The definition of DWC3_DCTL_HIRD_THRES macro is
completely wrong. It will only work for when we
want to read the register's contents for that bitfield.
Change the macro so that it can be used to writing to
the register, and when we need to read, we can add
extra right shift of 24 bits.
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
[ balbi@ti.com: add a commit log ]
Signed-off-by: Felipe Balbi <balbi@ti.com>
When mmc_host is not spi mode, mmc/sd is doing mmc_deselect_cards().
mmc_deselect_cards could be returned error.
If returned error, we can know something wrong when enter suspend.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Acked-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Commit fa550189 (mmc: core: Prevent eMMC VCC supply to be cut from late init)
slightly affected timings for how things are done for the omap MMC driver
causing the MMC cards not getting detected any longer.
Turns out this was caused by buggy reg_shift initialization in the omap
MMC driver that was happening after mmc_add_host() was being called.
Fix this by initializing reg_shift before mmc_add_host() is called.
Signed-off-by: Tony Lindgren <tony@atomide.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Chris Ball <cjb@laptop.org>
Commit b01a4f1c (mmc: omap: convert to per instance workqueue) initializes
the workqueue too late causing the following:
Unable to handle kernel NULL pointer dereference at virtual address 00000000
pgd = c0004000
[00000000] *pgd=00000000
Internal error: Oops: 5 [#1] SMP ARM
Modules linked in:
CPU: 0 Not tainted (3.4.0-08218-gb48b2c3 #158)
PC is at __queue_work+0x8/0x46c
LR is at queue_work_on+0x38/0x40
pc : [<c005bb4c>] lr : [<c005c00c>] psr: 60000193
sp : c0691e1c ip : 00000000 fp : c07374ac
r10: c7aae400 r9 : c0395700 r8 : 00000100
r7 : c0691e70 r6 : 00000000 r5 : 00000000 r4 : c7aae440
r3 : 00000001 r2 : c7aae440 r1 : 00000000 r0 : 00000000
Flags: nZCv IRQs off FIQs on Mode SVC_32 ISA ARM Segment kernel
Control: 00c5387d Table: 80004000 DAC: 00000017
Process swapper/0 (pid: 0, stack limit = 0xc06902f8)
Stack: (0xc0691e1c to 0xc0692000)
Fix this by initializing the workqueue before mmc_omap_remove_slot()
get called. Tested on n770, looks like n800 at least still has some
other issue with MMC.
Signed-off-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Commit b6e0703b (mmc: omap: make it behave well as a module) made some
__devinit changes but missed one function causing a section warning:
WARNING: vmlinux.o(.devinit.text+0x8604): Section mismatch in reference
from the function mmc_omap_probe)
The function __devinit mmc_omap_probe() references a function
__init mmc_omap_new_slot().
Signed-off-by: Tony Lindgren <tony@atomide.com>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
In case of "host->bus_hz < slot->clock", divider value is
miscalculated. And clock divider register value is multiple of 2. If
calculated divider value is odd number, result can be over-clocking.
Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Will Newton <will.newton@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
Setting host->data to NULL is incorrect sequence in
dw_mci_command_complete. This early setting makes the skip of
dma_unmap_sg in dw_mci_dma_cleanup.
Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Will Newton <will.newton@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
IDMAC may not be cleaned in driver probe if it has been already used in
boot time. So IDMAC needs sw reset newly. And DMA interface reset
precedes the internal DMAC reset. Additionally SDMMC_IDMAC_SWRESET is
replaced with magic code.
Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Acked-by: Will Newton <will.newton@gmail.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
DTO interrupt can be later than transmit interrupt(IDMAC) in case of
write. Current handling of IDMAC interrupt sets EVENT_DATA_COMPLETE as
well as EVENT_XFER_COMPLETE regardless of DTO rising. This makes the
current request finish in tasklet and permits the next request even
though current data transfer is still in progress. As a result, sequence
is broken and lock-up happens. Setting EVENT_DATA_COMPLETE is not proper
after IDMAC interrupt. It should be taken after DTO interrupt is
generated.
Reported-by: Dmitry Shmidt <dimitrysh@android.com>
Signed-off-by: Seungwon Jeon <tgih.jun@samsung.com>
Signed-off-by: Hyeonsu Kim <hyeonsu.kim@samsung.com>
Acked-by: Will Newton <will.newton@imgtec.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
SDIO_CCCR_IF[1:0] in SDIO card is used for card data bus width
setting as below:
00b: 1-bit bus
01b: Reserved
10b: 4-bit bus
11b: 8-bit bus (only for embedded SDIO)
And sdio_enable_wide is for setting data bus width as 4-bit.
But currently, it first reads the register, second OR' 1b with
SDIO_CCCR_IF[1], and then writes it back.
As we can see, this is based on such assumption that the
SDIO_CCCR_IF[0] is always 0. Apparently, this is not right.
Signed-off-by: Yong Ding <yongd@marvell.com>
Acked-by: Philip Rakity <prakity@marvell.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
The use of DMA slave config operation requires that the burst size
(aka chunk size) is specified through this interface.
Modify atmel-mci slave driver to use this specification on its side.
Signed-off-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
There's no need to declare the opmode ops
as extern since they're now dynamically
registered.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Since the RF config is done for all devices,
there's no need to keep a separate function
that is called for all devices, move it into
the general NIC config function.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Reviewed-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
We should also configure the PHY version in the
CSR_HW_IF_CONFIG_REG register for 1000 series
devices, not just for the other devices.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Reviewed-by: Wey-Yi W Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Commit 7c5ba4a830 ("iwlwifi: move queue
watchdog into transport") introduced the named constant
'IWL_WATCHHDOG_DISABLED'. Rename it to 'IWL_WATCHDOG_DISABLED'.
Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The EEPROM reading/parsing code is all mixed in
the driver today, and the EEPROM is parsed only
when we access data from it. This is problematic
because the NVM needs to be parsed and that is
independent of reading it. Also, the NVM format
for new devices will be different and probably
require a new parser.
Therefore refactor the reading and parsing and
create two independent components. Reading the
EEPROM requires direct hardware accesses and
therefore access to the transport, but parsing
is independent and can be done on an NVM blob.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This was missing. Fix the mask of the REV_TYPE on the way.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Johannes noticed this was completely messed up.
We got confused between masks and bit position.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
In a few cases we need to set a value in
a certain mask inside a register, add the
function iwl_set_bits_mask() to make such
code easy.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Different transports will have different needs: New tranports
need headroom for their own use before the Tx cmd. So allocate
the Tx cmd pool in the transport and give it a unique name
based on dev_name.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Although we don't use bit 24, this bit is valid, but bit 23
is not. Update the mask accordingly.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
We need to be able to enable / disable Tx queues in HW
dynamically. So this function is no longer related to AGG
only. It can do the job for any queue, even AC ones. Change
the name to better reflect its role.
Also use the new function to configure the AC / CMD queues
in tx_start.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
We need to be able to enable / disable Tx queues in HW
dynamically. So this function is no longer related to AGG
only. It can do the job for any queue, even AC ones. Change
the name to better reflect its role.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
This functions does the job so use it instead of duplicating
the code.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The DMA channels of the FH should be activated after the
configuration of the SCD queues too.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The scheduler can issue an interrupt when moving the read
pointer. To get this interrupt, the driver needs to
configure what queue can issue an interrupt when its read
pointer moves in the scheduler: this is the SCD_INT_MSK.
The driver also needs to enable the interrupt in
CSR_INT_MASK (bit CSR_INT_BIT_SCD).
Since we don't enable the scheduler interrupt in
CSR_INT_MASK, there is no point in requesting an interrupt
from the scheduler: it will be masked anyway. So don't
configure the scheduler to issue interrupts at all.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
There's no need to dynamically fill the HT40
band bitmap as it's a device parameter, just
put it into the HT configuration.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
There's no need to copy the same code for all
devices since none of the 5000 series devices
(that don't have the RX SISO override) don't
set the rx_with_siso_diversity variable.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Since we're working on another mode/driver
inside iwlwifi, move the current one into a
subdirectory to more cleanly separate the
code. While at it, rename all the files.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When mac80211 asks us to do HT40, it'll not do so
on a channel that we marked as not having HT40+/-
with IEEE80211_CHAN_NO_HT40PLUS (or MINUS). Thus,
there's no need to verify it again in the driver.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Structure the code a bit more and move all PCIe code
including the hardware configuration files into a
PCIe specific subdirectory.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
The STATUS_SCAN_HW is set before calling iwlagn_set_pan_params
(used as an input to calculate slot time allocation). The bit needs
to be cleared in case sending the command fails.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
We know whether we received a frame in GF format
or not, add it to the radiotap information.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
In channel switch, instead of relying on our internal
channel database, just use the mac80211 channel that
we filled with that information on startup.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Instead of iterating our own channel list,
use the mac80211 channel list since that's
already processed per band and thus makes
for less code.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
mac80211 will never set, switch to, or scan on an
invalid channel, so remove the code to validate
the channels against the driver channel list.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
mac80211 guarantees that the channel pointer is
always valid, so we can use that instead of our
own channel list.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
A lot of functions were temporarily made non-static
for experimental work, make them static again now.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
When a remain on channel request from mac80211 is followed by
a request to tx a mgmt frame offchannel, it is possible that
the remain on channel expires before the device reported the
tx status for the frame. This causes a race condition in
mac80211.
To fix this, delay the ROC notification to mac80211 until the
device reported the Tx status for all frames in the aux queue.
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
It is possible that the BSS context is not active (for example
when the current mode is set to GO), or that the vif->type is
different than station. In such a case we cannot
call mac80211 to report the average rssi for the interface
(the function assumes that the vif is valid and that the type
is station).
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Change the ns_timeout parameter of the wait ioctl to a signed value.
Doing this allows the kernel to provide an infinite wait when a timeout
of less than 0 is provided. This mimics select/poll.
Initially the parameter was meant to match up with the GL spec 1:1, but
after being made aware of how much 2^64 - 1 nanoseconds actually is, I
do not think anyone will ever notice the loss of 1 bit.
The infinite timeout on waiting is similar to the existing i915
userspace interface with the exception that struct_mutex is dropped
while doing the wait in this ioctl.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The data timeout timer was configured after mmc_add_host call. So, with bad
timings, it was possible to have a mmc request causing mod_timer call on a
non setup timer.
Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
3bdc9ba892 ("mmc: use really long write timeout to deal with crappy
cards") in 3.4 increased the write timeout that the core sends to host
drivers to 3 seconds. This makes sdhci's "requested timeout too large"
warning trigger on every write; so, change this pr_warning() to a DBG().
Reported-by: Adrian Hunter <adrian.hunter@intel.com>
Signed-off-by: Chris Ball <cjb@laptop.org>
By the time we're disabling the endpoint, HW
could already have posted more events to our
event buffer. In that case, we will receive
endpoint events for a disabled endpoint.
In order to protect ourselves from that situation,
we simply ignore endpoint interrupts whenever
the endpoint is disabled.
Tested-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
In case we get disconnected, we will call gadget
driver's disconnect method, which should make
sure to disable all endpoints. At that point
we will call stop_active_transfers() to make
sure we didn't leave any pending request on the
controller.
Tested-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
That IRQ is causing way too much trouble. We have
a different handling which was agreed with IP
provider and has been tested with FPGA and OMAP5.
Tested-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
We never set CMDIOC bit for Start Transfer
command, so that code will never be used.
Tested-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>