We can't use dev->mod_index for selecting the interrupt routing entry,
because it's not an index into interrupt routing table. It will be even
wrong on a machine with 2 CPUs (4 cores). But all needed information is
contained in the PAT entries for the serial ports. mod[0] contains the
iosapic address and mod_info has some indications for the interrupt
input (at least it looks like it). This patch implements the searching
for the right iosapic and uses this interrupt input information.
The parisc architecture does not have a pte special bit. As a result,
special mappings are handled with the VM_PFNMAP and VM_MIXEDMAP flags.
VM_MIXEDMAP mappings may or may not have a "struct page" backing. When
pfn_valid() is false, there is no "struct page" backing. Otherwise, they
are treated as normal pages.
The FireGL driver uses the VM_MIXEDMAP without a backing "struct page".
This treatment caused a panic due to a TLB data miss in
update_mmu_cache. This appeared to be in the code generated for
page_address(). We were in fact using a very circular bit of code to
determine the physical address of the PFN in various cache routines.
This wasn't valid when there was no "struct page" backing. The needed
address can in fact be determined simply from the PFN itself without
using the "struct page".
The attached patch updates update_mmu_cache(), flush_cache_mm(),
flush_cache_range() and flush_cache_page() to check pfn_valid() and to
directly compute the PFN physical and virtual addresses.
Signed-off-by: John David Anglin <dave.anglin@bell.net> Signed-off-by: Helge Deller <deller@gmx.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The system doesn't more fail to bind the memory, and hence not falling
back to the PCI mode (if other failures aren't detected).
This is just a simple write down from the following patches:
agp/amd-k7: Allow binding user memory to the AGP GART
agp/hp-agp: Allow binding user memory to the AGP GART
When an associativity level change is found for one thread, the
siblings threads need to be updated as well. This is done today
for PRRN in stage_topology_update() but is missing for VPHN in
update_cpu_associativity_changes_mask(). This patch will correctly
update all thread siblings during a topology change.
Without this patch a topology update can result in a CPU in
init_sched_groups_power() getting stuck indefinitely in a loop.
This loop is built in build_sched_groups(). As a result of the thread
moving to a node separate from its siblings the struct sched_group will
have its next pointer set to point to itself rather than the sched_group
struct of the next thread. This happens because we have a domain without
the SD_OVERLAP flag, which is correct, and a topology that doesn't conform
with reality (threads on the same core assigned to different numa nodes).
When this list is traversed by init_sched_groups_power() it will reach
the thread's sched_group structure and loop indefinitely; the cpu will
be stuck at this point.
The bug was exposed when VPHN was enabled in commit b7abef0 (v3.9).
Reported-by: Jan Stancek <jstancek@redhat.com> Signed-off-by: Robert Jennings <rcj@linux.vnet.ibm.com> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
a.out support on ARM requires that argc, argv and envp are passed in
r0-r2 respectively, which requires hacking load_aout_binary to
prevent argc being clobbered by the return code. Whilst mainline kernels
do set the registers up in start_thread, the aout loader has never
carried the hack in mainline.
Initialising the registers in this way actually goes against the libc
expectations for ELF binaries, where argc, argv and envp are passed on
the stack, with r0 being used to hold a pointer to an exit function for
cleaning up after the dynamic linker if required. If the pointer is
NULL, then it is ignored. When execing an ELF binary, Linux currently
zeroes r0, then sets it to argc and then finally clobbers it with the
return value of the execve syscall, so we actually end up with:
libc treats r1 and r2 as undefined. The clobbering of r0 by sys_execve
works for user-spawned threads, but when executing an ELF binary from a
kernel thread (via call_usermodehelper), the execve is performed on the
ret_from_fork path, which restores r0 from the saved pt_regs, resulting
in argc being presented to the C library. This has horrible consequences
when the application exits, since we have an exit function registered
using argc, resulting in a jump to hyperspace.
This patch solves the problem by removing the partial a.out support from
arch/arm/ altogether.
Signed-off-by: Will Deacon <will.deacon@arm.com> Cc: Ashish Sangwan <ashishsangwan2@gmail.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As of commit b9d4d42ad9 (ARM: Remove __ARCH_WANT_INTERRUPTS_ON_CTXSW on
pre-ARMv6 CPUs), the mm switching on VIVT processors is done in the
finish_arch_post_lock_switch() function to avoid whole cache flushing
with interrupts disabled. The need for deferred mm switch is stored as a
thread flag (TIF_SWITCH_MM). However, with preemption enabled, we can
have another thread switch before finish_arch_post_lock_switch(). If the
new thread has the same mm as the previous 'next' thread, the scheduler
will not call switch_mm() and the TIF_SWITCH_MM flag won't be set for
the new thread.
This patch moves the switch pending flag to the mm_context_t structure
since this is specific to the mm rather than thread.
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com> Reported-by: Marc Kleine-Budde <mkl@pengutronix.de> Tested-by: Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit ae8a8b9553bd ("ARM: 7691/1: mm: kill unused TLB_CAN_READ_FROM_L1_CACHE
and use ALT_SMP instead") added early function returns for page table
cache flushing operations on ARMv7 SMP CPUs.
Unfortunately, when targetting Thumb-2, these `mov pc, lr' sequences
assemble to 2 bytes which can lead to corruption of the instruction
stream after code patching.
This patch fixes the alternates to use wide (32-bit) instructions for
Thumb-2, therefore ensuring that the patching code works correctly.
Signed-off-by: Will Deacon <will.deacon@arm.com> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
slots-fan on G5 Xserve is always running at full speed with windfarm_rm31
driver, resulting in a very high acoustic noise level. It seems the fan
parameters are incorrect, and have been copied from the Drive Bay fan
(RPM, not present on rm31) of the legacy therm_pm72 driver. This patch
changes the parameters to match the Slots fan (PWM) of therm_pm72. With
the patch, slots-fan speed drops from 99% to 19% during normal use,
and slots-temp settle to ~42'C.
Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
arch/arm/kernel/signal.c: In function 'setup_return':
arch/arm/kernel/signal.c:413:25: error: 'mm_context_t' has no member named 'sigpage'
This shows one of the evilnesses of IS_ENABLED(). Get rid of it here
and replace it with #ifdef's - and as no noMMU platform can make use
of sigpage, depend on CONIFG_MMU not CONFIG_ARM_MPU.
Reported-by: Olof Johansson <olof@lixom.net> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Unfortunately, I never committed the fix to a nasty oops which can
occur as a result of that commit:
------------[ cut here ]------------
kernel BUG at /home/olof/work/batch/include/linux/mm.h:414!
Internal error: Oops - BUG: 0 [#1] PREEMPT SMP ARM
Modules linked in:
CPU: 0 PID: 490 Comm: killall5 Not tainted 3.11.0-rc3-00288-gabe0308 #53
task: e90acac0 ti: e9be8000 task.ti: e9be8000
PC is at special_mapping_fault+0xa4/0xc4
LR is at __do_fault+0x68/0x48c
This doesn't show up unless you do quite a bit of testing; a simple
boot test does not do this, so all my nightly tests were passing fine.
The reason for this is that install_special_mapping() expects the
page array to stick around, and as this was only inserting one page
which was stored on the kernel stack, that's why this was blowing up.
Reported-by: Olof Johansson <olof@lixom.net> Tested-by: Olof Johansson <olof@lixom.net> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If kuser helpers are not provided by the kernel, disable user access to
the vectors page. With the kuser helpers gone, there is no reason for
this page to be visible to userspace.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Move the signal handlers into a VDSO page rather than keeping them in
the vectors page. This allows us to place them randomly within this
page, and also map the page at a random location within userspace
further protecting these code fragments from ROP attacks. The new
VDSO page is also poisoned in the same way as the vector page.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Provide a kernel configuration option to allow the kernel user helpers
to be removed from the vector page, thereby preventing their use with
ROP (return orientated programming) attacks. This option is only
visible for CPU architectures which natively support all the operations
which kernel user helpers would normally provide, and must be enabled
with caution.
Acked-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
FIQ should no longer copy the FIQ code into the user visible vector
page. Instead, it should use the hidden page. This change makes
that happen.
Acked-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use linker magic to create the vectors and vector stubs: we can tell the
linker to place them at an appropriate VMA, but keep the LMA within the
kernel. This gets rid of some unnecessary symbol manipulation, and
have the linker calculate the relocations appropriately.
Acked-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Move the machine vector stubs into the page above the vector page,
which we can prevent from being visible to userspace. Also move
the reset stub, and place the swi vector at a location that the
'ldr' can get to it.
This hides pointers into the kernel which could give valuable
information to attackers, and reduces the number of exploitable
instructions at a fixed address.
Acked-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Poison the memory between each kuser helper. This ensures that any
branch between the kuser helpers will be appropriately trapped.
Acked-by: Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fill the empty regions of the vectors page with an exception generating
instruction. This ensures that any inappropriate branch to the vector
page is appropriately trapped, rather than just encountering some code
to execute. (The vectors page was filled with zero before, which
corresponds with the "andeq r0, r0, r0" instruction - a no-op.)
Acked-by Nicolas Pitre <nico@linaro.org> Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We have problems in arch/x86/kernel/cpu/mtrr/generic.c::generic_get_mtrr().
1. for base, we miss cast base_lo to 64bit before shifting.
Fix that by adding u64 casting.
2. for size, it only can handle 44 bits aka 32bits + page_shift
Fix that with 64bit mask instead of 32bit mask_lo, then range could be
more than 44bits.
At the same time, we need to update size_or_mask for old cpus that does
support cpuid 0x80000008 to get phys_addr. Need to set high 32bits
to all 1s, otherwise will not get correct size for them.
Also fix mtrr_add_page: it should check base and (base + size - 1)
instead of base and size, as base and size could be small but
base + size could bigger enough to be out of boundary. We can
use boot_cpu_data.x86_phys_bits directly to avoid size_or_mask.
So When are we going to have size more than 44bits? that is 16TiB.
attempts to aid the user by tapping into an existing error message,
as described in the commit message:
... Also fix an issue where _get_attempt was called with only
one argument. This prevented the error message from printing
the name of the variable that can be used to fix the problem.
However, The "missing" argument was in fact missing on purpose; it's
absence is a signal that the error message should be skipped, because
the failure would be due to the default value, not any user-supplied
value. This can be seen in how `_ge_attempt' uses `gea_err' (in the
config/utilities.mak file):
_ge_attempt = $(if $(get-executable),$(get-executable),$(_gea_warn)$(call _gea_err,$(2)))
_gea_warn = $(warning The path '$(1)' is not executable.)
_gea_err = $(if $(1),$(error Please set '$(1)' appropriately))
That is, because the argument is no longer missing, the value `$(1)'
(associated with `_gea_err') always evaluates to true, thus always
triggering the error condition that is meant to be reserved for
only the case when a user explicitly supplies an invalid value.
Concretely, the result is a regression in the Makefile's configuration
of python support; rather than gracefully disable support when the
relevant executables cannot be found according to default values, the
build process halts in error as though the user explicitly supplied
the values.
This new commit simply reverts the offending one-line change.
This patch changes iscsit_add_reject() + iscsit_add_reject_from_cmd()
usage to not sleep on iscsi_cmd->reject_comp to address a free-after-use
usage bug in v3.10 with iser-target code.
It saves ->reject_reason for use within iscsit_build_reject() so the
correct value for both transport cases. It also drops the legacy
fail_conn parameter usage throughput iscsi-target code and adds
two iscsit_add_reject_cmd() and iscsit_reject_cmd helper functions,
along with various small cleanups.
(v2: Re-enable target_put_sess_cmd() to be called from
iscsit_add_reject_from_cmd() for rejects invoked after
target_get_sess_cmd() has been called)
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org> Cc: Or Gerlitz <ogerlitz@mellanox.com> Cc: Mike Christie <michaelc@cs.wisc.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix a warning from lockdep caused by calling flush_work() for
uninitialized hotplug work. Initialize hotplug_work, audio_work
and reset_work upon successful radeon_irq_kms_init() completion
and thus perform hotplug flush_work only when rdev->irq.installed
is true.
[ 4.790019] [drm] Loading CEDAR Microcode
[ 4.790943] r600_cp: Failed to load firmware "radeon/CEDAR_smc.bin"
[ 4.791152] [drm:evergreen_startup] *ERROR* Failed to load firmware!
[ 4.791330] radeon 0000:01:00.0: disabling GPU acceleration
[ 4.792633] INFO: trying to register non-static key.
[ 4.792792] the code is fine but needs lockdep annotation.
[ 4.792953] turning off the locking correctness validator.
Unbinding an event channel (either with the ioctl or when the evtchn
device is closed) may deadlock because disable_irq() is called with
port_user_lock held which is also locked by the interrupt handler.
Think of the IOCTL_EVTCHN_UNBIND is being serviced, the routine has
just taken the lock, and an interrupt happens. The evtchn_interrupt
is invoked, tries to take the lock and spins forever.
A quick glance at the code shows that the spinlock is a local IRQ
variant. Unfortunately that does not help as "disable_irq() waits for
the interrupt handler on all CPUs to stop running. If the irq occurs
on another VCPU, it tries to take port_user_lock and can't because
the unbind ioctl is holding it." (from David). Hence we cannot
depend on the said spinlock to protect us. We could make it a system
wide IRQ disable spinlock but there is a better way.
We can piggyback on the fact that the existence of the spinlock is
to make get_port_user() checks be up-to-date. And we can alter those
checks to not depend on the spin lock (as it's protected by u->bind_mutex
in the ioctl) and can remove the unnecessary locking (this is
IOCTL_EVTCHN_UNBIND) path.
In the interrupt handler we cannot use the mutex, but we do not
need it.
"The unbind disables the irq before making the port user stale, so when
you clear it you are guaranteed that the interrupt handler that might
use that port cannot be running." (from David).
Hence this patch removes the spinlock usage on the teardown path
and piggybacks on disable_irq happening before we muck with the
get_port_user() data. This ensures that the interrupt handler will
never run on stale data.
Signed-off-by: David Vrabel <david.vrabel@citrix.com> Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
[v1: Expanded the commit description a bit] Signed-off-by: Jonghwan Choi <jhbird.choi@samsung.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Eric Sandeen has found a nasty livelock in sget() - take a mount(2) about
to fail. The superblock is on ->fs_supers, ->s_umount is held exclusive,
->s_active is 1. Along comes two more processes, trying to mount the same
thing; sget() in each is picking that superblock, bumping ->s_count and
trying to grab ->s_umount. ->s_active is 3 now. Original mount(2)
finally gets to deactivate_locked_super() on failure; ->s_active is 2,
superblock is still ->fs_supers because shutdown will *not* happen until
->s_active hits 0. ->s_umount is dropped and now we have two processes
chasing each other:
s_active = 2, A acquired ->s_umount, B blocked
A sees that the damn thing is stillborn, does deactivate_locked_super()
s_active = 1, A drops ->s_umount, B gets it
A restarts the search and finds the same superblock. And bumps it ->s_active.
s_active = 2, B holds ->s_umount, A blocked on trying to get it
... and we are in the earlier situation with A and B switched places.
The root cause, of course, is that ->s_active should not grow until we'd
got MS_BORN. Then failing ->mount() will have deactivate_locked_super()
shut the damn thing down. Fortunately, it's easy to do - the key point
is that grab_super() is called only for superblocks currently on ->fs_supers,
so it can bump ->s_count and grab ->s_umount first, then check MS_BORN and
bump ->s_active; we must never increment ->s_count for superblocks past
->kill_sb(), but grab_super() is never called for those.
The bug is pretty old; we would've caught it by now, if not for accidental
exclusion between sget() for block filesystems; the things like cgroup or
e.g. mtd-based filesystems don't have anything of that sort, so they get
bitten. The right way to deal with that is obviously to fix sget()...
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The name of udc state attribute file under sysfs is registered as
"state", while usb_gadget_set_state take it as "status" when it's
going to update. This patch fixes the typo.
Signed-off-by: Rong Wang <Rong.Wang@csr.com> Signed-off-by: Barry Song <Baohua.Song@csr.com> Signed-off-by: Felipe Balbi <balbi@ti.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
RT Systems makes many usb serial cables based on the ftdi_sio driver for
programming various amateur radios. This patch is a full listing of
their current product offerings and should allow these cables to all
be recognized.
Signed-off-by: Rick Farina (Zero_Chaos) <zerochaos@gentoo.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a269913c5 entitled "rtlwifi: Rework rtl_lps_leave() and
rtl_lps_enter() to use work queue" has two bugs for USB drivers.
Firstly, the work queue in question was not initialized. Secondly,
the callback routine used by this queue is contained within the
file used for PCI devices. As a result, it is not available for
architectures without PCI hardware.
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net> Reported-by: Richard Genoud <richard.genoud@gmail.com> Tested-by: Richard Genoud <richard.genoud@gmail.com> Cc: Richard Genoud <richard.genoud@gmail.com> Signed-off-by: John W. Linville <linville@tuxdriver.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The ProcessAuxChannel table on some rv635 boards assumes
the divmul members are initialized to 0 otherwise we get
an invalid fb offset since it has a bad mask set when
setting the fb base. While here initialize all the
atom interpretor elements to 0.
We need to set the dto source before setting the
dividers otherwise we may get stability problems
with the dto leading to audio playback problems.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Upon some code refactoring, a hunk was missed. This was fixed for
next, but missed the current trees, and hasn't yet been merged by Dave
Airlie. It is fixed in:
commit 907b28c56ea40629aa6595ddfa414ec2fc7da41c
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Fri Jul 19 20:36:52 2013 +0100
drm/i915: Colocate all GT access routines in the same file
Reported-by: Dave Jones <davej@redhat.com> Signed-off-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Dave Airlie <airlied@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/i915: fix long-standing SNB regression in power consumption after resume
unintentionally also changed the init sequence ordering between
gt_init and gt_reset - we need to reset BIOS damage like leftover
forcewake references before we run our own code. Otherwise we can get
nasty dmesg noise like
[drm:__gen6_gt_force_wake_mt_get] *ERROR* Timed out waiting for forcewake old ack to clear.
again. Since _reset suggests that we first need to have stuff
initialized (which isn't the case here) call it sanitze instead.
While at it also block out the rps disable introduced by the above
commit on ilk: We don't have any knowledge of ilk rps being broken in
similar ways. And the disable functions uses the default hw state
which is only read out when we're enabling rps. So essentially we've
been writing random grabage into that register.
Reported-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Konstantin Khlebnikov <khlebnikov@openvz.org> Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Tested-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In theory, the different register blocks were meant to be only ever
touched when holding either the struct_mutex, mode_config.lock or even a
specific localised lock. This does not seem to be the case, and the
hardware reacts extremely badly if we attempt to concurrently access two
registers within the same cacheline.
The HSD suggests that we only need to do this workaround for display
range registers. However, upon review we need to serialize the multiple
stages in our register write functions - if only for preemption
protection.
Irrespective of the hardware requirements, the current io functions are
a little too loose with respect to the combination of pre- and
post-condition testing that we do in conjunction with the actual io. As
a result, we may be pre-empted and generate both false-postive and
false-negative errors.
Note well that this is a "90%" solution, there remains a few direct
users of ioread/iowrite which will be fixed up in the next few patches.
Since they are more invasive and that this simple change will prevent
almost all lockups on Haswell, we kept this patch simple to facilitate
backporting to stable.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=63914 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Tested-by: Eric Griffith <EGriffith92@gmail.com> Tested-by: Kent Baxley <kent.baxley@canonical.com> Signed-off-by: Kamal Mostafa <kamal@canonical.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To avoid stalls we delay tiling changes and especially hold of
committing the new fence state for as long as possible.
Synchronization points are in the execbuf code and in our gtt fault
handler.
Unfortunately we've missed that tricky detail when adding proper fence
restore code in
drm/i915: Restore fences after resume and GPU resets
The result was that we've restored fences for objects with no tiling,
since the object<->fence link still existed after resume. Now that
wouldn't have been too bad since any subsequent access would have
fixed things up, but if we've changed from tiled to untiled real havoc
happened:
The tiling stride is stored -1 in the fence register, so a stride of 0
resulted in all 1s in the top 32bits, and so a completely bogus fence
spanning everything from the start of the object to the top of the
GTT. The tell-tale in the register dumps looks like:
FENCE START 2: 0x0214d001
FENCE END 2: 0xfffff3ff
Bit 11 isn't set since the hw doesn't store it, even when writing all
1s (at least on my snb here).
To prevent such a gaffle in the future add a sanity check for fences
with an untiled object attached in i915_gem_write_fence.
v2: Fix the WARN, spotted by Chris.
v3: Trying to reuse get_fences looked ugly and obfuscated the code.
Instead reuse update_fence and to make it really dtrt also move the
fence dirty state clearing into update_fence.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=60530 Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Stéphane Marchesin <marcheu@chromium.org> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Tested-by: Matthew Garrett <matthew.garrett@nebula.com> Tested-by: Björn Bidar <theodorstormgrade@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/i915: force full modeset if the connector is in DPMS OFF mode
a new function was added that walked over the set of connectors to see
if any of the currently associated CRTC was switched off. This function
walked an array of connectors, rather than the array of pointers to
connectors contained in the drm_mode_set - i.e. it was dereferencing far
past the end of the first connector. This only becomes an issue if we
attempt to use a clone mode (i.e. more than one connector per CRTC) such
that set->num_connectors > 1.
Reported-by: Timo Aaltonen <tjaalton@ubuntu.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=65927 Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Imre Deak <imre.deak@intel.com> Cc: Egbert Eich <eich@suse.de> Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This patch fixes regression in power consumtion of sandy bridge gpu, which
exists since v3.6 Sometimes after resuming from s2ram gpu starts thinking that
it's extremely busy. After that it never reaches rc6 state.
For some reason RC6 is already enabled at the beginning of resuming process.
Following initliaztion breaks some internal state and confuses RPS engine.
This patch disables RC6 at the beginnig of resume and initialization.
I've rearranged initialization sequence, because intel_disable_gt_powersave()
needs initialized force_wake_get/put and some locks from the dev_priv.
Note: The culprit in the initialization sequence seems to be the write
to MBCTL added in the above mentioned commit. The first version of
this patch just held a forcewake reference across the clock gating
init functions, which seems to have been enought to gather quite a few
positive test reports. But since that smelled a bit like ad-hoc
duct-tape v2 now just disables rps/rc6 across the entire hw setup.
[danvet: Add note about v1 vs. v2 of this patch and use standard
layout for the commit citation. Also add the tested-bys from v1 and a cc:
stable.]
drm/i915: Workaround incoherence between fences and LLC across multiple CPUs
Thanks to further investigation by Jon Bloomfield, he realised that
the 64-bit register might be broken up by the hardware into two 32-bit
writes (a problem we have encountered elsewhere). This non-atomicity
would then cause an issue where a second thread would see an
intermediate register state (new high dword, old low dword), and this
register would randomly be used in preference to its own thread register.
This would cause the second thread to read from and write into a fairly
random tiled location. Breaking the operation into 3 explicit 32-bit
updates (first disable the fence, poke the upper bits, then poke the lower
bits and enable) ensures that, given proper serialisation between the
32-bit register write and the memory transfer, that the fence value is
always consistent.
Armed with this knowledge, we can explain how the previous workaround
work. The key to the corruption is that a second thread sees an
erroneous fence register that conflicts and overrides its own. By
serialising the fence update across all CPUs, we have a small window
where no GTT access is occurring and so hide the potential corruption.
This also leads to the conclusion that the earlier workaround was
incomplete.
v2: Be overly paranoid about the order in which fence updates become
visible to the GPU to make really sure that we turn the fence off before
doing the update, and then only switch the fence on afterwards.
Signed-off-by: Jon Bloomfield <jon.bloomfield@intel.com> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Carsten Emde <C.Emde@osadl.org> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The original commit results in repeated 'Timed out waiting for forcewake old
ack to clear' messages on a Supermicro C7H61 board (BIOS version 2.00 and 2.00b)
with i7-3770K CPU. It ultimately results in a hangup if the system is highly
loaded. Reverting the commit for IvyBridge CPUs fixes the issue.
Issue a warning if the CPU is IvyBridge and mt forcewake is disabled, since
this condition can result in secondary issues.
v2: Only revert patch for Ivybridge CPUs
Issue info message if mt forcewake is disabled on Ivybridge
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=60541 Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Cc: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Mika Kuoppala <mika.kuoppala@intel.com> Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=66139 Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Daniel noticed a problem where is we wrote to an object with ring A in
the middle of a very long running batch, then executed a quick batch on
ring B before a batch that reads from the same object, its obj->ring would
now point to ring B, but its last_write_seqno would be still relative to
ring A. This would allow for the user to read from the object before the
GPU had completed the write, as set_domain would only check that ring B
had passed the last_write_seqno.
To fix this simply (and inelegantly), we bump the last_write_seqno when
switching rings so that the last_write_seqno is always relative to the
current obj->ring.
This fixes igt/tests/gem_write_read_ring_switch.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
[danvet: Add note about the newly created igt which exercises this bug.] Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drm/i915: Implement workaround for broken CS tlb on i830/845
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=64610 Cc: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Noticed that my old Radeon 7500 hung after printing
drm: GPU not posted. posting now...
when it wasn't selected as the primary card the BIOS. Some digging
revealed that it was hanging in combios_parse_mmio_table() while
parsing the ASIC INIT 3 table. Looking at the BIOS ROM for the card,
it becomes obvious that there is no ASIC INIT 3 table in the BIOS.
The code is just processing random garbage. No surprise it hangs!
Why do I say that there is no ASIC INIT 3 table is the BIOS? This
table is found through the MISC INFO table. The MISC INFO table can
be found at offset 0x5e in the COMBIOS header. But the header is
smaller than that. The COMBIOS header starts at offset 0x126. The
standard PCI Data Structure (the bit that starts with 'PCIR') lives at
offset 0x180. That means that the COMBIOS header can not be larger
than 0x5a bytes and therefore cannot contain a MISC INFO table.
I looked at a dozen or so BIOS images, some my own, some downloaded from:
It is fairly obvious that the size of the COMBIOS header can be found
at offset 0x6 of the header. Not sure if it is a 16-bit number or
just an 8-bit number, but that doesn't really matter since the tables
seems to be always smaller than 256 bytes.
So I think combios_get_table_offset() should check if the requested
table is present. This can be done by checking the offset against the
size of the header. See the diff below. The diff is against the WIP
OpenBSD codebase that roughly corresponds to Linux 3.8.13 at this
point. But I don't think this bit of the code changed much since
then.
For what it is worth:
Signed-off-by: Mark Kettenis <kettenis@openbsd.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The patch below fixes the problem for this card.
But I don't like the blacklist, couldn't some heuristic be used instead?
The interesting thing is that the manufacturer is the same as the other card
needing the same quirk. I wonder how many different types are broken this way.
The "wrong" ps2_pdac_adj value that comes from BIOS on this card is 0x300.
====================
drm/radeon: Add primary dac adj quirk for Sapphire Radeon VE 7000 DDR
Values from BIOS are wrong, causing too bright colors.
Use default values instead.
Signed-off-by: Ondrej Zary <linux@rainbow-software.org> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently doesn't matter cause we allocate the fence in the
lower 265MB anyway.
Reported-by: Frank Huang <FrankR.Huang@amd.com> Signed-off-by: Christian König <christian.koenig@amd.com> Signed-off-by: Alex Deucher <alexander.deucher@amd.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Fix a boundary condition that caused failure for certain device sizes.
The problem is reported at
http://code.google.com/p/cryptsetup/issues/detail?id=160
For certain device sizes the number of hashes at a specific level was
calculated incorrectly.
It happens for example for a device with data and metadata block size 4096
that has 16385 blocks and algorithm sha256.
The user can test if he is affected by this bug by running the
"veritysetup verify" command and also by activating the dm-verity kernel
driver and reading the whole block device. If it passes without an error,
then the user is not affected.
The condition for the bug is:
Split the total number of data blocks (data_block_bits) into bit strings,
each string has hash_per_block_bits bits. hash_per_block_bits is
rounddown(log2(metadata_block_size/hash_digest_size)). Equivalently, you
can say that you convert data_blocks_bits to 2^hash_per_block_bits base.
If there some zero bit string below the most significant bit string and at
least one bit below this zero bit string is set, then the bug happens.
The same bug exists in the userspace veritysetup tool, so you must use
fixed veritysetup too if you want to use devices that are affected by
this boundary condition.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com> Cc: Milan Broz <gmazyland@gmail.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Set noio flag while calling __vmalloc() because it doesn't fully respect
gfp flags to avoid a possible deadlock (see commit 502624bdad3dba45dfaacaf36b7d83e39e74b2d2).
This should be backported to stable kernels 3.8 and newer. The kernel 3.8
doesn't have memalloc_noio_save(), so we should set and restore process
flag PF_MEMALLOC instead.
When multipath needs to retry an ioctl the reference to the
current live table needs to be dropped. Otherwise a deadlock
occurs when all paths are down:
- dm_blk_ioctl takes a reference to the current table
and spins in multipath_ioctl().
- A new table is being loaded, but upon resume the process
hangs in dm_table_destroy() waiting for references to
drop to zero.
With this patch the reference to the old table is dropped
prior to retry, thereby avoiding the deadlock.
Signed-off-by: Hannes Reinecke <hare@suse.de> Cc: Mike Snitzer <snitzer@redhat.com> Signed-off-by: Alasdair G Kergon <agk@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The BIOS of FUjitsu E753 reports an incorrect initial backlight value
for WIN8 compatible OS, causing backlight to be dark during startup.
This change causes the incorrect initial value from BIOS to be ignored.
References: https://bugzilla.kernel.org/show_bug.cgi?id=60161 Reported-and-tested-by: Jan Hinnerk Stosch <janhinnerk.stosch@gmail.com> Signed-off-by: Lan Tianyu <tianyu.lan@intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
device->driver_data needs to be cleared when releasing its data,
mem_device, in an error path of acpi_memory_device_add().
The function evaluates the _CRS of memory device objects, and fails
when it gets an unexpected resource or cannot allocate memory. A
kernel crash or data corruption may occur when the kernel accesses
the stale pointer.
Signed-off-by: Toshi Kani <toshi.kani@hp.com> Reviewed-by: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In acpi_bus_device_attach(), if there is an ACPI device object
for the given handle and that device object has a scan handler
attached to it already, there's nothing more to do for that handle.
Moreover, if acpi_scan_attach_handler() is called then, it may
execute the .attach() callback of the ACPI scan handler already
attached to the device object and that may lead to interesting
breakage.
For this reason, make acpi_bus_device_attach() return success
immediately when the handle's device object has a scan handler
attached to it.
Reported-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
An ACPI_NOTIFY_BUS_CHECK notification means that we should scan the
entire namespace starting from the given handle even if the device
represented by that handle is present (other devices below it may
just have appeared).
For this reason, modify acpi_scan_bus_device_check() to always run
acpi_bus_scan() if the notification being handled is of type
ACPI_NOTIFY_BUS_CHECK.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> Acked-by: Toshi Kani <toshi.kani@hp.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit f8bd822cb ("regmap: cache: Factor out block sync") made
regcache_rbtree_sync() call regmap_async_complete(), which in turn does
not check for map->bus before dereferencing it.
This causes a NULL pointer dereference on bus-less maps.
Signed-off-by: Daniel Mack <zonque@gmail.com> Signed-off-by: Mark Brown <broonie@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When we are posting pressure status, we may get interrupted and handle
the un-balloon operation. In this case just don't post the status as we
know the pressure status is stale.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
As we hot-add 128 MB chunks of memory, we wait to ensure that the memory
is onlined before attempting to hot-add the next chunk. If the udev rule for
memory hot-add is not executed within the allowed time, we would rollback the
state and abort further hot-add. Since the hot-add has succeeded and the only
failure is that the memory is not onlined within the allowed time, we should not
be rolling back the state. Fix this bug.
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com> Acked-by: Jason Wang <jasowang@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
can result in the following state:
------------------------------------------------------------
struct nfs4_file {
...
fi_fds = {0xffff880c1fa65c80, 0xffffffffffffffe6, 0x0},
fi_access = {{
counter = 0x1
}, {
counter = 0x0
}},
...
------------------------------------------------------------
1) First time around, in nfs4_get_vfs_file() fp->fi_fds[O_WRONLY] is
NULL, hence nfsd_open() is called where we get status set to an error
and fp->fi_fds[O_WRONLY] to -ETXTBSY. Thus we do not reach
nfs4_file_get_access() and fi_access[O_WRONLY] is not incremented.
2) Second time around, in nfs4_get_vfs_file() fp->fi_fds[O_WRONLY] is
NOT NULL (-ETXTBSY), so nfsd_open() is NOT called, but
nfs4_file_get_access() IS called and fi_access[O_WRONLY] is incremented.
Thus we leave a landmine in the form of the nfs4_file data structure in
an incorrect state.
Module CRCs are implemented as absolute symbols that get resolved by
a linker script. We build an intermediate .o that contains an
unresolved symbol for each CRC. genksysms parses this .o, calculates
the CRCs and writes a linker script that "resolves" the symbols to
the calculated CRC.
Unfortunately the ppc64 relocatable kernel sees these CRCs as symbols
that need relocating and relocates them at boot. Commit d4703aef
(module: handle ppc64 relocating kcrctabs when CONFIG_RELOCATABLE=y)
added a hook to reverse the bogus relocations. Part of this patch
created a symbol at 0x0:
This reloc_start symbol is causing lots of confusion to perf. It
thinks reloc_start is a massive function that stretches from 0x0 to
0xc000000000000000 and we get various cryptic errors out of perf,
including:
problem incrementing symbol count, skipping event
This patch removes the reloc_start linker script label and instead
defines it as PHYSICAL_START. We also need to wrap it with
CONFIG_PPC64 because the ppc32 kernel can set a non zero
PHYSICAL_START at compile time and we wouldn't want to subtract
it from the CRCs in that case.
Signed-off-by: Anton Blanchard <anton@samba.org> Acked-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When kernel is compiled with CONFIG_SLUB_DEBUG=y and
CRYPTO_MANAGER_DISABLE_TESTS=n, during kernel bootup, the kernel
reports error given below. The root cause is that in function
hash_digest_key(), for allocating descriptor, insufficient memory was
being allocated. The required number of descriptor words apart from
input and output pointers are 8 (instead of 6).
My static checker marks everything from ntohl() as untrusted and it
complains we could have an underflow problem doing:
return (u32 *)&ary->wc_array[nchunks];
Also on 32 bit systems the upper bound check could overflow.
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
sata_inic162x never reached a state where it's reliable enough for
production use and data corruption is a relatively common occurrence.
Make the driver generate warning about the issues and mark the Kconfig
option as experimental.
If the situation doesn't improve, we'd be better off making it depend
on CONFIG_BROKEN. Let's wait for several cycles and see if the kernel
message draws any attention.
Signed-off-by: Tejun Heo <tj@kernel.org> Reported-by: Martin Braure de Calignon <braurede@free.fr> Reported-by: Ben Hutchings <ben@decadent.org.uk> Reported-by: risc4all@yahoo.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a patch b55f84e2d527182e7c611d466cd0bb6ddce201de "ata_piix: Fix DVD
not dectected at some Haswell platforms" to fix an issue of DVD not
recognized on Haswell Desktop platform with Lynx Point.
Recently, it is also found the same issue at some platformas with Wellsburg PCH.
So deliver a similar patch to fix it by disables 32bit PIO in IDE mode.
We always need to be careful when calling generic_make_request, as it
can start a chain of events which might free something that we are
using.
Here is one place I wasn't careful enough. If the wbio2 is not in
use, then it might get freed at the first generic_make_request call.
So perform all necessary tests first.
This bug was introduced in 3.3-rc3 (24afd80d99) and can cause an
oops, so fix is suitable for any -stable since then.
If a device in a RAID4/5/6 is being replaced while another is being
recovered, then the writes to the replacement device currently don't
happen, resulting in corruption when the replacement completes and the
new drive takes over.
This is because the replacement writes are only triggered when
's.replacing' is set and not when the similar 's.sync' is set (which
is the case during resync and recovery - it means all devices need to
be read).
So schedule those writes when s.replacing is set as well.
In this case we cannot use "STRIPE_INSYNC" to record that the
replacement has happened as that is needed for recording that any
parity calculation is complete. So introduce STRIPE_REPLACED to
record if the replacement has happened.
For safety we should also check that STRIPE_COMPUTE_RUN is not set.
This has a similar effect to the "s.locked == 0" test. The latter
ensure that now IO has been flagged but not started. The former
checks if any parity calculation has been flagged by not started.
We must wait for both of these to complete before triggering the
'replace'.
Add a similar test to the subsequent check for "are we finished yet".
This possibly isn't needed (is subsumed in the STRIPE_INSYNC test),
but it makes it more obvious that the REPLACE will happen before we
think we are finished.
Finally if a NeedReplace device is not UPTODATE then that is an
error. We really must trigger a warning.
This bug was introduced in commit 9a3e1101b827a59ac9036a672f5fa8d5279d0fe2
(md/raid5: detect and handle replacements during recovery.)
which introduced replacement for raid5.
That was in 3.3-rc3, so any stable kernel since then would benefit
from this fix.
Recent change to use bio_copy_data() in raid1 when repairing
an array is faulty.
The underlying may have changed the bio in various ways using
bio_advance and these need to be undone not just for the 'sbio' which
is being copied to, but also the 'pbio' (primary) which is being
copied from.
So perform the reset on all bios that were read from and do it early.
This also ensure that the sbio->bi_io_vec[j].bv_len passed to
memcmp is correct.
This fixes a crash during a 'check' of a RAID1 array. The crash was
introduced in 3.10 so this is suitable for 3.10-stable.
Reported-by: Joe Lawrence <joe.lawrence@stratus.com> Signed-off-by: NeilBrown <neilb@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
allowed a bit more than just that. It also allows devices to be added
to a read-write array and to end up skipping recovery.
This patch removes the offending piece of code pending a rewrite for a
subsequent release.
More specifically:
If the array has a bitmap, then the device will still need a bitmap
based resync ('saved_raid_disk' is set under different conditions
is a bitmap is present).
If the array doesn't have a bitmap, then this is correct as long as
nothing has been written to the array since the metadata was checked
by ->validate_super. However there is no locking to ensure that there
was no write.
Bug was introduced in 3.10 and causes data corruption so
patch is suitable for 3.10-stable.
Reported-by: Joe Lawrence <joe.lawrence@stratus.com> Signed-off-by: NeilBrown <neilb@suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since the IDT is referenced from a fixmap, make sure it is page aligned.
This avoids the risk of the IDT ever being moved in the bss and having
the mapping be offset, resulting in calling incorrect handlers. In the
current upstream kernel this is not a manifested bug, but heavily patched
kernels (such as those using the PaX patch series) did encounter this bug.
Signed-off-by: Kees Cook <keescook@chromium.org> Reported-by: PaX Team <pageexec@gmail.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Yinghai Lu <yinghai@kernel.org> Cc: Seiji Aguchi <seiji.aguchi@hds.com> Cc: Fenghua Yu <fenghua.yu@intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are CPUs which have errata causing RDMSR of a nonexistent MSR to
not fault. We would then try to WRMSR to restore the value of that
MSR, causing a crash. Specifically, some Pentium M variants would
have this problem trying to save and restore the non-existent EFER,
causing a crash on resume.
Work around this by making sure we can write back the result at
suspend time.
Huge thanks to Christian Sünkenberg for finding the offending erratum
that finally deciphered the mystery.
Reported-and-tested-by: Johan Heinrich <onny@project-insanity.org> Debugged-by: Christian Sünkenberg <christian.suenkenberg@student.kit.edu> Acked-by: Rafael J. Wysocki <rjw@sisk.pl> Link: http://lkml.kernel.org/r/51DDC972.3010005@student.kit.edu Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We need to make sure that the device is not RO or that
the request is not past the number of sectors we want to
issue the DISCARD operation for.
This fixes CVE-2013-2140.
Acked-by: Jan Beulich <JBeulich@suse.com> Acked-by: Ian Campbell <Ian.Campbell@citrix.com>
[v1: Made it pr_warn instead of pr_debug] Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Due to commit 3683243b ("xen-netfront: use __pskb_pull_tail to ensure
linear area is big enough on RX") xennet_fill_frags() may end up
filling MAX_SKB_FRAGS + 1 fragments in a receive skb, and only reduce
the fragment count subsequently via __pskb_pull_tail(). That's a
result of xennet_get_responses() allowing a maximum of one more slot to
be consumed (and intermediately transformed into a fragment) if the
head slot has a size less than or equal to RX_COPY_THRESHOLD.
Hence we need to adjust xennet_fill_frags() to pull earlier if we
reached the maximum fragment count - due to the described behavior of
xennet_get_responses() this guarantees that at least the first fragment
will get completely consumed, and hence the fragment count reduced.
In order to not needlessly call __pskb_pull_tail() twice, make the
original call conditional upon the pull target not having been reached
yet, and defer the newly added one as much as possible (an alternative
would have been to always call the function right before the call to
xennet_fill_frags(), but that would imply more frequent cases of
needing to call it twice).
Signed-off-by: Jan Beulich <jbeulich@suse.com> Acked-by: Wei Liu <wei.liu2@citrix.com> Cc: Ian Campbell <ian.campbell@citrix.com> Acked-by: Ian Campbell <ian.campbell@citrix.com> Signed-off-by: David S. Miller <davem@davemloft.net> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If we stop dropping a root for whatever reason we need to add it back to the
dead root list so that we will re-start the dropping next transaction commit.
The other case this happens is if we recover a drop because we will add a root
without adding it to the fs radix tree, so we can leak it's root and commit root
extent buffer, adding this to the dead root list makes this cleanup happen.
Thanks,
Reported-by: Alex Lyakas <alex.btrfs@zadarastorage.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We aren't setting path->locks[level] when we resume a snapshot deletion which
means we won't unlock the buffer when we free the path. This causes deadlocks
if we happen to re-allocate the block before we've evicted the extent buffer
from cache. Thanks,
Reported-by: Alex Lyakas <alex.btrfs@zadarastorage.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The filesystem was corrupted after we did a device replace.
Steps to reproduce:
# mkfs.btrfs -f -m single -d raid10 <device0>..<device3>
# mount <device0> <mnt>
# btrfs replace start -rfB 1 <device4> <mnt>
# umount <mnt>
# btrfsck <device4>
The reason for the issue is that we changed the write offset by mistake,
introduced by commit 625f1c8dc.
We read the data from the source device at first, and then write the
data into the corresponding place of the new device. In order to
implement the "-r" option, the source location is remapped using
btrfs_map_block(). The read takes place on the mapped location, and
the write needs to take place on the unmapped location. Currently
the write is using the mapped location, and this commit changes it
back by undoing the change to the write address that the aforementioned
commit added by mistake.
Reported-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the situation that a writer fails to copy data from userspace it will reset
the write offset to the value it had before it went to sleep. This discarding
any messages written while aquiring the mutex.
Therefore the reset offset needs to be retrieved after acquiring the mutex.
Comedi devices can do blocking read() or write() (or poll()) if an
asynchronous command has been set up, blocking for data (for read()) or
buffer space (for write()). Various events associated with the
asynchronous command will wake up the blocked reader or writer (or
poller). It is also possible to force the asynchronous command to
terminate by issuing a `COMEDI_CANCEL` ioctl. That shuts down the
asynchronous command, but does not currently wake up the blocked reader
or writer (or poller). If the blocked task could be woken up, it would
see that the command is no longer active and return. The caller of the
`COMEDI_CANCEL` ioctl could attempt to wake up the blocked task by
sending a signal, but that's a nasty workaround.
Change `do_cancel_ioctl()` to wake up the wait queue after it returns
from `do_cancel()`. `do_cancel()` can propagate an error return value
from the low-level comedi driver's cancel routine, but it always shuts
the command down regardless, so `do_cancel_ioctl()` can wake up he wait
queue regardless of the return value from `do_cancel()`.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
`do_cmd_ioctl()` is called with the comedi device's mutex locked to
process the `COMEDI_CMD` ioctl to set up comedi's asynchronous command
handling on a comedi subdevice. `comedi_read()` and `comedi_write()`
are the `read` and `write` handlers for the comedi device, but do not
lock the mutex (for performance reasons, as some things can hold the
mutex for quite a long time).
There is a race condition if `comedi_read()` or `comedi_write()` is
running at the same time and for the same file object and comedi
subdevice as `do_cmd_ioctl()`. `do_cmd_ioctl()` sets the subdevice's
`busy` pointer to the file object way before it sets the `SRF_RUNNING` flag
in the subdevice's `runflags` member. `comedi_read() and
`comedi_write()` check the subdevice's `busy` pointer is pointing to the
current file object, then if the `SRF_RUNNING` flag is not set, will call
`do_become_nonbusy()` to shut down the asyncronous command. Bad things
can happen if the asynchronous command is being shutdown and set up at
the same time.
To prevent the race, don't set the `busy` pointer until
after the `SRF_RUNNING` flag has been set. Also, make sure the mutex is
held in `comedi_read()` and `comedi_write()` while calling
`do_become_nonbusy()` in order to avoid moving the race condition to a
point within that function.
Change some error handling `goto cleanup` statements in `do_cmd_ioctl()`
to simple `return -ERRFOO` statements as a result of changing when the
`busy` pointer is set.
Signed-off-by: Ian Abbott <abbotti@mev.co.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The hub driver was recently changed to use "global" suspend for system
suspend transitions on non-SuperSpeed buses. This means that we don't
suspend devices individually by setting the suspend feature on the
upstream hub port; instead devices all go into suspend automatically
when the root hub stops transmitting packets. The idea was to save
time and to avoid certain kinds of wakeup races.
Now it turns out that many hubs are buggy; they don't relay wakeup
requests from a downstream port to their upstream port if the
downstream port's suspend feature is not set (depending on the speed
of the downstream port, whether or not the hub is enabled for remote
wakeup, and possibly other factors).
We can't have hubs dropping wakeup requests. Therefore this patch
goes partway back to the old policy: It sets the suspend feature for a
port if the device attached to that port or any of its descendants is
enabled for wakeup. People will still be able to benefit from the
time savings if they don't care about wakeup and leave it disabled on
all their devices.
In order to accomplish this, the patch adds a new field to the usb_hub
structure: wakeup_enabled_descendants is a count of how many devices
below a suspended hub are enabled for remote wakeup. A corresponding
new subroutine determines the number of wakeup-enabled devices at or
below an arbitrary suspended USB device.
Control transfers have both IN and OUT (or SETUP) packets, so when
clearing TT buffers for a control transfer it's necessary to send
two HUB_CLEAR_TT_BUFFER requests to the hub.
Signed-off-by: William Gulland <wgulland@google.com> Acked-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The driver failed to take the dynamic ids into account when determining
the device type and therefore all devices were detected as 2-port
devices when using the dynamic-id interface.
Match on the usb-serial-driver field instead of doing redundant id-table
searches.
Reported-by: Anders Hammarquist <iko@iko.pp.se> Signed-off-by: Johan Hovold <jhovold@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Per dwc3 2.50a spec, the is_devspec bit is used to distinguish the
Device Endpoint-Specific Event or Device-Specific Event (DEVT). If the
bit is 1, the event is represented Device-Specific Event, then use
[7:1] bits as Device Specific Event to marked the type. It has 7 bits,
and we can see the reserved8_31 variable name which means from 8 to 31
bits marked reserved, actually there are 24 bits not 25 bits between
that. And 1 + 7 + 24 = 32, the event size is 4 byes.
So in dwc3_event_type, the bit mask should be:
is_devspec [0] 1 bit
type [7:1] 7 bits
reserved8_31 [31:8] 24 bits
This patch should be backported to kernels as old as 3.2, that contain
the commit 72246da40f3719af3bfd104a2365b32537c27d83 "usb: Introduce
DesignWare USB3 DRD Driver".
When the host controller fails to respond to an Enable Slot command, and
the host fails to respond to the register write to abort the command
ring, the xHCI driver will assume the host is dead, and call
usb_hc_died().
The USB device's slot_id is still set to zero, and the pointer stored at
xhci->devs[0] will always be NULL. The call to xhci_check_args in
xhci_free_dev should have caught the NULL virt_dev pointer.
However, xhci_free_dev is designed to free the xhci_virt_device
structures, even if the host is dead, so that we don't leak kernel
memory. xhci_free_dev checks the return value from the generic
xhci_check_args function. If the return value is -ENODEV, it carries on
trying to free the virtual device.
The issue is that xhci_check_args looks at the host controller state
before it looks at the xhci_virt_device pointer. It will return -ENIVAL
because the host is dead, and xhci_free_dev will ignore the return
value, and happily dereference the NULL xhci_virt_device pointer.
The fix is to make sure that xhci_check_args checks the xhci_virt_device
pointer before it checks the host state.
See https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1203453 for
further details. This patch doesn't solve the underlying issue, but
will ensure we don't see any more NULL pointer dereferences because of
the issue.
This patch should be backported to kernels as old as 3.1, that
contain the commit 7bd89b4017f46a9b92853940fd9771319acb578a "xhci: Don't
submit commands or URBs to halted hosts."
Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com> Reported-by: Vincent Thiele <vincentthiele@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
in some cases where device is attched to xhci port and do not responding,
for example ath9k_htc with stalled firmware, kernel will
crash on ring_doorbell_for_active_rings.
This patch check if pointer exist before it is used.
This patch should be backported to kernels as old as 2.6.35, that
contain the commit e9df17eb1408cfafa3d1844bfc7f22c7237b31b8 "USB: xhci:
Correct assumptions about number of rings per endpoint"
Xhci controllers with hci_version > 0.96 gives spurious success
events on short packet completion. During webcam capture the
"ERROR Transfer event TRB DMA ptr not part of current TD" was observed.
The same application works fine with synopsis controllers hci_version 0.96.
The same issue is seen with Intel Pantherpoint xhci controller. So enabling
this quirk in xhci_gen_setup if controller verion is greater than 0.96.
For xhci-pci move the quirk to much generic place xhci_gen_setup.
Note from Sarah:
The xHCI 1.0 spec changed how hardware handles short packets. The HW
will notify SW of the TRB where the short packet occurred, and it will
also give a successful status for the last TRB in a TD (the one with the
IOC flag set). On the second successful status, that warning will be
triggered in the driver.
Software is now supposed to not assume the TD is not completed until it
gets that last successful status. That means we have a slight race
condition, although it should have little practical impact. This patch
papers over that issue.
It's on my long-term to-do list to fix this race condition, but it is a
much more involved patch that will probably be too big for stable. This
patch is needed for stable to avoid serious log spam.
This patch should be backported to kernels as old as 3.0, that
contain the commit ad808333d8201d53075a11bc8dd83b81f3d68f0b "Intel xhci:
Ignore spurious successful event."
The patch will have to be modified for kernels older than 3.2, since
that kernel added the xhci_gen_setup function for xhci platform devices.
The correct conflict resolution for kernels older than 3.2 is to set
XHCI_SPURIOUS_SUCCESS in xhci_pci_quirks for all xHCI 1.0 hosts.
Signed-off-by: George Cherian <george.cherian@ti.com> Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Commit a82274151af "tracing: Protect ftrace_trace_arrays list in trace_events.c"
added taking the trace_types_lock mutex in trace_events.c as there were
several locations that needed it for protection. Unfortunately, it also
encapsulated a call to tracing_reset_all_online_cpus() which also takes
the trace_types_lock, causing a deadlock.
This happens when a module has tracepoints and has been traced. When the
module is removed, the trace events module notifier will grab the
trace_types_lock, do a bunch of clean ups, and also clears the buffer
by calling tracing_reset_all_online_cpus. This doesn't happen often
which explains why it wasn't caught right away.
Commit a82274151af was marked for stable, which means this must be
sent to stable too.
Link: http://lkml.kernel.org/r/51EEC646.7070306@broadcom.com Reported-by: Arend van Spril <arend@broadcom.com> Tested-by: Arend van Spriel <arend@broadcom.com> Cc: Alexander Z Lam <azl@google.com> Cc: Vaibhav Nagarnaik <vnagarnaik@google.com> Cc: David Sharp <dhsharp@google.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
tracing_buffers_open() does trace_array_get() and then it wrongly
inrcements tr->ref again under trace_types_lock. This means that
every caller leaks trace_array:
# cd /sys/kernel/debug/tracing/
# mkdir instances/X
# true < instances/X/per_cpu/cpu0/trace_pipe_raw
# rmdir instances/X
rmdir: failed to remove `instances/X': Device or resource busy