linux_dsm_epyc7002/drivers
Chris Wilson a679f58d05 drm/i915: Flush pages on acquisition
When we return pages to the system, we ensure that they are marked as
being in the CPU domain since any external access is uncontrolled and we
must assume the worst. This means that we need to always flush the pages
on acquisition if we need to use them on the GPU, and from the beginning
have used set-domain. Set-domain is overkill for the purpose as it is a
general synchronisation barrier, but our intent is to only flush the
pages being swapped in. If we move that flush into the pages acquisition
phase, we know then that when we have obj->mm.pages, they are coherent
with the GPU and need only maintain that status without resorting to
heavy handed use of set-domain.

The principle knock-on effect for userspace is through mmap-gtt
pagefaulting. Our uAPI has always implied that the GTT mmap was async
(especially as when any pagefault occurs is unpredicatable to userspace)
and so userspace had to apply explicit domain control itself
(set-domain). However, swapping is transparent to the kernel, and so on
first fault we need to acquire the pages and make them coherent for
access through the GTT. Our use of set-domain here leaks into the uABI
that the first pagefault was synchronous. This is unintentional and
baring a few igt should be unoticed, nevertheless we bump the uABI
version for mmap-gtt to reflect the change in behaviour.

Another implication of the change is that gem_create() is presumed to
create an object that is coherent with the CPU and is in the CPU write
domain, so a set-domain(CPU) following a gem_create() would be a minor
operation that merely checked whether we could allocate all pages for
the object. On applying this change, a set-domain(CPU) causes a clflush
as we acquire the pages. This will have a small impact on mesa as we move
the clflush here on !llc from execbuf time to create, but that should
have minimal performance impact as the same clflush exists but is now
done early and because of the clflush issue, userspace recycles bo and
so should resist allocating fresh objects.

Internally, the presumption that objects are created in the CPU
write-domain and remain so through writes to obj->mm.mapping is more
prevalent than I expected; but easy enough to catch and apply a manual
flush.

For the future, we should push the page flush from the central
set_pages() into the callers so that we can more finely control when it
is applied, but for now doing it one location is easier to validate, at
the cost of sometimes flushing when there is no need.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.william.auld@gmail.com>
Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com>
Cc: Antonio Argenziano <antonio.argenziano@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Reviewed-by: Matthew Auld <matthew.william.auld@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190321161908.8007-1-chris@chris-wilson.co.uk
2019-03-21 17:28:12 +00:00
..
accessibility
acpi Linux 5.0-rc7 2019-02-18 13:27:15 +10:00
amba
android binderfs: remove separate device_initcall() 2019-02-01 15:50:26 +01:00
ata libata: Add NOLPM quirk for SAMSUNG MZ7TE512HMHP-000L1 SSD 2019-02-06 12:47:09 -07:00
atm atm: he: fix sign-extension overflow on large shift 2019-01-17 11:27:00 -08:00
auxdisplay auxdisplay: ht16k33: fix potential user-after-free on module unload 2019-02-15 19:48:39 +01:00
base Merge drm/drm-next into drm-intel-next-queued 2019-03-11 13:09:20 +02:00
bcma
block floppy: check_events callback should not return a negative number 2019-02-12 09:13:18 -07:00
bluetooth
bus Merge branch 'pwm-dmtimer-fixes' into omap-for-v5.0/fixes-v2 2019-01-29 07:53:47 -08:00
cdrom
char Char/Misc driver fixes for 5.0-rc4 2019-01-25 13:03:34 -10:00
clk clk: at91: fix masterck name 2019-02-20 11:40:21 -08:00
clocksource Merge branch 'pwm-dmtimer-fixes' into omap-for-v5.0/fixes-v2 2019-01-29 07:53:47 -08:00
connector
cpufreq cpufreq: scmi: Fix use-after-free in scmi_cpufreq_exit() 2019-02-19 11:17:37 +01:00
cpuidle cpuidle: poll_state: Fix default time limit 2019-01-30 22:57:42 +01:00
crypto Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6 2019-02-28 09:05:18 -08:00
dax
dca
devfreq
dio
dma dmaengine-fix-5.0-rc6 2019-02-10 10:39:37 -08:00
dma-buf drm-misc-next for 5.1: 2019-01-10 05:58:52 +10:00
edac EDAC, altera: Fix S10 persistent register offset 2019-01-24 17:13:59 +01:00
eisa
extcon
firewire scsi: communicate max segment size to the DMA mapping code 2019-01-22 20:40:59 -05:00
firmware Merge branch 'efi-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2019-02-17 09:22:01 -08:00
fmc
fpga fpga: stratix10-soc: fix wrong of_node_put() in init function 2019-01-31 16:19:48 +01:00
fsi
gnss
gpio gpio: pxa: avoid attempting to set pin direction via pinctrl on MMP2 2019-02-14 10:06:03 +01:00
gpu drm/i915: Flush pages on acquisition 2019-03-21 17:28:12 +00:00
hid HID: debug: fix the ring buffer implementation 2019-01-29 12:09:11 +01:00
hsi
hv vmbus: fix subchannel removal 2019-01-09 19:20:31 -05:00
hwmon hwmon: (nct6775) Fix fan6 detection for NCT6793D 2019-01-27 18:55:49 -08:00
hwspinlock
hwtracing
i2c i2c: bcm2835: Clear current buffer pointers and counts after a transfer 2019-02-15 09:45:05 +01:00
i3c i3c: master: dw: fix deadlock 2019-01-26 11:14:25 +01:00
ide ide: ensure atapi sense request aren't preempted 2019-01-31 08:25:09 -07:00
idle
iio First set of IIO fixes for the 5.0 cycle. 2019-02-03 13:10:41 +01:00
infiniband iw_cxgb4: cq/qp mask depends on bar2 pages in a host page 2019-02-15 09:39:39 -07:00
input Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input 2019-02-17 08:30:35 -08:00
iommu IOMMU Fix for Linux v5.0-rc8 2019-03-01 09:13:04 -08:00
ipack
irqchip Merge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip 2019-02-10 09:54:19 -08:00
isdn mISDN: fix a race in dev_expire_timer() 2019-02-05 16:39:29 -08:00
leds leds: lp5523: fix a missing check of return value of lp55xx_read 2019-01-17 22:27:39 +01:00
lightnvm
macintosh
mailbox mailbox: bcm-flexrm-mailbox: Fix FlexRM ring flush timeout issue 2019-02-18 10:40:58 -06:00
mcb
md for-linus-20190215 2019-02-15 09:12:28 -08:00
media media: vim2m: only cancel work if it is for right context 2019-01-16 11:13:25 -05:00
memory
memstick
message
mfd mfd: Fix unmet dependency warning for MFD_TPS68470 2019-01-29 10:55:34 +01:00
misc mic: vop: Fix crash on remove 2019-02-01 15:53:54 +01:00
mmc mmc: sdhci-esdhc-imx: correct the fix of ERR004536 2019-02-28 12:18:31 +01:00
mtd mtd: powernv_flash: Fix device registration error 2019-02-13 14:19:40 +01:00
mux
net net: dsa: mv88e6xxx: Fix statistics on mv88e6161 2019-03-02 00:45:04 -08:00
nfc
ntb cross-tree: phase out dma_zalloc_coherent() 2019-01-08 07:58:37 -05:00
nubus
nvdimm libnvdimm/security: Require nvdimm_security_setup_events() to succeed 2019-01-21 09:57:43 -08:00
nvme nvme-pci: add missing unlock for reset error 2019-02-12 09:29:07 +01:00
nvmem
of OF: properties: add missing of_node_put 2019-01-16 12:49:53 -06:00
opp
oprofile
parisc
parport
pci pci-v5.0-fixes-4 2019-02-08 15:32:10 -08:00
pcmcia
perf
phy Linux 5.0-rc7 2019-02-18 13:27:15 +10:00
pinctrl pinctrl: meson: meson8b: fix the sdxc_a data 1..3 pins 2019-02-17 22:24:32 +01:00
platform platform/x86: Fix unmet dependency warning for SAMSUNG_Q10 2019-01-29 10:59:07 +01:00
pnp
power
powercap
pps
ps3
ptp ptp: check that rsv field is zero in struct ptp_sys_offset_extended 2019-01-08 16:22:56 -05:00
pwm
rapidio cross-tree: phase out dma_zalloc_coherent() 2019-01-08 07:58:37 -05:00
ras
regulator
remoteproc virtio: don't allocate vqs when names[i] = NULL 2019-01-14 20:15:19 -05:00
reset reset: uniphier-glue: Add AHCI reset control support in glue layer 2019-01-07 16:38:51 +01:00
rpmsg
rtc
s390 s390 update with bug fixes for 5.0-rc6 2019-02-11 10:28:48 -08:00
sbus
scsi SCSI fixes on 20190302 2019-03-02 11:39:54 -08:00
sfi
sh
siox
slimbus
sn
soc soc/fsl fixes for v5.0 2019-01-30 11:14:04 +01:00
soundwire
spi cross-tree: phase out dma_zalloc_coherent() 2019-01-08 07:58:37 -05:00
spmi
ssb
staging Linux 5.0-rc7 2019-02-18 13:27:15 +10:00
target scsi: target: make the pi_prot_format ConfigFS path readable 2019-02-04 21:40:32 -05:00
tc
tee tee: optee: add missing of_node_put after of_device_is_available 2019-02-28 10:23:04 +01:00
thermal thermal: cpu_cooling: Clarify error message 2019-02-05 15:50:13 -08:00
thunderbolt
tty TTY/Serial fixes for 5.0-rc6 2019-02-08 10:49:55 -08:00
uio
usb usb: typec: tcpm: Correct the PPS out_volt calculation 2019-01-31 09:14:00 +01:00
uwb
vfio vfio-pci/nvlink2: Fix ancient gcc warnings 2019-01-23 08:20:43 -07:00
vhost vhost: correctly check the return value of translate_desc() in log_used() 2019-02-19 13:14:45 -08:00
video TTY/Serial driver fixes for 5.0-rc4 2019-01-25 12:58:40 -10:00
virt
virtio virtio: drop internal struct from UAPI 2019-02-05 15:29:48 -05:00
visorbus
vlynq
vme
w1
watchdog watchdog: tqmx86: Fix a couple IS_ERR() vs NULL bugs 2019-01-07 10:10:35 +01:00
xen arm64/xen: fix xen-swiotlb cache flushing 2019-01-23 22:14:56 +01:00
zorro
Kconfig
Makefile