- 12 Apr, 2023 27 commits
-
-
Dave Jiang authored
Add counters per opened file for the char device in order to keep track how many completion record faults occurred and how many of those faults failed the writeback by the driver after attempt to fault in the page. The counters are managed by xarray that associates the PASID with struct idxd_user_context. Tested-by:
Tony Zhu <tony.zhu@intel.com> Signed-off-by:
Dave Jiang <dave.jiang@intel.com> Co-developed-by:
Fenghua Yu <fenghua.yu@intel.com> Signed-off-by:
Vinod Koul <vkoul@kernel.org>
-
Dave Jiang authored
Add event log processing for faulting of user batch descriptor completion record. When encountering an event log entry for a page fault on a completion record, the driver is expected to do the following: 1. If the "first error in batch" bit in event log entry error info is set, discard any previously recorded errors associated with the "batch identifier". 2. Fix the page fault according to the fault address in the event log. If successful, write the completion record to the fault address in user space. 3. If an error is encountered while writing the completion record and it is associated to a descriptor in the batch, the driver associates the error with the batch identifier of the event log entry and tracks it until the event log entry for the corresponding batch desc is encountered. While processing an event log entry for a batch descriptor with error indicating that one or more descs in the batch had event log entries, the driver will do the following before writing the batch completion record: 1. If the status field of the completion record is 0x1, the driver will change it to error code 0x5 (one or more operations in batch completed with status not successful) and changes the result field to 1. 2. If the status is error code 0x6 (page fault on batch descriptor list address), change the result field to 1. 3. If status is any other value, the completion record is not changed. 4. Clear the recorded error in preparation for next batch with same batch identifier. The result field is for user software to determine whether to set the "Batch Error" flag bit in the descriptor for continuation of partial batch descriptor completion. See DSA spec 2.0 for additional information. If no error has been recorded for the batch, the batch completion record is written to user space as is. Tested-by:
-
Dave Jiang authored
The descs_completed field for a completion record is part of a batch descriptor completion record. It takes the same location as bytes_completed in a normal descriptor field. Add to expose to user. Tested-by:
-
Dave Jiang authored
DSA supports page fault handling through PRS. However, the DMA engine that's processing the descriptor is blocked until the PRS response is received. Other workqueues sharing the engine are also blocked. Page fault handing by the driver with PRS disabled can be used to mitigate the stalling. With PRS disabled while ATS remain enabled, DSA handles page faults on a completion record by reporting an event in the event log. In this instance, the descriptor is completed and the event log contains the completion record address and the contents of the completion record. Add support to the event log handling code to fault in the completion record and copy the content of the completion record to user memory. A bitmap is introduced to keep track of discarded event log entries. When the user process initiates ->release() of the char device, it no longer is interested in any remaining event log entries tied to the relevant wq and PASID. The driver will mark the event log entry index in the bitmap. Upon encountering the entries during processing, the event log handler will just clear the bitmap bit and skip the entry rather than attempt to process the event log entry. Tested-by:
-
Fenghua Yu authored
Define idxd_copy_cr() to copy completion record to fault address in user address that is found by work queue (wq) and PASID. It will be used to write the user's completion record that the hardware device is not able to write due to user completion record page fault. An xarray is added to associate the PASID and mm with the struct idxd_user_context so mm can be found by PASID and wq. It is called when handling the completion record fault in a kernel thread context. Switch to the mm using kthread_use_vm() and copy the completion record to the mm via copy_to_user(). Once the copy is completed, switch back to the current mm using kthread_unuse_mm(). Suggested-by:
Christoph Hellwig <hch@infradead.org> Suggested-by:
Jason Gunthorpe <jgg@nvidia.com> Suggested-by:
Tony Luck <tony.luck@intel.com> Tested-by:
-
Dave Jiang authored
Add a kmem cache per device for allocating event log fault context. The context allows an event log entry to be copied and passed to a software workqueue to be processed. Due to each device can have different sized event log entry depending on device type, it's not possible to have a global kmem cache. Tested-by:
-
Dave Jiang authored
Add a workqueue for user submitted completion record fault processing. The workqueue creation and destruction lifetime will be tied to the user sub-driver since it will only be used when the wq is a user type. Tested-by:
-
Dave Jiang authored
Add debugfs entry to dump the content of the event log for debugging. The function will dump all non-zero entries in the event log. It will note which entries are processed and which entries are still pending processing at the time of the dump. The entries may not always be in chronological order due to the log is a circular buffer. Tested-by:
-
Dave Jiang authored
An event log interrupt is raised in the misc interrupt INTCAUSE register when an event is written by the hardware. Add basic event log processing support to the interrupt handler. The event log is a ring where the hardware owns the tail and the software owns the head. The hardware will advance the tail index when an additional event has been pushed to memory. The software will process the log entry and then advances the head. The log is full when (tail + 1) % log_size = head. The hardware will stop writing when the log is full. The user is expected to create a log size large enough to handle all the expected events. Tested-by:
-
Dave Jiang authored
Add setup of event log feature for supported device. Event log addresses error reporting that was lacking in gen 1 DSA devices where a second error event does not get reported when a first event is pending software handling. The event log allows a circular buffer that the device can push error events to. It is up to the user to create a large enough event log ring in order to capture the expected events. The evl size can be set in the device sysfs attribute. By default 64 entries are supported as minimal when event log is enabled. Tested-by:
-
Dave Jiang authored
Add support for changing of the event log size. Event log is a feature added to DSA 2.0 hardware to improve error reporting. It supersedes the SWERROR register on DSA 1.0 hardware and hope to prevent loss of reported errors. The error log size determines how many error entries supported for the device. It can be configured by the user via sysfs attribute. Tested-by:
-
Dave Jiang authored
Current code continuously processes the interrupt as long as the hardware is setting the status bit. There's no reason to do that since the threaded handler will get called again if another interrupt is asserted. Also through testing, it has shown that if a misprogrammed (or malicious) agent can continuously submit descriptors with bad completion record and causes errors to be reported via the misc interrupt. Continuous processing by the thread can cause software hang watchdog to kick off since the thread isn't giving up the CPU. Reported-by:
Sanjay Kumar <sanjay.k.kumar@intel.com> Tested-by:
-
Walker Chen authored
The DMA controller needs two reset items to work properly on JH7110 SoC, so there is need to constrain the items' value to 2, other platforms have 1 reset item at most. Reviewed-by:
Rob Herring <robh@kernel.org> Signed-off-by:
Walker Chen <walker.chen@starfivetech.com> Link: https://lore.kernel.org/r/20230322094820.24738-2-walker.chen@starfivetech.comSigned-off-by:
-
Rob Herring authored
Cleanup bindings dropping unneeded quotes. Once all these are fixed, checking for this can be enabled in yamllint. Signed-off-by:
Peter Ujfalusi <peter.ujfalusi@gmail.com> Reviewed-by:
Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org> Link: https://lore.kernel.org/r/20230331182141.1900348-1-robh@kernel.orgSigned-off-by:
-
Claudiu Beznea authored
Align the declaration of ret in atmel_xdmac_resume() with the rest of variables. Do this by adding ret to the line with declaration for i variable. Signed-off-by:
Claudiu Beznea <claudiu.beznea@microchip.com> Link: https://lore.kernel.org/r/20230214151827.1050280-8-claudiu.beznea@microchip.comSigned-off-by:
-
Claudiu Beznea authored
Add a warning message on suspend to let the user that there are channels not paused by their consumers. Signed-off-by:
-
Claudiu Beznea authored
Do not global enable all the cyclic channels in at_xdmac_resume(). Instead save the global status in at_xdmac_suspend() and re-enable the cyclic channel only if it was active before suspend. Fixes: e1f7c9ee ("dmaengine: at_xdmac: creation of the atmel eXtended DMA Controller driver") Signed-off-by:
-
Claudiu Beznea authored
In case the system suspends to a deep sleep state where power to DMA controller is cut-off we need to restore the content of GRWS register. This is a write only register and writing bit X tells the controller to suspend read and write requests for channel X. Thus set GRWS before restoring the content of GE (Global Enable) regiter. Fixes: e1f7c9ee ("dmaengine: at_xdmac: creation of the atmel eXtended DMA Controller driver") Signed-off-by:
-
Claudiu Beznea authored
In case there are DMA channels not paused by consumers in suspend process (valid on AT91 SoCs for serial driver when no_console_suspend) the driver pauses them (using at_xdmac_device_pause() which is also the same function called by dmaengine_pause()) and then in the resume process the driver resumes them calling at_xdmac_device_resume() which is the same function called by dmaengine_resume()). This is good for DMA channels not paused by consumers but for drivers that calls dmaengine_pause()/dmaegine_resume() on suspend/resume path this may lead to DMA channel being enabled before the IP is enabled. For IPs that needs strict ordering with regards to DMA channel enablement this will lead to wrong behavior. To fix this add a new set of functions at_xdmac_device_pause_internal()/at_xdmac_device_resume_internal() to be called only on suspend/resume. Fixes: e1f7c9ee ("dmaengine: at_xdmac: creation of the atmel eXtended DMA Controller driver") Signed-off-by:
-
Claudiu Beznea authored
In case there are channels not paused during suspend (which on AT91 case is valid for serial driver when no_console_suspend boot argument is used) the at_xdmac_runtime_suspend_descriptors() was called more than one time due to at_xdmac_off(). To fix this add a new argument to at_xdmac_off() to specify if runtime PM reference counter needs to be decremented for queued active descriptors. Along with it moved the at_xdmac_runtime_suspend_descriptors() call under at_xdmac_chan_is_paused() check on suspend path as for the rest of channels the suspend is delayed by atmel_xdmac_prepare() in case channel is enabled. Same approach has been applied on resume path. Fixes: 650b0e99 ("dmaengine: at_xdmac: add runtime pm support") Signed-off-by:
-
Claudiu Beznea authored
Runtime PM APIs for at_xdmac just plays with clk_enable()/clk_disable() letting aside the clk_prepare()/clk_unprepare() that needs to be executed as the clock is also prepared on probe. Thus instead of using runtime PM force suspend/resume APIs use clk_disable_unprepare() + pm_runtime_put_noidle() on suspend and clk_prepare_enable() + pm_runtime_get_noresume() on resume. This approach as been chosen instead of using runtime PM force suspend/resume with clk_unprepare()/clk_prepare() as it looks simpler and the final code is better. While at it added the missing pm_runtime_mark_last_busy() on suspend before decrementing the reference counter. Fixes: 650b0e99 ("dmaengine: at_xdmac: add runtime pm support") Signed-off-by:
-
Walker Chen authored
The bit DMAC_CHEN[0] is automatically cleared by hardware to disable the channel after the last AMBA transfer of the DMA transfer to the destination has completed. Software can therefore poll this bit to determine when this channel is free for a new DMA transfer. This time requires at least 40 milliseconds on JH7110 SoC, otherwise an error message 'failed to stop' will be reported. Signed-off-by:
-
Walker Chen authored
Add DMA reset operation in device probe and use different configuration on CH_CFG registers according to match data. Update all uses of of_device_is_compatible with of_device_get_match_data. Signed-off-by:
Emil Renner Berthing <emil.renner.berthing@canonical.com> Link: https://lore.kernel.org/r/20230322094820.24738-3-walker.chen@starfivetech.comSigned-off-by:
-
Shunsuke Mie authored
The issue_pending request is ignored while driver is processing a DMA request. Fix to issue the pending requests on any dma channel status. Fixes: e63d79d1 ("dmaengine: Add Synopsys eDMA IP core driver") Signed-off-by:
Shunsuke Mie <mie@igel.co.jp> Link: https://lore.kernel.org/r/20230411101758.438472-2-mie@igel.co.jpSigned-off-by:
-
Shunsuke Mie authored
The dw-edma driver stops after processing a DMA request even if a request remains in the issued queue, which is not the expected behavior. The DMA engine API requires continuous processing. Add a trigger to start after one processing finished if there are requests remain. Fixes: e63d79d1 ("dmaengine: Add Synopsys eDMA IP core driver") Signed-off-by:
-
Rob Herring authored
qcom_hidma uses of_dma_configure() which is declared in of_device.h. platform_device.h and of_device.h get implicitly included by of_platform.h, but that is going to be removed soon. Signed-off-by:
Sinan Kaya <okaya@kernel.org> Link: https://lore.kernel.org/r/20230410232654.1561462-1-robh@kernel.orgSigned-off-by:
-
Dmitry Baryshkov authored
gpi_ch_init() doesn't lock the ctrl_lock mutex, so there is no need to unlock it too. Instead the mutex is handled by the function gpi_alloc_chan_resources(), which properly locks and unlocks the mutex. ===================================== WARNING: bad unlock balance detected! 6.3.0-rc5-00253-g99792582ded1-dirty #15 Not tainted ------------------------------------- kworker/u16:0/9 is trying to release lock (&gpii->ctrl_lock) at: [<ffffb99d04e1284c>] gpi_alloc_chan_resources+0x108/0x5bc but there are no more locks to release! other info that might help us debug this: 6 locks held by kworker/u16:0/9: #0: ffff575740010938 ((wq_completion)events_unbound){+.+.}-{0:0}, at: process_one_work+0x220/0x594 #1: ffff80000809bdd0 (deferred_probe_work){+.+.}-{0:0}, at: process_one_work+0x220/0x594 #2: ffff575740f2a0f8 (&dev->mutex){....}-{3:3}, at: __device_attach+0x38/0x188 #3: ffff57574b5570f8 (&dev->mutex){....}-{3:3}, at: __device_attach+0x38/0x188 #4: ffffb99d06a2f180 (of_dma_lock){+.+.}-{3:3}, at: of_dma_request_slave_channel+0x138/0x280 #5: ffffb99d06a2ee20 (dma_list_mutex){+.+.}-{3:3}, at: dma_get_slave_channel+0x28/0x10c stack backtrace: CPU: 7 PID: 9 Comm: kworker/u16:0 Not tainted 6.3.0-rc5-00253-g99792582ded1-dirty #15 Hardware name: Google Pixel 3 (DT) Workqueue: events_unbound deferred_probe_work_func Call trace: dump_backtrace+0xa0/0xfc show_stack+0x18/0x24 dump_stack_lvl+0x60/0xac dump_stack+0x18/0x24 print_unlock_imbalance_bug+0x130/0x148 lock_release+0x270/0x300 __mutex_unlock_slowpath+0x48/0x2cc mutex_unlock+0x20/0x2c gpi_alloc_chan_resources+0x108/0x5bc dma_chan_get+0x84/0x188 dma_get_slave_channel+0x5c/0x10c gpi_of_dma_xlate+0x110/0x1a0 of_dma_request_slave_channel+0x174/0x280 dma_request_chan+0x3c/0x2d4 geni_i2c_probe+0x544/0x63c platform_probe+0x68/0xc4 really_probe+0x148/0x2ac __driver_probe_device+0x78/0xe0 driver_probe_device+0x3c/0x160 __device_attach_driver+0xb8/0x138 bus_for_each_drv+0x84/0xe0 __device_attach+0x9c/0x188 device_initial_probe+0x14/0x20 bus_probe_device+0xac/0xb0 device_add+0x60c/0x7d8 of_device_add+0x44/0x60 of_platform_device_create_pdata+0x90/0x124 of_platform_bus_create+0x15c/0x3c8 of_platform_populate+0x58/0xf8 devm_of_platform_populate+0x58/0xbc geni_se_probe+0xf0/0x164 platform_probe+0x68/0xc4 really_probe+0x148/0x2ac __driver_probe_device+0x78/0xe0 driver_probe_device+0x3c/0x160 __device_attach_driver+0xb8/0x138 bus_for_each_drv+0x84/0xe0 __device_attach+0x9c/0x188 device_initial_probe+0x14/0x20 bus_probe_device+0xac/0xb0 deferred_probe_work_func+0x8c/0xc8 process_one_work+0x2bc/0x594 worker_thread+0x228/0x438 kthread+0x108/0x10c ret_from_fork+0x10/0x20 Fixes: 5d0c3533 ("dmaengine: qcom: Add GPI dma driver") Signed-off-by:Dmitry Baryshkov <dmitry.baryshkov@linaro.org> Link: https://lore.kernel.org/r/20230409233355.453741-1-dmitry.baryshkov@linaro.orgSigned-off-by:
-
- 31 Mar, 2023 13 commits
-
-
Apurva Nandan authored
Add psil thread IDs for J784s4 and include J784s4 in the set of "k3_soc_devices" in k3-psil.c. Signed-off-by:
Apurva Nandan <a-nandan@ti.com> Signed-off-by:
Hari Nagalla <hnagalla@ti.com> Signed-off-by:
Siddharth Vadapalli <s-vadapalli@ti.com> Signed-off-by:
Vignesh Raghavendra <vigneshr@ti.com> [vaishnav.a@ti.com: add MCSPI-TX and 3rd CSI2RX instance entries] Signed-off-by:
Vaishnav Achath <vaishnav.a@ti.com> [j-choudhary@ti.com: add sa2ul entries, minor cleanups] Signed-off-by:
Jayesh Choudhary <j-choudhary@ti.com> Acked-by:
-
Rob Herring authored
Replace of_get_address() and of_translate_address() calls with single call to of_address_to_resource(). Signed-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/dma/dw-edma/dw-edma-v0-core.c:162:20: error: unused function 'writeq_ch' [-Werror,-Wunused-function] static inline void writeq_ch(struct dw_edma *dw, enum dw_edma_dir dir, u16 ch, ^ drivers/dma/dw-edma/dw-edma-v0-core.c:185:19: error: unused function 'readq_ch' [-Werror,-Wunused-function] static inline u64 readq_ch(struct dw_edma *dw, enum dw_edma_dir dir, u16 ch, ^ These functions and their wrapping macros are not used, so remove them. Signed-off-by:Tom Rix <trix@redhat.com> Link: https://lore.kernel.org/r/20230320234906.1730308-1-trix@redhat.comSigned-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/dma/ti/edma.c:321:20: error: unused function 'edma_and' [-Werror,-Wunused-function] static inline void edma_and(struct edma_cc *ecc, int offset, unsigned and) ^ This function is not used, so remove it. Signed-off-by: -
Harini Katakam authored
Align ZDMA DMA as well as coherent memory masks to 44 bit. This is required when using >32 bit memory regions. Signed-off-by:
Harini Katakam <harini.katakam@amd.com> Acked-by:
Michal Simek <michal.simek@amd.com> Link: https://lore.kernel.org/r/20230316093318.6722-1-harini.katakam@amd.comSigned-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/dma/tegra20-apb-dma.c:236:19: error: unused function 'tdma_read' [-Werror,-Wunused-function] static inline u32 tdma_read(struct tegra_dma *tdma, u32 reg) ^ This function is not used so remove it. Signed-off-by:Jon Hunter <jonathanh@nvidia.com> Link: https://lore.kernel.org/r/20230322121001.2569909-1-trix@redhat.comSigned-off-by:
-
Vignesh Raghavendra authored
Per [1], UDMA TR15 transactions may hang if ICNT0 is less than 64B Work around is to set EOL flag is to 1 for ICNT0. Since, there is no performance penalty / side effects of setting EOL flag event ICNTO > 64B, just set the flag for all UDMAP TR15 descriptors. [1] https://www.ti.com/lit/er/sprz455a/sprz455a.pdf Errata doc for J721E DRA829/TDA4VM Processors Silicon Revision 1.1/1.0 (Rev. A) Signed-off-by:
-
Geert Uytterhoeven authored
The src_word_size and dst_word_size members of the rz_dmac_chan structure were never used, so they can be removed. Signed-off-by:
Geert Uytterhoeven <geert+renesas@glider.be> Reviewed-by:
Biju Das <biju.das.jz@bp.renesas.com> Link: https://lore.kernel.org/r/021bdf56f1716276a55bcfb1ea81bba5f1d42b3d.1679910274.git.geert+renesas@glider.beSigned-off-by:
-
Dave Jiang authored
Add IAA (IAX) capability mask sysfs attribute to expose to applications. The mask provides application knowledge of what capabilities this IAA device supports. This mask is available for IAA 2.0 device or later. Signed-off-by:
-
Dave Jiang authored
SWERROR register is 4 64bit wide registers. Currently the sysfs attribute just outputs 4 64bit hex integers. Convert to output with %*pb format specifier. Signed-off-by:
-
Fenghua Yu authored
The translation fetch operation (0x0A) fetches address translations for the address range specified in the descriptor by issuing address translation (ATS) requests to the IOMMU. Add descriptor definitions for the operation so that user can use DSA to accelerate translation fetch. Signed-off-by:
-
Fenghua Yu authored
The Data Integrity Extension (DIX) generate operation (0x17) computes the Data Integrity Field (DIF) on the source data and writes only the computed DIF for each source block to the PI destination address. Add descriptor definitions for this operation so that user can use DSA to accelerate DIX generate operation. Signed-off-by:
-
Fenghua Yu authored
The memory fill operation (0x04) can fill in memory with either 8 bytes or 16 bytes of pattern. To fill in memory with 16 bytes of pattern, the first 8 bytes are provided in pattern lower in bytes 16-23 and the next 8 bytes are in pattern upper in bytes 40-47 in the descriptor. Currently only 8 bytes of pattern is enabled. Add descriptor definitions for pattern lower and pattern upper so that user can use 16 bytes of pattern to fill memory. Signed-off-by:
-
