All the code has now been adjusted such that we can do away with the
old gbuf structure.

Three unused references remained in "greybus.h", so those are deleted.

Other than that most of the changes were done by simple global
substitution.  The gb_message structure incorporates the fields that
were previously found its embedded gbuf structure.  A few names have
been changed in the process:
    gbuf->transfer_buffer       message->buffer
    gbuf->transfer_buffer_size  message->buffer_size
    gbuf->hcd_data;             message->cookie
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Rework gb_connection_operation_recv() to be more oriented toward an
operation message, and to no longer use a struct gbuf local variable.
Rename it to be a little more wieldy.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Rework greybus_kill_gbuf() to be oriented toward an operation
message rather than a gbuf, and rename it.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Rework greybus_submit_gbuf() to be oriented toward an operation
message rather than a gbuf, and rename it accordingly.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Rework gb_opreation_message_init() so it doesn't use a struct gbuf
local variable.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Change the method that sends messages so that it sends "raw" buffers
rather than gbufs.  To do this, we supply the host device and
destination CPort when sending.  As with other recent patches,
change the name of the method to reflect that we're no longer
dealing with gbufs.

The interface has changed as well.  Now this routine will return a
"cookie" value.  The cookie is used to represent the outgoing
request, and is supplied by the caller if necessary to cancel a
previously-sent buffer.  We'll store the result in gbuf->hcd_data
for now (which produces the same result as before...).
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This changes the context value stashed with each USB URB so that it
is always the host device pointer.

In cport_out_callback() this allows us to get away with *not*
requiring the gbuf for handling completions any more.  We are
(currently) ignoring the gbuf status value returned anyway, so
we'll skip setting it altogether.

Greg's comments in cport_out_callback() point out that ignoring
this was misguided, and handling send errors will be put in
place in an upcoming patch.

The context is set to the host device pointer for SVC receive and
CPort receive URBs for consistency--because we can.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Change the interface for canceling in-flight buffers to take a magic
cookie value as argument rather than a gbuf.  Right now we pass the
gbuf->hcd_data pointer that's assumed to have been set by the submit
routine.  But the next patch will change the submit routine to
return the cookie to be used, and the caller will be responsible for
keeping track of it.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Switch the host driver free routine to take a pointer to the
previously-allocated buffer that should be freed.

Rename the method to reflect it's no longer tied to a gbuf.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This begins the transition to buffer allocation that does not rely
on the gbuf construct.

The host driver allocation routine will return a pointer to the
buffer to be used, and the caller will be responsible for keeping
track of that pointer, as well as the requested buffer size.

Rename the allocation method to reflect it's no longer tied to a
gbuf.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

For ES1 we need to insert the destination CPort id before the data
to be sent over UniPro.  Currently this is done at the time the
buffer is created, but there's no need to do so until we're actually
going to send the content of the buffer.

Move the setting of that destination information into submit_gbuf().
Note that this allows us to defer initializing a few other gbuf
fields until after we know the buffer allocation has succeeded.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

For ES1 we need to insert the destination CPort id in whatever we
supply for sending over UniPro.  Currently we allocate one extra
byte supply the caller with an address that's offset by one from
the beginning of the allocated space.

As a result we always return a poorly-aligned buffer pointer.

Instead, allocate enough space so that we can return a better
aligned buffer to the caller.

Notes:
    - It may be that it's more important to supply an aligned
      address to the hardware.
    - We probably need to be more careful about writing into
      these buffers at unaligned offsets anyway.  (E.g., writing
      a 2-byte value at an odd offset can't be assumed to work.)
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Treat communication buffers for both inbound and outbound data the
same way, prepending a "destination cport id" byte before the data
in the buffer.  Currently this is done only for outbound data
buffers.

This isn't needed for inbound data, but handling it this way
allows the free routine to work without knowing whether the
buffer was used for sending or receiving.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

We allocate message buffers with GFP_KERNEL allocation flags if
possible.  However when an incoming request message is received we
can be in interrupt context, so we must use GFP_ATOMIC in that case.

The computation of gfp_flags in gb_operation_message_init() is
wrong.  It is needlessly using GFP_ATOMIC when allocating outbound
response buffers.  Fix the flawed logic.

Change the name of "data_out" to be "outbound" to be consistent with
usage elsewhere.  (Data/messages are "inbound" or "outbound";
requests are "incoming" or "outgoing".)
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Only three functions remain in "gbuf.c".  Move one of them into
"connection.c" and the other two into "operation.c".

Some more cleanup is coming that will further straighten out gbufs
but for now there's no sense in drawing this out any longer.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Nobody dynamically allocates gbufs any more, so we can get rid of
the allocation and free routines, as as the slab cache and its
related code.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Embed the gbuf structures for operation messages into the message
structure rather than pointing to a dynamically allocated one.

Use a null gbuf->transfer_buffer pointer rather than a null gbuf
pointer to indicate an unused gbuf.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Make sure gbuf->transfer_buffer gets reset to NULL when the buffer
is freed.  We can leverage that to do a little extra error checking.
We'll also use a null transfer buffer in the next patch to indicate
an unused gbuf.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Change greybus_alloc_gbuf() so all it does is allocate the gbuf data
structure.  Move all of the initialization of the gbuf structure in
the caller.  Do the inverse in the caller prior to freeing the gbuf
structure via greybus_free_gbuf().  Use a null gbuf->transfer_buffer
pointer rather than a null gbuf pointer to indicate an unused gbuf.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This converts some of the operation code to start leveraging the
new gb_message type.  Instead of creating the request and response
gbufs, we initialize (and tear down with a new function) the
request and response message structures.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

A Greybus buffer (gbuf) is a generic buffer used for data transfer
over a Greybus interconnect.  We only ever use gbufs in operations,
which always involve exactly two of them.  The lifetime of a gbuf is
therefore directly connected to the lifetime of an operation, so
there no real need to manage gbufs separate from operations.

This patch begins the process of removing the gbuf abstraction, on
favor of a new data type, gb_message.  The purpose of a gb_message
is--like a gbuf--to represent data to be transferred over Greybus.
However a gb_message is oriented toward the more restrictive way
we do Greybus transfers--as operation messages (either a request or
a response).

This patch simply defines the structure in its initial form, and
defines the request and response fields in a Greybus operation
structure as embedded instances of that type.  The gbuf pointer
is defined within the gb_message structure, and as a result lots
of code needs to be tweaked to reference the request and response
gbufs as subfields of the request and response structures.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

We no longer need struct gbuf defined in "greybus.h".  An upcoming
patch will embed a gbuf struct (not a pointer) into the operation
structure, and to do that we'll need the struct defined prior to the
operation.  Just move the gbuf definition into "operation.h".
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Since there is only ever one reference to a gbuf, we don't need a
kref to figure out when it can be freed.  Get rid of the kref and
its supporting code.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

These functions are never used, so we can get rid of them.
Since there's no reference-getting function any more, we no
longer need "gbuf_mutex" to avoid racing gets and puts, so
get rid of that too.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The protocol values had gotten out of sync with the Greybus Protocol
specification document, so bring them back into sync by changing a few
values, and adding the missing values.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The only time we get a cport id is when setting up a new connection.
We already have a (coarser-grained) spin lock that's used to protect
the connection lists, and we can use that same lock for protecting
the hd's connection id map.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

First of all, there's a bug in _gb_hd_connection_insert, which
Viresh found.  But pointing out that problem just called attention
to the fact that I have planning to to remove the affected block of
code.

The set of connections associated with a host device is currently
maintained in a red-black tree.  The number of connections we're
likely to have is on the order of a hundred, and at least for now
isn't even going to approach that.  When this code first went in,
Greg asserted that using a list is speedier than a red-black tree
for smallish numbers of elements (maybe up to a few hundred?).

So this patch just removes the host device's red-black tree of
connections, using a simple list instead.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

The only thing we now use the gbuf->operation pointer for is
to get access to its connection's host device.

Record the host device pointer directly in the gbuf, rather
than keeping a pointer to the operation.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

If the buffer allocated in the ES1 alloc_gbuf_data() routine is for
outbound data, we are getting the destination CPort id from the
connection.  Switch to using the copy of the destination cport id
we now have in the gbuf instead.

Check for a valid CPort id there only if we're inserting it into
the buffer.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Rather than indicating whether a gbuf is intended for outbound data,
record its destination CPort id.  That's what's really needed by
the ES1 host driver.  Use CPORT_ID_BAD when the buffer is intended
for inbound data.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Add a reference counter to the operations structure.  We'll
need this when operations are actually allowed to complete
asynchronously.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This patch does some cleanup of gb_connection_operation_recv().
    - Improve the header comments
    - Verify message is big enough for header before interpreting
      beginning of the message as a header
    - Verify at buffer creation time rather than receive time that
      no operation buffer is bigger than the maximum allowed.  We
      can then compare the incoming data size against the buffer.
    - When a response message arrives, record its status in the
      operation result, not in the buffer status.
    - Record a buffer overflow as an operation error.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

It's possible this function was destined to do something important,
but at this point it's pretty pointless.  Get rid of it.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This gets rid of a block of unnecessary forward declarations in
"greybus.h".
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Move the cancel_delayed_work() call so it's done separate from the
removing the operation from the pending list.

This should have been part of this commit:
    d3809f7 greybus: move timeout out of gb_operation_insert()
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Flush out the Greybus UART driver to actually implement greybus
requests.  The number of Greybus Protocol operations has been reduced
down to a managable number, and, if you look closely, you will notice it
follows the CDC ACM USB specification, which can drive UART devices
quite well, no need for complex UART state changes, leave all of that
logic up to the firmware, if it wants/needs it.

The Greybus Protocol spec has been updated to match the driver.

TODO: There are 2 requests from the device to the host that need to be
implemented.  As this isn't fully hooked up in the Greybus core, that is
not implemented here yet either.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

Implement a skeleton for the uevent framework, to be filled in later
when we figure out what type of module "matching" we want to do for
things (connections, interfaces, modules, etc.)

Based on a patch from Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This reverts commit 4d1529e6687d53878b71cdcd646e28e10d62c2e8.

Alex reports that this causes problems.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>

This reverts commit b8ba855506906de71df5b12b50cdbbf7259a930c.

needed to revert an older change.
Signed-off-by: Greg Kroah-Hartman <greg@kroah.com>