The numa_node field of the tag_set struct has to be explicitly
initialized, otherwise it stays as 0, which is a valid numa node id and
cause memory allocation failure if node 0 is offline.
Acked-by: Sebastian Ott <sebott@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
The numa_node field of the tag_set struct has to be explicitly
initialized, otherwise it stays as 0, which is a valid numa node id and
cause memory allocation failure if node 0 is offline.
Acked-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Reduce the default values for the number of hardware queues and queue depth
to significantly reduce the memory footprint of a DASD device.
The memory consumption per DASD device reduces from approximately 40MB to
approximately 1.5MB.
This is necessary to build systems with a large number of DASD devices and
a reasonable amount of memory.
Performance measurements showed that good performance results are possible
with the new default values even on systems with lots of CPUs and lots of
alias devices.
Fixes: e443343e50 ("s390/dasd: blk-mq conversion")
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Dan Carpenter reported an integer underflow issue in the rtl8188eu driver.
This is also needed for the length (signed integer) in rtl8723bs, as it is
later converted to an unsigned integer and used in a memcpy operation.
Original issue is at https://patchwork.kernel.org/patch/9796371/
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Murray McAllister <murray.mcallister@insomniasec.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The dev section was opening curly bracket, but not adding ident, which
resulted in two times "}," after each other with same indentation. Add
ident at the right place fixes this problem.
This formatting issue is not detectable by checkpatch.pl
Signed-off-by: Radek Dostál <radek.dostal@streamunlimited.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Write multiple single line comments as block comments to
follow kernel coding style and improve readability.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The return type of a function should be on the same line as the
definition.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reorder the cases of a switch statement to be in ascending order.
Remove unrequired break from default case.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Simplify if else statement to a single function call
by passing the variable.
Suggested-by: Andreas Schwab <schwab@linux-m68k.org>
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use if(x) and if(!x) instead of comparsion to true/false.
Reported by checkpatch.
Remove unrequired braces from single if else statement.
Add missing space after else: else{ -> else {
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The file includes the file rtl819x_HT.h, which has already been included by
the previously included file ieee80211.h
Signed-off-by: John Whitmore <johnfwhitmore@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Put data to skb, decrypt with lib80211_crypt_wep, and place back to tx buffer.
Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Whether any of the conditions is true or not, the return variable
is always set to rtw_resume_process(padapter). Replace the if else
construct with a single call to rtw_resume_process(). Also remove
the now unused local variable pwrpriv.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Remove unnecessary parentheses as reported by checkpatch.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix line over 80 characters by replacing tab with space.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The header rtw_ioctl_rtl.h is not used anywhere.
Running 'grep -r rtw_ioctl_rtl *' from kernel root
directory returns nothing, remove the file.
Signed-off-by: Michael Straube <straube.linux@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
gcc 8.1.0 generates the following warnings.
drivers/staging/speakup/kobjects.c: In function 'punc_store':
drivers/staging/speakup/kobjects.c:522:2: warning:
'strncpy' output truncated before terminating nul
copying as many bytes from a string as its length
drivers/staging/speakup/kobjects.c:504:6: note: length computed here
drivers/staging/speakup/kobjects.c: In function 'synth_store':
drivers/staging/speakup/kobjects.c:391:2: warning:
'strncpy' output truncated before terminating nul
copying as many bytes from a string as its length
drivers/staging/speakup/kobjects.c:388:8: note: length computed here
Using strncpy() is indeed less than perfect since the length of data to
be copied has already been determined with strlen(). Replace strncpy()
with memcpy() to address the warning and optimize the code a little.
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When this was introduced in
commit a519435a96
Author: Christian König <christian.koenig@amd.com>
Date: Tue Oct 20 16:34:16 2015 +0200
dma-buf/fence: add fence_wait_any_timeout function v2
there was a restriction added that this only works if the dma-fence
uses the dma_fence_default_wait hook. Which works for amdgpu, which is
the only caller. Well, until you share some buffers with e.g. i915,
then you get an -EINVAL.
But there's really no reason for this, because all drivers must
support callbacks. The special ->wait hook is only as an optimization;
if the driver needs to create a worker thread for an active callback,
then it can avoid to do that if it knows that there's a process
context available already. So ->wait is just an optimization, just
using the logic in dma_fence_default_wait() should work for all
drivers.
Let's remove this restriction.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Cc: Christian König <christian.koenig@amd.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180503142603.28513-4-daniel.vetter@ffwll.ch
Many drivers have a trivial implementation for ->enable_signaling.
Let's make it optional by assuming that signalling is already
available when the callback isn't present.
v2: Don't do the trick to set the ENABLE_SIGNAL_BIT
unconditionally, it results in an expensive spinlock take for
everyone. Instead just check if the callback is present. Suggested by
Maarten.
Also move misplaced kerneldoc hunk to the right patch.
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com> (v1)
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Cc: Sumit Semwal <sumit.semwal@linaro.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: linux-media@vger.kernel.org
Cc: linaro-mm-sig@lists.linaro.org
Link: https://patchwork.freedesktop.org/patch/msgid/20180504141034.27727-1-daniel.vetter@ffwll.ch
Noticed while I was typing docs. Entirely unused.
v2: Remove reference in @timeline_value_str too. While at it clarify
why timeline_value_str has a fence parameter - we don't have an
explicit timeline structure unfortunately.
Cc: Eric Anholt <eric@anholt.net>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Christian König <christian.koenig@amd.com> (v1)
Cc: Christian König <christian.koenig@amd.com> (v1)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20180503142603.28513-2-daniel.vetter@ffwll.ch
Since the following commit:
cd77849a69 ("objtool: Fix GCC 8 cold subfunction detection for aliased functions")
... if the kernel is built with EXTRA_CFLAGS='-fno-reorder-functions',
objtool can get stuck in an infinite loop.
That flag causes the new GCC 8 cold subfunctions to be placed in .text
instead of .text.unlikely. But it also has an unfortunate quirk: in the
symbol table, the subfunction (e.g., nmi_panic.cold.7) is nested inside
the parent (nmi_panic).
That function overlap confuses objtool, and causes it to get into an
infinite loop in next_insn_same_func(). Here's Allan's description of
the loop:
"Objtool iterates through the instructions in nmi_panic using
next_insn_same_func. Once it reaches the end of nmi_panic at 0x534 it
jumps to 0x528 as that's the start of nmi_panic.cold.7. However, since
the instructions starting at 0x528 are still associated with nmi_panic
objtool will get stuck in a loop, continually jumping back to 0x528
after reaching 0x534."
Fix it by shortening the length of the parent function so that the
functions no longer overlap.
Reported-and-analyzed-by: Allan Xavier <allan.x.xavier@oracle.com>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Allan Xavier <allan.x.xavier@oracle.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/9e704c52bee651129b036be14feda317ae5606ae.1530136978.git.jpoimboe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Some Micron NAND chips (MT29F1G08ABAFAWP-ITE:F) report 00 00 for the
revision number field of the ONFI parameter page. Rather than rejecting
these outright assume ONFI version 1.0 if the revision number is 00 00.
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Add defines for the ONFI version bits and use them in
nand_flash_detect_onfi().
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
This is called after the ONFI parameter page checksum is verified
and allows us to override the contents of the parameter page.
Suggested-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
>From the controllers point of view this is the same as no or
software only ECC.
Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Add support for the NAND flash controller found on NVIDIA
Tegra 2 SoCs. This implementation does not make use of the
command queue feature. Regular operations using ->exec_op()
use PIO mode for data transfers. Raw, ECC and OOB read/writes
make use of the DMA mode for data transfer.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
This adds the devicetree binding for the Tegra 2 NAND flash
controller.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Allow to define a NAND chip as a boot device. This can be helpful
for the selection of the ECC algorithm and strength in case the boot
ROM supports only a subset of controller provided options.
Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The Gasket (Google ASIC Software, Kernel Extensions, and Tools) kernel
framework is a generic, flexible system that supports thin kernel
drivers. Gasket kernel drivers are expected to handle opening and
closing devices, mmap'ing BAR space as requested, a small selection of
ioctls, and handling page table translation (covered below). Any other
functions should be handled by userspace code.
The Gasket common module is not enough to run a device. In order to
customize the Gasket code for a given piece of hardware, a device
specific module must be created. At a minimum, this module must define a
struct gasket_driver_desc containing the device-specific data for use by
the framework; in addition, the module must declare an __init function
that calls gasket_register_device with the module's gasket_driver_desc
struct. Finally, the driver must define an exit function that calls
gasket_unregister_device with the module's gasket_driver_desc struct.
One of the core assumptions of the Gasket framework is that precisely
one process is allowed to have an open write handle to the device node
at any given time. (That process may, once it has one write handle, open
any number of additional write handles.) This is accomplished by
tracking open and close data for each driver instance.
Signed-off-by: Rob Springer <rspringer@google.com>
Signed-off-by: John Joseph <jnjoseph@google.com>
Signed-off-by: Simon Que <sque@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add Reed-Solomon (RS) to the enumeration of ECC algorithms.
Signed-off-by: Stefan Agner <stefan@agner.ch>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
The datasheet of the H27UCG8T2BTR states that this chip has a page size
of "16,384 + 1,280(Spare) bytes". The description of the "4th Byte of
Device Identifier Description" indicates that bits 6, 3 and 2 are
encoding the "Redundant Area Size / 8KB", where 640 bytes is a value of
0x6 (110 in binary notation).
hynix_nand_extract_oobsize decodes an OOB size of 640 bytes for this
chip. Kernel boot log extract before this patch:
nand: Could not find valid ONFI parameter page; aborting
nand: device found, Manufacturer ID: 0xad, Chip ID: 0xde
nand: Hynix NAND 8GiB 3,3V 8-bit
nand: 8192 MiB, MLC, erase size: 4096 KiB, page size: 16384,
OOB size: 640
However, based on the description in the datasheet we need to multiply
the OOB size with 2, because it's "640 spare bytes per 8192 bytes page
size" and this NAND chip has a page size of 16384 (= 2 * 8192). After
this patch the kernel boot log reports:
nand: Could not find valid ONFI parameter page; aborting
nand: device found, Manufacturer ID: 0xad, Chip ID: 0xde
nand: Hynix NAND 8GiB 3,3V 8-bit
nand: 8192 MiB, MLC, erase size: 4096 KiB, page size: 16384,
OOB size: 1280
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Even if we can't update ecc_stats.corrected with an accurate value we
should at least increase the number of bitflips so that MTD users can
know that there was some bitflips.
Just add chip->ecc.strength to mtd->ecc_stats.corrected which should
account for the worst case situation.
Signed-off-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
On a sama5d31 with a Full-HD dual LVDS panel (132MHz pixel clock) NAND
flash accesses have a tendency to cause display disturbances. Add a
module param to disable DMA from the NAND controller, since that fixes
the display problem for me.
Signed-off-by: Peter Rosin <peda@axentia.se>
Reviewed-by: Boris Brezillon <boris.brezillon@bootlin.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
It's been four years since this was added. In the interim, skein has
not seen any mainstream adoption. Same with the threefish block cipher
upon which it's based.
In the discussion over which hash algorithm will replace SHA1 in git,
it's not one of the contenders.
There's absolutely no reason to think that there is anything wrong with
Skein or Threefish. The only reason for this removal is a lack of
adoption.
If a real user comes forward, I'd be happy to assist with integrating
this code into mainline.
Signed-off-by: Jason Cooper <jason@lakedaemon.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Add missing include of sizes.h.
drivers/rpmsg/qcom_smd.c: In function ‘qcom_smd_channel_open’:
drivers/rpmsg/qcom_smd.c:809:36: error: ‘SZ_4K’ undeclared (first use in this function)
bb_size = min(channel->fifo_size, SZ_4K);
^~~~~
Signed-off-by: Niklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
changed gvt display transcode DDI mode from DP_SST to
DVI to address below calltrace issue during guest booting
up which is caused by zero dotclock initial value with DP_SST
mode. transcode DVI mode emulation also align with native with DP
connection.
[drm:drm_calc_timestamping_constants]
ERROR crtc 41: Can't calculate constants, dotclock = 0!
WARNING: at drivers/gpu/drm/drm_vblank.c:620
drm_calc_vbltimestamp_from_scanoutpos
Call Trace:
? drm_calc_timestamping_constants+0x144/0x150 [drm]
drm_get_last_vbltimestamp+0x54/0x90 [drm]
drm_reset_vblank_timestamp+0x59/0xd0 [drm]
drm_crtc_vblank_on+0x7b/0xd0 [drm]
intel_modeset_setup_hw_state+0xb67/0xfd0 [i915]
? gen2_read32+0x110/0x110 [i915]
? drm_modeset_lock+0x30/0xa0 [drm]
intel_modeset_init+0x794/0x19d0 [i915]
? intel_setup_gmbus+0x232/0x2e0 [i915]
i915_driver_load+0xb4a/0xf40 [i915]
Signed-off-by: Xiaolin Zhang <xiaolin.zhang@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
when guest writes ggtt entries, it could write 8 bytes a time if
gtt_entry_size is 8. But, qemu could split the 8 bytes into 2 consecutive
4-byte writes.
If each 4-byte partial write could trigger a host ggtt write, it is very
possible that a wrong combination is written to the host ggtt. E.g.
the higher 4 bytes is the old value, but the lower 4 bytes is the new
value, and this 8-byte combination is wrong but written to the ggtt, thus
causing bugs.
To handle this condition, we just record the first 4-byte write, then wait
until the second 4-byte write comes and write the combined 64-bit data to
host ggtt table.
To save memory space and to spot partial write as early as possible, we
don't keep this information for every ggtt index. Instread, we just record
the last ggtt write position, and assume the two 4-byte writes come in
consecutively for each vgpu.
This assumption is right based on the characteristic of ggtt entry which
stores memory address. When gtt_entry_size is 8, the guest memory physical
address should be 64 bits, so any sane guest driver should write 8-byte
long data at a time, so 2 consecutive 4-byte writes at the same ggtt index
should be trapped in gvt.
v2:
when incomplete ggtt entry write is located, e.g.
1. guest only writes 4 bytes at a ggtt offset and no long writes the
rest 4 bytes.
2. guest writes 4 bytes of a ggtt offset, then write at other ggtt
offsets, then return back to write the left 4 bytes of the first
ggtt offset.
add error handling logic to remap host entry to scratch page, and mark
guest virtual ggtt entry as not present. (zhenyu wang)
Signed-off-by: Zhao Yan <yan.y.zhao@intel.com>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_device. The resulting code is more aligned with the rest
of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_gem_object. The resulting code is more aligned with the
rest of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
This patch unifies the naming of DRM functions for reference counting
of struct drm_framebuffer. The resulting code is more aligned with the
rest of the Linux kernel interfaces.
Signed-off-by: Thomas Zimmermann <tdz@users.sourceforge.net>
Signed-off-by: Inki Dae <inki.dae@samsung.com>
Amritha Nambiar says:
====================
Symmetric queue selection using XPS for Rx queues
This patch series implements support for Tx queue selection based on
Rx queue(s) map. This is done by configuring Rx queue(s) map per Tx-queue
using sysfs attribute. If the user configuration for Rx queues does
not apply, then the Tx queue selection falls back to XPS using CPUs and
finally to hashing.
XPS is refactored to support Tx queue selection based on either the
CPUs map or the Rx-queues map. The config option CONFIG_XPS needs to be
enabled. By default no receive queues are configured for the Tx queue.
- /sys/class/net/<dev>/queues/tx-*/xps_rxqs
A set of receive queues can be mapped to a set of transmit queues (many:many),
although the common use case is a 1:1 mapping. This will enable sending
packets on the same Tx-Rx queue association as this is useful for busy polling
multi-threaded workloads where it is not possible to pin the threads to
a CPU. This is a rework of Sridhar's patch for symmetric queueing via
socket option:
https://www.spinics.net/lists/netdev/msg453106.html
Testing Hints:
Kernel: Linux 4.17.0-rc7+
Interface:
driver: ixgbe
version: 5.1.0-k
firmware-version: 0x00015e0b
Configuration:
ethtool -L $iface combined 16
ethtool -C $iface rx-usecs 1000
sysctl net.core.busy_poll=1000
ATR disabled:
ethtool -K $iface ntuple on
Workload:
Modified memcached that changes the thread selection policy to be based
on the incoming rx-queue of a connection using SO_INCOMING_NAPI_ID socket
option. The default is round-robin.
Default: No rxqs_map configured
Symmetric queues: Enable rxqs_map for all queues 1:1 mapped to Tx queue
System:
Architecture: x86_64
CPU(s): 72
Model name: Intel(R) Xeon(R) CPU E5-2699 v3 @ 2.30GHz
16 threads 400K requests/sec
=============================
-------------------------------------------------------------------------------
Default Symmetric queues
-------------------------------------------------------------------------------
RTT min/avg/max 4/51/2215 2/30/5163
(usec)
intr/sec 26655 18606
contextswitch/sec 5145 4044
insn per cycle 0.43 0.72
cache-misses 6.919 4.310
(% of all cache refs)
L1-dcache-load- 4.49 3.29
-misses
(% of all L1-dcache hits)
LLC-load-misses 13.26 8.96
(% of all LL-cache hits)
-------------------------------------------------------------------------------
32 threads 400K requests/sec
=============================
-------------------------------------------------------------------------------
Default Symmetric queues
-------------------------------------------------------------------------------
RTT min/avg/max 10/112/5562 9/46/4637
(usec)
intr/sec 30456 27666
contextswitch/sec 7552 5133
insn per cycle 0.41 0.49
cache-misses 9.357 2.769
(% of all cache refs)
L1-dcache-load- 4.09 3.98
-misses
(% of all L1-dcache hits)
LLC-load-misses 12.96 3.96
(% of all LL-cache hits)
-------------------------------------------------------------------------------
16 threads 800K requests/sec
=============================
-------------------------------------------------------------------------------
Default Symmetric queues
-------------------------------------------------------------------------------
RTT min/avg/max 5/151/4989 9/69/2611
(usec)
intr/sec 35686 22907
contextswitch/sec 25522 12281
insn per cycle 0.67 0.74
cache-misses 8.652 6.38
(% of all cache refs)
L1-dcache-load- 3.19 2.86
-misses
(% of all L1-dcache hits)
LLC-load-misses 16.53 11.99
(% of all LL-cache hits)
-------------------------------------------------------------------------------
32 threads 800K requests/sec
=============================
-------------------------------------------------------------------------------
Default Symmetric queues
-------------------------------------------------------------------------------
RTT min/avg/max 6/163/6152 8/88/4209
(usec)
intr/sec 47079 26548
contextswitch/sec 42190 39168
insn per cycle 0.45 0.54
cache-misses 8.798 4.668
(% of all cache refs)
L1-dcache-load- 6.55 6.29
-misses
(% of all L1-dcache hits)
LLC-load-misses 13.91 10.44
(% of all LL-cache hits)
-------------------------------------------------------------------------------
v6:
- Changed the names of some functions to begin with net_if.
- Cleaned up sk_tx_queue_set/sk_rx_queue_set functions.
- Added sk_rx_queue_clear to make it consistent with tx_queue_mapping
initialization.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Extend transmit queue sysfs attribute to configure Rx queue(s) map
per Tx queue. By default no receive queues are configured for the
Tx queue.
- /sys/class/net/eth0/queues/tx-*/xps_rxqs
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support to pick Tx queue based on the Rx queue(s) map
configuration set by the admin through the sysfs attribute
for each Tx queue. If the user configuration for receive queue(s) map
does not apply, then the Tx queue selection falls back to CPU(s) map
based selection and finally to hashing.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds a new field to sock_common 'skc_rx_queue_mapping'
which holds the receive queue number for the connection. The Rx queue
is marked in tcp_finish_connect() to allow a client app to do
SO_INCOMING_NAPI_ID after a connect() call to get the right queue
association for a socket. Rx queue is also marked in tcp_conn_request()
to allow syn-ack to go on the right tx-queue associated with
the queue on which syn is received.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Change 'skc_tx_queue_mapping' field in sock_common structure from
'int' to 'unsigned short' type with ~0 indicating unset and
other positive queue values being set. This will accommodate adding
a new 'unsigned short' field in sock_common in the next patch for
rx_queue_mapping.
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
