OSDN Git Service

sagit-ice-cold/kernel_xiaomi_msm8998.git
6 years agofutex: Replace barrier() in unqueue_me() with READ_ONCE()
Jianyu Zhan [Mon, 7 Mar 2016 01:32:24 +0000 (09:32 +0800)]
futex: Replace barrier() in unqueue_me() with READ_ONCE()

commit 29b75eb2d56a714190a93d7be4525e617591077a upstream.

Commit e91467ecd1ef ("bug in futex unqueue_me") introduced a barrier() in
unqueue_me() to prevent the compiler from rereading the lock pointer which
might change after a check for NULL.

Replace the barrier() with a READ_ONCE() for the following reasons:

1) READ_ONCE() is a weaker form of barrier() that affects only the specific
   load operation, while barrier() is a general compiler level memory barrier.
   READ_ONCE() was not available at the time when the barrier was added.

2) Aside of that READ_ONCE() is descriptive and self explainatory while a
   barrier without comment is not clear to the casual reader.

No functional change.

[ tglx: Massaged changelog ]

Signed-off-by: Jianyu Zhan <nasa4836@gmail.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Darren Hart <dvhart@linux.intel.com>
Cc: dave@stgolabs.net
Cc: peterz@infradead.org
Cc: linux@rasmusvillemoes.dk
Cc: akpm@linux-foundation.org
Cc: fengguang.wu@intel.com
Cc: bigeasy@linutronix.de
Link: http://lkml.kernel.org/r/1457314344-5685-1-git-send-email-nasa4836@gmail.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agolocks: don't check for race with close when setting OFD lock
Jeff Layton [Fri, 8 Jan 2016 12:30:43 +0000 (07:30 -0500)]
locks: don't check for race with close when setting OFD lock

commit 0752ba807b04ccd69cb4bc8bbf829a80ee208a3c upstream.

We don't clean out OFD locks on close(), so there's no need to check
for a race with them here. They'll get cleaned out at the same time
that flock locks are.

Signed-off-by: Jeff Layton <jeff.layton@primarydata.com>
Acked-by: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
6 years agozswap: don't param_set_charp while holding spinlock
Dan Streetman [Mon, 27 Feb 2017 22:26:53 +0000 (14:26 -0800)]
zswap: don't param_set_charp while holding spinlock

commit fd5bb66cd934987e49557455b6497fc006521940 upstream.

Change the zpool/compressor param callback function to release the
zswap_pools_lock spinlock before calling param_set_charp, since that
function may sleep when it calls kmalloc with GFP_KERNEL.

While this problem has existed for a while, I wasn't able to trigger it
using a tight loop changing either/both the zpool and compressor params; I
think it's very unlikely to be an issue on the stable kernels, especially
since most zswap users will change the compressor and/or zpool from sysfs
only one time each boot - or zero times, if they add the params to the
kernel boot.

Fixes: c99b42c3529e ("zswap: use charp for zswap param strings")
Link: http://lkml.kernel.org/r/20170126155821.4545-1-ddstreet@ieee.org
Signed-off-by: Dan Streetman <dan.streetman@canonical.com>
Reported-by: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Minchan Kim <minchan@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomm/zswap: use workqueue to destroy pool
Dan Streetman [Fri, 20 May 2016 23:59:54 +0000 (16:59 -0700)]
mm/zswap: use workqueue to destroy pool

commit 200867af4dedfe7cb707f96773684de1d1fd21e6 upstream.

Add a work_struct to struct zswap_pool, and change __zswap_pool_empty to
use the workqueue instead of using call_rcu().

When zswap destroys a pool no longer in use, it uses call_rcu() to
perform the destruction/freeing.  Since that executes in softirq
context, it must not sleep.  However, actually destroying the pool
involves freeing the per-cpu compressors (which requires locking the
cpu_add_remove_lock mutex) and freeing the zpool, for which the
implementation may sleep (e.g.  zsmalloc calls kmem_cache_destroy, which
locks the slab_mutex).  So if either mutex is currently taken, or any
other part of the compressor or zpool implementation sleeps, it will
result in a BUG().

It's not easy to reproduce this when changing zswap's params normally.
In testing with a loaded system, this does not fail:

  $ cd /sys/module/zswap/parameters
  $ echo lz4 > compressor ; echo zsmalloc > zpool

nor does this:

  $ while true ; do
  > echo lzo > compressor ; echo zbud > zpool
  > sleep 1
  > echo lz4 > compressor ; echo zsmalloc > zpool
  > sleep 1
  > done

although it's still possible either of those might fail, depending on
whether anything else besides zswap has locked the mutexes.

However, changing a parameter with no delay immediately causes the
schedule while atomic BUG:

  $ while true ; do
  > echo lzo > compressor ; echo lz4 > compressor
  > done

This is essentially the same as Yu Zhao's proposed patch to zsmalloc,
but moved to zswap, to cover compressor and zpool freeing.

Fixes: f1c54846ee45 ("zswap: dynamic pool creation")
Signed-off-by: Dan Streetman <ddstreet@ieee.org>
Reported-by: Yu Zhao <yuzhao@google.com>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Dan Streetman <dan.streetman@canonical.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomm/page-writeback: fix dirty_ratelimit calculation
Andrey Ryabinin [Tue, 15 Mar 2016 21:55:27 +0000 (14:55 -0700)]
mm/page-writeback: fix dirty_ratelimit calculation

commit d59b1087a98e402ed9a7cc577f4da435f9a555f5 upstream.

Calculation of dirty_ratelimit sometimes is not correct.  E.g.  initial
values of dirty_ratelimit == INIT_BW and step == 0, lead to the
following result:

   UBSAN: Undefined behaviour in ../mm/page-writeback.c:1286:7
   shift exponent 25600 is too large for 64-bit type 'long unsigned int'

The fix is straightforward - make step 0 if the shift exponent is too
big.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Tejun Heo <tj@kernel.org>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomm/compaction: pass only pageblock aligned range to pageblock_pfn_to_page
Joonsoo Kim [Tue, 15 Mar 2016 21:57:48 +0000 (14:57 -0700)]
mm/compaction: pass only pageblock aligned range to pageblock_pfn_to_page

commit e1409c325fdc1fef7b3d8025c51892355f065d15 upstream.

pageblock_pfn_to_page() is used to check there is valid pfn and all
pages in the pageblock is in a single zone.  If there is a hole in the
pageblock, passing arbitrary position to pageblock_pfn_to_page() could
cause to skip whole pageblock scanning, instead of just skipping the
hole page.  For deterministic behaviour, it's better to always pass
pageblock aligned range to pageblock_pfn_to_page().  It will also help
further optimization on pageblock_pfn_to_page() in the following patch.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rik van Riel <riel@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomm/compaction: fix invalid free_pfn and compact_cached_free_pfn
Joonsoo Kim [Tue, 15 Mar 2016 21:57:45 +0000 (14:57 -0700)]
mm/compaction: fix invalid free_pfn and compact_cached_free_pfn

commit 623446e4dc45b37740268165107cc63abb3022f0 upstream.

free_pfn and compact_cached_free_pfn are the pointer that remember
restart position of freepage scanner.  When they are reset or invalid,
we set them to zone_end_pfn because freepage scanner works in reverse
direction.  But, because zone range is defined as [zone_start_pfn,
zone_end_pfn), zone_end_pfn is invalid to access.  Therefore, we should
not store it to free_pfn and compact_cached_free_pfn.  Instead, we need
to store zone_end_pfn - 1 to them.  There is one more thing we should
consider.  Freepage scanner scan reversely by pageblock unit.  If
free_pfn and compact_cached_free_pfn are set to middle of pageblock, it
regards that sitiation as that it already scans front part of pageblock
so we lose opportunity to scan there.  To fix-up, this patch do
round_down() to guarantee that reset position will be pageblock aligned.

Note that thanks to the current pageblock_pfn_to_page() implementation,
actual access to zone_end_pfn doesn't happen until now.  But, following
patch will change pageblock_pfn_to_page() so this patch is needed from
now on.

Signed-off-by: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Aaron Lu <aaron.lu@intel.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Rik van Riel <riel@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/acpi: Reduce code duplication in mp_override_legacy_irq()
Vikas C Sajjan [Thu, 16 Nov 2017 16:13:45 +0000 (21:43 +0530)]
x86/acpi: Reduce code duplication in mp_override_legacy_irq()

commit 4ee2ec1b122599f7b10c849fa7915cebb37b7edb upstream.

The new function mp_register_ioapic_irq() is a subset of the code in
mp_override_legacy_irq().

Replace the code duplication by invoking mp_register_ioapic_irq() from
mp_override_legacy_irq().

Signed-off-by: Vikas C Sajjan <vikas.cha.sajjan@hpe.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: linux-pm@vger.kernel.org
Cc: kkamagui@gmail.com
Cc: linux-acpi@vger.kernel.org
Link: https://lkml.kernel.org/r/1510848825-21965-3-git-send-email-vikas.cha.sajjan@hpe.com
Cc: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: aloop: Fix racy hw constraints adjustment
Takashi Iwai [Thu, 4 Jan 2018 16:38:54 +0000 (17:38 +0100)]
ALSA: aloop: Fix racy hw constraints adjustment

commit 898dfe4687f460ba337a01c11549f87269a13fa2 upstream.

The aloop driver tries to update the hw constraints of the connected
target on the cable of the opened PCM substream.  This is done by
adding the extra hw constraints rules referring to the substream
runtime->hw fields, while the other substream may update the runtime
hw of another side on the fly.

This is, however, racy and may result in the inconsistent values when
both PCM streams perform the prepare concurrently.  One of the reason
is that it overwrites the other's runtime->hw field; which is not only
racy but also broken when it's called before the open of another side
finishes.  And, since the reference to runtime->hw isn't protected,
the concurrent write may give the partial value update and become
inconsistent.

This patch is an attempt to fix and clean up:
- The prepare doesn't change the runtime->hw of other side any longer,
  but only update the cable->hw that is referred commonly.
- The extra rules refer to the loopback_pcm object instead of the
  runtime->hw.  The actual hw is deduced from cable->hw.
- The extra rules take the cable_lock to protect against the race.

Fixes: b1c73fc8e697 ("ALSA: snd-aloop: Fix hw_params restrictions and checking")
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: aloop: Fix inconsistent format due to incomplete rule
Takashi Iwai [Fri, 5 Jan 2018 15:15:33 +0000 (16:15 +0100)]
ALSA: aloop: Fix inconsistent format due to incomplete rule

commit b088b53e20c7d09b5ab84c5688e609f478e5c417 upstream.

The extra hw constraint rule for the formats the aloop driver
introduced has a slight flaw, where it doesn't return a positive value
when the mask got changed.  It came from the fact that it's basically
a copy&paste from snd_hw_constraint_mask64().  The original code is
supposed to be a single-shot and it modifies the mask bits only once
and never after, while what we need for aloop is the dynamic hw rule
that limits the mask bits.

This difference results in the inconsistent state, as the hw_refine
doesn't apply the dependencies fully.  The worse and surprisingly
result is that it causes a crash in OSS emulation when multiple
full-duplex reads/writes are performed concurrently (I leave why it
triggers Oops to readers as a homework).

For fixing this, replace a few open-codes with the standard
snd_mask_*() macros.

Reported-by: syzbot+3902b5220e8ca27889ca@syzkaller.appspotmail.com
Fixes: b1c73fc8e697 ("ALSA: snd-aloop: Fix hw_params restrictions and checking")
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: aloop: Release cable upon open error path
Takashi Iwai [Fri, 5 Jan 2018 15:09:47 +0000 (16:09 +0100)]
ALSA: aloop: Release cable upon open error path

commit 9685347aa0a5c2869058ca6ab79fd8e93084a67f upstream.

The aloop runtime object and its assignment in the cable are left even
when opening a substream fails.  This doesn't mean any memory leak,
but it still keeps the invalid pointer that may be referred by the
another side of the cable spontaneously, which is a potential Oops
cause.

Clean up the cable assignment and the empty cable upon the error path
properly.

Fixes: 597603d615d2 ("ALSA: introduce the snd-aloop module for the PCM loopback")
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: pcm: Allow aborting mutex lock at OSS read/write loops
Takashi Iwai [Mon, 8 Jan 2018 13:03:53 +0000 (14:03 +0100)]
ALSA: pcm: Allow aborting mutex lock at OSS read/write loops

commit 900498a34a3ac9c611e9b425094c8106bdd7dc1c upstream.

PCM OSS read/write loops keep taking the mutex lock for the whole
read/write, and this might take very long when the exceptionally high
amount of data is given.  Also, since it invokes with mutex_lock(),
the concurrent read/write becomes unbreakable.

This patch tries to address these issues by replacing mutex_lock()
with mutex_lock_interruptible(), and also splits / re-takes the lock
at each read/write period chunk, so that it can switch the context
more finely if requested.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: pcm: Abort properly at pending signal in OSS read/write loops
Takashi Iwai [Mon, 8 Jan 2018 12:58:31 +0000 (13:58 +0100)]
ALSA: pcm: Abort properly at pending signal in OSS read/write loops

commit 29159a4ed7044c52e3e2cf1a9fb55cec4745c60b upstream.

The loops for read and write in PCM OSS emulation have no proper check
of pending signals, and they keep processing even after user tries to
break.  This results in a very long delay, often seen as RCU stall
when a huge unprocessed bytes remain queued.  The bug could be easily
triggered by syzkaller.

As a simple workaround, this patch adds the proper check of pending
signals and aborts the loop appropriately.

Reported-by: syzbot+993cb4cfcbbff3947c21@syzkaller.appspotmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: pcm: Add missing error checks in OSS emulation plugin builder
Takashi Iwai [Thu, 4 Jan 2018 15:39:27 +0000 (16:39 +0100)]
ALSA: pcm: Add missing error checks in OSS emulation plugin builder

commit 6708913750344a900f2e73bfe4a4d6dbbce4fe8d upstream.

In the OSS emulation plugin builder where the frame size is parsed in
the plugin chain, some places miss the possible errors returned from
the plugin src_ or dst_frames callback.

This patch papers over such places.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoALSA: pcm: Remove incorrect snd_BUG_ON() usages
Takashi Iwai [Mon, 1 Jan 2018 08:50:50 +0000 (09:50 +0100)]
ALSA: pcm: Remove incorrect snd_BUG_ON() usages

commit fe08f34d066f4404934a509b6806db1a4f700c86 upstream.

syzkaller triggered kernel warnings through PCM OSS emulation at
closing a stream:
  WARNING: CPU: 0 PID: 3502 at sound/core/pcm_lib.c:1635
  snd_pcm_hw_param_first+0x289/0x690 sound/core/pcm_lib.c:1635
  Call Trace:
  ....
   snd_pcm_hw_param_near.constprop.27+0x78d/0x9a0 sound/core/oss/pcm_oss.c:457
   snd_pcm_oss_change_params+0x17d3/0x3720 sound/core/oss/pcm_oss.c:969
   snd_pcm_oss_make_ready+0xaa/0x130 sound/core/oss/pcm_oss.c:1128
   snd_pcm_oss_sync+0x257/0x830 sound/core/oss/pcm_oss.c:1638
   snd_pcm_oss_release+0x20b/0x280 sound/core/oss/pcm_oss.c:2431
   __fput+0x327/0x7e0 fs/file_table.c:210
   ....

This happens while it tries to open and set up the aloop device
concurrently.  The warning above (invoked from snd_BUG_ON() macro) is
to detect the unexpected logical error where snd_pcm_hw_refine() call
shouldn't fail.  The theory is true for the case where the hw_params
config rules are static.  But for an aloop device, the hw_params rule
condition does vary dynamically depending on the connected target;
when another device is opened and changes the parameters, the device
connected in another side is also affected, and it caused the error
from snd_pcm_hw_refine().

That is, the simplest "solution" for this is to remove the incorrect
assumption of static rules, and treat such an error as a normal error
path.  As there are a couple of other places using snd_BUG_ON()
incorrectly, this patch removes these spurious snd_BUG_ON() calls.

Reported-by: syzbot+6f11c7e2a1b91d466432@syzkaller.appspotmail.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoiommu/arm-smmu-v3: Don't free page table ops twice
Jean-Philippe Brucker [Thu, 14 Dec 2017 11:03:01 +0000 (11:03 +0000)]
iommu/arm-smmu-v3: Don't free page table ops twice

commit 57d72e159b60456c8bb281736c02ddd3164037aa upstream.

Kasan reports a double free when finalise_stage_fn fails: the io_pgtable
ops are freed by arm_smmu_domain_finalise and then again by
arm_smmu_domain_free. Prevent this by leaving pgtbl_ops empty on failure.

Fixes: 48ec83bcbcf5 ("iommu/arm-smmu: Add initial driver support for ARM SMMUv3 devices")
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe.brucker@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/acpi: Handle SCI interrupts above legacy space gracefully
Vikas C Sajjan [Thu, 16 Nov 2017 16:13:44 +0000 (21:43 +0530)]
x86/acpi: Handle SCI interrupts above legacy space gracefully

commit 252714155f04c5d16989cb3aadb85fd1b5772f99 upstream.

Platforms which support only IOAPIC mode, pass the SCI information above
the legacy space (0-15) via the FADT mechanism and not via MADT.

In such cases mp_override_legacy_irq() which is invoked from
acpi_sci_ioapic_setup() to register SCI interrupts fails for interrupts
greater equal 16, since it is meant to handle only the legacy space and
emits error "Invalid bus_irq %u for legacy override".

Add a new function to handle SCI interrupts >= 16 and invoke it
conditionally in acpi_sci_ioapic_setup().

The code duplication due to this new function will be cleaned up in a
separate patch.

Co-developed-by: Sunil V L <sunil.vl@hpe.com>
Signed-off-by: Vikas C Sajjan <vikas.cha.sajjan@hpe.com>
Signed-off-by: Sunil V L <sunil.vl@hpe.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Abdul Lateef Attar <abdul-lateef.attar@hpe.com>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: linux-pm@vger.kernel.org
Cc: kkamagui@gmail.com
Cc: linux-acpi@vger.kernel.org
Link: https://lkml.kernel.org/r/1510848825-21965-2-git-send-email-vikas.cha.sajjan@hpe.com
Cc: Jean Delvare <jdelvare@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/vsdo: Fix build on PARAVIRT_CLOCK=y, KVM_GUEST=n
Andy Lutomirski [Wed, 30 Dec 2015 04:12:18 +0000 (20:12 -0800)]
x86/vsdo: Fix build on PARAVIRT_CLOCK=y, KVM_GUEST=n

commit 8705d603edd49f1cff165cd3b7998f4c7f098d27 upstream.

arch/x86/built-in.o: In function `arch_setup_additional_pages':
 (.text+0x587): undefined reference to `pvclock_pvti_cpu0_va'

KVM_GUEST selects PARAVIRT_CLOCK, so we can make pvclock_pvti_cpu0_va depend
on KVM_GUEST.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Tested-by: Borislav Petkov <bp@alien8.de>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Kees Cook <keescook@chromium.org>
Link: http://lkml.kernel.org/r/444d38a9bcba832685740ea1401b569861d09a72.1451446564.git.luto@kernel.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: James Dingwall <james@dingwall.me.uk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokvm: vmx: Scrub hardware GPRs at VM-exit
Jim Mattson [Wed, 3 Jan 2018 22:31:38 +0000 (14:31 -0800)]
kvm: vmx: Scrub hardware GPRs at VM-exit

commit 0cb5b30698fdc8f6b4646012e3acb4ddce430788 upstream.

Guest GPR values are live in the hardware GPRs at VM-exit.  Do not
leave any guest values in hardware GPRs after the guest GPR values are
saved to the vcpu_vmx structure.

This is a partial mitigation for CVE 2017-5715 and CVE 2017-5753.
Specifically, it defeats the Project Zero PoC for CVE 2017-5715.

Suggested-by: Eric Northup <digitaleric@google.com>
Signed-off-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Eric Northup <digitaleric@google.com>
Reviewed-by: Benjamin Serebrin <serebrin@google.com>
Reviewed-by: Andrew Honig <ahonig@google.com>
[Paolo: Add AMD bits, Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agonet/mac80211/debugfs.c: prevent build failure with CONFIG_UBSAN=y
Andrey Ryabinin [Wed, 20 Jan 2016 23:00:51 +0000 (15:00 -0800)]
net/mac80211/debugfs.c: prevent build failure with CONFIG_UBSAN=y

commit 68920c973254c5b71a684645c5f6f82d6732c5d6 upstream.

With upcoming CONFIG_UBSAN the following BUILD_BUG_ON in
net/mac80211/debugfs.c starts to trigger:

  BUILD_BUG_ON(hw_flag_names[NUM_IEEE80211_HW_FLAGS] != (void *)0x1);

It seems, that compiler instrumentation causes some code
deoptimizations.  Because of that GCC is not being able to resolve
condition in BUILD_BUG_ON() at compile time.

We could make size of hw_flag_names array unspecified and replace the
condition in BUILD_BUG_ON() with following:

  ARRAY_SIZE(hw_flag_names) != NUM_IEEE80211_HW_FLAGS

That will have the same effect as before (adding new flag without
updating array will trigger build failure) except it doesn't fail with
CONFIG_UBSAN.  As a bonus this patch slightly decreases size of
hw_flag_names array.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[Daniel: backport to 4.4.]
Signed-off-by: Daniel Wagner <daniel.wagner@siemens.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Disallow outsized PTRACE_SETREGSET NT_PRFPREG regset accesses
Maciej W. Rozycki [Mon, 11 Dec 2017 22:56:54 +0000 (22:56 +0000)]
MIPS: Disallow outsized PTRACE_SETREGSET NT_PRFPREG regset accesses

commit c8c5a3a24d395b14447a9a89d61586a913840a3b upstream.

Complement commit c23b3d1a5311 ("MIPS: ptrace: Change GP regset to use
correct core dump register layout") and also reject outsized
PTRACE_SETREGSET requests to the NT_PRFPREG regset, like with the
NT_PRSTATUS regset.

Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: c23b3d1a5311 ("MIPS: ptrace: Change GP regset to use correct core dump register layout")
Cc: James Hogan <james.hogan@mips.com>
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Alex Smith <alex@alex-smith.me.uk>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17930/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Also verify sizeof `elf_fpreg_t' with PTRACE_SETREGSET
Maciej W. Rozycki [Mon, 11 Dec 2017 22:55:40 +0000 (22:55 +0000)]
MIPS: Also verify sizeof `elf_fpreg_t' with PTRACE_SETREGSET

commit 006501e039eec411842bb3150c41358867d320c2 upstream.

Complement commit d614fd58a283 ("mips/ptrace: Preserve previous
registers for short regset write") and like with the PTRACE_GETREGSET
ptrace(2) request also apply a BUILD_BUG_ON check for the size of the
`elf_fpreg_t' type in the PTRACE_SETREGSET request handler.

Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: d614fd58a283 ("mips/ptrace: Preserve previous registers for short regset write")
Cc: James Hogan <james.hogan@mips.com>
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Alex Smith <alex@alex-smith.me.uk>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17929/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Fix an FCSR access API regression with NT_PRFPREG and MSA
Maciej W. Rozycki [Mon, 11 Dec 2017 22:54:33 +0000 (22:54 +0000)]
MIPS: Fix an FCSR access API regression with NT_PRFPREG and MSA

commit be07a6a1188372b6d19a3307ec33211fc9c9439d upstream.

Fix a commit 72b22bbad1e7 ("MIPS: Don't assume 64-bit FP registers for
FP regset") public API regression, then activated by commit 1db1af84d6df
("MIPS: Basic MSA context switching support"), that caused the FCSR
register not to be read or written for CONFIG_CPU_HAS_MSA kernel
configurations (regardless of actual presence or absence of the MSA
feature in a given processor) with ptrace(2) PTRACE_GETREGSET and
PTRACE_SETREGSET requests nor recorded in core dumps.

This is because with !CONFIG_CPU_HAS_MSA configurations the whole of
`elf_fpregset_t' array is bulk-copied as it is, which includes the FCSR
in one half of the last, 33rd slot, whereas with CONFIG_CPU_HAS_MSA
configurations array elements are copied individually, and then only the
leading 32 FGR slots while the remaining slot is ignored.

Correct the code then such that only FGR slots are copied in the
respective !MSA and MSA helpers an then the FCSR slot is handled
separately in common code.  Use `ptrace_setfcr31' to update the FCSR
too, so that the read-only mask is respected.

Retrieving a correct value of FCSR is important in debugging not only
for the human to be able to get the right interpretation of the
situation, but for correct operation of GDB as well.  This is because
the condition code bits in FSCR are used by GDB to determine the
location to place a breakpoint at when single-stepping through an FPU
branch instruction.  If such a breakpoint is placed incorrectly (i.e.
with the condition reversed), then it will be missed, likely causing the
debuggee to run away from the control of GDB and consequently breaking
the process of investigation.

Fortunately GDB continues using the older PTRACE_GETFPREGS ptrace(2)
request which is unaffected, so the regression only really hits with
post-mortem debug sessions using a core dump file, in which case
execution, and consequently single-stepping through branches is not
possible.  Of course core files created by buggy kernels out there will
have the value of FCSR recorded clobbered, but such core files cannot be
corrected and the person using them simply will have to be aware that
the value of FCSR retrieved is not reliable.

Which also means we can likely get away without defining a replacement
API which would ensure a correct value of FSCR to be retrieved, or none
at all.

This is based on previous work by Alex Smith, extensively rewritten.

Signed-off-by: Alex Smith <alex@alex-smith.me.uk>
Signed-off-by: James Hogan <james.hogan@mips.com>
Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: 72b22bbad1e7 ("MIPS: Don't assume 64-bit FP registers for FP regset")
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17928/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Consistently handle buffer counter with PTRACE_SETREGSET
Maciej W. Rozycki [Mon, 11 Dec 2017 22:53:14 +0000 (22:53 +0000)]
MIPS: Consistently handle buffer counter with PTRACE_SETREGSET

commit 80b3ffce0196ea50068885d085ff981e4b8396f4 upstream.

Update commit d614fd58a283 ("mips/ptrace: Preserve previous registers
for short regset write") bug and consistently consume all data supplied
to `fpr_set_msa' with the ptrace(2) PTRACE_SETREGSET request, such that
a zero data buffer counter is returned where insufficient data has been
given to fill a whole number of FP general registers.

In reality this is not going to happen, as the caller is supposed to
only supply data covering a whole number of registers and it is verified
in `ptrace_regset' and again asserted in `fpr_set', however structuring
code such that the presence of trailing partial FP general register data
causes `fpr_set_msa' to return with a non-zero data buffer counter makes
it appear that this trailing data will be used if there are subsequent
writes made to FP registers, which is going to be the case with the FCSR
once the missing write to that register has been fixed.

Fixes: d614fd58a283 ("mips/ptrace: Preserve previous registers for short regset write")
Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Cc: James Hogan <james.hogan@mips.com>
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Alex Smith <alex@alex-smith.me.uk>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17927/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Guard against any partial write attempt with PTRACE_SETREGSET
Maciej W. Rozycki [Mon, 11 Dec 2017 22:52:15 +0000 (22:52 +0000)]
MIPS: Guard against any partial write attempt with PTRACE_SETREGSET

commit dc24d0edf33c3e15099688b6bbdf7bdc24bf6e91 upstream.

Complement commit d614fd58a283 ("mips/ptrace: Preserve previous
registers for short regset write") and ensure that no partial register
write attempt is made with PTRACE_SETREGSET, as we do not preinitialize
any temporaries used to hold incoming register data and consequently
random data could be written.

It is the responsibility of the caller, such as `ptrace_regset', to
arrange for writes to span whole registers only, so here we only assert
that it has indeed happened.

Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: 72b22bbad1e7 ("MIPS: Don't assume 64-bit FP registers for FP regset")
Cc: James Hogan <james.hogan@mips.com>
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Alex Smith <alex@alex-smith.me.uk>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17926/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Factor out NT_PRFPREG regset access helpers
Maciej W. Rozycki [Mon, 11 Dec 2017 22:51:35 +0000 (22:51 +0000)]
MIPS: Factor out NT_PRFPREG regset access helpers

commit a03fe72572c12e98f4173f8a535f32468e48b6ec upstream.

In preparation to fix a commit 72b22bbad1e7 ("MIPS: Don't assume 64-bit
FP registers for FP regset") FCSR access regression factor out
NT_PRFPREG regset access helpers for the non-MSA and the MSA variants
respectively, to avoid having to deal with excessive indentation in the
actual fix.

No functional change, however use `target->thread.fpu.fpr[0]' rather
than `target->thread.fpu.fpr[i]' for FGR holding type size determination
as there's no `i' variable to refer to anymore, and for the factored out
`i' variable declaration use `unsigned int' rather than `unsigned' as
its type, following the common style.

Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: 72b22bbad1e7 ("MIPS: Don't assume 64-bit FP registers for FP regset")
Cc: James Hogan <james.hogan@mips.com>
Cc: Paul Burton <Paul.Burton@mips.com>
Cc: Alex Smith <alex@alex-smith.me.uk>
Cc: Dave Martin <Dave.Martin@arm.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17925/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMIPS: Validate PR_SET_FP_MODE prctl(2) requests against the ABI of the task
Maciej W. Rozycki [Mon, 27 Nov 2017 09:33:03 +0000 (09:33 +0000)]
MIPS: Validate PR_SET_FP_MODE prctl(2) requests against the ABI of the task

commit b67336eee3fcb8ecedc6c13e2bf88aacfa3151e2 upstream.

Fix an API loophole introduced with commit 9791554b45a2 ("MIPS,prctl:
add PR_[GS]ET_FP_MODE prctl options for MIPS"), where the caller of
prctl(2) is incorrectly allowed to make a change to CP0.Status.FR or
CP0.Config5.FRE register bits even if CONFIG_MIPS_O32_FP64_SUPPORT has
not been enabled, despite that an executable requesting the mode
requested via ELF file annotation would not be allowed to run in the
first place, or for n64 and n64 ABI tasks which do not have non-default
modes defined at all.  Add suitable checks to `mips_set_process_fp_mode'
and bail out if an invalid mode change has been requested for the ABI in
effect, even if the FPU hardware or emulation would otherwise allow it.

Always succeed however without taking any further action if the mode
requested is the same as one already in effect, regardless of whether
any mode change, should it be requested, would actually be allowed for
the task concerned.

Signed-off-by: Maciej W. Rozycki <macro@mips.com>
Fixes: 9791554b45a2 ("MIPS,prctl: add PR_[GS]ET_FP_MODE prctl options for MIPS")
Reviewed-by: Paul Burton <paul.burton@mips.com>
Cc: James Hogan <james.hogan@mips.com>
Cc: linux-mips@linux-mips.org
Cc: linux-kernel@vger.kernel.org
Patchwork: https://patchwork.linux-mips.org/patch/17800/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoIB/srpt: Disable RDMA access by the initiator
Bart Van Assche [Wed, 3 Jan 2018 21:39:15 +0000 (13:39 -0800)]
IB/srpt: Disable RDMA access by the initiator

commit bec40c26041de61162f7be9d2ce548c756ce0f65 upstream.

With the SRP protocol all RDMA operations are initiated by the target.
Since no RDMA operations are initiated by the initiator, do not grant
the initiator permission to submit RDMA reads or writes to the target.

Signed-off-by: Bart Van Assche <bart.vanassche@wdc.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocan: gs_usb: fix return value of the "set_bittiming" callback
Wolfgang Grandegger [Wed, 13 Dec 2017 18:52:23 +0000 (19:52 +0100)]
can: gs_usb: fix return value of the "set_bittiming" callback

commit d5b42e6607661b198d8b26a0c30969605b1bf5c7 upstream.

The "set_bittiming" callback treats a positive return value as error!
For that reason "can_changelink()" will quit silently after setting
the bittiming values without processing ctrlmode, restart-ms, etc.

Signed-off-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoKVM: Fix stack-out-of-bounds read in write_mmio
Wanpeng Li [Fri, 15 Dec 2017 01:40:50 +0000 (17:40 -0800)]
KVM: Fix stack-out-of-bounds read in write_mmio

commit e39d200fa5bf5b94a0948db0dae44c1b73b84a56 upstream.

Reported by syzkaller:

  BUG: KASAN: stack-out-of-bounds in write_mmio+0x11e/0x270 [kvm]
  Read of size 8 at addr ffff8803259df7f8 by task syz-executor/32298

  CPU: 6 PID: 32298 Comm: syz-executor Tainted: G           OE    4.15.0-rc2+ #18
  Hardware name: LENOVO ThinkCentre M8500t-N000/SHARKBAY, BIOS FBKTC1AUS 02/16/2016
  Call Trace:
   dump_stack+0xab/0xe1
   print_address_description+0x6b/0x290
   kasan_report+0x28a/0x370
   write_mmio+0x11e/0x270 [kvm]
   emulator_read_write_onepage+0x311/0x600 [kvm]
   emulator_read_write+0xef/0x240 [kvm]
   emulator_fix_hypercall+0x105/0x150 [kvm]
   em_hypercall+0x2b/0x80 [kvm]
   x86_emulate_insn+0x2b1/0x1640 [kvm]
   x86_emulate_instruction+0x39a/0xb90 [kvm]
   handle_exception+0x1b4/0x4d0 [kvm_intel]
   vcpu_enter_guest+0x15a0/0x2640 [kvm]
   kvm_arch_vcpu_ioctl_run+0x549/0x7d0 [kvm]
   kvm_vcpu_ioctl+0x479/0x880 [kvm]
   do_vfs_ioctl+0x142/0x9a0
   SyS_ioctl+0x74/0x80
   entry_SYSCALL_64_fastpath+0x23/0x9a

The path of patched vmmcall will patch 3 bytes opcode 0F 01 C1(vmcall)
to the guest memory, however, write_mmio tracepoint always prints 8 bytes
through *(u64 *)val since kvm splits the mmio access into 8 bytes. This
leaks 5 bytes from the kernel stack (CVE-2017-17741).  This patch fixes
it by just accessing the bytes which we operate on.

Before patch:

syz-executor-5567  [007] .... 51370.561696: kvm_mmio: mmio write len 3 gpa 0x10 val 0x1ffff10077c1010f

After patch:

syz-executor-13416 [002] .... 51302.299573: kvm_mmio: mmio write len 3 gpa 0x10 val 0xc1010f

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Reviewed-by: Marc Zyngier <marc.zyngier@arm.com>
Tested-by: Marc Zyngier <marc.zyngier@arm.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agodm bufio: fix shrinker scans when (nr_to_scan < retain_target)
Suren Baghdasaryan [Wed, 6 Dec 2017 17:27:30 +0000 (09:27 -0800)]
dm bufio: fix shrinker scans when (nr_to_scan < retain_target)

commit fbc7c07ec23c040179384a1f16b62b6030eb6bdd upstream.

When system is under memory pressure it is observed that dm bufio
shrinker often reclaims only one buffer per scan. This change fixes
the following two issues in dm bufio shrinker that cause this behavior:

1. ((nr_to_scan - freed) <= retain_target) condition is used to
terminate slab scan process. This assumes that nr_to_scan is equal
to the LRU size, which might not be correct because do_shrink_slab()
in vmscan.c calculates nr_to_scan using multiple inputs.
As a result when nr_to_scan is less than retain_target (64) the scan
will terminate after the first iteration, effectively reclaiming one
buffer per scan and making scans very inefficient. This hurts vmscan
performance especially because mutex is acquired/released every time
dm_bufio_shrink_scan() is called.
New implementation uses ((LRU size - freed) <= retain_target)
condition for scan termination. LRU size can be safely determined
inside __scan() because this function is called after dm_bufio_lock().

2. do_shrink_slab() uses value returned by dm_bufio_shrink_count() to
determine number of freeable objects in the slab. However dm_bufio
always retains retain_target buffers in its LRU and will terminate
a scan when this mark is reached. Therefore returning the entire LRU size
from dm_bufio_shrink_count() is misleading because that does not
represent the number of freeable objects that slab will reclaim during
a scan. Returning (LRU size - retain_target) better represents the
number of freeable objects in the slab. This way do_shrink_slab()
returns 0 when (LRU size < retain_target) and vmscan will not try to
scan this shrinker avoiding scans that will not reclaim any memory.

Test: tested using Android device running
<AOSP>/system/extras/alloc-stress that generates memory pressure
and causes intensive shrinker scans

Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoLinux 4.4.111
Greg Kroah-Hartman [Wed, 10 Jan 2018 08:27:15 +0000 (09:27 +0100)]
Linux 4.4.111

6 years agoFix build error in vma.c
Greg Kroah-Hartman [Tue, 9 Jan 2018 09:24:02 +0000 (10:24 +0100)]
Fix build error in vma.c

This fixes the following much-reported build issue:

arch/x86/entry/vdso/vma.c: In function ‘map_vdso’:
arch/x86/entry/vdso/vma.c:175:9: error:
        implicit declaration of function ‘pvclock_pvti_cpu0_va’

on some arches and configurations.

Thanks to Guenter for being persistent enough to get it fixed :)

Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoMap the vsyscall page with _PAGE_USER
Borislav Petkov [Thu, 4 Jan 2018 16:42:45 +0000 (17:42 +0100)]
Map the vsyscall page with _PAGE_USER

This needs to happen early in kaiser_pagetable_walk(), before the
hierarchy is established so that _PAGE_USER permission can be really
set.

A proper fix would be to teach kaiser_pagetable_walk() to update those
permissions but the vsyscall page is the only exception here so ...

Signed-off-by: Borislav Petkov <bp@suse.de>
Acked-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoproc: much faster /proc/vmstat
Alexey Dobriyan [Sat, 8 Oct 2016 00:02:14 +0000 (17:02 -0700)]
proc: much faster /proc/vmstat

commit 68ba0326b4e14988f9e0c24a6e12a85cf2acd1ca upstream.

Every current KDE system has process named ksysguardd polling files
below once in several seconds:

$ strace -e trace=open -p $(pidof ksysguardd)
Process 1812 attached
open("/etc/mtab", O_RDONLY|O_CLOEXEC)   = 8
open("/etc/mtab", O_RDONLY|O_CLOEXEC)   = 8
open("/proc/net/dev", O_RDONLY)         = 8
open("/proc/net/wireless", O_RDONLY)    = -1 ENOENT (No such file or directory)
open("/proc/stat", O_RDONLY)            = 8
open("/proc/vmstat", O_RDONLY)          = 8

Hell knows what it is doing but speed up reading /proc/vmstat by 33%!

Benchmark is open+read+close 1.000.000 times.

BEFORE
$ perf stat -r 10 taskset -c 3 ./proc-vmstat

 Performance counter stats for 'taskset -c 3 ./proc-vmstat' (10 runs):

      13146.768464      task-clock (msec)         #    0.960 CPUs utilized            ( +-  0.60% )
                15      context-switches          #    0.001 K/sec                    ( +-  1.41% )
                 1      cpu-migrations            #    0.000 K/sec                    ( +- 11.11% )
               104      page-faults               #    0.008 K/sec                    ( +-  0.57% )
    45,489,799,349      cycles                    #    3.460 GHz                      ( +-  0.03% )
     9,970,175,743      stalled-cycles-frontend   #   21.92% frontend cycles idle     ( +-  0.10% )
     2,800,298,015      stalled-cycles-backend    #   6.16% backend cycles idle       ( +-  0.32% )
    79,241,190,850      instructions              #    1.74  insn per cycle
                                                  #    0.13  stalled cycles per insn  ( +-  0.00% )
    17,616,096,146      branches                  # 1339.956 M/sec                    ( +-  0.00% )
       176,106,232      branch-misses             #    1.00% of all branches          ( +-  0.18% )

      13.691078109 seconds time elapsed                                          ( +-  0.03% )
      ^^^^^^^^^^^^

AFTER
$ perf stat -r 10 taskset -c 3 ./proc-vmstat

 Performance counter stats for 'taskset -c 3 ./proc-vmstat' (10 runs):

       8688.353749      task-clock (msec)         #    0.950 CPUs utilized            ( +-  1.25% )
                10      context-switches          #    0.001 K/sec                    ( +-  2.13% )
                 1      cpu-migrations            #    0.000 K/sec
               104      page-faults               #    0.012 K/sec                    ( +-  0.56% )
    30,384,010,730      cycles                    #    3.497 GHz                      ( +-  0.07% )
    12,296,259,407      stalled-cycles-frontend   #   40.47% frontend cycles idle     ( +-  0.13% )
     3,370,668,651      stalled-cycles-backend    #  11.09% backend cycles idle       ( +-  0.69% )
    28,969,052,879      instructions              #    0.95  insn per cycle
                                                  #    0.42  stalled cycles per insn  ( +-  0.01% )
     6,308,245,891      branches                  #  726.058 M/sec                    ( +-  0.00% )
       214,685,502      branch-misses             #    3.40% of all branches          ( +-  0.26% )

       9.146081052 seconds time elapsed                                          ( +-  0.07% )
       ^^^^^^^^^^^

vsnprintf() is slow because:

1. format_decode() is busy looking for format specifier: 2 branches
   per character (not in this case, but in others)

2. approximately million branches while parsing format mini language
   and everywhere

3.  just look at what string() does /proc/vmstat is good case because
   most of its content are strings

Link: http://lkml.kernel.org/r/20160806125455.GA1187@p183.telecom.by
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Joe Perches <joe@perches.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Mel Gorman <mgorman@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomodule: Issue warnings when tainting kernel
Libor Pechacek [Wed, 13 Apr 2016 01:36:12 +0000 (11:06 +0930)]
module: Issue warnings when tainting kernel

commit 3205c36cf7d96024626f92d65f560035df1abcb2 upstream.

While most of the locations where a kernel taint bit is set are accompanied
with a warning message, there are two which set their bits silently.  If
the tainting module gets unloaded later on, it is almost impossible to tell
what was the reason for setting the flag.

Signed-off-by: Libor Pechacek <lpechacek@suse.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agomodule: keep percpu symbols in module's symtab
Miroslav Benes [Thu, 26 Nov 2015 02:48:06 +0000 (13:18 +1030)]
module: keep percpu symbols in module's symtab

commit e0224418516b4d8a6c2160574bac18447c354ef0 upstream.

Currently, percpu symbols from .data..percpu ELF section of a module are
not copied over and stored in final symtab array of struct module.
Consequently such symbol cannot be returned via kallsyms API (for
example kallsyms_lookup_name). This can be especially confusing when the
percpu symbol is exported. Only its __ksymtab et al. are present in its
symtab.

The culprit is in layout_and_allocate() function where SHF_ALLOC flag is
dropped for .data..percpu section. There is in fact no need to copy the
section to final struct module, because kernel module loader allocates
extra percpu section by itself. Unfortunately only symbols from
SHF_ALLOC sections are copied due to a check in is_core_symbol().

The patch changes is_core_symbol() function to copy over also percpu
symbols (their st_shndx points to .data..percpu ELF section). We do it
only if CONFIG_KALLSYMS_ALL is set to be consistent with the rest of the
function (ELF section is SHF_ALLOC but !SHF_EXECINSTR). Finally
elf_type() returns type 'a' for a percpu symbol because its address is
absolute.

Signed-off-by: Miroslav Benes <mbenes@suse.cz>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agogenksyms: Handle string literals with spaces in reference files
Michal Marek [Wed, 9 Dec 2015 14:08:21 +0000 (15:08 +0100)]
genksyms: Handle string literals with spaces in reference files

commit a78f70e8d65e88b9f631d073f68cb26dcd746298 upstream.

The reference files use spaces to separate tokens, however, we must
preserve spaces inside string literals. Currently the only case in the
tree is struct edac_raw_error_desc in <linux/edac.h>:

$ KBUILD_SYMTYPES=1 make -s drivers/edac/amd64_edac.symtypes
$ mv drivers/edac/amd64_edac.{symtypes,symref}
$ KBUILD_SYMTYPES=1 make -s drivers/edac/amd64_edac.symtypes
drivers/edac/amd64_edac.c:527: warning: amd64_get_dram_hole_info: modversion changed because of changes in struct edac_raw_error_desc

Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/tlb: Drop the _GPL from the cpu_tlbstate export
Thomas Gleixner [Thu, 4 Jan 2018 21:19:04 +0000 (22:19 +0100)]
x86/tlb: Drop the _GPL from the cpu_tlbstate export

commit 1e5476815fd7f98b888e01a0f9522b63085f96c9 upstream.

The recent changes for PTI touch cpu_tlbstate from various tlb_flush
inlines. cpu_tlbstate is exported as GPL symbol, so this causes a
regression when building out of tree drivers for certain graphics cards.

Aside of that the export was wrong since it was introduced as it should
have been EXPORT_PER_CPU_SYMBOL_GPL().

Use the correct PER_CPU export and drop the _GPL to restore the previous
state which allows users to utilize the cards they payed for.

As always I'm really thrilled to make this kind of change to support the
#friends (or however the hot hashtag of today is spelled) from that closet
sauce graphics corp.

Fixes: 1e02ce4cccdc ("x86: Store a per-cpu shadow copy of CR4")
Fixes: 6fd166aae78c ("x86/mm: Use/Fix PCID to optimize user/kernel switches")
Reported-by: Kees Cook <keescook@google.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Thomas Backlund <tmb@mageia.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoparisc: Fix alignment of pa_tlb_lock in assembly on 32-bit SMP kernel
Helge Deller [Tue, 2 Jan 2018 19:36:44 +0000 (20:36 +0100)]
parisc: Fix alignment of pa_tlb_lock in assembly on 32-bit SMP kernel

commit 88776c0e70be0290f8357019d844aae15edaa967 upstream.

Qemu for PARISC reported on a 32bit SMP parisc kernel strange failures
about "Not-handled unaligned insn 0x0e8011d6 and 0x0c2011c9."

Those opcodes evaluate to the ldcw() assembly instruction which requires
(on 32bit) an alignment of 16 bytes to ensure atomicity.

As it turns out, qemu is correct and in our assembly code in entry.S and
pacache.S we don't pay attention to the required alignment.

This patch fixes the problem by aligning the lock offset in assembly
code in the same manner as we do in our C-code.

Signed-off-by: Helge Deller <deller@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/microcode/AMD: Add support for fam17h microcode loading
Tom Lendacky [Thu, 30 Nov 2017 22:46:40 +0000 (16:46 -0600)]
x86/microcode/AMD: Add support for fam17h microcode loading

commit f4e9b7af0cd58dd039a0fb2cd67d57cea4889abf upstream.

The size for the Microcode Patch Block (MPB) for an AMD family 17h
processor is 3200 bytes.  Add a #define for fam17h so that it does
not default to 2048 bytes and fail a microcode load/update.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Borislav Petkov <bp@alien8.de>
Link: https://lkml.kernel.org/r/20171130224640.15391.40247.stgit@tlendack-t1.amdoffice.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Alice Ferrazzi <alicef@gentoo.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoInput: elantech - add new icbody type 15
Aaron Ma [Sun, 26 Nov 2017 00:48:41 +0000 (16:48 -0800)]
Input: elantech - add new icbody type 15

commit 10d900303f1c3a821eb0bef4e7b7ece16768fba4 upstream.

The touchpad of Lenovo Thinkpad L480 reports it's version as 15.

Signed-off-by: Aaron Ma <aaron.ma@canonical.com>
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoARC: uaccess: dont use "l" gcc inline asm constraint modifier
Vineet Gupta [Fri, 8 Dec 2017 16:26:58 +0000 (08:26 -0800)]
ARC: uaccess: dont use "l" gcc inline asm constraint modifier

commit 79435ac78d160e4c245544d457850a56f805ac0d upstream.

This used to setup the LP_COUNT register automatically, but now has been
removed.

There was an earlier fix 3c7c7a2fc8811 which fixed instance in delay.h but
somehow missed this one as gcc change had not made its way into
production toolchains and was not pedantic as it is now !

Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokernel/signal.c: remove the no longer needed SIGNAL_UNKILLABLE check in complete_signal()
Oleg Nesterov [Fri, 17 Nov 2017 23:30:08 +0000 (15:30 -0800)]
kernel/signal.c: remove the no longer needed SIGNAL_UNKILLABLE check in complete_signal()

commit 426915796ccaf9c2bd9bb06dc5702225957bc2e5 upstream.

complete_signal() checks SIGNAL_UNKILLABLE before it starts to destroy
the thread group, today this is wrong in many ways.

If nothing else, fatal_signal_pending() should always imply that the
whole thread group (except ->group_exit_task if it is not NULL) is
killed, this check breaks the rule.

After the previous changes we can rely on sig_task_ignored();
sig_fatal(sig) && SIGNAL_UNKILLABLE can only be true if we actually want
to kill this task and sig == SIGKILL OR it is traced and debugger can
intercept the signal.

This should hopefully fix the problem reported by Dmitry.  This
test-case

static int init(void *arg)
{
for (;;)
pause();
}

int main(void)
{
char stack[16 * 1024];

for (;;) {
int pid = clone(init, stack + sizeof(stack)/2,
CLONE_NEWPID | SIGCHLD, NULL);
assert(pid > 0);

assert(ptrace(PTRACE_ATTACH, pid, 0, 0) == 0);
assert(waitpid(-1, NULL, WSTOPPED) == pid);

assert(ptrace(PTRACE_DETACH, pid, 0, SIGSTOP) == 0);
assert(syscall(__NR_tkill, pid, SIGKILL) == 0);
assert(pid == wait(NULL));
}
}

triggers the WARN_ON_ONCE(!(task->jobctl & JOBCTL_STOP_PENDING)) in
task_participate_group_stop().  do_signal_stop()->signal_group_exit()
checks SIGNAL_GROUP_EXIT and return false, but task_set_jobctl_pending()
checks fatal_signal_pending() and does not set JOBCTL_STOP_PENDING.

And his should fix the minor security problem reported by Kyle,
SECCOMP_RET_TRACE can miss fatal_signal_pending() the same way if the
task is the root of a pid namespace.

Link: http://lkml.kernel.org/r/20171103184246.GD21036@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Reported-by: Kyle Huey <me@kylehuey.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Tested-by: Kyle Huey <me@kylehuey.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokernel/signal.c: protect the SIGNAL_UNKILLABLE tasks from !sig_kernel_only() signals
Oleg Nesterov [Fri, 17 Nov 2017 23:30:04 +0000 (15:30 -0800)]
kernel/signal.c: protect the SIGNAL_UNKILLABLE tasks from !sig_kernel_only() signals

commit ac25385089f673560867eb5179228a44ade0cfc1 upstream.

Change sig_task_ignored() to drop the SIG_DFL && !sig_kernel_only()
signals even if force == T.  This simplifies the next change and this
matches the same check in get_signal() which will drop these signals
anyway.

Link: http://lkml.kernel.org/r/20171103184227.GC21036@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Tested-by: Kyle Huey <me@kylehuey.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokernel/signal.c: protect the traced SIGNAL_UNKILLABLE tasks from SIGKILL
Oleg Nesterov [Fri, 17 Nov 2017 23:30:01 +0000 (15:30 -0800)]
kernel/signal.c: protect the traced SIGNAL_UNKILLABLE tasks from SIGKILL

commit 628c1bcba204052d19b686b5bac149a644cdb72e upstream.

The comment in sig_ignored() says "Tracers may want to know about even
ignored signals" but SIGKILL can not be reported to debugger and it is
just wrong to return 0 in this case: SIGKILL should only kill the
SIGNAL_UNKILLABLE task if it comes from the parent ns.

Change sig_ignored() to ignore ->ptrace if sig == SIGKILL and rely on
sig_task_ignored().

SISGTOP coming from within the namespace is not really right too but at
least debugger can intercept it, and we can't drop it here because this
will break "gdb -p 1": ptrace_attach() won't work.  Perhaps we will add
another ->ptrace check later, we will see.

Link: http://lkml.kernel.org/r/20171103184206.GB21036@redhat.com
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Tested-by: Kyle Huey <me@kylehuey.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokernel: make groups_sort calling a responsibility group_info allocators
Thiago Rafael Becker [Thu, 14 Dec 2017 23:33:12 +0000 (15:33 -0800)]
kernel: make groups_sort calling a responsibility group_info allocators

commit bdcf0a423ea1c40bbb40e7ee483b50fc8aa3d758 upstream.

In testing, we found that nfsd threads may call set_groups in parallel
for the same entry cached in auth.unix.gid, racing in the call of
groups_sort, corrupting the groups for that entry and leading to
permission denials for the client.

This patch:
 - Make groups_sort globally visible.
 - Move the call to groups_sort to the modifiers of group_info
 - Remove the call to groups_sort from set_groups

Link: http://lkml.kernel.org/r/20171211151420.18655-1-thiago.becker@gmail.com
Signed-off-by: Thiago Rafael Becker <thiago.becker@gmail.com>
Reviewed-by: Matthew Wilcox <mawilcox@microsoft.com>
Reviewed-by: NeilBrown <neilb@suse.com>
Acked-by: "J. Bruce Fields" <bfields@fieldses.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agofscache: Fix the default for fscache_maybe_release_page()
David Howells [Tue, 2 Jan 2018 10:02:19 +0000 (10:02 +0000)]
fscache: Fix the default for fscache_maybe_release_page()

commit 98801506552593c9b8ac11021b0cdad12cab4f6b upstream.

Fix the default for fscache_maybe_release_page() for when the cookie isn't
valid or the page isn't cached.  It mustn't return false as that indicates
the page cannot yet be freed.

The problem with the default is that if, say, there's no cache, but a
network filesystem's pages are using up almost all the available memory, a
system can OOM because the filesystem ->releasepage() op will not allow
them to be released as fscache_maybe_release_page() incorrectly prevents
it.

This can be tested by writing a sequence of 512MiB files to an AFS mount.
It does not affect NFS or CIFS because both of those wrap the call in a
check of PG_fscache and it shouldn't bother Ceph as that only has
PG_private set whilst writeback is in progress.  This might be an issue for
9P, however.

Note that the pages aren't entirely stuck.  Removing a file or unmounting
will clear things because that uses ->invalidatepage() instead.

Fixes: 201a15428bd5 ("FS-Cache: Handle pages pending storage that get evicted under OOM conditions")
Reported-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Jeff Layton <jlayton@redhat.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Tested-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agosunxi-rsb: Include OF based modalias in device uevent
Stefan Brüns [Mon, 27 Nov 2017 19:05:34 +0000 (20:05 +0100)]
sunxi-rsb: Include OF based modalias in device uevent

commit e2bf801ecd4e62222a46d1ba9e57e710171d29c1 upstream.

Include the OF-based modalias in the uevent sent when registering devices
on the sunxi RSB bus, so that user space has a chance to autoload the
kernel module for the device.

Fixes a regression caused by commit 3f241bfa60bd ("arm64: allwinner: a64:
pine64: Use dcdc1 regulator for mmc0"). When the axp20x-rsb module for
the AXP803 PMIC is built as a module, it is not loaded and the system
ends up with an disfunctional MMC controller.

Fixes: d787dcdb9c8f ("bus: sunxi-rsb: Add driver for Allwinner Reduced Serial Bus")
Acked-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Stefan Brüns <stefan.bruens@rwth-aachen.de>
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocrypto: pcrypt - fix freeing pcrypt instances
Eric Biggers [Wed, 20 Dec 2017 22:28:25 +0000 (14:28 -0800)]
crypto: pcrypt - fix freeing pcrypt instances

commit d76c68109f37cb85b243a1cf0f40313afd2bae68 upstream.

pcrypt is using the old way of freeing instances, where the ->free()
method specified in the 'struct crypto_template' is passed a pointer to
the 'struct crypto_instance'.  But the crypto_instance is being
kfree()'d directly, which is incorrect because the memory was actually
allocated as an aead_instance, which contains the crypto_instance at a
nonzero offset.  Thus, the wrong pointer was being kfree()'d.

Fix it by switching to the new way to free aead_instance's where the
->free() method is specified in the aead_instance itself.

Reported-by: syzbot <syzkaller@googlegroups.com>
Fixes: 0496f56065e0 ("crypto: pcrypt - Add support for new AEAD interface")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocrypto: chacha20poly1305 - validate the digest size
Eric Biggers [Mon, 11 Dec 2017 20:15:17 +0000 (12:15 -0800)]
crypto: chacha20poly1305 - validate the digest size

commit e57121d08c38dabec15cf3e1e2ad46721af30cae upstream.

If the rfc7539 template was instantiated with a hash algorithm with
digest size larger than 16 bytes (POLY1305_DIGEST_SIZE), then the digest
overran the 'tag' buffer in 'struct chachapoly_req_ctx', corrupting the
subsequent memory, including 'cryptlen'.  This caused a crash during
crypto_skcipher_decrypt().

Fix it by, when instantiating the template, requiring that the
underlying hash algorithm has the digest size expected for Poly1305.

Reproducer:

    #include <linux/if_alg.h>
    #include <sys/socket.h>
    #include <unistd.h>

    int main()
    {
            int algfd, reqfd;
            struct sockaddr_alg addr = {
                    .salg_type = "aead",
                    .salg_name = "rfc7539(chacha20,sha256)",
            };
            unsigned char buf[32] = { 0 };

            algfd = socket(AF_ALG, SOCK_SEQPACKET, 0);
            bind(algfd, (void *)&addr, sizeof(addr));
            setsockopt(algfd, SOL_ALG, ALG_SET_KEY, buf, sizeof(buf));
            reqfd = accept(algfd, 0, 0);
            write(reqfd, buf, 16);
            read(reqfd, buf, 16);
    }

Reported-by: syzbot <syzkaller@googlegroups.com>
Fixes: 71ebc4d1b27d ("crypto: chacha20poly1305 - Add a ChaCha20-Poly1305 AEAD construction, RFC7539")
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agocrypto: n2 - cure use after free
Jan Engelhardt [Tue, 19 Dec 2017 18:09:07 +0000 (19:09 +0100)]
crypto: n2 - cure use after free

commit 203f45003a3d03eea8fa28d74cfc74c354416fdb upstream.

queue_cache_init is first called for the Control Word Queue
(n2_crypto_probe). At that time, queue_cache[0] is NULL and a new
kmem_cache will be allocated. If the subsequent n2_register_algs call
fails, the kmem_cache will be released in queue_cache_destroy, but
queue_cache_init[0] is not set back to NULL.

So when the Module Arithmetic Unit gets probed next (n2_mau_probe),
queue_cache_init will not allocate a kmem_cache again, but leave it
as its bogus value, causing a BUG() to trigger when queue_cache[0] is
eventually passed to kmem_cache_zalloc:

n2_crypto: Found N2CP at /virtual-devices@100/n2cp@7
n2_crypto: Registered NCS HVAPI version 2.0
called queue_cache_init
n2_crypto: md5 alg registration failed
n2cp f028687c: /virtual-devices@100/n2cp@7: Unable to register algorithms.
called queue_cache_destroy
n2cp: probe of f028687c failed with error -22
n2_crypto: Found NCP at /virtual-devices@100/ncp@6
n2_crypto: Registered NCS HVAPI version 2.0
called queue_cache_init
kernel BUG at mm/slab.c:2993!
Call Trace:
 [0000000000604488] kmem_cache_alloc+0x1a8/0x1e0
                  (inlined) kmem_cache_zalloc
                  (inlined) new_queue
                  (inlined) spu_queue_setup
                  (inlined) handle_exec_unit
 [0000000010c61eb4] spu_mdesc_scan+0x1f4/0x460 [n2_crypto]
 [0000000010c62b80] n2_mau_probe+0x100/0x220 [n2_crypto]
 [000000000084b174] platform_drv_probe+0x34/0xc0

Signed-off-by: Jan Engelhardt <jengelh@inai.de>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokernel/acct.c: fix the acct->needcheck check in check_free_space()
Oleg Nesterov [Fri, 5 Jan 2018 00:17:49 +0000 (16:17 -0800)]
kernel/acct.c: fix the acct->needcheck check in check_free_space()

commit 4d9570158b6260f449e317a5f9ed030c2504a615 upstream.

As Tsukada explains, the time_is_before_jiffies(acct->needcheck) check
is very wrong, we need time_is_after_jiffies() to make sys_acct() work.

Ignoring the overflows, the code should "goto out" if needcheck >
jiffies, while currently it checks "needcheck < jiffies" and thus in the
likely case check_free_space() does nothing until jiffies overflow.

In particular this means that sys_acct() is simply broken, acct_on()
sets acct->needcheck = jiffies and expects that check_free_space()
should set acct->active = 1 after the free-space check, but this won't
happen if jiffies increments in between.

This was broken by commit 32dc73086015 ("get rid of timer in
kern/acct.c") in 2011, then another (correct) commit 795a2f22a8ea
("acct() should honour the limits from the very beginning") made the
problem more visible.

Link: http://lkml.kernel.org/r/20171213133940.GA6554@redhat.com
Fixes: 32dc73086015 ("get rid of timer in kern/acct.c")
Reported-by: TSUKADA Koutaro <tsukada@ascade.co.jp>
Suggested-by: TSUKADA Koutaro <tsukada@ascade.co.jp>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kasan: Write protect kasan zero shadow
Andrey Ryabinin [Mon, 11 Jan 2016 12:51:19 +0000 (15:51 +0300)]
x86/kasan: Write protect kasan zero shadow

commit 063fb3e56f6dd29b2633b678b837e1d904200e6f upstream.

After kasan_init() executed, no one is allowed to write to kasan_zero_page,
so write protect it.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reviewed-by: Borislav Petkov <bp@suse.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Luis R. Rodriguez <mcgrof@suse.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/1452516679-32040-3-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoLinux 4.4.110
Greg Kroah-Hartman [Fri, 5 Jan 2018 14:44:27 +0000 (15:44 +0100)]
Linux 4.4.110

6 years agokaiser: Set _PAGE_NX only if supported
Guenter Roeck [Thu, 4 Jan 2018 21:41:55 +0000 (13:41 -0800)]
kaiser: Set _PAGE_NX only if supported

This resolves a crash if loaded under qemu + haxm under windows.
See https://www.spinics.net/lists/kernel/msg2689835.html for details.
Here is a boot log (the log is from chromeos-4.4, but Tao Wu says that
the same log is also seen with vanilla v4.4.110-rc1).

[    0.712750] Freeing unused kernel memory: 552K
[    0.721821] init: Corrupted page table at address 57b029b332e0
[    0.722761] PGD 80000000bb238067 PUD bc36a067 PMD bc369067 PTE 45d2067
[    0.722761] Bad pagetable: 000b [#1] PREEMPT SMP
[    0.722761] Modules linked in:
[    0.722761] CPU: 1 PID: 1 Comm: init Not tainted 4.4.96 #31
[    0.722761] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
rel-1.7.5.1-0-g8936dbb-20141113_115728-nilsson.home.kraxel.org 04/01/2014
[    0.722761] task: ffff8800bc290000 ti: ffff8800bc28c000 task.ti: ffff8800bc28c000
[    0.722761] RIP: 0010:[<ffffffff83f4129e>]  [<ffffffff83f4129e>] __clear_user+0x42/0x67
[    0.722761] RSP: 0000:ffff8800bc28fcf8  EFLAGS: 00010202
[    0.722761] RAX: 0000000000000000 RBX: 00000000000001a4 RCX: 00000000000001a4
[    0.722761] RDX: 0000000000000000 RSI: 0000000000000008 RDI: 000057b029b332e0
[    0.722761] RBP: ffff8800bc28fd08 R08: ffff8800bc290000 R09: ffff8800bb2f4000
[    0.722761] R10: ffff8800bc290000 R11: ffff8800bb2f4000 R12: 000057b029b332e0
[    0.722761] R13: 0000000000000000 R14: 000057b029b33340 R15: ffff8800bb1e2a00
[    0.722761] FS:  0000000000000000(0000) GS:ffff8800bfb00000(0000) knlGS:0000000000000000
[    0.722761] CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
[    0.722761] CR2: 000057b029b332e0 CR3: 00000000bb2f8000 CR4: 00000000000006e0
[    0.722761] Stack:
[    0.722761]  000057b029b332e0 ffff8800bb95fa80 ffff8800bc28fd18 ffffffff83f4120c
[    0.722761]  ffff8800bc28fe18 ffffffff83e9e7a1 ffff8800bc28fd68 0000000000000000
[    0.722761]  ffff8800bc290000 ffff8800bc290000 ffff8800bc290000 ffff8800bc290000
[    0.722761] Call Trace:
[    0.722761]  [<ffffffff83f4120c>] clear_user+0x2e/0x30
[    0.722761]  [<ffffffff83e9e7a1>] load_elf_binary+0xa7f/0x18f7
[    0.722761]  [<ffffffff83de2088>] search_binary_handler+0x86/0x19c
[    0.722761]  [<ffffffff83de389e>] do_execveat_common.isra.26+0x909/0xf98
[    0.722761]  [<ffffffff844febe0>] ? rest_init+0x87/0x87
[    0.722761]  [<ffffffff83de40be>] do_execve+0x23/0x25
[    0.722761]  [<ffffffff83c002e3>] run_init_process+0x2b/0x2d
[    0.722761]  [<ffffffff844fec4d>] kernel_init+0x6d/0xda
[    0.722761]  [<ffffffff84505b2f>] ret_from_fork+0x3f/0x70
[    0.722761]  [<ffffffff844febe0>] ? rest_init+0x87/0x87
[    0.722761] Code: 86 84 be 12 00 00 00 e8 87 0d e8 ff 66 66 90 48 89 d8 48 c1
eb 03 4c 89 e7 83 e0 07 48 89 d9 be 08 00 00 00 31 d2 48 85 c9 74 0a <48> 89 17
48 01 f7 ff c9 75 f6 48 89 c1 85 c9 74 09 88 17 48 ff
[    0.722761] RIP  [<ffffffff83f4129e>] __clear_user+0x42/0x67
[    0.722761]  RSP <ffff8800bc28fcf8>
[    0.722761] ---[ end trace def703879b4ff090 ]---
[    0.722761] BUG: sleeping function called from invalid context at /mnt/host/source/src/third_party/kernel/v4.4/kernel/locking/rwsem.c:21
[    0.722761] in_atomic(): 0, irqs_disabled(): 1, pid: 1, name: init
[    0.722761] CPU: 1 PID: 1 Comm: init Tainted: G      D         4.4.96 #31
[    0.722761] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.7.5.1-0-g8936dbb-20141113_115728-nilsson.home.kraxel.org 04/01/2014
[    0.722761]  0000000000000086 dcb5d76098c89836 ffff8800bc28fa30 ffffffff83f34004
[    0.722761]  ffffffff84839dc2 0000000000000015 ffff8800bc28fa40 ffffffff83d57dc9
[    0.722761]  ffff8800bc28fa68 ffffffff83d57e6a ffffffff84a53640 0000000000000000
[    0.722761] Call Trace:
[    0.722761]  [<ffffffff83f34004>] dump_stack+0x4d/0x63
[    0.722761]  [<ffffffff83d57dc9>] ___might_sleep+0x13a/0x13c
[    0.722761]  [<ffffffff83d57e6a>] __might_sleep+0x9f/0xa6
[    0.722761]  [<ffffffff84502788>] down_read+0x20/0x31
[    0.722761]  [<ffffffff83cc5d9b>] __blocking_notifier_call_chain+0x35/0x63
[    0.722761]  [<ffffffff83cc5ddd>] blocking_notifier_call_chain+0x14/0x16
[    0.800374] usb 1-1: new full-speed USB device number 2 using uhci_hcd
[    0.722761]  [<ffffffff83cefe97>] profile_task_exit+0x1a/0x1c
[    0.802309]  [<ffffffff83cac84e>] do_exit+0x39/0xe7f
[    0.802309]  [<ffffffff83ce5938>] ? vprintk_default+0x1d/0x1f
[    0.802309]  [<ffffffff83d7bb95>] ? printk+0x57/0x73
[    0.802309]  [<ffffffff83c46e25>] oops_end+0x80/0x85
[    0.802309]  [<ffffffff83c7b747>] pgtable_bad+0x8a/0x95
[    0.802309]  [<ffffffff83ca7f4a>] __do_page_fault+0x8c/0x352
[    0.802309]  [<ffffffff83eefba5>] ? file_has_perm+0xc4/0xe5
[    0.802309]  [<ffffffff83ca821c>] do_page_fault+0xc/0xe
[    0.802309]  [<ffffffff84507682>] page_fault+0x22/0x30
[    0.802309]  [<ffffffff83f4129e>] ? __clear_user+0x42/0x67
[    0.802309]  [<ffffffff83f4127f>] ? __clear_user+0x23/0x67
[    0.802309]  [<ffffffff83f4120c>] clear_user+0x2e/0x30
[    0.802309]  [<ffffffff83e9e7a1>] load_elf_binary+0xa7f/0x18f7
[    0.802309]  [<ffffffff83de2088>] search_binary_handler+0x86/0x19c
[    0.802309]  [<ffffffff83de389e>] do_execveat_common.isra.26+0x909/0xf98
[    0.802309]  [<ffffffff844febe0>] ? rest_init+0x87/0x87
[    0.802309]  [<ffffffff83de40be>] do_execve+0x23/0x25
[    0.802309]  [<ffffffff83c002e3>] run_init_process+0x2b/0x2d
[    0.802309]  [<ffffffff844fec4d>] kernel_init+0x6d/0xda
[    0.802309]  [<ffffffff84505b2f>] ret_from_fork+0x3f/0x70
[    0.802309]  [<ffffffff844febe0>] ? rest_init+0x87/0x87
[    0.830559] Kernel panic - not syncing: Attempted to kill init!  exitcode=0x00000009
[    0.830559]
[    0.831305] Kernel Offset: 0x2c00000 from 0xffffffff81000000 (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
[    0.831305] ---[ end Kernel panic - not syncing: Attempted to kill init!  exitcode=0x00000009

The crash part of this problem may be solved with the following patch
(thanks to Hugh for the hint). There is still another problem, though -
with this patch applied, the qemu session aborts with "VCPU Shutdown
request", whatever that means.

Cc: lepton <ytht.net@gmail.com>
Signed-off-by: Guenter Roeck <groeck@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kasan: Clear kasan_zero_page after TLB flush
Andrey Ryabinin [Mon, 11 Jan 2016 12:51:18 +0000 (15:51 +0300)]
x86/kasan: Clear kasan_zero_page after TLB flush

commit 69e0210fd01ff157d332102219aaf5c26ca8069b upstream.

Currently we clear kasan_zero_page before __flush_tlb_all(). This
works with current implementation of native_flush_tlb[_global]()
because it doesn't cause do any writes to kasan shadow memory.
But any subtle change made in native_flush_tlb*() could break this.
Also current code seems doesn't work for paravirt guests (lguest).

Only after the TLB flush we can be sure that kasan_zero_page is not
used as early shadow anymore (instrumented code will not write to it).
So it should cleared it only after the TLB flush.

Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reviewed-by: Borislav Petkov <bp@suse.de>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Luis R. Rodriguez <mcgrof@suse.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/1452516679-32040-2-git-send-email-aryabinin@virtuozzo.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Jamie Iles <jamie.iles@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/vdso: Get pvclock data from the vvar VMA instead of the fixmap
Andy Lutomirski [Fri, 11 Dec 2015 03:20:20 +0000 (19:20 -0800)]
x86/vdso: Get pvclock data from the vvar VMA instead of the fixmap

commit dac16fba6fc590fa7239676b35ed75dae4c4cd2b upstream.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/9d37826fdc7e2d2809efe31d5345f97186859284.1449702533.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Jamie Iles <jamie.iles@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86, vdso, pvclock: Simplify and speed up the vdso pvclock reader
Andy Lutomirski [Fri, 11 Dec 2015 03:20:19 +0000 (19:20 -0800)]
x86, vdso, pvclock: Simplify and speed up the vdso pvclock reader

commit 6b078f5de7fc0851af4102493c7b5bb07e49c4cb upstream.

The pvclock vdso code was too abstracted to understand easily
and excessively paranoid.  Simplify it for a huge speedup.

This opens the door for additional simplifications, as the vdso
no longer accesses the pvti for any vcpu other than vcpu 0.

Before, vclock_gettime using kvm-clock took about 45ns on my
machine. With this change, it takes 29ns, which is almost as
fast as the pure TSC implementation.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/6b51dcc41f1b101f963945c5ec7093d72bdac429.1449702533.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Jamie Iles <jamie.iles@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoKPTI: Report when enabled
Kees Cook [Wed, 3 Jan 2018 18:43:32 +0000 (10:43 -0800)]
KPTI: Report when enabled

Make sure dmesg reports when KPTI is enabled.

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoKPTI: Rename to PAGE_TABLE_ISOLATION
Kees Cook [Wed, 3 Jan 2018 18:43:15 +0000 (10:43 -0800)]
KPTI: Rename to PAGE_TABLE_ISOLATION

This renames CONFIG_KAISER to CONFIG_PAGE_TABLE_ISOLATION.

Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kaiser: Move feature detection up
Borislav Petkov [Mon, 25 Dec 2017 12:57:16 +0000 (13:57 +0100)]
x86/kaiser: Move feature detection up

... before the first use of kaiser_enabled as otherwise funky
things happen:

  about to get started...
  (XEN) d0v0 Unhandled page fault fault/trap [#14, ec=0000]
  (XEN) Pagetable walk from ffff88022a449090:
  (XEN)  L4[0x110] = 0000000229e0e067 0000000000001e0e
  (XEN)  L3[0x008] = 0000000000000000 ffffffffffffffff
  (XEN) domain_crash_sync called from entry.S: fault at ffff82d08033fd08
  entry.o#create_bounce_frame+0x135/0x14d
  (XEN) Domain 0 (vcpu#0) crashed on cpu#0:
  (XEN) ----[ Xen-4.9.1_02-3.21  x86_64  debug=n   Not tainted ]----
  (XEN) CPU:    0
  (XEN) RIP:    e033:[<ffffffff81007460>]
  (XEN) RFLAGS: 0000000000000286   EM: 1   CONTEXT: pv guest (d0v0)

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: disabled on Xen PV
Jiri Kosina [Tue, 2 Jan 2018 13:19:49 +0000 (14:19 +0100)]
kaiser: disabled on Xen PV

Kaiser cannot be used on paravirtualized MMUs (namely reading and writing CR3).
This does not work with KAISER as the CR3 switch from and to user space PGD
would require to map the whole XEN_PV machinery into both.

More importantly, enabling KAISER on Xen PV doesn't make too much sense, as PV
guests use distinct %cr3 values for kernel and user already.

Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kaiser: Reenable PARAVIRT
Borislav Petkov [Tue, 2 Jan 2018 13:19:49 +0000 (14:19 +0100)]
x86/kaiser: Reenable PARAVIRT

Now that the required bits have been addressed, reenable
PARAVIRT.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/paravirt: Dont patch flush_tlb_single
Thomas Gleixner [Mon, 4 Dec 2017 14:07:30 +0000 (15:07 +0100)]
x86/paravirt: Dont patch flush_tlb_single

commit a035795499ca1c2bd1928808d1a156eda1420383 upstream

native_flush_tlb_single() will be changed with the upcoming
PAGE_TABLE_ISOLATION feature. This requires to have more code in
there than INVLPG.

Remove the paravirt patching for it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Josh Poimboeuf <jpoimboe@redhat.com>
Reviewed-by: Juergen Gross <jgross@suse.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bpetkov@suse.de>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: David Laight <David.Laight@aculab.com>
Cc: Denys Vlasenko <dvlasenk@redhat.com>
Cc: Eduardo Valentin <eduval@amazon.com>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Rik van Riel <riel@redhat.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: aliguori@amazon.com
Cc: daniel.gruss@iaik.tugraz.at
Cc: hughd@google.com
Cc: keescook@google.com
Cc: linux-mm@kvack.org
Cc: michael.schwarz@iaik.tugraz.at
Cc: moritz.lipp@iaik.tugraz.at
Cc: richard.fellner@student.tugraz.at
Link: https://lkml.kernel.org/r/20171204150606.828111617@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: kaiser_flush_tlb_on_return_to_user() check PCID
Hugh Dickins [Sun, 5 Nov 2017 01:43:06 +0000 (18:43 -0700)]
kaiser: kaiser_flush_tlb_on_return_to_user() check PCID

Let kaiser_flush_tlb_on_return_to_user() do the X86_FEATURE_PCID
check, instead of each caller doing it inline first: nobody needs
to optimize for the noPCID case, it's clearer this way, and better
suits later changes.  Replace those no-op X86_CR3_PCID_KERN_FLUSH lines
by a BUILD_BUG_ON() in load_new_mm_cr3(), in case something changes.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: asm/tlbflush.h handle noPGE at lower level
Hugh Dickins [Sun, 5 Nov 2017 01:23:24 +0000 (18:23 -0700)]
kaiser: asm/tlbflush.h handle noPGE at lower level

I found asm/tlbflush.h too twisty, and think it safer not to avoid
__native_flush_tlb_global_irq_disabled() in the kaiser_enabled case,
but instead let it handle kaiser_enabled along with cr3: it can just
use __native_flush_tlb() for that, no harm in re-disabling preemption.

(This is not the same change as Kirill and Dave have suggested for
upstream, flipping PGE in cr4: that's neat, but needs a cpu_has_pge
check; cr3 is enough for kaiser, and thought to be cheaper than cr4.)

Also delete the X86_FEATURE_INVPCID invpcid_flush_all_nonglobals()
preference from __native_flush_tlb(): unlike the invpcid_flush_all()
preference in __native_flush_tlb_global(), it's not seen in upstream
4.14, and was recently reported to be surprisingly slow.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: drop is_atomic arg to kaiser_pagetable_walk()
Hugh Dickins [Sun, 29 Oct 2017 18:36:19 +0000 (11:36 -0700)]
kaiser: drop is_atomic arg to kaiser_pagetable_walk()

I have not observed a might_sleep() warning from setup_fixmap_gdt()'s
use of kaiser_add_mapping() in our tree (why not?), but like upstream
we have not provided a way for that to pass is_atomic true down to
kaiser_pagetable_walk(), and at startup it's far from a likely source
of trouble: so just delete the walk's is_atomic arg and might_sleep().

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: use ALTERNATIVE instead of x86_cr3_pcid_noflush
Hugh Dickins [Wed, 4 Oct 2017 03:49:04 +0000 (20:49 -0700)]
kaiser: use ALTERNATIVE instead of x86_cr3_pcid_noflush

Now that we're playing the ALTERNATIVE game, use that more efficient
method: instead of user-mapping an extra page, and reading an extra
cacheline each time for x86_cr3_pcid_noflush.

Neel has found that __stringify(bts $X86_CR3_PCID_NOFLUSH_BIT, %rax)
is a working substitute for the "bts $63, %rax" in these ALTERNATIVEs;
but the one line with $63 in looks clearer, so let's stick with that.

Worried about what happens with an ALTERNATIVE between the jump and
jump label in another ALTERNATIVE?  I was, but have checked the
combinations in SWITCH_KERNEL_CR3_NO_STACK at entry_SYSCALL_64,
and it does a good job.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kaiser: Check boottime cmdline params
Borislav Petkov [Tue, 2 Jan 2018 13:19:48 +0000 (14:19 +0100)]
x86/kaiser: Check boottime cmdline params

AMD (and possibly other vendors) are not affected by the leak
KAISER is protecting against.

Keep the "nopti" for traditional reasons and add pti=<on|off|auto>
like upstream.

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/kaiser: Rename and simplify X86_FEATURE_KAISER handling
Borislav Petkov [Tue, 2 Jan 2018 13:19:48 +0000 (14:19 +0100)]
x86/kaiser: Rename and simplify X86_FEATURE_KAISER handling

Concentrate it in arch/x86/mm/kaiser.c and use the upstream string "nopti".

Signed-off-by: Borislav Petkov <bp@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: add "nokaiser" boot option, using ALTERNATIVE
Hugh Dickins [Sun, 24 Sep 2017 23:59:49 +0000 (16:59 -0700)]
kaiser: add "nokaiser" boot option, using ALTERNATIVE

Added "nokaiser" boot option: an early param like "noinvpcid".
Most places now check int kaiser_enabled (#defined 0 when not
CONFIG_KAISER) instead of #ifdef CONFIG_KAISER; but entry_64.S
and entry_64_compat.S are using the ALTERNATIVE technique, which
patches in the preferred instructions at runtime.  That technique
is tied to x86 cpu features, so X86_FEATURE_KAISER is fabricated.

Prior to "nokaiser", Kaiser #defined _PAGE_GLOBAL 0: revert that,
but be careful with both _PAGE_GLOBAL and CR4.PGE: setting them when
nokaiser like when !CONFIG_KAISER, but not setting either when kaiser -
neither matters on its own, but it's hard to be sure that _PAGE_GLOBAL
won't get set in some obscure corner, or something add PGE into CR4.
By omitting _PAGE_GLOBAL from __supported_pte_mask when kaiser_enabled,
all page table setup which uses pte_pfn() masks it out of the ptes.

It's slightly shameful that the same declaration versus definition of
kaiser_enabled appears in not one, not two, but in three header files
(asm/kaiser.h, asm/pgtable.h, asm/tlbflush.h).  I felt safer that way,
than with #including any of those in any of the others; and did not
feel it worth an asm/kaiser_enabled.h - kernel/cpu/common.c includes
them all, so we shall hear about it if they get out of synch.

Cleanups while in the area: removed the silly #ifdef CONFIG_KAISER
from kaiser.c; removed the unused native_get_normal_pgd(); removed
the spurious reg clutter from SWITCH_*_CR3 macro stubs; corrected some
comments.  But more interestingly, set CR4.PSE in secondary_startup_64:
the manual is clear that it does not matter whether it's 0 or 1 when
4-level-pts are enabled, but I was distracted to find cr4 different on
BSP and auxiliaries - BSP alone was adding PSE, in probe_page_size_mask().

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: fix unlikely error in alloc_ldt_struct()
Hugh Dickins [Tue, 5 Dec 2017 04:13:35 +0000 (20:13 -0800)]
kaiser: fix unlikely error in alloc_ldt_struct()

An error from kaiser_add_mapping() here is not at all likely, but
Eric Biggers rightly points out that __free_ldt_struct() relies on
new_ldt->size being initialized: move that up.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: _pgd_alloc() without __GFP_REPEAT to avoid stalls
Hugh Dickins [Fri, 13 Oct 2017 19:10:00 +0000 (12:10 -0700)]
kaiser: _pgd_alloc() without __GFP_REPEAT to avoid stalls

Synthetic filesystem mempressure testing has shown softlockups, with
hour-long page allocation stalls, and pgd_alloc() trying for order:1
with __GFP_REPEAT in one of the backtraces each time.

That's _pgd_alloc() going for a Kaiser double-pgd, using the __GFP_REPEAT
common to all page table allocations, but actually having no effect on
order:0 (see should_alloc_oom() and should_continue_reclaim() in this
tree, but beware that ports to another tree might behave differently).

Order:1 stack allocation has been working satisfactorily without
__GFP_REPEAT forever, and page table allocation only asks __GFP_REPEAT
for awkward occasions in a long-running process: it's not appropriate
at fork or exec time, and seems to be doing much more harm than good:
getting those contiguous pages under very heavy mempressure can be
hard (though even without it, Kaiser does generate more mempressure).

Mask out that __GFP_REPEAT inside _pgd_alloc().  Why not take it out
of the PGALLOG_GFP altogether, as v4.7 commit a3a9a59d2067 ("x86: get
rid of superfluous __GFP_REPEAT") did?  Because I think that might
make a difference to our page table memcg charging, which I'd prefer
not to interfere with at this time.

hughd adds: __alloc_pages_slowpath() in the 4.4.89-stable tree handles
__GFP_REPEAT a little differently than in prod kernel or 3.18.72-stable,
so it may not always be exactly a no-op on order:0 pages, as said above;
but I think still appropriate to omit it from Kaiser or non-Kaiser pgd.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: paranoid_entry pass cr3 need to paranoid_exit
Hugh Dickins [Wed, 27 Sep 2017 01:43:07 +0000 (18:43 -0700)]
kaiser: paranoid_entry pass cr3 need to paranoid_exit

Neel Natu points out that paranoid_entry() was wrong to assume that
an entry that did not need swapgs would not need SWITCH_KERNEL_CR3:
paranoid_entry (used for debug breakpoint, int3, double fault or MCE;
though I think it's only the MCE case that is cause for concern here)
can break in at an awkward time, between cr3 switch and swapgs, but
its handling always needs kernel gs and kernel cr3.

Easy to fix in itself, but paranoid_entry() also needs to convey to
paranoid_exit() (and my reading of macro idtentry says paranoid_entry
and paranoid_exit are always paired) how to restore the prior state.
The swapgs state is already conveyed by %ebx (0 or 1), so extend that
also to convey when SWITCH_USER_CR3 will be needed (2 or 3).

(Yes, I'd much prefer that 0 meant no swapgs, whereas it's the other
way round: and a convention shared with error_entry() and error_exit(),
which I don't want to touch.  Perhaps I should have inverted the bit
for switch cr3 too, but did not.)

paranoid_exit() would be straightforward, except for TRACE_IRQS: it
did TRACE_IRQS_IRETQ when doing swapgs, but TRACE_IRQS_IRETQ_DEBUG
when not: which is it supposed to use when SWITCH_USER_CR3 is split
apart from that?  As best as I can determine, commit 5963e317b1e9
("ftrace/x86: Do not change stacks in DEBUG when calling lockdep")
missed the swapgs case, and should have used TRACE_IRQS_IRETQ_DEBUG
there too (the discrepancy has nothing to do with the liberal use
of _NO_STACK and _UNSAFE_STACK hereabouts: TRACE_IRQS_OFF_DEBUG has
just been used in all cases); discrepancy lovingly preserved across
several paranoid_exit() cleanups, but I'm now removing it.

Neel further indicates that to use SWITCH_USER_CR3_NO_STACK there in
paranoid_exit() is now not only unnecessary but unsafe: might corrupt
syscall entry's unsafe_stack_register_backup of %rax.  Just use
SWITCH_USER_CR3: and delete SWITCH_USER_CR3_NO_STACK altogether,
before we make the mistake of using it again.

hughd adds: this commit fixes an issue in the Kaiser-without-PCIDs
part of the series, and ought to be moved earlier, if you decided
to make a release of Kaiser-without-PCIDs.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: x86_cr3_pcid_noflush and x86_cr3_pcid_user
Hugh Dickins [Sun, 27 Aug 2017 23:24:27 +0000 (16:24 -0700)]
kaiser: x86_cr3_pcid_noflush and x86_cr3_pcid_user

Mostly this commit is just unshouting X86_CR3_PCID_KERN_VAR and
X86_CR3_PCID_USER_VAR: we usually name variables in lower-case.

But why does x86_cr3_pcid_noflush need to be __aligned(PAGE_SIZE)?
Ah, it's a leftover from when kaiser_add_user_map() once complained
about mapping the same page twice.  Make it __read_mostly instead.
(I'm a little uneasy about all the unrelated data which shares its
page getting user-mapped too, but that was so before, and not a big
deal: though we call it user-mapped, it's not mapped with _PAGE_USER.)

And there is a little change around the two calls to do_nmi().
Previously they set the NOFLUSH bit (if PCID supported) when
forcing to kernel context before do_nmi(); now they also have the
NOFLUSH bit set (if PCID supported) when restoring context after:
nothing done in do_nmi() should require a TLB to be flushed here.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: PCID 0 for kernel and 128 for user
Hugh Dickins [Sat, 9 Sep 2017 02:26:30 +0000 (19:26 -0700)]
kaiser: PCID 0 for kernel and 128 for user

Why was 4 chosen for kernel PCID and 6 for user PCID?
No good reason in a backport where PCIDs are only used for Kaiser.

If we continue with those, then we shall need to add Andy Lutomirski's
4.13 commit 6c690ee1039b ("x86/mm: Split read_cr3() into read_cr3_pa()
and __read_cr3()"), which deals with the problem of read_cr3() callers
finding stray bits in the cr3 that they expected to be page-aligned;
and for hibernation, his 4.14 commit f34902c5c6c0 ("x86/hibernate/64:
Mask off CR3's PCID bits in the saved CR3").

But if 0 is used for kernel PCID, then there's no need to add in those
commits - whenever the kernel looks, it sees 0 in the lower bits; and
0 for kernel seems an obvious choice.

And I naughtily propose 128 for user PCID.  Because there's a place
in _SWITCH_TO_USER_CR3 where it takes note of the need for TLB FLUSH,
but needs to reset that to NOFLUSH for the next occasion.  Currently
it does so with a "movb $(0x80)" into the high byte of the per-cpu
quadword, but that will cause a machine without PCID support to crash.
Now, if %al just happened to have 0x80 in it at that point, on a
machine with PCID support, but 0 on a machine without PCID support...

(That will go badly wrong once the pgd can be at a physical address
above 2^56, but even with 5-level paging, physical goes up to 2^52.)

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: load_new_mm_cr3() let SWITCH_USER_CR3 flush user
Hugh Dickins [Thu, 17 Aug 2017 22:00:37 +0000 (15:00 -0700)]
kaiser: load_new_mm_cr3() let SWITCH_USER_CR3 flush user

We have many machines (Westmere, Sandybridge, Ivybridge) supporting
PCID but not INVPCID: on these load_new_mm_cr3() simply crashed.

Flushing user context inside load_new_mm_cr3() without the use of
invpcid is difficult: momentarily switch from kernel to user context
and back to do so?  I'm not sure whether that can be safely done at
all, and would risk polluting user context with kernel internals,
and kernel context with stale user externals.

Instead, follow the hint in the comment that was there: change
X86_CR3_PCID_USER_VAR to be a per-cpu variable, then load_new_mm_cr3()
can leave a note in it, for SWITCH_USER_CR3 on return to userspace to
flush user context TLB, instead of default X86_CR3_PCID_USER_NOFLUSH.

Which works well enough that there's no need to do it this way only
when invpcid is unsupported: it's a good alternative to invpcid here.
But there's a couple of inlines in asm/tlbflush.h that need to do the
same trick, so it's best to localize all this per-cpu business in
mm/kaiser.c: moving that part of the initialization from setup_pcid()
to kaiser_setup_pcid(); with kaiser_flush_tlb_on_return_to_user() the
function for noting an X86_CR3_PCID_USER_FLUSH.  And let's keep a
KAISER_SHADOW_PGD_OFFSET in there, to avoid the extra OR on exit.

I did try to make the feature tests in asm/tlbflush.h more consistent
with each other: there seem to be far too many ways of performing such
tests, and I don't have a good grasp of their differences.  At first
I converted them all to be static_cpu_has(): but that proved to be a
mistake, as the comment in __native_flush_tlb_single() hints; so then
I reversed and made them all this_cpu_has().  Probably all gratuitous
change, but that's the way it's working at present.

I am slightly bothered by the way non-per-cpu X86_CR3_PCID_KERN_VAR
gets re-initialized by each cpu (before and after these changes):
no problem when (as usual) all cpus on a machine have the same
features, but in principle incorrect.  However, my experiment
to per-cpu-ify that one did not end well...

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: enhanced by kernel and user PCIDs
Dave Hansen [Wed, 30 Aug 2017 23:23:00 +0000 (16:23 -0700)]
kaiser: enhanced by kernel and user PCIDs

Merged performance improvements to Kaiser, using distinct kernel
and user Process Context Identifiers to minimize the TLB flushing.

Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: vmstat show NR_KAISERTABLE as nr_overhead
Hugh Dickins [Sun, 10 Sep 2017 04:27:32 +0000 (21:27 -0700)]
kaiser: vmstat show NR_KAISERTABLE as nr_overhead

The kaiser update made an interesting choice, never to free any shadow
page tables.  Contention on global spinlock was worrying, particularly
with it held across page table scans when freeing.  Something had to be
done: I was going to add refcounting; but simply never to free them is
an appealing choice, minimizing contention without complicating the code
(the more a page table is found already, the less the spinlock is used).

But leaking pages in this way is also a worry: can we get away with it?
At the very least, we need a count to show how bad it actually gets:
in principle, one might end up wasting about 1/256 of memory that way
(1/512 for when direct-mapped pages have to be user-mapped, plus 1/512
for when they are user-mapped from the vmalloc area on another occasion
(but we don't have vmalloc'ed stacks, so only large ldts are vmalloc'ed).

Add per-cpu stat NR_KAISERTABLE: including 256 at startup for the
shared pgd entries, and 1 for each intermediate page table added
thereafter for user-mapping - but leave out the 1 per mm, for its
shadow pgd, because that distracts from the monotonic increase.
Shown in /proc/vmstat as nr_overhead (0 if kaiser not enabled).

In practice, it doesn't look so bad so far: more like 1/12000 after
nine hours of gtests below; and movable pageblock segregation should
tend to cluster the kaiser tables into a subset of the address space
(if not, they will be bad for compaction too).  But production may
tell a different story: keep an eye on this number, and bring back
lighter freeing if it gets out of control (maybe a shrinker).

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: delete KAISER_REAL_SWITCH option
Hugh Dickins [Mon, 4 Sep 2017 01:30:43 +0000 (18:30 -0700)]
kaiser: delete KAISER_REAL_SWITCH option

We fail to see what CONFIG_KAISER_REAL_SWITCH is for: it seems to be
left over from early development, and now just obscures tricky parts
of the code.  Delete it before adding PCIDs, or nokaiser boot option.

(Or if there is some good reason to keep the option, then it needs
a help text - and a "depends on KAISER", so that all those without
KAISER are not asked the question.)

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: name that 0x1000 KAISER_SHADOW_PGD_OFFSET
Hugh Dickins [Sun, 10 Sep 2017 00:31:18 +0000 (17:31 -0700)]
kaiser: name that 0x1000 KAISER_SHADOW_PGD_OFFSET

There's a 0x1000 in various places, which looks better with a name.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: cleanups while trying for gold link
Hugh Dickins [Tue, 22 Aug 2017 03:11:43 +0000 (20:11 -0700)]
kaiser: cleanups while trying for gold link

While trying to get our gold link to work, four cleanups:
matched the gdt_page declaration to its definition;
in fiddling unsuccessfully with PERCPU_INPUT(), lined up backslashes;
lined up the backslashes according to convention in percpu-defs.h;
deleted the unused irq_stack_pointer addition to irq_stack_union.

Sad to report that aligning backslashes does not appear to help gold
align to 8192: but while these did not help, they are worth keeping.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: kaiser_remove_mapping() move along the pgd
Hugh Dickins [Mon, 2 Oct 2017 17:57:24 +0000 (10:57 -0700)]
kaiser: kaiser_remove_mapping() move along the pgd

When removing the bogus comment from kaiser_remove_mapping(),
I really ought to have checked the extent of its bogosity: as
Neel points out, there is nothing to stop unmap_pud_range_nofree()
from continuing beyond the end of a pud (and starting in the wrong
position on the next).

Fix kaiser_remove_mapping() to constrain the extent and advance pgd
pointer correctly: use pgd_addr_end() macro as used throughout base
mm (but don't assume page-rounded start and size in this case).

But this bug was very unlikely to trigger in this backport: since
any buddy allocation is contained within a single pud extent, and
we are not using vmapped stacks (and are only mapping one page of
stack anyway): the only way to hit this bug here would be when
freeing a large modified ldt.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: tidied up kaiser_add/remove_mapping slightly
Hugh Dickins [Mon, 4 Sep 2017 02:23:08 +0000 (19:23 -0700)]
kaiser: tidied up kaiser_add/remove_mapping slightly

Yes, unmap_pud_range_nofree()'s declaration ought to be in a
header file really, but I'm not sure we want to use it anyway:
so for now just declare it inside kaiser_remove_mapping().
And there doesn't seem to be such a thing as unmap_p4d_range(),
even in a 5-level paging tree.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: tidied up asm/kaiser.h somewhat
Hugh Dickins [Mon, 4 Sep 2017 02:18:07 +0000 (19:18 -0700)]
kaiser: tidied up asm/kaiser.h somewhat

Mainly deleting a surfeit of blank lines, and reflowing header comment.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: ENOMEM if kaiser_pagetable_walk() NULL
Hugh Dickins [Mon, 4 Sep 2017 01:48:02 +0000 (18:48 -0700)]
kaiser: ENOMEM if kaiser_pagetable_walk() NULL

kaiser_add_user_map() took no notice when kaiser_pagetable_walk() failed.
And avoid its might_sleep() when atomic (though atomic at present unused).

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: fix perf crashes
Hugh Dickins [Wed, 23 Aug 2017 21:21:14 +0000 (14:21 -0700)]
kaiser: fix perf crashes

Avoid perf crashes: place debug_store in the user-mapped per-cpu area
instead of allocating, and use page allocator plus kaiser_add_mapping()
to keep the BTS and PEBS buffers user-mapped (that is, present in the
user mapping, though visible only to kernel and hardware).  The PEBS
fixup buffer does not need this treatment.

The need for a user-mapped struct debug_store showed up before doing
any conscious perf testing: in a couple of kernel paging oopses on
Westmere, implicating the debug_store offset of the per-cpu area.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: fix regs to do_nmi() ifndef CONFIG_KAISER
Hugh Dickins [Fri, 22 Sep 2017 03:39:56 +0000 (20:39 -0700)]
kaiser: fix regs to do_nmi() ifndef CONFIG_KAISER

pjt has observed that nmi's second (nmi_from_kernel) call to do_nmi()
adjusted the %rdi regs arg, rightly when CONFIG_KAISER, but wrongly
when not CONFIG_KAISER.

Although the minimal change is to add an #ifdef CONFIG_KAISER around
the addq line, that looks cluttered, and I prefer how the first call
to do_nmi() handled it: prepare args in %rdi and %rsi before getting
into the CONFIG_KAISER block, since it does not touch them at all.

And while we're here, place the "#ifdef CONFIG_KAISER" that follows
each, to enclose the "Unconditionally restore CR3" comment: matching
how the "Unconditionally use kernel CR3" comment above is enclosed.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: KAISER depends on SMP
Hugh Dickins [Wed, 13 Sep 2017 21:03:10 +0000 (14:03 -0700)]
kaiser: KAISER depends on SMP

It is absurd that KAISER should depend on SMP, but apparently nobody
has tried a UP build before: which breaks on implicit declaration of
function 'per_cpu_offset' in arch/x86/mm/kaiser.c.

Now, you would expect that to be trivially fixed up; but looking at
the System.map when that block is #ifdef'ed out of kaiser_init(),
I see that in a UP build __per_cpu_user_mapped_end is precisely at
__per_cpu_user_mapped_start, and the items carefully gathered into
that section for user-mapping on SMP, dispersed elsewhere on UP.

So, some other kind of section assignment will be needed on UP,
but implementing that is not a priority: just make KAISER depend
on SMP for now.

Also inserted a blank line before the option, tidied up the
brief Kconfig help message, and added an "If unsure, Y".

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: fix build and FIXME in alloc_ldt_struct()
Hugh Dickins [Mon, 4 Sep 2017 00:09:44 +0000 (17:09 -0700)]
kaiser: fix build and FIXME in alloc_ldt_struct()

Include linux/kaiser.h instead of asm/kaiser.h to build ldt.c without
CONFIG_KAISER.  kaiser_add_mapping() does already return an error code,
so fix the FIXME.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: stack map PAGE_SIZE at THREAD_SIZE-PAGE_SIZE
Hugh Dickins [Mon, 4 Sep 2017 01:57:03 +0000 (18:57 -0700)]
kaiser: stack map PAGE_SIZE at THREAD_SIZE-PAGE_SIZE

Kaiser only needs to map one page of the stack; and
kernel/fork.c did not build on powerpc (no __PAGE_KERNEL).
It's all cleaner if linux/kaiser.h provides kaiser_map_thread_stack()
and kaiser_unmap_thread_stack() wrappers around asm/kaiser.h's
kaiser_add_mapping() and kaiser_remove_mapping().  And use
linux/kaiser.h in init/main.c to avoid the #ifdefs there.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: do not set _PAGE_NX on pgd_none
Hugh Dickins [Tue, 5 Sep 2017 19:05:01 +0000 (12:05 -0700)]
kaiser: do not set _PAGE_NX on pgd_none

native_pgd_clear() uses native_set_pgd(), so native_set_pgd() must
avoid setting the _PAGE_NX bit on an otherwise pgd_none() entry:
usually that just generated a warning on exit, but sometimes
more mysterious and damaging failures (our production machines
could not complete booting).

The original fix to this just avoided adding _PAGE_NX to
an empty entry; but eventually more problems surfaced with kexec,
and EFI mapping expected to be a problem too.  So now instead
change native_set_pgd() to update shadow only if _PAGE_USER:

A few places (kernel/machine_kexec_64.c, platform/efi/efi_64.c for sure)
use set_pgd() to set up a temporary internal virtual address space, with
physical pages remapped at what Kaiser regards as userspace addresses:
Kaiser then assumes a shadow pgd follows, which it will try to corrupt.

This appears to be responsible for the recent kexec and kdump failures;
though it's unclear how those did not manifest as a problem before.
Ah, the shadow pgd will only be assumed to "follow" if the requested
pgd is on an even-numbered page: so I suppose it was going wrong 50%
of the time all along.

What we need is a flag to set_pgd(), to tell it we're dealing with
userspace.  Er, isn't that what the pgd's _PAGE_USER bit is saying?
Add a test for that.  But we cannot do the same for pgd_clear()
(which may be called to clear corrupted entries - set aside the
question of "corrupt in which pgd?" until later), so there just
rely on pgd_clear() not being called in the problematic cases -
with a WARN_ON_ONCE() which should fire half the time if it is.

But this is getting too big for an inline function: move it into
arch/x86/mm/kaiser.c (which then demands a boot/compressed mod);
and de-void and de-space native_get_shadow/normal_pgd() while here.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agokaiser: merged update
Dave Hansen [Wed, 30 Aug 2017 23:23:00 +0000 (16:23 -0700)]
kaiser: merged update

Merged fixes and cleanups, rebased to 4.4.89 tree (no 5-level paging).

Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoKAISER: Kernel Address Isolation
Richard Fellner [Thu, 4 May 2017 12:26:50 +0000 (14:26 +0200)]
KAISER: Kernel Address Isolation

This patch introduces our implementation of KAISER (Kernel Address Isolation to
have Side-channels Efficiently Removed), a kernel isolation technique to close
hardware side channels on kernel address information.

More information about the patch can be found on:

        https://github.com/IAIK/KAISER

From: Richard Fellner <richard.fellner@student.tugraz.at>
From: Daniel Gruss <daniel.gruss@iaik.tugraz.at>
X-Subject: [RFC, PATCH] x86_64: KAISER - do not map kernel in user mode
Date: Thu, 4 May 2017 14:26:50 +0200
Link: http://marc.info/?l=linux-kernel&m=149390087310405&w=2
Kaiser-4.10-SHA1: c4b1831d44c6144d3762ccc72f0c4e71a0c713e5

To: <linux-kernel@vger.kernel.org>
To: <kernel-hardening@lists.openwall.com>
Cc: <clementine.maurice@iaik.tugraz.at>
Cc: <moritz.lipp@iaik.tugraz.at>
Cc: Michael Schwarz <michael.schwarz@iaik.tugraz.at>
Cc: Richard Fellner <richard.fellner@student.tugraz.at>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: <kirill.shutemov@linux.intel.com>
Cc: <anders.fogh@gdata-adan.de>
After several recent works [1,2,3] KASLR on x86_64 was basically
considered dead by many researchers. We have been working on an
efficient but effective fix for this problem and found that not mapping
the kernel space when running in user mode is the solution to this
problem [4] (the corresponding paper [5] will be presented at ESSoS17).

With this RFC patch we allow anybody to configure their kernel with the
flag CONFIG_KAISER to add our defense mechanism.

If there are any questions we would love to answer them.
We also appreciate any comments!

Cheers,
Daniel (+ the KAISER team from Graz University of Technology)

[1] http://www.ieee-security.org/TC/SP2013/papers/4977a191.pdf
[2] https://www.blackhat.com/docs/us-16/materials/us-16-Fogh-Using-Undocumented-CPU-Behaviour-To-See-Into-Kernel-Mode-And-Break-KASLR-In-The-Process.pdf
[3] https://www.blackhat.com/docs/us-16/materials/us-16-Jang-Breaking-Kernel-Address-Space-Layout-Randomization-KASLR-With-Intel-TSX.pdf
[4] https://github.com/IAIK/KAISER
[5] https://gruss.cc/files/kaiser.pdf

[patch based also on
https://raw.githubusercontent.com/IAIK/KAISER/master/KAISER/0001-KAISER-Kernel-Address-Isolation.patch]

Signed-off-by: Richard Fellner <richard.fellner@student.tugraz.at>
Signed-off-by: Moritz Lipp <moritz.lipp@iaik.tugraz.at>
Signed-off-by: Daniel Gruss <daniel.gruss@iaik.tugraz.at>
Signed-off-by: Michael Schwarz <michael.schwarz@iaik.tugraz.at>
Acked-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/boot: Add early cmdline parsing for options with arguments
Tom Lendacky [Mon, 17 Jul 2017 21:10:33 +0000 (16:10 -0500)]
x86/boot: Add early cmdline parsing for options with arguments

commit e505371dd83963caae1a37ead9524e8d997341be upstream.

Add a cmdline_find_option() function to look for cmdline options that
take arguments. The argument is returned in a supplied buffer and the
argument length (regardless of whether it fits in the supplied buffer)
is returned, with -1 indicating not found.

Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andrey Ryabinin <aryabinin@virtuozzo.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Brijesh Singh <brijesh.singh@amd.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Larry Woodman <lwoodman@redhat.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matt Fleming <matt@codeblueprint.co.uk>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Toshimitsu Kani <toshi.kani@hpe.com>
Cc: kasan-dev@googlegroups.com
Cc: kvm@vger.kernel.org
Cc: linux-arch@vger.kernel.org
Cc: linux-doc@vger.kernel.org
Cc: linux-efi@vger.kernel.org
Cc: linux-mm@kvack.org
Link: http://lkml.kernel.org/r/36b5f97492a9745dce27682305f990fc20e5cf8a.1500319216.git.thomas.lendacky@amd.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agoLinux 4.4.109
Greg Kroah-Hartman [Tue, 2 Jan 2018 19:33:28 +0000 (20:33 +0100)]
Linux 4.4.109

6 years agomm/vmstat: Make NR_TLB_REMOTE_FLUSH_RECEIVED available even on UP
Andy Lutomirski [Mon, 5 Jun 2017 14:40:25 +0000 (07:40 -0700)]
mm/vmstat: Make NR_TLB_REMOTE_FLUSH_RECEIVED available even on UP

commit 5dd0b16cdaff9b94da06074d5888b03235c0bf17 upstream.

This fixes CONFIG_SMP=n, CONFIG_DEBUG_TLBFLUSH=y without introducing
further #ifdef soup.  Caught by a Kbuild bot randconfig build.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Fixes: ce4a4e565f52 ("x86/mm: Remove the UP asm/tlbflush.h code, always use the (formerly) SMP code")
Link: http://lkml.kernel.org/r/76da9a3cc4415996f2ad2c905b93414add322021.1496673616.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agon_tty: fix EXTPROC vs ICANON interaction with TIOCINQ (aka FIONREAD)
Linus Torvalds [Thu, 21 Dec 2017 01:57:06 +0000 (17:57 -0800)]
n_tty: fix EXTPROC vs ICANON interaction with TIOCINQ (aka FIONREAD)

commit 966031f340185eddd05affcf72b740549f056348 upstream.

We added support for EXTPROC back in 2010 in commit 26df6d13406d ("tty:
Add EXTPROC support for LINEMODE") and the intent was to allow it to
override some (all?) ICANON behavior.  Quoting from that original commit
message:

         There is a new bit in the termios local flag word, EXTPROC.
         When this bit is set, several aspects of the terminal driver
         are disabled.  Input line editing, character echo, and mapping
         of signals are all disabled.  This allows the telnetd to turn
         off these functions when in linemode, but still keep track of
         what state the user wants the terminal to be in.

but the problem turns out that "several aspects of the terminal driver
are disabled" is a bit ambiguous, and you can really confuse the n_tty
layer by setting EXTPROC and then causing some of the ICANON invariants
to no longer be maintained.

This fixes at least one such case (TIOCINQ) becoming unhappy because of
the confusion over whether ICANON really means ICANON when EXTPROC is set.

This basically makes TIOCINQ match the case of read: if EXTPROC is set,
we ignore ICANON.  Also, make sure to reset the ICANON state ie EXTPROC
changes, not just if ICANON changes.

Fixes: 26df6d13406d ("tty: Add EXTPROC support for LINEMODE")
Reported-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Reported-by: syzkaller <syzkaller@googlegroups.com>
Cc: Jiri Slaby <jslaby@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
6 years agox86/smpboot: Remove stale TLB flush invocations
Thomas Gleixner [Sat, 30 Dec 2017 21:13:53 +0000 (22:13 +0100)]
x86/smpboot: Remove stale TLB flush invocations

commit 322f8b8b340c824aef891342b0f5795d15e11562 upstream.

smpboot_setup_warm_reset_vector() and smpboot_restore_warm_reset_vector()
invoke local_flush_tlb() for no obvious reason.

Digging in history revealed that the original code in the 2.1 era added
those because the code manipulated a swapper_pg_dir pagetable entry. The
pagetable manipulation was removed long ago in the 2.3 timeframe, but the
TLB flush invocations stayed around forever.

Remove them along with the pointless pr_debug()s which come from the same 2.1
change.

Reported-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Linus Torvalds <torvalds@linuxfoundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20171230211829.586548655@linutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>