freebsd-nq/sys/kern/subr_prf.c

1125 lines
24 KiB
C
Raw Normal View History

1994-05-24 10:09:53 +00:00
/*-
* Copyright (c) 1986, 1988, 1991, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* 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.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
*
* @(#)subr_prf.c 8.3 (Berkeley) 1/21/94
*/
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_ddb.h"
#include "opt_printf.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/lock.h>
#include <sys/kdb.h>
#include <sys/mutex.h>
#include <sys/sx.h>
#include <sys/kernel.h>
1994-05-24 10:09:53 +00:00
#include <sys/msgbuf.h>
#include <sys/malloc.h>
#include <sys/priv.h>
1994-05-24 10:09:53 +00:00
#include <sys/proc.h>
#include <sys/stddef.h>
#include <sys/sysctl.h>
1994-05-24 10:09:53 +00:00
#include <sys/tty.h>
#include <sys/syslog.h>
#include <sys/cons.h>
#include <sys/uio.h>
#include <sys/ctype.h>
1994-05-24 10:09:53 +00:00
#ifdef DDB
#include <ddb/ddb.h>
#endif
1994-05-24 10:09:53 +00:00
/*
* Note that stdarg.h and the ANSI style va_start macro is used for both
* ANSI and traditional C compilers.
*/
#include <machine/stdarg.h>
#define TOCONS 0x01
#define TOTTY 0x02
#define TOLOG 0x04
/* Max number conversion buffer length: a u_quad_t in base 2, plus NUL byte. */
#define MAXNBUF (sizeof(intmax_t) * NBBY + 1)
1994-05-24 10:09:53 +00:00
struct putchar_arg {
int flags;
int pri;
struct tty *tty;
char *p_bufr;
size_t n_bufr;
char *p_next;
size_t remain;
};
struct snprintf_arg {
char *str;
size_t remain;
};
extern int log_open;
static void msglogchar(int c, int pri);
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
static void msglogstr(char *str, int pri, int filter_cr);
2002-03-19 21:25:46 +00:00
static void putchar(int ch, void *arg);
static char *ksprintn(char *nbuf, uintmax_t num, int base, int *len, int upper);
2002-03-19 21:25:46 +00:00
static void snprintf_func(int ch, void *arg);
1994-05-24 10:09:53 +00:00
static int msgbufmapped; /* Set when safe to use msgbuf */
int msgbuftrigger;
1994-05-24 10:09:53 +00:00
static int log_console_output = 1;
TUNABLE_INT("kern.log_console_output", &log_console_output);
SYSCTL_INT(_kern, OID_AUTO, log_console_output, CTLFLAG_RW,
&log_console_output, 0, "Duplicate console output to the syslog.");
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
/*
* See the comment in log_console() below for more explanation of this.
*/
static int log_console_add_linefeed = 0;
TUNABLE_INT("kern.log_console_add_linefeed", &log_console_add_linefeed);
SYSCTL_INT(_kern, OID_AUTO, log_console_add_linefeed, CTLFLAG_RW,
&log_console_add_linefeed, 0, "log_console() adds extra newlines.");
static int always_console_output = 0;
TUNABLE_INT("kern.always_console_output", &always_console_output);
SYSCTL_INT(_kern, OID_AUTO, always_console_output, CTLFLAG_RW,
&always_console_output, 0, "Always output to console despite TIOCCONS.");
1994-05-24 10:09:53 +00:00
/*
* Warn that a system table is full.
*/
void
tablefull(const char *tab)
1994-05-24 10:09:53 +00:00
{
log(LOG_ERR, "%s: table is full\n", tab);
}
/*
* Uprintf prints to the controlling terminal for the current process.
*/
int
1994-05-24 10:09:53 +00:00
uprintf(const char *fmt, ...)
{
va_list ap;
struct putchar_arg pca;
struct proc *p;
struct thread *td;
int retval;
td = curthread;
if (TD_IS_IDLETHREAD(td))
return (0);
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
sx_slock(&proctree_lock);
p = td->td_proc;
PROC_LOCK(p);
if ((p->p_flag & P_CONTROLT) == 0) {
PROC_UNLOCK(p);
sx_sunlock(&proctree_lock);
return (0);
1994-05-24 10:09:53 +00:00
}
SESS_LOCK(p->p_session);
pca.tty = p->p_session->s_ttyp;
SESS_UNLOCK(p->p_session);
PROC_UNLOCK(p);
if (pca.tty == NULL) {
sx_sunlock(&proctree_lock);
return (0);
}
pca.flags = TOTTY;
pca.p_bufr = NULL;
va_start(ap, fmt);
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
tty_lock(pca.tty);
sx_sunlock(&proctree_lock);
retval = kvprintf(fmt, putchar, &pca, 10, ap);
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
tty_unlock(pca.tty);
va_end(ap);
return (retval);
1994-05-24 10:09:53 +00:00
}
/*
* tprintf and vtprintf print on the controlling terminal associated with the
* given session, possibly to the log as well.
1994-05-24 10:09:53 +00:00
*/
void
tprintf(struct proc *p, int pri, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
vtprintf(p, pri, fmt, ap);
va_end(ap);
}
void
vtprintf(struct proc *p, int pri, const char *fmt, va_list ap)
1994-05-24 10:09:53 +00:00
{
struct tty *tp = NULL;
int flags = 0;
struct putchar_arg pca;
struct session *sess = NULL;
1994-05-24 10:09:53 +00:00
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
sx_slock(&proctree_lock);
if (pri != -1)
flags |= TOLOG;
if (p != NULL) {
PROC_LOCK(p);
if (p->p_flag & P_CONTROLT && p->p_session->s_ttyvp) {
sess = p->p_session;
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
sess_hold(sess);
PROC_UNLOCK(p);
tp = sess->s_ttyp;
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
if (tp != NULL && tty_checkoutq(tp))
flags |= TOTTY;
else
tp = NULL;
} else
PROC_UNLOCK(p);
1994-05-24 10:09:53 +00:00
}
pca.pri = pri;
pca.tty = tp;
pca.flags = flags;
pca.p_bufr = NULL;
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
if (pca.tty != NULL)
tty_lock(pca.tty);
sx_sunlock(&proctree_lock);
kvprintf(fmt, putchar, &pca, 10, ap);
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
if (pca.tty != NULL)
tty_unlock(pca.tty);
if (sess != NULL)
Integrate the new MPSAFE TTY layer to the FreeBSD operating system. The last half year I've been working on a replacement TTY layer for the FreeBSD kernel. The new TTY layer was designed to improve the following: - Improved driver model: The old TTY layer has a driver model that is not abstract enough to make it friendly to use. A good example is the output path, where the device drivers directly access the output buffers. This means that an in-kernel PPP implementation must always convert network buffers into TTY buffers. If a PPP implementation would be built on top of the new TTY layer (still needs a hooks layer, though), it would allow the PPP implementation to directly hand the data to the TTY driver. - Improved hotplugging: With the old TTY layer, it isn't entirely safe to destroy TTY's from the system. This implementation has a two-step destructing design, where the driver first abandons the TTY. After all threads have left the TTY, the TTY layer calls a routine in the driver, which can be used to free resources (unit numbers, etc). The pts(4) driver also implements this feature, which means posix_openpt() will now return PTY's that are created on the fly. - Improved performance: One of the major improvements is the per-TTY mutex, which is expected to improve scalability when compared to the old Giant locking. Another change is the unbuffered copying to userspace, which is both used on TTY device nodes and PTY masters. Upgrading should be quite straightforward. Unlike previous versions, existing kernel configuration files do not need to be changed, except when they reference device drivers that are listed in UPDATING. Obtained from: //depot/projects/mpsafetty/... Approved by: philip (ex-mentor) Discussed: on the lists, at BSDCan, at the DevSummit Sponsored by: Snow B.V., the Netherlands dcons(4) fixed by: kan
2008-08-20 08:31:58 +00:00
sess_release(sess);
msgbuftrigger = 1;
1994-05-24 10:09:53 +00:00
}
/*
* Ttyprintf displays a message on a tty; it should be used only by
* the tty driver, or anything that knows the underlying tty will not
* be revoke(2)'d away. Other callers should use tprintf.
*/
int
1994-05-24 10:09:53 +00:00
ttyprintf(struct tty *tp, const char *fmt, ...)
{
va_list ap;
struct putchar_arg pca;
int retval;
1994-05-24 10:09:53 +00:00
va_start(ap, fmt);
pca.tty = tp;
pca.flags = TOTTY;
pca.p_bufr = NULL;
retval = kvprintf(fmt, putchar, &pca, 10, ap);
1994-05-24 10:09:53 +00:00
va_end(ap);
return (retval);
1994-05-24 10:09:53 +00:00
}
static int
_vprintf(int level, int flags, const char *fmt, va_list ap)
1994-05-24 10:09:53 +00:00
{
struct putchar_arg pca;
int retval;
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
#ifdef PRINTF_BUFR_SIZE
char bufr[PRINTF_BUFR_SIZE];
#endif
1994-05-24 10:09:53 +00:00
pca.tty = NULL;
pca.pri = level;
pca.flags = flags;
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
#ifdef PRINTF_BUFR_SIZE
pca.p_bufr = bufr;
pca.p_next = pca.p_bufr;
pca.n_bufr = sizeof(bufr);
pca.remain = sizeof(bufr);
*pca.p_next = '\0';
#else
/* Don't buffer console output. */
pca.p_bufr = NULL;
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
#endif
retval = kvprintf(fmt, putchar, &pca, 10, ap);
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
#ifdef PRINTF_BUFR_SIZE
/* Write any buffered console/log output: */
if (*pca.p_bufr != '\0') {
if (pca.flags & TOLOG)
msglogstr(pca.p_bufr, level, /*filter_cr*/1);
if (pca.flags & TOCONS)
cnputs(pca.p_bufr);
}
#endif
return (retval);
}
/*
* Log writes to the log buffer, and guarantees not to sleep (so can be
* called by interrupt routines). If there is no process reading the
* log yet, it writes to the console also.
*/
void
log(int level, const char *fmt, ...)
{
va_list ap;
va_start(ap, fmt);
(void)_vprintf(level, log_open ? TOLOG : TOCONS, fmt, ap);
va_end(ap);
msgbuftrigger = 1;
1994-05-24 10:09:53 +00:00
}
#define CONSCHUNK 128
1994-05-24 10:09:53 +00:00
void
log_console(struct uio *uio)
{
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
int c, error, nl;
char *consbuffer;
int pri;
if (!log_console_output)
return;
pri = LOG_INFO | LOG_CONSOLE;
uio = cloneuio(uio);
consbuffer = malloc(CONSCHUNK, M_TEMP, M_WAITOK);
nl = 0;
while (uio->uio_resid > 0) {
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
c = imin(uio->uio_resid, CONSCHUNK - 1);
error = uiomove(consbuffer, c, uio);
if (error != 0)
break;
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
/* Make sure we're NUL-terminated */
consbuffer[c] = '\0';
if (consbuffer[c - 1] == '\n')
nl = 1;
else
nl = 0;
msglogstr(consbuffer, pri, /*filter_cr*/ 1);
}
/*
* The previous behavior in log_console() is preserved when
* log_console_add_linefeed is non-zero. For that behavior, if an
* individual console write came in that was not terminated with a
* line feed, it would add a line feed.
*
* This results in different data in the message buffer than
* appears on the system console (which doesn't add extra line feed
* characters).
*
* A number of programs and rc scripts write a line feed, or a period
* and a line feed when they have completed their operation. On
* the console, this looks seamless, but when displayed with
* 'dmesg -a', you wind up with output that looks like this:
*
* Updating motd:
* .
*
* On the console, it looks like this:
* Updating motd:.
*
* We could add logic to detect that situation, or just not insert
* the extra newlines. Set the kern.log_console_add_linefeed
* sysctl/tunable variable to get the old behavior.
*/
if (!nl && log_console_add_linefeed) {
consbuffer[0] = '\n';
consbuffer[1] = '\0';
msglogstr(consbuffer, pri, /*filter_cr*/ 1);
}
msgbuftrigger = 1;
free(uio, M_IOV);
free(consbuffer, M_TEMP);
return;
1994-05-24 10:09:53 +00:00
}
1996-01-29 03:18:05 +00:00
int
1994-05-24 10:09:53 +00:00
printf(const char *fmt, ...)
{
va_list ap;
1996-01-29 03:18:05 +00:00
int retval;
1994-05-24 10:09:53 +00:00
va_start(ap, fmt);
retval = vprintf(fmt, ap);
1994-05-24 10:09:53 +00:00
va_end(ap);
return (retval);
1994-05-24 10:09:53 +00:00
}
int
vprintf(const char *fmt, va_list ap)
{
int retval;
retval = _vprintf(-1, TOCONS | TOLOG, fmt, ap);
if (!panicstr)
msgbuftrigger = 1;
return (retval);
}
static void
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
putbuf(int c, struct putchar_arg *ap)
{
/* Check if no console output buffer was provided. */
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
if (ap->p_bufr == NULL) {
/* Output direct to the console. */
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
if (ap->flags & TOCONS)
cnputc(c);
if (ap->flags & TOLOG)
msglogchar(c, ap->pri);
} else {
/* Buffer the character: */
*ap->p_next++ = c;
ap->remain--;
/* Always leave the buffer zero terminated. */
*ap->p_next = '\0';
/* Check if the buffer needs to be flushed. */
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
if (ap->remain == 2 || c == '\n') {
if (ap->flags & TOLOG)
msglogstr(ap->p_bufr, ap->pri, /*filter_cr*/1);
if (ap->flags & TOCONS) {
if ((panicstr == NULL) && (constty != NULL))
msgbuf_addstr(&consmsgbuf, -1,
ap->p_bufr, /*filter_cr*/ 0);
if ((constty == NULL) ||(always_console_output))
cnputs(ap->p_bufr);
}
ap->p_next = ap->p_bufr;
ap->remain = ap->n_bufr;
*ap->p_next = '\0';
}
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
/*
* Since we fill the buffer up one character at a time,
* this should not happen. We should always catch it when
* ap->remain == 2 (if not sooner due to a newline), flush
* the buffer and move on. One way this could happen is
* if someone sets PRINTF_BUFR_SIZE to 1 or something
* similarly silly.
*/
KASSERT(ap->remain > 2, ("Bad buffer logic, remain = %zd",
ap->remain));
}
}
/*
* Print a character on console or users terminal. If destination is
* the console then the last bunch of characters are saved in msgbuf for
* inspection later.
*/
static void
putchar(int c, void *arg)
{
struct putchar_arg *ap = (struct putchar_arg*) arg;
struct tty *tp = ap->tty;
int flags = ap->flags;
/* Don't use the tty code after a panic or while in ddb. */
if (kdb_active) {
if (c != '\0')
cnputc(c);
return;
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
}
if ((flags & TOTTY) && tp != NULL && panicstr == NULL)
tty_putchar(tp, c);
if ((flags & (TOCONS | TOLOG)) && c != '\0')
putbuf(c, ap);
}
/*
* Scaled down version of sprintf(3).
*/
int
sprintf(char *buf, const char *cfmt, ...)
{
int retval;
va_list ap;
va_start(ap, cfmt);
retval = kvprintf(cfmt, NULL, (void *)buf, 10, ap);
buf[retval] = '\0';
va_end(ap);
return (retval);
}
/*
* Scaled down version of vsprintf(3).
*/
int
vsprintf(char *buf, const char *cfmt, va_list ap)
{
int retval;
retval = kvprintf(cfmt, NULL, (void *)buf, 10, ap);
buf[retval] = '\0';
return (retval);
}
/*
* Scaled down version of snprintf(3).
*/
int
snprintf(char *str, size_t size, const char *format, ...)
{
int retval;
va_list ap;
va_start(ap, format);
retval = vsnprintf(str, size, format, ap);
va_end(ap);
return(retval);
}
/*
* Scaled down version of vsnprintf(3).
*/
int
vsnprintf(char *str, size_t size, const char *format, va_list ap)
{
struct snprintf_arg info;
int retval;
info.str = str;
info.remain = size;
retval = kvprintf(format, snprintf_func, &info, 10, ap);
if (info.remain >= 1)
*info.str++ = '\0';
return (retval);
}
/*
* Kernel version which takes radix argument vsnprintf(3).
*/
int
vsnrprintf(char *str, size_t size, int radix, const char *format, va_list ap)
{
struct snprintf_arg info;
int retval;
info.str = str;
info.remain = size;
retval = kvprintf(format, snprintf_func, &info, radix, ap);
if (info.remain >= 1)
*info.str++ = '\0';
return (retval);
}
static void
snprintf_func(int ch, void *arg)
{
struct snprintf_arg *const info = arg;
if (info->remain >= 2) {
*info->str++ = ch;
info->remain--;
}
}
/*
* Put a NUL-terminated ASCII number (base <= 36) in a buffer in reverse
* order; return an optional length and a pointer to the last character
* written in the buffer (i.e., the first character of the string).
* The buffer pointed to by `nbuf' must have length >= MAXNBUF.
*/
static char *
ksprintn(char *nbuf, uintmax_t num, int base, int *lenp, int upper)
{
char *p, c;
p = nbuf;
*p = '\0';
do {
c = hex2ascii(num % base);
*++p = upper ? toupper(c) : c;
} while (num /= base);
if (lenp)
*lenp = p - nbuf;
return (p);
}
1994-05-24 10:09:53 +00:00
/*
* Scaled down version of printf(3).
*
* Two additional formats:
*
* The format %b is supported to decode error registers.
* Its usage is:
*
* printf("reg=%b\n", regval, "<base><arg>*");
*
* where <base> is the output base expressed as a control character, e.g.
* \10 gives octal; \20 gives hex. Each arg is a sequence of characters,
* the first of which gives the bit number to be inspected (origin 1), and
* the next characters (up to a control character, i.e. a character <= 32),
* give the name of the register. Thus:
*
* kvprintf("reg=%b\n", 3, "\10\2BITTWO\1BITONE\n");
1994-05-24 10:09:53 +00:00
*
* would produce output:
*
* reg=3<BITTWO,BITONE>
*
* XXX: %D -- Hexdump, takes pointer and separator string:
* ("%6D", ptr, ":") -> XX:XX:XX:XX:XX:XX
* ("%*D", len, ptr, " " -> XX XX XX XX ...
1994-05-24 10:09:53 +00:00
*/
int
kvprintf(char const *fmt, void (*func)(int, void*), void *arg, int radix, va_list ap)
1994-05-24 10:09:53 +00:00
{
#define PCHAR(c) {int cc=(c); if (func) (*func)(cc,arg); else *d++ = cc; retval++; }
char nbuf[MAXNBUF];
char *d;
const char *p, *percent, *q;
u_char *up;
int ch, n;
uintmax_t num;
int base, lflag, qflag, tmp, width, ladjust, sharpflag, neg, sign, dot;
int cflag, hflag, jflag, tflag, zflag;
int dwidth, upper;
1994-05-24 10:09:53 +00:00
char padc;
int stop = 0, retval = 0;
num = 0;
if (!func)
d = (char *) arg;
else
d = NULL;
if (fmt == NULL)
fmt = "(fmt null)\n";
1996-01-22 13:21:33 +00:00
if (radix < 2 || radix > 36)
1996-01-22 13:21:33 +00:00
radix = 10;
1994-05-24 10:09:53 +00:00
for (;;) {
padc = ' ';
width = 0;
while ((ch = (u_char)*fmt++) != '%' || stop) {
2002-06-02 21:55:58 +00:00
if (ch == '\0')
return (retval);
PCHAR(ch);
1994-05-24 10:09:53 +00:00
}
percent = fmt - 1;
qflag = 0; lflag = 0; ladjust = 0; sharpflag = 0; neg = 0;
sign = 0; dot = 0; dwidth = 0; upper = 0;
cflag = 0; hflag = 0; jflag = 0; tflag = 0; zflag = 0;
reswitch: switch (ch = (u_char)*fmt++) {
1996-01-16 18:08:57 +00:00
case '.':
dot = 1;
goto reswitch;
case '#':
sharpflag = 1;
goto reswitch;
case '+':
sign = 1;
goto reswitch;
case '-':
ladjust = 1;
goto reswitch;
case '%':
PCHAR(ch);
break;
case '*':
1996-01-19 21:05:52 +00:00
if (!dot) {
width = va_arg(ap, int);
if (width < 0) {
ladjust = !ladjust;
width = -width;
}
} else {
dwidth = va_arg(ap, int);
}
goto reswitch;
1994-05-24 10:09:53 +00:00
case '0':
1996-01-18 10:23:02 +00:00
if (!dot) {
padc = '0';
goto reswitch;
}
1994-05-24 10:09:53 +00:00
case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
1996-01-18 10:23:02 +00:00
for (n = 0;; ++fmt) {
n = n * 10 + ch - '0';
ch = *fmt;
if (ch < '0' || ch > '9')
break;
}
if (dot)
dwidth = n;
else
width = n;
1994-05-24 10:09:53 +00:00
goto reswitch;
case 'b':
num = (u_int)va_arg(ap, int);
1994-05-24 10:09:53 +00:00
p = va_arg(ap, char *);
for (q = ksprintn(nbuf, num, *p++, NULL, 0); *q;)
PCHAR(*q--);
1994-05-24 10:09:53 +00:00
if (num == 0)
1994-05-24 10:09:53 +00:00
break;
for (tmp = 0; *p;) {
n = *p++;
if (num & (1 << (n - 1))) {
PCHAR(tmp ? ',' : '<');
1994-05-24 10:09:53 +00:00
for (; (n = *p) > ' '; ++p)
PCHAR(n);
1994-05-24 10:09:53 +00:00
tmp = 1;
} else
for (; *p > ' '; ++p)
continue;
}
if (tmp)
PCHAR('>');
1994-05-24 10:09:53 +00:00
break;
case 'c':
PCHAR(va_arg(ap, int));
1994-05-24 10:09:53 +00:00
break;
case 'D':
up = va_arg(ap, u_char *);
p = va_arg(ap, char *);
if (!width)
width = 16;
while(width--) {
PCHAR(hex2ascii(*up >> 4));
PCHAR(hex2ascii(*up & 0x0f));
up++;
if (width)
for (q=p;*q;q++)
PCHAR(*q);
}
break;
1994-05-24 10:09:53 +00:00
case 'd':
case 'i':
1994-05-24 10:09:53 +00:00
base = 10;
sign = 1;
goto handle_sign;
case 'h':
if (hflag) {
hflag = 0;
cflag = 1;
} else
hflag = 1;
goto reswitch;
case 'j':
jflag = 1;
goto reswitch;
case 'l':
if (lflag) {
lflag = 0;
qflag = 1;
} else
lflag = 1;
goto reswitch;
case 'n':
if (jflag)
*(va_arg(ap, intmax_t *)) = retval;
else if (qflag)
*(va_arg(ap, quad_t *)) = retval;
else if (lflag)
*(va_arg(ap, long *)) = retval;
else if (zflag)
*(va_arg(ap, size_t *)) = retval;
else if (hflag)
*(va_arg(ap, short *)) = retval;
else if (cflag)
*(va_arg(ap, char *)) = retval;
else
*(va_arg(ap, int *)) = retval;
break;
case 'o':
1994-05-24 10:09:53 +00:00
base = 8;
goto handle_nosign;
case 'p':
base = 16;
sharpflag = (width == 0);
sign = 0;
num = (uintptr_t)va_arg(ap, void *);
goto number;
case 'q':
qflag = 1;
goto reswitch;
case 'r':
base = radix;
if (sign)
goto handle_sign;
goto handle_nosign;
case 's':
p = va_arg(ap, char *);
if (p == NULL)
p = "(null)";
1996-01-16 18:08:57 +00:00
if (!dot)
n = strlen (p);
else
1996-01-18 10:23:02 +00:00
for (n = 0; n < dwidth && p[n]; n++)
1996-01-16 18:08:57 +00:00
continue;
1996-01-18 10:23:02 +00:00
1996-01-16 18:08:57 +00:00
width -= n;
1996-01-18 10:23:02 +00:00
if (!ladjust && width > 0)
while (width--)
PCHAR(padc);
1996-01-16 18:08:57 +00:00
while (n--)
PCHAR(*p++);
if (ladjust && width > 0)
while (width--)
PCHAR(padc);
break;
case 't':
tflag = 1;
goto reswitch;
1994-05-24 10:09:53 +00:00
case 'u':
base = 10;
goto handle_nosign;
case 'X':
upper = 1;
case 'x':
1994-05-24 10:09:53 +00:00
base = 16;
goto handle_nosign;
case 'y':
base = 16;
sign = 1;
goto handle_sign;
case 'z':
zflag = 1;
goto reswitch;
handle_nosign:
sign = 0;
if (jflag)
num = va_arg(ap, uintmax_t);
else if (qflag)
num = va_arg(ap, u_quad_t);
else if (tflag)
num = va_arg(ap, ptrdiff_t);
else if (lflag)
num = va_arg(ap, u_long);
else if (zflag)
num = va_arg(ap, size_t);
else if (hflag)
num = (u_short)va_arg(ap, int);
else if (cflag)
num = (u_char)va_arg(ap, int);
else
num = va_arg(ap, u_int);
goto number;
handle_sign:
if (jflag)
num = va_arg(ap, intmax_t);
else if (qflag)
num = va_arg(ap, quad_t);
else if (tflag)
num = va_arg(ap, ptrdiff_t);
else if (lflag)
num = va_arg(ap, long);
else if (zflag)
num = va_arg(ap, ssize_t);
else if (hflag)
num = (short)va_arg(ap, int);
else if (cflag)
num = (char)va_arg(ap, int);
else
num = va_arg(ap, int);
number:
if (sign && (intmax_t)num < 0) {
neg = 1;
num = -(intmax_t)num;
}
p = ksprintn(nbuf, num, base, &n, upper);
tmp = 0;
if (sharpflag && num != 0) {
if (base == 8)
tmp++;
else if (base == 16)
tmp += 2;
}
if (neg)
tmp++;
if (!ladjust && padc == '0')
dwidth = width - tmp;
width -= tmp + imax(dwidth, n);
dwidth -= n;
if (!ladjust)
while (width-- > 0)
PCHAR(' ');
if (neg)
PCHAR('-');
if (sharpflag && num != 0) {
if (base == 8) {
PCHAR('0');
} else if (base == 16) {
PCHAR('0');
PCHAR('x');
}
}
while (dwidth-- > 0)
PCHAR('0');
while (*p)
PCHAR(*p--);
if (ladjust)
while (width-- > 0)
PCHAR(' ');
1994-05-24 10:09:53 +00:00
break;
default:
while (percent < fmt)
PCHAR(*percent++);
/*
* Since we ignore a formatting argument it is no
* longer safe to obey the remaining formatting
* arguments as the arguments will no longer match
* the format specs.
*/
stop = 1;
break;
1994-05-24 10:09:53 +00:00
}
}
#undef PCHAR
1994-05-24 10:09:53 +00:00
}
/*
* Put character in log buffer with a particular priority.
1994-05-24 10:09:53 +00:00
*/
static void
msglogchar(int c, int pri)
1994-05-24 10:09:53 +00:00
{
static int lastpri = -1;
static int dangling;
char nbuf[MAXNBUF];
char *p;
1994-05-24 10:09:53 +00:00
if (!msgbufmapped)
return;
if (c == '\0' || c == '\r')
return;
if (pri != -1 && pri != lastpri) {
if (dangling) {
msgbuf_addchar(msgbufp, '\n');
dangling = 0;
}
msgbuf_addchar(msgbufp, '<');
for (p = ksprintn(nbuf, (uintmax_t)pri, 10, NULL, 0); *p;)
msgbuf_addchar(msgbufp, *p--);
msgbuf_addchar(msgbufp, '>');
lastpri = pri;
}
msgbuf_addchar(msgbufp, c);
if (c == '\n') {
dangling = 0;
lastpri = -1;
} else {
dangling = 1;
}
}
Fix apparent garbage in the message buffer. While we have had a fix in place (options PRINTF_BUFR_SIZE=128) to fix scrambled console output, the message buffer and syslog were still getting log messages one character at a time. While all of the characters still made it into the log (courtesy of atomic operations), they were often interleaved when there were multiple threads writing to the buffer at the same time. This fixes message buffer accesses to use buffering logic as well, so that strings that are less than PRINTF_BUFR_SIZE will be put into the message buffer atomically. So now dmesg output should look the same as console output. subr_msgbuf.c: Convert most message buffer calls to use a new spin lock instead of atomic variables in some places. Add a new routine, msgbuf_addstr(), that adds a NUL-terminated string to a message buffer. This takes a priority argument, which allows us to eliminate some races (at least in the the string at a time case) that are present in the implementation of msglogchar(). (dangling and lastpri are static variables, and are subject to races when multiple callers are present.) msgbuf_addstr() also allows the caller to request that carriage returns be stripped out of the string. This matches the behavior of msglogchar(), but in testing so far it doesn't appear that any newlines are being stripped out. So the carriage return removal functionality may be a candidate for removal later on if further analysis shows that it isn't necessary. subr_prf.c: Add a new msglogstr() routine that calls msgbuf_logstr(). Rename putcons() to putbuf(). This now handles buffered output to the message log as well as the console. Also, remove the logic in putcons() (now putbuf()) that added a carriage return before a newline. The console path was the only path that needed it, and cnputc() (called by cnputs()) already adds a carriage return. So this duplication resulted in kernel-generated console output lines ending in '\r''\r''\n'. Refactor putchar() to handle the new buffering scheme. Add buffering to log(). Change log_console() to use msglogstr() instead of msglogchar(). Don't add extra newlines by default in log_console(). Hide that behavior behind a tunable/sysctl (kern.log_console_add_linefeed) for those who would like the old behavior. The old behavior led to the insertion of extra newlines for log output for programs that print out a string, and then a trailing newline on a separate write. (This is visible with dmesg -a.) msgbuf.h: Add a prototype for msgbuf_addstr(). Add three new fields to struct msgbuf, msg_needsnl, msg_lastpri and msg_lock. The first two are needed for log message functionality previously handled by msglogchar(). (Which is still active if buffering isn't enabled.) Include sys/lock.h and sys/mutex.h for the new mutex. Reviewed by: gibbs
2011-05-31 17:29:58 +00:00
static void
msglogstr(char *str, int pri, int filter_cr)
{
if (!msgbufmapped)
return;
msgbuf_addstr(msgbufp, pri, str, filter_cr);
}
void
msgbufinit(void *ptr, int size)
{
char *cp;
static struct msgbuf *oldp = NULL;
size -= sizeof(*msgbufp);
cp = (char *)ptr;
msgbufp = (struct msgbuf *)(cp + size);
msgbuf_reinit(msgbufp, cp, size);
if (msgbufmapped && oldp != msgbufp)
msgbuf_copy(oldp, msgbufp);
msgbufmapped = 1;
oldp = msgbufp;
}
static int unprivileged_read_msgbuf = 1;
SYSCTL_INT(_security_bsd, OID_AUTO, unprivileged_read_msgbuf,
CTLFLAG_RW, &unprivileged_read_msgbuf, 0,
"Unprivileged processes may read the kernel message buffer");
/* Sysctls for accessing/clearing the msgbuf */
static int
sysctl_kern_msgbuf(SYSCTL_HANDLER_ARGS)
{
char buf[128];
u_int seq;
int error, len;
if (!unprivileged_read_msgbuf) {
error = priv_check(req->td, PRIV_MSGBUF);
if (error)
return (error);
}
/* Read the whole buffer, one chunk at a time. */
mtx_lock(&msgbuf_lock);
msgbuf_peekbytes(msgbufp, NULL, 0, &seq);
for (;;) {
len = msgbuf_peekbytes(msgbufp, buf, sizeof(buf), &seq);
mtx_unlock(&msgbuf_lock);
if (len == 0)
return (0);
error = sysctl_handle_opaque(oidp, buf, len, req);
if (error)
return (error);
mtx_lock(&msgbuf_lock);
}
}
SYSCTL_PROC(_kern, OID_AUTO, msgbuf,
CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE,
NULL, 0, sysctl_kern_msgbuf, "A", "Contents of kernel message buffer");
static int msgbuf_clearflag;
static int
sysctl_kern_msgbuf_clear(SYSCTL_HANDLER_ARGS)
{
int error;
error = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req);
if (!error && req->newptr) {
mtx_lock(&msgbuf_lock);
msgbuf_clear(msgbufp);
mtx_unlock(&msgbuf_lock);
msgbuf_clearflag = 0;
}
return (error);
}
SYSCTL_PROC(_kern, OID_AUTO, msgbuf_clear,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE | CTLFLAG_MPSAFE,
&msgbuf_clearflag, 0, sysctl_kern_msgbuf_clear, "I",
"Clear kernel message buffer");
#ifdef DDB
DB_SHOW_COMMAND(msgbuf, db_show_msgbuf)
{
int i, j;
if (!msgbufmapped) {
db_printf("msgbuf not mapped yet\n");
return;
}
db_printf("msgbufp = %p\n", msgbufp);
db_printf("magic = %x, size = %d, r= %u, w = %u, ptr = %p, cksum= %u\n",
msgbufp->msg_magic, msgbufp->msg_size, msgbufp->msg_rseq,
msgbufp->msg_wseq, msgbufp->msg_ptr, msgbufp->msg_cksum);
for (i = 0; i < msgbufp->msg_size && !db_pager_quit; i++) {
j = MSGBUF_SEQ_TO_POS(msgbufp, i + msgbufp->msg_rseq);
db_printf("%c", msgbufp->msg_ptr[j]);
}
db_printf("\n");
}
#endif /* DDB */
void
2005-04-06 10:14:13 +00:00
hexdump(const void *ptr, int length, const char *hdr, int flags)
{
int i, j, k;
int cols;
2005-04-06 10:14:13 +00:00
const unsigned char *cp;
char delim;
if ((flags & HD_DELIM_MASK) != 0)
delim = (flags & HD_DELIM_MASK) >> 8;
else
delim = ' ';
if ((flags & HD_COLUMN_MASK) != 0)
cols = flags & HD_COLUMN_MASK;
else
cols = 16;
cp = ptr;
for (i = 0; i < length; i+= cols) {
if (hdr != NULL)
printf("%s", hdr);
if ((flags & HD_OMIT_COUNT) == 0)
printf("%04x ", i);
if ((flags & HD_OMIT_HEX) == 0) {
for (j = 0; j < cols; j++) {
k = i + j;
if (k < length)
printf("%c%02x", delim, cp[k]);
else
printf(" ");
}
}
if ((flags & HD_OMIT_CHARS) == 0) {
printf(" |");
for (j = 0; j < cols; j++) {
k = i + j;
if (k >= length)
printf(" ");
else if (cp[k] >= ' ' && cp[k] <= '~')
printf("%c", cp[k]);
else
printf(".");
}
printf("|");
}
printf("\n");
}
}