Since __ATTR_PREALLOC was introduced in v3.19-rc1~78^2~18
it can now be used by md.
This ensure that writing to these sysfs attributes will never
block due to a memory allocation.
Such blocking could become a deadlock if mdmon is trying to
reconfigure an array after a failure prior to re-enabling writes.
Signed-off-by: NeilBrown <neilb@suse.de>
When we have more than 1 drive failure, it's possible we start
rebuild one drive while leaving another faulty drive in array.
To determine whether array will be optimal after building, current
code only check whether a drive is missing, which could potentially
lead to data corruption. This patch is to add checking Faulty flag.
Signed-off-by: NeilBrown <neilb@suse.de>
When a drive is marked write-mostly it should only be the
target of reads if there is no other option.
This behaviour was broken by
commit 9dedf60313
md/raid1: read balance chooses idlest disk for SSD
which causes a write-mostly device to be *preferred* is some cases.
Restore correct behaviour by checking and setting
best_dist_disk and best_pending_disk rather than best_disk.
We only need to test one of these as they are both changed
from -1 or >=0 at the same time.
As we leave min_pending and best_dist unchanged, any non-write-mostly
device will appear better than the write-mostly device.
Reported-by: Tomáš Hodek <tomas.hodek@volny.cz>
Reported-by: Dark Penguin <darkpenguin@yandex.ru>
Signed-off-by: NeilBrown <neilb@suse.de>
Link: http://marc.info/?l=linux-raid&m=135982797322422
Fixes: 9dedf60313
Cc: stable@vger.kernel.org (3.6+)
Add Byte Queue Limits (BQL) support to via-rhine driver.
[edumazet] tweaked patch and changed TX_RING_SIZE from 16 to 64
Signed-off-by: Tino Reichardt <milky-kernel@mcmilk.de>
Tested-by: Jamie Gloudon <jamie.gloudon@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The IP6_TNL_TRACE macro is no longer used anywhere in the code so remove definition.
Signed-off-by: Ian Morris <ipm@chirality.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
If the pending indexes are released /after/ pushing the Tx response
then a stale pending index may be used if a new Tx request is
immediately pushed by the frontend. The may cause various WARNINGs or
BUGs if the stale pending index is actually still in use.
Fix this by releasing the pending index before pushing the Tx
response.
The full barrier for the pending ring update is not required since the
the Tx response push already has a suitable write barrier.
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Wei Liu <wei.liu2@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Before da413eec72 ("packet: Fixed TPACKET V3 to signal poll when block is
closed rather than every packet") poll listening for an af_packet socket was
not signaled if there was no packets to process. After the patch poll is
signaled evety time when block retire timer expires. That happens because
af_packet closes the current block on timeout even if the block is empty.
Passing empty blocks to the user not only wastes CPU but also wastes ring
buffer space increasing probability of packets dropping on small timeouts.
Signed-off-by: Alexander Drozdov <al.drozdov@gmail.com>
Cc: Dan Collins <dan@dcollins.co.nz>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Guy Harris <guy@alum.mit.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mahesh Bandewar says:
====================
(Shortened) Bonding patch series
I'm shortening the earlier patch series to just two patches at this moment.
Andy is going to integrate remaining 3 patches with his netlink changes
and post the completed patches.
These two patches are the same one from the previous patch series with
some styling changes (Dave Miller's suggestions) for the churn-machine patch.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
The Churn Detection machines detect the situation where a port is operable,
but the Actor and Partner have not attached the link to an Aggregator and
brought the link into operation within a bound time period. Under normal
operation of the LACP, agreement between Actor and Partner should be reached
very rapidly. Continued failure to reach agreement can be symptomatic of
device failure.
Actor-churn-detection state-machine
Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com>
===================================
BEGIN=True + PortEnable=False
|
v
+------------------------+ ActorPort.Sync=True +------------------+
| ACTOR_CHURN_MONITOR | ---------------------> | NO_ACTOR_CHURN |
|========================| |==================|
| ActorChurn=False | ActorPort.Sync=False | ActorChurn=False |
| ActorChurn.Timer=Start | <--------------------- | |
+------------------------+ +------------------+
| ^
| |
ActorChurn.Timer=Expired |
| ActorPort.Sync=True
| |
| +-----------------+ |
| | ACTOR_CHURN | |
| |=================| |
+--------------> | ActorChurn=True | ------------+
| |
+-----------------+
Similar for the Partner-churn-detection.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 802.1AX standard states:
"The DA in LACPDUs is the Slow_Protocols_Multicast address."
This patch enforces that and drops LACPDUs with destination MAC
addresses other than Slow_Protocols_Multicast address
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Arrays (when not in a struct) "shall have a value greater than zero".
GCC complains when it's not the case here.
Fixes: ba7d49b1f0 ("rtnetlink: provide api for getting and setting slave info")
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 06d961a8e2 ("mac80211/minstrel: use the new rate control API")
inverted the condition 'if (msr->sample_limit != 0)' to
'if (!msr->sample_limit != 0)'. But it is confusing both to people and
compilers (gcc5):
net/mac80211/rc80211_minstrel.c: In function 'minstrel_get_rate':
net/mac80211/rc80211_minstrel.c:376:26: warning: logical not is only applied to the left hand side of comparison
if (!msr->sample_limit != 0)
^
Let there be only 'if (!msr->sample_limit)'.
Fixes: 06d961a8e2 ("mac80211/minstrel: use the new rate control API")
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Swap interrupt disable and thermal zone unregistration in the error and
remove paths, to make them more symmetrical with the initialization
path.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
As soon as the interrupt has been enabled by devm_request_irq(), the
interrupt routine may be called, depending on the current status of the
hardware.
However, at that point rcar_thermal_common hasn't been initialized
complely yet. E.g. rcar_thermal_common.base is still NULL, causing a
NULL pointer dereference:
Unable to handle kernel NULL pointer dereference at virtual address 0000000c
pgd = c0004000
[0000000c] *pgd=00000000
Internal error: Oops: 5 [#1] SMP ARM
CPU: 0 PID: 1 Comm: swapper/0 Not tainted 3.19.0-rc7-ape6evm-04564-gb6e46cb7cbe82389 #30
Hardware name: Generic R8A73A4 (Flattened Device Tree)
task: ee8953c0 ti: ee896000 task.ti: ee896000
PC is at rcar_thermal_irq+0x1c/0xf0
LR is at _raw_spin_lock_irqsave+0x48/0x54
Postpone the call to devm_request_irq() until all initialization has
been done to fix this.
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
When CONFIG_THERMAL is not enabled, it is better to introduce
equivalent dummy functions in the exported header than to
introduce #ifdeffery in drivers using the function.
This will prevent issues such as that reported in:
http://www.spinics.net/lists/linux-next/msg31573.html
While at it switch over to IS_ENABLED for thermal macros
to allow for thermal framework to be built as framework
and relevant APIs be usable by relevant drivers as a result.
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Fixes: http://bugs.elinux.org/issues/127
the bb.org community was seeing random reboots before this change.
Signed-off-by: Robert Nelson <robertcnelson@gmail.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
In commit 87517d26d8 ("ARM: dts: dra7-evm: Add extcon nodes for USB")
we enabled Extcon USB gpio to tackle the USB ID pin and get
peripheral mode to work.
But the extcon-gpio-usb driver [1] didn't make it into v4.0
and this makes the USB driver defer probe indefinitely breaking
USB Host functionality.
As a temporary fix we remove the extcon handle from the
USB controller and add it back when the extcon driver
merges in v4.1.
[1] - https://lkml.org/lkml/2015/2/2/187
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
SATA operation depends on PIPE3 PHY and if we want
to boot from SATA drives, we have to have the PIPE3 PHY
driver built-in.
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
According to the Documentation/devicetree/bindings/dma/dma.txt the
dma-channels and dma-requests property should not have '#'.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
According to the Documentation/devicetree/bindings/dma/dma.txt the
dma-channels and dma-requests property should not have '#'.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
According to the Documentation/devicetree/bindings/dma/dma.txt the
dma-channels and dma-requests property should not have '#'.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
According to the Documentation/devicetree/bindings/dma/dma.txt the
dma-channels and dma-requests property should not have '#'.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
According to the Documentation/devicetree/bindings/dma/dma.txt the
dma-channels and dma-requests property should not have '#'.
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
we have i2c0 sleep pinctrl state but were passing
default state anyhow. Fix that.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
This regulator is used on AM437x Industrial Development Kit.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
As it turns out, tps62362 is actually on I2C bus0,
not bus1. This has gone unnoticed because Linux
doesn't use (as of now) that regulator at all, it's
setup by the bootloader and left as is.
While at that, also add missing reg property for
our regulator.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Offset for smc91x must be zero otherwise smc91x linux kernel driver does not
detect smc91x ethernet hardware in qemu N900 machine.
The 0x300 offset was used to supress a warning the smsc911x
driver produces about non-standard offset as 0x300 seems to
be the EEPROM default. As only three address lines are
connected both 0 and 0x300 will work just fine with 0 being
correct. The warning about the non-standard offset can be
fixed by writing to EEPROM as that's needed in any case to
set the MAC address.
Signed-off-by: Pali Rohár <pali.rohar@gmail.com>
[tony@atomide.com: updated comments, just use 0 instead of 0x0]
Signed-off-by: Tony Lindgren <tony@atomide.com>
With legacy boot i2c buses on Nokia N900 are numbered i2c1, i2c2 and i2c3.
Commit 20b80942ef ("ARM: dts: OMAP3+: Add i2c aliases") fixed the
numbering with DT boot, but introduced a regression on N900 - aliases
become i2c0, i2c1 and i2c2. Fix that by providing the correct aliases in
the board dts.
Signed-off-by: Ivaylo Dimitrov <ivo.g.dimitrov.75@gmail.com>
Tested-by: Pali Rohár <pali.rohar@gmail.com>
Acked-by: Nishanth Menon <nm@ti.com>
[tony@atomide.com: this is needed for legacy user space to work]
Signed-off-by: Tony Lindgren <tony@atomide.com>
The cpufreq_cooling_unregister() function tests whether its argument is NULL
and then returns immediately. Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
Fix following build warning if CONFIG_PM_SLEEP is not set:
drivers/thermal/ti-soc-thermal/ti-bandgap.c:1478:12: warning: 'ti_bandgap_suspend' defined but not used [-Wunused-function]
static int ti_bandgap_suspend(struct device *dev)
^
drivers/thermal/ti-soc-thermal/ti-bandgap.c:1492:12: warning: 'ti_bandgap_resume' defined but not used [-Wunused-function]
static int ti_bandgap_resume(struct device *dev)
^
Acked-by: Nishanth Menon <nm@ti.com>
Signed-off-by: Grygorii Strashko <Grygorii.Strashko@linaro.org>
Signed-off-by: Eduardo Valentin <edubezval@gmail.com>
- Fix two regression introduced in 4.0-rc1 affecting PV/PVH guests in
certain configurations.
- Prevent pvscsi frontends bypassing backend checks.
- Allow privcmd hypercalls to be preempted even on kernel with
voluntary preemption. This fixes soft-lockups with long running
toolstack hypercalls (e.g., when creating/destroying large domains).
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
iQEcBAABAgAGBQJU7GDMAAoJEFxbo/MsZsTRj7AIAIFD2Dgu9p56nx6rUA6JOWP9
2O1VlINEkggUFNCekTJ3HkAjHreC2Ir67nk1PTTfL3ZN+PqxEv+IoUF1XKJxE/G3
0KchszK0MVnc7UBrUJWRUEpGo92iIC9QBc8MyDN6dGbfV5lT4aHB5ug/aSZqG8HV
8ITOftfwnVE8ukOK9bSRPZRejnoRMbGUmqpr0X17sE4Zb8+LEf0iO5Yh88NEjzxM
OD5hL3TGN5uXh6rYHS+U82M1g83G4Q7v10WIN5O96QhA6e5rVbVNsohHpL8IHY5q
kazw7wBY//SKGCGniACYnh+vZDeNddwCz8chj6S49h272zanMleyolrUjlcMGzQ=
=xmcy
-----END PGP SIGNATURE-----
Merge tag 'stable/for-linus-4.0-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip
Pull xen bugfixes from David Vrabel:
"Xen regression and bug fixes for 4.0-rc1
- Fix two regressions introduced in 4.0-rc1 affecting PV/PVH guests
in certain configurations.
- Prevent pvscsi frontends bypassing backend checks.
- Allow privcmd hypercalls to be preempted even on kernel with
voluntary preemption. This fixes soft-lockups with long running
toolstack hypercalls (e.g., when creating/destroying large
domains)"
* tag 'stable/for-linus-4.0-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
x86/xen: Initialize cr4 shadow for 64-bit PV(H) guests
xen-scsiback: mark pvscsi frontend request consumed only after last read
x86/xen: allow privcmd hypercalls to be preempted
x86/xen: Make sure X2APIC_ENABLE bit of MSR_IA32_APICBASE is not set
Pull livepatching fixes from Jiri Kosina:
"Two tiny fixes for livepatching infrastructure:
- extending RCU critical section to cover all accessess to
RCU-protected variable, by Petr Mladek
- proper format string passing to kobject_init_and_add(), by Jiri
Kosina"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/livepatching:
livepatch: RCU protect struct klp_func all the time when used in klp_ftrace_handler()
livepatch: fix format string in kobject_init_and_add()
Commit 7800064ba5 ("ARM: dts: Add basic dm816x device tree
configuration") added basic devices for dm816x, but I was not able
to test the USB completely because of an unconfigured USB phy, and
I only tested it to make sure the Mentor chips are detected and
clocked without a phy.
After testing the USB with actual devices I noticed a few issues
that should be fixed to avoid confusion:
- The USB id pin on dm8168-evm is hardwired and can be changed
only by software. As there are two USB-A type connectors, let's
start both in host mode instead of otg.
- The Mentor core is configured in such a way on dm8168-evm that
it's not capable of multipoint at least on revision c board
that I have.
- We need ranges for the syscon to properly set up the phy as
children of the SCM syscon area.
- Let's not disable the second interface, the board specific
dts files can do that if really needed. Most boards should
just keep it enabled to ensure the device is idled properly.
Note that also a phy and several musb fixes are still needed to
make the USB to work properly in addition to this fix.
Cc: Brian Hutchinson <b.hutchman@gmail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates 2015-02-23
This series contains updates to e1000e, igbvf, i40e and i40evf.
David adds support for i219 devices to the e1000e driver.
Jeff (me) provides two patches to cleanup igbvf, first cleans up the
whitespace issues found and the second cleans up the usage of msleep(),
min() and max() with usleep_range(), min_t() and max_t() respectively.
Kamil updates the shadow RAM read/write functions by refactoring them
to prepare for future work.
Shannon renames the debugfs command "clear_stats pf" to clear_stats port"
to clarify what the function really does.
Mitch refactors the receive routine, by splitting the receive hot path
code into two, one for packet split and one for single buffer, which
improves receive performance. Disables NAPI polling sooner when closing
the interface to fix an occasional panic during close which was
caused by the driver trying to delete and clean rings at the same time.
Also refactors reset for i40evf, since a recent change to the shutdown
flow messed up the reset flow. Since i40evf_down() now holds the
critical section lock, we cannot call it from the reset handler, which
also holds the lock.
Nicholas restricts the virtual channel opcodes should remain consistent
between updates to the opcode enum.
Neerav converts the VSI connection type to use a #define instead of
using a magic number.
Anjali updates the registers file to remove registers no longer available.
Also fixes the EMPR interrupt handling, so that we won't trigger another
EMPR when we receive an EMPR event.
Catherine cleans up the variable an_enable since it was set and never
used.
Greg fixes the netdev op that allows the operator to turn MAC/VLAN
spoof checking on and off so that it includes the flag for VLAN spoof
checking.
v2: Updated patch #10 in the series to use test_and_clear_bit() as
suggested by Sergei Shtylyov
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Petri Gynther says:
====================
net: bcmgenet: Tx init improvements
Four small patches to improve bcmgenet Tx init:
1. bcmgenet_init_tx_ring() cleanup
2. rework Tx queue init
3. precalculate TxCB->bd_addr
4. rename bcmgenet_hw_params->bds_cnt and GENET_DEFAULT_BD_CNT
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
bcmgenet_hw_params->bds_cnt and GENET_DEFAULT_BD_CNT are used only in Tx init.
Rename them accordingly:
- bcmgenet_hw_params->bds_cnt => bcmgenet_hw_params->tx_bds_per_q
- GENET_DEFAULT_BD_CNT => GENET_Q16_TX_BD_CNT
Signed-off-by: Petri Gynther <pgynther@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There is 1-to-1 mapping between TxCBs and TxBDs. Precalculate TxCB->bd_addr
once in bcmgenet_init_dma() instead of doing it over and over needlessly in
bcmgenet_get_txcb().
Signed-off-by: Petri Gynther <pgynther@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1. Rename bcmgenet_init_multiq() to bcmgenet_init_tx_queues()
2. Fix bcmgenet_init_tx_queues() function description
3. Move Tx default queue init inside bcmgenet_init_tx_queues()
4. Modify bcmgenet_init_dma() to call bcmgenet_init_tx_queues()
Signed-off-by: Petri Gynther <pgynther@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1. Simplify function description
2. Rename function parameter write_ptr to start_ptr to better indicate use
3. Remove unnecessary local variable first_bd
4. Remove out-of-place comment "Unclassified traffic goes to ring 16"
5. Fix TDMA_WRITE_PTR register init
Signed-off-by: Petri Gynther <pgynther@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull HID fixes from Jiri Kosina:
- a few fixes to Sony driver (rmmod breakage, spinlock initialization),
by Antonio Ospite, Frank Praznik and Jiri Kosina
- fix for wMaxInputLength handling regression in i2c-hid, by Seth
Forshee
- IRQ safety spinlock fix in sensor hub driver, by Srinivas Pandruvada
- IRQ level sensitivity fix to i2c-hid to be compliant with the spec,
by Mika Westerberg
- a couple device ID additions piggy-backing on top of that
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
HID: microsoft: Add ID for NE7K wireless keyboard
HID: i2c-hid: Limit reads to wMaxInputLength bytes for input events
HID: sony: fix uninitialized per-controller spinlock
HID: sony: initialize sony_dev_list_lock properly
HID: sony: Fix a WARNING shown when rmmod-ing the driver
HID: sensor-hub: correct dyn_callback_lock IRQ-safe change
HID: hid-sensor-hub: Correct documentation
HID: saitek: add USB ID for older R.A.T. 7
HID: i2c-hid: The interrupt should be level sensitive
HID: wacom: Add missing ABS_MISC event and feature declaration for 27QHD
The sata_ref_clk is a reference clock to the SATA phy.
This fixes SATA malfunction across suspend/resume or when
SATA driver is used as a module.
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
The sata_ref_clk is a reference clock to the SATA phy.
This fixes SATA malfunction across suspend/resume or when
SATA driver is used as a module.
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Some bios really like to joke and start the planes at an offset ...
hooray!
Align start and end to fix this.
v2: Fixup calculation of size, spotted by Chris Wilson.
v3: Fix serious fumble I've just spotted.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=86883
Cc: stable@vger.kernel.org
Cc: Johannes W <jargon@molb.org>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Reported-and-tested-by: Johannes W <jargon@molb.org>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
[Jani: split WARN_ONs, rebase on v4.0-rc1]
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Atm, it's possible that the interrupt handler is called when the device
is in D3 or some other low-power state. It can be due to another device
that is still in D0 state and shares the interrupt line with i915, or on
some platforms there could be spurious interrupts even without sharing
the interrupt line. The latter case was reported by Klaus Ethgen using a
Lenovo x61p machine (gen 4). He noticed this issue via a system
suspend/resume hang and bisected it to the following commit:
commit e11aa36230
Author: Jesse Barnes <jbarnes@virtuousgeek.org>
Date: Wed Jun 18 09:52:55 2014 -0700
drm/i915: use runtime irq suspend/resume in freeze/thaw
This is a problem, since in low-power states IIR will always read
0xffffffff resulting in an endless IRQ servicing loop.
Fix this by handling interrupts only when the driver explicitly enables
them and so it's guaranteed that the interrupt registers return a valid
value.
Note that this issue existed even before the above commit, since during
runtime suspend/resume we never unregistered the handler.
v2:
- clarify the purpose of smp_mb() vs. synchronize_irq() in the
code comment (Chris)
v3:
- no need for an explicit smp_mb(), we can assume that synchronize_irq()
and the mmio read/writes in the install hooks provide for this (Daniel)
- remove code comment as the remaining synchronize_irq() is self
explanatory (Daniel)
v4:
- drm_irq_uninstall() implies synchronize_irq(), so no need to call it
explicitly (Daniel)
Reference: https://lkml.org/lkml/2015/2/11/205
Reported-and-bisected-by: Klaus Ethgen <Klaus@Ethgen.ch>
Cc: stable@vger.kernel.org
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
When we walk the list of vma, or even for protecting against concurrent
framebuffer creation, we must hold the struct_mutex or else a second
thread can corrupt the list as we walk it.
Fixes regression from
commit d7f46fc4e7
Author: Ben Widawsky <benjamin.widawsky@intel.com>
Date: Fri Dec 6 14:10:55 2013 -0800
drm/i915: Make pin count per VMA
References: https://bugs.freedesktop.org/show_bug.cgi?id=89085
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: stable@vger.kernel.org
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
When converting from implicitly tracked execlist queue items to ref counted
requests, not all frees of requests were replaced with unrefs, and extraneous
refs/unrefs of contexts were added.
Correct the unbalanced refcount & replace the frees.
Remove a noisy warning when hitting the request creation path.
drm_i915_gem_request and intel_context are both kref reference counted
structures. Upon allocation, drm_i915_gem_request's ref count should be
bumped using kref_init. When a context is assigned to the request,
the context's reference count should be bumped using i915_gem_context_reference.
i915_gem_request_reference will reduce the context reference count when
the request is freed.
Problem introduced in
commit 6d3d8274bc
Author: Nick Hoath <nicholas.hoath@intel.com>
AuthorDate: Thu Jan 15 13:10:39 2015 +0000
drm/i915: Subsume intel_ctx_submit_request in to drm_i915_gem_request
v2: Added comments explaining how the ctx pointer and the request object should
be ref-counted. Removed noisy warning.
v3: Cleaned up the language used in the commit & the header
description (Thanks David Gordon)
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=88652
Signed-off-by: Nick Hoath <nicholas.hoath@intel.com>
Reviewed-by: Thomas Daniel <thomas.daniel@intel.com>
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
