This adds devicetree documentation for the bindings of the
ads7828 driver.
Signed-off-by: Denis Carikli <denis@eukrea.com>
Acked-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Cadence 1.0 version has bugs which have been fixed in the cadence 1.4 version.
This patch removes the quirks present in the driver for cadence 1.4 version.
Signed-off-by: Anurag Kumar Vulisha <anuragku@xilinx.com>
[wsa: fixed indentation issues in r1p10_i2c_def]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Tegra I2C controller required to configure the clock divisor
register inside controller to different value based on the clock
speed. The recommended clock divisor for the I2C controller for
standard/fast mode is 0x19 and for fast-mode plus is 0x10.
Add support to configure clock divisor register of I2C controller
based on bus clock rate.
This clock divisor is supported form T114 onwards.
Signed-off-by: Chaitanya Bandi <bandik@nvidia.com>
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Once the new configuration is set on the conifg register of
I2C controller, it is require to update the CONFIG_LOAD register
to transfer the new SW configuration to actual HW internal
registers that would be used in the actual logic.
It is like, SW is programming only shadow registers through
regular configuration and when these load_config bit fields
are set to 1, it causes the regular/shadows registers
configuration transferred to the HW internal active registers.
So SW has to set these bit fields at the end of all regular
registers configuration. And these config_load bits are HW
auto-clear bits. HW clears these bit fields once the register
configuration is moved to HW internal active registers. So SW
has to wait until these bits are auto-cleared before going
for any further programming
This mechanism is supported on T124 and after this SoCs.
Signed-off-by: Chaitanya Bandi <bandik@nvidia.com>
Signed-off-by: Laxman Dewangan <ldewangan@nvidia.com>
Tested-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Patch "i2c: omap: abolish variable name confusion" triggered a
coccinelle warning which we fix here:
drivers/i2c/busses/i2c-omap.c:1333:5-24: pm_runtime_get_sync returns < 0 as error. Unecessary IS_ERR_VALUE at line 1334
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
In case of error conditions like Arbitration lost or NACK lets signal
the waiting process.
Handle error cases in the Rx path
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Currently only one interrupt is serviced in the isr.
In case the multiple interrupts happen simultenously we service and ack
only one of them. Check for all the causes in the isr and service them.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Currently there is no slave mode support in the driver
also in the isr we just ack it and do nothing.
So disable the AAS interrupt.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Remove the busy loop while waiting for bus busy.
Instead let the processor sleep.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Currently before every transfer the interrupts are disabled.
So incase the slave nacks in the middle of the transfer the
current transfer is not aborted. Upon enabling the interrupts
conditions like NACK , arbitration lost will not be masked.
Remove the disabling of the interrupts.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Currently before every transfer the controller is reinitialised.
We are already resetting the controller upon errors so upon every
transfer is a performance kill.
Remove the same.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
The xiic_process is a 154 line code that runs in isr context currently
move it to thread context. Also the name xiic_process suggests that the
intension was to run in process context.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
[wsa: initialized irqreturn_t to IRQ_NONE instead of IRQ_HANDLED]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Currently the interrupts are disabled at the start of the
isr and enabled at the end of the isr. Remove the same.
In case the slave device NACKs the transaction while in the isr
the transfer will continue and the NACK interrupt will arrive
only after the isr is serviced.
Signed-off-by: Shubhrajyoti Datta <shubhraj@xilinx.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
This patch cleans up i2c_pxa_probe() function,
- Use devm_ variants wherever
This will clean both probe exit and i2c_pxa_remove() functions
- Check platform resource before parsing any other data from DT/platform
- Use dev_err on failure from i2c_add_numbered_adapter()
- Use pr_info instead of printk for KERN_INFO
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
[wsa: removed unneeded error prinout after devm_ioremap_resource]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Update i2c_pxa_scream_blue_murder() fn to print more information
in case of error.
Also, use dev_err variants instead of printk.
Signed-off-by: Jett.Zhou <jtzhou@marvell.com>
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Cc: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Fix below warning message, coming from 64 bit toolchain.
drivers/i2c/busses/i2c-pxa.c:1237:15:
warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
Signed-off-by: Yipeng Yao <ypyao@marvell.com>
[vaibhav.hiremath@linaro.org: Updated Changelog]
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Cc: Wolfram Sang <wsa@the-dreams.de>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
In case of timeout in pio mode of operation return I2C_RETRY.
This behavior will be same as interrupt mode of operation.
Signed-off-by: Shouming Wang <wangshm@marvell.com>
[vaibhav.hiremath@linaro.org: Updated changelog]
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Normally i2c controller works as master, so slave addr is not needed, or
it will impact some slave device (eg. ST NFC chip) i2c accesses, because
it has the same i2c address with controller.
For example,
On the pxa1928 based platform, where PMIC (88pm860) is present @0x30
address on TWSI0 interface, and if we set 0x30 as a slave address in
pxa1928 TWSI0 module, all the transactions towards PMIC would go for toss.
Signed-off-by: Jett.Zhou <jtzhou@marvell.com>
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Acked-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
During suspend there may still be some i2c access happening, as the
interrupt is shared between multiple drivers.
And if we don't keep i2c irq ON, there may be i2c access timeout if
i2c is in irq mode of operation.
Signed-off-by: Raul Xiong <xjian@marvell.com>
Signed-off-by: Xiaofan Tian <tianxf@marvell.com>
[vaibhav.hiremath@linaro.org: updated Changelog]
Signed-off-by: Vaibhav Hiremath <vaibhav.hiremath@linaro.org>
Cc: Wolfram Sang <wsa@the-dreams.de>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
we're about to remove the module, so we can't
really schedule a PM transition in the future,
we must wait for it to finish.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
struct device pointers are usually called
dev. Calling our struct omap_i2c_dev pointers
also dev has caused enough confusion.
This is the result of a few simple sed rules
to convert all struct omap_i2c_dev pointers
to be called omap instead.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
there's no need to fetch the platform_device
in order to dereference it back to the dev
pointer to access drvdata, we can use
dev_get_drvdata() instead.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Acked-by: Alexander Sverdlin <alexander.sverdlin@nokia.com>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Start using pr_fmt and convert all remaining printk to use
pr_* family of macros.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
[wsa: remove print on kzalloc failure]
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Now parport is using device model so use dev_* macros instead of printk.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Wolfram Sang <wsa@the-dreams.de>
Pull input subsystem fixes from Dmitry Torokhov:
"Just small ALPS and Elan touchpads, and other driver fixups"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/dtor/input:
Input: elantech - add special check for fw_version 0x470f01 touchpad
Input: twl4030-vibra - fix ERROR: Bad of_node_put() warning
Input: alps - only Dell laptops have separate button bits for v2 dualpoint sticks
Input: axp20x-pek - add module alias
Input: turbografx - fix potential out of bound access
Pull MIPS fixes from Ralf Baechle:
"Another round of MIPS fixes for 4.2. No area does particularly stand
out but we have a two unpleasant ones:
- Kernel ptes are marked with a global bit which allows the kernel to
share kernel TLB entries between all processes. For this to work
both entries of an adjacent even/odd pte pair need to have the
global bit set. There has been a subtle race in setting the other
entry's global bit since ~ 2000 but it take particularly
pathological workloads that essentially do mostly vmalloc/vfree to
trigger this.
This pull request fixes the 64-bit case but leaves the case of 32
bit CPUs with 64 bit ptes unsolved for now. The unfixed cases
affect hardware that is not available in the field yet.
- Instruction emulation requires loading instructions from user space
but the current fast but simplistic approach will fail on pages
that are PROT_EXEC but !PROT_READ. For this reason we temporarily
do not permit this permission and will map pages with PROT_EXEC |
PROT_READ.
The remainder of this pull request is more or less across the field
and the short log explains them well"
* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus:
MIPS: Make set_pte() SMP safe.
MIPS: Replace add and sub instructions in relocate_kernel.S with addiu
MIPS: Flush RPS on kernel entry with EVA
Revert "MIPS: BCM63xx: Provide a plat_post_dma_flush hook"
MIPS: BMIPS: Delete unused Kconfig symbol
MIPS: Export get_c0_perfcount_int()
MIPS: show_stack: Fix stack trace with EVA
MIPS: do_mcheck: Fix kernel code dump with EVA
MIPS: SMP: Don't increment irq_count multiple times for call function IPIs
MIPS: Partially disable RIXI support.
MIPS: Handle page faults of executable but unreadable pages correctly.
MIPS: Malta: Don't reinitialise RTC
MIPS: unaligned: Fix build error on big endian R6 kernels
MIPS: Fix sched_getaffinity with MT FPAFF enabled
MIPS: Fix build with CONFIG_OF=y for non OF-enabled targets
CPUFREQ: Loongson2: Fix broken build due to incorrect include.
Pull btrfs fix from Chris Mason:
"We have a btrfs quota regression fix.
I merged this one on Thursday and have run it through tests against
current master.
Normally I wouldn't have sent this while you were finalizing rc6, but
I'm feeding mosquitoes in the adirondacks next week, so I wanted to
get this one out before leaving. I'll leave longer tests running and
check on things during the week, but I don't expect any problems"
* 'for-linus-4.2' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
btrfs: qgroup: Fix a regression in qgroup reserved space.
Pull thermal management fixes from Zhang Rui:
"Specifics:
- fix an error that "weight_attr" sysfs attribute is not removed
while unbinding. From: Viresh Kumar.
- fix power allocator governor tracing to return the real request.
From Javi Merino.
- remove redundant owner assignment of hisi platform thermal driver.
From Krzysztof Kozlowski.
- a couple of small fixes of Exynos thermal driver. From Krzysztof
Kozlowski and Chanwoo Choi"
* 'for-rc' of git://git.kernel.org/pub/scm/linux/kernel/git/rzhang/linux:
thermal: Drop owner assignment from platform_driver
thermal: exynos: Remove unused code related to platform_data on probe()
thermal: exynos: Add the dependency of CONFIG_THERMAL_OF instead of CONFIG_OF
thermal: exynos: Disable the regulator on probe failure
thermal: power_allocator: trace the real requested power
thermal: remove dangling 'weight_attr' device file
Pull ARC fixes from Vineet Gupta:
"Here's a late pull request for accumulated ARC fixes which came out of
extended testing of the new ARCv2 port with LTP etc. llock/scond
livelock workaround has been reviewed by PeterZ. The changes look a
lot but I've crafted them into finer grained patches for better
tracking later.
I have some more fixes (ARC Futex backend) ready to go but those will
have to wait for tglx to return from vacation.
Summary:
- Enable a reduced config of HS38 (w/o div-rem, ll64...)
- Add software workaround for LLOCK/SCOND livelock
- Fallout of a recent pt_regs update"
* tag 'arc-v4.2-rc6-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc:
ARCv2: spinlock/rwlock/atomics: reduce 1 instruction in exponential backoff
ARC: Make pt_regs regs unsigned
ARCv2: spinlock/rwlock: Reset retry delay when starting a new spin-wait cycle
ARCv2: spinlock/rwlock/atomics: Delayed retry of failed SCOND with exponential backoff
ARC: LLOCK/SCOND based rwlock
ARC: LLOCK/SCOND based spin_lock
ARC: refactor atomic inline asm operands with symbolic names
Revert "ARCv2: STAR 9000837815 workaround hardware exclusive transactions livelock"
ARCv2: [axs103_smp] Reduce clk for Quad FPGA configs
ARCv2: Fix the peripheral address space detection
ARCv2: allow selection of page size for MMUv4
ARCv2: lib: memset: Don't assume 64-bit load/stores
ARCv2: lib: memcpy: Missing PREFETCHW
ARCv2: add knob for DIV_REV in Kconfig
ARC/time: Migrate to new 'set-state' interface
Pull virtio fix from Michael Tsirkin:
"A last minute fix for the new virtio input driver. It seems pretty
obvious, and the problem it's fixing would be quite hard to debug"
* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
virtio-input: reset device and detach unused during remove
Pull device mapper fixes from Mike Snitzer:
- stable fix for a dm_merge_bvec() regression on 32 bit Fedora systems.
- fix for a 4.2 DM thinp discard regression due to inability to
properly delete a range of blocks in a data mapping btree.
* tag 'dm-4.2-fixes-4' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
dm btree remove: fix bug in remove_one()
dm: fix dm_merge_bvec regression on 32 bit systems
Pull sound fixes from Takashi Iwai:
"The only bulk changes in this request is ABI updates for ASoC topology
API. It's a new API that was introduced in 4.2, and we'd like to
avoid ABI change after the release, so it's taken now. As there is no
real in-tree user for this API, it should be fairly safe.
Other than that, the usual small fixes are found in various drivers:
ASoC cs4265, rt5645, intel-sst, firewire, oxygen and HD-audio"
* tag 'sound-4.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound:
ASoC: topology: Add private data type and bump ABI version to 3
ASoC: topology: Add ops support to byte controls UAPI
ASoC: topology: Update TLV support so we can support more TLV types
ASoC: topology: add private data to manifest
ASoC: topology: Add subsequence in topology
ALSA: hda - one Dell machine needs the headphone white noise fixup
ALSA: fireworks/firewire-lib: add support for recent firmware quirk
Revert "ALSA: fireworks: add support for AudioFire2 quirk"
ASoC: topology: fix typo in soc_tplg_kcontrol_bind_io()
ALSA: HDA: Dont check return for snd_hdac_chip_readl
ALSA: HDA: Fix stream assignment for host in decoupled mode
ASoC: rt5645: Fix lost pin setting for DMIC1
ALSA: oxygen: Fix logical-not-parentheses warning
ASoC: Intel: sst_byt: fix initialize 'NULL device *' issue
ASoC: Intel: haswell: fix initialize 'NULL device *' issue
ASoC: cs4265: Fix setting dai format for Left/Right Justified
Pull hwmon fixes from Guenter Roeck:
- Export module alias information in g762 and nct7904 to support
auto-loading.
- Blacklist Dell Studio XPS 8100 in dell-smm to fix fan control
problems.
* tag 'hwmon-for-linus-v4.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
hwmon: (g762) Export OF module alias information
hwmon: (nct7904) Export I2C module alias information
hwmon: (dell-smm) Blacklist Dell Studio XPS 8100
Pull USB fixes from Greg KH:
"Here are some USB and PHY fixes for 4.2-rc6 that resolve some reported
issues.
All of these have been in the linux-next tree for a while, full
details on the patches are in the shortlog below"
* tag 'usb-4.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
ARM: dts: dra7: Add syscon-pllreset syscon to SATA PHY
drivers/usb: Delete XHCI command timer if necessary
xhci: fix off by one error in TRB DMA address boundary check
usb: udc: core: add device_del() call to error pathway
phy: ti-pipe3: i783 workaround for SATA lockup after dpll unlock/relock
phy-sun4i-usb: Add missing EXPORT_SYMBOL_GPL for sun4i_usb_phy_set_squelch_detect
USB: sierra: add 1199:68AB device ID
usb: gadget: f_printer: actually limit the number of instances
usb: gadget: f_hid: actually limit the number of instances
usb: gadget: f_uac2: fix calculation of uac2->p_interval
usb: gadget: bdc: fix a driver crash on disconnect
usb: chipidea: ehci_init_driver is intended to call one time
USB: qcserial: Add support for Dell Wireless 5809e 4G Modem
USB: qcserial/option: make AT URCs work for Sierra Wireless MC7305/MC7355
Pull staging driver fixes from Greg KH:
"Here are three bugfixes for some staging driver issues that have been
reported. All have been in the linux-next tree for a while"
* tag 'staging-4.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: lustre: Include unaligned.h instead of access_ok.h
staging: vt6655: vnt_bss_info_changed check conf->beacon_rate is not NULL
staging: comedi: das1800: add missing break in switch
Pull char/misc fixes from Greg KH:
"Here are some extcon fixes for 4.2-rc6 that resolve some reported
problems.
All have been in linux-next for a while"
* tag 'char-misc-4.2-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
extcon: Fix extcon_cable_get_state() from getting old state after notification
extcon: Fix hang and extcon_get/set_cable_state().
extcon: palmas: Fix NULL pointer error
Pull drm fixes from Daniel Vetter:
"One i915 regression fix and a drm core one since Dave's not around,
both introduced in 4.2 so not cc: stable.
The fix for the warning Ted reported isn't in here yet since he didn't
yet supply a tested-by and I can't repro this one myself (it's in
fixup code that needs firmware doing something i915 wouldn't do)"
* tag 'drm-intel-fixes-2015-08-07' of git://anongit.freedesktop.org/drm-intel:
drm/vblank: Use u32 consistently for vblank counters
drm/i915: Allow parsing of variable size child device entries from VBT
It is no need to check the packet[0] for sanity check when doing
elantech_packet_check_v4() function for fw_version = 0x470f01 touchpad.
Signed-off by: Duson Lin <dusonlin@emc.com.tw>
Reviewed-by: Ulrik De Bie <ulrik.debie-os@e2big.org>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
remove_one() was not incrementing the key for the beginning of the
range, so not all entries were being removed. This resulted in
discards that were not unmapping all blocks.
Fixes: 4ec331c3ea ("dm btree: add dm_btree_remove_leaves()")
Signed-off-by: Joe Thornber <ejt@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
In
commit 99264a61df
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date: Wed Apr 15 19:34:43 2015 +0200
drm/vblank: Fixup and document timestamp update/read barriers
I've switched vblank->count from atomic_t to unsigned long and
accidentally created an integer comparison bug in
drm_vblank_count_and_time since vblanke->count might overflow the u32
local copy and hence the retry loop never succeed.
Fix this by consistently using u32.
Cc: Michel Dänzer <michel@daenzer.net>
Reported-by: Michel Dänzer <michel@daenzer.net>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
ASoC: Fixes for v4.2
There are a couple of small driver specific fixes here but the
overwhelming bulk of these changes are fixes to the topology ABI that
has been newly introduced in v4.2. Once this makes it into a release we
will have to firm this up but for now getting enhancements in before
they've made it into a release is the most expedient thing.
Merge fixes from Andrew Morton:
"21 fixes"
* emailed patches from Andrew Morton <akpm@linux-foundation.org>: (21 commits)
writeback: fix initial dirty limit
mm/memory-failure: set PageHWPoison before migrate_pages()
mm: check __PG_HWPOISON separately from PAGE_FLAGS_CHECK_AT_*
mm/memory-failure: give up error handling for non-tail-refcounted thp
mm/memory-failure: fix race in counting num_poisoned_pages
mm/memory-failure: unlock_page before put_page
ipc: use private shmem or hugetlbfs inodes for shm segments.
mm: initialize hotplugged pages as reserved
ocfs2: fix shift left overflow
kthread: export kthread functions
fsnotify: fix oops in fsnotify_clear_marks_by_group_flags()
lib/iommu-common.c: do not use 0xffffffffffffffffl for computing align_mask
mm/slub: allow merging when SLAB_DEBUG_FREE is set
signalfd: fix information leak in signalfd_copyinfo
signal: fix information leak in copy_siginfo_to_user
signal: fix information leak in copy_siginfo_from_user32
ocfs2: fix BUG in ocfs2_downconvert_thread_do_work()
fs, file table: reinit files_stat.max_files after deferred memory initialisation
mm, meminit: replace rwsem with completion
mm, meminit: allow early_pfn_to_nid to be used during runtime
...
If we have a series of events from userpsace, with %fprs=FPRS_FEF,
like follows:
ETRAP
ETRAP
VIS_ENTRY(fprs=0x4)
VIS_EXIT
RTRAP (kernel FPU restore with fpu_saved=0x4)
RTRAP
We will not restore the user registers that were clobbered by the FPU
using kernel code in the inner-most trap.
Traps allocate FPU save slots in the thread struct, and FPU using
sequences save the "dirty" FPU registers only.
This works at the initial trap level because all of the registers
get recorded into the top-level FPU save area, and we'll return
to userspace with the FPU disabled so that any FPU use by the user
will take an FPU disabled trap wherein we'll load the registers
back up properly.
But this is not how trap returns from kernel to kernel operate.
The simplest fix for this bug is to always save all FPU register state
for anything other than the top-most FPU save area.
Getting rid of the optimized inner-slot FPU saving code ends up
making VISEntryHalf degenerate into plain VISEntry.
Longer term we need to do something smarter to reinstate the partial
save optimizations. Perhaps the fundament error is having trap entry
and exit allocate FPU save slots and restore register state. Instead,
the VISEntry et al. calls should be doing that work.
This bug is about two decades old.
Reported-by: James Y Knight <jyknight@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
