aborted. In
this case, uprobe_deny_signal() returns true and get_signal() ends up
returning 0. However, in do_signal(), we are not looking at the return
value, but depending on ksig.sig for further action, all with an
uninitialized ksig that is not touched in this scenario. Fix the same
by initializing ksig.sig to 0.
Fixes: 129b69df9c ("powerpc: Use get_signal() signal_setup_done()")
Cc: stable@vger.kernel.org # v3.17+
Reported-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently sysfs store handlers in fadump use if buf[0] == 'char'.
This means input "100foo" is interpreted as '1' and "01" as '0'.
Change to kstrtoint so leading zeroes and the like is handled in
expected way.
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Acked-by: Hari Bathini <hbathini@linux.vnet.ibm.com>
Signed-off-by: Michal Suchanek <a class="moz-txt-link-rfc2396E" href="mailto:msuchanek@suse.de"><msuchanek@suse.de></a></pre>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Implement the architecture specific portitions of the UACCESS_FLUSHCACHE
API. This provides functions for the copy_user_flushcache iterator that
ensure that when the copy is finished the destination buffer contains
a copy of the original and that the destination buffer is clean in the
processor caches.
Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Implement the architecture specific cache maintence functions that make
up the "PMEM API". Currently the writeback and invalidate functions
are the same since the function of the DCBST (data cache block store)
instruction is typically interpreted as "writeback to the point of
coherency" rather than to memory. As a result implementing the API
requires a full cache flush rather than just a cache write back. This
will probably change in the not-too-distant future.
Signed-off-by: Oliver O'Halloran <oohall@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The nest mmu required an explicit flush as a tlbi would not flush it in the
same way as the core. However an alternate firmware fix exists which should
eliminate the need for this flush, so instead add a device-tree property
(ibm,nmmu-flush) on the NVLink2 PHB to enable it only if required.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Reviewed-by: Frederic Barrat <fbarrat@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
With the optimisations introduced by commit a46cc7a908 ("powerpc/mm/radix:
Improve TLB/PWC flushes"), flush_tlb_mm() no longer flushes the page walk
cache with radix. Switch to using flush_all_mm() to ensure the pwc and tlb
are properly flushed on the nmmu.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
On PowerNV platforms, firmware provides exit latency and
target residency for each of the idle states in nano
seconds. Cpuidle framework expects the values in micro
seconds. Round up to nearest micro seconds to avoid errors
in cases where the values are defined as fractional micro
seconds.
Default idle state of 'snooze' has exit latency of zero. If
other states have fractional micro second exit latency, they
would get rounded down to zero micro second and make cpuidle
framework choose deeper idle state when snooze loop is the
right choice.
Reported-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Vaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Use safer string manipulation functions when dealing with a
user-provided string in kprobe_lookup_name().
Reported-by: David Laight <David.Laight@ACULAB.COM>
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Commit 3cdfcbfd32 ("powerpc: Change analyse_instr so it doesn't
modify *regs") introduced emulate_update_regs() to perform part of what
emulate_step() was doing earlier. However, this function was not added
to the kprobes blacklist. Add it so as to prevent it from being probed.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Per Documentation/kprobes.txt, we don't necessarily need to disable
interrupts before invoking the kprobe handlers. Masami submitted
similar changes for x86 via commit a19b2e3d78 ("kprobes/x86: Remove
IRQ disabling from ftrace-based/optimized kprobes"). Do the same for
powerpc.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Per Documentation/kprobes.txt, probe handlers need to be invoked with
preemption disabled. Update optimized_callback() to do so. Also move
get_kprobe_ctlblk() invocation post preemption disable, since it
accesses pre-cpu data.
This was not an issue so far since optprobes wasn't selected if
CONFIG_PREEMPT was enabled. Commit a30b85df7d ("kprobes: Use
synchronize_rcu_tasks() for optprobe with CONFIG_PREEMPT=y") changes
this.
Signed-off-by: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Commit 78adf6c214 ("powerpc/64s: Implement system reset idle wakeup
reason"), added a call to ppc_save_regs() in the book3s code.
ppc_save_regs() is only built if XMON and/or KEXEC_CORE are enabled,
which is usually the case, however if they're not enabled then the
build breaks.
Fix it by making the Makefile check also build ppc_save_regs.o if
CONFIG_PPC_BOOK3S is enabled.
Fixes: 78adf6c214 ("powerpc/64s: Implement system reset idle wakeup reason")
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
[mpe: Write change log]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
When using the radix MMU on Power9 DD1, to work around a hardware
problem, radix__pte_update() is required to do a two stage update of
the PTE. First we write a zero value into the PTE, then we flush the
TLB, and then we write the new PTE value.
In the normal case that works OK, but it does not work if we're
updating the PTE that maps the code we're executing, because the
mapping is removed by the TLB flush and we can no longer execute from
it. Unfortunately the STRICT_RWX code needs to do exactly that.
The exact symptoms when we hit this case vary, sometimes we print an
oops and then get stuck after that, but I've also seen a machine just
get stuck continually page faulting with no oops printed. The variance
is presumably due to the exact layout of the text and the page size
used for the mappings. In all cases we are unable to boot to a shell.
There are possible solutions such as creating a second mapping of the
TLB flush code, executing from that, and then jumping back to the
original. However we don't want to add that level of complexity for a
DD1 work around.
So just detect that we're running on Power9 DD1 and refrain from
changing the permissions, effectively disabling STRICT_RWX on Power9
DD1.
Fixes: 7614ff3272 ("powerpc/mm/radix: Implement STRICT_RWX/mark_rodata_ro() for Radix")
Cc: stable@vger.kernel.org # v4.13+
Reported-by: Andrew Jeffery <andrew@aj.id.au>
[Changelog as suggested by Michael Ellerman <mpe@ellerman.id.au>]
Signed-off-by: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
We are using percpu send window on P9 NX (powerNV) instead of opening
/ closing per each crypto session. Means txwin is removed from
workmem. So we do not need to initialize workmem for each request.
Signed-off-by: Haren Myneni <haren@us.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
For P9 NX, the send window is opened for each crypto session and
closed upon free. But VAS supports 64K windows per chip for all
coprocessors including in user space support. So there is a
possibility of not getting the window for kernel requests.
This patch reserves windows for each coprocessor type (NX842) and are
available forever for kernel requests, Opens each window for each CPU
on the corresponding chip during driver initialization. So then use
the percpu txwin for NX requests depends on the CPU on which the
process is executing.
Signed-off-by: Haren Myneni <haren@us.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Add support for user space receive window (for the Fast thread-wakeup
coprocessor type)
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Define an interface to return a system-wide unique id for a given VAS
window.
The vas_win_id() will be used in a follow-on patch to generate an unique
handle for a user space receive window. Applications can use this handle
to pair send and receive windows for fast thread-wakeup.
The hardware refers to this system-wide unique id as a Partition Send
Window ID which is expected to be used during fault handling. Hence the
"pswid" in the function names.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Define an interface that the NX drivers can use to find the physical
paste address of a send window. This interface is expected to be used
with the mmap() operation of the NX driver's device. i.e the user space
process can use driver's mmap() operation to map the send window's paste
address into their address space and then use copy and paste instructions
to submit the CRBs to the NX engine.
Note that kernel drivers will use vas_paste_crb() directly and don't need
this interface.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
A CP_ABORT instruction is required in processes that have mapped a VAS
"paste address" with the intention of using COPY/PASTE instructions.
But since CP_ABORT is expensive, we want to restrict it to only
processes that use/intend to use COPY/PASTE.
Define an interface, set_thread_uses_vas(), that VAS can use to
indicate that the current process opened a send window. During context
switch, issue CP_ABORT only for processes that have the flag set.
Thanks for input from Nick Piggin, Michael Ellerman.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
[mpe: Fix to not use new_thread after _switch() returns]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
We need the SPRN_TIDR to be set for use with fast thread-wakeup (core-
to-core wakeup) and also with CAPI.
Each thread in a process needs to have a unique id within the process.
But for now, we assign globally unique thread ids to all threads in
the system.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Philippe Bergheaud <felix@linux.vnet.ibm.com>
Signed-off-by: Christophe Lombard <clombard@linux.vnet.ibm.com>
[mpe: Simplify tidr clearing on fork() and ctx switch code]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Export the VAS Window context information to debugfs.
We need to hold a mutex when closing the window to prevent a race
with the debugfs read(). Rather than introduce a per-instance mutex,
we use the global vas_mutex for now, since it is not heavily contended.
The window->cop field is only relevant to a receive window so we were
not setting it for a send window (which is is paired to a receive window
anyway). But to simplify reporting in debugfs, set the 'cop' field for the
send window also.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Define a helper, chip_to_vas_id() to map a given chip id to corresponding
vas id.
Normally, callers of vas_rx_win_open() and vas_tx_win_open() want the VAS
window to be on the same chip where the calling thread is executing. These
callers can pass in -1 for the VAS id.
This interface will be useful if a thread running on one chip wants to open
a window on another chip (like the NX-842 driver does during start up).
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Create a cpu to vasid mapping so callers can specify -1 instead of
trying to find a VAS id.
Changelog[v2]
[Michael Ellerman] Use per-cpu variables to simplify code.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Normally, the NX driver waits for the CRBs to be processed before closing
the window. But it is better to ensure that the credits are returned before
the window gets reassigned later.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Save the configured max window credits for a window in the vas_window
structure. We will need this when polling for return of window credits.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
A VAS window is normally in "busy" state for only a short duration.
Reduce the time we wait for the window to go to "not-busy" state to
speed-up vas_win_close() a bit.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Use a helper to have the hardware unpin and mark a window closed.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Polling for window cast out is listed in the spec, but turns out that
it is not strictly necessary and slows down window close. Making it a
stub for now.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Clean up vas.h and the debug code around ifdef vas_debug.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
NX-842, the only user of VAS, sets the window credits to default values
but VAS should check the credits against the possible max values.
The VAS_WCREDS_MIN is not needed and can be dropped.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Initialize a few missing window context fields from the window attributes
specified by the caller. These fields are currently set to their default
values by the caller (NX-842), but would be good to apply them anyway.
Signed-off-by: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Take the DSCR value set by firmware as the dscr_default value,
rather than zero.
POWER9 recommends DSCR default to a non-zero value.
Signed-off-by: From: Nicholas Piggin <npiggin@gmail.com>
[mpe: Make record_spr_defaults() __init]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
OPAL boot does not insert secondaries at 0x60 to wait at the secondary
hold spinloop. Instead they are started later, and inserted at
generic_secondary_smp_init(), which is after the secondary hold
spinloop.
Avoid waiting on this spinloop when booting with OPAL firmware. This
wait always times out that case.
This saves 100ms boot time on powernv, and 10s of seconds of real time
when booting on the simulator in SMP.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Unmaps that free page tables always flush the entire PID, which is
sub-optimal. Provide TLB range flushing with an additional PWC flush
that can be use for va range invalidations with PWC flush.
Time to munmap N pages of memory including last level page table
teardown (after mmap, touch), local invalidate:
N 1 2 4 8 16 32 64
vanilla 3.2us 3.3us 3.4us 3.6us 4.1us 5.2us 7.2us
patched 1.4us 1.5us 1.7us 1.9us 2.6us 3.7us 6.2us
Global invalidate:
N 1 2 4 8 16 32 64
vanilla 2.2us 2.3us 2.4us 2.6us 3.2us 4.1us 6.2us
patched 2.1us 2.5us 3.4us 5.2us 8.7us 15.7us 6.2us
Local invalidates get much better across the board. Global ones have
the same issue where multiple tlbies for va flush do get slower than
the single tlbie to invalidate the PID. None of this test captures
the TLB benefits of avoiding killing everything.
Global gets worse, but it is brought in to line with global invalidate
for munmap()s that do not free page tables.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
The single page flush ceiling is the cut-off point at which we switch
from invalidating individual pages, to invalidating the entire process
address space in response to a range flush.
Introduce a local variant of this heuristic because local and global
tlbie have significantly different properties:
- Local tlbiel requires 128 instructions to invalidate a PID, global
tlbie only 1 instruction.
- Global tlbie instructions are expensive broadcast operations.
The local ceiling has been made much higher, 2x the number of
instructions required to invalidate the entire PID (i.e., 256 pages).
Time to mprotect N pages of memory (after mmap, touch), local invalidate:
N 32 34 64 128 256 512
vanilla 7.4us 9.0us 14.6us 26.4us 50.2us 98.3us
patched 7.4us 7.8us 13.8us 26.4us 51.9us 98.3us
The behaviour of both is identical at N=32 and N=512. Between there,
the vanilla kernel does a PID invalidate and the patched kernel does
a va range invalidate.
At N=128, these require the same number of tlbiel instructions, so
the patched version can be sen to be cheaper when < 128, and more
expensive when > 128. However this does not well capture the cost
of invalidated TLB.
The additional cost at 256 pages does not seem prohibitive. It may
be the case that increasing the limit further would continue to be
beneficial to avoid invalidating all of the process's TLB entries.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Currently for radix, flush_tlb_range flushes the entire PID, because
the Linux mm code does not tell us about page size here for THP vs
regular pages. This is quite sub-optimal for small mremap / mprotect
/ change_protection.
So implement va range flushes with two flush passes, one for each
page size (regular and THP). The second flush has an order of matnitude
fewer tlbie instructions than the first, so it is a relatively small
additional cost.
There is still room for improvement here with some changes to generic
APIs, particularly if there are mostly THP pages to be invalidated,
the small page flushes could be reduced.
Time to mprotect 1 page of memory (after mmap, touch):
vanilla 2.9us 1.8us
patched 1.2us 1.6us
Time to mprotect 30 pages of memory (after mmap, touch):
vanilla 8.2us 7.2us
patched 6.9us 17.9us
Time to mprotect 34 pages of memory (after mmap, touch):
vanilla 9.1us 8.0us
patched 9.0us 8.0us
34 pages is the point at which the invalidation switches from va
to entire PID, which tlbie can do in a single instruction. This is
why in the case of 30 pages, the new code runs slower for this test.
This is a deliberate tradeoff already present in the unmap and THP
promotion code, the idea is that the benefit from avoiding flushing
entire TLB for this PID on all threads in the system.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Move the barriers and range iteration down into the _tlbie* level,
which improves readability.
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Short range flushes issue a sequences of tlbie(l) instructions for
individual effective addresses. These do not all require individual
barrier sequences, only one covering all tlbie(l) instructions.
Commit f7327e0ba3 ("powerpc/mm/radix: Remove unnecessary ptesync")
made a similar optimization for tlbiel for PID flushing.
For tlbie, the ISA says:
The tlbsync instruction provides an ordering function for the
effects of all tlbie instructions executed by the thread executing
the tlbsync instruction, with respect to the memory barrier
created by a subsequent ptesync instruction executed by the same
thread.
Time to munmap 30 pages of memory (after mmap, touch):
local global
vanilla 10.9us 22.3us
patched 3.4us 14.4us
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
We have some dependencies & conflicts between patches in fixes and
things to go in next, both in the radix TLB flush code and the IMC PMU
driver. So merge fixes into next.
Add a self test to check if FP/VEC/VSX registers are sane (restored
correctly) after a FP/VEC/VSX unavailable exception is caught during a
transaction.
This test checks all possibilities in a thread regarding the combination
of MSR.[FP|VEC] states in a thread and for each scenario raises a
FP/VEC/VSX unavailable exception in transactional state, verifying if
vs0 and vs32 registers, which are representatives of FP/VEC/VSX reg
sets, are not corrupted.
Signed-off-by: Gustavo Romero <gromero@linux.vnet.ibm.com>
Signed-off-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Cyril Bur <cyrilbur@gmail.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
