Pull x86 MDS mitigations from Thomas Gleixner:
"Microarchitectural Data Sampling (MDS) is a hardware vulnerability
which allows unprivileged speculative access to data which is
available in various CPU internal buffers. This new set of misfeatures
has the following CVEs assigned:
CVE-2018-12126 MSBDS Microarchitectural Store Buffer Data Sampling
CVE-2018-12130 MFBDS Microarchitectural Fill Buffer Data Sampling
CVE-2018-12127 MLPDS Microarchitectural Load Port Data Sampling
CVE-2019-11091 MDSUM Microarchitectural Data Sampling Uncacheable Memory
MDS attacks target microarchitectural buffers which speculatively
forward data under certain conditions. Disclosure gadgets can expose
this data via cache side channels.
Contrary to other speculation based vulnerabilities the MDS
vulnerability does not allow the attacker to control the memory target
address. As a consequence the attacks are purely sampling based, but
as demonstrated with the TLBleed attack samples can be postprocessed
successfully.
The mitigation is to flush the microarchitectural buffers on return to
user space and before entering a VM. It's bolted on the VERW
instruction and requires a microcode update. As some of the attacks
exploit data structures shared between hyperthreads, full protection
requires to disable hyperthreading. The kernel does not do that by
default to avoid breaking unattended updates.
The mitigation set comes with documentation for administrators and a
deeper technical view"
* 'x86-mds-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (23 commits)
x86/speculation/mds: Fix documentation typo
Documentation: Correct the possible MDS sysfs values
x86/mds: Add MDSUM variant to the MDS documentation
x86/speculation/mds: Add 'mitigations=' support for MDS
x86/speculation/mds: Print SMT vulnerable on MSBDS with mitigations off
x86/speculation/mds: Fix comment
x86/speculation/mds: Add SMT warning message
x86/speculation: Move arch_smt_update() call to after mitigation decisions
x86/speculation/mds: Add mds=full,nosmt cmdline option
Documentation: Add MDS vulnerability documentation
Documentation: Move L1TF to separate directory
x86/speculation/mds: Add mitigation mode VMWERV
x86/speculation/mds: Add sysfs reporting for MDS
x86/speculation/mds: Add mitigation control for MDS
x86/speculation/mds: Conditionally clear CPU buffers on idle entry
x86/kvm/vmx: Add MDS protection when L1D Flush is not active
x86/speculation/mds: Clear CPU buffers on exit to user
x86/speculation/mds: Add mds_clear_cpu_buffers()
x86/kvm: Expose X86_FEATURE_MD_CLEAR to guests
x86/speculation/mds: Add BUG_MSBDS_ONLY
...
Pull turbostat utility updates for 5.1 from Len Brown:
"Misc fixes and updates."
* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: update version number
tools/power turbostat: Warn on bad ACPI LPIT data
tools/power turbostat: Add checks for failure of fgets() and fscanf()
tools/power turbostat: Also read package power on AMD F17h (Zen)
tools/power turbostat: Add support for AMD Fam 17h (Zen) RAPL
tools/power turbostat: Do not display an error on systems without a cpufreq driver
tools/power turbostat: Add Die column
tools/power turbostat: Add Icelake support
tools/power turbostat: Cleanup CNL-specific code
tools/power turbostat: Cleanup CC3-skip code
tools/power turbostat: Restore ability to execute in topology-order
On some systems /sys/devices/system/cpu/cpuidle/low_power_idle_cpu_residency_us
or /sys/devices/system/cpu/cpuidle/low_power_idle_system_residency_us
return a file error because of bad ACPI LPIT data from a misconfigured BIOS.
turbostat interprets this failure as a fatal error and outputs
turbostat: CPU LPI: No data available
If the ACPI LPIT sysfs files return an error output a warning instead of
a fatal error, disable the ACPI LPIT evaluation code, and continue.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Most calls to fgets() and fscanf() are followed by error checks.
Add an exit-on-error in the remaining cases.
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: Len Brown <len.brown@intel.com>
The package power can also be read from an MSR. It's not clear exactly
what is included, and whether it's aggregated over all nodes or
reported separately.
It does look like this is reported separately per CCX (I get a single
value on the Ryzen R7 1700), but it might be reported separately per-
die (node?) on larger processors. If that's the case, it would have to
be recorded per node and aggregated for the socket.
Note that although Zen has these MSRs reporting power, it looks like
the actual RAPL configuration (power limits, configured TDP) is done
through PCI configuration space. I have not yet found any public
documentation for this.
Signed-off-by: Calvin Walton <calvin.walton@kepstin.ca>
Signed-off-by: Len Brown <len.brown@intel.com>
Based on the Open-Source Register Reference for AMD Family 17h
Processors Models 00h-2Fh:
https://support.amd.com/TechDocs/56255_OSRR.pdf
These processors report RAPL support in bit 14 of CPUID 0x80000007 EDX,
and the following MSRs are present:
0xc0010299 (RAPL_PWR_UNIT), like Intel's RAPL_POWER_UNIT
0xc001029a (CORE_ENERGY_STAT), kind of like Intel's PP0_ENERGY_STATUS
0xc001029b (PKG_ENERGY_STAT), like Intel's PKG_ENERGY_STATUS
A notable difference from the Intel implementation is that AMD reports
the "Cores" energy usage separately for each core, rather than a
per-package total. The code has been adjusted to handle either case in a
generic way.
I haven't yet enabled collection of package power, due to being unable
to test it on multi-node systems (TR, EPYC).
Signed-off-by: Calvin Walton <calvin.walton@kepstin.ca>
Signed-off-by: Len Brown <len.brown@intel.com>
Running without a cpufreq driver is a valid case so warnings output in
this case should not be to stderr.
Use outf instead of stderr for these warnings.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat executes on CPUs in "topology order".
This is an optimization for measuring profoundly idle systems --
as the closest hardware is woken next...
Fix a typo that was added with the sub-die-node support,
that broke topology ordering on multi-node systems.
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat failed to return a non-zero exit status even though the
supplied command (turbostat <command>) failed. Currently when turbostat
forks a command it returns zero instead of the actual exit status of the
command. Modify the code to return the exit status.
Signed-off-by: David Arcari <darcari@redhat.com>
Acked-by: Len Brown <len.brown@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Greg pointed out that speculation related bit defines are using (1 << N)
format instead of BIT(N). Aside of that (1 << N) is wrong as it should use
1UL at least.
Clean it up.
[ Josh Poimboeuf: Fix tools build ]
Reported-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Jon Masters <jcm@redhat.com>
Tested-by: Jon Masters <jcm@redhat.com>
perf c2c:
Jiri Olsa:
- Change the default coalesce setup to from '--coalesce pid,iaddr' to just '--coalesce iaddr'.
- Increase the HITM ratio limit for displayed cachelines.
perf script:
Andi Kleen:
- Fix LBR skid dump problems in brstackinsn.
perf trace:
Arnaldo Carvalho de Melo:
- Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter.
- Do not hardcode the size of the tracepoint common_ fields.
- Beautify USBDEFFS_ ioctl commands.
Colin Ian King:
- Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*".
perf python:
Jiri Olsa:
- Do not force closing original perf descriptor in evlist.get_pollfd().
tools misc:
Jiri Olsa:
- Allow overriding CFLAGS and LDFLAGS.
perf build:
Stanislav Fomichev:
- Don't unconditionally link the libbfd feature test to -liberty and -lz
thread-stack:
Adrian Hunter:
- Fix processing for the idle task, having a stack per cpu.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
-----BEGIN PGP SIGNATURE-----
iHUEABYIAB0WIQR2GiIUctdOfX2qHhGyPKLppCJ+JwUCXC4BOgAKCRCyPKLppCJ+
Jy1BAP0Vr6fC+mv/ul0x3WC4dlF0UG9p9+GxBoKsXPpG5vojCgEAqX7F+Pmx+6HK
FIBjbOWIL5NYRViskwPlQy5+qkKmJgA=
=I4gE
-----END PGP SIGNATURE-----
Merge tag 'perf-core-for-mingo-4.21-20190103' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent
Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:
perf c2c:
Jiri Olsa:
- Change the default coalesce setup to from '--coalesce pid,iaddr' to just '--coalesce iaddr'.
- Increase the HITM ratio limit for displayed cachelines.
perf script:
Andi Kleen:
- Fix LBR skid dump problems in brstackinsn.
perf trace:
Arnaldo Carvalho de Melo:
- Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter.
- Do not hardcode the size of the tracepoint common_ fields.
- Beautify USBDEFFS_ ioctl commands.
Colin Ian King:
- Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*".
perf python:
Jiri Olsa:
- Do not force closing original perf descriptor in evlist.get_pollfd().
tools misc:
Jiri Olsa:
- Allow overriding CFLAGS and LDFLAGS.
perf build:
Stanislav Fomichev:
- Don't unconditionally link the libbfd feature test to -liberty and -lz
thread-stack:
Adrian Hunter:
- Fix processing for the idle task, having a stack per cpu.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
So that the user can specify outside CFLAGS/LDFLAGS values.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Herton Krzesinski <herton@redhat.com>
Cc: Len Brown <len.brown@intel.com>
Link: http://lkml.kernel.org/r/20181212102537.25902-5-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
So user could specify outside CFLAGS/LDFLAGS values.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Herton Krzesinski <herton@redhat.com>
Cc: Len Brown <len.brown@intel.com>
Link: https://lkml.kernel.org/r/20181212102537.25902-2-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Pull turbostat utility updates for v4.21 from Len Brown:
"A couple of random fixes that were sitting in the queue."
* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: consolidate duplicate model numbers
tools/power turbostat: fix goldmont C-state limit decoding
tools/power turbostat: reduce debug output
tools/power turbosat: fix AMD APIC-id output
This script is supposed to be allowed to run with regular user
privileges if a previously captured trace is being post processed.
Commit fbe313884d (tools/power/x86/intel_pstate_tracer: Free the
trace buffer memory) introduced a bug that breaks that option.
Commit 35459105de (tools/power/x86/intel_pstate_tracer: Add
optional setting of trace buffer memory allocation) moved the code
but kept the bug.
This patch fixes the issue.
Fixes: 35459105de (tools/power/x86/intel_pstate_tracer: Add optional ...)
Signed-off-by: Doug Smythies <dsmythies@telus.net>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Often a new processor gets a new model number, but from a turbostat
point of view, it is the same as a previous model. Support duplicates
with 1-line updates, rather than error-prone scattering of model #'s.
Signed-off-by: Len Brown <len.brown@intel.com>
When the C-state limit is 8 on Goldmont, PC10 is enabled.
Previously turbostat saw this as "undefined", and thus assumed
it should not show some counters, such as pc3, pc6, pc7.
Signed-off-by: Len Brown <len.brown@intel.com>
A recent turbostat release increased topo.max_cpu_num
to make it convenient to handle sysfs bitmaps of 32-cpus.
But users, who regularly make use of "--debug", then saw a bunch of output
for cpus that were not present.
Remove that extra output by checking a cpu is online before dumping its info.
Signed-off-by: Len Brown <len.brown@intel.com>
Cc: Prarit Bhargava <prarit@redhat.com>
turbostat recently gained a feature adding APIC and X2APIC columns.
While they are disabled by-default, they are enabled with --debug
or when explicitly requested, eg.
$ sudo turbostat --quiet --show Package,Node,Core,CPU,APIC,X2APIC date
But these columns erroneously showed zeros on AMD hardware.
This patch corrects the APIC and X2APIC [sic] columns on AMD.
Signed-off-by: Len Brown <len.brown@intel.com>
Going primarily by:
https://en.wikipedia.org/wiki/List_of_Intel_Atom_microprocessors
with additional information gleaned from other related pages; notably:
- Bonnell shrink was called Saltwell
- Moorefield is the Merriefield refresh which makes it Airmont
The general naming scheme is: FAM6_ATOM_UARCH_SOCTYPE
for i in `git grep -l FAM6_ATOM` ; do
sed -i -e 's/ATOM_PINEVIEW/ATOM_BONNELL/g' \
-e 's/ATOM_LINCROFT/ATOM_BONNELL_MID/' \
-e 's/ATOM_PENWELL/ATOM_SALTWELL_MID/g' \
-e 's/ATOM_CLOVERVIEW/ATOM_SALTWELL_TABLET/g' \
-e 's/ATOM_CEDARVIEW/ATOM_SALTWELL/g' \
-e 's/ATOM_SILVERMONT1/ATOM_SILVERMONT/g' \
-e 's/ATOM_SILVERMONT2/ATOM_SILVERMONT_X/g' \
-e 's/ATOM_MERRIFIELD/ATOM_SILVERMONT_MID/g' \
-e 's/ATOM_MOOREFIELD/ATOM_AIRMONT_MID/g' \
-e 's/ATOM_DENVERTON/ATOM_GOLDMONT_X/g' \
-e 's/ATOM_GEMINI_LAKE/ATOM_GOLDMONT_PLUS/g' ${i}
done
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Vince Weaver <vincent.weaver@maine.edu>
Cc: dave.hansen@linux.intel.com
Cc: len.brown@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull turbostat utility fixes for 4.18 from Len Brown:
"Three of them are for regressions since Linux-4.17"
* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: version 18.07.27
tools/power turbostat: Read extended processor family from CPUID
tools/power turbostat: Fix logical node enumeration to allow for non-sequential physical nodes
tools/power turbostat: fix x2apic debug message output file
tools/power turbostat: fix bogus summary values
tools/power turbostat: fix -S on UP systems
tools/power turbostat: Update turbostat(8) RAPL throttling column description
This fixes the reported family on modern AMD processors (e.g. Ryzen,
which is family 0x17). Previously these processors all showed up as
family 0xf.
See the document
https://support.amd.com/TechDocs/56255_OSRR.pdf
section CPUID_Fn00000001_EAX for how to calculate the family
from the BaseFamily and ExtFamily values.
This matches the code in arch/x86/lib/cpu.c
Signed-off-by: Calvin Walton <calvin.walton@kepstin.ca>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat fails on some multi-package topologies because the logical node
enumeration assumes that the nodes are sequentially numbered,
which causes the logical numa nodes to not be enumerated, or enumerated incorrectly.
Use a more robust enumeration algorithm which allows for non-seqential physical nodes.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
This patch fixes a regression introduced in
commit 8cb48b32a5 ("tools/power turbostat: track thread ID in cpu_topology")
Turbostat uses incorrect cores number ('topo.num_cores') - its value is count
of logical CPUs, instead of count of physical cores. So it is twice as large as
it should be on a typical Intel system. For example, on a 6 core Xeon system
'topo.num_cores' is 12, and on a 52 core Xeon system 'topo.num_cores' is 104.
And interestingly, on a 68-core Knights Landing Intel system 'topo.num_cores'
is 272, because this system has 4 logical CPUs per core.
As a result, some of the turbostat calculations are incorrect. For example,
on idle 52-core Xeon system when all cores are ~99% in Core C6 (CPU%c6), the
summary (very first) line shows ~48% Core C6, while it should be ~99%.
This patch fixes the problem by fixing 'topo.num_cores' calculation.
Was:
1. Init 'thread_id' for all CPUs to -1
2. Run 'get_thread_siblings()' which sets it to 0 or 1
3. Increment 'topo.num_cores' when thread_id != -1 (bug!)
Now:
1. Init 'thread_id' for all CPUs to -1
2. Run 'get_thread_siblings()' which sets it to 0 or 1
3. Increment 'topo.num_cores' when thread_id is not 0
I did not have a chance to test this on an AMD machine, and only tested on a
couple of Intel Xeons (6 and 52 cores).
Reported-by: Vladislav Govtva <vladislav.govtva@intel.com>
Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The -S (system summary) option failed to print any data on a 1-processor system.
Reported-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Pull turbostat utility changes for 4.18-rc2 from Len Brown.
"This includes two regression fixes, plus a couple more random, but
worthy, patches."
* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
tools/power turbostat: version 18.06.20
tools/power turbostat: add the missing command line switches
tools/power turbostat: add single character tokens to help
tools/power turbostat: alphabetize the help output
tools/power turbostat: fix segfault on 'no node' machines
tools/power turbostat: add optional APIC X2APIC columns
tools/power turbostat: decode cpuid.1.HT
tools/power turbostat: fix show/hide issues resulting from mis-merge
Document the missing command line tokens in the help() function.
Signed-off-by: Nathan Ciobanu <nathan.d.ciobanu@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Improve the help() output by adding the single character
tokens (e.g -a).
Signed-off-by: Nathan Ciobanu <nathan.d.ciobanu@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Sort the command line arguments output of help() in
alphabetical order in line with other linux tools.
Signed-off-by: Nathan Ciobanu <nathan.d.ciobanu@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Running turbostat on machines that don't expose nodes
in sysfs (no /sys/bus/node) causes a segfault or a -nan
value diesplayed in the log. This is caused by
physical_node_id being reported as -1 and logical_node_id
being calculated as a negative number resulting in the new
GET_THREAD/GET_CORE returning an incorrect address.
Signed-off-by: Nathan Ciobanu <nathan.d.ciobanu@linux.intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Add APIC and X2APIC columns to the topology section.
They are disabled-by-default -- enable like so:
--debug
or
--enable APIC,X2APIC
Signed-off-by: Len Brown <len.brown@intel.com>
The --show and --hide options failed on "Node", which was listed as "Node%".
The --show and --hide options were generally fouled-up do due to come
content merges that scrambled the list of column name indexes.
Signed-off-by: Len Brown <len.brown@intel.com>
Output a Node column if there is more than one node/socket.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The previous patches have added node information to turbostat, but the
counters code does not take it into account.
Add node information from cpu_topology calculations to turbostat
counters.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Cleanup, remove num_ from num_nodes_per_pkg, num_cores_per_node, and
num_threads_per_node.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
turbostat incorrectly assumes that there is one node per package. As a
result num_cores_per_pkg is not correctly named and is actually
num_cores_per_node.
Rename num_cores_per_pkg to num_cores_per_node.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The code can be simplified if the cpu_topology *cpus tracks the thread
IDs. This removes an additional file lookup and simplifies the counter
initialization code.
Add thread ID to cpu_topology information and cleanup the counter
initialization code.
v2: prevent thread_id from being overwritten
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
The code currently assumes each package has exactly one node. This is not
the case for AMD systems and Intel systems with COD. AMD systems also
may re-enumerate each node's core IDs starting at 0 (for example, an AMD
processor may have two nodes, each with core IDs from 0 to 7). In order
to properly enumerate the cores we need to track both the physical and
logical node IDs.
Add physical_node_id to track the node ID assigned by the kernel, and
logical_node_id used by turbostat to track the nodes per package ie) a
0-based count within the package.
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Len Brown <len.brown@intel.com>
Linus Torvalds