Apart from DVM, all firmware uses the same base API, and there's
code outside iwlmvm that needs to interact with it. Reflect this
in the source better and reorganize the firmware API to a new
fw/api/ directory.

While at it, split the already pretty large fw-api.h file into a
number of smaller files, going from almost 3k lines in there to
a maximum number of lines less than 1k.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Before TVQM, all TX queues were allocated straight at init.
With TVQM, queues are allocated on demand and hence we need
to check if a queue exists before dereferencing it.

Fixes: 66128fa08806 ("iwlwifi: move to TVQM mode")
Signed-off-by: NMordechai Goodstein <mordechay.goodstein@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In case we need to move the scheduler write pointer by
steps of 0x40, 0x80 or 0xc0, the scheduler gets stuck.
This leads to hardware error interrupts with status:
0x5A5A5A5A or alike.

In order to work around this, detect in the transport
layer that we are going to hit this case and tell iwlmvm
to increment the sequence number of the packets. This
allows to keep the requirement that the WiFi sequence
number is in sync with the index in the scheduler Tx queue
and it also allows to avoid the problematic sequence.
This means that from time to time, we will start a queue
from ssn + 1, but that shouldn't be a problem since we
don't switch to new queues for AMPDU now that we have
DQA which allows to keep the same queue while toggling
the AMPDU state.

This bug has been fixed on 9000 devices and up.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When the command name is printed on command completion, the wrong
group is used, leading to the wrong name being printed. Fix this
by using the group ID without inappropriately mangling it through
iwl_cmd_groupid() - it's already a u8. Also, while at it, use it
from the same place as the command ID, everything else is just
confusing.

Fixes: ab02165c ("iwlwifi: add wide firmware command infrastructure for TX")
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When sending non-linear SKBs that should be included in the regular
TX tracing completely (and not be pushed into the tx_data tracing),
the (tracing) code didn't correctly take the fact that they were
non-linear into account and added only the skb head portion.

This probably never really triggered, since those frames we want
traced fully are most likely linear anyway, but the code gets easier
to understand and we lose an argument to the tracing function, so
overall fixing this is better.

Fixes: 206eea78 ("iwlwifi: pcie: support frag SKBs")
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

There's no need to calculate the data_len outside of the tracepoint,
since it's always skb->len - hdr_len, which are both available inside.
Simplify the callers and move the calculation in.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

When toggling the RF-kill pin quickly in succession, the driver can
get rather confused because it might be in the process of shutting
down, expecting all commands to go through quickly due to rfkill,
but the transport already thinks the device is accessible again,
even though it previously shut it down. This leads to bugs, and I
even observed a kernel panic.

Avoid this by making the PCIe code only report that the radio is
enabled again after the higher layers actually decided to shut it
off.

This also pulls out this common RF-kill checking code into a common
function called by both transport generations and also moves it to
the direct method - in the internal helper we don't really care
about the RF-kill status anymore since we won't report it up until
the stop anyway.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Letting the preprocessor/compiler generate the shift/mask by itself
is a win for readability, so use bitfield.h for some registers.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

A common pattern with skb_put() is to just want to memcpy()
some data into the new space, introduce skb_put_data() for
this.

An spatch similar to the one for skb_put_zero() converts many
of the places using it:

    @@
    identifier p, p2;
    expression len, skb, data;
    type t, t2;
    @@
    (
    -p = skb_put(skb, len);
    +p = skb_put_data(skb, data, len);
    |
    -p = (t)skb_put(skb, len);
    +p = skb_put_data(skb, data, len);
    )
    (
    p2 = (t2)p;
    -memcpy(p2, data, len);
    |
    -memcpy(p, data, len);
    )

    @@
    type t, t2;
    identifier p, p2;
    expression skb, data;
    @@
    t *p;
    ...
    (
    -p = skb_put(skb, sizeof(t));
    +p = skb_put_data(skb, data, sizeof(t));
    |
    -p = (t *)skb_put(skb, sizeof(t));
    +p = skb_put_data(skb, data, sizeof(t));
    )
    (
    p2 = (t2)p;
    -memcpy(p2, data, sizeof(*p));
    |
    -memcpy(p, data, sizeof(*p));
    )

    @@
    expression skb, len, data;
    @@
    -memcpy(skb_put(skb, len), data, len);
    +skb_put_data(skb, data, len);

(again, manually post-processed to retain some comments)
Reviewed-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add two new device families to differentiate them from 8000.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This is essentially the same code as gen1, except that it uses
gen2 functions and SW checksum is not included.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In TVQM mode the queue ID is assigned after enablement.
Get rid of assuming pre-defined TX queue ID in functions
that will be used by TVQM allocation path.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Change queue allocation to be dynamic. On transport init only
the command queue is being allocated. Other queues are allocated
on demand.
This is due to the huge amount of queues we will soon enable (512)
and as a preparation for TX Virtual Queue Manager feature (TVQM),
where firmware will assign the actual queue number on demand.
This includes also allocation of the byte count table per queue
and not as a contiguous chunk of memory.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In a000 transport we will allocate queues dynamically.
Right now queue are allocated as one big chunk of memory
and accessed as such.
The dynamic allocation of the queues will require accessing
the queues as pointers.
In order to keep simplicity of pre-a000 tx queues handling,
keep allocating and freeing the memory in the same style,
but move to access the queues in the various functions as
individual pointers.
Dynamic allocation for the a000 devices will be in a separate
patch.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Code is basically the same, with a cleanups of old narrow host
command, ampg workarounds, some cosmetic stuff, and usage of
TFH functions when accessing TFD queues.
This enables also the cleanup of iwl_pcie_tfd_set_tb() since
now it won't be called anywhere in the a000 data path
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Cleanup code that is irrelevant for a000 devices.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This is just a copy-paste in order to make changes tracking
easier.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In a000 devices the TX handling is different in a few ways:
* Queues are allocated dynamically
* DQA is enabled by default
* Driver shouldn't access TFH registers - ucode configures it
  all in SCD_QUEUE_CFG command

Support all this in a new API with op mode, where op mode sends
the command, transport will allocate the queue dynamically, fill
in DMA properties, send the command to FW and get the ID back.
Current implementation only sets the new transport API and fills
the DMA properties.
Future patches will complete the other parts.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Context information structure is going to be used in a000
devices for firmware self init.

The self init includes firmware self loading from DRAM by
ROM.
This means the TFH relevant firmware loading can be cleaned up.

The firmware loading includes the paging memory as well, so op
mode can stop initializing the paging and sending the DRAM_BLOCK_CMD.

Firmware is doing RFH, TFH and SCD configuration, while driver
only fills the required configurations and addresses in the
context information structure.

The only remaining access to RFH is the write pointer, which
is updated upon alive interrupt after FW configured the RFH.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Use iwl_get_dma_hi_addr() instead of open-coding it.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We already have queue_used in the transport - we can
use it instead.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This reverts commit 8aacf4b7 ("iwlwifi: introduce trans API
to get byte count table").
The commit is not needed as a better approach will be taken.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

We don't need to print so much data in the kernel log.
Limit the data to be printed to the queue that actually
got stuck in case of a TFD queue hang, and stop dumping
all the CSR and FH registers. Over the course of time, the
CSR and FH values haven't proven themselves to be really
useful for debugging, and they are now in the firmware dump
anyway.

This comes as a preparation to the addition of more data
required to be printed by the firwmare team.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Instead of setting the tx_cmd length in the mvm code, which is
complicated by the fact that DQA may want to temporarily store
the SKB on the side, adjust the length in the PCIe code which
also knows about this since it's responsible for duplicating
all those headers that are account for in this code.

As the PCIe code already relies on the tx_cmd->len field, this
doesn't really introduce any new dependencies.

To make this possible we need to move the memcpy() of the TX
command until after it was updated.

This does even simplify the code though, since the PCIe code
already does a lot of manipulations to build A-MSDUs correctly
and changing the length becomes a simple operation to see how
much was added/removed, rather than predicting it.

Fixes: 24afba76 ("iwlwifi: mvm: support bss dynamic alloc/dealloc of queues")
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Since offset is zero, it's not necessary to use set function. Reset
function is straightforward, and will remove the unnecessary add
operation in set function.
Signed-off-by: NZhang Shengju <zhangshengju@cmss.chinamobile.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Emmanuel reports that when CMD_WANT_ASYNC_CALLBACK is used by mvm,
the callback will be called with the command queue lock held, and
mvm will try to stop all (other) TX queues, which acquires their
locks - this caused a false lockdep recursive locking report.

Suppress this report by marking the command queue lock with a new,
separate, lock class so lockdep can tell the difference between
the two types of queues.

Fixes: 156f92f2 ("iwlwifi: block the queues when we send ADD_STA for uAPSD")
Reported-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The various TFD/TB helpers have two code paths depending on the
type of TFD supported, with variable shadowing due to the new if
branches. Move the fall-through code into else branches to avoid
variable shadowing. While doing so, rename some of the variables
and do some other cleanups (like removing void * casts of void *
pointers.)
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Due to firmware design considerations, move to wide ID for
all commands.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The original intent was to have the general iwl_queue shared
between RX and TX queues, but it is not the actual status.
Since it is not shared with any struct but iwl_txq, it adds
unnecessary complexity. Merge those structs.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Since TFD was enlarged to 256 bytes, the fetch of the TFD
itself is very expensive.
To make DRAM to SRAM more efficient, bits 12-13 will indicate
the number of 64 byte chunks that should be transferred to
SRAM.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Previous patch introduced the new formats. This patch
allocates the new structures and adjusts code accordingly.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

In future HW the byte count table address will be configured
by ucode per queue. Add API to expose the byte count table to
the opmode
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

New hardware supports bigger TFDs and TBs.
Introduce the new formats and adjust defines and code
relying on old format.
Changing the actual TFD allocation is trickier and
deferred to the next patch.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Up till now we accessed SCD configuration only for initial
configuration and for enabling command queue.
For a000 generation the command queue is open by default
and firmware configures the rest. No driver SCD accesses
are expected. Make sure this is the case.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Centralize the logging of SCD status. The motivation is
that for a000 devices we will have new SCD HW, but this
code was duplicate anyway, so it is a proper cleanup.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

For a000 device the FH was replaced by the TFH.
This is the first patch in a series introducing the
changes stemming from this change.
This patch initializes the TFQ queue table with the new
64 bit register and the relevant TFH configuration
registers.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Currently the scratch buffer is set to 16 bytes and indicates
the size of the bi-directional DMA.
However, next HW generation will perform additional offloading,
and will write the result in the key location of the TX command,
so the size of the bi-directional consistent memory should grow
accordingly - increase it to 40.
Generalize the code to get rid of now irrelevant scratch references.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

The PCIe transport needs to store two pointers in each TX SKB, and
currently assumes mac80211's ieee80211_tx_info is present in the CB
to do that.

In order to remove that assumption, have the opmodes pass in the
offset to where the pointers can be stored in the CB and use the
offset in the PCIe code.

To make the disentanglement complete, remove mac80211.h includes
from everywhere in the generic iwlwifi code. This required adding
an include of cfg80211.h in one place.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

This fixes a pretty ancient bug that hasn't manifested itself
until now.
The scratchbuf for command queue is allocated only for 32 slots
but is accessed with the queue write pointer - which can be
up to 256.
Since the scratch buf size was 16 and there are up to 256 TFDs
we never passed a page boundary when accessing the scratch buffer,
but when attempting to increase the size of the scratch buffer a
panic was quick to follow when trying to access the address resulted
in a page boundary.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Fixes: 38c0f334 ("iwlwifi: use coherent DMA memory for command header")
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>

Support DQA queue sharing when no free queue exists for
allocation to a STA that already exists. This means that
a single queue will serve more than a single TID (although
the RA will be the same for all TIDs served).

We try to choose the lowest AC possible, to ensure the
shared queues have the lowest possible combined AC
requirements. The queue to share is chosen only from the
same RA's DATA queues as follows (in descending priority):
 1. An AC_BE queue
 2. Same AC queue
 3. Highest AC queue that is lower than new AC
 4. Any existing AC (there always is at least 1 DATA queue)

If any aggregations existed for any of the TIDs of the
shared queue - they are stopped (the FW is notified), but
no delBA is sent.
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>