Index: head/share/man/man4/ddb.4 =================================================================== --- head/share/man/man4/ddb.4 (revision 290428) +++ head/share/man/man4/ddb.4 (revision 290429) @@ -1,1484 +1,1484 @@ .\" .\" Mach Operating System .\" Copyright (c) 1991,1990 Carnegie Mellon University .\" Copyright (c) 2007 Robert N. M. Watson .\" All Rights Reserved. .\" .\" Permission to use, copy, modify and distribute this software and its .\" documentation is hereby granted, provided that both the copyright .\" notice and this permission notice appear in all copies of the .\" software, derivative works or modified versions, and any portions .\" thereof, and that both notices appear in supporting documentation. .\" .\" CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS" .\" CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND FOR .\" ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE. .\" .\" Carnegie Mellon requests users of this software to return to .\" .\" Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU .\" School of Computer Science .\" Carnegie Mellon University .\" Pittsburgh PA 15213-3890 .\" .\" any improvements or extensions that they make and grant Carnegie Mellon .\" the rights to redistribute these changes. .\" .\" changed a \# to #, since groff choked on it. .\" .\" HISTORY .\" ddb.4,v .\" Revision 1.1 1993/07/15 18:41:02 brezak .\" Man page for DDB .\" .\" Revision 2.6 92/04/08 08:52:57 rpd .\" Changes from OSF. .\" [92/01/17 14:19:22 jsb] .\" Changes for OSF debugger modifications. .\" [91/12/12 tak] .\" .\" Revision 2.5 91/06/25 13:50:22 rpd .\" Added some watchpoint explanation. .\" [91/06/25 rpd] .\" .\" Revision 2.4 91/06/17 15:47:31 jsb .\" Added documentation for continue/c, match, search, and watchpoints. .\" I've not actually explained what a watchpoint is; maybe Rich can .\" do that (hint, hint). .\" [91/06/17 10:58:08 jsb] .\" .\" Revision 2.3 91/05/14 17:04:23 mrt .\" Correcting copyright .\" .\" Revision 2.2 91/02/14 14:10:06 mrt .\" Changed to new Mach copyright .\" [91/02/12 18:10:12 mrt] .\" .\" Revision 2.2 90/08/30 14:23:15 dbg .\" Created. .\" [90/08/30 dbg] .\" .\" $FreeBSD$ .\" -.Dd March 16, 2015 +.Dd November 5, 2015 .Dt DDB 4 .Os .Sh NAME .Nm ddb .Nd interactive kernel debugger .Sh SYNOPSIS In order to enable kernel debugging facilities include: .Bd -ragged -offset indent .Cd options KDB .Cd options DDB .Ed .Pp To prevent activation of the debugger on kernel .Xr panic 9 : .Bd -ragged -offset indent .Cd options KDB_UNATTENDED .Ed .Pp In order to print a stack trace of the current thread on the console for a panic: .Bd -ragged -offset indent .Cd options KDB_TRACE .Ed .Pp To print the numerical value of symbols in addition to the symbolic representation, define: .Bd -ragged -offset indent .Cd options DDB_NUMSYM .Ed .Pp To enable the .Xr gdb 1 backend, so that remote debugging with .Xr kgdb 1 is possible, include: .Bd -ragged -offset indent .Cd options GDB .Ed .Sh DESCRIPTION The .Nm kernel debugger is an interactive debugger with a syntax inspired by .Xr gdb 1 . If linked into the running kernel, it can be invoked locally with the .Ql debug .Xr keymap 5 action. The debugger is also invoked on kernel .Xr panic 9 if the .Va debug.debugger_on_panic .Xr sysctl 8 MIB variable is set non-zero, which is the default unless the .Dv KDB_UNATTENDED option is specified. .Pp The current location is called .Va dot . The .Va dot is displayed with a hexadecimal format at a prompt. The commands .Ic examine and .Ic write update .Va dot to the address of the last line examined or the last location modified, and set .Va next to the address of the next location to be examined or changed. Other commands do not change .Va dot , and set .Va next to be the same as .Va dot . .Pp The general command syntax is: .Ar command Ns Op Li / Ns Ar modifier .Ar address Ns Op Li , Ns Ar count .Pp A blank line repeats the previous command from the address .Va next with count 1 and no modifiers. Specifying .Ar address sets .Va dot to the address. Omitting .Ar address uses .Va dot . A missing .Ar count is taken to be 1 for printing commands or infinity for stack traces. .Pp The .Nm debugger has a pager feature (like the .Xr more 1 command) for the output. If an output line exceeds the number set in the .Va lines variable, it displays .Dq Li --More-- and waits for a response. The valid responses for it are: .Pp .Bl -tag -compact -width ".Li SPC" .It Li SPC one more page .It Li RET one more line .It Li q abort the current command, and return to the command input mode .El .Pp Finally, .Nm provides a small (currently 10 items) command history, and offers simple .Nm emacs Ns -style command line editing capabilities. In addition to the .Nm emacs control keys, the usual .Tn ANSI arrow keys may be used to browse through the history buffer, and move the cursor within the current line. .Sh COMMANDS .Bl -tag -width indent -compact .It Ic examine .It Ic x Display the addressed locations according to the formats in the modifier. Multiple modifier formats display multiple locations. If no format is specified, the last format specified for this command is used. .Pp The format characters are: .Bl -tag -compact -width indent .It Cm b look at by bytes (8 bits) .It Cm h look at by half words (16 bits) .It Cm l look at by long words (32 bits) .It Cm g look at by quad words (64 bits) .It Cm a print the location being displayed .It Cm A print the location with a line number if possible .It Cm x display in unsigned hex .It Cm z display in signed hex .It Cm o display in unsigned octal .It Cm d display in signed decimal .It Cm u display in unsigned decimal .It Cm r display in current radix, signed .It Cm c display low 8 bits as a character. Non-printing characters are displayed as an octal escape code (e.g., .Ql \e000 ) . .It Cm s display the null-terminated string at the location. Non-printing characters are displayed as octal escapes. .It Cm m display in unsigned hex with character dump at the end of each line. The location is also displayed in hex at the beginning of each line. .It Cm i display as an instruction .It Cm I display as an instruction with possible alternate formats depending on the machine, but none of the supported architectures have an alternate format. .It Cm S display a symbol name for the pointer stored at the address .El .Pp .It Ic xf Examine forward: execute an .Ic examine command with the last specified parameters to it except that the next address displayed by it is used as the start address. .Pp .It Ic xb Examine backward: execute an .Ic examine command with the last specified parameters to it except that the last start address subtracted by the size displayed by it is used as the start address. .Pp .It Ic print Ns Op Li / Ns Cm acdoruxz .It Ic p Ns Op Li / Ns Cm acdoruxz Print .Ar addr Ns s according to the modifier character (as described above for .Cm examine ) . Valid formats are: .Cm a , x , z , o , d , u , r , and .Cm c . If no modifier is specified, the last one specified to it is used. The argument .Ar addr can be a string, in which case it is printed as it is. For example: .Bd -literal -offset indent print/x "eax = " $eax "\enecx = " $ecx "\en" .Ed .Pp will print like: .Bd -literal -offset indent eax = xxxxxx ecx = yyyyyy .Ed .Pp .It Xo .Ic write Ns Op Li / Ns Cm bhl .Ar addr expr1 Op Ar expr2 ... .Xc .It Xo .Ic w Ns Op Li / Ns Cm bhl .Ar addr expr1 Op Ar expr2 ... .Xc Write the expressions specified after .Ar addr on the command line at succeeding locations starting with .Ar addr . The write unit size can be specified in the modifier with a letter .Cm b (byte), .Cm h (half word) or .Cm l (long word) respectively. If omitted, long word is assumed. .Pp .Sy Warning : since there is no delimiter between expressions, strange things may happen. It is best to enclose each expression in parentheses. .Pp .It Ic set Li $ Ns Ar variable Oo Li = Oc Ar expr Set the named variable or register with the value of .Ar expr . Valid variable names are described below. .Pp .It Ic break Ns Op Li / Ns Cm u .It Ic b Ns Op Li / Ns Cm u Set a break point at .Ar addr . If .Ar count is supplied, continues .Ar count \- 1 times before stopping at the break point. If the break point is set, a break point number is printed with .Ql # . This number can be used in deleting the break point or adding conditions to it. .Pp If the .Cm u modifier is specified, this command sets a break point in user address space. Without the .Cm u option, the address is considered to be in the kernel space, and a wrong space address is rejected with an error message. This modifier can be used only if it is supported by machine dependent routines. .Pp .Sy Warning : If a user text is shadowed by a normal user space debugger, user space break points may not work correctly. Setting a break point at the low-level code paths may also cause strange behavior. .Pp .It Ic delete Ar addr .It Ic d Ar addr .It Ic delete Li # Ns Ar number .It Ic d Li # Ns Ar number Delete the break point. The target break point can be specified by a break point number with .Ql # , or by using the same .Ar addr specified in the original .Ic break command. .Pp .It Ic watch Ar addr Ns Li , Ns Ar size Set a watchpoint for a region. Execution stops when an attempt to modify the region occurs. The .Ar size argument defaults to 4. If you specify a wrong space address, the request is rejected with an error message. .Pp .Sy Warning : Attempts to watch wired kernel memory may cause unrecoverable error in some systems such as i386. Watchpoints on user addresses work best. .Pp .It Ic hwatch Ar addr Ns Li , Ns Ar size Set a hardware watchpoint for a region if supported by the architecture. Execution stops when an attempt to modify the region occurs. The .Ar size argument defaults to 4. .Pp .Sy Warning : The hardware debug facilities do not have a concept of separate address spaces like the watch command does. Use .Ic hwatch for setting watchpoints on kernel address locations only, and avoid its use on user mode address spaces. .Pp .It Ic dhwatch Ar addr Ns Li , Ns Ar size Delete specified hardware watchpoint. .Pp .It Ic step Ns Op Li / Ns Cm p .It Ic s Ns Op Li / Ns Cm p Single step .Ar count times (the comma is a mandatory part of the syntax). If the .Cm p modifier is specified, print each instruction at each step. Otherwise, only print the last instruction. .Pp .Sy Warning : depending on machine type, it may not be possible to single-step through some low-level code paths or user space code. On machines with software-emulated single-stepping (e.g., pmax), stepping through code executed by interrupt handlers will probably do the wrong thing. .Pp .It Ic continue Ns Op Li / Ns Cm c .It Ic c Ns Op Li / Ns Cm c Continue execution until a breakpoint or watchpoint. If the .Cm c modifier is specified, count instructions while executing. Some machines (e.g., pmax) also count loads and stores. .Pp .Sy Warning : when counting, the debugger is really silently single-stepping. This means that single-stepping on low-level code may cause strange behavior. .Pp .It Ic until Ns Op Li / Ns Cm p Stop at the next call or return instruction. If the .Cm p modifier is specified, print the call nesting depth and the cumulative instruction count at each call or return. Otherwise, only print when the matching return is hit. .Pp .It Ic next Ns Op Li / Ns Cm p .It Ic match Ns Op Li / Ns Cm p Stop at the matching return instruction. If the .Cm p modifier is specified, print the call nesting depth and the cumulative instruction count at each call or return. Otherwise, only print when the matching return is hit. .Pp .It Xo .Ic trace Ns Op Li / Ns Cm u .Op Ar pid | tid .Op Li , Ns Ar count .Xc .It Xo .Ic t Ns Op Li / Ns Cm u .Op Ar pid | tid .Op Li , Ns Ar count .Xc .It Xo .Ic where Ns Op Li / Ns Cm u .Op Ar pid | tid .Op Li , Ns Ar count .Xc .It Xo .Ic bt Ns Op Li / Ns Cm u .Op Ar pid | tid .Op Li , Ns Ar count .Xc Stack trace. The .Cm u option traces user space; if omitted, .Ic trace only traces kernel space. The optional argument .Ar count is the number of frames to be traced. If .Ar count is omitted, all frames are printed. .Pp .Sy Warning : User space stack trace is valid only if the machine dependent code supports it. .Pp .It Xo .Ic search Ns Op Li / Ns Cm bhl .Ar addr .Ar value .Op Ar mask .Op Li , Ns Ar count .Xc Search memory for .Ar value . This command might fail in interesting ways if it does not find the searched-for value. This is because .Nm does not always recover from touching bad memory. The optional .Ar count argument limits the search. .\" .Pp .It Xo .Ic findstack .Ar addr .Xc Prints the thread address for a thread kernel-mode stack of which contains the specified address. If the thread is not found, search the thread stack cache and prints the cached stack address. Otherwise, prints nothing. .Pp .It Ic show Cm all procs Ns Op Li / Ns Cm m .It Ic ps Ns Op Li / Ns Cm m Display all process information. The process information may not be shown if it is not supported in the machine, or the bottom of the stack of the target process is not in the main memory at that time. The .Cm m modifier will alter the display to show VM map addresses for the process and not show other information. .\" .Pp .It Ic show Cm all ttys Show all TTY's within the system. Output is similar to .Xr pstat 8 , but also includes the address of the TTY structure. .\" .Pp .It Ic show Cm allchains Show the same information like "show lockchain" does, but for every thread in the system. .\" .Pp .It Ic show Cm alllocks Show all locks that are currently held. This command is only available if .Xr witness 4 is included in the kernel. .\" .Pp .It Ic show Cm allpcpu The same as "show pcpu", but for every CPU present in the system. .\" .Pp .It Ic show Cm allrman Show information related with resource management, including -interrupt request lines, DMA request lines, I/O ports and I/O memory -addresses. +interrupt request lines, DMA request lines, I/O ports, I/O memory +addresses, and Resource IDs. .\" .Pp .It Ic show Cm apic Dump data about APIC IDT vector mappings. .\" .Pp .It Ic show Cm breaks Show breakpoints set with the "break" command. .\" .Pp .It Ic show Cm bio Ar addr Show information about the bio structure .Vt struct bio present at .Ar addr . See the .Pa sys/bio.h header file and .Xr g_bio 9 for more details on the exact meaning of the structure fields. .\" .Pp .It Ic show Cm buffer Ar addr Show information about the buf structure .Vt struct buf present at .Ar addr . See the .Pa sys/buf.h header file for more details on the exact meaning of the structure fields. .\" .Pp .It Ic show Cm cbstat Show brief information about the TTY subsystem. .\" .Pp .It Ic show Cm cdev Without argument, show the list of all created cdev's, consisting of devfs node name and struct cdev address. When address of cdev is supplied, show some internal devfs state of the cdev. .\" .Pp .It Ic show Cm conifhk Lists hooks currently waiting for completion in run_interrupt_driven_config_hooks(). .\" .Pp .It Ic show Cm cpusets Print numbered root and assigned CPU affinity sets. See .Xr cpuset 2 for more details. .\" .Pp .It Ic show Cm cyrixreg Show registers specific to the Cyrix processor. .\" .Pp .It Ic show Cm devmap Prints the contents of the static device mapping table. Currently only available on the ARM architecture. .\" .Pp .It Ic show Cm domain Ar addr Print protocol domain structure .Vt struct domain at address .Ar addr . See the .Pa sys/domain.h header file for more details on the exact meaning of the structure fields. .\" .Pp .It Ic show Cm ffs Op Ar addr Show brief information about ffs mount at the address .Ar addr , if argument is given. Otherwise, provides the summary about each ffs mount. .\" .Pp .It Ic show Cm file Ar addr Show information about the file structure .Vt struct file present at address .Ar addr . .\" .Pp .It Ic show Cm files Show information about every file structure in the system. .\" .Pp .It Ic show Cm freepages Show the number of physical pages in each of the free lists. .\" .Pp .It Ic show Cm geom Op Ar addr If the .Ar addr argument is not given, displays the entire GEOM topology. If .Ar addr is given, displays details about the given GEOM object (class, geom, provider or consumer). .\" .Pp .It Ic show Cm idt Show IDT layout. The first column specifies the IDT vector. The second one is the name of the interrupt/trap handler. Those functions are machine dependent. .\" .Pp .It Ic show Cm inodedeps Op Ar addr Show brief information about each inodedep structure. If .Ar addr is given, only inodedeps belonging to the fs located at the supplied address are shown. .\" .Pp .It Ic show Cm inpcb Ar addr Show information on IP Control Block .Vt struct in_pcb present at .Ar addr . .\" .Pp .It Ic show Cm intr Dump information about interrupt handlers. .\" .Pp .It Ic show Cm intrcnt Dump the interrupt statistics. .\" .Pp .It Ic show Cm irqs Show interrupt lines and their respective kernel threads. .\" .Pp .It Ic show Cm jails Show the list of .Xr jail 8 instances. In addition to what .Xr jls 8 shows, also list kernel internal details. .\" .Pp .It Ic show Cm lapic Show information from the local APIC registers for this CPU. .\" .Pp .It Ic show Cm lock Ar addr Show lock structure. The output format is as follows: .Bl -tag -width "flags" .It Ic class: Class of the lock. Possible types include .Xr mutex 9 , .Xr rmlock 9 , .Xr rwlock 9 , .Xr sx 9 . .It Ic name: Name of the lock. .It Ic flags: Flags passed to the lock initialization function. For exact possibilities see manual pages of possible lock types. .It Ic state: Current state of a lock. As well as .Ic flags it's lock-specific. .It Ic owner: Lock owner. .El .\" .Pp .It Ic show Cm lockchain Ar addr Show all threads a particular thread at address .Ar addr is waiting on based on non-sleepable and non-spin locks. .\" .Pp .It Ic show Cm lockedbufs Show the same information as "show buf", but for every locked .Vt struct buf object. .\" .Pp .It Ic show Cm lockedvnods List all locked vnodes in the system. .\" .Pp .It Ic show Cm locks Prints all locks that are currently acquired. This command is only available if .Xr witness 4 is included in the kernel. .\" .Pp .It Ic show Cm locktree .\" .Pp .It Ic show Cm malloc Prints .Xr malloc 9 memory allocator statistics. The output format is as follows: .Pp .Bl -tag -compact -offset indent -width "Requests" .It Ic Type Specifies a type of memory. It is the same as a description string used while defining the given memory type with .Xr MALLOC_DECLARE 9 . .It Ic InUse Number of memory allocations of the given type, for which .Xr free 9 has not been called yet. .It Ic MemUse Total memory consumed by the given allocation type. .It Ic Requests Number of memory allocation requests for the given memory type. .El .Pp The same information can be gathered in userspace with .Dq Nm vmstat Fl m . .\" .Pp .It Ic show Cm map Ns Oo Li / Ns Cm f Oc Ar addr Prints the VM map at .Ar addr . If the .Cm f modifier is specified the complete map is printed. .\" .Pp .It Ic show Cm msgbuf Print the system's message buffer. It is the same output as in the .Dq Nm dmesg case. It is useful if you got a kernel panic, attached a serial cable to the machine and want to get the boot messages from before the system hang. .\" .It Ic show Cm mount Displays short info about all currently mounted file systems. .Pp .It Ic show Cm mount Ar addr Displays details about the given mount point. .\" .Pp .It Ic show Cm object Ns Oo Li / Ns Cm f Oc Ar addr Prints the VM object at .Ar addr . If the .Cm f option is specified the complete object is printed. .\" .Pp .It Ic show Cm page Show statistics on VM pages. .\" .Pp .It Ic show Cm pageq Show statistics on VM page queues. .\" .Pp .It Ic show Cm pciregs Print PCI bus registers. The same information can be gathered in userspace by running .Dq Nm pciconf Fl lv . .\" .Pp .It Ic show Cm pcpu Print current processor state. The output format is as follows: .Pp .Bl -tag -compact -offset indent -width "spin locks held:" .It Ic cpuid Processor identifier. .It Ic curthread Thread pointer, process identifier and the name of the process. .It Ic curpcb Control block pointer. .It Ic fpcurthread FPU thread pointer. .It Ic idlethread Idle thread pointer. .It Ic APIC ID CPU identifier coming from APIC. .It Ic currentldt LDT pointer. .It Ic spin locks held Names of spin locks held. .El .\" .Pp .It Ic show Cm pgrpdump Dump process groups present within the system. .\" .Pp .It Ic show Cm proc Op Ar addr If no .Op Ar addr is specified, print information about the current process. Otherwise, show information about the process at address .Ar addr . .\" .Pp .It Ic show Cm procvm Show process virtual memory layout. .\" .Pp .It Ic show Cm protosw Ar addr Print protocol switch structure .Vt struct protosw at address .Ar addr . .\" .Pp .It Ic show Cm registers Ns Op Li / Ns Cm u Display the register set. If the .Cm u modifier is specified, it displays user registers instead of kernel registers or the currently saved one. .Pp .Sy Warning : The support of the .Cm u modifier depends on the machine. If not supported, incorrect information will be displayed. .\" .Pp .It Ic show Cm rman Ar addr Show resource manager object .Vt struct rman at address .Ar addr . Addresses of particular pointers can be gathered with "show allrman" command. .\" .Pp .It Ic show Cm rtc Show real time clock value. Useful for long debugging sessions. .\" .Pp .It Ic show Cm sleepchain Show all the threads a particular thread is waiting on based on sleepable locks. .\" .Pp .It Ic show Cm sleepq .It Ic show Cm sleepqueue Both commands provide the same functionality. They show sleepqueue .Vt struct sleepqueue structure. Sleepqueues are used within the .Fx kernel to implement sleepable synchronization primitives (thread holding a lock might sleep or be context switched), which at the time of writing are: .Xr condvar 9 , .Xr sx 9 and standard .Xr msleep 9 interface. .\" .Pp .It Ic show Cm sockbuf Ar addr .It Ic show Cm socket Ar addr Those commands print .Vt struct sockbuf and .Vt struct socket objects placed at .Ar addr . Output consists of all values present in structures mentioned. For exact interpretation and more details, visit .Pa sys/socket.h header file. .\" .Pp .It Ic show Cm sysregs Show system registers (e.g., .Li cr0-4 on i386.) Not present on some platforms. .\" .Pp .It Ic show Cm tcpcb Ar addr Print TCP control block .Vt struct tcpcb lying at address .Ar addr . For exact interpretation of output, visit .Pa netinet/tcp.h header file. .\" .Pp .It Ic show Cm thread Op Ar addr If no .Ar addr is specified, show detailed information about current thread. Otherwise, information about thread at .Ar addr is printed. .\" .Pp .It Ic show Cm threads Show all threads within the system. Output format is as follows: .Pp .Bl -tag -compact -offset indent -width "Second column" .It Ic First column Thread identifier (TID) .It Ic Second column Thread structure address .It Ic Third column Backtrace. .El .\" .Pp .It Ic show Cm tty Ar addr Display the contents of a TTY structure in a readable form. .\" .Pp .It Ic show Cm turnstile Ar addr Show turnstile .Vt struct turnstile structure at address .Ar addr . Turnstiles are structures used within the .Fx kernel to implement synchronization primitives which, while holding a specific type of lock, cannot sleep or context switch to another thread. Currently, those are: .Xr mutex 9 , .Xr rwlock 9 , .Xr rmlock 9 . .\" .Pp .It Ic show Cm uma Show UMA allocator statistics. Output consists five columns: .Pp .Bl -tag -compact -offset indent -width "Requests" .It Cm "Zone" Name of the UMA zone. The same string that was passed to .Xr uma_zcreate 9 as a first argument. .It Cm "Size" Size of a given memory object (slab). .It Cm "Used" Number of slabs being currently used. .It Cm "Free" Number of free slabs within the UMA zone. .It Cm "Requests" Number of allocations requests to the given zone. .El .Pp The very same information might be gathered in the userspace with the help of .Dq Nm vmstat Fl z . .\" .Pp .It Ic show Cm unpcb Ar addr Shows UNIX domain socket private control block .Vt struct unpcb present at the address .Ar addr . .\" .Pp .It Ic show Cm vmochk Prints, whether the internal VM objects are in a map somewhere and none have zero ref counts. .\" .Pp .It Ic show Cm vmopag This is supposed to show physical addresses consumed by a VM object. Currently, it is not possible to use this command when .Xr witness 4 is compiled in the kernel. .\" .Pp .It Ic show Cm vnode Op Ar addr Prints vnode .Vt struct vnode structure lying at .Op Ar addr . For the exact interpretation of the output, look at the .Pa sys/vnode.h header file. .\" .Pp .It Ic show Cm vnodebufs Ar addr Shows clean/dirty buffer lists of the vnode located at .Ar addr . .\" .Pp .It Ic show Cm watches Displays all watchpoints. Shows watchpoints set with "watch" command. .\" .Pp .It Ic show Cm witness Shows information about lock acquisition coming from the .Xr witness 4 subsystem. .\" .Pp .It Ic gdb Toggles between remote GDB and DDB mode. In remote GDB mode, another machine is required that runs .Xr gdb 1 using the remote debug feature, with a connection to the serial console port on the target machine. Currently only available on the i386 architecture. .Pp .It Ic halt Halt the system. .Pp .It Ic kill Ar sig pid Send signal .Ar sig to process .Ar pid . The signal is acted on upon returning from the debugger. This command can be used to kill a process causing resource contention in the case of a hung system. See .Xr signal 3 for a list of signals. Note that the arguments are reversed relative to .Xr kill 2 . .Pp .It Ic reboot Op Ar seconds .It Ic reset Op Ar seconds Hard reset the system. If the optional argument .Ar seconds is given, the debugger will wait for this long, at most a week, before rebooting. .Pp .It Ic help Print a short summary of the available commands and command abbreviations. .Pp .It Ic capture on .It Ic capture off .It Ic capture reset .It Ic capture status .Nm supports a basic output capture facility, which can be used to retrieve the results of debugging commands from userspace using .Xr sysctl 2 . .Ic capture on enables output capture; .Ic capture off disables capture. .Ic capture reset will clear the capture buffer and disable capture. .Ic capture status will report current buffer use, buffer size, and disposition of output capture. .Pp Userspace processes may inspect and manage .Nm capture state using .Xr sysctl 8 : .Pp .Dv debug.ddb.capture.bufsize may be used to query or set the current capture buffer size. .Pp .Dv debug.ddb.capture.maxbufsize may be used to query the compile-time limit on the capture buffer size. .Pp .Dv debug.ddb.capture.bytes may be used to query the number of bytes of output currently in the capture buffer. .Pp .Dv debug.ddb.capture.data returns the contents of the buffer as a string to an appropriately privileged process. .Pp This facility is particularly useful in concert with the scripting and .Xr textdump 4 facilities, allowing scripted debugging output to be captured and committed to disk as part of a textdump for later analysis. The contents of the capture buffer may also be inspected in a kernel core dump using .Xr kgdb 1 . .Pp .It Ic run .It Ic script .It Ic scripts .It Ic unscript Run, define, list, and delete scripts. See the .Sx SCRIPTING section for more information on the scripting facility. .Pp .It Ic textdump dump .It Ic textdump set .It Ic textdump status .It Ic textdump unset Use the .Ic textdump dump command to immediately perform a textdump. More information may be found in .Xr textdump 4 . The .Ic textdump set command may be used to force the next kernel core dump to be a textdump rather than a traditional memory dump or minidump. .Ic textdump status reports whether a textdump has been scheduled. .Ic textdump unset cancels a request to perform a textdump as the next kernel core dump. .El .Sh VARIABLES The debugger accesses registers and variables as .Li $ Ns Ar name . Register names are as in the .Dq Ic show Cm registers command. Some variables are suffixed with numbers, and may have some modifier following a colon immediately after the variable name. For example, register variables can have a .Cm u modifier to indicate user register (e.g., .Dq Li $eax:u ) . .Pp Built-in variables currently supported are: .Pp .Bl -tag -width ".Va tabstops" -compact .It Va radix Input and output radix. .It Va maxoff Addresses are printed as .Dq Ar symbol Ns Li + Ns Ar offset unless .Ar offset is greater than .Va maxoff . .It Va maxwidth The width of the displayed line. .It Va lines The number of lines. It is used by the built-in pager. .It Va tabstops Tab stop width. .It Va work Ns Ar xx Work variable; .Ar xx can take values from 0 to 31. .El .Sh EXPRESSIONS Most expression operators in C are supported except .Ql ~ , .Ql ^ , and unary .Ql & . Special rules in .Nm are: .Bl -tag -width ".No Identifiers" .It Identifiers The name of a symbol is translated to the value of the symbol, which is the address of the corresponding object. .Ql \&. and .Ql \&: can be used in the identifier. If supported by an object format dependent routine, .Sm off .Oo Ar filename : Oc Ar func : lineno , .Sm on .Oo Ar filename : Oc Ns Ar variable , and .Oo Ar filename : Oc Ns Ar lineno can be accepted as a symbol. .It Numbers Radix is determined by the first two letters: .Ql 0x : hex, .Ql 0o : octal, .Ql 0t : decimal; otherwise, follow current radix. .It Li \&. .Va dot .It Li + .Va next .It Li .. address of the start of the last line examined. Unlike .Va dot or .Va next , this is only changed by .Ic examine or .Ic write command. .It Li ' last address explicitly specified. .It Li $ Ns Ar variable Translated to the value of the specified variable. It may be followed by a .Ql \&: and modifiers as described above. .It Ar a Ns Li # Ns Ar b A binary operator which rounds up the left hand side to the next multiple of right hand side. .It Li * Ns Ar expr Indirection. It may be followed by a .Ql \&: and modifiers as described above. .El .Sh SCRIPTING .Nm supports a basic scripting facility to allow automating tasks or responses to specific events. Each script consists of a list of DDB commands to be executed sequentially, and is assigned a unique name. Certain script names have special meaning, and will be automatically run on various .Nm events if scripts by those names have been defined. .Pp The .Ic script command may be used to define a script by name. Scripts consist of a series of .Nm commands separated with the .Ql \&; character. For example: .Bd -literal -offset indent script kdb.enter.panic=bt; show pcpu script lockinfo=show alllocks; show lockedvnods .Ed .Pp The .Ic scripts command lists currently defined scripts. .Pp The .Ic run command execute a script by name. For example: .Bd -literal -offset indent run lockinfo .Ed .Pp The .Ic unscript command may be used to delete a script by name. For example: .Bd -literal -offset indent unscript kdb.enter.panic .Ed .Pp These functions may also be performed from userspace using the .Xr ddb 8 command. .Pp Certain scripts are run automatically, if defined, for specific .Nm events. The follow scripts are run when various events occur: .Bl -tag -width kdb.enter.powerfail .It Dv kdb.enter.acpi The kernel debugger was entered as a result of an .Xr acpi 4 event. .It Dv kdb.enter.bootflags The kernel debugger was entered at boot as a result of the debugger boot flag being set. .It Dv kdb.enter.break The kernel debugger was entered as a result of a serial or console break. .It Dv kdb.enter.cam The kernel debugger was entered as a result of a .Xr CAM 4 event. .It Dv kdb.enter.mac The kernel debugger was entered as a result of an assertion failure in the .Xr mac_test 4 module of the TrustedBSD MAC Framework. .It Dv kdb.enter.ndis The kernel debugger was entered as a result of an .Xr ndis 4 breakpoint event. .It Dv kdb.enter.netgraph The kernel debugger was entered as a result of a .Xr netgraph 4 event. .It Dv kdb.enter.panic .Xr panic 9 was called. .It Dv kdb.enter.powerfail The kernel debugger was entered as a result of a powerfail NMI on the sparc64 platform. .It Dv kdb.enter.powerpc The kernel debugger was entered as a result of an unimplemented interrupt type on the powerpc platform. .It Dv kdb.enter.sysctl The kernel debugger was entered as a result of the .Dv debug.kdb.enter sysctl being set. .It Dv kdb.enter.trapsig The kernel debugger was entered as a result of a trapsig event on the sparc64 platform. .It Dv kdb.enter.unionfs The kernel debugger was entered as a result of an assertion failure in the union file system. .It Dv kdb.enter.unknown The kernel debugger was entered, but no reason has been set. .It Dv kdb.enter.vfslock The kernel debugger was entered as a result of a VFS lock violation. .It Dv kdb.enter.watchdog The kernel debugger was entered as a result of a watchdog firing. .It Dv kdb.enter.witness The kernel debugger was entered as a result of a .Xr witness 4 violation. .El .Pp In the event that none of these scripts is found, .Nm will attempt to execute a default script: .Bl -tag -width kdb.enter.powerfail .It Dv kdb.enter.default The kernel debugger was entered, but a script exactly matching the reason for entering was not defined. This can be used as a catch-all to handle cases not specifically of interest; for example, .Dv kdb.enter.witness might be defined to have special handling, and .Dv kdb.enter.default might be defined to simply panic and reboot. .El .Sh HINTS On machines with an ISA expansion bus, a simple NMI generation card can be constructed by connecting a push button between the A01 and B01 (CHCHK# and GND) card fingers. Momentarily shorting these two fingers together may cause the bridge chipset to generate an NMI, which causes the kernel to pass control to .Nm . Some bridge chipsets do not generate a NMI on CHCHK#, so your mileage may vary. The NMI allows one to break into the debugger on a wedged machine to diagnose problems. Other bus' bridge chipsets may be able to generate NMI using bus specific methods. .Sh FILES Header files mentioned in this manual page can be found below .Pa /usr/include directory. .Pp .Bl -dash -compact .It .Pa sys/buf.h .It .Pa sys/domain.h .It .Pa netinet/in_pcb.h .It .Pa sys/socket.h .It .Pa sys/vnode.h .El .Sh SEE ALSO .Xr gdb 1 , .Xr kgdb 1 , .Xr acpi 4 , .Xr CAM 4 , .Xr mac_test 4 , .Xr ndis 4 , .Xr netgraph 4 , .Xr textdump 4 , .Xr witness 4 , .Xr ddb 8 , .Xr sysctl 8 , .Xr panic 9 .Sh HISTORY The .Nm debugger was developed for Mach, and ported to .Bx 386 0.1 . This manual page translated from .Xr man 7 macros by .An Garrett Wollman . .Pp .An Robert N. M. Watson added support for .Nm output capture, .Xr textdump 4 and scripting in .Fx 7.1 . Index: head/sys/kern/subr_rman.c =================================================================== --- head/sys/kern/subr_rman.c (revision 290428) +++ head/sys/kern/subr_rman.c (revision 290429) @@ -1,1092 +1,1093 @@ /*- * Copyright 1998 Massachusetts Institute of Technology * * Permission to use, copy, modify, and distribute this software and * its documentation for any purpose and without fee is hereby * granted, provided that both the above copyright notice and this * permission notice appear in all copies, that both the above * copyright notice and this permission notice appear in all * supporting documentation, and that the name of M.I.T. not be used * in advertising or publicity pertaining to distribution of the * software without specific, written prior permission. M.I.T. makes * no representations about the suitability of this software for any * purpose. It is provided "as is" without express or implied * warranty. * * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE, * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. */ /* * The kernel resource manager. This code is responsible for keeping track * of hardware resources which are apportioned out to various drivers. * It does not actually assign those resources, and it is not expected * that end-device drivers will call into this code directly. Rather, * the code which implements the buses that those devices are attached to, * and the code which manages CPU resources, will call this code, and the * end-device drivers will make upcalls to that code to actually perform * the allocation. * * There are two sorts of resources managed by this code. The first is * the more familiar array (RMAN_ARRAY) type; resources in this class * consist of a sequence of individually-allocatable objects which have * been numbered in some well-defined order. Most of the resources * are of this type, as it is the most familiar. The second type is * called a gauge (RMAN_GAUGE), and models fungible resources (i.e., * resources in which each instance is indistinguishable from every * other instance). The principal anticipated application of gauges * is in the context of power consumption, where a bus may have a specific * power budget which all attached devices share. RMAN_GAUGE is not * implemented yet. * * For array resources, we make one simplifying assumption: two clients * sharing the same resource must use the same range of indices. That * is to say, sharing of overlapping-but-not-identical regions is not * permitted. */ #include "opt_ddb.h" #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include /* XXX debugging */ #include #include #include #ifdef DDB #include #endif /* * We use a linked list rather than a bitmap because we need to be able to * represent potentially huge objects (like all of a processor's physical * address space). That is also why the indices are defined to have type * `unsigned long' -- that being the largest integral type in ISO C (1990). * The 1999 version of C allows `long long'; we may need to switch to that * at some point in the future, particularly if we want to support 36-bit * addresses on IA32 hardware. */ struct resource_i { struct resource r_r; TAILQ_ENTRY(resource_i) r_link; LIST_ENTRY(resource_i) r_sharelink; LIST_HEAD(, resource_i) *r_sharehead; u_long r_start; /* index of the first entry in this resource */ u_long r_end; /* index of the last entry (inclusive) */ u_int r_flags; void *r_virtual; /* virtual address of this resource */ struct device *r_dev; /* device which has allocated this resource */ struct rman *r_rm; /* resource manager from whence this came */ int r_rid; /* optional rid for this resource. */ }; static int rman_debug = 0; SYSCTL_INT(_debug, OID_AUTO, rman_debug, CTLFLAG_RWTUN, &rman_debug, 0, "rman debug"); #define DPRINTF(params) if (rman_debug) printf params static MALLOC_DEFINE(M_RMAN, "rman", "Resource manager"); struct rman_head rman_head; static struct mtx rman_mtx; /* mutex to protect rman_head */ static int int_rman_release_resource(struct rman *rm, struct resource_i *r); static __inline struct resource_i * int_alloc_resource(int malloc_flag) { struct resource_i *r; r = malloc(sizeof *r, M_RMAN, malloc_flag | M_ZERO); if (r != NULL) { r->r_r.__r_i = r; } return (r); } int rman_init(struct rman *rm) { static int once = 0; if (once == 0) { once = 1; TAILQ_INIT(&rman_head); mtx_init(&rman_mtx, "rman head", NULL, MTX_DEF); } if (rm->rm_start == 0 && rm->rm_end == 0) rm->rm_end = ~0ul; if (rm->rm_type == RMAN_UNINIT) panic("rman_init"); if (rm->rm_type == RMAN_GAUGE) panic("implement RMAN_GAUGE"); TAILQ_INIT(&rm->rm_list); rm->rm_mtx = malloc(sizeof *rm->rm_mtx, M_RMAN, M_NOWAIT | M_ZERO); if (rm->rm_mtx == NULL) return ENOMEM; mtx_init(rm->rm_mtx, "rman", NULL, MTX_DEF); mtx_lock(&rman_mtx); TAILQ_INSERT_TAIL(&rman_head, rm, rm_link); mtx_unlock(&rman_mtx); return 0; } int rman_manage_region(struct rman *rm, u_long start, u_long end) { struct resource_i *r, *s, *t; int rv = 0; DPRINTF(("rman_manage_region: <%s> request: start %#lx, end %#lx\n", rm->rm_descr, start, end)); if (start < rm->rm_start || end > rm->rm_end) return EINVAL; r = int_alloc_resource(M_NOWAIT); if (r == NULL) return ENOMEM; r->r_start = start; r->r_end = end; r->r_rm = rm; mtx_lock(rm->rm_mtx); /* Skip entries before us. */ TAILQ_FOREACH(s, &rm->rm_list, r_link) { if (s->r_end == ULONG_MAX) break; if (s->r_end + 1 >= r->r_start) break; } /* If we ran off the end of the list, insert at the tail. */ if (s == NULL) { TAILQ_INSERT_TAIL(&rm->rm_list, r, r_link); } else { /* Check for any overlap with the current region. */ if (r->r_start <= s->r_end && r->r_end >= s->r_start) { rv = EBUSY; goto out; } /* Check for any overlap with the next region. */ t = TAILQ_NEXT(s, r_link); if (t && r->r_start <= t->r_end && r->r_end >= t->r_start) { rv = EBUSY; goto out; } /* * See if this region can be merged with the next region. If * not, clear the pointer. */ if (t && (r->r_end + 1 != t->r_start || t->r_flags != 0)) t = NULL; /* See if we can merge with the current region. */ if (s->r_end + 1 == r->r_start && s->r_flags == 0) { /* Can we merge all 3 regions? */ if (t != NULL) { s->r_end = t->r_end; TAILQ_REMOVE(&rm->rm_list, t, r_link); free(r, M_RMAN); free(t, M_RMAN); } else { s->r_end = r->r_end; free(r, M_RMAN); } } else if (t != NULL) { /* Can we merge with just the next region? */ t->r_start = r->r_start; free(r, M_RMAN); } else if (s->r_end < r->r_start) { TAILQ_INSERT_AFTER(&rm->rm_list, s, r, r_link); } else { TAILQ_INSERT_BEFORE(s, r, r_link); } } out: mtx_unlock(rm->rm_mtx); return rv; } int rman_init_from_resource(struct rman *rm, struct resource *r) { int rv; if ((rv = rman_init(rm)) != 0) return (rv); return (rman_manage_region(rm, r->__r_i->r_start, r->__r_i->r_end)); } int rman_fini(struct rman *rm) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (r->r_flags & RF_ALLOCATED) { mtx_unlock(rm->rm_mtx); return EBUSY; } } /* * There really should only be one of these if we are in this * state and the code is working properly, but it can't hurt. */ while (!TAILQ_EMPTY(&rm->rm_list)) { r = TAILQ_FIRST(&rm->rm_list); TAILQ_REMOVE(&rm->rm_list, r, r_link); free(r, M_RMAN); } mtx_unlock(rm->rm_mtx); mtx_lock(&rman_mtx); TAILQ_REMOVE(&rman_head, rm, rm_link); mtx_unlock(&rman_mtx); mtx_destroy(rm->rm_mtx); free(rm->rm_mtx, M_RMAN); return 0; } int rman_first_free_region(struct rman *rm, u_long *start, u_long *end) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (!(r->r_flags & RF_ALLOCATED)) { *start = r->r_start; *end = r->r_end; mtx_unlock(rm->rm_mtx); return (0); } } mtx_unlock(rm->rm_mtx); return (ENOENT); } int rman_last_free_region(struct rman *rm, u_long *start, u_long *end) { struct resource_i *r; mtx_lock(rm->rm_mtx); TAILQ_FOREACH_REVERSE(r, &rm->rm_list, resource_head, r_link) { if (!(r->r_flags & RF_ALLOCATED)) { *start = r->r_start; *end = r->r_end; mtx_unlock(rm->rm_mtx); return (0); } } mtx_unlock(rm->rm_mtx); return (ENOENT); } /* Shrink or extend one or both ends of an allocated resource. */ int rman_adjust_resource(struct resource *rr, u_long start, u_long end) { struct resource_i *r, *s, *t, *new; struct rman *rm; /* Not supported for shared resources. */ r = rr->__r_i; if (r->r_flags & RF_SHAREABLE) return (EINVAL); /* * This does not support wholesale moving of a resource. At * least part of the desired new range must overlap with the * existing resource. */ if (end < r->r_start || r->r_end < start) return (EINVAL); /* * Find the two resource regions immediately adjacent to the * allocated resource. */ rm = r->r_rm; mtx_lock(rm->rm_mtx); #ifdef INVARIANTS TAILQ_FOREACH(s, &rm->rm_list, r_link) { if (s == r) break; } if (s == NULL) panic("resource not in list"); #endif s = TAILQ_PREV(r, resource_head, r_link); t = TAILQ_NEXT(r, r_link); KASSERT(s == NULL || s->r_end + 1 == r->r_start, ("prev resource mismatch")); KASSERT(t == NULL || r->r_end + 1 == t->r_start, ("next resource mismatch")); /* * See if the changes are permitted. Shrinking is always allowed, * but growing requires sufficient room in the adjacent region. */ if (start < r->r_start && (s == NULL || (s->r_flags & RF_ALLOCATED) || s->r_start > start)) { mtx_unlock(rm->rm_mtx); return (EBUSY); } if (end > r->r_end && (t == NULL || (t->r_flags & RF_ALLOCATED) || t->r_end < end)) { mtx_unlock(rm->rm_mtx); return (EBUSY); } /* * While holding the lock, grow either end of the resource as * needed and shrink either end if the shrinking does not require * allocating a new resource. We can safely drop the lock and then * insert a new range to handle the shrinking case afterwards. */ if (start < r->r_start || (start > r->r_start && s != NULL && !(s->r_flags & RF_ALLOCATED))) { KASSERT(s->r_flags == 0, ("prev is busy")); r->r_start = start; if (s->r_start == start) { TAILQ_REMOVE(&rm->rm_list, s, r_link); free(s, M_RMAN); } else s->r_end = start - 1; } if (end > r->r_end || (end < r->r_end && t != NULL && !(t->r_flags & RF_ALLOCATED))) { KASSERT(t->r_flags == 0, ("next is busy")); r->r_end = end; if (t->r_end == end) { TAILQ_REMOVE(&rm->rm_list, t, r_link); free(t, M_RMAN); } else t->r_start = end + 1; } mtx_unlock(rm->rm_mtx); /* * Handle the shrinking cases that require allocating a new * resource to hold the newly-free region. We have to recheck * if we still need this new region after acquiring the lock. */ if (start > r->r_start) { new = int_alloc_resource(M_WAITOK); new->r_start = r->r_start; new->r_end = start - 1; new->r_rm = rm; mtx_lock(rm->rm_mtx); r->r_start = start; s = TAILQ_PREV(r, resource_head, r_link); if (s != NULL && !(s->r_flags & RF_ALLOCATED)) { s->r_end = start - 1; free(new, M_RMAN); } else TAILQ_INSERT_BEFORE(r, new, r_link); mtx_unlock(rm->rm_mtx); } if (end < r->r_end) { new = int_alloc_resource(M_WAITOK); new->r_start = end + 1; new->r_end = r->r_end; new->r_rm = rm; mtx_lock(rm->rm_mtx); r->r_end = end; t = TAILQ_NEXT(r, r_link); if (t != NULL && !(t->r_flags & RF_ALLOCATED)) { t->r_start = end + 1; free(new, M_RMAN); } else TAILQ_INSERT_AFTER(&rm->rm_list, r, new, r_link); mtx_unlock(rm->rm_mtx); } return (0); } #define SHARE_TYPE(f) (f & (RF_SHAREABLE | RF_PREFETCHABLE)) struct resource * rman_reserve_resource_bound(struct rman *rm, u_long start, u_long end, u_long count, u_long bound, u_int flags, struct device *dev) { u_int new_rflags; struct resource_i *r, *s, *rv; u_long rstart, rend, amask, bmask; rv = NULL; DPRINTF(("rman_reserve_resource_bound: <%s> request: [%#lx, %#lx], " "length %#lx, flags %u, device %s\n", rm->rm_descr, start, end, count, flags, dev == NULL ? "" : device_get_nameunit(dev))); KASSERT((flags & RF_FIRSTSHARE) == 0, ("invalid flags %#x", flags)); new_rflags = (flags & ~RF_FIRSTSHARE) | RF_ALLOCATED; mtx_lock(rm->rm_mtx); for (r = TAILQ_FIRST(&rm->rm_list); r && r->r_end < start + count - 1; r = TAILQ_NEXT(r, r_link)) ; if (r == NULL) { DPRINTF(("could not find a region\n")); goto out; } amask = (1ul << RF_ALIGNMENT(flags)) - 1; KASSERT(start <= ULONG_MAX - amask, ("start (%#lx) + amask (%#lx) would wrap around", start, amask)); /* If bound is 0, bmask will also be 0 */ bmask = ~(bound - 1); /* * First try to find an acceptable totally-unshared region. */ for (s = r; s; s = TAILQ_NEXT(s, r_link)) { DPRINTF(("considering [%#lx, %#lx]\n", s->r_start, s->r_end)); /* * The resource list is sorted, so there is no point in * searching further once r_start is too large. */ if (s->r_start > end - (count - 1)) { DPRINTF(("s->r_start (%#lx) + count - 1> end (%#lx)\n", s->r_start, end)); break; } if (s->r_start > ULONG_MAX - amask) { DPRINTF(("s->r_start (%#lx) + amask (%#lx) too large\n", s->r_start, amask)); break; } if (s->r_flags & RF_ALLOCATED) { DPRINTF(("region is allocated\n")); continue; } rstart = ulmax(s->r_start, start); /* * Try to find a region by adjusting to boundary and alignment * until both conditions are satisfied. This is not an optimal * algorithm, but in most cases it isn't really bad, either. */ do { rstart = (rstart + amask) & ~amask; if (((rstart ^ (rstart + count - 1)) & bmask) != 0) rstart += bound - (rstart & ~bmask); } while ((rstart & amask) != 0 && rstart < end && rstart < s->r_end); rend = ulmin(s->r_end, ulmax(rstart + count - 1, end)); if (rstart > rend) { DPRINTF(("adjusted start exceeds end\n")); continue; } DPRINTF(("truncated region: [%#lx, %#lx]; size %#lx (requested %#lx)\n", rstart, rend, (rend - rstart + 1), count)); if ((rend - rstart + 1) >= count) { DPRINTF(("candidate region: [%#lx, %#lx], size %#lx\n", rstart, rend, (rend - rstart + 1))); if ((s->r_end - s->r_start + 1) == count) { DPRINTF(("candidate region is entire chunk\n")); rv = s; rv->r_flags = new_rflags; rv->r_dev = dev; goto out; } /* * If s->r_start < rstart and * s->r_end > rstart + count - 1, then * we need to split the region into three pieces * (the middle one will get returned to the user). * Otherwise, we are allocating at either the * beginning or the end of s, so we only need to * split it in two. The first case requires * two new allocations; the second requires but one. */ rv = int_alloc_resource(M_NOWAIT); if (rv == NULL) goto out; rv->r_start = rstart; rv->r_end = rstart + count - 1; rv->r_flags = new_rflags; rv->r_dev = dev; rv->r_rm = rm; if (s->r_start < rv->r_start && s->r_end > rv->r_end) { DPRINTF(("splitting region in three parts: " "[%#lx, %#lx]; [%#lx, %#lx]; [%#lx, %#lx]\n", s->r_start, rv->r_start - 1, rv->r_start, rv->r_end, rv->r_end + 1, s->r_end)); /* * We are allocating in the middle. */ r = int_alloc_resource(M_NOWAIT); if (r == NULL) { free(rv, M_RMAN); rv = NULL; goto out; } r->r_start = rv->r_end + 1; r->r_end = s->r_end; r->r_flags = s->r_flags; r->r_rm = rm; s->r_end = rv->r_start - 1; TAILQ_INSERT_AFTER(&rm->rm_list, s, rv, r_link); TAILQ_INSERT_AFTER(&rm->rm_list, rv, r, r_link); } else if (s->r_start == rv->r_start) { DPRINTF(("allocating from the beginning\n")); /* * We are allocating at the beginning. */ s->r_start = rv->r_end + 1; TAILQ_INSERT_BEFORE(s, rv, r_link); } else { DPRINTF(("allocating at the end\n")); /* * We are allocating at the end. */ s->r_end = rv->r_start - 1; TAILQ_INSERT_AFTER(&rm->rm_list, s, rv, r_link); } goto out; } } /* * Now find an acceptable shared region, if the client's requirements * allow sharing. By our implementation restriction, a candidate * region must match exactly by both size and sharing type in order * to be considered compatible with the client's request. (The * former restriction could probably be lifted without too much * additional work, but this does not seem warranted.) */ DPRINTF(("no unshared regions found\n")); if ((flags & RF_SHAREABLE) == 0) goto out; for (s = r; s && s->r_end <= end; s = TAILQ_NEXT(s, r_link)) { if (SHARE_TYPE(s->r_flags) == SHARE_TYPE(flags) && s->r_start >= start && (s->r_end - s->r_start + 1) == count && (s->r_start & amask) == 0 && ((s->r_start ^ s->r_end) & bmask) == 0) { rv = int_alloc_resource(M_NOWAIT); if (rv == NULL) goto out; rv->r_start = s->r_start; rv->r_end = s->r_end; rv->r_flags = new_rflags; rv->r_dev = dev; rv->r_rm = rm; if (s->r_sharehead == NULL) { s->r_sharehead = malloc(sizeof *s->r_sharehead, M_RMAN, M_NOWAIT | M_ZERO); if (s->r_sharehead == NULL) { free(rv, M_RMAN); rv = NULL; goto out; } LIST_INIT(s->r_sharehead); LIST_INSERT_HEAD(s->r_sharehead, s, r_sharelink); s->r_flags |= RF_FIRSTSHARE; } rv->r_sharehead = s->r_sharehead; LIST_INSERT_HEAD(s->r_sharehead, rv, r_sharelink); goto out; } } /* * We couldn't find anything. */ out: mtx_unlock(rm->rm_mtx); return (rv == NULL ? NULL : &rv->r_r); } struct resource * rman_reserve_resource(struct rman *rm, u_long start, u_long end, u_long count, u_int flags, struct device *dev) { return (rman_reserve_resource_bound(rm, start, end, count, 0, flags, dev)); } int rman_activate_resource(struct resource *re) { struct resource_i *r; struct rman *rm; r = re->__r_i; rm = r->r_rm; mtx_lock(rm->rm_mtx); r->r_flags |= RF_ACTIVE; mtx_unlock(rm->rm_mtx); return 0; } int rman_deactivate_resource(struct resource *r) { struct rman *rm; rm = r->__r_i->r_rm; mtx_lock(rm->rm_mtx); r->__r_i->r_flags &= ~RF_ACTIVE; mtx_unlock(rm->rm_mtx); return 0; } static int int_rman_release_resource(struct rman *rm, struct resource_i *r) { struct resource_i *s, *t; if (r->r_flags & RF_ACTIVE) r->r_flags &= ~RF_ACTIVE; /* * Check for a sharing list first. If there is one, then we don't * have to think as hard. */ if (r->r_sharehead) { /* * If a sharing list exists, then we know there are at * least two sharers. * * If we are in the main circleq, appoint someone else. */ LIST_REMOVE(r, r_sharelink); s = LIST_FIRST(r->r_sharehead); if (r->r_flags & RF_FIRSTSHARE) { s->r_flags |= RF_FIRSTSHARE; TAILQ_INSERT_BEFORE(r, s, r_link); TAILQ_REMOVE(&rm->rm_list, r, r_link); } /* * Make sure that the sharing list goes away completely * if the resource is no longer being shared at all. */ if (LIST_NEXT(s, r_sharelink) == NULL) { free(s->r_sharehead, M_RMAN); s->r_sharehead = NULL; s->r_flags &= ~RF_FIRSTSHARE; } goto out; } /* * Look at the adjacent resources in the list and see if our * segment can be merged with any of them. If either of the * resources is allocated or is not exactly adjacent then they * cannot be merged with our segment. */ s = TAILQ_PREV(r, resource_head, r_link); if (s != NULL && ((s->r_flags & RF_ALLOCATED) != 0 || s->r_end + 1 != r->r_start)) s = NULL; t = TAILQ_NEXT(r, r_link); if (t != NULL && ((t->r_flags & RF_ALLOCATED) != 0 || r->r_end + 1 != t->r_start)) t = NULL; if (s != NULL && t != NULL) { /* * Merge all three segments. */ s->r_end = t->r_end; TAILQ_REMOVE(&rm->rm_list, r, r_link); TAILQ_REMOVE(&rm->rm_list, t, r_link); free(t, M_RMAN); } else if (s != NULL) { /* * Merge previous segment with ours. */ s->r_end = r->r_end; TAILQ_REMOVE(&rm->rm_list, r, r_link); } else if (t != NULL) { /* * Merge next segment with ours. */ t->r_start = r->r_start; TAILQ_REMOVE(&rm->rm_list, r, r_link); } else { /* * At this point, we know there is nothing we * can potentially merge with, because on each * side, there is either nothing there or what is * there is still allocated. In that case, we don't * want to remove r from the list; we simply want to * change it to an unallocated region and return * without freeing anything. */ r->r_flags &= ~RF_ALLOCATED; r->r_dev = NULL; return 0; } out: free(r, M_RMAN); return 0; } int rman_release_resource(struct resource *re) { int rv; struct resource_i *r; struct rman *rm; r = re->__r_i; rm = r->r_rm; mtx_lock(rm->rm_mtx); rv = int_rman_release_resource(rm, r); mtx_unlock(rm->rm_mtx); return (rv); } uint32_t rman_make_alignment_flags(uint32_t size) { int i; /* * Find the hightest bit set, and add one if more than one bit * set. We're effectively computing the ceil(log2(size)) here. */ for (i = 31; i > 0; i--) if ((1 << i) & size) break; if (~(1 << i) & size) i++; return(RF_ALIGNMENT_LOG2(i)); } void rman_set_start(struct resource *r, u_long start) { r->__r_i->r_start = start; } u_long rman_get_start(struct resource *r) { return (r->__r_i->r_start); } void rman_set_end(struct resource *r, u_long end) { r->__r_i->r_end = end; } u_long rman_get_end(struct resource *r) { return (r->__r_i->r_end); } u_long rman_get_size(struct resource *r) { return (r->__r_i->r_end - r->__r_i->r_start + 1); } u_int rman_get_flags(struct resource *r) { return (r->__r_i->r_flags); } void rman_set_virtual(struct resource *r, void *v) { r->__r_i->r_virtual = v; } void * rman_get_virtual(struct resource *r) { return (r->__r_i->r_virtual); } void rman_set_bustag(struct resource *r, bus_space_tag_t t) { r->r_bustag = t; } bus_space_tag_t rman_get_bustag(struct resource *r) { return (r->r_bustag); } void rman_set_bushandle(struct resource *r, bus_space_handle_t h) { r->r_bushandle = h; } bus_space_handle_t rman_get_bushandle(struct resource *r) { return (r->r_bushandle); } void rman_set_rid(struct resource *r, int rid) { r->__r_i->r_rid = rid; } int rman_get_rid(struct resource *r) { return (r->__r_i->r_rid); } void rman_set_device(struct resource *r, struct device *dev) { r->__r_i->r_dev = dev; } struct device * rman_get_device(struct resource *r) { return (r->__r_i->r_dev); } int rman_is_region_manager(struct resource *r, struct rman *rm) { return (r->__r_i->r_rm == rm); } /* * Sysctl interface for scanning the resource lists. * * We take two input parameters; the index into the list of resource * managers, and the resource offset into the list. */ static int sysctl_rman(SYSCTL_HANDLER_ARGS) { int *name = (int *)arg1; u_int namelen = arg2; int rman_idx, res_idx; struct rman *rm; struct resource_i *res; struct resource_i *sres; struct u_rman urm; struct u_resource ures; int error; if (namelen != 3) return (EINVAL); if (bus_data_generation_check(name[0])) return (EINVAL); rman_idx = name[1]; res_idx = name[2]; /* * Find the indexed resource manager */ mtx_lock(&rman_mtx); TAILQ_FOREACH(rm, &rman_head, rm_link) { if (rman_idx-- == 0) break; } mtx_unlock(&rman_mtx); if (rm == NULL) return (ENOENT); /* * If the resource index is -1, we want details on the * resource manager. */ if (res_idx == -1) { bzero(&urm, sizeof(urm)); urm.rm_handle = (uintptr_t)rm; if (rm->rm_descr != NULL) strlcpy(urm.rm_descr, rm->rm_descr, RM_TEXTLEN); urm.rm_start = rm->rm_start; urm.rm_size = rm->rm_end - rm->rm_start + 1; urm.rm_type = rm->rm_type; error = SYSCTL_OUT(req, &urm, sizeof(urm)); return (error); } /* * Find the indexed resource and return it. */ mtx_lock(rm->rm_mtx); TAILQ_FOREACH(res, &rm->rm_list, r_link) { if (res->r_sharehead != NULL) { LIST_FOREACH(sres, res->r_sharehead, r_sharelink) if (res_idx-- == 0) { res = sres; goto found; } } else if (res_idx-- == 0) goto found; } mtx_unlock(rm->rm_mtx); return (ENOENT); found: bzero(&ures, sizeof(ures)); ures.r_handle = (uintptr_t)res; ures.r_parent = (uintptr_t)res->r_rm; ures.r_device = (uintptr_t)res->r_dev; if (res->r_dev != NULL) { if (device_get_name(res->r_dev) != NULL) { snprintf(ures.r_devname, RM_TEXTLEN, "%s%d", device_get_name(res->r_dev), device_get_unit(res->r_dev)); } else { strlcpy(ures.r_devname, "nomatch", RM_TEXTLEN); } } else { ures.r_devname[0] = '\0'; } ures.r_start = res->r_start; ures.r_size = res->r_end - res->r_start + 1; ures.r_flags = res->r_flags; mtx_unlock(rm->rm_mtx); error = SYSCTL_OUT(req, &ures, sizeof(ures)); return (error); } static SYSCTL_NODE(_hw_bus, OID_AUTO, rman, CTLFLAG_RD, sysctl_rman, "kernel resource manager"); #ifdef DDB static void dump_rman_header(struct rman *rm) { if (db_pager_quit) return; db_printf("rman %p: %s (0x%lx-0x%lx full range)\n", rm, rm->rm_descr, rm->rm_start, rm->rm_end); } static void dump_rman(struct rman *rm) { struct resource_i *r; const char *devname; if (db_pager_quit) return; TAILQ_FOREACH(r, &rm->rm_list, r_link) { if (r->r_dev != NULL) { devname = device_get_nameunit(r->r_dev); if (devname == NULL) devname = "nomatch"; } else devname = NULL; - db_printf(" 0x%lx-0x%lx ", r->r_start, r->r_end); + db_printf(" 0x%lx-0x%lx (RID=%d) ", + r->r_start, r->r_end, r->r_rid); if (devname != NULL) db_printf("(%s)\n", devname); else db_printf("----\n"); if (db_pager_quit) return; } } DB_SHOW_COMMAND(rman, db_show_rman) { if (have_addr) { dump_rman_header((struct rman *)addr); dump_rman((struct rman *)addr); } } DB_SHOW_COMMAND(rmans, db_show_rmans) { struct rman *rm; TAILQ_FOREACH(rm, &rman_head, rm_link) { dump_rman_header(rm); } } DB_SHOW_ALL_COMMAND(rman, db_show_all_rman) { struct rman *rm; TAILQ_FOREACH(rm, &rman_head, rm_link) { dump_rman_header(rm); dump_rman(rm); } } DB_SHOW_ALIAS(allrman, db_show_all_rman); #endif