kernel/linux.git/drivers/soc/xilinx, branch v6.1.168

drivers: soc: xilinx: add the missing kfree in xlnx_add_cb_for_suspend()

2024-12-14T18:53:20+00:00

[ Upstream commit 44ed4f90a97ff6f339e50ac01db71544e0990efc ] If we fail to allocate memory for cb_data by kmalloc, the memory allocation for eve_data is never freed, add the missing kfree() in the error handling path. Fixes: 05e5ba40ea7a ("driver: soc: xilinx: Add support of multiple callbacks for same event in event management driver") Signed-off-by: Gaosheng Cui Link: https://lore.kernel.org/r/20240706065155.452764-1-cuigaosheng1@huawei.com Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

drivers: soc: xilinx: check return status of get_api_version()

2024-08-03T06:49:35+00:00

commit 9b003e14801cf85a8cebeddc87bc9fc77100fdce upstream. Currently return status is not getting checked for get_api_version and because of that for x86 arch we are getting below smatch error. CC drivers/soc/xilinx/zynqmp_power.o drivers/soc/xilinx/zynqmp_power.c: In function 'zynqmp_pm_probe': drivers/soc/xilinx/zynqmp_power.c:295:12: warning: 'pm_api_version' is used uninitialized [-Wuninitialized] 295 | if (pm_api_version < ZYNQMP_PM_VERSION) | ^ CHECK drivers/soc/xilinx/zynqmp_power.c drivers/soc/xilinx/zynqmp_power.c:295 zynqmp_pm_probe() error: uninitialized symbol 'pm_api_version'. So, check return status of pm_get_api_version and return error in case of failure to avoid checking uninitialized pm_api_version variable. Fixes: b9b3a8be28b3 ("firmware: xilinx: Remove eemi ops for get_api_version") Signed-off-by: Jay Buddhabhatti Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20240515112345.24673-1-jay.buddhabhatti@amd.com Signed-off-by: Michal Simek Signed-off-by: Greg Kroah-Hartman

soc: xilinx: rename cpu_number1 to dummy_cpu_number

2024-08-03T06:48:58+00:00

[ Upstream commit 4a95449dd975e2ea6629a034f3e74b46c9634916 ] The per cpu variable cpu_number1 is passed to xlnx_event_handler as argument "dev_id", but it is not used in this function. So drop the initialization of this variable and rename it to dummy_cpu_number. This patch is to fix the following call trace when the kernel option CONFIG_DEBUG_ATOMIC_SLEEP is enabled: BUG: sleeping function called from invalid context at include/linux/sched/mm.h:274 in_atomic(): 1, irqs_disabled(): 0, non_block: 0, pid: 1, name: swapper/0 preempt_count: 1, expected: 0 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 6.1.0 #53 Hardware name: Xilinx Versal vmk180 Eval board rev1.1 (QSPI) (DT) Call trace: dump_backtrace+0xd0/0xe0 show_stack+0x18/0x40 dump_stack_lvl+0x7c/0xa0 dump_stack+0x18/0x34 __might_resched+0x10c/0x140 __might_sleep+0x4c/0xa0 __kmem_cache_alloc_node+0xf4/0x168 kmalloc_trace+0x28/0x38 __request_percpu_irq+0x74/0x138 xlnx_event_manager_probe+0xf8/0x298 platform_probe+0x68/0xd8 Fixes: daed80ed0758 ("soc: xilinx: Fix for call trace due to the usage of smp_processor_id()") Signed-off-by: Jay Buddhabhatti Link: https://lore.kernel.org/r/20240408110610.15676-1-jay.buddhabhatti@amd.com Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

soc: xilinx: fix unhandled SGI warning message

2024-02-05T20:12:50+00:00

[ Upstream commit 9c6724abf969251af53cdae525ad8100ec78d3c2 ] Xen broadcasts SGI to each VM when multiple VMs run on Xen hypervisor. In such case spurious SGI is expected if one event is registered by one VM and not registered by another VM. We let users know that Unhandled SGI is not error and expected if kernel is running on Xen hypervisor. Signed-off-by: Tanmay Shah Signed-off-by: Radhey Shyam Pandey Link: https://lore.kernel.org/r/1698431039-2734260-1-git-send-email-radhey.shyam.pandey@amd.com Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

soc: xilinx: Fix for call trace due to the usage of smp_processor_id()

2024-02-05T20:12:50+00:00

[ Upstream commit daed80ed07580e5adc0e6d8bc79933a35154135a ] When preemption is enabled in kernel and if any task which can be preempted should not use smp_processor_id() directly, since CPU switch can happen at any time, the previous value of cpu_id differs with current cpu_id. As a result we see the below call trace during xlnx_event_manager_probe. [ 6.140197] dump_backtrace+0x0/0x190 [ 6.143884] show_stack+0x18/0x40 [ 6.147220] dump_stack_lvl+0x7c/0xa0 [ 6.150907] dump_stack+0x18/0x34 [ 6.154241] check_preemption_disabled+0x124/0x134 [ 6.159068] debug_smp_processor_id+0x20/0x2c [ 6.163453] xlnx_event_manager_probe+0x48/0x250 To protect cpu_id, It is recommended to use get_cpu()/put_cpu() to disable preemption, get the cpu_id and enable preemption respectively. (For Reference, Documentation/locking/preempt-locking.rst and Documentation/kernel-hacking/hacking.rst) Use preempt_disable()/smp_processor_id()/preempt_enable() API's to achieve the same. Signed-off-by: HariBabu Gattem Signed-off-by: Jay Buddhabhatti Link: https://lore.kernel.org/r/20231027055622.21544-1-jay.buddhabhatti@amd.com Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

driver: soc: xilinx: use _safe loop iterator to avoid a use after free

2023-07-19T14:21:17+00:00

[ Upstream commit c58da0ba3e5c86e51e2c1557afaf6f71e00c4533 ] The hash_for_each_possible() loop dereferences "eve_data" to get the next item on the list. However the loop frees eve_data so it leads to a use after free. Use hash_for_each_possible_safe() instead. Fixes: c7fdb2404f66 ("drivers: soc: xilinx: add xilinx event management driver") Signed-off-by: Dan Carpenter Link: https://lore.kernel.org/r/761e0e4a-4caf-4a71-8f47-1c6ad908a848@kili.mountain Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

driver: soc: xilinx: fix memory leak in xlnx_add_cb_for_notify_event()

2023-03-11T12:55:16+00:00

[ Upstream commit 1bea534991b9b35c41848a397666ada436456beb ] The kfree() should be called when memory fails to be allocated for cb_data in xlnx_add_cb_for_notify_event(), otherwise there will be a memory leak, so add kfree() to fix it. Fixes: 05e5ba40ea7a ("driver: soc: xilinx: Add support of multiple callbacks for same event in event management driver") Signed-off-by: Gaosheng Cui Acked-by: Michal Simek Link: https://lore.kernel.org/r/20221129010146.1026685-1-cuigaosheng1@huawei.com Signed-off-by: Michal Simek Signed-off-by: Sasha Levin

firmware: xilinx: Add TF_A_PM_REGISTER_SGI SMC call

2022-06-29T12:46:22+00:00

SGI interrupt register and reset is performed by EEMI ioctl IOCTL_REGISTER_SGI. However, this is not correct use of EEMI call. SGI registration functionality does not qualify as energy management activity and so shouldn't be mapped to EEMI call. This new call will replace IOCTL_REGISTER_SGI and will be handled by TF-A specific handler in TF-A. To maintain backward compatibility for a while firmware driver will still use IOCTL_REGISTER_SGI as fallback strategy if new call fails or is not supported by TF-A. This new design also helps to make TF-A as pass through layer for EEMI calls. So we don't have to maintain PM_IOCTL as EEMI API ID in TF-A. Signed-off-by: Tanmay Shah Acked-by: Michal Simek Link: https://lore.kernel.org/r/20220607224253.54919-1-tanmay.shah@xilinx.com Signed-off-by: Michal Simek

driver: soc: xilinx: Update function prototype for xlnx_unregister_event

2022-05-19T16:57:28+00:00

As per the current implementation only single callback data gets saved per event, driver is throwing an error if try to register multiple callback for same event. So at time of unregistration of any event required things are event details and callback handler as parameter of xlnx_unregister_event(). As part of adding support of multiple callbacks for same event also require change in prototype of xlnx_unregister_event(). During unregistration of any events, now required things are event details, callback handler and agent's private data as parameter of xlnx_unregister_event(). Also modify the usage of xlnx_unregister_event() in xilinx/zynqmp_power.c driver as per new implementation. Signed-off-by: Abhyuday Godhasara Link: https://lore.kernel.org/r/20220427074803.19009-3-abhyuday.godhasara@xilinx.com Signed-off-by: Greg Kroah-Hartman

driver: soc: xilinx: Add support of multiple callbacks for same event in event management driver

2022-05-19T16:57:28+00:00

As per the current implementation of only single callback data gets saved per event, driver is throwing an error if try to register multiple callback for same event. Add support of multiple callbacks data for same event. So agent can register for same event with multiple callbacks. Here event management driver will store the callbacks as list in Hash table entry for that event. Here each callback data contain 2 element as callback handler and private data of agent driver. Signed-off-by: Abhyuday Godhasara Link: https://lore.kernel.org/r/20220427074803.19009-2-abhyuday.godhasara@xilinx.com Signed-off-by: Greg Kroah-Hartman