Jia-Ju Bai [Tue, 10 Apr 2018 12:24:23 +0000 (20:24 +0800)]
staging: rtl8192u: Replace GFP_ATOMIC with GFP_KERNEL in ieee80211_softmac_init
ieee80211_softmac_init() is never called in atomic context.
The call chains ending up at ieee80211_softmac_init() is:
[1] ieee80211_softmac_init() <- alloc_ieee80211_rsl() <-
rtl8192_usb_probe()
rtl8192_usb_probe() is set as ".probe" in struct usb_driver.
Despite never getting called from atomic context,
ieee80211_softmac_init() calls kzalloc() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jia-Ju Bai [Tue, 10 Apr 2018 12:23:54 +0000 (20:23 +0800)]
staging: rtl8192u: Replace GFP_ATOMIC with GFP_KERNEL in prism2_wep_init
prism2_wep_init() is never called in atomic context.
prism2_wep_init() is only set as ".init" in
struct ieee80211_crypto_ops.
The call chains ending up at "->init" function are:
[1] ->init() <- ieee80211_wpa_set_encryption() <-
ieee80211_wpa_supplicant_ioctl()
[2] ->init() <- ieee80211_wx_set_encode_ext_rsl() <-
r8192_wx_set_enc_ext()
[3] ->init() <- ieee80211_wx_set_encode_rsl() <-
r8192_wx_set_enc()
ieee80211_wpa_supplicant_ioctl(), r8192_wx_set_enc_ext() and
r8192_wx_set_enc() call mutex_lock(), which indicates these functions
are not called in atomic context.
Despite never getting called from atomic context,
prism2_wep_init() calls kzalloc() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jia-Ju Bai [Tue, 10 Apr 2018 12:23:29 +0000 (20:23 +0800)]
staging: rtl8192u: Replace GFP_ATOMIC with GFP_KERNEL in ieee80211_tkip_init
ieee80211_tkip_init() is never called in atomic context.
ieee80211_tkip_init() is only set as ".init" in
struct ieee80211_crypto_ops.
The call chains ending up at "->init" function are:
[1] ->init() <- ieee80211_wpa_set_encryption() <-
ieee80211_wpa_supplicant_ioctl()
[2] ->init() <- ieee80211_wx_set_encode_ext_rsl() <-
r8192_wx_set_enc_ext()
[3] ->init() <- ieee80211_wx_set_encode_rsl() <-
r8192_wx_set_enc()
ieee80211_wpa_supplicant_ioctl(), r8192_wx_set_enc_ext() and
r8192_wx_set_enc() call mutex_lock(), which indicates these functions
are not called in atomic context.
Despite never getting called from atomic context,
ieee80211_tkip_init() calls kzalloc() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Jia-Ju Bai [Tue, 10 Apr 2018 12:23:00 +0000 (20:23 +0800)]
staging: rtl8192u: Replace GFP_ATOMIC with GFP_KERNEL in ieee80211_ccmp_init
ieee80211_ccmp_init() is never called in atomic context.
ieee80211_ccmp_init() is only set as ".init" in
struct ieee80211_crypto_ops.
The call chains ending up at "->init" function are:
[1] ->init() <- ieee80211_wpa_set_encryption() <-
ieee80211_wpa_supplicant_ioctl()
[2] ->init() <- ieee80211_wx_set_encode_ext_rsl() <-
r8192_wx_set_enc_ext()
[3] ->init() <- ieee80211_wx_set_encode_rsl() <-
r8192_wx_set_enc()
ieee80211_wpa_supplicant_ioctl(), r8192_wx_set_enc_ext() and
r8192_wx_set_enc() call mutex_lock(), which indicates these functions
are not called in atomic context.
Despite never getting called from atomic context,
ieee80211_ccmp_init() calls kzalloc() with GFP_ATOMIC,
which does not sleep for allocation.
GFP_ATOMIC is not necessary and can be replaced with GFP_KERNEL,
which can sleep and improve the possibility of sucessful allocation.
This is found by a static analysis tool named DCNS written by myself.
And I also manually check it.
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: comedi: cb_pcidas64: fix alignment of function parameters
Fix most checkpatch.pl issues of type:
CHECK: Alignment should match open parenthesis
Signed-off-by: Gabriel Francisco Mandaji <gfmandaji@gmail.com> Reviewed-by: Ian Abbott <abbotti@mev.co.uk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: vc04_services: Remove depends on HAS_DMA in case of platform dependency
Remove dependencies on HAS_DMA where a Kconfig symbol depends on another
symbol that implies HAS_DMA, and, optionally, on "|| COMPILE_TEST".
In most cases this other symbol is an architecture or platform specific
symbol, or PCI.
Generic symbols and drivers without platform dependencies keep their
dependencies on HAS_DMA, to prevent compiling subsystems or drivers that
cannot work anyway.
This simplifies the dependencies, and allows to improve compile-testing.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Reviewed-by: Mark Brown <broonie@kernel.org> Acked-by: Robin Murphy <robin.murphy@arm.com> Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stefan Wahren [Sat, 31 Mar 2018 20:09:39 +0000 (22:09 +0200)]
staging: vchiq_core: remove BITSET_ZERO
Hiding memset behind a macro isn't the best, because it relies on that
the parameter is not a pointer. Luckily all user has been removed, so
we can remove BITSET_ZERO too.
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stefan Wahren [Sat, 31 Mar 2018 20:09:37 +0000 (22:09 +0200)]
staging: vchiq_core: Fix missing semaphore release in error case
The bail out branch in case of a invalid tx_pos missed a semaphore
release. Dan Carpenter found this with a static checker.
Fixes: d1eab9dec610 ("staging: vchiq_core: Bail out in case of invalid tx_pos") Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
pi433_write requires locking due to multiple writers. After acquiring
the lock check if enough free space is available in the kfifo to write
the whole message. This check should prevent partial writes to tx_fifo
so kfifo_reset is not needed anymore.
pi433_tx_thread is the only reader so it does not require locking
after kfifo_reset is removed.
Signed-off-by: Valentin Vidic <Valentin.Vidic@CARNet.hr> Reviewed-by: Marcus Wolf <marcus.wolf@wolf-entwicklungen.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Change default transfer mode to DMA
The current default transfer is to use DMA or not depending on the
size of the data. The upstream driver mtk-sd uses DMA all times,
change the standard mode here to DMA for testing, if there are any
performance problems with DMA for small data sizes. If not, the option
for transfer mode should be removed in the future,
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The msdc_init_gpd_bd function is currently hard to read, because of
old, commented out code and a while loop, where a for loop is much
easier to read. Refactor it to make the code more readable.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove unused field xfersz from msdc_dma
The xfersz field of msdc_dma is only set to host->xfer_size and this
is also available at the (few) positions, where dma.xfersz is used, so
it is removed.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The flags of msdc_dma are set to a constant value, so block and dword
padding are disabled and checksum is enabled. By removing the field
the code is easier to read and the difference to the mtk-sd driver is
smaller.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove unused fields from msdc_dma
The fields used_gpd|bd in struct msdc_dma are only set and reset, but
never used, so remove them. The other removed fields are never used,
so they were removed either.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Start cleanup of msdc_dma_config
Currently the msdc_dma_config function has some variables, that are
not needed, uses the macro msdc_init_bd that is only used here and
does not use the for_each_sg iterator. That last fact could cause a
bug if the scatterlist is chained.
The function is changed to remove these things, but none of the
changes should change the behavior of it (despite the case of a linked
scatterlist).
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove unused function msdc_dma_dump
The msdc_dma_dump function is disabled by a preprocessor macro, was
only used in the msdc_dma_setup function. The code also contains no
information about the device, it only dumps some memory for debug.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove unused field burstsz from msdc_dma
The burstsz field in msdc_dma is set to a constant value and later
only used two times, using the constant directly in that places allows
to remove the field and improves readability.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Relax cpu while waiting for stable clock
Current code just busy waits with nop for clock becoming stable,
relaxing the cpu here improves readability and portability and also
removes a unnecessary difference with the mtk-sd driver.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Refactor and rename msdc_reset to msdc_reset_hw
In the current code the hardware reset is realised via a macro and
does only wait for a certain ammount of time for the controller to
come out of reset state, but it should wait until it IS out of reset
state. So it is refactored to improve this, it is also changed from a
macro to a function and renamed to msdc_reset_hw to match the mtk-sd
driver.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The enable|disable_sdio_irq() fields of msdc_hw are never set, so when
they are called, they are null pointer. In case of the
MSDC_EXT_SDIO_IRQ flag set in msdc0_hw, this happens. But because
these fields are never set, they can simply be removed.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The opcode check in msdc_command_start is redundant in case of
MMC_SET_RELATIVE_ADDR, this is 3 like SD_SEND_RELATIVE_ADDR, so the
second check can be removed.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Reviewed-by: NeilBrown <neil@brown.name> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove error message in debug
If the debug file in proc fs is not successfully created current code
prints an error message, this is removed. de is also not an error
pointer, so the IS_ERR() call leads to a static checker warning. The
unused return value of the msdc_debug_proc_init function is also removed.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Reviewed-by: NeilBrown <neil@brown.name> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The functions msdc_eirq_cd and msdc_eirq_sdio are defined but never
used, so they are removed. This also fixes two compiler warnings:
drivers/staging/mt7621-mmc/sd.c:517:13: warning: ‘msdc_eirq_cd’ defined but
not used [-Wunused-function]
static void msdc_eirq_cd(void *data)
^~~~~~~~~~~~
drivers/staging/mt7621-mmc/sd.c:507:13: warning: ‘msdc_eirq_sdio’ defined but
not used [-Wunused-function]
static void msdc_eirq_sdio(void *data)
staging: mt7621-mmc: Fix power management by removing old api
The mmc_suspend|resume_host functions have been decrepated in the
3.11 [1] release and were completely removed in 3.13 [2]. Removing the
calls to this functions as it was done for other mmc host drivers
before the old api was removed (eg. [3]) fixes compile errors in this
driver while CONFIG_PM is enabled.
The rest of the removed code is simply dead, the MSDC_SYS_SUSPEND flag
is always set and will be removed in a later patch.
[1] commit 986892ca78ee ("mmc: core: Initiate suspend|resume from mmc bus instead of mmc host")
[2] commit 3c0d22e8180b ("mmc: core: Remove deprecated mmc_suspend|resume_host APIs")
[3] commit 1e63d4854202 ("mmc: au1xmmc: Move away from using deprecated APIs")
The request of the irq line is not correctly handled by the current
code, platform_get_irq may fail and if the irq is requested via
devm_request_irq, its release is automatically done.
staging: mt7621-mmc: Remove dead codepath from probe function
The probe function contains lots of code, that is never used, because
the fields it checks for in the msdc_hw structure are never
initialized. Removing them should be safe.
staging: mt7621-mmc: Fix different caps for different devices
Current code saves the need for polling in a global variable, that
leeds to problems, when the driver should handle multiple devices with
different capabilities. By looking up the capability in the mmc_host
capabilities, they now have per device scope.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Remove unnecessary setting of dev.platform_data
In the probe function, pdev->dev.platform_data is set to &msdc0_hw,
but the only get access to that is also in the probe function. So that
assignment is removed. The msdc0_hw structure is still available to the
driver as part of the msdc_host structure.
staging: mt7621-mmc: Fix bug on dma_alloc_coherent fail
In case of dma_alloc_coherent failing the current code just called
BUG_ON. By adding error handling for that case this can be avoided.
This also fixes a memory leek in case of a fail later on in the probe
function.
Current code calls release_mem_regin on driver remove without
requesting it explicit first. The region is only requested via
devm_ioremap_resource and that releases it automatically. Removing the
release_mem_region calls fixes this.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-mmc: Fix memory leek in case of error in probe
If the base address is not successfully obtained in the probe
function, then the mmc_host struct is not freed. Adding an exit for
that case fixes the bug.
staging: mt7621-mmc: Remove code for different soc types
Remove code that is intended for soc other then the mt7621 because
this driver is only for that soc. This also removes some conditions.
However the remaining code is also at the wrong place, it just selects
the mode of the pins, this is a job for pinctl not for the sdxc driver.
Signed-off-by: Christian Lütke-Stetzkamp <christian@lkamp.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
staging: mt7621-pci: Hack 2 more PCI interrupts for gnubee1
The mt7621-pci support 3 pci devices and has 3 interrupts.
Each of these need to be enabled by the same sort of hack to
map hwirq number to virq number.
This is a hack which will go as soon as I understand how this is
supposed to work.
Ji-Hun Kim [Thu, 5 Apr 2018 07:09:20 +0000 (16:09 +0900)]
staging: vt6655: add handling memory leak on vnt_start()
There was no code for handling memory leaks of device_init_rings() and
request_irq(). It needs to free allocated memory in the device_init_rings()
, when request_irq() would be failed. Add freeing sequences of irq and
device init rings.
Signed-off-by: Ji-Hun Kim <ji_hun.kim@samsung.com> Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>