staging: hv: fix the return status of netvsc_start_xmit()

Fix the return status, so the upper layer will retry if transmission
fails.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: fix counting of available buffer slots when send fails

Because the number of available buffer slots doesn't decrease for failed
sends, we should not call netvsc_xmit_completion(), which increase the
count of available slots. In this failed case, just free the memory is
enough.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: fix counting of #outstanding-sends in failed sends

If the packet failed to be sent, we shouldn't count it as the
number of outstanding sends.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: re-order the code in netvsc_probe()

Re-order the code in netvsc_probe() to prevent a guest crash caused by
packets possibly received from NetVSP before call to register_netdev().

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: fix a kernel warning in netvsc_linkstatus_callback()

netif_notify_peers() caused a kernel warning in netvsc_linkstatus_callback(),
because netvsc_linkstatus_callback() is within IRQ context. So we move
the first call to netif_notify_peers() into queued work as well, but with
zero delay.

In addition to "staging-next", this should also be back-ported to stable
kernels 2.6.32 and later.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: convert dev_<loglevel> to netdev_<loglevel> in netvsc

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: add newline to log messages in netvsc

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: hv: remove unnecessary includes in netvsc

hyperv.h is included by hyperv_net.h already, so no need to include it
again in these C files.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Cleanup the code in process_chn_event()

A channel in Hyper-V is equivalent to a device. Thus, a channel is
persistent once it is presented to the guest, even if the driver
managing this device is unloaded. By checking and invoking the driver
specific callback function under the protection of the channel
inbound_lock, we can properly deal with racing driver unloads since an
unloading driver sets the callback to NULL under the protection of this
inbound_lock.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Check for events before messages

The Windows team has informed us that on Windows guests on Hyper-V,
they check for events before messages. They also recommended that we do
the same. This patch addresses this.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Fix a bug in error handling in vmbus_bus_init()

Fix a bug in error handling in vmbus_bus_init().

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: util: Deal with driver register failures

Properly deal with vmbus_driver_register() failures.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: lirc_sasem: fix NULL pointer dereference in sasem_probe

If any memory allocation failed, goto alloc_status_switch
leads to mutex_unlock(&context->ctx_lock) while context is NULL.
The patch moves alloc_status_switch to handle error conditions
in correct way.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: lirc: fix mismatch in mutex lock-unlock in imon_probe

If kzalloc failed for allocation of context, goto alloc_status_switch
leads to unlock of unacquired driver_lock. The patch moves
mutex_lock(&driver_lock) before allocation of context.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: fix rtl8192e build errors

drivers/staging/rtl8192e/rtllib.h:2329: error: field 'wx_sem' has incomplete type
drivers/staging/rtl8192e/rtllib.h:2330: error: field 'scan_sem' has incomplete type
drivers/staging/rtl8192e/rtllib.h:2331: error: field 'ips_sem' has incomplete type
drivers/staging/rtl8192e/rtllib_wx.c:267: error: implicit declaration of function 'down'
drivers/staging/rtl8192e/rtllib_wx.c:290: error: implicit declaration of function 'up'
Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete IF_DEF_ScaleLCD

Delete one more disabled feature.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete IF_DEF_PWD

Delete already disabled code paths.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete IF_DEF_OEMUtil

Delete already disabled code paths.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete Win9xDOSMode

Delete another useless flag.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete IF_DEF_VideoCapture

Delete already disabled code paths.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated timing data

Delete redundant timing data tables. Replace as follows:

EnlargeTap4Timing ==> xgifb_tap4_timing
NoScaleTap4Timing ==> xgifb_tap4_timing
NTSCTap4Timing ==> xgifb_ntsc_525_tap4_timing
YPbPr525iTap4Timing ==> xgifb_ntsc_525_tap4_timing
YPbPr525pTap4Timing ==> xgifb_ntsc_525_tap4_timing

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated TV data tables

Delete redundant TV data tables. Replace as follows:

XGI_EPLCHTVDataPtr ==> xgifb_chrontel_tv
XGI_EPLCHTVRegPtr ==> xgifb_chrontel_tv

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated LCD table

XGI_EPLLCDCRT1Ptr_H and XGI_EPLLCDCRT1Ptr_V tables are identical. Replace
as follows:

XGI_EPLLCDCRT1Ptr_H ==> xgifb_epllcd_crt1
XGI_EPLLCDCRT1Ptr_V ==> xgifb_epllcd_crt1

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated LCD data

Delete duplicated LCD data tables. Replace as follows:

XGI_ExtLCD1400x1050Data ==> xgifb_lcd_1400x1050
XGI_StLCD1400x1050Data ==> xgifb_lcd_1400x1050
XGI_ExtLCD1280x1024x75Data ==> xgifb_lcd_1280x1024x75
XGI_StLCD1280x1024x75Data ==> xgifb_lcd_1280x1024x75
XGI_ExtLCDDes1400x1050Data ==> xgifb_lcddes_1400x1050
XGI_StLCDDes1400x1050Data ==> xgifb_lcddes_1400x1050
XGI_ExtLCDDLDes1400x1050Data ==> xgifb_lcddldes_1400x1050
XGI_StLCDDLDes1400x1050Data ==> xgifb_lcddldes_1400x1050
XGI_ExtLCDDes1024x768x75Data ==> xgifb_lcddes_1024x768x75
XGI_StLCDDes1024x768x75Data ==> xgifb_lcddes_1024x768x75
XGI_ExtLCDDes1280x1024x75Data ==> xgifb_lcddes_1280x1024x75
XGI_StLCDDes1280x1024x75Data ==> xgifb_lcddes_1280x1024x75
XGI_ExtLCDDLDes1280x1024x75Data ==> xgifb_lcddldes_1280x1024x75
XGI_StLCDDLDes1280x1024x75Data ==> xgifb_lcddldes_1280x1024x75

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated YFilter1 data

PAL-M and PAL-N "YFilter1" tables are identical. Replace as follows:

PALMYFilter1 ==> xgifb_palmn_yfilter1
PALNYFilter1 ==> xgifb_palmn_yfilter1

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated YFilter2 data

All four "YFilter2" tables are identical. Replace as follows:

NTSCYFilter2 ==> xgifb_yfilter2;
PALMYFilter2 ==> xgifb_yfilter2;
PALNYFilter2 ==> xgifb_yfilter2;
PALYFilter2 ==> xgifb_yfilter2;

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: do not initialize static data to 0

Remove redundant initializations.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: eliminate #ifdef Tap4

Eliminate #define Tap4 and related checks.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete commented-out routines

Delete routines which have been commented out.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: refactor XGI_SetXG21LVDSPara() and XGI_SetXG27LVDSPara()

The code is almost the same for XG21 and XG27, a single routine can
cover both.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: refactor XGI_SetXG21LCD() and XGI_SetXG27LCD()

The code is almost the same for XG21 and XG27, a single routine can
cover both.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete redundant Chrontel TV data

Delete redundant Chrontel TV data.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete support for Chrontel 7007 TV output

IF_DEF_CH7007 is hard-coded to 0, so this is all just dead code that
can be deleted.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete support for Chrontel 7005 TV output

IF_DEF_CH7005 is hard-coded to 0, so this is all just dead code that
can be deleted.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete support for Chrontel 7017 TV output

IF_DEF_CH7017 is hard-coded to 0, so this is all just dead code that
can be deleted.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete XGI_OpenCRTC()

Delete redundant function. Also eliminates the following sparse warning:

drivers/staging/xgifb/vb_setmode.c:8033:6: warning: symbol 'XGI_OpenCRTC' was not declared. Should it be static?

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: use common init for module and built-in

Use a common init function for module and built-in driver.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

stating: xgifb: update module parameter descriptions

Remove incorrect/misleading text from parameter descriptions.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete all references to mode "none"

Delete all references to mode "none". It should be handled like any
other invalid mode.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: correct the default mode comment

According to the comment, the default mode should be 800x600x8 while
the index points actually to 800x600x16. Change the comment and move
the definition instead of changing the current behaviour.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: simplify default mode setup

The default mode is the same regardless of the display type, except for
XG21/LCD which is determined dynamically.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: fail the probe if no supported LCD video mode found

Fail the probe if the LCD resolution described in card firmware does
not match any of the supported modes.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated default video mode setup code

The code for determining the default video mode is unnecessarily
duplicated in XGIfb_search_mode() and XGIfb_search_vesamode() functions.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: assume the default video mode also when module

The same default video mode should be assumed regardless if the driver
is a module or not. This won't change the current behaviour, but it will
simplify further changes.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: XGIfb_search_mode(): delete unreachable code

XGIfb_search_mode() is always called with a valid string pointer.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: search video mode later in probe

When searching the video mode, some HW details needs to be
known. Therefore it's not possible to do that reliably before probe().

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: fix LVDS LCD resolution data setup

If the driver fails to get LVDS LCD data from the firmware, it falls
back to using driver's hardcoded values. However, this is done too
early in the probe. The selected video mode index is not yet known,
and also in some cases it could be defaulting to -1 resulting in a read
outside the array. Fix this by moving the array lookup after the video
mode validation.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: check and report invalid option values

Check option values with kstrtoul(). This will also eliminate some
checkpatch warnings about simple_strtoul() usage.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: XGIfb_setup(): delete redundant assignment

The global data is already initialized to zero.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: XGIfb_setup(): clean up driver messages

Print options only if they are given. Delete other messages.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete "off" option handling

It's possible to disable the built-in xgifb driver with the kernel
command line option:

video=xgifb:off

This is already handled by the generic framebuffer layer (fb_get_options()
will return TRUE resulting in init failure), so no need to process the
option in the driver.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete duplicated code

Delete redundant branches for "mode" and "vesa" options.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: fix ypan disable option

Setting any of the FB options will disable ypan as a side effect. It
should be disabled only by the "noypan" option.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: xgifb: delete unused module parameters and options

Delete parameters which have no effect.

Signed-off-by: Aaro Koskinen <aaro.koskinen@iki.fi>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: serqt_usb2: remove ssu100 from supported devices

The serqt_usb2 driver will not work properly with the ssu100 device
even though it claims to support it.  The ssu100 is supported by the
ssu100 driver in mainline so there is no need to have it claimed by
serqt_usb2.

Signed-off-by: Bill Pemberton <wfp5p@virginia.edu>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging/easycap: fix mismatch in easycap_poll() mutex lock-unlock

In case if condition (kd != isdongle(peasycap)) becomes true,
easycap_poll() returns without releasing easycapdc60_dongle[kd].mutex_video.
The patch adds mutex_unlock() before that return.

Found by Linux Driver Verification project (linuxtesting.org).

Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Acked-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: et131x: Use phy-device, mii_bus and ethtool_ops

Adding some basic ethtool ops and supporting functionality using a phy device.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: et131x: Remove module_param et131x_speed_set

Manual setting of speed/duplex will be achieved using ethtool.
Remove the driver specific module_param that also does this.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: et131x: Remove redundant replica loopback code

A mechanism used to set the phy in loopback mode is not present in the driver, making associated checks and variables redundant. Removing them.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: et131x: Remove unused xcvr_id in struct ce_stats

xcvr_id holds the phy ID which is stored but never used in the driver.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

MAINAINERS: Add details for drivers/staging/et131x

Adding myself as a maintainer for this driver, as I appear to have been the only interested party for some time.
Not sure if Olaf Hartman is still interested? So will leave it up to him to add his name to the list.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

staging: et131x: Further tidy up of 131x_pci_setup()

* Removed unused bool variable.
* Eliminated mid-function returns, used gotos instead

Signed-off-by: Mark Einon <mark.einon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: return negative error codes

There was a typo here and we changed the -ENOMEM into +ENOMEM.  The
error codes aren't used, so this is just a cleanup.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: mousevsc: Fix checkpatch errors and warnings

Fix checkpatch errors and warnings.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Fix checkpatch warnings in connection.c

Fix checkpatch warnings in connection.c.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Fix a checkpatch warning in ring_buffer.c

Fix a checkpatch warning in ring_buffer.c

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Get rid of an unnecessary check in vmbus_connect()

Get rid of an unnecessary check in vmbus_connect().

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Get rid of some dated/redundant comments

Get rid of some dated/redundant comments in vmbus_drv.c

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Fixup indentation in vmbus_acpi_add()

Fixup indentation in vmbus_acpi_add().

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Do not enable auto eoi

Linux interrupt handling code generates the eoi; don't enable auto eoi.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Get rid of vmbus_on_isr() by inlining the code

Get rid of vmbus_on_isr() by inlining the code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: Fix a bug in vmbus_match()

The recent checkin that add a private pointer to hv_vmbus_device_id
introduced this bug in vmbus_match; fix it.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Properly deal with de-registering channel callback

Ensure that we correctly handle racing invocations of the channel callback
when the channel is being closed. We do this using the channel's inbound_lock.
A side-effect of this strategy is that we avoid repeatedly picking up this lock
as we drain the inbound ring-buffer.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: vmbus: Check before invoking the channel callback

When we close a channel, we set the corresponding callback function to NULL.
Check before invoking the channel callback.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: util: Adjust guest time in a process context

The current code was adjusting guest time in interrupt context; do this
in process context since we may have to initiate cross-processor
interrupts as part of setting time.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: util: Forcefully shutdown when shutdown is requested

When the host requests a "shutdown", make sure we shutdown!

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: No need to copy from bounce buffer in case of a failure

No need to copy from bounce buffer in case of a failure; cleanup the code
accordingly.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: In case of scsi errors offline the device

When we do get fatal errors from the host, offline the device since the
host has already tried all possible recovery actions.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Fix a typo

Fix a typo in a function name.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Fixup srb and scsi status for INQUIRY and MODE_SENSE

The current VHD handler on the Windows Host does not correctly handle
INQUIRY and MODE_SENSE commands with some options. Fixup srb_status
in these cases since the failure is not fatal.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Cleanup storvsc_drv.c after adding the contents of hyperv_storage.h

Cleanup storvsc_drv.c after adding the contents of hyperv_storage.h.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Add the contents of hyperv_storage.h to storvsc_drv.c

As part of further cleanup of storvsc, add the contents of hyperv_storage.h
to storvsc_drv.c and do the necessary adjustments.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Cleanup storvsc_drv.c after adding the contents of storvsc.c

Cleanup storvsc_drv.c after adding the contents of storvsc.c.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Include storvsc.c in storvsc_drv.c

As part of further cleanup of our storage drivers, include the content
of storvsc.c into storvsc_drv.c and delete storvsc.c and do the necessary
adjustments.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Optimize the bounce buffer handling in the "read" case

Optimize the bounce buffer handling in the "read" case.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Optimize bounce buffer handling for the "write" case

Optimize bounce buffer handling for the "write" case.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: blkvsc: Get rid of blkvsc_drv.c as this code is not used

Now that blkvsc driver is no longer needed, remove blkvsc_drv.c

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Handle IDE devices using the storvsc driver

Now, enable handling of all IDE devices by extending the storvsc
device id table to handle IDE guid. As part of this cleanup Kconfig
and Hyper-V Makefile to not build the IDE driver (blkvsc).

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Add code to handle IDE devices using the storvsc driver

Add code to handle IDE devices using the storvsc driver. The storvsc_probe()
is modified so that the storvsc driver can surface all disks presented to the
guest as scsi devices using generic upper level Linux scsi drivers.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Get rid of the refcnt field in struct netvsc_device

Get rid of the refcnt field in struct netvsc_device. We implement the following
logic to manage the life cycle of the device: If the device is being destroyed,
we do not allow any outgoing traffic. Furthermore, if the device is being
destroyed, we allow incoming traffic only to drain outgoing traffic. Note that
the driver may send some book keeping messages to the host not known to
upper level Linux code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Get rid of release_inbound_net_device() by inlining the code

Get rid of release_inbound_net_device() by inlining the code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Get rid of release_outbound_net_device() by inlining the code

Get rid of release_outbound_net_device() by inlining the code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Prevent outgoing traffic when netvsc dev is destroyed

Prevent outgoing traffic when netvsc dev is destroyed.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Introduce state to manage the lifecycle of net device

Introduce state to manage the lifecycle of net device.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Cleanup alloc_net_device()

Cleanup alloc_net_device(); we can directly set the refcnt.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: netvsc: Inline the code for free_net_device()

Inline the code for free_net_device().

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Get rid of the reference counting in struct storvsc_device

Get rid of the reference counting in struct storvsc_device. We manage the lifecycle with
the following logic: If the device is marked for destruction, we dot allow any
outgoing traffic on the device. Incoming traffic is allowed only to drain pending
outgoing traffic. Note that while the upper level code in Linux deals with outstanding
I/Os, we may have situations on Hyper-V where some book keeping messages are sent out
that the upper level Linux code may not be aware of.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Get rid of final_release_stor_device() by inlining code

Get rid of final_release_stor_device() by inlining code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Get rid of release_stor_device() by inlining the code

Get rid of release_stor_device() by inlining the code.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>

Staging: hv: storvsc: Prevent outgoing traffic when stor dev is being destroyed

Prevent outgoing traffic when stor dev is destroyed.

Signed-off-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>