commit 94788af4ed upstream.
I was doing some experiments with I2C and noticed that Tegra APB DMA
driver crashes sometime after I2C DMA transfer termination. The crash
happens because tegra_dma_terminate_all() bails out immediately if pending
list is empty, and thus, it doesn't release the half-completed descriptors
which are getting re-used before ISR tasklet kicks-in.
tegra-i2c 7000c400.i2c: DMA transfer timeout
elants_i2c 0-0010: elants_i2c_irq: failed to read data: -110
------------[ cut here ]------------
WARNING: CPU: 0 PID: 142 at lib/list_debug.c:45 __list_del_entry_valid+0x45/0xac
list_del corruption, ddbaac44->next is LIST_POISON1 (00000100)
Modules linked in:
CPU: 0 PID: 142 Comm: kworker/0:2 Not tainted 5.5.0-rc2-next-20191220-00175-gc3605715758d-dirty #538
Hardware name: NVIDIA Tegra SoC (Flattened Device Tree)
Workqueue: events_freezable_power_ thermal_zone_device_check
[<c010e5c5>] (unwind_backtrace) from [<c010a1c5>] (show_stack+0x11/0x14)
[<c010a1c5>] (show_stack) from [<c0973925>] (dump_stack+0x85/0x94)
[<c0973925>] (dump_stack) from [<c011f529>] (__warn+0xc1/0xc4)
[<c011f529>] (__warn) from [<c011f7e9>] (warn_slowpath_fmt+0x61/0x78)
[<c011f7e9>] (warn_slowpath_fmt) from [<c042497d>] (__list_del_entry_valid+0x45/0xac)
[<c042497d>] (__list_del_entry_valid) from [<c047a87f>] (tegra_dma_tasklet+0x5b/0x154)
[<c047a87f>] (tegra_dma_tasklet) from [<c0124799>] (tasklet_action_common.constprop.0+0x41/0x7c)
[<c0124799>] (tasklet_action_common.constprop.0) from [<c01022ab>] (__do_softirq+0xd3/0x2a8)
[<c01022ab>] (__do_softirq) from [<c0124683>] (irq_exit+0x7b/0x98)
[<c0124683>] (irq_exit) from [<c0168c19>] (__handle_domain_irq+0x45/0x80)
[<c0168c19>] (__handle_domain_irq) from [<c043e429>] (gic_handle_irq+0x45/0x7c)
[<c043e429>] (gic_handle_irq) from [<c0101aa5>] (__irq_svc+0x65/0x94)
Exception stack(0xde2ebb90 to 0xde2ebbd8)
Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Acked-by: Jon Hunter <jonathanh@nvidia.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20200209163356.6439-2-digetx@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit bd73dfabdd ]
Illegal memory will be touch if SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V3
(41) exceed the size of structure sdma_script_start_addrs(40),
thus cause memory corrupt such as slob block header so that kernel
trap into while() loop forever in slob_free(). Please refer to below
code piece in imx-sdma.c:
for (i = 0; i < sdma->script_number; i++)
if (addr_arr[i] > 0)
saddr_arr[i] = addr_arr[i]; /* memory corrupt here */
That issue was brought by commit a572460be9 ("dmaengine: imx-sdma: Add
support for version 3 firmware") because SDMA_SCRIPT_ADDRS_ARRAY_SIZE_V3
(38->41 3 scripts added) not align with script number added in
sdma_script_start_addrs(2 scripts).
Fixes: a572460be9 ("dmaengine: imx-sdma: Add support for version 3 firmware")
Cc: stable@vger.kernel
Link: https://www.spinics.net/lists/arm-kernel/msg754895.html
Signed-off-by: Robin Gong <yibin.gong@nxp.com>
Reported-by: Jurgen Lambrecht <J.Lambrecht@TELEVIC.com>
Link: https://lore.kernel.org/r/1569347584-3478-1-git-send-email-yibin.gong@nxp.com
[vkoul: update the patch title]
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c24a5c735f ]
The commit
080edf75d3 ("dmaengine: hsu: set HSU_CH_MTSR to memory width")
has been mistakenly submitted. The further investigations show that
the original code does better job since the memory side transfer size
has never been configured by DMA users.
As per latest revision of documentation: "Channel minimum transfer size
(CHnMTSR)... For IOSF UART, maximum value that can be programmed is 64 and
minimum value that can be programmed is 1."
This reverts commit 080edf75d3.
Fixes: 080edf75d3 ("dmaengine: hsu: set HSU_CH_MTSR to memory width")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit b53611fb1c ]
Commit f33e7bb3eb ("dmaengine: tegra210-adma: restore channel status")
added support to save and restore the DMA channel registers when runtime
suspending the ADMA. This change is causing the kernel to crash when
probing the ADMA, if the device is probed deferred when looking up the
channel interrupts. The crash occurs because not all of the channel base
addresses have been setup at this point and in the clean-up path of the
probe, pm_runtime_suspend() is called invoking its callback which
expects all the channel base addresses to be initialised.
Although this could be fixed by simply checking for a NULL address, on
further review of the driver it seems more appropriate that we only call
pm_runtime_get_sync() after all the channel interrupts and base
addresses have been configured. Therefore, fix this crash by moving the
calls to pm_runtime_enable(), pm_runtime_get_sync() and
tegra_adma_init() after the DMA channels have been initialised.
Fixes: f33e7bb3eb ("dmaengine: tegra210-adma: restore channel status")
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f33e7bb3eb ]
Status of ADMA channel registers is not saved and restored during system
suspend. During active playback if system enters suspend, this results in
wrong state of channel registers during system resume and playback fails
to resume properly. Fix this by saving following channel registers in
runtime suspend and restore during runtime resume.
* ADMA_CH_LOWER_SRC_ADDR
* ADMA_CH_LOWER_TRG_ADDR
* ADMA_CH_FIFO_CTRL
* ADMA_CH_CONFIG
* ADMA_CH_CTRL
* ADMA_CH_CMD
* ADMA_CH_TC
Runtime PM calls will be inovked during system resume path if a playback
or capture needs to be resumed. Hence above changes work fine for system
suspend case.
Fixes: f46b195799 ("dmaengine: tegra-adma: Add support for Tegra210 ADMA")
Signed-off-by: Sameer Pujar <spujar@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 648865a79d ]
In 2D transfers (for the AXI DMAC), the number of frames (numf) represents
Y_LENGTH, and the length of a frame is X_LENGTH. 2D transfers are useful
for video transfers where screen resolutions ( X * Y ) are typically
aligned for X, but not for Y.
There is no requirement for Y_LENGTH to be aligned to the bus-width (or
anything), and this is also true for AXI DMAC.
Checking the Y_LENGTH for alignment causes false errors when initiating DMA
transfers. This change fixes this by checking only that the Y_LENGTH is
non-zero.
Fixes: 0e3b67b348 ("dmaengine: Add support for the Analog Devices AXI-DMAC DMA controller")
Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 3e5daee5ec ]
Using dma_dev->dev for mappings before it's assigned with the correct
device is unlikely to work as expected, and with future dma-direct
changes, passing a NULL device may end up crashing entirely. I don't
know enough about this hardware or the mv_xor_prep_dma_interrupt()
operation to implement the appropriate error-handling logic that would
have revealed those dma_map_single() calls failing on arm64 for as long
as the driver has been enabled there, but moving the assignment earlier
will at least make the current code operate as intended.
Fixes: 22843545b2 ("dma: mv_xor: Add support for DMA_INTERRUPT")
Reported-by: John David Anglin <dave.anglin@bell.net>
Tested-by: John David Anglin <dave.anglin@bell.net>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Tested-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 627469e444 ]
The function coh901318_alloc_chan_resources() calls spin_lock_irqsave()
before calling coh901318_config().
But coh901318_config() calls spin_lock_irqsave() again in its
definition, which may cause a double-lock bug.
Because coh901318_config() is only called by
coh901318_alloc_chan_resources(), the bug fix is to remove the
calls to spin-lock and -unlock functions in coh901318_config().
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 5e621f5d53 ]
Clang warns when implicitly converting from one enumerated type to
another. Avoid this by using the equivalent value from the expected
type.
drivers/dma/timb_dma.c:548:27: warning: implicit conversion from
enumeration type 'enum dma_transfer_direction' to different enumeration
type 'enum dma_data_direction' [-Wenum-conversion]
td_desc->desc_list_len, DMA_MEM_TO_DEV);
^~~~~~~~~~~~~~
1 warning generated.
Reported-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 83ef4fb755 ]
Func jz4780_dma_desc_residue() expects the index to the next hw
descriptor as its last parameter. Caller func jz4780_dma_tx_status(),
however, applied modulus before passing it. When the current hw
descriptor was last in the list, the index passed became zero.
The resulting excess of reported residue especially caused problems
with cyclic DMA transfer clients, i.e. ALSA AIC audio output, which
rely on this for determining current DMA location within buffer.
Combined with the recent and related residue-reporting fixes, spurious
ALSA audio underruns on jz4770 hardware are now fixed.
Signed-off-by: Daniel Silsby <dansilsby@gmail.com>
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Tested-by: Mathieu Malaterre <malat@debian.org>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit c558ecd21c ]
If we make this driver depend on MACH_JZ4780, that means it can be
enabled only if we're building a kernel specially crafted for a
JZ4780-based board, while most GNU/Linux distributions will want one
generic MIPS kernel that works on multiple boards.
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 7667819385 upstream.
bam_dma_terminate_all() will leak resources if any of the transactions are
committed to the hardware (present in the desc fifo), and not complete.
Since bam_dma_terminate_all() does not cause the hardware to be updated,
the hardware will still operate on any previously committed transactions.
This can cause memory corruption if the memory for the transaction has been
reassigned, and will cause a sync issue between the BAM and its client(s).
Fix this by properly updating the hardware in bam_dma_terminate_all().
Fixes: e7c0fe2a5c ("dmaengine: add Qualcomm BAM dma driver")
Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@gmail.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20191017152606.34120-1-jeffrey.l.hugo@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit bacdcb6675 upstream.
Yegor Yefremov <yegorslists@googlemail.com> reported that musb and ftdi
uart can fail for the first open of the uart unless connected using
a hub.
This is because the first dma call done by musb_ep_program() must wait
if cppi41 is PM runtime suspended. Otherwise musb_ep_program() continues
with other non-dma packets before the DMA transfer is started causing at
least ftdi uarts to fail to receive data.
Let's fix the issue by waking up cppi41 with PM runtime calls added to
cppi41_dma_prep_slave_sg() and return NULL if still idled. This way we
have musb_ep_program() continue with PIO until cppi41 is awake.
Fixes: fdea2d09b9 ("dmaengine: cppi41: Add basic PM runtime support")
Reported-by: Yegor Yefremov <yegorslists@googlemail.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Cc: stable@vger.kernel.org # v4.9+
Link: https://lore.kernel.org/r/20191023153138.23442-1-tony@atomide.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 00c9755524 ]
When compile-testing on other architectures, we get lots of warnings
about incorrect format strings, like:
drivers/dma/iop-adma.c: In function 'iop_adma_alloc_slots':
drivers/dma/iop-adma.c:307:6: warning: format '%x' expects argument of type 'unsigned int', but argument 6 has type 'dma_addr_t {aka long long unsigned int}' [-Wformat=]
drivers/dma/iop-adma.c: In function 'iop_adma_prep_dma_memcpy':
>> drivers/dma/iop-adma.c:518:40: warning: format '%u' expects argument of type 'unsigned int', but argument 5 has type 'size_t {aka long unsigned int}' [-Wformat=]
Use %zu for printing size_t as required, and cast the dma_addr_t
arguments to 'u64' for printing with %llx. Ideally this should use
the %pad format string, but that requires an lvalue argument that
doesn't work here.
Link: https://lore.kernel.org/r/20190809163334.489360-3-arnd@arndb.de
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 5d6fb56072 ]
clang-9 points out that there are two variables that depending on the
configuration may only be used in an ARRAY_SIZE() expression but not
referenced:
drivers/dma/ste_dma40.c:145:12: error: variable 'd40_backup_regs' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration]
static u32 d40_backup_regs[] = {
^
drivers/dma/ste_dma40.c:214:12: error: variable 'd40_backup_regs_chan' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration]
static u32 d40_backup_regs_chan[] = {
Mark these __maybe_unused to shut up the warning.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20190712091357.744515-1-arnd@arndb.de
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 78efb76ab4 ]
While the .device_prep_slave_sg() callback rejects empty scatterlists,
it still accepts single-entry scatterlists with a zero-length segment.
These may happen if a driver calls dmaengine_prep_slave_single() with a
zero len parameter. The corresponding DMA request will never complete,
leading to messages like:
rcar-dmac e7300000.dma-controller: Channel Address Error happen
and DMA timeouts.
Although requesting a zero-length DMA request is a driver bug, rejecting
it early eases debugging. Note that the .device_prep_dma_memcpy()
callback already rejects requests to copy zero bytes.
Reported-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Analyzed-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 2b8066c3de ]
If probe() fails anywhere beyond the point where
sdma_get_firmware() is called, then a kernel oops may occur.
Problematic sequence of events:
1. probe() calls sdma_get_firmware(), which schedules the
firmware callback to run when firmware becomes available,
using the sdma instance structure as the context
2. probe() encounters an error, which deallocates the
sdma instance structure
3. firmware becomes available, firmware callback is
called with deallocated sdma instance structure
4. use after free - kernel oops !
Solution: only attempt to load firmware when we're certain
that probe() will succeed. This guarantees that the firmware
callback's context will remain valid.
Note that the remove() path is unaffected by this issue: the
firmware loader will increment the driver module's use count,
ensuring that the module cannot be unloaded while the
firmware callback is pending or running.
Signed-off-by: Sven Van Asbroeck <TheSven73@gmail.com>
Reviewed-by: Robin Gong <yibin.gong@nxp.com>
[vkoul: fixed braces for if condition]
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 3f93a4f297 upstream.
It is possible for an irq triggered by channel0 to be received later
after clks are disabled once firmware loaded during sdma probe. If
that happens then clearing them by writing to SDMA_H_INTR won't work
and the kernel will hang processing infinite interrupts. Actually,
don't need interrupt triggered on channel0 since it's pollling
SDMA_H_STATSTOP to know channel0 done rather than interrupt in
current code, just clear BD_INTR to disable channel0 interrupt to
avoid the above case.
This issue was brought by commit 1d069bfa3c ("dmaengine: imx-sdma:
ack channel 0 IRQ in the interrupt handler") which didn't take care
the above case.
Fixes: 1d069bfa3c ("dmaengine: imx-sdma: ack channel 0 IRQ in the interrupt handler")
Cc: stable@vger.kernel.org #5.0+
Signed-off-by: Robin Gong <yibin.gong@nxp.com>
Reported-by: Sven Van Asbroeck <thesven73@gmail.com>
Tested-by: Sven Van Asbroeck <thesven73@gmail.com>
Reviewed-by: Michael Olbrich <m.olbrich@pengutronix.de>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 5ba846b1ee ]
Intel IOMMU, when enabled, tries to find the domain of the device,
assuming it's a PCI one, during DMA operations, such as mapping or
unmapping. Since we are splitting the actual PCI device to couple of
children via MFD framework (see drivers/mfd/intel-lpss.c for details),
the DMA device appears to be a platform one, and thus not an actual one
that performs DMA. In a such situation IOMMU can't find or allocate
a proper domain for its operations. As a result, all DMA operations are
failed.
In order to fix this, supply parent of the platform device
to the DMA engine framework and fix filter functions accordingly.
We may rely on the fact that parent is a real PCI device, because no
other configuration is present in the wild.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mark Brown <broonie@kernel.org>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> [for tty parts]
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit f6ed6491d5 ]
adma driver is using pm_clk_*() interface for managing clock resources.
With this it is observed that clocks remain ON always. This happens on
Tegra devices which use BPMP co-processor to manage clock resources,
where clocks are enabled during prepare phase. This is necessary because
clocks to BPMP are always blocking. When pm_clk_*() interface is used on
such Tegra devices, clock prepare count is not balanced till remove call
happens for the driver and hence clocks are seen ON always. Thus this
patch replaces pm_clk_*() with devm_clk_*() framework.
Suggested-by: Mohan Kumar D <mkumard@nvidia.com>
Reviewed-by: Jonathan Hunter <jonathanh@nvidia.com>
Signed-off-by: Sameer Pujar <spujar@nvidia.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e2c114c06d ]
Even if this case shouldn't happen when controller is properly programmed,
it's still better to avoid dumping a kernel Oops for this.
As the sequence may happen only for debugging purposes, log the error and
just finish the tasklet call.
Signed-off-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Acked-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 2da254cc79 ]
This patch kill instructs the DMAC to immediately terminate
execution of a thread. and then clear the interrupt status,
at last, stop generating interrupts for DMA_SEV. to guarantee
the next dma start is clean. otherwise, one interrupt maybe leave
to next start and make some mistake.
we can reporduce the problem as follows:
DMASEV: modify the event-interrupt resource, and if the INTEN sets
function as interrupt, the DMAC will set irq<event_num> HIGH to
generate interrupt. write INTCLR to clear interrupt.
DMA EXECUTING INSTRUCTS DMA TERMINATE
| |
| |
... _stop
| |
| spin_lock_irqsave
DMASEV |
| |
| mask INTEN
| |
| DMAKILL
| |
| spin_unlock_irqrestore
in above case, a interrupt was left, and if we unmask INTEN, the DMAC
will set irq<event_num> HIGH to generate interrupt.
to fix this, do as follows:
DMA EXECUTING INSTRUCTS DMA TERMINATE
| |
| |
... _stop
| |
| spin_lock_irqsave
DMASEV |
| |
| DMAKILL
| |
| clear INTCLR
| mask INTEN
| |
| spin_unlock_irqrestore
Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit e486df3930 ]
The dma_desc->bytes_transferred counter tracks the number of bytes
moved by the DMA channel. This is then used to calculate the information
passed back in the in the tegra_dma_tx_status callback, which is usually
fine.
When the DMA channel is configured as continous, then the bytes_transferred
counter will increase over time and eventually overflow to become negative
so the residue count will become invalid and the ALSA sound-dma code will
report invalid hardware pointer values to the application. This results in
some users becoming confused about the playout position and putting audio
data in the wrong place.
To fix this issue, always ensure the bytes_transferred field is modulo the
size of the request. We only do this for the case of the cyclic transfer
done ISR as anyone attempting to move 2GiB of DMA data in one transfer
is unlikely.
Note, we don't fix the issue that we should /never/ transfer a negative
number of bytes so we could make those fields unsigned.
Reviewed-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Acked-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 546c054755 ]
When dma_cookie_complete() is called in hidma_process_completed(),
dma_cookie_status() will return DMA_COMPLETE in hidma_tx_status(). Then,
hidma_txn_is_success() will be called to use channel cookie
mchan->last_success to do additional DMA status check. Current code
assigns mchan->last_success after dma_cookie_complete(). This causes
a race condition of dma_cookie_status() returns DMA_COMPLETE before
mchan->last_success is assigned correctly. The race will cause
hidma_tx_status() return DMA_ERROR but the transaction is actually a
success. Moreover, in async_tx case, it will cause a timeout panic
in async_tx_quiesce().
Kernel panic - not syncing: async_tx_quiesce: DMA error waiting for
transaction
...
Call trace:
[<ffff000008089994>] dump_backtrace+0x0/0x1f4
[<ffff000008089bac>] show_stack+0x24/0x2c
[<ffff00000891e198>] dump_stack+0x84/0xa8
[<ffff0000080da544>] panic+0x12c/0x29c
[<ffff0000045d0334>] async_tx_quiesce+0xa4/0xc8 [async_tx]
[<ffff0000045d03c8>] async_trigger_callback+0x70/0x1c0 [async_tx]
[<ffff0000048b7d74>] raid_run_ops+0x86c/0x1540 [raid456]
[<ffff0000048bd084>] handle_stripe+0x5e8/0x1c7c [raid456]
[<ffff0000048be9ec>] handle_active_stripes.isra.45+0x2d4/0x550 [raid456]
[<ffff0000048beff4>] raid5d+0x38c/0x5d0 [raid456]
[<ffff000008736538>] md_thread+0x108/0x168
[<ffff0000080fb1cc>] kthread+0x10c/0x138
[<ffff000008084d34>] ret_from_fork+0x10/0x18
Cc: Joey Zheng <yu.zheng@hxt-semitech.com>
Reviewed-by: Sinan Kaya <okaya@kernel.org>
Signed-off-by: Shunyong Yang <shunyong.yang@hxt-semitech.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit 9227ab5643 ]
The warning got introduced by commit 930507c183 ("arm64: add basic
Kconfig symbols for i.MX8"). Since it got enabled for arm64. The warning
haven't been seen before since size_t was 'unsigned int' when built on
arm32.
../drivers/dma/imx-dma.c: In function ‘imxdma_sg_next’:
../include/linux/kernel.h:846:29: warning: comparison of distinct pointer types lacks a cast
(!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
^~
../include/linux/kernel.h:860:4: note: in expansion of macro ‘__typecheck’
(__typecheck(x, y) && __no_side_effects(x, y))
^~~~~~~~~~~
../include/linux/kernel.h:870:24: note: in expansion of macro ‘__safe_cmp’
__builtin_choose_expr(__safe_cmp(x, y), \
^~~~~~~~~~
../include/linux/kernel.h:879:19: note: in expansion of macro ‘__careful_cmp’
#define min(x, y) __careful_cmp(x, y, <)
^~~~~~~~~~~~~
../drivers/dma/imx-dma.c:288:8: note: in expansion of macro ‘min’
now = min(d->len, sg_dma_len(sg));
^~~
Rework so that we use min_t and pass in the size_t that returns the
minimum of two values, using the specified type.
Signed-off-by: Anders Roxell <anders.roxell@linaro.org>
Acked-by: Olof Johansson <olof@lixom.net>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
[ Upstream commit dc3f595b66 ]
atchan->status variable is used to store two different information:
- pass channel interrupts status from interrupt handler to tasklet;
- channel information like whether it is cyclic or paused;
This causes a bug when device_terminate_all() is called,
(AT_XDMAC_CHAN_IS_CYCLIC cleared on atchan->status) and then a late End
of Block interrupt arrives (AT_XDMAC_CIS_BIS), which sets bit 0 of
atchan->status. Bit 0 is also used for AT_XDMAC_CHAN_IS_CYCLIC, so when
a new descriptor for a cyclic transfer is created, the driver reports
the channel as in use:
if (test_and_set_bit(AT_XDMAC_CHAN_IS_CYCLIC, &atchan->status)) {
dev_err(chan2dev(chan), "channel currently used\n");
return NULL;
}
This patch fixes the bug by adding a different struct member to keep
the interrupts status separated from the channel status bits.
Fixes: e1f7c9eee7 ("dmaengine: at_xdmac: creation of the atmel eXtended DMA Controller driver")
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@microchip.com>
Acked-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 341198eda7 upstream.
Once the "ld_queue" list is not empty, next descriptor will migrate
into "ld_active" list. The "desc" variable will be overwritten
during that transition. And later the dmaengine_desc_get_callback_invoke()
will use it as an argument. As result we invoke wrong callback.
That behaviour was in place since:
commit fcaaba6c71 ("dmaengine: imx-dma: fix callback path in tasklet").
But after commit 4cd13c21b2 ("softirq: Let ksoftirqd do its job")
things got worse, since possible delay between tasklet_schedule()
from DMA irq handler and actual tasklet function execution got bigger.
And that gave more time for new DMA request to be submitted and
to be put into "ld_queue" list.
It has been noticed that DMA issue is causing problems for "mxc-mmc"
driver. While stressing the system with heavy network traffic and
writing/reading to/from sd card simultaneously the timeout may happen:
10013000.sdhci: mxcmci_watchdog: read time out (status = 0x30004900)
That often lead to file system corruption.
Signed-off-by: Leonid Iziumtsev <leonid.iziumtsev@gmail.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 9e528c799d upstream.
There are multiple issues with bcm2835_dma_abort() (which is called on
termination of a transaction):
* The algorithm to abort the transaction first pauses the channel by
clearing the ACTIVE flag in the CS register, then waits for the PAUSED
flag to clear. Page 49 of the spec documents the latter as follows:
"Indicates if the DMA is currently paused and not transferring data.
This will occur if the active bit has been cleared [...]"
https://www.raspberrypi.org/app/uploads/2012/02/BCM2835-ARM-Peripherals.pdf
So the function is entering an infinite loop because it is waiting for
PAUSED to clear which is always set due to the function having cleared
the ACTIVE flag. The only thing that's saving it from itself is the
upper bound of 10000 loop iterations.
The code comment says that the intention is to "wait for any current
AXI transfer to complete", so the author probably wanted to check the
WAITING_FOR_OUTSTANDING_WRITES flag instead. Amend the function
accordingly.
* The CS register is only read at the beginning of the function. It
needs to be read again after pausing the channel and before checking
for outstanding writes, otherwise writes which were issued between
the register read at the beginning of the function and pausing the
channel may not be waited for.
* The function seeks to abort the transfer by writing 0 to the NEXTCONBK
register and setting the ABORT and ACTIVE flags. Thereby, the 0 in
NEXTCONBK is sought to be loaded into the CONBLK_AD register. However
experimentation has shown this approach to not work: The CONBLK_AD
register remains the same as before and the CS register contains
0x00000030 (PAUSED | DREQ_STOPS_DMA). In other words, the control
block is not aborted but merely paused and it will be resumed once the
next DMA transaction is started. That is absolutely not the desired
behavior.
A simpler approach is to set the channel's RESET flag instead. This
reliably zeroes the NEXTCONBK as well as the CS register. It requires
less code and only a single MMIO write. This is also what popular
user space DMA drivers do, e.g.:
https://github.com/metachris/RPIO/blob/master/source/c_pwm/pwm.c
Note that the spec is contradictory whether the NEXTCONBK register
is writeable at all. On the one hand, page 41 claims:
"The value loaded into the NEXTCONBK register can be overwritten so
that the linked list of Control Block data structures can be
dynamically altered. However it is only safe to do this when the DMA
is paused."
On the other hand, page 40 specifies:
"Only three registers in each channel's register set are directly
writeable (CS, CONBLK_AD and DEBUG). The other registers (TI,
SOURCE_AD, DEST_AD, TXFR_LEN, STRIDE & NEXTCONBK), are automatically
loaded from a Control Block data structure held in external memory."
Fixes: 96286b5766 ("dmaengine: Add support for BCM2835")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org # v3.14+
Cc: Frank Pavlic <f.pavlic@kunbus.de>
Cc: Martin Sperl <kernel@martin.sperl.org>
Cc: Florian Meier <florian.meier@koalo.de>
Cc: Clive Messer <clive.m.messer@gmail.com>
Cc: Matthias Reichl <hias@horus.com>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Acked-by: Florian Kauer <florian.kauer@koalo.de>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit f7da7782ab upstream.
If IRQ handlers are threaded (either because CONFIG_PREEMPT_RT_BASE is
enabled or "threadirqs" was passed on the command line) and if system
load is sufficiently high that wakeup latency of IRQ threads degrades,
SPI DMA transactions on the BCM2835 occasionally break like this:
ks8851 spi0.0: SPI transfer timed out
bcm2835-dma 3f007000.dma: DMA transfer could not be terminated
ks8851 spi0.0 eth2: ks8851_rdfifo: spi_sync() failed
The root cause is an assumption made by the DMA driver which is
documented in a code comment in bcm2835_dma_terminate_all():
/*
* Stop DMA activity: we assume the callback will not be called
* after bcm_dma_abort() returns (even if it does, it will see
* c->desc is NULL and exit.)
*/
That assumption falls apart if the IRQ handler bcm2835_dma_callback() is
threaded: A client may terminate a descriptor and issue a new one
before the IRQ handler had a chance to run. In fact the IRQ handler may
miss an *arbitrary* number of descriptors. The result is the following
race condition:
1. A descriptor finishes, its interrupt is deferred to the IRQ thread.
2. A client calls dma_terminate_async() which sets channel->desc = NULL.
3. The client issues a new descriptor. Because channel->desc is NULL,
bcm2835_dma_issue_pending() immediately starts the descriptor.
4. Finally the IRQ thread runs and writes BCM2835_DMA_INT to the CS
register to acknowledge the interrupt. This clears the ACTIVE flag,
so the newly issued descriptor is paused in the middle of the
transaction. Because channel->desc is not NULL, the IRQ thread
finalizes the descriptor and tries to start the next one.
I see two possible solutions: The first is to call synchronize_irq()
in bcm2835_dma_issue_pending() to wait until the IRQ thread has
finished before issuing a new descriptor. The downside of this approach
is unnecessary latency if clients desire rapidly terminating and
re-issuing descriptors and don't have any use for an IRQ callback.
(The SPI TX DMA channel is a case in point.)
A better alternative is to make the IRQ thread recognize that it has
missed descriptors and avoid finalizing the newly issued descriptor.
So first of all, set the ACTIVE flag when acknowledging the interrupt.
This keeps a newly issued descriptor running.
If the descriptor was finished, the channel remains idle despite the
ACTIVE flag being set. However the ACTIVE flag can then no longer be
used to check whether the channel is idle, so instead check whether
the register containing the current control block address is zero
and finalize the current descriptor only if so.
That way, there is no impact on latency and throughput if the client
doesn't care for the interrupt: Only minimal additional overhead is
introduced for non-cyclic descriptors as one further MMIO read is
necessary per interrupt to check for idleness of the channel. Cyclic
descriptors are sped up slightly by removing one MMIO write per
interrupt.
Fixes: 96286b5766 ("dmaengine: Add support for BCM2835")
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Cc: stable@vger.kernel.org # v3.14+
Cc: Frank Pavlic <f.pavlic@kunbus.de>
Cc: Martin Sperl <kernel@martin.sperl.org>
Cc: Florian Meier <florian.meier@koalo.de>
Cc: Clive Messer <clive.m.messer@gmail.com>
Cc: Matthias Reichl <hias@horus.com>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Acked-by: Florian Kauer <florian.kauer@koalo.de>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit aeaebcc17c ]
Clang warns:
drivers/dma/xilinx/zynqmp_dma.c:166:4: warning: attribute 'aligned' is
ignored, place it after "struct" to apply attribute to type declaration
[-Wignored-attributes]
}; __aligned(64)
^
./include/linux/compiler_types.h:200:38: note: expanded from macro
'__aligned'
^
1 warning generated.
As Nick pointed out in the previous version of this patch, the author
likely intended for this struct to be 8-byte (64-bit) aligned, not
64-byte, which is the default. Remove the hanging __aligned attribute.
Fixes: b0cc417c16 ("dmaengine: Add Xilinx zynqmp dma engine driver support")
Reported-by: Nick Desaulniers <ndesaulniers@google.com>
Suggested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
commit 59861547ec upstream.
The driver defines three states for a cppi channel.
- idle: .chan_busy == 0 && not in .pending list
- pending: .chan_busy == 0 && in .pending list
- busy: .chan_busy == 1 && not in .pending list
There are cases in which the cppi channel could be in the pending state
when cppi41_dma_issue_pending() is called after cppi41_runtime_suspend()
is called.
cppi41_stop_chan() has a bug for these cases to set channels to idle state.
It only checks the .chan_busy flag, but not the .pending list, then later
when cppi41_runtime_resume() is called the channels in .pending list will
be transitioned to busy state.
Removing channels from the .pending list solves the problem.
Fixes: 975faaeb99 ("dma: cppi41: start tear down only if channel is busy")
Cc: stable@vger.kernel.org # v3.15+
Signed-off-by: Bin Liu <b-liu@ti.com>
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 77e75fda94 upstream.
of_dma_controller_free() was not called on module onloading.
This lead to a soft lockup:
watchdog: BUG: soft lockup - CPU#0 stuck for 23s!
Modules linked in: at_hdmac [last unloaded: at_hdmac]
when of_dma_request_slave_channel() tried to call ofdma->of_dma_xlate().
Cc: stable@vger.kernel.org
Fixes: bbe89c8e3d ("at_hdmac: move to generic DMA binding")
Acked-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
commit 98f5f93225 upstream.
The leak was found when opening/closing a serial port a great number of
time, increasing kmalloc-32 in slabinfo.
Each time the port was opened, dma_request_slave_channel() was called.
Then, in at_dma_xlate(), atslave was allocated with devm_kzalloc() and
never freed. (Well, it was free at module unload, but that's not what we
want).
So, here, kzalloc is more suited for the job since it has to be freed in
atc_free_chan_resources().
Cc: stable@vger.kernel.org
Fixes: bbe89c8e3d ("at_hdmac: move to generic DMA binding")
Reported-by: Mario Forner <m.forner@be4energy.com>
Suggested-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Acked-by: Alexandre Belloni <alexandre.belloni@bootlin.com>
Acked-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 54f919a04c ]
The driver calls clk_get() with the clock name set to NULL, which means
that the driver could only work when probed from devicetree. From now
on, we explicitly require the driver to be probed from devicetree.
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Tested-by: Mathieu Malaterre <malat@debian.org>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit cfb03be6c7 ]
The following lockdep splat was observed:
[ 1222.241750] ======================================================
[ 1222.271301] WARNING: possible circular locking dependency detected
[ 1222.301060] 4.16.0-10.el8+5.x86_64+debug #1 Not tainted
[ 1222.326659] ------------------------------------------------------
[ 1222.356565] systemd-shutdow/1 is trying to acquire lock:
[ 1222.382660] ((&ioat_chan->timer)){+.-.}, at: [<00000000f71e1a28>] del_timer_sync+0x5/0xf0
[ 1222.422928]
[ 1222.422928] but task is already holding lock:
[ 1222.451743] (&(&ioat_chan->prep_lock)->rlock){+.-.}, at: [<000000008ea98b12>] ioat_shutdown+0x86/0x100 [ioatdma]
:
[ 1223.524987] Chain exists of:
[ 1223.524987] (&ioat_chan->timer) --> &(&ioat_chan->cleanup_lock)->rlock --> &(&ioat_chan->prep_lock)->rlock
[ 1223.524987]
[ 1223.594082] Possible unsafe locking scenario:
[ 1223.594082]
[ 1223.622630] CPU0 CPU1
[ 1223.645080] ---- ----
[ 1223.667404] lock(&(&ioat_chan->prep_lock)->rlock);
[ 1223.691535] lock(&(&ioat_chan->cleanup_lock)->rlock);
[ 1223.728657] lock(&(&ioat_chan->prep_lock)->rlock);
[ 1223.765122] lock((&ioat_chan->timer));
[ 1223.784095]
[ 1223.784095] *** DEADLOCK ***
[ 1223.784095]
[ 1223.813492] 4 locks held by systemd-shutdow/1:
[ 1223.834677] #0: (reboot_mutex){+.+.}, at: [<0000000056d33456>] SYSC_reboot+0x10f/0x300
[ 1223.873310] #1: (&dev->mutex){....}, at: [<00000000258dfdd7>] device_shutdown+0x1c8/0x660
[ 1223.913604] #2: (&dev->mutex){....}, at: [<0000000068331147>] device_shutdown+0x1d6/0x660
[ 1223.954000] #3: (&(&ioat_chan->prep_lock)->rlock){+.-.}, at: [<000000008ea98b12>] ioat_shutdown+0x86/0x100 [ioatdma]
In the ioat_shutdown() function:
spin_lock_bh(&ioat_chan->prep_lock);
set_bit(IOAT_CHAN_DOWN, &ioat_chan->state);
del_timer_sync(&ioat_chan->timer);
spin_unlock_bh(&ioat_chan->prep_lock);
According to the synchronization rule for the del_timer_sync() function,
the caller must not hold locks which would prevent completion of the
timer's handler.
The timer structure has its own lock that manages its synchronization.
Setting the IOAT_CHAN_DOWN bit should prevent other CPUs from
trying to use that device anyway, there is probably no need to call
del_timer_sync() while holding the prep_lock. So the del_timer_sync()
call is now moved outside of the prep_lock critical section to prevent
the circular lock dependency.
Signed-off-by: Waiman Long <longman@redhat.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
