Problem and motivation: Once a breakpoint perf event (PERF_TYPE_BREAKPOINT)
is created, there is no flexibility to change the breakpoint type
(bp_type), breakpoint address (bp_addr), or breakpoint length (bp_len). The
only option is to close the perf event and configure a new breakpoint
event. This inflexibility has a significant performance overhead. For
example, sampling-based, lightweight performance profilers (and also
concurrency bug detection tools), monitor different addresses for a short
duration using PERF_TYPE_BREAKPOINT and change the address (bp_addr) to
another address or change the kind of breakpoint (bp_type) from "write" to
a "read" or vice-versa or change the length (bp_len) of the address being
monitored. The cost of these modifications is prohibitive since it involves
unmapping the circular buffer associated with the perf event, closing the
perf event, opening another perf event and mmaping another circular buffer.
Solution: The new ioctl flag for perf events,
PERF_EVENT_IOC_MODIFY_ATTRIBUTES, introduced in this patch takes a pointer
to a struct perf_event_attr as an argument to update an old breakpoint
event with new address, type, and size. This facility allows retaining a
previous mmaped perf events ring buffer and avoids having to close and
reopen another perf event.
This patch supports only changing PERF_TYPE_BREAKPOINT event type; future
implementations can extend this feature. The patch replicates some of its
functionality of modify_user_hw_breakpoint() in
kernel/events/hw_breakpoint.c. modify_user_hw_breakpoint cannot be called
directly since perf_event_ctx_lock() is already held in _perf_ioctl().
Evidence: Experiments show that the baseline (not able to modify an already
created breakpoint) costs an order of magnitude (~10x) more than the
suggested optimization (having the ability to dynamically modifying a
configured breakpoint via ioctl). When the breakpoints typically do not
trap, the speedup due to the suggested optimization is ~10x; even when the
breakpoints always trap, the speedup is ~4x due to the suggested
optimization.
Testing: tests posted at
https://github.com/linux-contrib/perf_event_modify_bp demonstrate the
performance significance of this patch. Tests also check the functional
correctness of the patch.
Signed-off-by: Milind Chabbi <chabbi.milind@gmail.com>
[ Using modify_user_hw_breakpoint_check function. ]
[ Reformated PERF_EVENT_IOC_*, so the values are all in one column. ]
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Hari Bathini <hbathini@linux.vnet.ibm.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Oleg Nesterov <onestero@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sukadev Bhattiprolu <sukadev@linux.vnet.ibm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Will Deacon <will.deacon@arm.com>
Link: http://lkml.kernel.org/r/20180312134548.31532-8-jolsa@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch supports to recognize hot file extension in f2fs, so that we
can allocate proper hot segment location for its data, which can lead to
better hot/cold seperation in filesystem.
In addition, we changes a bit on query/add/del operation method for
extension_list sysfs entry as below:
- Query: cat /sys/fs/f2fs/<disk>/extension_list
- Add: echo 'extension' > /sys/fs/f2fs/<disk>/extension_list
- Del: echo '!extension' > /sys/fs/f2fs/<disk>/extension_list
- Add: echo '[h/c]extension' > /sys/fs/f2fs/<disk>/extension_list
- Del: echo '[h/c]!extension' > /sys/fs/f2fs/<disk>/extension_list
- [h] means add/del hot file extension
- [c] means add/del cold file extension
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Previously, we will store all nat version bitmap in checkpoint pack block,
so our total node entry number has a limitation which caused total node
number can not exceed (3900 * 8) block * 455 node/block = 14196000. So
that once user wants to create more nodes in large size image, it becomes
a bottleneck, that's unreasonable.
This patch detects the new layout of nat/sit version bitmap in image in
order to enable supporting large nat bitmap.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Previous dentry page uses highmem, which will cause panic in platforms
using highmem (such as arm), since the address space of dentry pages
from highmem directly goes into the decryption path via the function
fscrypt_fname_disk_to_usr. But sg_init_one assumes the address is not
from highmem, and then cause panic since it doesn't call kmap_high but
kunmap_high is triggered at the end. To fix this problem in a simple
way, this patch avoids to put dentry page in pagecache into highmem.
Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
[Jaegeuk Kim: fix coding style]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
When a divider clock has CLK_DIVIDER_READ_ONLY set, it means that the
register shall be left un-touched, but it does not mean the clock
should stop rate propagation if CLK_SET_RATE_PARENT is set
This is properly handled in qcom clk-regmap-divider but it was not in
the generic divider
To fix this situation, introduce a new helper function
divider_ro_round_rate, on the same model as divider_round_rate.
Fixes: e6d5e7d90b ("clk-divider: Fix READ_ONLY when divider > 1")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Tested-By: David Lechner <david@lechnology.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
The mux documentation mentions the non-existing parameter width instead
of mask, so just sed this.
The table field is missing in the documentation of clk_mux.
Add a small blurb explaining what it is
Fixes: 9d9f78ed9a ("clk: basic clock hardware types")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Add helper functions for the translation between parent index and
register value in the generic multiplexer function. The purpose of
this change is avoid duplicating the code in other clock providers,
using the same generic logic.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Export clk_div_mask() in clk-provider header so every clock providers
derived from the generic clock divider may share the definition instead
of redefining it.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Michael Turquette <mturquette@baylibre.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Pablo Neira Ayuso says:
====================
Netfilter fixes for net
The following patchset contains Netfilter fixes for your net tree, they are:
1) Fixed hashtable representation doesn't support timeout flag, skip it
otherwise rules to add elements from the packet fail bogusly fail with
EOPNOTSUPP.
2) Fix bogus error with 32-bits ebtables userspace and 64-bits kernel,
patch from Florian Westphal.
3) Sanitize proc names in several x_tables extensions, also from Florian.
4) Add sanitization to ebt_among wormhash logic, from Florian.
5) Missing release of hook array in flowtable.
====================
This flag was added by fe0f07d08e ("direct-io: only inc/deci
inode->i_dio_count for file systems") as means to optimise the atomic
modificaiton of the variable for blockdevices. However with the advent
of 542ff7bf18 ("block: new direct I/O implementation") it became
unused. So let's remove it.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This flag was added by 6039257378 ("direct-io: add flag to allow aio
writes beyond i_size") to support XFS. However, with the rework of
XFS' DIO's path to use iomap in acdda3aae1 ("xfs: use iomap_dio_rw")
it became redundant. So let's remove it.
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Now when using 'ss' in iproute, kernel would try to load all _diag
modules, which also causes corresponding family and proto modules
to be loaded as well due to module dependencies.
Like after running 'ss', sctp, dccp, af_packet (if it works as a module)
would be loaded.
For example:
$ lsmod|grep sctp
$ ss
$ lsmod|grep sctp
sctp_diag 16384 0
sctp 323584 5 sctp_diag
inet_diag 24576 4 raw_diag,tcp_diag,sctp_diag,udp_diag
libcrc32c 16384 3 nf_conntrack,nf_nat,sctp
As these family and proto modules are loaded unintentionally, it
could cause some problems, like:
- Some debug tools use 'ss' to collect the socket info, which loads all
those diag and family and protocol modules. It's noisy for identifying
issues.
- Users usually expect to drop sctp init packet silently when they
have no sense of sctp protocol instead of sending abort back.
- It wastes resources (especially with multiple netns), and SCTP module
can't be unloaded once it's loaded.
...
In short, it's really inappropriate to have these family and proto
modules loaded unexpectedly when just doing debugging with inet_diag.
This patch is to introduce sock_load_diag_module() where it loads
the _diag module only when it's corresponding family or proto has
been already registered.
Note that we can't just load _diag module without the family or
proto loaded, as some symbols used in _diag module are from the
family or proto module.
v1->v2:
- move inet proto check to inet_diag to avoid a compiling err.
v2->v3:
- define sock_load_diag_module in sock.c and export one symbol
only.
- improve the changelog.
Reported-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In 664fcf123a (net: phy: Threaded interrupts allow some simplification)
the phy_interrupt system was changed to use a traditional threaded
interrupt scheme instead of a workqueue approach.
With this change, the phy status check moved into phy_change, which
did not report back to the caller whether or not the interrupt was
handled. This means that, in the case of a shared phy interrupt,
only the first phydev's interrupt registers are checked (since
phy_interrupt() would always return IRQ_HANDLED). This leads to
interrupt storms when it is a secondary device that's actually the
interrupt source.
Signed-off-by: Brad Mouring <brad.mouring@ni.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull "This is the pxa changes for v4.17 cycle" from Robert Jarzmik:
- minor changes for property API
- clock API fix for ULPI driver warning
It exceptionally contains a merge from the mtd tree from Boris
to prevent any merge conflicts in the PXA tree.
* tag 'pxa-for-4.17' of https://github.com/rjarzmik/linux:
ARM: pxa/raumfeld: use PROPERTY_ENTRY_U32() directly
ARM: pxa: ulpi: fix ulpi timeout and slowpath warn
ARM: pxa: cm-x300: remove inline directive
ARM: pxa: fix static checker warning in pxa3xx-ulpi
MAINTAINERS: remove entry for deleted pxa3xx_nand driver
arm: dts: pxa: use reworked NAND controller driver
dt-bindings: mtd: remove pxa3xx NAND controller documentation
mtd: nand: remove useless fields from pxa3xx NAND platform data
mtd: nand: remove deprecated pxa3xx_nand driver
mtd: nand: use Marvell reworked NAND controller driver with all platforms
This adds two new disk triggers for triggering on reads
and writes respectively, named "disk-read" and "disk-write".
The use case comes from working on the D-Link DNS-313 NAS
box. This features an RGB LED for disk activity. with
these two triggers I can couple the green LED to read
activity and the red LED to write activity, which gives
the appropriate user feedback about what is happening
on the disk. When tested it gave exactly the feedback
desired.
The in-kernel interface is simply changed to pass a bool
indicating if the activity is write activity and update
each trigger (and the composite "disk-activity" trigger)
depending on what is passed in.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Reviewed-by: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Acked-by: Pavel Machek <pavel@ucw.cz>
Acked-by: Tejun Heo <tj@kernel.org>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Jacek Anaszewski <jacek.anaszewski@gmail.com>
Pull RCU updates from Paul E. McKenney:
- Miscellaneous fixes, perhaps most notably removing obsolete
code whose only purpose in life was to gather information for
the now-removed RCU debugfs facility. Other notable changes
include removing NO_HZ_FULL_ALL in favor of the nohz_full kernel
boot parameter, minor optimizations for expedited grace periods,
some added tracing, creating an RCU-specific workqueue using Tejun's
new WQ_MEM_RECLAIM flag, and several cleanups to code and comments.
- SRCU cleanups and optimizations.
- Torture-test updates, perhaps most notably the adding of ARMv8
support, but also including numerous cleanups and usability fixes.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The ring-buffer code has recusion protection in case tracing ends up tracing
itself, the ring-buffer will detect that it was called at the same context
(normal, softirq, interrupt or NMI), and not continue to record the event.
With the histogram synthetic events, they are called while tracing another
event at the same context. The recusion protection triggers because it
detects tracing at the same context and stops it.
Add ring_buffer_nest_start() and ring_buffer_nest_end() that will notify the
ring buffer that a trace is about to happen within another trace and that it
is intended, and not to trigger the recursion blocking.
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
RINGBUF_TYPE_TIME_STAMP is defined but not used, and from what I can
gather was reserved for something like an absolute timestamp feature
for the ring buffer, if not a complete replacement of the current
time_delta scheme.
This code redefines RINGBUF_TYPE_TIME_STAMP to implement absolute time
stamps. Another way to look at it is that it essentially forces
extended time_deltas for all events.
The motivation for doing this is to enable time_deltas that aren't
dependent on previous events in the ring buffer, making it feasible to
use the ring_buffer_event timetamps in a more random-access way, for
purposes other than serial event printing.
To set/reset this mode, use tracing_set_timestamp_abs() from the
previous interface patch.
Link: http://lkml.kernel.org/r/477b362dba1ce7fab9889a1a8e885a62c472f041.1516069914.git.tom.zanussi@linux.intel.com
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
When the length of the NTP tick changes significantly, e.g. when an
NTP/PTP application is correcting the initial offset of the clock, a
large value may accumulate in the NTP error before the multiplier
converges to the correct value. It may then take a very long time (hours
or even days) before the error is corrected. This causes the clock to
have an unstable frequency offset, which has a negative impact on the
stability of synchronization with precise time sources (e.g. NTP/PTP
using hardware timestamping or the PTP KVM clock).
Use division to determine the correct multiplier directly from the NTP
tick length and replace the iterative approach. This removes the last
major source of the NTP error. The only remaining source is now limited
resolution of the multiplier, which is corrected by adding 1 to the
multiplier when the system clock is behind the NTP time.
Signed-off-by: Miroslav Lichvar <mlichvar@redhat.com>
Signed-off-by: John Stultz <john.stultz@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Cc: Stephen Boyd <stephen.boyd@linaro.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1520620971-9567-3-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull PCI fixes from Bjorn Helgaas:
- fix sparc build issue when OF_IRQ not enabled (Guenter Roeck)
- fix enumeration of devices below switches on DesignWare-based
controllers (Koen Vandeputte)
* tag 'pci-v4.16-fixes-3' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
PCI: dwc: Fix enumeration end when reaching root subordinate
PCI: Move of_irq_parse_and_map_pci() declaration under OF_IRQ
Some network devices - notably ipvlan slave - are not compatible with
any kind of rx_handler. Currently the hook can be installed but any
configuration (bridge, bond, macsec, ...) is nonfunctional.
This change allocates a priv_flag bit to mark such devices and explicitly
forbid installing a rx_handler if such bit is set. The new bit is used
by ipvlan slave device.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the new PHY wrapper in place we can now handle multiple PHYs.
Remove the code which handles only one generic PHY as this is now
covered (with support for multiple PHYs as well as suspend/resume
support) by the new PHY wrapper.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Tested-by: Neil Armstrong <narmstrong@baylibre.con>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This integrates the PHY wrapper into the core hcd infrastructure.
Multiple PHYs which are part of the HCD's device tree node are now
managed (= powered on/off when needed), by the new usb_phy_roothub code.
Suspend and resume is also supported, however not for
runtime/auto-suspend (which is triggered for example when no devices are
connected to the USB bus). This is needed on some SoCs (for example
Amlogic Meson GXL) because if the PHYs are disabled during auto-suspend
then devices which are plugged in afterwards are not seen by the host.
One example where this is required is the Amlogic GXL and GXM SoCs:
They are using a dwc3 USB controller with up to three ports enabled on
the internal roothub. Each port has it's own PHY which must be enabled
(if one of the PHYs is left disabled then none of the USB ports works at
all).
The new logic works on the Amlogic GXL and GXM SoCs because the dwc3
driver internally creates a xhci-hcd which then registers a HCD which
then triggers our new PHY wrapper.
USB controller drivers can opt out of this by setting
"skip_phy_initialization" in struct usb_hcd to true. This is identical
to how it works for a single USB PHY, so the "multiple PHY" handling is
disabled for drivers that opted out of the management logic of a single
PHY.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Acked-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
Tested-by: Yixun Lan <yixun.lan@amlogic.com>
Tested-by: Neil Armstrong <narmstrong@baylibre.con>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The USB HCD core driver parses the device-tree node for "phys" and
"usb-phys" properties. It also manages the power state of these PHYs
automatically.
However, drivers may opt-out of this behavior by setting "phy" or
"usb_phy" in struct usb_hcd to a non-null value. An example where this
is required is the "Qualcomm USB2 controller", implemented by the
chipidea driver. The hardware requires that the PHY is only powered on
after the "reset completed" event from the controller is received.
A follow-up patch will allow the USB HCD core driver to manage more than
one PHY. Add a new "skip_phy_initialization" bitflag to struct usb_hcd
so drivers can opt-out of any PHY management provided by the USB HCD
core driver.
This also updates the existing drivers so they use the new flag if they
want to opt out of the PHY management provided by the USB HCD core
driver. This means that for these drivers the new "multiple PHY"
handling (which will be added in a follow-up patch) will be disabled as
well.
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Acked-by: Peter Chen <peter.chen@nxp.com>
Tested-by: Neil Armstrong <narmstrong@baylibre.con>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After commit fc72d1d54d ("tuntap: XDP transmission"), we can
actually queueing XDP pointers in the pointer ring, so we should
examine the pointer type before freeing the pointer.
Fixes: fc72d1d54d ("tuntap: XDP transmission")
Reported-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.
These tunnels are also automatically created when a new network
namespace is created, at a great cost.
In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)
Add a new sysctl so that we can opt-out from this automatic creation.
Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.
Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
(for j in `seq 1 100` ; do unshare -n /bin/true >/dev/null ; done) &
done
wait
lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m37.521s
user 0m0.886s
sys 7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh
real 0m4.761s
user 0m0.851s
sys 1m8.343s
lpk43:~#
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This new security level works so that it creates one PCIe tunnel to the
connected Thunderbolt dock, removing PCIe links downstream of the dock.
This leaves only the internal USB controller visible.
Display Port tunnels are created normally.
While there make sure security sysfs attribute returns "unknown" for any
future security level.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Preboot ACL is a mechanism that allows connecting Thunderbolt devices
boot time in more secure way than the legacy Thunderbolt boot support.
As with the legacy boot option, this also needs to be enabled from the
BIOS before booting is allowed. Difference to the legacy mode is that
the userspace software explicitly adds device UUIDs by sending a special
message to the ICM firmware. Only the devices listed in the boot ACL are
connected automatically during the boot. This works in both "user" and
"secure" security levels.
We implement this in Linux by exposing a new sysfs attribute (boot_acl)
below each Thunderbolt domain. The userspace software can then update
the full list as needed.
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
This is needed by the new ICM interface to find xdomains by route string
instead of link and depth.
While there update existing tb_xdomain_find_* functions to use
tb_xdomain_get() instead of open-coding the same.
Signed-off-by: Radion Mirchevsky <radion.mirchevsky@intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
