Protected virtualization guests have to use shared pages for airq
notifier bit vectors, because the hypervisor needs to write these bits.
Let us make sure we allocate DMA memory for the notifier bit vectors by
replacing the kmem_cache with a dma_cache and kalloc() with
cio_dma_zalloc().
Signed-off-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Michael Mueller <mimu@linux.ibm.com>
Tested-by: Michael Mueller <mimu@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
As virtio-ccw devices are channel devices, we need to use the
dma area within the common I/O layer for any communication with
the hypervisor.
Note that we do not need to use that area for control blocks
directly referenced by instructions, e.g. the orb.
It handles neither QDIO in the common code, nor any device type specific
stuff (like channel programs constructed by the DASD driver).
An interesting side effect is that virtio structures are now going to
get allocated in 31 bit addressable storage.
Signed-off-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Michael Mueller <mimu@linux.ibm.com>
Tested-by: Michael Mueller <mimu@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
To support protected virtualization cio will need to make sure the
memory used for communication with the hypervisor is DMA memory.
Let us introduce one global pool for cio.
Our DMA pools are implemented as a gen_pool backed with DMA pages. The
idea is to avoid each allocation effectively wasting a page, as we
typically allocate much less than PAGE_SIZE.
Signed-off-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Sebastian Ott <sebott@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Michael Mueller <mimu@linux.ibm.com>
Tested-by: Michael Mueller <mimu@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
On s390, protected virtualization guests have to use bounced I/O
buffers. That requires some plumbing.
Let us make sure, any device that uses DMA API with direct ops correctly
is spared from the problems, that a hypervisor attempting I/O to a
non-shared page would bring.
Signed-off-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Reviewed-by: Michael Mueller <mimu@linux.ibm.com>
Tested-by: Michael Mueller <mimu@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Let's use the error value that is typically used if HW support is not
available when trying to load a module - this is also what systemd's
systemd-modules-load.service expects.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Let's use the error value that is typically used if HW support is not
available when trying to load a module - this is also what systemd's
systemd-modules-load.service expects.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Let's use the error value that is typically used if HW support is not
available when trying to load a module - this is also what systemd's
systemd-modules-load.service expects.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
systemd-modules-load.service automatically tries to load the pkey module
on systems that have MSA.
Pkey also requires the MSA3 facility and a bunch of subfunctions.
Failing with -EOPNOTSUPP makes "systemd-modules-load.service" fail on
any system that does not have all needed subfunctions. For example,
when running under QEMU TCG (but also on systems where protected keys
are disabled via the HMC).
Let's use -ENODEV, so systemd-modules-load.service properly ignores
failing to load the pkey module because of missing HW functionality.
While at it, also convert the -EOPNOTSUPP in pkey_clr2protkey() to -ENODEV.
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Adric Blake reported the following warning during suspend-resume:
Enabling non-boot CPUs ...
x86: Booting SMP configuration:
smpboot: Booting Node 0 Processor 1 APIC 0x2
unchecked MSR access error: WRMSR to 0x10f (tried to write 0x0000000000000000) \
at rIP: 0xffffffff8d267924 (native_write_msr+0x4/0x20)
Call Trace:
intel_set_tfa
intel_pmu_cpu_starting
? x86_pmu_dead_cpu
x86_pmu_starting_cpu
cpuhp_invoke_callback
? _raw_spin_lock_irqsave
notify_cpu_starting
start_secondary
secondary_startup_64
microcode: sig=0x806ea, pf=0x80, revision=0x96
microcode: updated to revision 0xb4, date = 2019-04-01
CPU1 is up
The MSR in question is MSR_TFA_RTM_FORCE_ABORT and that MSR is emulated
by microcode. The log above shows that the microcode loader callback
happens after the PMU restoration, leading to the conjecture that
because the microcode hasn't been updated yet, that MSR is not present
yet, leading to the #GP.
Add a microcode loader-specific hotplug vector which comes before
the PERF vectors and thus executes earlier and makes sure the MSR is
present.
Fixes: 400816f60c ("perf/x86/intel: Implement support for TSX Force Abort")
Reported-by: Adric Blake <promarbler14@gmail.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Cc: x86@kernel.org
Link: https://bugzilla.kernel.org/show_bug.cgi?id=203637
One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array. For example:
struct bio_map_data {
...
struct iovec iov[];
};
instance = kmalloc(sizeof(sizeof(struct bio_map_data) + sizeof(struct iovec) *
count, GFP_KERNEL);
Instead of leaving these open-coded and prone to type mistakes, we can
now use the new struct_size() helper:
instance = kmalloc(struct_size(instance, iov, count), GFP_KERNEL);
This code was detected with the help of Coccinelle.
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Make use of the struct_size() helper instead of an open-coded version
in order to avoid any potential type mistakes, in particular in the
context in which this code is being used.
So, replace the following form:
sizeof(*new_ptbl) + target * sizeof(new_ptbl->part[0])
with:
struct_size(new_ptbl, part, target)
Also, notice that variable size is unnecessary, hence it is removed.
This code was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
struct blk_rq_stat::mean is a u64 value, so use %llu
Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Andy reported that raid10 array with SSD disks has poor
read performance. Compared with raid1, RAID-1 can be 3x
faster than RAID-10 sometimes [1].
The thing is that raid10 chooses the low distance disk
for read request, however, the approach doesn't work
well for SSD device since it doesn't have spindle like
HDD, we should just read from the SSD which has less
pending IO like commit 9dedf60313 ("md/raid1: read
balance chooses idlest disk for SSD").
So this commit selects the idlest SSD disk for read if
array has none rotational disk, otherwise, read_balance
uses the previous distance priority algorithm. With the
change, the performance of raid10 gets increased largely
per Andy's test [2].
[1]. https://marc.info/?l=linux-raid&m=155915890004761&w=2
[2]. https://marc.info/?l=linux-raid&m=155990654223786&w=2
Tested-by: Andy Smith <andy@strugglers.net>
Signed-off-by: Guoqing Jiang <gqjiang@suse.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Avoiding duplicated code, since they just execute a kfree.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
One of the more common cases of allocation size calculations is finding
the size of a structure that has a zero-sized array at the end, along
with memory for some number of elements for that array. For example:
struct foo {
int stuff;
struct boo entry[];
};
instance = kmalloc(size, GFP_KERNEL);
Instead of leaving these open-coded and prone to type mistakes, we can
now use the new struct_size() helper:
instance = kmalloc(struct_size(instance, entry, count), GFP_KERNEL);
This code was detected with the help of Coccinelle.
Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This patch get rid of extra blank line and space, and
add necessary space for code.
Signed-off-by: Yufen Yu <yuyufen@huawei.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
This patch fix a spelling typo and add necessary space for code.
In addition, the patch get rid of the unnecessary 'if'.
Signed-off-by: Yufen Yu <yuyufen@huawei.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Commit c42d324099
("md: return -ENODEV if rdev has no mddev assigned") changed
rdev_attr_store to return -ENODEV when rdev->mddev is NULL, now do the
same to rdev_attr_show.
Signed-off-by: Marcos Paulo de Souza <marcos.souza.org@gmail.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
commit d5d885fd51 ("md: introduce new personality funciton start()")
splits the init job to two parts. The first part run() does the jobs that
do not require the md threads. The second part start() does the jobs that
require the md threads.
Now it just does run() in adding new journal device. It needs to do the
second part start() too.
Fixes: d5d885fd51 ("md: introduce new personality funciton start()")
Cc: stable@vger.kernel.org #v4.9+
Reported-by: Michal Soltys <soltys@ziu.info>
Signed-off-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <songliubraving@fb.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
The reg property represents the address and size on the bus that a
device lives, but for APR the parent is a rpmsg bus, which does not have
numerical addresses. Simply defining #address/#size-cells to 1 and 0,
respectively, to silence the compiler is not an appropriate solution.
Replace the use of "reg" with an APR specific property.
Reviewed-by: Srinivas Kandagatla <srinivas.kandagatla@linaro.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
It's common for the platform to replace the cache device nodes after a
migration. Since the cacheinfo code is never informed about this, it
never drops its references to the source system's cache nodes, causing
it to wind up in an inconsistent state resulting in warnings and oopses
as soon as CPU online/offline occurs after the migration, e.g.
cache for /cpus/l3-cache@3113(Unified) refers to cache for /cpus/l2-cache@200d(Unified)
WARNING: CPU: 15 PID: 86 at arch/powerpc/kernel/cacheinfo.c:176 release_cache+0x1bc/0x1d0
[...]
NIP release_cache+0x1bc/0x1d0
LR release_cache+0x1b8/0x1d0
Call Trace:
release_cache+0x1b8/0x1d0 (unreliable)
cacheinfo_cpu_offline+0x1c4/0x2c0
unregister_cpu_online+0x1b8/0x260
cpuhp_invoke_callback+0x114/0xf40
cpuhp_thread_fun+0x270/0x310
smpboot_thread_fn+0x2c8/0x390
kthread+0x1b8/0x1c0
ret_from_kernel_thread+0x5c/0x68
Using device tree notifiers won't work since we want to rebuild the
hierarchy only after all the removals and additions have occurred and
the device tree is in a consistent state. Call cacheinfo_teardown()
before processing device tree updates, and rebuild the hierarchy
afterward.
Fixes: 410bccf978 ("powerpc/pseries: Partition migration in the kernel")
Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
CPU online/offline code paths are sensitive to parts of the device
tree (various cpu node properties, cache nodes) that can be changed as
a result of a migration.
Prevent CPU hotplug while the device tree potentially is inconsistent.
Fixes: 410bccf978 ("powerpc/pseries: Partition migration in the kernel")
Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Allow external callers to force the cacheinfo code to release all its
references to cache nodes, e.g. before processing device tree updates
post-migration, and to rebuild the hierarchy afterward.
CPU online/offline must be blocked by callers; enforce this.
Fixes: 410bccf978 ("powerpc/pseries: Partition migration in the kernel")
Signed-off-by: Nathan Lynch <nathanl@linux.ibm.com>
Reviewed-by: Gautham R. Shenoy <ego@linux.vnet.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
According with Documentation/ABI/, the right tag to describe
an ABI symbol is "What:", and not "Where:".
Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix Kconfig warning and subsequent build errors that are caused
when PM_GENERIC_DOMAINS=y but CONFIG_PM is not set/enabled.
WARNING: unmet direct dependencies detected for PM_GENERIC_DOMAINS
Depends on [n]: PM [=n]
Selected by [m]:
- QCOM_AOSS_QMP [=m] && (ARCH_QCOM || COMPILE_TEST [=y]) && MAILBOX [=y] && COMMON_CLK [=y]
Fixes these build errors:
../drivers/base/power/domain.c: In function ‘genpd_queue_power_off_work’:
../drivers/base/power/domain.c:485:13: error: ‘pm_wq’ undeclared (first use in this function)
queue_work(pm_wq, &genpd->power_off_work);
../drivers/base/power/domain.c:485:13: note: each undeclared identifier is reported only once for each function it appears in
../drivers/base/power/domain.c: In function ‘genpd_dev_pm_qos_notifier’:
../drivers/base/power/domain.c:675:25: error: ‘struct dev_pm_info’ has no member named ‘ignore_children’
if (!dev || dev->power.ignore_children)
../drivers/base/power/domain.c: In function ‘rtpm_status_str’:
../drivers/base/power/domain.c:2754:16: error: ‘struct dev_pm_info’ has no member named ‘runtime_error’
if (dev->power.runtime_error)
../drivers/base/power/domain.c:2756:21: error: ‘struct dev_pm_info’ has no member named ‘disable_depth’
else if (dev->power.disable_depth)
../drivers/base/power/domain.c:2758:21: error: ‘struct dev_pm_info’ has no member named ‘runtime_status’
else if (dev->power.runtime_status < ARRAY_SIZE(status_lookup))
../drivers/base/power/domain.c:2759:31: error: ‘struct dev_pm_info’ has no member named ‘runtime_status’
p = status_lookup[dev->power.runtime_status];
../drivers/base/power/domain_governor.c: In function ‘default_suspend_ok’:
../drivers/base/power/domain_governor.c:82:17: error: ‘struct dev_pm_info’ has no member named ‘ignore_children’
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Bjorn Andersson <bjorn.andersson@sonymobile.com>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Andy Gross <andy.gross@linaro.org>
Cc: David Brown <david.brown@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Pull cgroup fixes from Tejun Heo:
"This has an unusually high density of tricky fixes:
- task_get_css() could deadlock when it races against a dying cgroup.
- cgroup.procs didn't list thread group leaders with live threads.
This could mislead readers to think that a cgroup is empty when
it's not. Fixed by making PROCS iterator include dead tasks. I made
a couple mistakes making this change and this pull request contains
a couple follow-up patches.
- When cpusets run out of online cpus, it updates cpusmasks of member
tasks in bizarre ways. Joel improved the behavior significantly"
* 'for-5.2-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cpuset: restore sanity to cpuset_cpus_allowed_fallback()
cgroup: Fix css_task_iter_advance_css_set() cset skip condition
cgroup: css_task_iter_skip()'d iterators must be advanced before accessed
cgroup: Include dying leaders with live threads in PROCS iterations
cgroup: Implement css_task_iter_skip()
cgroup: Call cgroup_release() before __exit_signal()
docs cgroups: add another example size for hugetlb
cgroup: Use css_tryget() instead of css_tryget_online() in task_get_css()
Pull drm fixes from Daniel Vetter:
"Nothing unsettling here, also not aware of anything serious still
pending.
The edid override regression fix took a bit longer since this seems to
be an area with an overabundance of bad options. But the fix we have
now seems like a good path forward.
Next week it should be back to Dave.
Summary:
- fix regression on amdgpu on SI
- fix edid override regression
- driver fixes: amdgpu, i915, mediatek, meson, panfrost
- fix writecombine for vmap in gem-shmem helper (used by panfrost)
- add more panel quirks"
* tag 'drm-fixes-2019-06-14' of git://anongit.freedesktop.org/drm/drm: (25 commits)
drm/amdgpu: return 0 by default in amdgpu_pm_load_smu_firmware
drm/amdgpu: Fix bounds checking in amdgpu_ras_is_supported()
drm: add fallback override/firmware EDID modes workaround
drm/edid: abstract override/firmware EDID retrieval
drm/i915/perf: fix whitelist on Gen10+
drm/i915/sdvo: Implement proper HDMI audio support for SDVO
drm/i915: Fix per-pixel alpha with CCS
drm/i915/dmc: protect against reading random memory
drm/i915/dsi: Use a fuzzy check for burst mode clock check
drm/amdgpu/{uvd,vcn}: fetch ring's read_ptr after alloc
drm/panfrost: Require the simple_ondemand governor
drm/panfrost: make devfreq optional again
drm/gem_shmem: Use a writecombine mapping for ->vaddr
drm: panel-orientation-quirks: Add quirk for GPD MicroPC
drm: panel-orientation-quirks: Add quirk for GPD pocket2
drm/meson: fix G12A primary plane disabling
drm/meson: fix primary plane disabling
drm/meson: fix G12A HDMI PLL settings for 4K60 1000/1001 variations
drm/mediatek: call mtk_dsi_stop() after mtk_drm_crtc_atomic_disable()
drm/mediatek: clear num_pipes when unbind driver
...
Pull gfs2 fix from Andreas Gruenbacher:
"Fix rounding error in gfs2_iomap_page_prepare"
* tag 'gfs2-v5.2.fixes2' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
gfs2: Fix rounding error in gfs2_iomap_page_prepare
Vivien Didelot says:
====================
net: dsa: use switchdev attr and obj handlers
This series reduces boilerplate in the handling of switchdev attribute and
object operations by using the switchdev_handle_* helpers, which check the
targeted devices and recurse into their lower devices.
This also brings back the ability to inspect operations targeting the bridge
device itself (where .orig_dev and .dev were originally the bridge device),
even though that is of no use yet and skipped by this series.
Changes in v2: Only VLAN and (non-host) MDB objects not directly targeting
the slave device are unsupported at the moment, so only skip these cases.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Get rid of the dsa_slave_switchdev_port_{attr_set,obj}_event functions
in favor of the switchdev_handle_port_{attr_set,obj_add,obj_del}
helpers which recurse into the lower devices of the target interface.
This has the benefit of being aware of the operations made on the
bridge device itself, where orig_dev is the bridge, and dev is the
slave. This can be used later to configure the hardware switches.
Only VLAN and (port) MDB objects not directly targeting the slave
device are unsupported at the moment, so skip this case in their
respective case statements.
Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The switchdev handle helpers make use of a device checking helper
requiring a const net_device. Make dsa_slave_dev_check compliant
to this.
Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A port may trigger operations on its dedicated CPU port, so using
cpu_dp as const will raise warnings. Make cpu_dp non const.
Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current DSA code handling switchdev objects does not recurse into
the lower devices thus is never called with an orig_dev member being
a bridge device, hence remove this useless check.
At the same time, remove the comments about the callers, which is
unlikely to be updated if the code changes and thus will be confusing.
Signed-off-by: Vivien Didelot <vivien.didelot@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eric Dumazet says:
====================
tcp: add three static keys
Recent addition of per TCP socket rx/tx cache brought
regressions for some workloads, as reported by Feng Tang.
It seems better to make them opt-in, before we adopt better
heuristics.
The last patch adds high_order_alloc_disable sysctl
to ask TCP sendmsg() to exclusively use order-0 allocations,
as mm layer has specific optimizations.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
>From linux-3.7, (commit 5640f76858 "net: use a per task frag
allocator") TCP sendmsg() has preferred using order-3 allocations.
While it gives good results for most cases, we had reports
that heavy uses of TCP over loopback were hitting a spinlock
contention in page allocations/freeing.
This commits adds a sysctl so that admins can opt-in
for order-0 allocations. Hopefully mm layer might optimize
order-3 allocations in the future since it could give us
a nice boost (see 8 lines of following benchmark)
The following benchmark shows a win when more than 8 TCP_STREAM
threads are running (56 x86 cores server in my tests)
for thr in {1..30}
do
sysctl -wq net.core.high_order_alloc_disable=0
T0=`./super_netperf $thr -H 127.0.0.1 -l 15`
sysctl -wq net.core.high_order_alloc_disable=1
T1=`./super_netperf $thr -H 127.0.0.1 -l 15`
echo $thr:$T0:$T1
done
1: 49979: 37267
2: 98745: 76286
3: 141088: 110051
4: 177414: 144772
5: 197587: 173563
6: 215377: 208448
7: 241061: 234087
8: 267155: 263373
9: 295069: 297402
10: 312393: 335213
11: 340462: 368778
12: 371366: 403954
13: 412344: 443713
14: 426617: 473580
15: 474418: 507861
16: 503261: 538539
17: 522331: 563096
18: 532409: 567084
19: 550824: 605240
20: 525493: 641988
21: 564574: 665843
22: 567349: 690868
23: 583846: 710917
24: 588715: 736306
25: 603212: 763494
26: 604083: 792654
27: 602241: 796450
28: 604291: 797993
29: 611610: 833249
30: 577356: 841062
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Feng Tang reported a performance regression after introduction
of per TCP socket tx/rx caches, for TCP over loopback (netperf)
There is high chance the regression is caused by a change on
how well the 32 KB per-thread page (current->task_frag) can
be recycled, and lack of pcp caches for order-3 pages.
I could not reproduce the regression myself, cpus all being
spinning on the mm spinlocks for page allocs/freeing, regardless
of enabling or disabling the per tcp socket caches.
It seems best to disable the feature by default, and let
admins enabling it.
MM layer either needs to provide scalable order-3 pages
allocations, or could attempt a trylock on zone->lock if
the caller only attempts to get a high-order page and is
able to fallback to order-0 ones in case of pressure.
Tests run on a 56 cores host (112 hyper threads)
- 35.49% netperf [kernel.vmlinux] [k] queued_spin_lock_slowpath
- 35.49% queued_spin_lock_slowpath
- 18.18% get_page_from_freelist
- __alloc_pages_nodemask
- 18.18% alloc_pages_current
skb_page_frag_refill
sk_page_frag_refill
tcp_sendmsg_locked
tcp_sendmsg
inet_sendmsg
sock_sendmsg
__sys_sendto
__x64_sys_sendto
do_syscall_64
entry_SYSCALL_64_after_hwframe
__libc_send
+ 17.31% __free_pages_ok
+ 31.43% swapper [kernel.vmlinux] [k] intel_idle
+ 9.12% netperf [kernel.vmlinux] [k] copy_user_enhanced_fast_string
+ 6.53% netserver [kernel.vmlinux] [k] copy_user_enhanced_fast_string
+ 0.69% netserver [kernel.vmlinux] [k] queued_spin_lock_slowpath
+ 0.68% netperf [kernel.vmlinux] [k] skb_release_data
+ 0.52% netperf [kernel.vmlinux] [k] tcp_sendmsg_locked
0.46% netperf [kernel.vmlinux] [k] _raw_spin_lock_irqsave
Fixes: 472c2e07ee ("tcp: add one skb cache for tx")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Feng Tang <feng.tang@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of relying on rps_needed, it is safer to use a separate
static key, since we do not want to enable TCP rx_skb_cache
by default. This feature can cause huge increase of memory
usage on hosts with millions of sockets.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Convert proc_dointvec_minmax_bpf_stats() into a more generic
helper, since we are going to use jump labels more often.
Note that sysctl_bpf_stats_enabled is removed, since
it is no longer needed/used.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jeff Kirsher says:
====================
40GbE Intel Wired LAN Driver Updates 2019-06-14
This series contains updates to i40e only.
Aleksandr adds stub functions for Energy Efficient Ethernet (EEE) to
currently report that it is not supported in i40e. Fixed up the Link
Layer Detection Protocol (LLDP) code to ensure we do not set the LLDP
flag too early before we ensure that we have a successful start. This
also will prevent needles restarting of the device if LLDP did not
change its state with an unsuccessful start.
Piotr bumps up the amount of VLANs that an untrusted VF can implement,
from 8 VLANs to 16. Adds checks to the Virtual Embedded Bridge (VEB)
and channel arrays so access does not exceed the boundary and ensure the
index is below the maximum. Fixed an issue in the driver where we were
not checking the response from the LLDP flag and were returned success
no matter what the value of the response was.
Mitch fixes a variable counter, which can be negative in value so make
it an integer instead of an unsigned-integer.
Doug improves the admin queue log granularity by making it possible to
log only the admin queue descriptors without the entire admin queue
message buffers.
Sergey fixes up the virtchnl code by removing duplicate checks, ensure
the variable type is correct when comparing integers, enhance error and
warning messages to include useful information.
Adam fixes a potential kernel panic when the i40e driver was being bound
to a non-i40e port by adding a check on the BAR size to ensure it is
large enough by reading the highest register.
Jake fixes a statistics error in the "transmit errors" stat, which was
being calculated twice.
Gustavo A. R. Silva adds a fall-through code comment to help with
compiler checks.
v2: Fixed the return values wrapped in parenthesis in patch 8 and
cleaned up the commit message in patch 12 so the Gustavo does
not repeat himself.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This was originally passed through to the VRF logic in compute_score().
But that logic has now been replaced by udp_sk_bound_dev_eq() and so
this code is no longer used or needed.
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Originally this was used by the VRF logic in compute_score(), but that
was later replaced by udp_sk_bound_dev_eq() and the parameter became
unused.
Note this change adds an 'unused variable' compiler warning that will be
removed in the next patch (I've split the removal in two to make review
slightly easier).
Signed-off-by: Tim Beale <timbeale@catalyst.net.nz>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If we want to set a EDT time for the skb we want to send
via ip_send_unicast_reply(), we have to pass a new parameter
and initialize ipc.sockc.transmit_time with it.
This fixes the EDT time for ACK/RST packets sent on behalf of
a TIME_WAIT socket.
Fixes: a842fe1425 ("tcp: add optional per socket transmit delay")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pointer members of an object with static storage duration, if not
explicitly initialized, will be initialized to a NULL pointer. The
net namespace API checks if this pointer is not NULL before using it,
it are safe to remove the function.
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jakub Kicinski says:
====================
nfp: flower: loosen L4 checks and add extack to flower offload
Pieter says:
This set allows the offload of filters that make use of an unknown
ip protocol, given that layer 4 is being wildcarded. The set then
aims to make use of extack messaging for flower offloads. It adds
about 70 extack messages to the driver.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
