It's possible for FW to panic during early boot.

The patch re-introduces support to detect and
print those crashes.

This introduces an additional irq handler that is
set for the duration of early boot and shutdown.
The handler is then overriden with regular
handlers upon hif start().
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Preparation for code re-use. Also use ioread/write
wrappers.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It's impossible to rely on disable_irq() and/or CE
interrupt masking with legacy shared interrupts.
Other devices sharing the same irq line may assert
it while ath10k is doing something that requires
no interrupts.

Irq handlers are now registered after all
preparations are complete so spurious/foreign
interrupts won't do any harm. The handlers are
unregistered when no interrupts are required (i.e.
during driver teardown).

This also removes the ability to receive FW early
indication (since interrupts are not registered
until early boot is complete). This is not mission
critical (it's more of a hint that early boot
failed due to unexpected FW crash) and will be
re-added in a follow up patch.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The function did a couple of things: it allocated
CE completions, registered CE callbacks and
enabled CE interrupts through HW registers.

This cannot be so. Split the function into one
that allocates CE completions and the other one
that starts off CE operation.

This is required for future legacy shared
interrupt handling.

This also fixes possible memory leak if post rx
failed.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It's not really necessary for interrupts to be
used for BMI. BMI already assumes there's only one
caller at a time and it works directly with CE.

Make BMI poll for CE completions instead of
waiting for interrupts. This makes disabling
interrupts during early boot possible.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Hardware waits until host signals whether it has
chosen MSI(-X) or shared legacy interrupts. It is
not required for the driver to register interrupt
handlers immediately.

This patch prepares the pci irq code for more
changes.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

ath10k assumed all interrupts were directed to it.
This isn't the case for legacy shared interrupts.
ath10k consumed interrupts for other devices.

Check device irq status and return IRQ_NONE when
appropriate.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This makes ath10k a lot less spammy by default.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Convert the MSI failure warnings to a debug message to make them less spammy.
Also convert the irq mode printout to a single print to make it easier to
show it only once.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This should make sure the device won't issue any
interrupts nor access any memory after the driver
is stopped/freed thus avoid memory corruption in
some cases.
Reported-By: NBen Greear <greearb@candelatech.com>
Reported-By: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Add missing error reporting and adjust other
prints to make everything more consistent.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This patch moves irq registering after necessary
structures have been allocated and initialized.
This should prevent interrupts from causing
tasklet access invalid memory pointers.
Reported-By: NBen Greear <greearb@candelatech.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In case firmware crashes it may report CE
completions for entries that were never
submitted/filled with meaningful data. This in
turn led to NULL dereferences.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This shouldn't be silenced. This will be necessary
for PCI init code reordering.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This removes some remaining direct use of the wake
register which could interfere with power state
tracking of the target device. This will allow
initialization code reordering.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The check doesn't make much sense. If the address
were to be 0x0000 the check would fail. In this
case a 0 address isn't wrong.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

What the function does is to actually wait for the
firmware indication bit to be set. Prerequisite
for this is having interrupts registered.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The function will soon be called from more than 1
place.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It was possible for FW error tasklet to be
executed during teardown. This could lead to
system crashes and/or memory corruption.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It wasn't really useful to have it to begin with.
This makes it a little simpler to re-arrange PCI
init code as some function depended on
ar_pci->ce_count being set.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

memory is malloced in ath10k_pci_probe() and should be freed before
leaving from the error handling cases, otherwise it will cause
memory leak.
Signed-off-by: NWei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

ath10k_pci_wait() didn't notify any errors to callers, it
just printed a warning so add proper error handling. This fixes
a crash Ben reported:

ath10k: MSI-X interrupt handling (8 intrs)
ath10k: Unable to wakeup target
ath10k: target took longer 5000 us to wake up (awake count 1)
ath10k: Failed to get pcie state addr: -16
ath10k: early firmware event indicated
BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
IP: [<ffffffffa06ae46c>] ath10k_ce_completed_send_next+0x47/0x122 [ath10k_pci]
Reported-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

To make it easier to access SOC registers. No functional
changes.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

If firmware crashes during FW probing it would try
to perform FW recovery which uses mac80211
workqueue before registering to mac80211.

Using internal workqueue solves the problem.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The value provided by num_sends_allowed is now
derived from CE source ringbuffer state.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It is completely pointless to keep this function
around. It doesn't do anything different than
ce_send except it introduces more overhead.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Commit e9bb0aa3 ("ath10k: delete struct ce_sendlist") broke
num_sends_allowed incrementing. num_sends_allowed
exceeded initial values and could overflow.

This code was supposed to replenish
num_sends_allowed for partial sendlist items (i.e.
before final sendlist item from a sendlist was
completed and could be processed by completion
handlers).

Fortunately it seems it did not cause any major breakage,
yet.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The driver core clears the driver data to NULL after device_release
or on probe failure. Thus, it is not needed to manually clear the
device driver data to NULL.
Signed-off-by: NJingoo Han <jg1.han@samsung.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

struct ce_sendlist is useless as we always add just one buffer onto it.
And most importantly, it's ugly as it doesn't use skb properly.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

To unify all boot related debug messages into one debug
level.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The double while() could end up running forever.
Inner while() would complete very fast. However
the completion processing could take enough time
for more completions to flow in. In that case the
outer while() would not terminate and run again,
and again. This could happen especially on a slow
host system.

This could lead to a system freeze during heavy
traffic. Note: this doesn't solve all known
starvation issues yet.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The indentation here implies this was meant to be
a multi-statement if, but it lacks the braces.

kvalo: add "ath10k: " prefix
Signed-off-by: NDave Jones <davej@fedoraproject.org>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

ath10k doesn't support qca988x hw1.0 boards anymore. Unfortunately
the PCI id is the same in hw1.0 and hw2.0 so ath10k tries to use
hw1.0 boards anyway. But without hw1.0 workarounds in place
ath10k just crashes horribly.

To avoid using hw1.0 boards at all add a chip id detection
and fail the probe if hw1.0 is detected:

[ 5265.786408] ath10k: ERROR: qca988x hw1.0 is not supported
[ 5265.786497] ath10k: Unsupported chip id 0x043200ff
[ 5265.786574] ath10k: could not register driver core (-95)
[ 5265.793191] ath10k_pci: probe of 0000:02:00.0 failed with error -95

Also add a warning if there's an unknown chip id but continue
the boot process normally anyway.
Reported-by: NZaki Bakar <zaki.bm@gmail.com>
Tested-by: NChristian Lamparter <chunkeey@googlemail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

We should not try to access hw if wakeup fails so add
proper error checking for that. Also add the timeout lenght
to the warning message.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This is consistent with all other functions.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Void pointers are bad, mmkay.

No functional changes.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It's easier to read t if the field names are visible.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It's much more readable if struct entries in host_ce_config_wlan
are explicitly set.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

cacheline_sz is not used anywhere and can be removed.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>