Pull upstream dma backport into rpi 4.4.y

[8192cu] Fixes

The tx_status hook is supposed to be safe to call from interrupt
context, but it wouldn't ever return completion for the last transfer,
meaning you couldn't poll for DMA completion with interrupts masked.

This fixes IRQ handling for bcm2835's DSI1, which requires using the
DMA engine to write its registers due to a bug in the AXI bridge.

Signed-off-by: Eric Anholt <eric@anholt.net>

The current cyclic DMA period splitting implementation can generate
very small chunks at the end of each period. For example a 65536 byte
period will be split into a 65532 byte chunk and a 4 byte chunk on
the "lite" DMA channels.

This increases pressure on the RAM controller as the DMA controller
needs to fetch two control blocks from RAM in quick succession and
could potentially cause latency issues if the RAM is tied up by other
devices.

We can easily avoid these situations by distributing the remaining
length evenly between the last-but-one and the last chunk, making
sure that split chunks will be at least half the maximum length the
DMA controller can handle.

This patch checks if the last chunk would be less than half of
the maximum DMA length and if yes distributes the max len+4...max_len*1.5
bytes evenly between the last 2 chunks. This results in chunk sizes
between max_len/2 and max_len*0.75 bytes.

Signed-off-by: Matthias Reichl <hias@horus.com>
Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Tested-by: Clive Messer <clive.messer@digitaldreamtime.co.uk>

The code responsible for splitting periods into chunks that
can be handled by the DMA controller missed to update total_len,
the number of bytes processed in the current period, when there
are more chunks to follow.

Therefore total_len was stuck at 0 and the code didn't work at all.
This resulted in a wrong control block layout and audio issues because
the cyclic DMA callback wasn't executing on period boundaries.

Fix this by adding the missing total_len update.

Signed-off-by: Matthias Reichl <hias@horus.com>
Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Tested-by: Clive Messer <clive.messer@digitaldreamtime.co.uk>

Enable the use of dma-channel 0 when using the vc4-kms-v3d overlay.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Add interrupt-names properties to dt and apply the correct
mapping between irq and dma channels.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Dma channel0 is used by the legacy api - to avoid confilcts this
needs to get masked out in the device-tree instead of hardcoding it
in the driver.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Load driver early since at least bcm2708_fb doesn't support deferred
probing and even if it did, we don't want the video driver deferred.

Support the legacy DMA API which is needed by bcm2708_fb
(but only using the dedicated dma channel 0).

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Signed-off-by: Martin Sperl <kernel@martin.sperl.org>

Use platform_get_irq_byname to allow for correct mapping of
interrupts to dma channels.

The currently implemented device tree is unfortunately
implemented with the wrong assumption, that each dma-channel
has its own dma channel, but dma-irq 11 is handling
dma-channel 11-14 and dma-irq 12 is actually a "catch all"
interrupt.

So here we use the byname variant and require that interrupts
are explicitly named via the interrupts-name property in the
device tree.

The use of shared interrupts is also implemented.

As a side-effect this means we can now use dma channels 12, 13 and 14
in a correct manner - also testing shows that onl using
channels 11 to 14 for spi and i2s works perfectly (when playing
some video)

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Acked-by: Eric Anholt <eric@anholt.net>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Also added check for an error condition in bcm2835_dma_create_cb_chain
that showed up during development of this patch.

Tested using dmatest for all enabled channels.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Add slave_sg support to bcm2835-dma using shared allocation
code for bcm2835_desc and DMA-control blocks already used by
dma_cyclic.

Note that bcm2835_dma_callback had to get modified to support
both modes of operation (cyclic and non-cyclic).

Tested using:
* Hifiberry I2S card (using cyclic DMA)
* fb_st7735r SPI-framebuffer (using slave_sg DMA via spi-bcm2835)
playing BigBuckBunny for audio and video.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The bcm2835 dma system has 2 basic types of dma-channels:
* "normal" channels
* "light" channels

Lite channels are limited in several aspects:
* internal data-structure is 128 bit (not 256)
* does not support BCM2835_DMA_TDMODE (2D)
* DMA length register is limited to 16 bit.
  so 0-65535 (not 0-65536 as mentioned in the official datasheet)
* BCM2835_DMA_S/D_IGNORE are not supported

The detection of the type of mode is implemented by looking at
the LITE bit in the DEBUG register for each channel.
This allows automatic detection.

Based on this the maximum block size is set to (64K - 4) or to 1G
and this limit is honored during generation of control block
chains. The effect is that when a LITE channel is used more
control blocks are used to do the same transfer (compared
to a normal channel).

As there are several sources/target DREQS that are 32 bit wide
we need to have the transfer to be a multiple of 4 as this would
break the transfer otherwise.

This is why the limit of (64K - 4) was chosen over the
alternative of (64K - 4K).

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

In preparation of adding slave_sg functionality this patch moves the
generation/allocation of bcm2835_desc and the building of
the corresponding DMA-control-block chain from bcm2835_dma_prep_dma_cyclic
into the newly created method bcm2835_dma_create_cb_chain.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

In preparation to consolidating code we move the cyclic member
into the bcm_2835_desc structure.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

Add additional defines describing the DMA registers
as well as adding some more documentation to those registers.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

The original patch contained 3 dma channels that were masked out.

These - as far as research and discussions show - are a
artefacts remaining from the downstream legacy dma-api.

Right now down-stream still includes a legacy api used only
in a single (downstream only) driver (bcm2708_fb) that requires
2D DMA for speedup (DMA-channel 0).
Formerly the sd-card support driver also was using this legacy
api (DMA-channel 2), but since has been moved over to use
dmaengine directly.

The DMA-channel 3 is already masked out in the devicetree in
the default property "brcm,dma-channel-mask = <0x7f35>;"

So we can remove the whole masking of DMA channels.

Signed-off-by: Martin Sperl <kernel@martin.sperl.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Eric Anholt <eric@anholt.net>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>

This reverts commit 8a349301.

This reverts commit 6f56fff3.

This reverts commit 052c2005.

This reverts commit ec2e48fd.

[8192cu] Enables cfg80211

Fixed MCP23017 overlay description in README

Old description could have led to the misunderstanding that it is an
i2c port expander, but in fact it is a gpio expander.

See: https://github.com/raspberrypi/linux/pull/1520

commit 7d6a13f0 upstream.

When we do dquot readahead in log recovery, we do not use a verifier
as the underlying buffer may not have dquots in it. e.g. the
allocation operation hasn't yet been replayed. Hence we do not want
to fail recovery because we detect an operation to be replayed has
not been run yet. This problem was addressed for inodes in commit
d8914002

 ("xfs: inode buffers may not be valid during recovery
readahead") but the problem was not recognised to exist for dquots
and their buffers as the dquot readahead did not have a verifier.

The result of not using a verifier is that when the buffer is then
next read to replay a dquot modification, the dquot buffer verifier
will only be attached to the buffer if *readahead is not complete*.
Hence we can read the buffer, replay the dquot changes and then add
it to the delwri submission list without it having a verifier
attached to it. This then generates warnings in xfs_buf_ioapply(),
which catches and warns about this case.

Fix this and make it handle the same readahead verifier error cases
as for inode buffers by adding a new readahead verifier that has a
write operation as well as a read operation that marks the buffer as
not done if any corruption is detected.  Also make sure we don't run
readahead if the dquot buffer has been marked as cancelled by
recovery.

This will result in readahead either succeeding and the buffer
having a valid write verifier, or readahead failing and the buffer
state requiring the subsequent read to resubmit the IO with the new
verifier.  In either case, this will result in the buffer always
ending up with a valid write verifier on it.

Note: we also need to fix the inode buffer readahead error handling
to mark the buffer with EIO. Brian noticed the code I copied from
there wrong during review, so fix it at the same time. Add comments
linking the two functions that handle readahead verifier errors
together so we don't forget this behavioural link in future.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 233135b7

 upstream.

This adds a name to each buf_ops structure, so that if
a verifier fails we can print the type of verifier that
failed it.  Should be a slight debugging aid, I hope.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Cc: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 7d3aa7fe

 upstream.

We don't write back stale inodes so we should skip them in
xfs_iflush_cluster, too.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 51b07f30 upstream.

Some careless idiot(*) wrote crap code in commit 1a3e8f3d

 ("xfs:
convert inode cache lookups to use RCU locking") back in late 2010,
and so xfs_iflush_cluster checks the wrong inode for whether it is
still valid under RCU protection. Fix it to lock and check the
correct inode.

(*) Careless-idiot: Dave Chinner <dchinner@redhat.com>

Discovered-by: Brain Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit b1438f47

 upstream.

When a failure due to an inode buffer occurs, the error handling
fails to abort the inode writeback correctly. This can result in the
inode being reclaimed whilst still in the AIL, leading to
use-after-free situations as well as filesystems that cannot be
unmounted as the inode log items left in the AIL never get removed.

Fix this by ensuring fatal errors from xfs_imap_to_bp() result in
the inode flush being aborted correctly.

Reported-by: Shyam Kaushik <shyam@zadarastorage.com>
Diagnosed-by: Shyam Kaushik <shyam@zadarastorage.com>
Tested-by: Shyam Kaushik <shyam@zadarastorage.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit ad747e3b upstream.

Commit 96f859d5

 ("libxfs: pack the agfl header structure so
XFS_AGFL_SIZE is correct") allowed the freelist to use the empty
slot at the end of the freelist on 64 bit systems that was not
being used due to sizeof() rounding up the structure size.

This has caused versions of xfs_repair prior to 4.5.0 (which also
has the fix) to report this as a corruption once the filesystem has
been grown. Older kernels can also have problems (seen from a whacky
container/vm management environment) mounting filesystems grown on a
system with a newer kernel than the vm/container it is deployed on.

To avoid this problem, change the initial free list indexes not to
wrap across the end of the AGFL, hence avoiding the initialisation
of agf_fllast to the last index in the AGFL.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit d0a58e83

 upstream.

Today, a kernel which refuses to mount a filesystem read-write
due to unknown ro-compat features can still transition to read-write
via the remount path.  The old kernel is most likely none the wiser,
because it's unaware of the new feature, and isn't using it.  However,
writing to the filesystem may well corrupt metadata related to that
new feature, and moving to a newer kernel which understand the feature
will have problems.

Right now the only ro-compat feature we have is the free inode btree,
which showed up in v3.16.  It would be good to push this back to
all the active stable kernels, I think, so that if anyone is using
newer mkfs (which enables the finobt feature) with older kernel
releases, they'll be protected.

Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Bill O'Donnell <billodo@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit c87bf431 upstream.

Enabling CONFIG_GCOV_PROFILE_ALL produces us a lot of warnings like

lib/lz4/lz4hc_compress.c: In function 'lz4_compresshcctx':
lib/lz4/lz4hc_compress.c:514:1: warning: the frame size of 1504 bytes is larger than 1024 bytes [-Wframe-larger-than=]

After some investigation, I found that this behavior started with gcc-4.9,
and opened https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69702

.
A suggested workaround for it is to use the -fno-tree-loop-im
flag that turns off one of the optimization stages in gcc, so the
code runs a little slower but does not use excessive amounts
of stack.

We could make this conditional on the gcc version, but I could not
find an easy way to do this in Kbuild and the benefit would be
fairly small, given that most of the gcc version in production are
affected now.

I'm marking this for 'stable' backports because it addresses a bug
with code generation in gcc that exists in all kernel versions
with the affected gcc releases.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Peter Oberparleiter <oberpar@linux.vnet.ibm.com>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 65a9f31c upstream.

After commit 21a59991 ("scripts/package/Makefile: rpmbuild is needed
for rpm targets"), it is no longer possible to specify RPMOPTS.
For example, we can no longer able to control _topdir using the following
make command.
make RPMOPTS="--define '_topdir /home/xyz/workspace/'" binrpm-pkg

Fixes: 21a59991

 ("scripts/package/Makefile: rpmbuild is needed for rpm targets")
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Michal Marek <mmarek@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 3017cd63 upstream.

With netconsole (at least) the pr_err("...  disablingn") call can
recurse back into the dma-debug code, where it'll try to grab
free_entries_lock again.  Avoid the problem by doing the printk after
dropping the lock.

Link: http://lkml.kernel.org/r/1463678421-18683-1-git-send-email-ville.syrjala@linux.intel.com


Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.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>

commit 3a17fb32

 upstream.

Grygorii Strashko reports:

 The PM runtime will be left disabled for the device if its
 .suspend_late() callback fails and async suspend is not allowed
 for this device. In this case device will not be added in
 dpm_late_early_list and dpm_resume_early() will ignore this
 device, as result PM runtime will be disabled for it forever
 (side effect: after 8 subsequent failures for the same device
 the PM runtime will be reenabled due to disable_depth overflow).

To fix this problem, add devices to dpm_late_early_list regardless
of whether or not device_suspend_late() returns errors for them.

That will ensure failures in there to be handled consistently for
all devices regardless of their async suspend/resume status.

Reported-by: Grygorii Strashko <grygorii.strashko@ti.com>
Tested-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

commit 935244cd upstream.

Currently, in ext4_mb_init(), there's a loop like the following:

  do {
    ...
    offset += 1 << (sb->s_blocksize_bits - i);
    i++;
  } while (i <= sb->s_blocksize_bits + 1);

Note that the updated offset is used in the loop's next iteration only.

However, at the last iteration, that is at i == sb->s_blocksize_bits + 1,
the shift count becomes equal to (unsigned)-1 > 31 (c.f. C99 6.5.7(3))
and UBSAN reports

  UBSAN: Undefined behaviour in fs/ext4/mballoc.c:2621:15
  shift exponent 4294967295 is too large for 32-bit type 'int'
  [...]
  Call Trace:
   [<ffffffff818c4d25>] dump_stack+0xbc/0x117
   [<ffffffff818c4c69>] ? _atomic_dec_and_lock+0x169/0x169
   [<ffffffff819411ab>] ubsan_epilogue+0xd/0x4e
   [<ffffffff81941cac>] __ubsan_handle_shift_out_of_bounds+0x1fb/0x254
   [<ffffffff81941ab1>] ? __ubsan_handle_load_invalid_value+0x158/0x158
   [<ffffffff814b6dc1>] ? kmem_cache_alloc+0x101/0x390
   [<ffffffff816fc13b>] ? ext4_mb_init+0x13b/0xfd0
   [<ffffffff814293c7>] ? create_cache+0x57/0x1f0
   [<ffffffff8142948a>] ? create_cache+0x11a/0x1f0
   [<ffffffff821c2168>] ? mutex_lock+0x38/0x60
   [<ffffffff821c23ab>] ? mutex_unlock+0x1b/0x50
   [<ffffffff814c26ab>] ? put_online_mems+0x5b/0xc0
   [<ffffffff81429677>] ? kmem_cache_create+0x117/0x2c0
   [<ffffffff816fcc49>] ext4_mb_init+0xc49/0xfd0
   [...]

Observe that the mentioned shift exponent, 4294967295, equals (unsigned)-1.

Unless compilers start to do some fancy transformations (which at least
GCC 6.0.0 doesn't currently do), the issue is of cosmetic nature only: the
such calculated value of offset is never used again.

Silence UBSAN by introducing another variable, offset_incr, holding the
next increment to apply to offset and adjust that one by right shifting it
by one position per loop iteration.

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=114701
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=112161



Signed-off-by: Nicolai Stange <nicstange@gmail.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>