Many BAU functions have different implementations depending on the UV
version. Rather than switching on the uvhub_version throughout the driver,
we can define a set of operations for each version. This is especially
beneficial for UV4, which will require many new MMR read/write functions.
Currently, the set of abstracted functions are the same for UV1, UV2, and
UV3. The functions were chosen because each one will have a different
implementation for UV4. Other functions will be added as needed to handle
new implementations or to cleanup the existing differences between UV1,
UV2, and UV3, i.e. read_status and wait_completion.
Signed-off-by: Andrew Banman <abanman@sgi.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Mike Travis <travis@sgi.com>
Acked-by: Dimitri Sivanich <sivanich@sgi.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: akpm@linux-foundation.org
Cc: rja@sgi.com
Link: http://lkml.kernel.org/r/1474474161-265604-6-git-send-email-abanman@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
For several years, the common practice has been to boot UVs with the
"nobau" parameter on the command line, to disable the BAU. We've
decided that it makes more sense to just disable the BAU by default in
the kernel, and provide the option to turn it on, if desired.
For now, having the on/off switch doesn't buy us any more than just
reversing the logic would, but we're working towards having the BAU
enabled by default on UV4. When those changes are in place, having the
on/off switch will make more sense than an enable flag, since the
default behavior will be different depending on the system version.
I've also added a bit of documentation for the new parameter to
Documentation/kernel-parameters.txt.
Signed-off-by: Alex Thorlton <athorlton@sgi.com>
Reviewed-by: Hedi Berriche <hedi@sgi.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1459451909-121845-1-git-send-email-athorlton@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
UV: NMI: insert this_cpu_read accessor function on uv_hub_nmi.
On SGI UV systems a 'power nmi' command from the CMC causes
all processors to drop into uv_handle_nmi(). With the 4.0
kernel this results in
BUG: unable to handle kernel paging request
The bug is caused by the current code trying to use the PER_CPU
variable uv_cpu_nmi.hub without an appropriate accessor
function. That oversight occurred in
commit e16321709c ("uv: Replace __get_cpu_var")
Author: Christoph Lameter <cl@linux.com>
Date: Sun Aug 17 12:30:41 2014 -0500
This patch inserts this_cpu_read() in the uv_hub_nmi macro
restoring the intended functionality.
Signed-off-by: George Beshers <gbeshers@sgi.com>
Acked-by: Mike Travis <travis@sgi.com>
Cc: Alex Thorlton <athorlton@sgi.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Cc: Hedi Berriche <hedi@sgi.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russ Anderson <rja@sgi.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
__get_cpu_var() is used for multiple purposes in the kernel source. One of
them is address calculation via the form &__get_cpu_var(x). This calculates
the address for the instance of the percpu variable of the current processor
based on an offset.
Other use cases are for storing and retrieving data from the current
processors percpu area. __get_cpu_var() can be used as an lvalue when
writing data or on the right side of an assignment.
__get_cpu_var() is defined as :
#define __get_cpu_var(var) (*this_cpu_ptr(&(var)))
__get_cpu_var() always only does an address determination. However, store
and retrieve operations could use a segment prefix (or global register on
other platforms) to avoid the address calculation.
this_cpu_write() and this_cpu_read() can directly take an offset into a
percpu area and use optimized assembly code to read and write per cpu
variables.
This patch converts __get_cpu_var into either an explicit address
calculation using this_cpu_ptr() or into a use of this_cpu operations that
use the offset. Thereby address calculations are avoided and less registers
are used when code is generated.
Transformations done to __get_cpu_var()
1. Determine the address of the percpu instance of the current processor.
DEFINE_PER_CPU(int, y);
int *x = &__get_cpu_var(y);
Converts to
int *x = this_cpu_ptr(&y);
2. Same as #1 but this time an array structure is involved.
DEFINE_PER_CPU(int, y[20]);
int *x = __get_cpu_var(y);
Converts to
int *x = this_cpu_ptr(y);
3. Retrieve the content of the current processors instance of a per cpu
variable.
DEFINE_PER_CPU(int, y);
int x = __get_cpu_var(y)
Converts to
int x = __this_cpu_read(y);
4. Retrieve the content of a percpu struct
DEFINE_PER_CPU(struct mystruct, y);
struct mystruct x = __get_cpu_var(y);
Converts to
memcpy(&x, this_cpu_ptr(&y), sizeof(x));
5. Assignment to a per cpu variable
DEFINE_PER_CPU(int, y)
__get_cpu_var(y) = x;
Converts to
__this_cpu_write(y, x);
6. Increment/Decrement etc of a per cpu variable
DEFINE_PER_CPU(int, y);
__get_cpu_var(y)++
Converts to
__this_cpu_inc(y)
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: x86@kernel.org
Acked-by: H. Peter Anvin <hpa@linux.intel.com>
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Update of TLB shootdown code for UV3.
Kernel function native_flush_tlb_others() calls
uv_flush_tlb_others() on UV to invalidate tlb page definitions
on remote cpus. The UV systems have a hardware 'broadcast assist
unit' which can be used to broadcast shootdown messages to all
cpu's of selected nodes.
The behavior of the BAU has changed only slightly with UV3:
- UV3 is recognized with is_uv3_hub().
- UV2 functions and structures (uv2_xxx) are in most cases
simply renamed to uv2_3_xxx.
- Some UV2 error workarounds are not needed for UV3.
(see uv_bau_message_interrupt and enable_timeouts)
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/E1WkgWh-0001yJ-3K@eag09.americas.sgi.com
[ Removed a few linebreak uglies. ]
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The value of n_lshift for UV is currently set based on the
socket m_val.
For UV3, set the n_lshift value based on the GAM_GR_CONFIG MMR.
This will allow bios to control the n_lshift value independent
of the socket m_val. Then n_lshift can be assigned a fixed value
across a multi-partition system, allowing for a fixed common
global physical address format that is independent of socket
m_val.
Cleanup unneeded macros.
Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
Link: http://lkml.kernel.org/r/20140331143700.GB29916@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The current UV NMI handler has not been updated for the changes
in the system NMI handler and the perf operations. The UV NMI
handler reads an MMR in the UV Hub to check to see if the NMI
event was caused by the external 'system NMI' that the operator
can initiate on the System Mgmt Controller.
The problem arises when the perf tools are running, causing
millions of perf events per second on very large CPU count
systems. Previously this was okay because the perf NMI handler
ran at a higher priority on the NMI call chain and if the NMI
was a perf event, it would stop calling other NMI handlers
remaining on the NMI call chain.
Now the system NMI handler calls all the handlers on the NMI
call chain including the UV NMI handler. This causes the UV NMI
handler to read the MMRs at the same millions per second rate.
This can lead to significant performance loss and possible
system failures. It also can cause thousands of 'Dazed and
Confused' messages being sent to the system console. This
effectively makes perf tools unusable on UV systems.
To avoid this excessive overhead when perf tools are running,
this code has been optimized to minimize reading of the MMRs as
much as possible, by moving to the NMI_UNKNOWN notifier chain.
This chain is called only when all the users on the standard
NMI_LOCAL call chain have been called and none of them have
claimed this NMI.
There is an exception where the NMI_LOCAL notifier chain is
used. When the perf tools are in use, it's possible that the UV
NMI was captured by some other NMI handler and then either
ignored or mistakenly processed as a perf event. We set a
per_cpu ('ping') flag for those CPUs that ignored the initial
NMI, and then send them an IPI NMI signal. The NMI_LOCAL
handler on each cpu does not need to read the MMR, but instead
checks the in memory flag indicating it was pinged. There are
two module variables, 'ping_count' indicating how many requested
NMI events occurred, and 'ping_misses' indicating how many stray
NMI events. These most likely are perf events so it shows the
overhead of the perf NMI interrupts and how many MMR reads were avoided.
This patch also minimizes the reads of the MMRs by having the
first cpu entering the NMI handler on each node set a per HUB
in-memory atomic value. (Having a per HUB value avoids sending
lock traffic over NumaLink.) Both types of UV NMIs from the SMI
layer are supported.
Signed-off-by: Mike Travis <travis@sgi.com>
Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Reviewed-by: Hedi Berriche <hedi@sgi.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Jason Wessel <jason.wessel@windriver.com>
Link: http://lkml.kernel.org/r/20130923212500.353547733@asylum.americas.sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
[Purpose of this patch]
As Vaibhav explained in the thread below, tracepoints for irq vectors
are useful.
http://www.spinics.net/lists/mm-commits/msg85707.html
<snip>
The current interrupt traces from irq_handler_entry and irq_handler_exit
provide when an interrupt is handled. They provide good data about when
the system has switched to kernel space and how it affects the currently
running processes.
There are some IRQ vectors which trigger the system into kernel space,
which are not handled in generic IRQ handlers. Tracing such events gives
us the information about IRQ interaction with other system events.
The trace also tells where the system is spending its time. We want to
know which cores are handling interrupts and how they are affecting other
processes in the system. Also, the trace provides information about when
the cores are idle and which interrupts are changing that state.
<snip>
On the other hand, my usecase is tracing just local timer event and
getting a value of instruction pointer.
I suggested to add an argument local timer event to get instruction pointer before.
But there is another way to get it with external module like systemtap.
So, I don't need to add any argument to irq vector tracepoints now.
[Patch Description]
Vaibhav's patch shared a trace point ,irq_vector_entry/irq_vector_exit, in all events.
But there is an above use case to trace specific irq_vector rather than tracing all events.
In this case, we are concerned about overhead due to unwanted events.
So, add following tracepoints instead of introducing irq_vector_entry/exit.
so that we can enable them independently.
- local_timer_vector
- reschedule_vector
- call_function_vector
- call_function_single_vector
- irq_work_entry_vector
- error_apic_vector
- thermal_apic_vector
- threshold_apic_vector
- spurious_apic_vector
- x86_platform_ipi_vector
Also, introduce a logic switching IDT at enabling/disabling time so that a time penalty
makes a zero when tracepoints are disabled. Detailed explanations are as follows.
- Create trace irq handlers with entering_irq()/exiting_irq().
- Create a new IDT, trace_idt_table, at boot time by adding a logic to
_set_gate(). It is just a copy of original idt table.
- Register the new handlers for tracpoints to the new IDT by introducing
macros to alloc_intr_gate() called at registering time of irq_vector handlers.
- Add checking, whether irq vector tracing is on/off, into load_current_idt().
This has to be done below debug checking for these reasons.
- Switching to debug IDT may be kicked while tracing is enabled.
- On the other hands, switching to trace IDT is kicked only when debugging
is disabled.
In addition, the new IDT is created only when CONFIG_TRACING is enabled to avoid being
used for other purposes.
Signed-off-by: Seiji Aguchi <seiji.aguchi@hds.com>
Link: http://lkml.kernel.org/r/51C323ED.5050708@hds.com
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Cc: Steven Rostedt <rostedt@goodmis.org>
This patch trims the MMR register definitions after the updates for the
SGI UV3 system have been applied. Note that because these definitions
are automatically generated from the RTL we cannot control the length
of the names. Therefore there are lines that exceed 80 characters.
Signed-off-by: Mike Travis <travis@sgi.com>
Link: http://lkml.kernel.org/r/20130211194509.173026880@gulag1.americas.sgi.com
Acked-by: Russ Anderson <rja@sgi.com>
Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
This patch updates the MMR register definitions for the SGI UV3 system.
Note that because these definitions are automatically generated from
the RTL we cannot control the length of the names. Therefore there are
lines that exceed 80 characters.
All the new MMR definitions are added in this patch. The patches that
follow then update the references. The last patch is a "trim" patch
which reduces the size of the MMR definitions file by about a third.
This keeps "bi-sectability" in place as the intermediate patches would
not compile correctly if the trimmed MMR defines were done first.
Signed-off-by: Mike Travis <travis@sgi.com>
Link: http://lkml.kernel.org/r/20130211194508.326204556@gulag1.americas.sgi.com
Acked-by: Russ Anderson <rja@sgi.com>
Reviewed-by: Dimitri Sivanich <sivanich@sgi.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
The flush tlb optimization code has logical issue on UV
platform. It doesn't flush the full range at all, since it
simply ignores its 'end' parameter (and hence also the "all"
indicator) in uv_flush_tlb_others() function.
Cliff's notes:
| I tested the patch on a UV. It has the effect of either
| clearing 1 or all TLBs in a cpu. I added some debugging to
| test for the cases when clearing all TLBs is overkill, and in
| practice it happens very seldom.
Reported-by: Jan Beulich <jbeulich@suse.com>
Signed-off-by: Alex Shi <alex.shi@intel.com>
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Tested-by: Cliff Wickman <cpw@sgi.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull x86/mm changes from Peter Anvin:
"The big change here is the patchset by Alex Shi to use INVLPG to flush
only the affected pages when we only need to flush a small page range.
It also removes the special INVALIDATE_TLB_VECTOR interrupts (32
vectors!) and replace it with an ordinary IPI function call."
Fix up trivial conflicts in arch/x86/include/asm/apic.h (added code next
to changed line)
* 'x86-mm-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/tlb: Fix build warning and crash when building for !SMP
x86/tlb: do flush_tlb_kernel_range by 'invlpg'
x86/tlb: replace INVALIDATE_TLB_VECTOR by CALL_FUNCTION_VECTOR
x86/tlb: enable tlb flush range support for x86
mm/mmu_gather: enable tlb flush range in generic mmu_gather
x86/tlb: add tlb_flushall_shift knob into debugfs
x86/tlb: add tlb_flushall_shift for specific CPU
x86/tlb: fall back to flush all when meet a THP large page
x86/flush_tlb: try flush_tlb_single one by one in flush_tlb_range
x86/tlb_info: get last level TLB entry number of CPU
x86: Add read_mostly declaration/definition to variables from smp.h
x86: Define early read-mostly per-cpu macros
x86 has no flush_tlb_range support in instruction level. Currently the
flush_tlb_range just implemented by flushing all page table. That is not
the best solution for all scenarios. In fact, if we just use 'invlpg' to
flush few lines from TLB, we can get the performance gain from later
remain TLB lines accessing.
But the 'invlpg' instruction costs much of time. Its execution time can
compete with cr3 rewriting, and even a bit more on SNB CPU.
So, on a 512 4KB TLB entries CPU, the balance points is at:
(512 - X) * 100ns(assumed TLB refill cost) =
X(TLB flush entries) * 100ns(assumed invlpg cost)
Here, X is 256, that is 1/2 of 512 entries.
But with the mysterious CPU pre-fetcher and page miss handler Unit, the
assumed TLB refill cost is far lower then 100ns in sequential access. And
2 HT siblings in one core makes the memory access more faster if they are
accessing the same memory. So, in the patch, I just do the change when
the target entries is less than 1/16 of whole active tlb entries.
Actually, I have no data support for the percentage '1/16', so any
suggestions are welcomed.
As to hugetlb, guess due to smaller page table, and smaller active TLB
entries, I didn't see benefit via my benchmark, so no optimizing now.
My micro benchmark show in ideal scenarios, the performance improves 70
percent in reading. And in worst scenario, the reading/writing
performance is similar with unpatched 3.4-rc4 kernel.
Here is the reading data on my 2P * 4cores *HT NHM EP machine, with THP
'always':
multi thread testing, '-t' paramter is thread number:
with patch unpatched 3.4-rc4
./mprotect -t 1 14ns 24ns
./mprotect -t 2 13ns 22ns
./mprotect -t 4 12ns 19ns
./mprotect -t 8 14ns 16ns
./mprotect -t 16 28ns 26ns
./mprotect -t 32 54ns 51ns
./mprotect -t 128 200ns 199ns
Single process with sequencial flushing and memory accessing:
with patch unpatched 3.4-rc4
./mprotect 7ns 11ns
./mprotect -p 4096 -l 8 -n 10240
21ns 21ns
[ hpa: http://lkml.kernel.org/r/1B4B44D9196EFF41AE41FDA404FC0A100BFF94@SHSMSX101.ccr.corp.intel.com
has additional performance numbers. ]
Signed-off-by: Alex Shi <alex.shi@intel.com>
Link: http://lkml.kernel.org/r/1340845344-27557-3-git-send-email-alex.shi@intel.com
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
On SGI's UV2 the BAU (Broadcast Assist Unit) driver can hang
under a heavy load. To cure this:
- Disable the UV2 extended status mode (see UV2_EXT_SHFT), as
this mode changes BAU behavior in more ways then just delivering
an extra bit of status. Revert status to just two meaningful bits,
like UV1.
- Use no IPI-style resets on UV2. Just give up the request for
whatever the reason it failed and let it be accomplished with
the legacy IPI method.
- Use no alternate sending descriptor (the former UV2 workaround
bcp->using_desc and handle_uv2_busy() stuff). Just disable the
use of the BAU for a period of time in favor of the legacy IPI
method when the h/w bug leaves a descriptor busy.
-- new tunable: giveup_limit determines the threshold at which a hub is
so plugged that it should do all requests with the legacy IPI method for a
period of time
-- generalize disable_for_congestion() (renamed disable_for_period()) for
use whenever a hub should avoid using the BAU for a period of time
Also:
- Fix find_another_by_swack(), which is part of the UV2 bug workaround
- Correct and clarify the statistics (new stats s_overipilimit, s_giveuplimit,
s_enters, s_ipifordisabled, s_plugged, s_congested)
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/20120622131459.GC31884@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch enables the BAU to be turned on or off dynamically.
echo "on" > /proc/sgi_uv/ptc_statistics
echo "off" > /proc/sgi_uv/ptc_statistics
The system may be booted with or without the nobau option.
Whether the system currently has the BAU off can be seen in
the /proc file -- normally with the baustats script.
Each cpu will have a 1 in the bauoff field if the BAU was turned
off, so baustats will give a count of cpus that have it off.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/20120622131330.GB31884@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The SGI Altix UV2 BAU (Broadcast Assist Unit) as used for
tlb-shootdown (selective broadcast mode) always uses UV2
broadcast descriptor format. There is no need to clear the
'legacy' (UV1) mode, because the hardware always uses UV2 mode
for selective broadcast.
But the BIOS uses general broadcast and legacy mode, and the
hardware pays attention to the legacy mode bit for general
broadcast. So the kernel must not clear that mode bit.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Cc: <stable@kernel.org>
Link: http://lkml.kernel.org/r/E1SccoO-0002Lh-Cb@eag09.americas.sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch implements a workaround for a UV2 hardware bug.
The bug is a non-atomic update of a memory-mapped register. When
hardware message delivery and software message acknowledge occur
simultaneously the pending message acknowledge for the arriving
message may be lost. This causes the sender's message status to
stay busy.
Part of the workaround is to not acknowledge a completed message
until it is verified that no other message is actually using the
resource that is mistakenly recorded in the completed message.
Part of the workaround is to test for long elapsed time in such
a busy condition, then handle it by using a spare sending
descriptor. The stay-busy condition is eventually timed out by
hardware, and then the original sending descriptor can be
re-used. Most of that logic change is in keeping track of the
current descriptor and the state of the spares.
The occurrences of the workaround are added to the BAU
statistics.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/20120116211947.GC5767@sgi.com
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Update the use of the Broadcast Assist Unit on SGI Altix UV2 to
the use of native UV2 mode on new hardware (not the legacy mode).
UV2 native mode has a different format for a broadcast message.
We also need quick differentiaton between UV1 and UV2.
Signed-off-by: Cliff Wickman <cpw@sgi.com>
Link: http://lkml.kernel.org/r/20120116211750.GA5767@sgi.com
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There was a mixup when the SGI UV2 hub chip was sent to be
fabricated, and it ended up with the wrong part number in the
HRP_NODE_ID mmr. Future versions of the chip will (may) have the
correct part number. Change the UV infrastructure to recognize
both part numbers as valid IDs of a UV2 hub chip.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Link: http://lkml.kernel.org/r/20111129210058.GA20452@sgi.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This is a workaround for a UV2 hub bug that affects the format of system
global addresses.
The GRU API for UV2 was inadvertently broken by a hardware change. The
format of the physical address used for TLB dropins and for addresses used
with instructions running in unmapped mode has changed. This change was
not documented and became apparent only when diags failed running on
system simulators.
For UV1, TLB and GRU instruction physical addresses are identical to
socket physical addresses (although high NASID bits must be OR'ed into the
address).
For UV2, socket physical addresses need to be converted. The NODE portion
of the physical address needs to be shifted so that the low bit is in bit
39 or bit 40, depending on an MMR value.
It is not yet clear if this bug will be fixed in a silicon respin. If it
is fixed, the hub revision will be incremented & the workaround disabled.
Signed-off-by: Jack Steiner <steiner@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
