Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This change moves the lock so that it protects a call to __skb_dequeue().

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

The incorrect placement of a lock meant that the TX ring queue was not
protected. On one system, this led to freezes when running rtl8192ee.
When checking the other drivers to see which ones are affected, it was
discovered that several of them used identical code. Accordingly, this
routine was moved into core.c where all the drivers could share a single
copy.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Conflicts:
	rtl8188ee/fw.c
	rtl8192c/fw_common.c
	rtl8192de/fw.c
	rtl8192ee/fw.c
	rtl8723ae/fw.c
	rtl8723be/fw.c
	rtl8821ae/fw.c

Conflicts:
	rtl8188ee/trx.c

In a number of places, kmalloc or valloc were used to acquire memory.
To ensure that these locations are correctly initialized, the calls were
changed to kzalloc and vzalloc. The change seems to have cleared a problem
that was causing HT operations to be cancelled.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Conflicts:
	rtl8723ae/hal_btc.c

In kernel 3.13.0-XX where XX is >= 32, Ubuntu has implemented the API
change that was introduced in mainline V3.15 by commit d8ca16db6bb2 ("mac80211:
add length check in ieee80211_is_robust_mgmt_frame()"). The result is that the
code in this repo gets it wrong for Ubuntu kernels. The result is
a kernel that crashes when using the corresponging driver. Thanks to user
"lenonk" at GitHub, there is a fix. I have not tested it as I do not use one
of the affected kernels; however, it looks correct.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Whe uninstalling the drivers, the saved versions are restored. Firmware is
not touched when uninstalling.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

…as never mapped

This fix is a kluge. I'm not sure that these buffers are mapped under any
circumstance. At least the splat goes away.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Conflicts:
	rtl8192ee/hw.c

----------------------
11A Throughput is very low
because ratr is not initialized

---------------------
this will affect some modes that issue beacon

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

…ation

The low-level routines involve the use of outb(), etc. These routines are not
available for all architectures, and some builds will fail. The high-level
routines are correct for all architectures.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Driver rtlwifi has been modified to call ieee80211_register_hw()
from the probe routine; however, the existing call in the callback
routine for deferred firmware loading was not removed.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

In 3.18, the kernel contains options to compress the modules. When this
option is selected, installation of modules from this repo resulted in a
mixture of old and new versions that would not load correctly.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Device RTL8812AE fails to work on these kernels if VHT is enabled because
of lack of implementation for add_chanctx.

Signed-off-by: troy-tan <troy_tan@realsil.com.cn>

----------------------
make cam code more easy to view
and prepare for concurrent function

-------------------------------
1.fix cannot parse C2H CMD
2.fix TX HANG when adhoc is setup
3.fix TX HANG sometimes
4.fix rx desc unavailable

 Author:    troy-tan <troy_tan@realsil.com.cn>
 Date:      Mon Jan 12 10:17:32 2015 +0800

This callback routine pointer was added to driver rtl8192ee to help
fix the stability of the connection. One problem with that fix is that
this pointer is undefined for the rest of the drivers. To prevent
crashes, we need to test for a non-NULL pointer before we call the
routine.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

These changes convert the various makefiles to be similar to those of
the kernel. Now, the build is much cleaner, and using the -jX switch
for multiple build processes is much better.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

The rates used are of the form DESC92C_RATEXXXX. Rename them to DESC_RATEXXXX.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

All hw driver components in the rtlwifi driver, except for the
rtl8192de component has this bug. They would clear BE bit in the
ACM_CTRL register instead of the VO bit when processing the VO queue.

Signed-off-by: Jes Sorensen <Jes.Sorensen@redhat.com>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This array would be uninitialized in the unlikely case that the device did not
boot from EFUSE, but this change silences any possible compiler warning.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Fix XZ compression

For devices using the new style of trx flow, when a new skb cannot be
allocated, pointer buffer_desc is left as NULL.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

When a CAM key is changed, the change is always logged. The log level is
changed from DBG_EMERG to DBG_LOUD.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

When a beacon is transmitted, the "own" bit is not cleared. As a result, most
of the implementations of is_tx_desc_closed() will return false. This causes
the TX interrupt service routine to quit early and leak an IOMMU mapping.
By testing for a beacon before calling the above routine, this leakage is
prevented.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Transmission of a beacon while in AP mode does not generate an interrupt, thus
the TX interrupt service routine is not called, and the cleanup is handled in
a tasklet completion routine. Unfortunately, this routine does not release the
DMA mapping before it frees the skb.

Signed-off-by: Shao Fu <shaofu@realtek.com>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

USB ID 2001:330d is used for a D-Link DWA-131.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: shaofu <shaofu@realtek.com>

Remove duplicate spin_lock in rtl8821ae_gpio_radio_on_off_checking.

It is merged error in commit below:
"rtlwifi_new: rtl8821ae: Change driver to match version in kernel 3.18"

Signed-off-by: shaofu <shaofu@realtek.com>

Symptom:
TX randomly blocked for 3~5 seconds while measuring tx throughput (iperf).

Root cause:
This issue happenes in rtl_pci_flush(). The function uses a while-loop to wait TX queue lenght to decrease to 0. TX queue length reflects the packets are queued in driver.

Driver relys on TX OK interrupt to return skb and reduce TX queue length.

The interrupt subroutine disables interupt, read it and then cleaning it in the beginning of _rtl_pci_interrupt(). After all interupts process are finished, driver invoke enable_interrupt() to enable interupt again. This is a normal interrupt subroutine.

But enable_interrupt() invokes clear_interrupt() again. This unexpected interrupt cleaning may clean some fresh TX OK interrupts. These missing interrupts cause TX queue lenght never reduce to 0. Finally caused rtl_pci_flush() stucks in unterminated while-loop.

This patch remove clear_interrupt() in enable_interrupt() to avoid unexpected interrupt cleaning.

Signed-off-by: vincent_fann <vincent_fann@realtek.com>
Signed-off-by: shaofu <shaofu@realtek.com>

Conflicts:
	rtl8821ae/hw.c

* 8188ee/8723be/8192ee/8821ae use eeprom_channelplan value defaultly
* Add channel_plan to country code function

Signed-off-by: shaofu <shaofu@realtek.com>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Conflicts:
	btcoexist/halbtc8812a_ext.c

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

New uses of /proc in the kernel are discouraged, thus the in-kernel versions
of these drivers do not include any inclusion of such code. As the code also
causes build problems with some kernel versions, all of it is deleted.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

In kernel 4.3, the rtlwifi directory has been moved. Add the code needed
to detect the change, and install these modules in the correct location.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

how to install instruction added for the debian based distro.

installation instruction added.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This driver has a number of errors in the module initialization. These
include the following:

Paramters sw_crypto and disable_watchdog were never stored in the final
locations, nor were they initialized properly.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This driver has some errors in the handling of module parameters. These
include missing initialization for parameters msi_support and
disable_watchdog. In addition, neither of these parameters nor sw_crypto
are transferred into the locations used by the driver. A final fix is
adding parameter msi to the module named and description macros.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This driver failed to copy parameters sw_crypto and disable_watchdog into
the locations actually used by the driver. In addition, msi_support was
initialized three times and one of them used the wrong variable. The
initialization of parameter int_clear was moved so that it is near that
of the rest of the parameters.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

In this driver, parameters disable_watchdog and sw_crypto are never
copied into the locations used in the main code. While modifying the
parameter handling, the copying of parameter msi_support is moved to
be with the rest.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Two of the module parameters are listed with incorrect default values.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Two of the module parameter descriptions show incorrect default values.
In addition the value for software encryption is not transferred to
the locations used by the driver.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

The module parameter for software encryption was never transferred to
the location used by the driver.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This driver fails to copy the module parameter for software encryption
to the locations used by the main code.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

The old code returned only allowable channels for the 2.4 GHz band.
Now it also permits 5G bands.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This log message is totally useless.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: timlee <timlee@realtek.com>
Signed-off-by: shaofu <shaofu@realtek.com>

The current code has pci_alloc_consistent() followed by memset().
Replace the two calls with a simple pci_zalloc_consistent().

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

…et_txpower_by_rate()

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

These changes make diff'ing with the kernel easier.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

This info is no longer used in the kernel. Removing it makes it easier
to compare this code with the kernal version.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

…w variable

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

…ces with kernel code

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Calling synchronize_irq() right before free_irq() is quite useless. On one
hand the IRQ can easily fire again before free_irq() is entered, on the
other hand free_irq() itself calls synchronize_irq() internally (in a race
condition free way), before any state associated with the IRQ is freed.

Patch was generated using the following semantic patch:
// <smpl>
@@
expression irq;
@@
-synchronize_irq(irq);
 free_irq(irq, ...);
// </smpl>

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>

Signed-off-by: Dmitry Tunin <hanipouspilot@gmail.com>

passing rtl_stats by value is inefficient; the structure is over 300
bytes in size and generally just one field (packet_report_type)
is being accessed, so the pass by value is a relatively large overhead.
This change just affects just the rx_command_packet calls.

Signed-off-by: Colin Ian King <colin.king@canonical.com>

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>