There is a mis-order in mlx4 log. Fix it.
Signed-off-by: Wengang Wang <wen.gang.wang@oracle.com>
Acked-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

The driver now exposes sufficient limits so we can
avoid having mlx4 specific work-around.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

mlx4 devices (ConnectX-2, ConnectX-3) has a limitation
where rdma read work queue entries cannot exceed 512 bytes.
A rdma_read wqe needs to fit in 512 bytes:
- wqe control segment (16 bytes)
- rdma segment (16 bytes)
- scatter elements (16 bytes each)

So max_sge_rd should be: (512 - 16 - 16) / 16 = 30.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Steve Wise <swise@opengridcomputing.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Receipt of CM MAD with other than the Send method for an attribute
other than the ClassPortInfo attribute is invalid.

CM attributes other than ClassPortInfo only use the send method.

The SRP initiator does not maintain a timeout policy for CM connect
requests relies on the CM layer to do that. The result was that
the SRP initiator hung as the connect request never completed.

A new SRP target has been observed to respond to Send CM REQ
with GetResp of CM REQ with bad status. This is non conformant
with IBA spec but exposes a vulnerability in the current MAD/CM
code which will respond to the incoming GetResp of CM REQ as if
it was a valid incoming Send of CM REQ rather than tossing
this on the floor. It also causes the MAD layer not to
retransmit the original REQ even though it has not received a REP.
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Hal Rosenstock <hal@mellanox.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Ensure that validate_ipv4_net_dev() calls rcu_read_unlock() if
fib_lookup() fails. Detected by sparse. Compile-tested only.

Fixes: "IB/cma: Validate routing of incoming requests" (commit f887f2ac).
Cc: Haggai Eran <haggaie@mellanox.com>
Cc: stable <stable@vger.kernel.org>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Haggai Eran <haggaie@mellanox.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

On 12/03/2015 01:18 AM, Christoph Hellwig wrote:
> The patch looks good to me, but while we touch this area, how about
> throwing in a few cosmetic fixes as well?

How about the patch below ? In that version of the ib_sg_to_pages() fix
these concerns have been addressed and additionally to more bugs have been fixed.

------------

[PATCH] IB core: Fix ib_sg_to_pages()

Fix the code for detecting gaps. A gap occurs not only if the
second or later scatterlist element is not aligned but also if
any scatterlist element other than the last does not end at a
page boundary.

In the code for coalescing contiguous elements, ensure that
mr->length is correct and that last_page_addr is up-to-date.

Ensure that this function returns a negative
error code instead of zero if the first set_page() call fails.

Fixes: commit 4c67e2bf ("IB/core: Introduce new fast registration API")
Reported-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>

After dma_map_sg() has been called the return value of that function
must be used as the number of elements in the scatterlist instead of
scsi_sg_count().

Fixes: commit f7f7aab1 ("IB/srp: Convert to new registration API")
Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: stable <stable@vger.kernel.org> # v4.4+
Cc: Sagi Grimberg <sagig@mellanox.com>
Cc: Sebastian Parschauer <sebastian.riemer@profitbricks.com>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Detected by sparse.

Fixes: commit 330179f2 ("IB/srp: Register the indirect data buffer descriptor")
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Cc: stable <stable@vger.kernel.org> # v4.3+
Cc: Sagi Grimberg <sagig@mellanox.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Sebastian Parschauer <sebastian.riemer@profitbricks.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Without this sg_dma_len will return 0 on architectures tha have
the dma_length field.

Fixes: commit f7f7aab1 ("IB/srp: Convert to new registration API")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

When using work request based memory registration (fast_reg)
we must reserve SQ entries for registration and invalidation
in addition to send operations. Each IO consumes 3 SQ entries
(registration, send, invalidation) so we need to allocate 3x
larger send-queue instead of 2x.
Signed-off-by: Sagi Grimberg <sagig@mellanox.com>
CC: Stable <stable@vger.kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>

If srp_connect_ch() returns a positive value then that is considered
by its caller as a connection failure but this does not result in a
scsi_host_put() call and additionally causes the srp_create_target()
function to return a positive value while it should return a negative
value. Avoid all this confusion and additionally fix a memory leak by
ensuring that srp_connect_ch() always returns a value that is <= 0.
This patch avoids that a rejected login triggers the following memory
leak:

unreferenced object 0xffff88021b24a220 (size 8):
  comm "srp_daemon", pid 56421, jiffies 4295006762 (age 4240.750s)
  hex dump (first 8 bytes):
    68 6f 73 74 35 38 00 a5                          host58..
  backtrace:
    [<ffffffff8151014a>] kmemleak_alloc+0x7a/0xc0
    [<ffffffff81165c1e>] __kmalloc_track_caller+0xfe/0x160
    [<ffffffff81260d2b>] kvasprintf+0x5b/0x90
    [<ffffffff81260e2d>] kvasprintf_const+0x8d/0xb0
    [<ffffffff81254b0c>] kobject_set_name_vargs+0x3c/0xa0
    [<ffffffff81337e3c>] dev_set_name+0x3c/0x40
    [<ffffffff81355757>] scsi_host_alloc+0x327/0x4b0
    [<ffffffffa03edc8e>] srp_create_target+0x4e/0x8a0 [ib_srp]
    [<ffffffff8133778b>] dev_attr_store+0x1b/0x20
    [<ffffffff811f27fa>] sysfs_kf_write+0x4a/0x60
    [<ffffffff811f1e8e>] kernfs_fop_write+0x14e/0x180
    [<ffffffff81176eef>] __vfs_write+0x2f/0xf0
    [<ffffffff811771e4>] vfs_write+0xa4/0x100
    [<ffffffff81177c64>] SyS_write+0x54/0xc0
    [<ffffffff8151b257>] entry_SYSCALL_64_fastpath+0x12/0x6f
Signed-off-by: Bart Van Assche <bart.vanassche@sandisk.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Cc: Sebastian Parschauer <sebastian.riemer@profitbricks.com>
Cc: stable <stable@vger.kernel.org>
Signed-off-by: Doug Ledford <dledford@redhat.com>

It was found by Saurabh Sengar that the netlink code tried to allocate
memory with GFP_KERNEL while holding a spinlock. While it is possible
to fix the issue by replacing GFP_KERNEL with GFP_ATOMIC, it is better
to get rid of the spinlock while sending the packet. However, in order
to protect against a race condition that a quick response may be received
before the request is put on the request list, we need to put the request
on the list first.
Signed-off-by: Kaike Wan <kaike.wan@intel.com>
Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reported-by: Saurabh Sengar <saurabh.truth@gmail.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

do_div is the wrong way to divide a sector_t, as it is less
efficient when sector_t is 32-bit wide. With the upcoming
do_div optimizations, the kernel starts warning about this:

drivers/infiniband/ulp/iser/iser_verbs.c:1296:4: note: in expansion of macro 'do_div'
include/asm-generic/div64.h:224:22: warning: passing argument 1 of '__div64_32' from incompatible pointer type

This changes the code to use sector_div instead, which always
produces optimal code.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Sagi Grimberg <sagig@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

The current implementation gets a spin_lock, and at any scale with
qib and hfi1 post send, the lock contention grows exponentially
with the number of QPs.

idr_find() is RCU compatibile, so read doesn't need the lock.

Change to use rcu_read_lock() and rcu_read_unlock() in
__idr_get_uobj().

kfree_rcu() is used to insure a grace period between the
idr removal and actual free.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Minor errors found via code inspection during future development.
SFF 8636 defines bit position 2 to hold the status indication of
QSFP memory paging. The mask used to test for the value was
incorrect and is fixed in this patch. Additionally, the dump
function had a mismatch between the field being printed out and
the field used to source the data which was fixed.
Reviewed-by: Mitko Haralanov <mitko.haralanov@intel.com>
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reported-by: Easwar Hariharan <easwar.hariharan@intel.com>
Signed-off-by: Easwar Hariharan <easwar.hariharan@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

Commit e622f2f4 ("IB: split struct ib_send_wr")
introduced a regression for HCAs whose user mode post
sends go through ib_uverbs_post_send().

The code didn't account for the fact that the first sge is
offset by an operation dependent length.  The allocation did,
but the pointer to the destination sge list is computed without
that knowledge.  The sge list copy_from_user() then corrupts
fields in the work request

Store the operation dependent length in a local variable and
compute the sge list copy_from_user() destination using that length.
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Doug Ledford <dledford@redhat.com>

struct qib_mr requires the mr member be the last because struct
qib_mregion contains a dynamic array at the end.  The additions
of members should have been placed before this structure as the
comment noted.

Failure to do so was causing random memory corruption.  Reproducing
this bug was easy to do by running the client and server of
ib_write_bw -s 8 -n 5 on the same node.

This BUG() was tripped in a slab debug kernel:

kernel BUG at mm/slab.c:2572!

Fixes: 38071a46 ("IB/qib: Support the new memory registration API")
Reviewed-by: Mike Marciniszyn <mike.marciniszyn@intel.com>
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

The ioctl IOC_LIBCFS_PING_TEST has not been used in ages.  The recent
nidstring changes which moved all the nidstring operations from libcfs
to the LNet layer but this ioctl code was still using an nidstring
operation that was causing a circular dependency loop between libcfs and
LNet.
Signed-off-by: James Simmons <jsimmons@infradead.org>
Signed-off-by: Oleg Drokin <green@linuxhacker.ru>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull vfs fixes from Al Viro:
 "A couple of fixes (-stable fodder) + dead code removal after the
  overlayfs fix.

  I agree that it's better to separate from the fix part to make
  backporting easier, but IMO it's not worth delaying said dead code
  removal until the next window"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  Don't reset ->total_link_count on nested calls of vfs_path_lookup()
  ovl: get rid of the dead code left from broken (and disabled) optimizations
  ovl: fix permission checking for setattr

we already zero it on outermost set_nameidata(), so initialization in
path_init() is pointless and wrong.  The same DoS exists on pre-4.2
kernels, but there a slightly different fix will be needed.

Cc: stable@vger.kernel.org # v4.2
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

[Al Viro] The bug is in being too enthusiastic about optimizing ->setattr()
away - instead of "copy verbatim with metadata" + "chmod/chown/utimes"
(with the former being always safe and the latter failing in case of
insufficient permissions) it tries to combine these two.  Note that copyup
itself will have to do ->setattr() anyway; _that_ is where the elevated
capabilities are right.  Having these two ->setattr() (one to set verbatim
copy of metadata, another to do what overlayfs ->setattr() had been asked
to do in the first place) combined is where it breaks.
Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Cc: <stable@vger.kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

Pull scheduler fixes from Thomas Gleixner:
 "This updates contains the following changes:

   - Fix a signal handling regression in the bit wait functions.

   - Avoid false positive warnings in the wakeup path.

   - Initialize the scheduler root domain properly.

   - Handle gtime calculations in proc/$PID/stat proper.

   - Add more documentation for the barriers in try_to_wake_up().

   - Fix a subtle race in try_to_wake_up() which might cause a task to
     be scheduled on two cpus

   - Compile static helper function only when it is used"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/core: Fix an SMP ordering race in try_to_wake_up() vs. schedule()
  sched/core: Better document the try_to_wake_up() barriers
  sched/cputime: Fix invalid gtime in proc
  sched/core: Clear the root_domain cpumasks in init_rootdomain()
  sched/core: Remove false-positive warning from wake_up_process()
  sched/wait: Fix signal handling in bit wait helpers
  sched/rt: Hide the push_irq_work_func() declaration

Pull x86 fixes from Thoma Gleixner:
 "Another round of fixes for x86:

   - Move the initialization of the microcode driver to late_initcall to
     make sure everything that init function needs is available.

   - Make sure that lockdep knows about interrupts being off in the
     entry code before calling into c-code.

   - Undo the cpu hotplug init delay regression.

   - Use the proper conditionals in the mpx instruction decoder.

   - Fixup restart_syscall for x32 tasks.

   - Fix the hugepage regression on PAE kernels which was introduced
     with the latest PAT changes"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/signal: Fix restart_syscall number for x32 tasks
  x86/mpx: Fix instruction decoder condition
  x86/mm: Fix regression with huge pages on PAE
  x86 smpboot: Re-enable init_udelay=0 by default on modern CPUs
  x86/entry/64: Fix irqflag tracing wrt context tracking
  x86/microcode: Initialize the driver late when facilities are up

Pull SCSI fixes from James Bottomley:
 "This is quite a bumper crop of fixes: three from Arnd correcting
  various build issues in some configurations, a lock recursion in
  qla2xxx.  Two potentially exploitable issues in hpsa and mvsas, a
  potential null deref in st, a revert of a bdi registration fix that
  turned out to cause even more problems, a set of fixes to allow people
  who only defined MPT2SAS to still work after the mpt2/mpt3sas merger
  and a couple of fixes for issues turned up by the hyper-v storvsc
  driver"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  mpt3sas: fix Kconfig dependency problem for mpt2sas back compatibility
  Revert "scsi: Fix a bdi reregistration race"
  mpt3sas: Add dummy Kconfig option for backwards compatibility
  Fix a memory leak in scsi_host_dev_release()
  block/sd: Fix device-imposed transfer length limits
  scsi_debug: fix prevent_allow+verify regressions
  MAINTAINERS: Add myself as co-maintainer of the SCSI subsystem.
  sd: Make discard granularity match logical block size when LBPRZ=1
  scsi: hpsa: select CONFIG_SCSI_SAS_ATTR
  scsi: advansys needs ISA dma api for ISA support
  scsi_sysfs: protect against double execution of __scsi_remove_device()
  st: fix potential null pointer dereference.
  scsi: report 'INQUIRY result too short' once per host
  advansys: fix big-endian builds
  qla2xxx: Fix rwlock recursion
  hpsa: logical vs bitwise AND typo
  mvsas: don't allow negative timeouts
  mpt3sas: Fix use sas_is_tlr_enabled API before enabling MPI2_SCSIIO_CONTROL_TLR_ON flag

Pull drm fixes from Dave Airlie:
 "A bunch of change across the board, the main things are some vblank
  fallout in radeon and nouveau required some work, but I think this
  should fix it all.  There is also one drm fix for an oops in vmwgfx
  with how we pass the drm master around.

  The rest is just some amdgpu, i915, imx and rockchip fixes.

  Probably more than I'd like at this point, but hopefully things settle
  down now"

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: (40 commits)
  drm/amdgpu: Fixup hw vblank counter/ts for new drm_update_vblank_count() (v3)
  drm/radeon: Fixup hw vblank counter/ts for new drm_update_vblank_count() (v2)
  drm/radeon: Retry DDC probing on DVI on failure if we got an HPD interrupt
  drm/amdgpu: add spin lock to protect freed list in vm (v2)
  drm/amdgpu: partially revert "drm/amdgpu: fix VM_CONTEXT*_PAGE_TABLE_END_ADDR" v2
  drm/amdgpu: take a BO reference for the user fence
  drm/amdgpu: take a BO reference in the display code
  drm/amdgpu: set snooped flags only on system addresses v2
  drm/nouveau: Fix pre-nv50 pageflip events (v4)
  drm: Fix an unwanted master inheritance v2
  drm/amdgpu: fix race condition in amd_sched_entity_push_job
  drm/amdgpu: add err check for pin userptr
  drm/i915: take a power domain reference while checking the HDMI live status
  drm/i915: add MISSING_CASE to a few port/aux power domain helpers
  drm/i915/ddi: fix intel_display_port_aux_power_domain() after HDMI detect
  drm/i915: Introduce a gmbus power domain
  drm/i915: Clean up AUX power domain handling
  drm/rockchip: Use CRTC vblank event interface
  drm/rockchip: Fix module autoload for OF platform driver
  drm/rockchip: vop: fix window origin calculation
  ...

Pull crypto fixes from Herbert Xu:
 "This fixes a couple of crypto drivers that were using memcmp to verify
  authentication tags.  They now use crypto_memneq instead"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: talitos - Fix timing leak in ESP ICV verification
  crypto: nx - Fix timing leak in GCM and CCM decryption

When restarting a syscall with regs->ax == -ERESTART_RESTARTBLOCK,
regs->ax is assigned to a restart_syscall number.  For x32 tasks, this
syscall number must have __X32_SYSCALL_BIT set, otherwise it will be
an x86_64 syscall number instead of a valid x32 syscall number. This
issue has been there since the introduction of x32.

Reported-by: strace/tests/restart_syscall.test
Reported-and-tested-by: Elvira Khabirova <lineprinter0@gmail.com>
Signed-off-by: Dmitry V. Levin <ldv@altlinux.org>
Cc: Elvira Khabirova <lineprinter0@gmail.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/20151130215436.GA25996@altlinux.orgSigned-off-by: Thomas Gleixner <tglx@linutronix.de>

MPX decodes instructions in order to tell which bounds register
was violated.  Part of this decoding involves looking at the "REX
prefix" which is a special instrucion prefix used to retrofit
support for new registers in to old instructions.

The X86_REX_*() macros are defined to return actual bit values:

	#define X86_REX_R(rex) ((rex) & 4)

*not* boolean values.  However, the MPX code was checking for
them like they were booleans.  This might have led to us
mis-decoding the "REX prefix" and giving false information out to
userspace about bounds violations.  X86_REX_B() actually is bit 1,
so this is really only broken for the X86_REX_X() case.

Fix the conditionals up to tolerate the non-boolean values.

Fixes: fcc7ffd6 "x86, mpx: Decode MPX instruction to get bound violation information"
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: x86@kernel.org
Cc: Dave Hansen <dave@sr71.net>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/20151201003113.D800C1E0@viggo.jf.intel.comSigned-off-by: Thomas Gleixner <tglx@linutronix.de>

A few more last minute fixes for 4.4 on top of my pull request from
earlier this week.  The big change here is a vblank regression fix due to
commit 4dfd6486 "drm: Use vblank timestamps to guesstimate how many vblanks
were missed".  Beyond that, a hotplug fix and a few VM fixes.

* 'drm-fixes-4.4' of git://people.freedesktop.org/~agd5f/linux:
  drm/amdgpu: Fixup hw vblank counter/ts for new drm_update_vblank_count() (v3)
  drm/radeon: Fixup hw vblank counter/ts for new drm_update_vblank_count() (v2)
  drm/radeon: Retry DDC probing on DVI on failure if we got an HPD interrupt
  drm/amdgpu: add spin lock to protect freed list in vm (v2)
  drm/amdgpu: partially revert "drm/amdgpu: fix VM_CONTEXT*_PAGE_TABLE_END_ADDR" v2
  drm/amdgpu: take a BO reference for the user fence
  drm/amdgpu: take a BO reference in the display code
  drm/amdgpu: set snooped flags only on system addresses v2
  drm/amdgpu: fix race condition in amd_sched_entity_push_job
  drm/amdgpu: add err check for pin userptr
  add blacklist for thinkpad T40p
  drm/amdgpu: fix VM page table reference counting
  drm/amdgpu: fix userptr flags check

Pull Ceph fix from Sage Weil:
 "This addresses a refcounting bug that leads to a use-after-free"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph-client:
  rbd: don't put snap_context twice in rbd_queue_workfn()

commit 4dfd6486 "drm: Use vblank timestamps to guesstimate how many
vblanks were missed" introduced in Linux 4.4-rc1 makes the drm core
more fragile to drivers which don't update hw vblank counters and
vblank timestamps in sync with firing of the vblank irq and
essentially at leading edge of vblank.

This exposed a problem with radeon-kms/amdgpu-kms which do not
satisfy above requirements:

The vblank irq fires a few scanlines before start of vblank, but
programmed pageflips complete at start of vblank and
vblank timestamps update at start of vblank, whereas the
hw vblank counter increments only later, at start of vsync.

This leads to problems like off by one errors for vblank counter
updates, vblank counters apparently going backwards or vblank
timestamps apparently having time going backwards. The net result
is stuttering of graphics in games, or little hangs, as well as
total failure of timing sensitive applications.

See bug #93147 for an example of the regression on Linux 4.4-rc:

https://bugs.freedesktop.org/show_bug.cgi?id=93147

This patch tries to align all above events better from the
viewpoint of the drm core / of external callers to fix the problem:

1. The apparent start of vblank is shifted a few scanlines earlier,
so the vblank irq now always happens after start of this extended
vblank interval and thereby drm_update_vblank_count() always samples
the updated vblank count and timestamp of the new vblank interval.

To achieve this, the reporting of scanout positions by
radeon_get_crtc_scanoutpos() now operates as if the vblank starts
radeon_crtc->lb_vblank_lead_lines before the real start of the hw
vblank interval. This means that the vblank timestamps which are based
on these scanout positions will now update at this earlier start of
vblank.

2. The driver->get_vblank_counter() function will bump the returned
vblank count as read from the hw by +1 if the query happens after
the shifted earlier start of the vblank, but before the real hw increment
at start of vsync, so the counter appears to increment at start of vblank
in sync with the timestamp update.

3. Calls from vblank irq-context and regular non-irq calls are now
treated identical, always simulating the shifted vblank start, to
avoid inconsistent results for queries happening from vblank irq vs.
happening from drm_vblank_enable() or vblank_disable_fn().

4. The radeon_flip_work_func will delay mmio programming a pageflip until
the start of the real vblank iff it happens to execute inside the shifted
earlier start of the vblank, so pageflips now also appear to execute at
start of the shifted vblank, in sync with vblank counter and timestamp
updates. This to avoid some races between updates of vblank count and
timestamps that are used for swap scheduling and pageflip execution which
could cause pageflips to execute before the scheduled target vblank.

The lb_vblank_lead_lines "fudge" value is calculated as the size of
the display controllers line buffer in scanlines for the given video
mode: Vblank irq's are triggered by the line buffer logic when the line
buffer refill for a video frame ends, ie. when the line buffer source read
position enters the hw vblank. This means that a vblank irq could fire at
most as many scanlines before the current reported scanout position of the
crtc timing generator as the number of scanlines the line buffer can
maximally hold for a given video mode.

This patch has been successfully tested on a RV730 card with DCE-3 display
engine and on a evergreen card with DCE-4 display engine, in single-display
and dual-display configuration, with different video modes.

A similar patch is needed for amdgpu-kms to fix the same problem.

Limitations:

- Maybe replace the udelay() in the flip_work_func() by a suitable
  usleep_range() for a bit better efficiency? Will try that.

- Line buffer sizes in pixels are hard-coded on < DCE-4 to a value
  i just guessed to be high enough to work ok, lacking info on the true
  sizes atm.

Probably fixes: fdo#93147

Port of Mario's radeon fix to amdgpu.
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
(v1) Reviewed-by: Mario Kleiner <mario.kleiner.de@gmail.com>

(v2) Refine amdgpu_flip_work_func() for better efficiency.

     In amdgpu_flip_work_func, replace the busy waiting udelay(5)
     with event lock held by a more performance and energy efficient
     usleep_range() until at least predicted true start of hw vblank,
     with some slack for scheduler happiness. Release the event lock
     during waits to not delay other outputs in doing their stuff, as
     the waiting can last up to 200 usecs in some cases.

     Also small fix to code comment and formatting in that function.

(v2) Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>

(v3) Fix crash in crtc disabled case

Pull libnvdimm fixes from Dan Williams:

 - NFIT parsing regression fixes from Linda.  The nvdimm hot-add
   implementation merged in 4.4-rc1 interpreted the specification in a
   way that breaks actual HPE platforms.  We are also closing the loop
   with the ACPI Working Group to get this clarification added to the
   spec.

 - Andy pointed out that his laptop without nvdimm resources is loading
   the e820-nvdimm module by default, fix that up to only load the
   module when an e820-type-12 range is present.

* 'libnvdimm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm:
  nfit: Adjust for different _FIT and NFIT headers
  nfit: Fix the check for a successful NFIT merge
  nfit: Account for table size length variation
  libnvdimm, e820: skip module loading when no type-12

Pull ARM KVM fixes from Paolo Bonzini:

 - a series of fixes to deal with the aliasing between the sp and xzr
   register

 - a fix for the cache flush fix that went in -rc3

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  ARM/arm64: KVM: correct PTE uncachedness check
  arm64: KVM: Get rid of old vcpu_reg()
  arm64: KVM: Correctly handle zero register in system register accesses
  arm64: KVM: Remove const from struct sys_reg_params
  arm64: KVM: Correctly handle zero register during MMIO

commit 4dfd6486 "drm: Use vblank timestamps to guesstimate how many
vblanks were missed" introduced in Linux 4.4-rc1 makes the drm core
more fragile to drivers which don't update hw vblank counters and
vblank timestamps in sync with firing of the vblank irq and
essentially at leading edge of vblank.

This exposed a problem with radeon-kms/amdgpu-kms which do not
satisfy above requirements:

The vblank irq fires a few scanlines before start of vblank, but
programmed pageflips complete at start of vblank and
vblank timestamps update at start of vblank, whereas the
hw vblank counter increments only later, at start of vsync.

This leads to problems like off by one errors for vblank counter
updates, vblank counters apparently going backwards or vblank
timestamps apparently having time going backwards. The net result
is stuttering of graphics in games, or little hangs, as well as
total failure of timing sensitive applications.

See bug #93147 for an example of the regression on Linux 4.4-rc:

https://bugs.freedesktop.org/show_bug.cgi?id=93147

This patch tries to align all above events better from the
viewpoint of the drm core / of external callers to fix the problem:

1. The apparent start of vblank is shifted a few scanlines earlier,
so the vblank irq now always happens after start of this extended
vblank interval and thereby drm_update_vblank_count() always samples
the updated vblank count and timestamp of the new vblank interval.

To achieve this, the reporting of scanout positions by
radeon_get_crtc_scanoutpos() now operates as if the vblank starts
radeon_crtc->lb_vblank_lead_lines before the real start of the hw
vblank interval. This means that the vblank timestamps which are based
on these scanout positions will now update at this earlier start of
vblank.

2. The driver->get_vblank_counter() function will bump the returned
vblank count as read from the hw by +1 if the query happens after
the shifted earlier start of the vblank, but before the real hw increment
at start of vsync, so the counter appears to increment at start of vblank
in sync with the timestamp update.

3. Calls from vblank irq-context and regular non-irq calls are now
treated identical, always simulating the shifted vblank start, to
avoid inconsistent results for queries happening from vblank irq vs.
happening from drm_vblank_enable() or vblank_disable_fn().

4. The radeon_flip_work_func will delay mmio programming a pageflip until
the start of the real vblank iff it happens to execute inside the shifted
earlier start of the vblank, so pageflips now also appear to execute at
start of the shifted vblank, in sync with vblank counter and timestamp
updates. This to avoid some races between updates of vblank count and
timestamps that are used for swap scheduling and pageflip execution which
could cause pageflips to execute before the scheduled target vblank.

The lb_vblank_lead_lines "fudge" value is calculated as the size of
the display controllers line buffer in scanlines for the given video
mode: Vblank irq's are triggered by the line buffer logic when the line
buffer refill for a video frame ends, ie. when the line buffer source read
position enters the hw vblank. This means that a vblank irq could fire at
most as many scanlines before the current reported scanout position of the
crtc timing generator as the number of scanlines the line buffer can
maximally hold for a given video mode.

This patch has been successfully tested on a RV730 card with DCE-3 display
engine and on a evergreen card with DCE-4 display engine, in single-display
and dual-display configuration, with different video modes.

A similar patch is needed for amdgpu-kms to fix the same problem.

Limitations:

- Line buffer sizes in pixels are hard-coded on < DCE-4 to a value
  i just guessed to be high enough to work ok, lacking info on the true
  sizes atm.

Fixes: fdo#93147
Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Cc: Alex Deucher <alexander.deucher@amd.com>
Cc: Michel Dänzer <michel.daenzer@amd.com>
Cc: Harry Wentland <Harry.Wentland@amd.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>

(v1) Tested-by: Dave Witbrodt <dawitbro@sbcglobal.net>

(v2) Refine radeon_flip_work_func() for better efficiency:

     In radeon_flip_work_func, replace the busy waiting udelay(5)
     with event lock held by a more performance and energy efficient
     usleep_range() until at least predicted true start of hw vblank,
     with some slack for scheduler happiness. Release the event lock
     during waits to not delay other outputs in doing their stuff, as
     the waiting can last up to 200 usecs in some cases.

     Retested on DCE-3 and DCE-4 to verify it still works nicely.

(v2) Signed-off-by: Mario Kleiner <mario.kleiner.de@gmail.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

HPD signals on DVI ports can be fired off before the pins required for
DDC probing actually make contact, due to the pins for HPD making
contact first. This results in a HPD signal being asserted but DDC
probing failing, resulting in hotplugging occasionally failing.

This is somewhat rare on most cards (depending on what angle you plug
the DVI connector in), but on some cards it happens constantly. The
Radeon R5 on the machine used for testing this patch for instance, runs
into this issue just about every time I try to hotplug a DVI monitor and
as a result hotplugging almost never works.

Rescheduling the hotplug work for a second when we run into an HPD
signal with a failing DDC probe usually gives enough time for the rest
of the connector's pins to make contact, and fixes this issue.
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Lyude <cpaul@redhat.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

there is a protection fault about freed list when OCL test.
add a spin lock to protect it.

v2: drop changes in vm_fini
Signed-off-by: JimQu <jim.qu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

The gtt_end is already inclusive, we don't need to subtract one here.

v2 (chk): keep the fix for the VM code, cause here it really applies.
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Anatoli Antonovitch <anatoli.antonovitch@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

No need for a GEM reference here.
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>