Commit Graph

441 Commits

Author SHA1 Message Date
Alex Elder
9654d8c462 net: ipa: determine replenish doorbell differently
Rather than tracking the number of receive buffer transactions that
have been submitted without a doorbell, just track the total number
of transactions that have been issued.  Then ring the doorbell when
that number modulo the replenish batch size is 0.

The effect is roughly the same, but the new count is slightly more
interesting, and this approach will someday allow the replenish
batch size to be tuned at runtime.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
5d6ac24fb1 net: ipa: replenish after delivering payload
Replenishing is now solely driven by whether transactions are
available for a channel, and it doesn't really matter whether
we replenish before or after we deliver received packets to the
network stack.

Replenishing before delivering the payload adds a little latency.
Eliminate that by requesting a replenish after the payload is
delivered.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
09b337deda net: ipa: kill replenish_backlog
We no longer use the replenish_backlog atomic variable to decide
when we've got work to do providing receive buffers to hardware.
Basically, we try to keep the hardware as full as possible, all the
time.  We keep supplying buffers until the hardware has no more
space for them.

As a result, we can get rid of the replenish_backlog field and the
atomic operations performed on it.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
5fc7f9ba2e net: ipa: introduce gsi_channel_trans_idle()
Create a new function that returns true if all transactions for a
channel are available for use.

Use it in ipa_endpoint_replenish_enable() to see whether to start
replenishing, and in ipa_endpoint_replenish() to determine whether
it's necessary after a failure to schedule delayed work to ensure a
future replenish attempt occurs.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
d0ac30e74e net: ipa: don't use replenish_backlog
Rather than determining when to stop replenishing using the
replenish backlog, just stop when we have exhausted all available
transactions.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
6a606b9015 net: ipa: allocate transaction in replenish loop
When replenishing, have ipa_endpoint_replenish() allocate a
transaction, and pass that to ipa_endpoint_replenish_one() to fill.
Then, if that produces no error, commit the transaction within the
replenish loop as well.  In this way we can distinguish between
transaction failures and buffer allocation/mapping failures.

Failure to allocate a transaction simply means the hardware already
has as many receive buffers as it can hold.  In that case we can
break out of the replenish loop because there's nothing more to do.

If we fail to allocate or map pages for the receive buffer, just
try again later.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
b9dbabc5ca net: ipa: decide on doorbell in replenish loop
Decide whether the doorbell should be signaled when committing a
replenish transaction in the main replenish loop, rather than in
ipa_endpoint_replenish_one().  This is a step to facilitate the
next patch.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:08 +00:00
Alex Elder
4b22d84195 net: ipa: increment backlog in replenish caller
Three spots call ipa_endpoint_replenish(), and just one of those
requests that the backlog be incremented after completing the
replenish operation.

Instead, have the caller increment the backlog, and get rid of the
add_one argument to ipa_endpoint_replenish().

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:07 +00:00
Alex Elder
b4061c136b net: ipa: allocate transaction before pages when replenishing
A transaction failure only occurs if no more transactions are
available for an endpoint.  It's a very cheap test.

When replenishing an RX endpoint buffer, there's no point in
allocating pages if transactions are exhausted.  So don't bother
doing so unless the transaction allocation succeeds.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:07 +00:00
Alex Elder
a9bec7ae70 net: ipa: kill replenish_saved
The replenish_saved field keeps track of the number of times a new
buffer is added to the backlog when replenishing is disabled.  We
don't really use it though, so there's no need for us to track it
separately.  Whether replenishing is enabled or not, we can simply
increment the backlog.

Get rid of replenish_saved, and initialize and increment the backlog
where it would have otherwise been used.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-02-04 10:16:07 +00:00
Jakub Kicinski
c59400a68c Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
No conflicts.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-03 17:36:16 -08:00
Alex Elder
34a081761e net: ipa: request IPA register values be retained
In some cases, the IPA hardware needs to request the always-on
subsystem (AOSS) to coordinate with the IPA microcontroller to
retain IPA register values at power collapse.  This is done by
issuing a QMP request to the AOSS microcontroller.  A similar
request ondoes that request.

We must get and hold the "QMP" handle early, because we might get
back EPROBE_DEFER for that.  But the actual request should be sent
while we know the IPA clock is active, and when we know the
microcontroller is operational.

Fixes: 1aac309d32 ("net: ipa: use autosuspend")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-03 08:03:43 -08:00
Alex Elder
33230aeb2e net: ipa: set IPA v4.11 AP<-modem RX buffer size to 32KB
Increase the receive buffer size used for data received from the
modem to 32KB, to improve download performance by allowing much
greater aggregation.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-02 21:13:45 -08:00
Alex Elder
ed23f02680 net: ipa: define per-endpoint receive buffer size
Allow RX endpoints to have differing receive buffer sizes.  Define
the receive buffer size in the configuration data, and use that
rather than IPA_RX_BUFFER_SIZE when configuring the endpoint.

Add verification in ipa_endpoint_data_valid_one() that the receive
buffer specified for AP RX endpoints is both big enough to handle at
least one full packet, and not so big in an aggregating endpoint
that its size can't be represented when programming the hardware.
Move aggr_byte_limit_max() up in "ipa_endpoint.c" so it can be used
earlier in the file without a forward-reference.

Initially we'll just keep the 8KB receive buffer size already in use
for all AP RX endpoints..

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2022-02-02 21:13:45 -08:00
Alex Elder
998c0bd2b3 net: ipa: prevent concurrent replenish
We have seen cases where an endpoint RX completion interrupt arrives
while replenishing for the endpoint is underway.  This causes another
instance of replenishing to begin as part of completing the receive
transaction.  If this occurs it can lead to transaction corruption.

Use a new flag to ensure only one replenish instance for an endpoint
executes at a time.

Fixes: 84f9bd12d4 ("soc: qcom: ipa: IPA endpoints")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-01-12 14:39:53 +00:00
Alex Elder
c1aaa01dbf net: ipa: use a bitmap for endpoint replenish_enabled
Define a new replenish_flags bitmap to contain Boolean flags
associated with an endpoint's replenishing state.  Replace the
replenish_enabled field with a flag in that bitmap.  This is to
prepare for the next patch, which adds another flag.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-01-12 14:39:53 +00:00
Alex Elder
6c0e3b5ce9 net: ipa: fix atomic update in ipa_endpoint_replenish()
In ipa_endpoint_replenish(), if an error occurs when attempting to
replenish a receive buffer, we just quit and try again later.  In
that case we increment the backlog count to reflect that the attempt
was unsuccessful.  Then, if the add_one flag was true we increment
the backlog again.

This second increment is not included in the backlog local variable
though, and its value determines whether delayed work should be
scheduled.  This is a bug.

Fix this by determining whether 1 or 2 should be added to the
backlog before adding it in a atomic_add_return() call.

Reviewed-by: Matthias Kaehlcke <mka@chromium.org>
Fixes: 84f9bd12d4 ("soc: qcom: ipa: IPA endpoints")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2022-01-12 14:39:53 +00:00
Alex Elder
97884b0712 net: ipa: fix IPA v4.5 interconnect data
Update the definition of the IPA interconnects for IPA v4.5 so
the path between IPA and system memory is represented by a single
"memory" interconnect.

Tested-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-12-13 12:49:26 +00:00
Jakub Kicinski
93d5404e89 Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
drivers/net/ipa/ipa_main.c
  8afc7e471a ("net: ipa: separate disabling setup from modem stop")
  76b5fbcd6b ("net: ipa: kill ipa_modem_init()")

Duplicated include, drop one.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-26 13:45:19 -08:00
Alex Elder
fe68c43ce3 net: ipa: support enhanced channel flow control
IPA v4.2 introduced GSI channel flow control, used instead of IPA
endpoint DELAY mode to prevent a TX channel from injecting packets
into the IPA core.  It used a new FLOW_CONTROLLED channel state
which could be entered using GSI generic commands.

IPA v4.11 extended the channel flow control model.  Rather than
having a distinct FLOW_CONTROLLED channel state, each channel has a
"flow control" property that can be enabled or not--independent of
the channel state.  The AP (or modem) can modify this property using
the same GSI generic commands as before.

The AP only uses channel flow control on modem TX channels, and only
when recovering from a modem crash.  The AP has no way to discover
the state of a modem channel, so the fact that (starting with IPA
v4.11) flow control no longer uses a distinct channel state is
invisible to the AP.  So enhanced flow control generally does not
change the way AP uses flow control.

There are a few small differences, however:
  - There is a notion of "primary" or "secondary" flow control, and
    when enabling or disabling flow control that must be specified
    in a new field in the GSI generic command register.  For now, we
    always specify 0 (meaning "primary").
  - When disabling flow control, it's possible a request will need
    to be retried.  We retry up to 5 times in this case.
  - Another new generic command allows the current flow control
    state to be queried.  We do not use this.

Other than the need for retries, the code essentially works the same
way as before.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 20:04:52 -08:00
Alex Elder
4c9d631adb net: ipa: introduce channel flow control
One quirk for certain versions of IPA is that endpoint DELAY mode
does not work properly.  IPA DELAY mode prevents any packets from
being delivered to the IPA core for processing on a TX endpoint.
The AP uses DELAY mode when the modem crashes, to prevent modem TX
endpoints from generating traffic during crash recovery.  Without
this, there is a chance the hardware will stall during recovery from
a modem crash.

To achieve a similar effect, a GSI FLOW_CONTROLLED channel state
was created.  A STARTED TX channel can be placed in FLOW_CONTROLLED
state, which prevents the transfer of any more packets.  A channel
in FLOW_CONTROLLED state can be either returned to STARTED state, or
can be transitioned to STOPPED state.

Because this operates on GSI channels, two generic commands were
added to allow the AP to control this state for modem channels
(similar to the ALLOCATE and HALT channel commands).

Previously the code assumed this quirk only applied to IPA v4.2.
In fact, channel flow control (rather than endpoint DELAY mode)
should be used for all versions *starting* with IPA v4.2.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 20:03:20 -08:00
Alex Elder
faa88ecead net: ipa: rearrange GSI structure fields
The dummy net_device is a large field in the GSI structure, but it
is not at all interesting from the perspective of debugging.  Move
it to the end of the GSI structure so the other fields are easier to
find in memory.

The channel and event ring arrays are also very large, so move them
near the end of the structure as well.

Swap the position of the result and completion fields to improve
structure packing.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:34 -08:00
Alex Elder
7ece9eaa3f net: ipa: GSI only needs one completion
A mutex ensures we never submit more than one GSI command of any
kind at once.  This means the per-channel and per-event ring
completion structures provide no benefit.  Instead, just use the
single (existing) GSI completion to signal the completion of GSI
commands of all types.

This makes gsi_evt_ring_init() a trivial function with no inverse,
so open-code it in its sole caller and get rid of the function.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:34 -08:00
Alex Elder
1b65bbcc9a net: ipa: skip SKB copy if no netdev
In ipa_endpoint_skb_copy(), a new socket buffer structure is
allocated so that some data can be copied into it.  However, after
doing this, if the endpoint has a null netdev pointer, we just drop
free the socket buffer.

Instead, check endpoint->netdev pointer first, and just return early
if it's null.  Also return early if the SKB allocation fails, to
avoid the deeper indentation in the normal path.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:34 -08:00
Alex Elder
01c36637ae net: ipa: explicitly disable HOLB drop during setup
During setup, ipa_endpoint_program() programs each endpoint with
various configuration parameters.  One of those registers defines
whether to drop packets when a head-of-line blocking condition is
detected on an RX endpoint.  We currently assume this is disabled;
instead, explicitly set it to be disabled.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:34 -08:00
Alex Elder
e6aab6b9b6 net: ipa: rework how HOL_BLOCK handling is specified
The head-of-line block (HOLB) drop timer is only meaningful when
dropping packets due to blocking is enabled.  Given that, redefine
the interface so the timer is specified when enabling HOLB drop, and
use a different function when disabling.

To enable and disable HOLB drop, these functions will now be used:
  ipa_endpoint_init_hol_block_enable(endpoint, milliseconds)
  ipa_endpoint_init_hol_block_disable(endpoint)

The existing ipa_endpoint_init_hol_block_enable() becomes a helper
function, renamed ipa_endpoint_init_hol_block_en(), and used with
ipa_endpoint_init_hol_block_timer() to enable HOLB block on an
endpoint.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:34 -08:00
Alex Elder
dc901505fd net: ipa: zero unused portions of filter table memory
Not all filter table entries are used.  Only certain endpoints
support filtering, and the table begins with a bitmap indicating
which endpoints use the "slots" that follow for filter rules.

Currently, unused filter table entries are not initialized.
Instead, zero-fill the entire unused portion of the filter table
memory regions, to make it more obvious that memory is unused (and
not subsequently modified).

This is not strictly necessary, but the result is reassuring when
looking at filter table memory.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:33 -08:00
Alex Elder
76b5fbcd6b net: ipa: kill ipa_modem_init()
A recent commit made disabling the SMP2P "setup ready" interrupt
unrelated to ipa_modem_stop().  Given that, it seems fitting to get
rid of ipa_modem_init() and ipa_modem_exit() (which are trivial
wrapper functions), and call ipa_smp2p_init() and ipa_smp2p_exit()
directly instead.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-25 19:37:33 -08:00
Alex Elder
e4e9bfb7c9 net: ipa: kill ipa_cmd_pipeline_clear()
Calling ipa_cmd_pipeline_clear() after stopping the channel
underlying the AP<-modem RX endpoint can lead to a deadlock.

This occurs in the ->runtime_suspend device power operation for the
IPA driver.  While this callback is in progress, any other requests
for power will block until the callback returns.

Stopping the AP<-modem RX channel does not prevent the modem from
sending another packet to this endpoint.  If a packet arrives for an
RX channel when the channel is stopped, an SUSPEND IPA interrupt
condition will be pending.  Handling an IPA interrupt requires
power, so ipa_isr_thread() calls pm_runtime_get_sync() first thing.

The problem occurs because a "pipeline clear" command will not
complete while such a SUSPEND interrupt condition exists.  So the
SUSPEND IPA interrupt handler won't proceed until it gets power;
that won't happen until the ->runtime_suspend callback (and its
"pipeline clear" command) completes; and that can't happen while
the SUSPEND interrupt condition exists.

It turns out that in this case there is no need to use the "pipeline
clear" command.  There are scenarios in which clearing the pipeline
is required while suspending, but those are not (yet) supported
upstream.  So a simple fix, avoiding the potential deadlock, is to
stop calling ipa_cmd_pipeline_clear() in ipa_endpoint_suspend().
This removes the only user of ipa_cmd_pipeline_clear(), so get rid
of that function.  It can be restored again whenever it's needed.

This is basically a manual revert along with an explanation for
commit 6cb63ea6a3 ("net: ipa: introduce ipa_cmd_tag_process()").

Fixes: 6cb63ea6a3 ("net: ipa: introduce ipa_cmd_tag_process()")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-23 12:26:40 +00:00
Alex Elder
8afc7e471a net: ipa: separate disabling setup from modem stop
The IPA setup_complete flag is set at the end of ipa_setup(), when
the setup phase of initialization has completed successfully.  This
occurs as part of driver probe processing, or (if "modem-init" is
specified in the DTS file) it is triggered by the "ipa-setup-ready"
SMP2P interrupt generated by the modem.

In the latter case, it's possible for driver shutdown (or remove) to
begin while setup processing is underway, and this can't be allowed.
The problem is that the setup_complete flag is not adequate to signal
that setup is underway.

If setup_complete is set, it will never be un-set, so that case is
not a problem.  But if setup_complete is false, there's a chance
setup is underway.

Because setup is triggered by an interrupt on a "modem-init" system,
there is a simple way to ensure the value of setup_complete is safe
to read.  The threaded handler--if it is executing--will complete as
part of a request to disable the "ipa-modem-ready" interrupt.  This
means that ipa_setup() (which is called from the handler) will run
to completion if it was underway, or will never be called otherwise.

The request to disable the "ipa-setup-ready" interrupt is currently
made within ipa_modem_stop().  Instead, disable the interrupt
outside that function in the two places it's called.  In the case of
ipa_remove(), this ensures the setup_complete flag is safe to read
before we read it.

Rename ipa_smp2p_disable() to be ipa_smp2p_irq_disable_setup(), to be
more specific about its effect.

Fixes: 530f9216a9 ("soc: qcom: ipa: AP/modem communications")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-23 12:06:40 +00:00
Alex Elder
33a153100b net: ipa: directly disable ipa-setup-ready interrupt
We currently maintain a "disabled" Boolean flag to determine whether
the "ipa-setup-ready" SMP2P IRQ handler does anything.  That flag
must be accessed under protection of a mutex.

Instead, disable the SMP2P interrupt when requested, which prevents
the interrupt handler from ever being called.  More importantly, it
synchronizes a thread disabling the interrupt with the completion of
the interrupt handler in case they run concurrently.

Use the IPA setup_complete flag rather than the disabled flag in the
handler to determine whether to ignore any interrupts arriving after
the first.

Rename the "disabled" flag to be "setup_disabled", to be specific
about its purpose.

Fixes: 530f9216a9 ("soc: qcom: ipa: AP/modem communications")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-23 12:06:40 +00:00
Christophe JAILLET
a6366b13c1 net: ipa: Use 'for_each_clear_bit' when possible
Use 'for_each_clear_bit()' instead of hand writing it. It is much less
version.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-19 11:19:11 +00:00
Alex Elder
816316caca net: ipa: disable HOLB drop when updating timer
The head-of-line blocking timer should only be modified when
head-of-line drop is disabled.

One of the steps in recovering from a modem crash is to enable
dropping of packets with timeout of 0 (immediate).  We don't know
how the modem configured its endpoints, so before we program the
timer, we need to ensure HOL_BLOCK is disabled.

Fixes: 84f9bd12d4 ("soc: qcom: ipa: IPA endpoints")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-15 13:25:45 +00:00
Alex Elder
6e228d8cbb net: ipa: HOLB register sometimes must be written twice
Starting with IPA v4.5, the HOL_BLOCK_EN register must be written
twice when enabling head-of-line blocking avoidance.

Fixes: 84f9bd12d4 ("soc: qcom: ipa: IPA endpoints")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-11-15 13:25:45 +00:00
Konrad Dybcio
27df68d579 net/ipa: ipa_resource: Fix wrong for loop range
The source group count was mistakenly assigned to both dst and src loops.
Fix it to make IPA probe and work again.

Fixes: 4fd704b360 ("net: ipa: record number of groups in data")
Acked-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@somainline.org>
Reviewed-by: Marijn Suijten <marijn.suijten@somainline.org>
Signed-off-by: Konrad Dybcio <konrad.dybcio@somainline.org>
Reviewed-by: Alex Elder <elder@linaro.org>
Link: https://lore.kernel.org/r/20211111183724.593478-1-konrad.dybcio@somainline.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2021-11-12 19:59:45 -08:00
Linus Torvalds
0dcf60d001 asm-generic: build fixes for v5.15
There is one build fix for Arm platforms that ended up impacting most
 architectures because of the way the drivers/firmware Kconfig file is
 wired up:
 
 The CONFIG_QCOM_SCM dependency have caused a number of randconfig
 regressions over time, and some still remain in v5.15-rc4. The
 fix we agreed on in the end is to make this symbol selected by any
 driver using it, and then building it even for non-Arm platforms with
 CONFIG_COMPILE_TEST.
 
 To make this work on all architectures, the drivers/firmware/Kconfig
 file needs to be included for all architectures to make the symbol
 itself visible.
 
 In a separate discussion, we found that a sound driver patch that is
 pending for v5.16 needs the same change to include this Kconfig file,
 so the easiest solution seems to have my Kconfig rework included in v5.15.
 
 There is a small merge conflict against an earlier partial fix for the
 QCOM_SCM dependency problems.
 
 Finally, the branch also includes a small unrelated build fix for NOMMU
 architectures.
 
 Link: https://lore.kernel.org/all/20210928153508.101208f8@canb.auug.org.au/
 Link: https://lore.kernel.org/all/20210928075216.4193128-1-arnd@kernel.org/
 Link: https://lore.kernel.org/all/20211007151010.333516-1-arnd@kernel.org/
 Signed-off-by: Arnd Bergmann <arnd@arndb.de>
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCgAdFiEEo6/YBQwIrVS28WGKmmx57+YAGNkFAmFgVp8ACgkQmmx57+YA
 GNlQoA/+O0ljtTy5D0MjRGmFDs11M5AtKNrfys82lm2GeEnc4lnxn722jLk8kR6s
 y6DSOWFs7w1bqhKExQNehZYtJO3sgW/9qiLMV9qfOx1Nc6WwhDPcYM9bMyGlpTmL
 M456nh8NopixV7slanNtfz1e0kbMKoK+4Ub7M5OHepK6x9FKQXQYQpeoBxaXHmWZ
 9eaRiL/CsRHO/cSkvpq1GtL7IVrudvij3FDHzxoDGFFjkCUm9LiN/8yrnVxHA9G7
 3EPyJazI559SsnxXJR32udGPJWZV1HZ7D5gbxDvzr5rZ9EX0JpyPGJsuXUR1wqlS
 UB2Y7AUTSxkwDiZ8UhPoXn6i67WAirzEsP2WmdS4v6NEbxlNloLGTIeGwcwkCRMU
 DBvMtDW8kKusgVu/OkEUgoC6MTRt+Mg+gZcQI/C4sp0MqZGaMY6c7abnYjqwEzBV
 ARS7bUYyME2GL6wNDPFB8esuD9jjdFXy96bGHATmzMxT3012K3X7ufFOzJZ+GOF9
 pan00fgoC17oiI+Xu/sZEHns6KvMTSE11Aw3uk+yhHxYtZbzWi2B5Nk+4tBdsOxF
 PAZdZ5qsyuEcBw+PyfbyZIHWOrlbvZkrmjiIsMJo63cIXuOtgraCjvRRAwe/ZwoU
 PXgPcUmrlAs06WjKhuQAZWt6bww7cEP2XyOYlDqwZ4Vj0dqav6g=
 =187C
 -----END PGP SIGNATURE-----

Merge tag 'asm-generic-fixes-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic

Pull asm-generic fixes from Arnd Bergmann:
 "There is one build fix for Arm platforms that ended up impacting most
  architectures because of the way the drivers/firmware Kconfig file is
  wired up:

  The CONFIG_QCOM_SCM dependency have caused a number of randconfig
  regressions over time, and some still remain in v5.15-rc4. The fix we
  agreed on in the end is to make this symbol selected by any driver
  using it, and then building it even for non-Arm platforms with
  CONFIG_COMPILE_TEST.

  To make this work on all architectures, the drivers/firmware/Kconfig
  file needs to be included for all architectures to make the symbol
  itself visible.

  In a separate discussion, we found that a sound driver patch that is
  pending for v5.16 needs the same change to include this Kconfig file,
  so the easiest solution seems to have my Kconfig rework included in
  v5.15.

  Finally, the branch also includes a small unrelated build fix for
  NOMMU architectures"

Link: https://lore.kernel.org/all/20210928153508.101208f8@canb.auug.org.au/
Link: https://lore.kernel.org/all/20210928075216.4193128-1-arnd@kernel.org/
Link: https://lore.kernel.org/all/20211007151010.333516-1-arnd@kernel.org/

* tag 'asm-generic-fixes-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/arnd/asm-generic:
  asm-generic/io.h: give stub iounmap() on !MMU same prototype as elsewhere
  qcom_scm: hide Kconfig symbol
  firmware: include drivers/firmware/Kconfig unconditionally
2021-10-08 11:57:54 -07:00
Arnd Bergmann
424953cf3c qcom_scm: hide Kconfig symbol
Now that SCM can be a loadable module, we have to add another
dependency to avoid link failures when ipa or adreno-gpu are
built-in:

aarch64-linux-ld: drivers/net/ipa/ipa_main.o: in function `ipa_probe':
ipa_main.c:(.text+0xfc4): undefined reference to `qcom_scm_is_available'

ld.lld: error: undefined symbol: qcom_scm_is_available
>>> referenced by adreno_gpu.c
>>>               gpu/drm/msm/adreno/adreno_gpu.o:(adreno_zap_shader_load) in archive drivers/built-in.a

This can happen when CONFIG_ARCH_QCOM is disabled and we don't select
QCOM_MDT_LOADER, but some other module selects QCOM_SCM. Ideally we'd
use a similar dependency here to what we have for QCOM_RPROC_COMMON,
but that causes dependency loops from other things selecting QCOM_SCM.

This appears to be an endless problem, so try something different this
time:

 - CONFIG_QCOM_SCM becomes a hidden symbol that nothing 'depends on'
   but that is simply selected by all of its users

 - All the stubs in include/linux/qcom_scm.h can go away

 - arm-smccc.h needs to provide a stub for __arm_smccc_smc() to
   allow compile-testing QCOM_SCM on all architectures.

 - To avoid a circular dependency chain involving RESET_CONTROLLER
   and PINCTRL_SUNXI, drop the 'select RESET_CONTROLLER' statement.
   According to my testing this still builds fine, and the QCOM
   platform selects this symbol already.

Acked-by: Kalle Valo <kvalo@codeaurora.org>
Acked-by: Alex Elder <elder@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
2021-10-07 16:51:57 +02:00
Alex Elder
b5c102238c net: ipa: initialize all filter table slots
There is an off-by-one problem in ipa_table_init_add(), when
initializing filter tables.

In that function, the number of filter table entries is determined
based on the number of set bits in the filter map.  However that
count does *not* include the extra "slot" in the filter table that
holds the filter map itself.  Meanwhile, ipa_table_addr() *does*
include the filter map in the memory it returns, but because the
count it's provided doesn't include it, it includes one too few
table entries.

Fix this by including the extra slot for the filter map in the count
computed in ipa_table_init_add().

Note: ipa_filter_reset_table() does not have this problem; it resets
filter table entries one by one, but does not overwrite the filter
bitmap.

Fixes: 2b9feef2b6 ("soc: qcom: ipa: filter and routing tables")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-09-08 12:10:33 +01:00
Alex Elder
2775cbc5af net: ipa: rename "ipa_clock.c"
Finally, rename "ipa_clock.c" to be "ipa_power.c" and "ipa_clock.h"
to be "ipa_power.h".

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-22 09:44:17 +01:00
Alex Elder
7aa0e8b8bd net: ipa: rename ipa_clock_* symbols
Rename a number of functions to clarify that there is no longer a
notion of an "IPA clock," but rather that the functions are more
generally related to IPA power management.

  ipa_clock_enable() -> ipa_power_enable()
  ipa_clock_disable() -> ipa_power_disable()
  ipa_clock_rate() -> ipa_core_clock_rate()
  ipa_clock_init() -> ipa_power_init()
  ipa_clock_exit() -> ipa_power_exit()

Rename the ipa_clock structure to be ipa_power.  Rename all
variables and fields using that structure type "power" rather
than "clock".

Rename the ipa_clock_data structure to be ipa_power_data, and more
broadly, just substitute "power" for "clock" in places that
previously represented things related to the "IPA clock".

Update comments throughout.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-22 09:44:17 +01:00
Alex Elder
1aac309d32 net: ipa: use autosuspend
Use runtime power management autosuspend.

Up until this point, we only suspended the IPA hardware for system
suspend; now we'll suspend it aggressively using runtime power
management, setting the initial autosuspend delay to half a second
of inactivity.

Replace pm_runtime_put() calls with pm_runtime_put_autosuspend(),
call pm_runtime_mark_last_busy() before each of those.  In places
where we're shutting things down, or decrementing power references
for errors, use pm_runtime_put_noidle() instead.

Finally, remove ipa_runtime_idle(), so the ->runtime_suspend
callback will occur if idle.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-22 09:44:17 +01:00
Alex Elder
c3f115aa5e net: ipa: kill ipa_clock_get()
The only remaining user of the ipa_clock_{get,put}() interface is
ipa_isr_thread().  Replace calls to ipa_clock_get() there calling
pm_runtime_get_sync() instead.  And call pm_runtime_put() there
rather than ipa_clock_put().  Warn if we ever get an error.

With that, we can get rid of ipa_clock_get() and ipa_clock_put().

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:45:47 +01:00
Alex Elder
724c2d7436 net: ipa: don't use ipa_clock_get() in "ipa_modem.c"
When we open or close the modem network device we need to ensure the
hardware is powered.  Replace the callers of ipa_clock_get() found
in ipa_open() and ipa_stop() with calls to pm_runtime_get_sync().
If an error is returned, simply return that error to the caller
(without any error or warning message).  This could conceivably
occur if the function was called while the system was suspended,
but that really shouldn't happen.  Replace corresponding calls to
ipa_clock_put() with pm_runtime_put() also.

If the modem crashes we also need to ensure the hardware is powered
to recover.  If getting power returns an error there's not much we
can do, but at least report the error.  (Ideally the remoteproc SSR
code would ensure the AP was not suspended when it sends the
notification, but that is not (yet) the case.)

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:45:47 +01:00
Alex Elder
799c5c24b7 net: ipa: don't use ipa_clock_get() in "ipa_uc.c"
Replace the ipa_clock_get() call in ipa_uc_clock() when taking the
"proxy" clock reference for the microcontroller with a call to
pm_runtime_get_sync().  Replace calls of ipa_clock_put() for the
microcontroller with pm_runtime_put() calls instead.

There is a chance we get an error when taking the microcontroller
power reference.  This is an unlikely scenario, where system suspend
is initiated just before we learn the modem is booting.  For now
we'll just accept that this could occur, and report it if it does.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:45:47 +01:00
Alex Elder
c43adc75dc net: ipa: don't use ipa_clock_get() in "ipa_smp2p.c"
If the "modem-init" Device Tree property is present for a platform,
the modem performs early IPA hardware initialization, and signals
this is complete with an "ipa-setup-ready" SMP2P interrupt.  This
triggers a call to ipa_setup(), which requires the hardware to be
powered.

Replace the call to ipa_clock_get() in this case with a call to
pm_runtime_get_sync().  And replace the corresponding calls to
ipa_clock_put() with calls to pm_runtime_put() instead.

There is a chance we get an error when taking this power reference.
This is an unlikely scenario, where system suspend is initiated just
before the modem signals it has finished initializing the IPA
hardware.  For now we'll just accept that this could occur, and
report it if it does.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:45:47 +01:00
Alex Elder
4c6a4da844 net: ipa: don't use ipa_clock_get() in "ipa_main.c"
We need the hardware to be powered starting at the config stage of
initialization when the IPA driver probes.  And we need it powered
when the driver is removed, at least until the deconfig stage has
completed.

Replace callers of ipa_clock_get() in ipa_probe() and ipa_exit(),
calling pm_runtime_get_sync() instead.  Replace the corresponding
callers of ipa_clock_put(), calling pm_runtime_put() instead.

The only error we expect when getting power would occur when the
system is suspended.  The ->probe and ->remove driver callbacks
won't be called when suspended, so issue a WARN() call if an error
is seen getting power.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:45:47 +01:00
Alex Elder
b8e36e13ea net: ipa: fix TX queue race
Jakub Kicinski pointed out a race condition in ipa_start_xmit() in a
recently-accepted series of patches:
  https://lore.kernel.org/netdev/20210812195035.2816276-1-elder@linaro.org/
We are stopping the modem TX queue in that function if the power
state is not active.  We restart the TX queue again once hardware
resume is complete.

  TX path                       Power Management
  -------                       ----------------
  pm_runtime_get(); no power    Start resume
  Stop TX queue                      ...
  pm_runtime_put()              Resume complete
  return NETDEV_TX_BUSY         Start TX queue

  pm_runtime_get()
  Power present, transmit
  pm_runtime_put()              (auto-suspend)

The issue is that the power management (resume) activity and the
network transmit activity can occur concurrently, and there's a
chance the queue will be stopped *after* it has been started again.

  TX path                       Power Management
  -------                       ----------------
                                Resume underway
  pm_runtime_get(); no power         ...
                                Resume complete
                                Start TX queue
  Stop TX queue       <-- No more transmits after this
  pm_runtime_put()
  return NETDEV_TX_BUSY

We address this using a STARTED flag to indicate when the TX queue
has been started from the resume path, and a spinlock to make the
flag and queue updates happen atomically.

  TX path                       Power Management
  -------                       ----------------
                                Resume underway
  pm_runtime_get(); no power    Resume complete
                                start TX queue     \
  If STARTED flag is *not* set:                     > atomic
      Stop TX queue             set STARTED flag   /
  pm_runtime_put()
  return NETDEV_TX_BUSY

A second flag is used to address a different race that involves
another path requesting power.

  TX path            Other path              Power Management
  -------            ----------              ----------------
                     pm_runtime_get_sync()   Resume
                                             Start TX queue   \ atomic
                                             Set STARTED flag /
                     (do its thing)
                     pm_runtime_put()
                                             (auto-suspend)
  pm_runtime_get()                           Mark delayed resume
  STARTED *is* set, so
    do *not* stop TX queue  <-- Queue should be stopped here
  pm_runtime_put()
  return NETDEV_TX_BUSY                      Suspend done, resume
                                             Resume complete
  pm_runtime_get()
  Stop TX queue
    (STARTED is *not* set)                   Start TX queue   \ atomic
  pm_runtime_put()                           Set STARTED flag /
  return NETDEV_TX_BUSY

So a STOPPED flag is set in the transmit path when it has stopped
the TX queue, and this pair of operations is also protected by the
spinlock.  The resume path only restarts the TX queue if the STOPPED
flag is set.  This case isn't a major problem, but it avoids the
"non-trivial amount of useless work" done by the networking stack
when NETDEV_TX_BUSY is returned.

Fixes: 6b51f802d6 ("net: ipa: ensure hardware has power in ipa_start_xmit()")
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-20 14:43:39 +01:00
Alex Elder
8dc181f2cd net: ipa: don't hold clock reference while netdev open
Currently a clock reference is taken whenever the ->ndo_open
callback for the modem netdev is called.  That reference is dropped
when the device is closed, in ipa_stop().

We no longer need this, because ipa_start_xmit() now handles the
situation where the hardware power state is not active.

Drop the clock reference in ipa_open() when we're done, and take a
new reference in ipa_stop() before we begin closing the interface.

Finally (and unrelated, but trivial), change the return type of
ipa_start_xmit() to be netdev_tx_t instead of int.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-14 14:13:39 +01:00
Alex Elder
8dcf8bb30f net: ipa: don't stop TX on suspend
Currently we stop the modem netdev transmit queue when suspending
the hardware.  For system suspend this ensured we'd never attempt
to transmit while attempting to suspend the modem endpoints.

For runtime suspend, the IPA hardware might get suspended while the
system is operating.  In that case we want an attempt to transmit a
packet to cause the hardware to resume if necessary.  But if we
disable the queue this cannot happen.

So stop disabling the queue on suspend.  In case we end up disabling
it in ipa_start_xmit() (see the previous commit), we still arrange
to start the TX queue on resume.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-14 14:13:38 +01:00
Alex Elder
6b51f802d6 net: ipa: ensure hardware has power in ipa_start_xmit()
We need to ensure the hardware is powered when we transmit a packet.
But if it's not, we can't block to wait for it.  So asynchronously
request power in ipa_start_xmit(), and only proceed if the return
value indicates the power state is active.

If the hardware is not active, a runtime resume request will have
been initiated.  In that case, stop the network stack from further
transmit attempts until the resume completes.  Return NETDEV_TX_BUSY,
to retry sending the packet once the queue is restarted.

If the power request returns an error (other than -EINPROGRESS,
which just means a resume requested elsewhere isn't complete), just
drop the packet.

Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2021-08-14 14:13:38 +01:00