AuxXxilium/linux_dsm_epyc7002
1
0
Fork 0
mirror of https://github.com/AuxXxilium/linux_dsm_epyc7002.git synced 2024-12-16 23:56:45 +07:00
Code Issues Actions Packages Projects Releases Wiki Activity
26720ab97f
linux_dsm_epyc7002/drivers
History
Tvrtko Ursulin 26720ab97f drm/i915: Move CSB MMIO reads out of the execlists lock 
By reading the CSB (slow MMIO accesses) into a temporary local
buffer we can decrease the duration of holding the execlist
lock.

Main advantage is that during heavy batch buffer submission we
reduce the execlist lock contention, which should decrease the
latency and CPU usage between the submitting userspace process
and interrupt handling.

Downside is that we need to grab and relase the forcewake twice,
but as the below numbers will show this is completely hidden
by the primary gains.

Testing with "gem_latency -n 100" (submit batch buffers with a
hundred nops each) shows more than doubling of the throughput
and more than halving of the dispatch latency, overall latency
and CPU time spend in the submitting process.

Submitting empty batches ("gem_latency -n 0") does not seem
significantly affected by this change with throughput and CPU
time improving by half a percent, and overall latency worsening
by the same amount.

Above tests were done in a hundred runs on a big core Broadwell.

v2:
  * Overflow protection to local CSB buffer.
  * Use closer dev_priv in execlists_submit_requests. (Chris Wilson)

v3: Rebase.

v4: Added commend about irq needed to be disabled in
    execlists_submit_request. (Chris Wilson)

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilsno <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1458219586-20452-1-git-send-email-tvrtko.ursulin@linux.intel.com
2016-03-18 10:25:56 +00:00
..
accessibility
acpi ACPICA: Revert "Parser: Fix for SuperName method invocation" 2016-03-08 22:58:38 +01:00
amba
android drivers: android: correct the size of struct binder_uintptr_t for BC_DEAD_BINDER_DONE 2016-02-20 15:43:56 -08:00
ata ata: ahci: don't mark HotPlugCapable Ports as external/removable 2016-02-29 16:17:57 -05:00
atm atm: solos-pci: use to_pci_dev() 2015-12-29 15:32:24 -05:00
auxdisplay
base device property: fwnode->secondary may contain ERR_PTR(-ENODEV) 2016-03-09 01:07:43 +01:00
bcma GPIO bulk updates for the v4.5 kernel cycle: 2016-01-17 12:32:01 -08:00
block null_blk: oops when initializing without lightnvm 2016-02-11 08:56:09 -07:00
bluetooth Bluetooth: btmrvl: don't send data to firmware while processing suspend 2016-01-06 16:37:14 +01:00
bus vexpress fixes for v4.5 2016-02-01 12:27:18 -08:00
cdrom cdrom: don't open-code memdup_user() 2016-01-06 08:25:24 -05:00
char Merge drm-fixes into drm-next. 2016-03-14 09:46:02 +10:00
clk clk: ti: omap3+: dpll: use non-locking version of clk_get_rate 2016-02-22 14:03:02 -08:00
clocksource Merge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2016-01-31 15:49:06 -08:00
connector connector: bump skb->users before callback invocation 2016-01-04 21:46:45 -05:00
cpufreq cpufreq: mediatek: allow building as a module 2016-03-01 02:43:05 +01:00
cpuidle Merge branches 'pm-cpuidle', 'pm-cpufreq', 'pm-domains' and 'pm-sleep' 2016-01-29 21:45:17 +01:00
crypto crypto: marvell/cesa - fix test in mv_cesa_dev_dma_init() 2016-02-06 15:23:56 +08:00
dca
devfreq PM / devfreq: tegra: Set freq in rate callback 2016-02-23 14:27:42 +09:00
dio
dma dmaengine: at_xdmac: fix residue computation 2016-03-10 16:32:36 +05:30
dma-buf dma-buf: Add ioctls to allow userspace to flush 2016-02-12 16:01:32 +01:00
edac EDAC, sb_edac: Fix logic when computing DIMM sizes on Xeon Phi 2016-03-07 19:07:40 +01:00
eisa
extcon extcon: add Maxim MAX3355 driver 2015-12-21 11:44:00 +09:00
firewire
firmware efi: Add pstore variables to the deletion whitelist 2016-02-16 12:48:18 +00:00
fmc
fpga
gpio gpio: rcar: Add Runtime PM handling for interrupts 2016-02-25 15:46:27 +01:00
gpu drm/i915: Move CSB MMIO reads out of the execlists lock 2016-03-18 10:25:56 +00:00
hid asm-generic changes for 4.5 2016-01-20 17:30:20 -08:00
hsi HSI: omap_ssi_port: fix handling of_get_named_gpio result 2016-01-07 16:07:54 +01:00
hv char/misc patches for 4.5-rc1 2016-01-13 10:23:36 -08:00
hwmon hwmon: (gpio-fan) Remove un-necessary speed_index lookup for thermal hook 2016-02-19 17:14:25 -08:00
hwspinlock drivers/hwspinlock: fix race between radix tree insertion and lookup 2016-02-03 08:28:43 -08:00
hwtracing coresight: Fix a typo in Kconfig 2015-12-21 13:31:31 -08:00
i2c i2c: brcmstb: allocate correct amount of memory for regmap 2016-03-01 19:16:45 +01:00
ide drivers/ide: make ide-scan-pci.c driver explicitly non-modular 2016-01-18 14:12:33 -05:00
idle
iio Second set of IIO fixes for the 4.5 cycle. These ones are mostly 2016-02-01 13:08:26 -08:00
infiniband Additional 4.5-rc6 fixes 2016-03-04 18:06:49 -08:00
input Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input 2016-02-10 12:21:57 -08:00
iommu iommu/vt-d: Use BUS_NOTIFY_REMOVED_DEVICE in hotplug path 2016-02-29 23:55:16 +01:00
ipack
irqchip irqchip/gicv3-its: Avoid cache flush beyond ITS_BASERn memory size 2016-02-17 17:39:05 +00:00
isdn ser_gigaset: use container_of() instead of detour 2016-02-19 15:52:41 -05:00
leds GPIO bulk updates for the v4.5 kernel cycle: 2016-01-17 12:32:01 -08:00
lguest lguest: Map switcher text R/O 2016-01-12 12:17:28 +01:00
lightnvm lightnvm: allow to force mm initialization 2016-02-04 09:19:45 -07:00
macintosh
mailbox mailbox: Fix dependencies for !HAS_IOMEM archs 2016-02-02 16:47:14 +05:30
mcb
md dm: fix dm_rq_target_io leak on faults with .request_fn DM w/ blk-mq paths 2016-02-21 20:27:50 -05:00
media Merge drm-fixes into drm-next. 2016-03-14 09:46:02 +10:00
memory ARM: SoC driver updates for v4.5 2016-01-20 18:42:30 -08:00
memstick memstick: use sector_div instead of do_div 2016-01-20 17:09:18 -08:00
message
mfd thermal: allow u8500-thermal driver to be a module 2016-02-09 14:18:23 -08:00
misc powerpc fixes for 4.5 #4 2016-03-06 11:08:06 -08:00
mmc mmc: omap_hsmmc: Fix PM regression with deferred probe for pm_runtime_reinit 2016-02-15 14:10:48 +01:00
mtd ubi: Fix out of bounds write in volume update code 2016-03-05 21:56:23 +01:00
net ppp: release rtnl mutex when interface creation fails 2016-03-07 16:11:31 -05:00
nfc nfc: pn544: Remove i2c client gpio irq configuration 2015-12-29 19:06:23 +01:00
ntb NTB: Fix macro parameter conflict with field name 2016-01-21 19:53:10 -05:00
nubus
nvdimm nvdimm: use 'u64' for pfn flags 2016-02-23 17:17:20 -08:00
nvme nvme: fix max_segments integer truncation 2016-03-03 14:43:10 -07:00
nvmem nvmem: qfprom: Specify LE device endianness 2016-02-07 23:09:13 -08:00
of of_mdio: fix kernel-doc for of_phy_connect() 2016-03-01 17:41:40 -05:00
oprofile wrappers for ->i_mutex access 2016-01-22 18:04:28 -05:00
parisc parisc: convert to dma_map_ops 2016-01-20 17:09:18 -08:00
parport parport: avoid assignment in if 2016-01-03 16:32:59 -08:00
pci PCI updates for v4.5: 2016-03-09 13:28:27 -08:00
pcmcia
perf
phy phy: twl4030-usb: Fix unbalanced pm_runtime_enable on module reload 2016-02-10 11:46:01 +05:30
pinctrl pinctrl: samsung: fix SMP race condition 2016-02-15 20:45:50 +01:00
platform Merge drm-fixes into drm-next. 2016-03-14 09:46:02 +10:00
pnp PNP: Add Haswell-ULT to Intel MCH size workaround 2016-02-03 01:00:29 +01:00
power power: bq27xxx_battery: Restore device name 2016-02-21 20:49:34 +01:00
powercap Merge branch 'powercap' 2016-01-12 01:12:40 +01:00
pps
ps3
ptp ptp: ixp46x: use helpers for converting ns to timespec 2016-01-29 12:38:59 -08:00
pwm pwm: Mark all devices as "might sleep" 2016-01-21 15:04:59 +01:00
rapidio rapidio: use kobj_to_dev() 2016-01-20 17:09:18 -08:00
ras
regulator regulator: Update for v4.5 2016-01-15 12:14:47 -08:00
remoteproc virtio: make find_vqs() checkpatch.pl-friendly 2016-01-12 20:47:06 +02:00
reset
rpmsg virtio: make find_vqs() checkpatch.pl-friendly 2016-01-12 20:47:06 +02:00
rtc RTC for 4.5 2016-01-18 12:10:45 -08:00
s390 s390/dasd: fix diag 0x250 inline assembly 2016-03-02 06:43:48 -06:00
sbus convert a bunch of open-coded instances of memdup_user_nul() 2016-01-04 10:26:58 -05:00
scsi ipr: Fix regression when loading firmware 2016-02-26 17:26:02 -05:00
sfi
sh drivers: sh: Restore legacy clock domain on SuperH platforms 2016-02-25 09:05:19 +09:00
sn
soc ARM: SoC support for Tegra platforms for v4.5 2016-01-22 17:30:52 -08:00
spi Merge remote-tracking branch 'spi/fix/rockchip' into spi-linus 2016-03-10 10:42:24 +07:00
spmi
ssb ssb: host_soc depends on sprom 2016-02-26 12:47:32 +02:00
staging media fixes for v4.5-rc7 2016-03-05 12:32:34 -08:00
target target/transport: add flag to indicate CPU Affinity is observed 2016-02-10 23:08:55 -08:00
tc
thermal thermal: cpu_cooling: fix out of bounds access in time_in_idle 2016-02-11 07:13:29 -08:00
thunderbolt
tty Revert "8250: uniphier: allow modular build with 8250 console" 2016-02-07 18:22:54 -08:00
uio
usb USB-serial fixes for v4.5-rc7 2016-03-03 12:37:21 -08:00
uwb
vfio vfio: fix ioctl error handling 2016-02-28 07:38:52 -07:00
vhost vhost: fix error path in vhost_init_used() 2016-03-02 17:01:49 +02:00
video Merge drm-fixes into drm-next. 2016-03-14 09:46:02 +10:00
virt
virtio virtio-pci: read the right virtio_pci_notify_cap field 2016-03-02 17:01:49 +02:00
vlynq
vme
w1
watchdog Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc 2016-03-01 08:57:34 -08:00
xen Xen bug fixes for 4.5-rc5 2016-02-22 13:57:01 -08:00
zorro
Kconfig
Makefile