Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
/*-
|
|
|
|
* Copyright (c) 2007 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, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS 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.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* DDB capture support: capture kernel debugger output into a fixed-size
|
|
|
|
* buffer for later dumping to disk or extraction from user space.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
2008-01-26 22:32:23 +00:00
|
|
|
#include "opt_ddb.h"
|
|
|
|
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/conf.h>
|
|
|
|
#include <sys/kernel.h>
|
|
|
|
#include <sys/kerneldump.h>
|
|
|
|
#include <sys/malloc.h>
|
|
|
|
#include <sys/msgbuf.h>
|
|
|
|
#include <sys/priv.h>
|
|
|
|
#include <sys/sx.h>
|
|
|
|
#include <sys/sysctl.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
|
|
|
|
#include <ddb/ddb.h>
|
|
|
|
#include <ddb/db_lex.h>
|
|
|
|
|
|
|
|
/*
|
|
|
|
* While it would be desirable to use a small block-sized buffer and dump
|
|
|
|
* incrementally to disk in fixed-size blocks, it's not possible to enter
|
|
|
|
* kernel dumper routines without restarting the kernel, which is undesirable
|
|
|
|
* in the midst of debugging. Instead, we maintain a large static global
|
|
|
|
* buffer that we fill from DDB's output routines.
|
2008-01-26 22:32:23 +00:00
|
|
|
*
|
|
|
|
* We enforce an invariant at runtime that buffer sizes are even multiples of
|
|
|
|
* the textdump block size, which is a design choice that we might want to
|
|
|
|
* reconsider.
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
*/
|
2008-01-26 13:55:52 +00:00
|
|
|
static MALLOC_DEFINE(M_DDB_CAPTURE, "ddb_capture", "DDB capture buffer");
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
|
2008-01-26 22:32:23 +00:00
|
|
|
#ifndef DDB_CAPTURE_DEFAULTBUFSIZE
|
2008-01-26 13:55:52 +00:00
|
|
|
#define DDB_CAPTURE_DEFAULTBUFSIZE 48*1024
|
2008-01-26 22:32:23 +00:00
|
|
|
#endif
|
|
|
|
#ifndef DDB_CAPTURE_MAXBUFSIZE
|
2008-01-26 23:02:14 +00:00
|
|
|
#define DDB_CAPTURE_MAXBUFSIZE 5*1024*1024
|
2008-01-26 22:32:23 +00:00
|
|
|
#endif
|
2008-01-26 13:55:52 +00:00
|
|
|
#define DDB_CAPTURE_FILENAME "ddb.txt" /* Captured DDB output. */
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
|
|
|
|
static char *db_capture_buf;
|
2008-01-26 13:55:52 +00:00
|
|
|
static u_int db_capture_bufsize = DDB_CAPTURE_DEFAULTBUFSIZE;
|
|
|
|
static u_int db_capture_maxbufsize = DDB_CAPTURE_MAXBUFSIZE; /* Read-only. */
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
static u_int db_capture_bufoff; /* Next location to write in buffer. */
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
static u_int db_capture_bufpadding; /* Amount of zero padding. */
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
static int db_capture_inpager; /* Suspend capture in pager. */
|
|
|
|
static int db_capture_inprogress; /* DDB capture currently in progress. */
|
|
|
|
|
|
|
|
struct sx db_capture_sx; /* Lock against user thread races. */
|
|
|
|
SX_SYSINIT(db_capture_sx, &db_capture_sx, "db_capture_sx");
|
|
|
|
|
|
|
|
static SYSCTL_NODE(_debug_ddb, OID_AUTO, capture, CTLFLAG_RW, 0,
|
|
|
|
"DDB capture options");
|
|
|
|
|
2008-04-25 13:23:36 +00:00
|
|
|
SYSCTL_UINT(_debug_ddb_capture, OID_AUTO, bufoff, CTLFLAG_RD,
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
&db_capture_bufoff, 0, "Bytes of data in DDB capture buffer");
|
|
|
|
|
|
|
|
SYSCTL_UINT(_debug_ddb_capture, OID_AUTO, maxbufsize, CTLFLAG_RD,
|
|
|
|
&db_capture_maxbufsize, 0,
|
|
|
|
"Maximum value for debug.ddb.capture.bufsize");
|
|
|
|
|
2008-04-25 13:23:36 +00:00
|
|
|
SYSCTL_UINT(_debug_ddb_capture, OID_AUTO, inprogress, CTLFLAG_RD,
|
|
|
|
&db_capture_inprogress, 0, "DDB output capture in progress");
|
|
|
|
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
/*
|
2008-01-26 22:32:23 +00:00
|
|
|
* Boot-time allocation of the DDB capture buffer, if any. Force all buffer
|
|
|
|
* sizes, including the maximum size, to be rounded to block sizes.
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_sysinit(__unused void *dummy)
|
|
|
|
{
|
|
|
|
|
|
|
|
TUNABLE_INT_FETCH("debug.ddb.capture.bufsize", &db_capture_bufsize);
|
2008-01-26 22:32:23 +00:00
|
|
|
db_capture_maxbufsize = roundup(db_capture_maxbufsize,
|
|
|
|
TEXTDUMP_BLOCKSIZE);
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
db_capture_bufsize = roundup(db_capture_bufsize, TEXTDUMP_BLOCKSIZE);
|
2008-01-26 22:32:23 +00:00
|
|
|
if (db_capture_bufsize > db_capture_maxbufsize)
|
|
|
|
db_capture_bufsize = db_capture_maxbufsize;
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
if (db_capture_bufsize != 0)
|
2008-01-26 13:55:52 +00:00
|
|
|
db_capture_buf = malloc(db_capture_bufsize, M_DDB_CAPTURE,
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
M_WAITOK);
|
|
|
|
}
|
|
|
|
SYSINIT(db_capture, SI_SUB_DDB_SERVICES, SI_ORDER_ANY, db_capture_sysinit,
|
|
|
|
NULL);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Run-time adjustment of the capture buffer.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
sysctl_debug_ddb_capture_bufsize(SYSCTL_HANDLER_ARGS)
|
|
|
|
{
|
|
|
|
u_int len, size;
|
|
|
|
char *buf;
|
|
|
|
int error;
|
|
|
|
|
|
|
|
size = db_capture_bufsize;
|
|
|
|
error = sysctl_handle_int(oidp, &size, 0, req);
|
|
|
|
if (error || req->newptr == NULL)
|
|
|
|
return (error);
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
size = roundup(size, TEXTDUMP_BLOCKSIZE);
|
2008-01-26 22:32:23 +00:00
|
|
|
if (size > db_capture_maxbufsize)
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
return (EINVAL);
|
|
|
|
sx_xlock(&db_capture_sx);
|
|
|
|
if (size != 0) {
|
|
|
|
/*
|
|
|
|
* Potentially the buffer is quite large, so if we can't
|
|
|
|
* allocate it, fail rather than waiting.
|
|
|
|
*/
|
2008-01-26 13:55:52 +00:00
|
|
|
buf = malloc(size, M_DDB_CAPTURE, M_NOWAIT);
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
if (buf == NULL) {
|
|
|
|
sx_xunlock(&db_capture_sx);
|
|
|
|
return (ENOMEM);
|
|
|
|
}
|
|
|
|
len = min(db_capture_bufoff, size);
|
|
|
|
} else {
|
|
|
|
buf = NULL;
|
|
|
|
len = 0;
|
|
|
|
}
|
|
|
|
if (db_capture_buf != NULL && buf != NULL)
|
|
|
|
bcopy(db_capture_buf, buf, len);
|
|
|
|
if (db_capture_buf != NULL)
|
2008-01-26 13:55:52 +00:00
|
|
|
free(db_capture_buf, M_DDB_CAPTURE);
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
db_capture_bufoff = len;
|
|
|
|
db_capture_buf = buf;
|
|
|
|
db_capture_bufsize = size;
|
|
|
|
sx_xunlock(&db_capture_sx);
|
|
|
|
|
|
|
|
KASSERT(db_capture_bufoff <= db_capture_bufsize,
|
|
|
|
("sysctl_debug_ddb_capture_bufsize: bufoff > bufsize"));
|
2008-01-26 22:32:23 +00:00
|
|
|
KASSERT(db_capture_bufsize <= db_capture_maxbufsize,
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
("sysctl_debug_ddb_capture_maxbufsize: bufsize > maxbufsize"));
|
|
|
|
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
SYSCTL_PROC(_debug_ddb_capture, OID_AUTO, bufsize, CTLTYPE_UINT|CTLFLAG_RW,
|
|
|
|
0, 0, sysctl_debug_ddb_capture_bufsize, "IU",
|
|
|
|
"Size of DDB capture buffer");
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Sysctl to read out the capture buffer from userspace. We require
|
|
|
|
* privilege as sensitive process/memory information may be accessed.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
sysctl_debug_ddb_capture_data(SYSCTL_HANDLER_ARGS)
|
|
|
|
{
|
|
|
|
int error;
|
|
|
|
char ch;
|
|
|
|
|
|
|
|
error = priv_check(req->td, PRIV_DDB_CAPTURE);
|
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
|
|
|
|
sx_slock(&db_capture_sx);
|
|
|
|
error = SYSCTL_OUT(req, db_capture_buf, db_capture_bufoff);
|
|
|
|
sx_sunlock(&db_capture_sx);
|
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
ch = '\0';
|
|
|
|
return (SYSCTL_OUT(req, &ch, sizeof(ch)));
|
|
|
|
}
|
|
|
|
SYSCTL_PROC(_debug_ddb_capture, OID_AUTO, data, CTLTYPE_STRING | CTLFLAG_RD,
|
|
|
|
NULL, 0, sysctl_debug_ddb_capture_data, "A", "DDB capture data");
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Routines for capturing DDB output into a fixed-size buffer. These are
|
|
|
|
* invoked from DDB's input and output routines. If we hit the limit on the
|
|
|
|
* buffer, we simply drop further data.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
db_capture_write(char *buffer, u_int buflen)
|
|
|
|
{
|
|
|
|
u_int len;
|
|
|
|
|
|
|
|
if (db_capture_inprogress == 0 || db_capture_inpager)
|
|
|
|
return;
|
|
|
|
len = min(buflen, db_capture_bufsize - db_capture_bufoff);
|
|
|
|
bcopy(buffer, db_capture_buf + db_capture_bufoff, len);
|
|
|
|
db_capture_bufoff += len;
|
|
|
|
|
|
|
|
KASSERT(db_capture_bufoff <= db_capture_bufsize,
|
|
|
|
("db_capture_write: bufoff > bufsize"));
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
db_capture_writech(char ch)
|
|
|
|
{
|
|
|
|
|
|
|
|
return (db_capture_write(&ch, sizeof(ch)));
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
db_capture_enterpager(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
db_capture_inpager = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
db_capture_exitpager(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
db_capture_inpager = 0;
|
|
|
|
}
|
|
|
|
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
/*
|
|
|
|
* Zero out any bytes left in the last block of the DDB capture buffer. This
|
|
|
|
* is run shortly before writing the blocks to disk, rather than when output
|
|
|
|
* capture is stopped, in order to avoid injecting nul's into the middle of
|
|
|
|
* output.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_zeropad(void)
|
|
|
|
{
|
|
|
|
u_int len;
|
|
|
|
|
|
|
|
len = min(TEXTDUMP_BLOCKSIZE, (db_capture_bufsize -
|
|
|
|
db_capture_bufoff) % TEXTDUMP_BLOCKSIZE);
|
|
|
|
bzero(db_capture_buf + db_capture_bufoff, len);
|
|
|
|
db_capture_bufpadding = len;
|
|
|
|
}
|
|
|
|
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
/*
|
|
|
|
* Reset capture state, which flushes buffers.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_reset(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
db_capture_inprogress = 0;
|
|
|
|
db_capture_bufoff = 0;
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
db_capture_bufpadding = 0;
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Start capture. Only one session is allowed at any time, but we may
|
|
|
|
* continue a previous session, so the buffer isn't reset.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_start(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (db_capture_inprogress) {
|
|
|
|
db_printf("Capture already started\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
db_capture_inprogress = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
* Terminate DDB output capture--real work is deferred to db_capture_dump,
|
|
|
|
* which executes outside of the DDB context. We don't zero pad here because
|
|
|
|
* capture may be started again before the dump takes place.
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_stop(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
if (db_capture_inprogress == 0) {
|
|
|
|
db_printf("Capture not started\n");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
db_capture_inprogress = 0;
|
|
|
|
}
|
|
|
|
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
/*
|
|
|
|
* Dump DDB(4) captured output (and resets capture buffers).
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
db_capture_dump(struct dumperinfo *di)
|
|
|
|
{
|
|
|
|
u_int offset;
|
|
|
|
|
|
|
|
if (db_capture_bufoff == 0)
|
|
|
|
return;
|
|
|
|
|
|
|
|
db_capture_zeropad();
|
2008-01-26 13:55:52 +00:00
|
|
|
textdump_mkustar(textdump_block_buffer, DDB_CAPTURE_FILENAME,
|
Add textdump(4) facility, which provides an alternative form of kernel
dump using mechanically generated/extracted debugging output rather than
a simple memory dump. Current sources of debugging output are:
- DDB output capture buffer, if there is captured output to save
- Kernel message buffer
- Kernel configuration, if included in kernel
- Kernel version string
- Panic message
Textdumps are stored in swap/dump partitions as with regular dumps, but
are laid out as ustar files in order to allow multiple parts to be stored
as a stream of sequentially written blocks. Blocks are written out in
reverse order, as the size of a textdump isn't known a priori. As with
regular dumps, they will be extracted using savecore(8).
One new DDB(4) command is added, "textdump", which accepts "set",
"unset", and "status" arguments. By default, normal kernel dumps are
generated unless "textdump set" is run in order to schedule a textdump.
It can be canceled using "textdump unset" to restore generation of a
normal kernel dump.
Several sysctls exist to configure aspects of textdumps;
debug.ddb.textdump.pending can be set to check whether a textdump is
pending, or set/unset in order to control whether the next kernel dump
will be a textdump from userspace.
While textdumps don't have to be generated as a result of a DDB script
run automatically as part of a kernel panic, this is a particular useful
way to use them, as instead of generating a complete memory dump, a
simple transcript of an automated DDB session can be captured using the
DDB output capture and textdump facilities. This can be used to
generate quite brief kernel bug reports rich in debugging information
but not dependent on kernel symbol tables or precisely synchronized
source code. Most textdumps I generate are less than 100k including
the full message buffer. Using textdumps with an interactive debugging
session is also useful, with capture being enabled/disabled in order to
record some but not all of the DDB session.
MFC after: 3 months
2007-12-26 11:32:33 +00:00
|
|
|
db_capture_bufoff);
|
|
|
|
(void)textdump_writenextblock(di, textdump_block_buffer);
|
|
|
|
for (offset = 0; offset < db_capture_bufoff + db_capture_bufpadding;
|
|
|
|
offset += TEXTDUMP_BLOCKSIZE)
|
|
|
|
(void)textdump_writenextblock(di, db_capture_buf + offset);
|
|
|
|
db_capture_bufoff = 0;
|
|
|
|
db_capture_bufpadding = 0;
|
|
|
|
}
|
|
|
|
|
Add a new DDB(4) facility, output capture. Input and output from DDB may be
captured to a memory buffer for later inspection using sysctl(8), or in the
future, to a textdump.
A new DDB command, "capture", is added, which accepts arguments "on", "off",
"reset", and "status".
A new DDB sysctl tree, debug.ddb.capture, is added, which can be used to
resize the capture buffer and extract buffer contents.
MFC after: 3 months
2007-12-25 23:06:51 +00:00
|
|
|
/*-
|
|
|
|
* DDB(4) command to manage capture:
|
|
|
|
*
|
|
|
|
* capture on - start DDB output capture
|
|
|
|
* capture off - stop DDB output capture
|
|
|
|
* capture reset - reset DDB capture buffer (also stops capture)
|
|
|
|
* capture status - print DDB output capture status
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
db_capture_usage(void)
|
|
|
|
{
|
|
|
|
|
|
|
|
db_error("capture [on|off|reset|status]\n");
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
db_capture_cmd(db_expr_t addr, boolean_t have_addr, db_expr_t count,
|
|
|
|
char *modif)
|
|
|
|
{
|
|
|
|
int t;
|
|
|
|
|
|
|
|
t = db_read_token();
|
|
|
|
if (t != tIDENT) {
|
|
|
|
db_capture_usage();
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
if (db_read_token() != tEOL)
|
|
|
|
db_error("?\n");
|
|
|
|
if (strcmp(db_tok_string, "on") == 0)
|
|
|
|
db_capture_start();
|
|
|
|
else if (strcmp(db_tok_string, "off") == 0)
|
|
|
|
db_capture_stop();
|
|
|
|
else if (strcmp(db_tok_string, "reset") == 0)
|
|
|
|
db_capture_reset();
|
|
|
|
else if (strcmp(db_tok_string, "status") == 0) {
|
|
|
|
db_printf("%u/%u bytes used\n", db_capture_bufoff,
|
|
|
|
db_capture_bufsize);
|
|
|
|
if (db_capture_inprogress)
|
|
|
|
db_printf("capture is on\n");
|
|
|
|
else
|
|
|
|
db_printf("capture is off\n");
|
|
|
|
} else
|
|
|
|
db_capture_usage();
|
|
|
|
}
|