- 29 Aug, 2021 21 commits
-
-
Ofir Bitton authored
State dump is relevant to the user in case of Sync Manager error, so we need to trigger it in that case as well. Signed-off-by:
Ofir Bitton <obitton@habana.ai> Reviewed-by:
Oded Gabbay <ogabbay@kernel.org> Signed-off-by:
-
Oded Gabbay authored
This is required from any device that is capable to perform DMA. Signed-off-by: -
Oded Gabbay authored
Each ASIC can have a different offset to add to a host dma address, to enable the ASIC to access that host memory. The usage for this can be common code so add this to the asic property structure. Signed-off-by: -
Oded Gabbay authored
Recently, the size parameter in userptr structure was change to u64. As a result, we need to change the type of the local range_size in device_va_to_pa() to u64 to avoid overflow. Signed-off-by: -
Tomer Tayar authored
If hard reset fails after the call to hw_fini and before loading the linux image to the device, a subsequent call to hw_fini should communicate via COMMS (or MSG_TO_CPU regs for old FW versions). However, the driver still tries in this case to communicate via the GIC, and thus no hard reset is actually done. To avoid that, the patch clears the linux_loaded flag after every call to hw_fini. Signed-off-by:
Tomer Tayar <ttayar@habana.ai> Reviewed-by:
-
Tomer Tayar authored
In case of host-resident MMU, when the page tables pool is destroyed, its pointer is not nullified correctly. As a result, on a device fini which happens after a failing reset, the already destroyed pool is accessed, which leads to a kernel panic. The patch fixes the setting of the pool pointer to NULL. Signed-off-by:
-
Zvika Yehudai authored
This function will be used for more mmap operations than just mmaping CBs. Signed-off-by:
Zvika Yehudai <zyehudai@habana.ai> Reviewed-by:
-
Ofir Bitton authored
missing mutex unlock once driver is giving up killing user processes. Signed-off-by:
-
Yuri Nudelman authored
At the first stage, only gaudi core dump shall be implemented, not including the status registers. Signed-off-by:
Yuri Nudelman <ynudelman@habana.ai> Reviewed-by:
-
Yuri Nudelman authored
With the infrastructure in place, monitors and fences dump shall be implemented. Signed-off-by:
-
Yuri Nudelman authored
To improve the user's ability to debug the case where a workload that is part of executing training/inference of a topology is getting stuck, we need to add a 'core dump' each time a CS times-out. The 'core dump' shall contain all relevant Sync Manager information and corresponding fence values. The most recent dumps shall be accessible via debugfs, under 'state_dump' node. Reading from the node will provide the oldest dump available. Writing an integer value X will discard X dumps, starting with the oldest one, i.e. subsequent read will now return newer dumps. Signed-off-by:
-
Oded Gabbay authored
The previous function we used, find_get_pid(), wasn't good in case the user process was run inside docker. As a result, we didn't had the PID and we couldn't kill the user process in case the device got stuck and we needed to reset the device. Signed-off-by: -
Oded Gabbay authored
Sometimes we may need to disable optimization of using huge pages in our memory management code. Add such a flag to the function that creates the list of physical pages that would be programmed into the device MMU. Signed-off-by: -
Oded Gabbay authored
Increase the size variable in the userptr structure to 64-bit. That variable describes the size of the memory allocation of the user that is now being mapped into the device. The mapping can be larger than 4GB, so we need to support it. Signed-off-by: -
Oded Gabbay authored
Same as we handle it in the regular wait for CS, we need to handle the case where the waiting for user interrupt was interrupted. In that case, we need to return correct error code to the user. Signed-off-by: -
Oded Gabbay authored
In device fini there was missing a call to release all pending user interrupts. That can cause a process to be stuck inside the driver's IOCTL of wait for interrupts, in case the device is removed or simulator is killed at the same time. In addition, also call to remove inactive codec job was missing. Moreover, to prevent such errors in the future (where code is added to reset path but not to device fini), we moved some common parts to two dedicated functions: cleanup_resources take_release_locks Signed-off-by: -
Oded Gabbay authored
In case user interrupt arrived but the completion value is less than the target value, we want to retry the wait. However, before the retry we must reinitialize the completion object, under spin-lock, so the wait function won't exit immediately because the completion object is already completed (from the previous interrupt). Signed-off-by: -
Oded Gabbay authored
We don't use typedefs so the enum name shouldn't end with _t Signed-off-by: -
Ofir Bitton authored
Update recent changes made in firmware header files, which contain a minor COMMS protocol change and new error status definitions. Signed-off-by:
-
Yuri Nudelman authored
A new user flag is required to make memory map hint mandatory, in contrast to the current situation where it is best effort. This is due to the requirement to map certain data to specific pre-determined device virtual address ranges. Signed-off-by:
-
farah kassabri authored
Add support for pre-determined driver-reserved device VA address ranges. This is needed for future ASIC support where some contents must be mapped into these pre-determined ranges because the H/W will be configured using these ranges. In case the user asks to map a VA without a hint address, avoid allocating the device VA from the reserved ranges. Make sure the validation checks of the hint address take into account situation where the DRAM page size is not pow of 2. Signed-off-by:
farah kassabri <fkassabri@habana.ai> Reviewed-by:
-
- 27 Aug, 2021 5 commits
-
-
Greg Kroah-Hartman authored
This reverts commit 0092a1e3 This should be reverted in the char-misc-next branch to make merging with Linus's branch possible due to issues with the mhi code that was found in the networking tree. Link: https://lore.kernel.org/r/20210827175852.GB15018@thinkpadReported-by:
Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Reported-by:
Stephen Rothwell <sfr@canb.auug.org.au> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Bhaumik Bhatt <bbhatt@codeaurora.org> Cc: Hemant Kumar <hemantk@codeaurora.org> Cc: Jakub Kicinski <kuba@kernel.org> Cc: Kalle Valo <kvalo@codeaurora.org> Cc: Loic Poulain <loic.poulain@linaro.org> Cc: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Mihai Carabas authored
Add again dev_set_drvdata(), but this time in devm_pvpanic_probe(), in order for dev_get_drvdata() to not return NULL. Fixes: 394febc9 ("misc/pvpanic: Make 'pvpanic_probe()' resource managed") Reviewed-by:
Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by:
Mihai Carabas <mihai.carabas@oracle.com> Link: https://lore.kernel.org/r/1629385946-4584-2-git-send-email-mihai.carabas@oracle.comSigned-off-by:
-
Wang Hai authored
I got a NULL pointer dereference report when doing fuzz test: Call Trace: qp_release_pages+0xae/0x130 qp_host_unregister_user_memory.isra.25+0x2d/0x80 vmci_qp_broker_unmap+0x191/0x320 ? vmci_host_do_alloc_queuepair.isra.9+0x1c0/0x1c0 vmci_host_unlocked_ioctl+0x59f/0xd50 ? do_vfs_ioctl+0x14b/0xa10 ? tomoyo_file_ioctl+0x28/0x30 ? vmci_host_do_alloc_queuepair.isra.9+0x1c0/0x1c0 __x64_sys_ioctl+0xea/0x120 do_syscall_64+0x34/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae When a queue pair is created by the following call, it will not register the user memory if the page_store is NULL, and the entry->state will be set to VMCIQPB_CREATED_NO_MEM. vmci_host_unlocked_ioctl vmci_host_do_alloc_queuepair vmci_qp_broker_alloc qp_broker_alloc qp_broker_create // set entry->state = VMCIQPB_CREATED_NO_MEM; When unmapping this queue pair, qp_host_unregister_user_memory() will be called to unregister the non-existent user memory, which will result in a null pointer reference. It will also change VMCIQPB_CREATED_NO_MEM to VMCIQPB_CREATED_MEM, which should not be present in this operation. Only when the qp broker has mem, it can unregister the user memory when unmapping the qp broker. Only when the qp broker has no mem, it can register the user memory when mapping the qp broker. Fixes: 06164d2b ("VMCI: queue pairs implementation.") Cc: stable <stable@vger.kernel.org> Reported-by:Hulk Robot <hulkci@huawei.com> Reviewed-by:
Jorgen Hansen <jhansen@vmware.com> Signed-off-by:
Wang Hai <wanghai38@huawei.com> Link: https://lore.kernel.org/r/20210818124845.488312-1-wanghai38@huawei.comSigned-off-by:
-
jing yangyang authored
Remove unneeded variables when "0" can be returned. Generated by: scripts/coccinelle/misc/returnvar.cocci Reported-by:
Zeal Robot <zealci@zte.com.cn> Signed-off-by:
jing yangyang <jing.yangyang@zte.com.cn> Link: https://lore.kernel.org/r/20210820021752.10927-1-jing.yangyang@zte.com.cnSigned-off-by:
-
Colin Ian King authored
The check for count appears to be incorrect since a non-zero count check occurs a couple of statements earlier. Currently the check is always false and the dev->port->irq != PARPORT_IRQ_NONE part of the check is never tested and the if statement is dead-code. Fix this by removing the check on count. Note that this code is pre-git history, so I can't find a sha for it. Acked-by:
Sudip Mukherjee <sudipm.mukherjee@gmail.com> Signed-off-by:
Colin Ian King <colin.king@canonical.com> Addresses-Coverity: ("Logically dead code") Link: https://lore.kernel.org/r/20210730100710.27405-1-colin.king@canonical.comSigned-off-by:
-
- 24 Aug, 2021 4 commits
-
-
Greg Kroah-Hartman authored
Merge tag 'soundwire-5.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire into char-misc-next Vinod writes: soundwire updates for 5.15-rc1 - Core has updates to support SoundWire mockup device (includes tag from asoc), improved error handling and slave status. - Drivers has update on Intel driver for new quriks and better handling of errors and suspend routines * tag 'soundwire-5.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/soundwire: soundwire: cadence: do not extend reset delay soundwire: intel: conditionally exit clock stop mode on system suspend soundwire: intel: skip suspend/resume/wake when link was not started soundwire: intel: fix potential race condition during power down soundwire: cadence: override PDI configurations to create loopback soundwire: cadence: add debugfs interface for PDI loopbacks soundwire: stream: don't program mockup device ports soundwire: bus: squelch error returned by mockup devices soundwire: add flag to ignore all command/control for mockup devices soundwire: stream: don't abort bank switch on Command_Ignored/-ENODATA soundwire: cadence: add paranoid check on self-clearing bits soundwire: dmi-quirks: add quirk for Intel 'Bishop County' NUC M15 soundwire: bus: update Slave status in sdw_clear_slave_status soundwire: cadence: Remove ret variable from sdw_cdns_irq() soundwire: bus: filter out more -EDATA errors on clock stop soundwire: dmi-quirks: add ull suffix for SoundWire _ADR values ASoC: Intel: boards: sof_sdw: add SoundWire mockup codecs for tests ASoC: soc-acpi: tgl: add table for SoundWire mockup devices ASoC: soc-acpi: cnl: add table for SoundWire mockup devices ASoC: codecs: add SoundWire mockup device support
-
Greg Kroah-Hartman authored
Merge tag 'phy-for-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy into char-misc-next Vinod writes: phy-for-5.15 - Updates: - Yaml conversion for Freescale imx8mq usb phy, TI AM654 SERDES phy, Cadence torrent phy - Updates for Amlogic Meson8b-usb2 phy, Samsung ufs phy - New support: - UFS phy for Qualcomm SM6115 - PCIe & USB/DP phy for Qualcomm sc8180x - USB3 PHY support for Qualcomm IPQ6018 - Renesas USB2.0 PHY for RZ/G2L * tag 'phy-for-5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy: (45 commits) phy: qcom-qmp: Add support for SM6115 UFS phy dt-bindings: phy: qcom,qmp: Add SM6115 UFS PHY bindings phy: qmp: Provide unique clock names for DP clocks phy: xilinx: zynqmp: skip PHY initialization and PLL lock for USB phy: amlogic: meson8b-usb2: don't log an error on -EPROBE_DEFER phy: amlogic: meson8b-usb2: Power off the PHY by putting it into reset mode phy: phy-mtk-mipi-dsi: convert to devm_platform_ioremap_resource phy: phy-mtk-mipi-dsi: remove dummy assignment of error number phy: phy-mtk-hdmi: convert to devm_platform_ioremap_resource phy: phy-mtk-ufs: use clock bulk to get clocks phy: phy-mtk-tphy: remove error log of ioremap failure phy: phy-mtk-tphy: print error log using child device phy: phy-mtk-tphy: support type switch by pericfg phy: phy-mtk-tphy: use clock bulk to get clocks dt-bindings: phy: mediatek: tphy: support type switch by pericfg phy: cadence-torrent: Check PIPE mode PHY status to be ready for operation phy: cadence-torrent: Add debug information for PHY configuration phy: cadence-torrent: Add separate functions for reusable code phy: cadence-torrent: Add PHY configuration for DP with 100MHz ref clock phy: cadence-torrent: Add PHY registers for DP in array format ...
-
Greg Kroah-Hartman authored
Merge tag 'icc-5.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/djakov/icc into char-misc-next Georgi writes: interconnect changes for 5.15 Here are changes for the 5.15-rc1 merge window consisting of interconnect core and driver updates. Framework change: - Add sanity check to detect if node is already added to provider. Driver changes: - RPMh drivers probe function consolidation - Add driver for SC8180x platforms - Add support for SC8180x OSM L3 - Use driver-specific naming in OSM L3 Signed-off-by:
Georgi Djakov <djakov@kernel.org> * tag 'icc-5.15-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/djakov/icc: interconnect: qcom: osm-l3: Use driver-specific naming interconnect: qcom: osm-l3: Add sc8180x support dt-bindings: interconnect: Add SC8180x to OSM L3 DT binding interconnect: qcom: Add SC8180x providers dt-bindings: interconnect: Add Qualcomm SC8180x DT bindings interconnect: Sanity check that node isn't already on list interconnect: qcom: icc-rpmh: Consolidate probe functions
-
Greg Kroah-Hartman authored
We need the char/misc fixes in here as well. Signed-off-by:
-
- 23 Aug, 2021 6 commits
-
-
Pierre-Louis Bossart authored
The duration of the hw_reset is defined as 4096 cycles. The Cadence IP allows for an additional delay which doesn't seem necessary in practice: the actual reset sequence duration is defined by the sync_go mechanism, not by the IP itself. Signed-off-by:
Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com> Reviewed-by:
Rander Wang <rander.wang@intel.com> Signed-off-by:
Bard Liao <yung-chuan.liao@linux.intel.com> Link: https://lore.kernel.org/r/20210818030130.17113-1-yung-chuan.liao@linux.intel.comSigned-off-by:
Vinod Koul <vkoul@kernel.org>
-
Pierre-Louis Bossart authored
Intel stress tests reported issues with the clock stop mode, specifically when trying to do a system suspend while the link is already pm_runtime suspended. In this case, we need to disable the shim wake, but when the PCI parent device is also pm_runtime suspended the SHIM registers are not accessible. Since this is an invalid corner case, this patch suggests a pm_runtime resume of the entire bus to full power (parent+child devices) before the system suspend so that the shim wake can be disabled. Unlike the suspend operation, the .prepare callbacks are propagated from root device to leaf devices. By adding a .prepare callback at the SoundWire link level, we can double-check the pm_runtime status of the device as well as its parent PCI device. When the problematic configuration is detected, the device is pm_runtime resumed - which by construction also resume its parent. An additional loop is added to resume all child devices. In theory we only need to restart the link, but doing so will also cause the physical devices to synchronize and re-initialize, while their Linux devices remain pm_runtime suspended. It's simpler to make sure the codec devices are fully resumed so that we don't have to deal with zombie states. This additional loop could have been avoided by adding a .prepare callback in SoundWire codec drivers. Functionally this would have been equivalent. The rationale for implementing a loop at the link level is only to reduce the amount of code required to deal at the codec level with an Intel corner case - in other words keep codec drivers independent from Intel platform-specific programming sequences. BugLink: https://github.com/thesofproject/linux/issues/2606Signed-off-by:
-
Pierre-Louis Bossart authored
The SoundWire Linux devices are created purely based on information provided by platform firmware (e.g. ACPI DSDT table). When the kernel finds a matching driver for the device address (_ADR), the probe will initialize required data structures and initialize pm ops. When the SoundWire link is started at a later point, the physical devices will synchronize on the SoundWire frames and report their attachment status, thereby triggering the enumeration and initialization of device registers. This two-step solution was a conscious design decision to allow e.g. a driver to use sideband mechanisms to turn power rails on. This can also allow OEMs to describe multiple platforms with the same DSDT table, the devices that are not physically present in hardware. The drawback of this approach is a bit of confusion, with more devices than are actually present in hardware. This results in 'ghost' devices, for which the driver successfully probes, but that will not generate any traffic on the bus. suspend-resume transitions are handled by drivers, and skipped when the devices are not physically present. This patch provides a work-around for a second-level of confusion in platform firmware: some platforms only use HDaudio links, but nevertheless expose SoundWire 'ghost' devices. This results in error messages in the Intel driver while trying to suspend/resume these links. The simplest solution is to add a boolean status flag to skip all suspend/resume/wake sequences if the link was never started. Signed-off-by:
-
Pierre-Louis Bossart authored
The power down sequence sets the link_up flag as false outside of the mutex_lock. This is potentially unsafe. In additional the flow in that sequence can be improved by first testing if the link was powered, setting the link_up flag as false and proceeding with the power down. In case the CPA bits cannot be cleared, we only flag an error since we cannot deal with interrupts any longer. Signed-off-by:
Ranjani Sridharan <ranjani.sridharan@linux.intel.com> Signed-off-by:
-
Iskren Chernev authored
Add the tables and constants for init sequences for UFS QMP phy found in SM4250/6115 SoC. The phy is a variation of the v2 phy, but is mistakenly labeled as v3-660 in downstream sources. QSERDES COM, RX, TX registers match fully existing v2 registers, with a few additions. PCS registers don't have much in common, but there are no clashes with existing ones so new registers were added to existing v2 PCS pack. Signed-off-by:
Iskren Chernev <iskren.chernev@gmail.com> Link: https://lore.kernel.org/r/20210821155657.893165-3-iskren.chernev@gmail.comSigned-off-by:
-
Iskren Chernev authored
Add the compatible strings for the UFS PHY found on SM4250/6115 SoC. Signed-off-by:
Rob Herring <robh@kernel.org> Link: https://lore.kernel.org/r/20210821155657.893165-2-iskren.chernev@gmail.comSigned-off-by:
-
- 22 Aug, 2021 2 commits
-
-
Linus Torvalds authored
-
git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linuxLinus Torvalds authored
Pull powerpc fixes from Michael Ellerman: - Fix random crashes on some 32-bit CPUs by adding isync() after locking/unlocking KUEP - Fix intermittent crashes when loading modules with strict module RWX - Fix a section mismatch introduce by a previous fix. Thanks to Christophe Leroy, Fabiano Rosas, Laurent Vivier, Murilo Opsfelder Araújo, Nathan Chancellor, and Stan Johnson. h# -----BEGIN PGP SIGNATURE----- * tag 'powerpc-5.14-6' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux: powerpc/mm: Fix set_memory_*() against concurrent accesses powerpc/32s: Fix random crashes by adding isync() after locking/unlocking KUEP powerpc/xive: Do not mark xive_request_ipi() as __init
-
- 21 Aug, 2021 2 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/clk/linuxLinus Torvalds authored
Pull clk driver fixes from Stephen Boyd: - Make the regulator state match the GDSC power domain state at boot on Qualcomm SoCs so that the regulator isn't turned off inadvertently. - Fix earlycon on i.MX6Q SoCs * tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux: clk: qcom: gdsc: Ensure regulator init state matches GDSC state clk: imx6q: fix uart earlycon unwork
-
git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-miscLinus Torvalds authored
Pull char/misc driver fixes from Greg KH: "Here are some small driver fixes for 5.14-rc7. They consist of: - revert for an interconnect patch that was found to have problems - ipack tpci200 driver fixes for reported problems - slimbus messaging and ngd fixes for reported problems All are small and have been in linux-next for a while with no reported issues" * tag 'char-misc-5.14-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: ipack: tpci200: fix memory leak in the tpci200_register ipack: tpci200: fix many double free issues in tpci200_pci_probe slimbus: ngd: reset dma setup during runtime pm slimbus: ngd: set correct device for pm slimbus: messaging: check for valid transaction id slimbus: messaging: start transaction ids from 1 instead of zero Revert "interconnect: qcom: icc-rpmh: Add BCMs to commit list in pre_aggregate"
-
