From 0d6526f82c3cdefcca47f73f5fc08dc6f335eac6 Mon Sep 17 00:00:00 2001 From: Tim Chen Date: Mon, 21 Sep 2026 17:37:22 -0700 Subject: sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling bug alb_break_llc() decides whether to break LLC preference during active load balance. It does so by testing that every runnable fair task on the source rq prefers its LLC: env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable But the two counters cover different sets. nr_pref_llc_running is updated in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued, so it follows queued tasks. h_nr_runnable is updated in set_delayed()/ clear_delayed() and drops delay-dequeued tasks. So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks, alb_break_llc() returns false, and active balance is free to pull a task off its preferred LLC. Active balance only moves runnable tasks, and this is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB skips the per-task test in can_migrate_task(). The runnable set is the one we want. Fix it on the counter side. A task should be counted in nr_pref_llc_running exactly while it is both queued on its preferred LLC (pref_llc_queued) and runnable (!sched_delayed). Define that membership once in task_pref_llc_runnable(), and adjust the counter only through pref_llc_running_inc()/pref_llc_running_dec() from the four sites that change either input: account_llc_enqueue(), account_llc_dequeue(), set_delayed() and clear_delayed(). Gating every update on the same predicate keeps the delay, wake and dequeue paths from double-counting or underflowing; see the comments at those sites for the ordering. nr_llc_running and sd->llc_counts are not touched and stay on queued semantics. Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC") Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/ Reported-by: Zhan Xusheng Suggested-by: Chen Yu Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Kayra Cizmeci Cc: # v7.2.x Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com --- drivers/crypto/ccp/tee-dev.h | 116 +++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 116 insertions(+) create mode 100644 drivers/crypto/ccp/tee-dev.h (limited to 'drivers/crypto/ccp/tee-dev.h') diff --git a/drivers/crypto/ccp/tee-dev.h b/drivers/crypto/ccp/tee-dev.h new file mode 100644 index 000000000..c23416cb7 --- /dev/null +++ b/drivers/crypto/ccp/tee-dev.h @@ -0,0 +1,116 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright (C) 2019,2021 Advanced Micro Devices, Inc. + * + * Author: Rijo Thomas + * Author: Devaraj Rangasamy + * + */ + +/* This file describes the TEE communication interface between host and AMD + * Secure Processor + */ + +#ifndef __TEE_DEV_H__ +#define __TEE_DEV_H__ + +#include +#include + +#define TEE_DEFAULT_CMD_TIMEOUT (10 * MSEC_PER_SEC) +#define TEE_DEFAULT_RING_TIMEOUT 10 +#define MAX_BUFFER_SIZE 988 + +/** + * struct tee_init_ring_cmd - Command to init TEE ring buffer + * @low_addr: bits [31:0] of the physical address of ring buffer + * @hi_addr: bits [63:32] of the physical address of ring buffer + * @size: size of ring buffer in bytes + */ +struct tee_init_ring_cmd { + u32 low_addr; + u32 hi_addr; + u32 size; +}; + +#define MAX_RING_BUFFER_ENTRIES 32 + +/** + * struct ring_buf_manager - Helper structure to manage ring buffer. + * @ring_start: starting address of ring buffer + * @ring_size: size of ring buffer in bytes + * @ring_pa: physical address of ring buffer + * @wptr: index to the last written entry in ring buffer + */ +struct ring_buf_manager { + struct mutex mutex; /* synchronizes access to ring buffer */ + void *ring_start; + u32 ring_size; + phys_addr_t ring_pa; + u32 wptr; +}; + +struct psp_tee_device { + struct device *dev; + struct psp_device *psp; + void __iomem *io_regs; + struct tee_vdata *vdata; + struct ring_buf_manager rb_mgr; +}; + +/** + * enum tee_cmd_state - TEE command states for the ring buffer interface + * @TEE_CMD_STATE_INIT: initial state of command when sent from host + * @TEE_CMD_STATE_PROCESS: command being processed by TEE environment + * @TEE_CMD_STATE_COMPLETED: command processing completed + */ +enum tee_cmd_state { + TEE_CMD_STATE_INIT, + TEE_CMD_STATE_PROCESS, + TEE_CMD_STATE_COMPLETED, +}; + +/** + * enum cmd_resp_state - TEE command's response status maintained by driver + * @CMD_RESPONSE_INVALID: initial state when no command is written to ring + * @CMD_WAITING_FOR_RESPONSE: driver waiting for response from TEE + * @CMD_RESPONSE_TIMEDOUT: failed to get response from TEE + * @CMD_RESPONSE_COPIED: driver has copied response from TEE + */ +enum cmd_resp_state { + CMD_RESPONSE_INVALID, + CMD_WAITING_FOR_RESPONSE, + CMD_RESPONSE_TIMEDOUT, + CMD_RESPONSE_COPIED, +}; + +/** + * struct tee_ring_cmd - Structure of the command buffer in TEE ring + * @cmd_id: refers to &enum tee_cmd_id. Command id for the ring buffer + * interface + * @cmd_state: refers to &enum tee_cmd_state + * @status: status of TEE command execution + * @res0: reserved region + * @pdata: private data (currently unused) + * @res1: reserved region + * @buf: TEE command specific buffer + * @flag: refers to &enum cmd_resp_state + */ +struct tee_ring_cmd { + u32 cmd_id; + u32 cmd_state; + u32 status; + u32 res0[1]; + u64 pdata; + u32 res1[2]; + u8 buf[MAX_BUFFER_SIZE]; + u32 flag; + + /* Total size: 1024 bytes */ +} __packed; + +int tee_dev_init(struct psp_device *psp); +void tee_dev_destroy(struct psp_device *psp); +int tee_restore(struct psp_device *psp); + +#endif /* __TEE_DEV_H__ */ -- cgit v1.3.1