linux-toradex.git/arch/x86/kernel/cpu, branch v3.10.41

perf/x86: Fix event scheduling

2014-03-07T05:30:08+00:00

commit 26e61e8939b1fe8729572dabe9a9e97d930dd4f6 upstream.

Vince "Super Tester" Weaver reported a new round of syscall fuzzing (Trinity) failures,
with perf WARN_ON()s triggering. He also provided traces of the failures.

This is I think the relevant bit:

	>    pec_1076_warn-2804  [000] d...   147.926153: x86_pmu_disable: x86_pmu_disable
	>    pec_1076_warn-2804  [000] d...   147.926153: x86_pmu_state: Events: {
	>    pec_1076_warn-2804  [000] d...   147.926156: x86_pmu_state:   0: state: .R config: ffffffffffffffff (          (null))
	>    pec_1076_warn-2804  [000] d...   147.926158: x86_pmu_state:   33: state: AR config: 0 (ffff88011ac99800)
	>    pec_1076_warn-2804  [000] d...   147.926159: x86_pmu_state: }
	>    pec_1076_warn-2804  [000] d...   147.926160: x86_pmu_state: n_events: 1, n_added: 0, n_txn: 1
	>    pec_1076_warn-2804  [000] d...   147.926161: x86_pmu_state: Assignment: {
	>    pec_1076_warn-2804  [000] d...   147.926162: x86_pmu_state:   0->33 tag: 1 config: 0 (ffff88011ac99800)
	>    pec_1076_warn-2804  [000] d...   147.926163: x86_pmu_state: }
	>    pec_1076_warn-2804  [000] d...   147.926166: collect_events: Adding event: 1 (ffff880119ec8800)

So we add the insn:p event (fd[23]).

At this point we should have:

  n_events = 2, n_added = 1, n_txn = 1

	>    pec_1076_warn-2804  [000] d...   147.926170: collect_events: Adding event: 0 (ffff8800c9e01800)
	>    pec_1076_warn-2804  [000] d...   147.926172: collect_events: Adding event: 4 (ffff8800cbab2c00)

We try and add the {BP,cycles,br_insn} group (fd[3], fd[4], fd[15]).
These events are 0:cycles and 4:br_insn, the BP event isn't x86_pmu so
that's not visible.

	group_sched_in()
	  pmu->start_txn() /* nop - BP pmu */
	  event_sched_in()
	     event->pmu->add()

So here we should end up with:

  0: n_events = 3, n_added = 2, n_txn = 2
  4: n_events = 4, n_added = 3, n_txn = 3

But seeing the below state on x86_pmu_enable(), the must have failed,
because the 0 and 4 events aren't there anymore.

Looking at group_sched_in(), since the BP is the leader, its
event_sched_in() must have succeeded, for otherwise we would not have
seen the sibling adds.

But since neither 0 or 4 are in the below state; their event_sched_in()
must have failed; but I don't see why, the complete state: 0,0,1:p,4
fits perfectly fine on a core2.

However, since we try and schedule 4 it means the 0 event must have
succeeded!  Therefore the 4 event must have failed, its failure will
have put group_sched_in() into the fail path, which will call:

	event_sched_out()
	  event->pmu->del()

on 0 and the BP event.

Now x86_pmu_del() will reduce n_events; but it will not reduce n_added;
giving what we see below:

 n_event = 2, n_added = 2, n_txn = 2

	>    pec_1076_warn-2804  [000] d...   147.926177: x86_pmu_enable: x86_pmu_enable
	>    pec_1076_warn-2804  [000] d...   147.926177: x86_pmu_state: Events: {
	>    pec_1076_warn-2804  [000] d...   147.926179: x86_pmu_state:   0: state: .R config: ffffffffffffffff (          (null))
	>    pec_1076_warn-2804  [000] d...   147.926181: x86_pmu_state:   33: state: AR config: 0 (ffff88011ac99800)
	>    pec_1076_warn-2804  [000] d...   147.926182: x86_pmu_state: }
	>    pec_1076_warn-2804  [000] d...   147.926184: x86_pmu_state: n_events: 2, n_added: 2, n_txn: 2
	>    pec_1076_warn-2804  [000] d...   147.926184: x86_pmu_state: Assignment: {
	>    pec_1076_warn-2804  [000] d...   147.926186: x86_pmu_state:   0->33 tag: 1 config: 0 (ffff88011ac99800)
	>    pec_1076_warn-2804  [000] d...   147.926188: x86_pmu_state:   1->0 tag: 1 config: 1 (ffff880119ec8800)
	>    pec_1076_warn-2804  [000] d...   147.926188: x86_pmu_state: }
	>    pec_1076_warn-2804  [000] d...   147.926190: x86_pmu_enable: S0: hwc->idx: 33, hwc->last_cpu: 0, hwc->last_tag: 1 hwc->state: 0

So the problem is that x86_pmu_del(), when called from a
group_sched_in() that fails (for whatever reason), and without x86_pmu
TXN support (because the leader is !x86_pmu), will corrupt the n_added
state.

Reported-and-Tested-by: Vince Weaver 
Signed-off-by: Peter Zijlstra 
Cc: Paul Mackerras 
Cc: Steven Rostedt 
Cc: Stephane Eranian 
Cc: Dave Jones 
Link: http://lkml.kernel.org/r/20140221150312.GF3104@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar 
Signed-off-by: Greg Kroah-Hartman

x86, smap: Don't enable SMAP if CONFIG_X86_SMAP is disabled

2014-02-22T20:41:27+00:00

commit 03bbd596ac04fef47ce93a730b8f086d797c3021 upstream.

If SMAP support is not compiled into the kernel, don't enable SMAP in
CR4 -- in fact, we should clear it, because the kernel doesn't contain
the proper STAC/CLAC instructions for SMAP support.

Found by Fengguang Wu's test system.

Reported-by: Fengguang Wu 
Link: http://lkml.kernel.org/r/20140213124550.GA30497@localhost
Signed-off-by: H. Peter Anvin 
Signed-off-by: Greg Kroah-Hartman

x86: mm: change tlb_flushall_shift for IvyBridge

2014-02-20T19:06:11+00:00

commit f98b7a772ab51b52ca4d2a14362fc0e0c8a2e0f3 upstream.

There was a large performance regression that was bisected to
commit 611ae8e3 ("x86/tlb: enable tlb flush range support for
x86").  This patch simply changes the default balance point
between a local and global flush for IvyBridge.

In the interest of allowing the tests to be reproduced, this
patch was tested using mmtests 0.15 with the following
configurations

	configs/config-global-dhp__tlbflush-performance
	configs/config-global-dhp__scheduler-performance
	configs/config-global-dhp__network-performance

Results are from two machines

Ivybridge   4 threads:  Intel(R) Core(TM) i3-3240 CPU @ 3.40GHz
Ivybridge   8 threads:  Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz

Page fault microbenchmark showed nothing interesting.

Ebizzy was configured to run multiple iterations and threads.
Thread counts ranged from 1 to NR_CPUS*2. For each thread count,
it ran 100 iterations and each iteration lasted 10 seconds.

Ivybridge 4 threads
                    3.13.0-rc7            3.13.0-rc7
                       vanilla           altshift-v3
Mean   1     6395.44 (  0.00%)     6789.09 (  6.16%)
Mean   2     7012.85 (  0.00%)     8052.16 ( 14.82%)
Mean   3     6403.04 (  0.00%)     6973.74 (  8.91%)
Mean   4     6135.32 (  0.00%)     6582.33 (  7.29%)
Mean   5     6095.69 (  0.00%)     6526.68 (  7.07%)
Mean   6     6114.33 (  0.00%)     6416.64 (  4.94%)
Mean   7     6085.10 (  0.00%)     6448.51 (  5.97%)
Mean   8     6120.62 (  0.00%)     6462.97 (  5.59%)

Ivybridge 8 threads
                     3.13.0-rc7            3.13.0-rc7
                        vanilla           altshift-v3
Mean   1      7336.65 (  0.00%)     7787.02 (  6.14%)
Mean   2      8218.41 (  0.00%)     9484.13 ( 15.40%)
Mean   3      7973.62 (  0.00%)     8922.01 ( 11.89%)
Mean   4      7798.33 (  0.00%)     8567.03 (  9.86%)
Mean   5      7158.72 (  0.00%)     8214.23 ( 14.74%)
Mean   6      6852.27 (  0.00%)     7952.45 ( 16.06%)
Mean   7      6774.65 (  0.00%)     7536.35 ( 11.24%)
Mean   8      6510.50 (  0.00%)     6894.05 (  5.89%)
Mean   12     6182.90 (  0.00%)     6661.29 (  7.74%)
Mean   16     6100.09 (  0.00%)     6608.69 (  8.34%)

Ebizzy hits the worst case scenario for TLB range flushing every
time and it shows for these Ivybridge CPUs at least that the
default choice is a poor on.  The patch addresses the problem.

Next was a tlbflush microbenchmark written by Alex Shi at
http://marc.info/?l=linux-kernel&m=133727348217113 .  It
measures access costs while the TLB is being flushed.  The
expectation is that if there are always full TLB flushes that
the benchmark would suffer and it benefits from range flushing

There are 320 iterations of the test per thread count.  The
number of entries is randomly selected with a min of 1 and max
of 512.  To ensure a reasonably even spread of entries, the full
range is broken up into 8 sections and a random number selected
within that section.

iteration 1, random number between 0-64
iteration 2, random number between 64-128 etc

This is still a very weak methodology.  When you do not know
what are typical ranges, random is a reasonable choice but it
can be easily argued that the opimisation was for smaller ranges
and an even spread is not representative of any workload that
matters.  To improve this, we'd need to know the probability
distribution of TLB flush range sizes for a set of workloads
that are considered "common", build a synthetic trace and feed
that into this benchmark.  Even that is not perfect because it
would not account for the time between flushes but there are
limits of what can be reasonably done and still be doing
something useful.  If a representative synthetic trace is
provided then this benchmark could be revisited and the shift values retuned.

Ivybridge 4 threads
                        3.13.0-rc7            3.13.0-rc7
                           vanilla           altshift-v3
Mean       1       10.50 (  0.00%)       10.50 (  0.03%)
Mean       2       17.59 (  0.00%)       17.18 (  2.34%)
Mean       3       22.98 (  0.00%)       21.74 (  5.41%)
Mean       5       47.13 (  0.00%)       46.23 (  1.92%)
Mean       8       43.30 (  0.00%)       42.56 (  1.72%)

Ivybridge 8 threads
                         3.13.0-rc7            3.13.0-rc7
                            vanilla           altshift-v3
Mean       1         9.45 (  0.00%)        9.36 (  0.93%)
Mean       2         9.37 (  0.00%)        9.70 ( -3.54%)
Mean       3         9.36 (  0.00%)        9.29 (  0.70%)
Mean       5        14.49 (  0.00%)       15.04 ( -3.75%)
Mean       8        41.08 (  0.00%)       38.73 (  5.71%)
Mean       13       32.04 (  0.00%)       31.24 (  2.49%)
Mean       16       40.05 (  0.00%)       39.04 (  2.51%)

For both CPUs, average access time is reduced which is good as
this is the benchmark that was used to tune the shift values in
the first place albeit it is now known *how* the benchmark was
used.

The scheduler benchmarks were somewhat inconclusive.  They
showed gains and losses and makes me reconsider how stable those
benchmarks really are or if something else might be interfering
with the test results recently.

Network benchmarks were inconclusive.  Almost all results were
flat except for netperf-udp tests on the 4 thread machine.
These results were unstable and showed large variations between
reboots.  It is unknown if this is a recent problems but I've
noticed before that netperf-udp results tend to vary.

Based on these results, changing the default for Ivybridge seems
like a logical choice.

Signed-off-by: Mel Gorman 
Tested-by: Davidlohr Bueso 
Reviewed-by: Alex Shi 
Reviewed-by: Rik van Riel 
Signed-off-by: Andrew Morton 
Cc: Linus Torvalds 
Cc: Peter Zijlstra 
Link: http://lkml.kernel.org/n/tip-cqnadffh1tiqrshthRj3Esge@git.kernel.org
Signed-off-by: Ingo Molnar 
Signed-off-by: Greg Kroah-Hartman

x86, cpu, amd: Add workaround for family 16h, erratum 793

2014-02-06T19:08:18+00:00

commit 3b56496865f9f7d9bcb2f93b44c63f274f08e3b6 upstream.

This adds the workaround for erratum 793 as a precaution in case not
every BIOS implements it.  This addresses CVE-2013-6885.

Erratum text:

[Revision Guide for AMD Family 16h Models 00h-0Fh Processors,
document 51810 Rev. 3.04 November 2013]

793 Specific Combination of Writes to Write Combined Memory Types and
Locked Instructions May Cause Core Hang

Description

Under a highly specific and detailed set of internal timing
conditions, a locked instruction may trigger a timing sequence whereby
the write to a write combined memory type is not flushed, causing the
locked instruction to stall indefinitely.

Potential Effect on System

Processor core hang.

Suggested Workaround

BIOS should set MSR
C001_1020[15] = 1b.

Fix Planned

No fix planned

[ hpa: updated description, fixed typo in MSR name ]

Signed-off-by: Borislav Petkov 
Link: http://lkml.kernel.org/r/20140114230711.GS29865@pd.tnic
Tested-by: Aravind Gopalakrishnan 
Signed-off-by: H. Peter Anvin 
Signed-off-by: Greg Kroah-Hartman

perf/x86/amd/ibs: Fix waking up from S3 for AMD family 10h

2014-01-25T16:27:11+00:00

commit bee09ed91cacdbffdbcd3b05de8409c77ec9fcd6 upstream.

On AMD family 10h we see following error messages while waking up from
S3 for all non-boot CPUs leading to a failed IBS initialization:

 Enabling non-boot CPUs ...
 smpboot: Booting Node 0 Processor 1 APIC 0x1
 [Firmware Bug]: cpu 1, try to use APIC500 (LVT offset 0) for vector 0x400, but the register is already in use for vector 0xf9 on another cpu
 perf: IBS APIC setup failed on cpu #1
 process: Switch to broadcast mode on CPU1
 CPU1 is up
 ...
 ACPI: Waking up from system sleep state S3

Reason for this is that during suspend the LVT offset for the IBS
vector gets lost and needs to be reinialized while resuming.

The offset is read from the IBSCTL msr. On family 10h the offset needs
to be 1 as offset 0 is used for the MCE threshold interrupt, but
firmware assings it for IBS to 0 too. The kernel needs to reprogram
the vector. The msr is a readonly node msr, but a new value can be
written via pci config space access. The reinitialization is
implemented for family 10h in setup_ibs_ctl() which is forced during
IBS setup.

This patch fixes IBS setup after waking up from S3 by adding
resume/supend hooks for the boot cpu which does the offset
reinitialization.

Marking it as stable to let distros pick up this fix.

Signed-off-by: Robert Richter 
Signed-off-by: Peter Zijlstra 
Cc: Linus Torvalds 
Link: http://lkml.kernel.org/r/1389797849-5565-1-git-send-email-rric.net@gmail.com
Signed-off-by: Ingo Molnar 
Signed-off-by: Greg Kroah-Hartman

x86 idle: Repair large-server 50-watt idle-power regression

2014-01-09T20:24:21+00:00

commit 40e2d7f9b5dae048789c64672bf3027fbb663ffa upstream.

Linux 3.10 changed the timing of how thread_info->flags is touched:

	x86: Use generic idle loop
	(7d1a941731fabf27e5fb6edbebb79fe856edb4e5)

This caused Intel NHM-EX and WSM-EX servers to experience a large number
of immediate MONITOR/MWAIT break wakeups, which caused cpuidle to demote
from deep C-states to shallow C-states, which caused these platforms
to experience a significant increase in idle power.

Note that this issue was already present before the commit above,
however, it wasn't seen often enough to be noticed in power measurements.

Here we extend an errata workaround from the Core2 EX "Dunnington"
to extend to NHM-EX and WSM-EX, to prevent these immediate
returns from MWAIT, reducing idle power on these platforms.

While only acpi_idle ran on Dunnington, intel_idle
may also run on these two newer systems.
As of today, there are no other models that are known
to need this tweak.

Link: http://lkml.kernel.org/r/CAJvTdK=%2BaNN66mYpCGgbHGCHhYQAKx-vB0kJSWjVpsNb_hOAtQ@mail.gmail.com
Signed-off-by: Len Brown 
Link: http://lkml.kernel.org/r/baff264285f6e585df757d58b17788feabc68918.1387403066.git.len.brown@intel.com
Signed-off-by: H. Peter Anvin 
Signed-off-by: Greg Kroah-Hartman

perf/x86: Fix intel QPI uncore event definitions

2013-08-20T15:43:02+00:00

commit c9601247f8f3fdc18aed7ed7e490e8dfcd07f122 upstream.

John McCalpin reports that the "drs_data" and "ncb_data" QPI
uncore events are missing the "extra bit" and always return zero
values unless the bit is properly set.

More details from him:

 According to the Xeon E5-2600 Product Family Uncore Performance
 Monitoring Guide, Table 2-94, about 1/2 of the QPI Link Layer events
 (including the ones that "perf" calls "drs_data" and "ncb_data") require
 that the "extra bit" be set.

 This was confusing for a while -- a note at the bottom of page 94 says
 that the "extra bit" is bit 16 of the control register.
 Unfortunately, Table 2-86 clearly says that bit 16 is reserved and must
 be zero.  Looking around a bit, I found that bit 21 appears to be the
 correct "extra bit", and further investigation shows that "perf" actually
 agrees with me:
	[root@c560-003.stampede]# cat /sys/bus/event_source/devices/uncore_qpi_0/format/event
	config:0-7,21

 So the command
	# perf -e "uncore_qpi_0/event=drs_data/"
 Is the same as
	# perf -e "uncore_qpi_0/event=0x02,umask=0x08/"
 While it should be
	# perf -e "uncore_qpi_0/event=0x102,umask=0x08/"

 I confirmed that this last version gives results that agree with the
 amount of data that I expected the STREAM benchmark to move across the QPI
 link in the second (cross-chip) test of the original script.

Reported-by: John McCalpin 
Signed-off-by: Vince Weaver 
Cc: zheng.z.yan@intel.com
Cc: Peter Zijlstra 
Cc: Arnaldo Carvalho de Melo 
Cc: Paul Mackerras 
Link: http://lkml.kernel.org/r/alpine.DEB.2.10.1308021037280.26119@vincent-weaver-1.um.maine.edu
Signed-off-by: Ingo Molnar 
Signed-off-by: Greg Kroah-Hartman

x86: Fix /proc/mtrr with base/size more than 44bits

2013-08-04T08:51:18+00:00

commit d5c78673b1b28467354c2c30c3d4f003666ff385 upstream.

On one sytem that mtrr range is more then 44bits, in dmesg we have
[    0.000000] MTRR default type: write-back
[    0.000000] MTRR fixed ranges enabled:
[    0.000000]   00000-9FFFF write-back
[    0.000000]   A0000-BFFFF uncachable
[    0.000000]   C0000-DFFFF write-through
[    0.000000]   E0000-FFFFF write-protect
[    0.000000] MTRR variable ranges enabled:
[    0.000000]   0 [000080000000-0000FFFFFFFF] mask 3FFF80000000 uncachable
[    0.000000]   1 [380000000000-38FFFFFFFFFF] mask 3F0000000000 uncachable
[    0.000000]   2 [000099000000-000099FFFFFF] mask 3FFFFF000000 write-through
[    0.000000]   3 [00009A000000-00009AFFFFFF] mask 3FFFFF000000 write-through
[    0.000000]   4 [381FFA000000-381FFBFFFFFF] mask 3FFFFE000000 write-through
[    0.000000]   5 [381FFC000000-381FFC0FFFFF] mask 3FFFFFF00000 write-through
[    0.000000]   6 [0000AD000000-0000ADFFFFFF] mask 3FFFFF000000 write-through
[    0.000000]   7 [0000BD000000-0000BDFFFFFF] mask 3FFFFF000000 write-through
[    0.000000]   8 disabled
[    0.000000]   9 disabled

but /proc/mtrr report wrong:
reg00: base=0x080000000 ( 2048MB), size= 2048MB, count=1: uncachable
reg01: base=0x80000000000 (8388608MB), size=1048576MB, count=1: uncachable
reg02: base=0x099000000 ( 2448MB), size=   16MB, count=1: write-through
reg03: base=0x09a000000 ( 2464MB), size=   16MB, count=1: write-through
reg04: base=0x81ffa000000 (8519584MB), size=   32MB, count=1: write-through
reg05: base=0x81ffc000000 (8519616MB), size=    1MB, count=1: write-through
reg06: base=0x0ad000000 ( 2768MB), size=   16MB, count=1: write-through
reg07: base=0x0bd000000 ( 3024MB), size=   16MB, count=1: write-through
reg08: base=0x09b000000 ( 2480MB), size=   16MB, count=1: write-combining

so bit 44 and bit 45 get cut off.

We have problems in arch/x86/kernel/cpu/mtrr/generic.c::generic_get_mtrr().
1. for base, we miss cast base_lo to 64bit before shifting.
Fix that by adding u64 casting.

2. for size, it only can handle 44 bits aka 32bits + page_shift
Fix that with 64bit mask instead of 32bit mask_lo, then range could be
more than 44bits.
At the same time, we need to update size_or_mask for old cpus that does
support cpuid 0x80000008 to get phys_addr. Need to set high 32bits
to all 1s, otherwise will not get correct size for them.

Also fix mtrr_add_page: it should check base and (base + size - 1)
instead of base and size, as base and size could be small but
base + size could bigger enough to be out of boundary. We can
use boot_cpu_data.x86_phys_bits directly to avoid size_or_mask.

So When are we going to have size more than 44bits? that is 16TiB.

after patch we have right ouput:
reg00: base=0x080000000 ( 2048MB), size= 2048MB, count=1: uncachable
reg01: base=0x380000000000 (58720256MB), size=1048576MB, count=1: uncachable
reg02: base=0x099000000 ( 2448MB), size=   16MB, count=1: write-through
reg03: base=0x09a000000 ( 2464MB), size=   16MB, count=1: write-through
reg04: base=0x381ffa000000 (58851232MB), size=   32MB, count=1: write-through
reg05: base=0x381ffc000000 (58851264MB), size=    1MB, count=1: write-through
reg06: base=0x0ad000000 ( 2768MB), size=   16MB, count=1: write-through
reg07: base=0x0bd000000 ( 3024MB), size=   16MB, count=1: write-through
reg08: base=0x09b000000 ( 2480MB), size=   16MB, count=1: write-combining

-v2: simply checking in mtrr_add_page according to hpa.

[ hpa: This probably wants to go into -stable only after having sat in
  mainline for a bit.  It is not a regression. ]

Signed-off-by: Yinghai Lu 
Link: http://lkml.kernel.org/r/1371162815-29931-1-git-send-email-yinghai@kernel.org
Signed-off-by: H. Peter Anvin 
Signed-off-by: Greg Kroah-Hartman

Merge branch 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

2013-06-21T16:33:48+00:00

Pull x86 fixes from Peter Anvin:
 "This series fixes a couple of build failures, and fixes MTRR cleanup
  and memory setup on very specific memory maps.

  Finally, it fixes triggering backtraces on all CPUs, which was
  inadvertently disabled on x86."

* 'x86/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/efi: Fix dummy variable buffer allocation
  x86: Fix trigger_all_cpu_backtrace() implementation
  x86: Fix section mismatch on load_ucode_ap
  x86: fix build error and kconfig for ia32_emulation and binfmt
  range: Do not add new blank slot with add_range_with_merge
  x86, mtrr: Fix original mtrr range get for mtrr_cleanup

perf/x86: Fix broken PEBS-LL support on SNB-EP/IVB-EP

2013-06-19T10:44:16+00:00

This patch fixes broken support of PEBS-LL on SNB-EP/IVB-EP.
For some reason, the LDLAT extra reg definition for snb_ep
showed up as duplicate in the snb table.

This patch moves the definition of LDLAT back into the
snb_ep table.

Thanks to Don Zickus for tracking this one down.

Signed-off-by: Stephane Eranian 
Signed-off-by: Peter Zijlstra 
Link: http://lkml.kernel.org/r/20130607212210.GA11849@quad
Signed-off-by: Ingo Molnar