Copyright (c) 2007-2009 Robert N. M. Watson All rights reserved. Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: 1. Redistributions of source code must retain the above copyright notice(s), this list of conditions and the following disclaimer as the first lines of this file unmodified other than the possible addition of one or more copyright notices. 2. Redistributions in binary form must ...
NAMEstack - kernel thread stack tracing routines
SYNOPSISIn sys/param.h In sys/stack.h
In the kernel configuration file: options DDB options STACK
Ft struct stack * Fn stack_create void Ft void Fn stack_destroy struct stack *st Ft int Fn stack_put struct stack *st vm_offset_t pc Ft void Fn stack_copy const struct stack *src struct stack dst Ft void Fn stack_zero struct stack *st Ft void Fn stack_print const struct stack *st Ft void Fn stack_print_ddb const struct stack *st Ft void Fn stack_print_short const struct stack *st Ft void Fn stack_print_short_ddb const struct stack *st Ft void Fn stack_sbuf_print struct sbuf sb* const struct stack *st Ft void Fn stack_sbuf_print_ddb struct sbuf sb* const struct stack *st Ft void Fn stack_save struct stack *st Ft void Fn stack_save_td struct stack *st struct thread *td Ft int Fn stack_save_td_running struct stack *st struct thread *td
DESCRIPTIONThe sysctl Cm net.inet.tcp.syncookies KPI allows querying of kernel stack trace information and the automated generation of kernel stack trace strings for the purposes of debugging and tracing. To use the KPI, at least one of options DDB and options STACK must be compiled into the kernel.
Each stack trace is described by a Vt struct stack . Before a trace may be created or otherwise manipulated, storage for the trace must be allocated with Fn stack_create , which may sleep. Memory associated with a trace is freed by calling Fn stack_destroy .
A trace of the current kernel thread's call stack may be captured using Fn stack_save . Fn stack_save_td and Fn stack_save_td_running can also be used to capture the stack of a caller-specified thread. Callers of these functions must own the thread lock of the specified thread. Fn stack_save_td can capture the stack of a kernel thread that is not running or swapped out at the time of the call. Fn stack_save_td_running can capture the stack of a running kernel thread.
Fn stack_print and Fn stack_print_short may be used to print a stack trace using the kernel printf(9), and may sleep as a result of acquiring sx(9) locks in the kernel linker while looking up symbol names. In locking-sensitive environments, the unsynchronized Fn stack_print_ddb and Fn stack_print_short_ddb variants may be invoked. This function bypasses kernel linker locking, making it usable in ddb(4), but not in a live system where linker data structures may change.
Fn stack_sbuf_print may be used to construct a human-readable string, including conversion (where possible) from a simple kernel instruction pointer to a named symbol and offset. The argument sb must be an initialized struct sbuf as described in sbuf(9). This function may sleep if an auto-extending struct sbuf is used, or due to kernel linker locking. In locking-sensitive environments, such as ddb(4), the unsynchronized Fn stack_sbuf_print_ddb variant may be invoked to avoid kernel linker locking; it should be used with a fixed-length sbuf.
The utility functions stack_zero stack_copy and stack_put may be used to manipulate stack data structures directly.
RETURN VALUESFn stack_put returns 0 on success. Otherwise the struct stack does not contain space to record additional frames, and a non-zero value is returned.
Fn stack_save_td_running returns 0 when the stack capture was successful and a non-zero error number otherwise. In particular, Er EAGAIN is returned if the thread was running in user mode at the time that the capture was attempted, and Er EOPNOTSUPP is returned if the operation is not implemented.