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
|
|
|
/*-
|
|
|
|
* Copyright (c) 2008 Ed Schouten <ed@FreeBSD.org>
|
2006-01-26 01:30:34 +00:00
|
|
|
* All rights reserved.
|
|
|
|
*
|
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
|
|
|
* Portions of this software were developed under sponsorship from Snow
|
|
|
|
* B.V., the Netherlands.
|
2006-01-26 01:30:34 +00:00
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*
|
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
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
2006-01-26 01:30:34 +00:00
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
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
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
2006-01-26 01:30:34 +00:00
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
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
|
|
|
/* Add compatibility bits for FreeBSD. */
|
|
|
|
#define PTS_COMPAT
|
2009-09-06 10:27:45 +00:00
|
|
|
/* Add pty(4) compat bits. */
|
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
|
|
|
#define PTS_EXTERNAL
|
|
|
|
/* Add bits to make Linux binaries work. */
|
|
|
|
#define PTS_LINUX
|
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/lock.h>
|
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
|
|
|
#include <sys/condvar.h>
|
|
|
|
#include <sys/conf.h>
|
2006-01-26 01:30:34 +00:00
|
|
|
#include <sys/fcntl.h>
|
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
|
|
|
#include <sys/file.h>
|
|
|
|
#include <sys/filedesc.h>
|
|
|
|
#include <sys/filio.h>
|
2006-01-26 01:30:34 +00:00
|
|
|
#include <sys/kernel.h>
|
2009-04-25 10:05:55 +00:00
|
|
|
#include <sys/limits.h>
|
2006-01-26 01:30:34 +00:00
|
|
|
#include <sys/malloc.h>
|
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
|
|
|
#include <sys/poll.h>
|
|
|
|
#include <sys/proc.h>
|
2011-04-02 15:02:42 +00:00
|
|
|
#include <sys/racct.h>
|
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
|
|
|
#include <sys/resourcevar.h>
|
|
|
|
#include <sys/serial.h>
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#include <sys/syscall.h>
|
|
|
|
#include <sys/syscallsubr.h>
|
2009-04-25 10:05:55 +00:00
|
|
|
#include <sys/sysctl.h>
|
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
|
|
|
#include <sys/sysent.h>
|
|
|
|
#include <sys/sysproto.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/tty.h>
|
|
|
|
#include <sys/ttycom.h>
|
2006-01-26 01:30:34 +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
|
|
|
#include <machine/stdarg.h>
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2009-04-25 10:05:55 +00:00
|
|
|
/*
|
|
|
|
* Our utmp(5) format is limited to 8-byte TTY line names. This means
|
|
|
|
* we can at most allocate 1000 pseudo-terminals ("pts/999"). Allow
|
|
|
|
* users to increase this number, assuming they have manually increased
|
|
|
|
* UT_LINESIZE.
|
|
|
|
*/
|
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
|
|
|
static struct unrhdr *pts_pool;
|
2006-01-26 01:30:34 +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
|
|
|
static MALLOC_DEFINE(M_PTS, "pts", "pseudo tty device");
|
2006-01-26 01:30:34 +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
|
|
|
/*
|
|
|
|
* Per-PTS structure.
|
2006-01-26 01:30:34 +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
|
|
|
* List of locks
|
|
|
|
* (t) locked by tty_lock()
|
|
|
|
* (c) const until freeing
|
2006-01-26 01:30:34 +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
|
|
|
struct pts_softc {
|
|
|
|
int pts_unit; /* (c) Device unit number. */
|
|
|
|
unsigned int pts_flags; /* (t) Device flags. */
|
2008-09-23 17:12:25 +00:00
|
|
|
#define PTS_PKT 0x1 /* Packet mode. */
|
|
|
|
#define PTS_FINISHED 0x2 /* Return errors on read()/write(). */
|
2008-09-04 16:39:02 +00:00
|
|
|
char pts_pkt; /* (t) Unread packet mode data. */
|
2006-01-26 01:30:34 +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
|
|
|
struct cv pts_inwait; /* (t) Blocking write() on master. */
|
|
|
|
struct selinfo pts_inpoll; /* (t) Select queue for write(). */
|
|
|
|
struct cv pts_outwait; /* (t) Blocking read() on master. */
|
|
|
|
struct selinfo pts_outpoll; /* (t) Select queue for read(). */
|
2006-01-26 01:30:34 +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
|
|
|
#ifdef PTS_EXTERNAL
|
|
|
|
struct cdev *pts_cdev; /* (c) Master device node. */
|
|
|
|
#endif /* PTS_EXTERNAL */
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2011-03-03 17:33:22 +00:00
|
|
|
struct ucred *pts_cred; /* (c) Resource limit. */
|
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
|
|
|
};
|
2006-01-26 01:30:34 +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
|
|
|
/*
|
|
|
|
* Controller-side file operations.
|
|
|
|
*/
|
2006-01-26 01:30:34 +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
|
|
|
static int
|
|
|
|
ptsdev_read(struct file *fp, struct uio *uio, struct ucred *active_cred,
|
|
|
|
int flags, struct thread *td)
|
|
|
|
{
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
2008-09-04 16:39:02 +00:00
|
|
|
int error = 0;
|
|
|
|
char pkt;
|
2006-01-26 01:30:34 +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
|
|
|
if (uio->uio_resid == 0)
|
|
|
|
return (0);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
tty_lock(tp);
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
/*
|
|
|
|
* Implement packet mode. When packet mode is turned on,
|
|
|
|
* the first byte contains a bitmask of events that
|
|
|
|
* occured (start, stop, flush, window size, etc).
|
|
|
|
*/
|
|
|
|
if (psc->pts_flags & PTS_PKT && psc->pts_pkt) {
|
|
|
|
pkt = psc->pts_pkt;
|
|
|
|
psc->pts_pkt = 0;
|
|
|
|
tty_unlock(tp);
|
|
|
|
|
|
|
|
error = ureadc(pkt, uio);
|
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
|
|
|
return (error);
|
2008-09-04 16:39:02 +00:00
|
|
|
}
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
/*
|
|
|
|
* Transmit regular data.
|
|
|
|
*
|
|
|
|
* XXX: We shouldn't use ttydisc_getc_poll()! Even
|
|
|
|
* though in this implementation, there is likely going
|
|
|
|
* to be data, we should just call ttydisc_getc_uio()
|
|
|
|
* and use its return value to sleep.
|
|
|
|
*/
|
|
|
|
if (ttydisc_getc_poll(tp)) {
|
|
|
|
if (psc->pts_flags & PTS_PKT) {
|
|
|
|
/*
|
|
|
|
* XXX: Small race. Fortunately PTY
|
|
|
|
* consumers aren't multithreaded.
|
|
|
|
*/
|
|
|
|
|
|
|
|
tty_unlock(tp);
|
|
|
|
error = ureadc(TIOCPKT_DATA, uio);
|
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
tty_lock(tp);
|
|
|
|
}
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
error = ttydisc_getc_uio(tp, uio);
|
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
|
|
|
break;
|
2008-09-04 16:39:02 +00:00
|
|
|
}
|
2006-01-26 01:30:34 +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
|
|
|
/* Maybe the device isn't used anyway. */
|
2008-09-23 17:12:25 +00:00
|
|
|
if (psc->pts_flags & PTS_FINISHED)
|
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
|
|
|
break;
|
2006-01-26 01:30:34 +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
|
|
|
/* Wait for more data. */
|
|
|
|
if (fp->f_flag & O_NONBLOCK) {
|
|
|
|
error = EWOULDBLOCK;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
error = cv_wait_sig(&psc->pts_outwait, tp->t_mtx);
|
|
|
|
if (error != 0)
|
|
|
|
break;
|
2006-01-26 01:30:34 +00:00
|
|
|
}
|
2008-09-04 16:39:02 +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
|
|
|
tty_unlock(tp);
|
2006-01-26 01:30:34 +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
|
|
|
return (error);
|
2006-01-26 01:30:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
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
|
|
|
ptsdev_write(struct file *fp, struct uio *uio, struct ucred *active_cred,
|
|
|
|
int flags, struct thread *td)
|
2006-01-26 01:30:34 +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
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
char ib[256], *ibstart;
|
|
|
|
size_t iblen, rintlen;
|
|
|
|
int error = 0;
|
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
if (uio->uio_resid == 0)
|
|
|
|
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
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
for (;;) {
|
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
|
|
|
ibstart = ib;
|
|
|
|
iblen = MIN(uio->uio_resid, sizeof ib);
|
|
|
|
error = uiomove(ib, iblen, uio);
|
2008-09-04 16:39:02 +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
|
|
|
tty_lock(tp);
|
2009-08-02 14:25:26 +00:00
|
|
|
if (error != 0) {
|
|
|
|
iblen = 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
|
|
|
goto done;
|
2009-08-02 14:25:26 +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
|
|
|
|
|
|
|
/*
|
|
|
|
* When possible, avoid the slow path. rint_bypass()
|
|
|
|
* copies all input to the input queue at once.
|
|
|
|
*/
|
2008-09-04 16:39:02 +00:00
|
|
|
MPASS(iblen > 0);
|
|
|
|
do {
|
2009-08-23 08:04:40 +00:00
|
|
|
rintlen = ttydisc_rint_simple(tp, ibstart, iblen);
|
|
|
|
ibstart += rintlen;
|
|
|
|
iblen -= rintlen;
|
|
|
|
if (iblen == 0) {
|
|
|
|
/* All data written. */
|
|
|
|
break;
|
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
|
|
|
}
|
|
|
|
|
|
|
|
/* Maybe the device isn't used anyway. */
|
2008-09-23 17:12:25 +00:00
|
|
|
if (psc->pts_flags & PTS_FINISHED) {
|
2008-08-22 10:40:21 +00:00
|
|
|
error = EIO;
|
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
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Wait for more data. */
|
|
|
|
if (fp->f_flag & O_NONBLOCK) {
|
|
|
|
error = EWOULDBLOCK;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Wake up users on the slave side. */
|
|
|
|
ttydisc_rint_done(tp);
|
|
|
|
error = cv_wait_sig(&psc->pts_inwait, tp->t_mtx);
|
|
|
|
if (error != 0)
|
|
|
|
goto done;
|
2008-09-04 16:39:02 +00:00
|
|
|
} while (iblen > 0);
|
|
|
|
|
|
|
|
if (uio->uio_resid == 0)
|
|
|
|
break;
|
|
|
|
tty_unlock(tp);
|
2006-01-26 01:30:34 +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
|
|
|
|
|
|
|
done: ttydisc_rint_done(tp);
|
|
|
|
tty_unlock(tp);
|
2009-08-02 14:25:26 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't account for the part of the buffer that we couldn't
|
|
|
|
* pass to the TTY.
|
|
|
|
*/
|
|
|
|
uio->uio_resid += iblen;
|
2006-01-26 01:30:34 +00:00
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
2008-12-11 21:44:02 +00:00
|
|
|
static int
|
|
|
|
ptsdev_truncate(struct file *fp, off_t length, struct ucred *active_cred,
|
|
|
|
struct thread *td)
|
|
|
|
{
|
|
|
|
|
|
|
|
return (EINVAL);
|
|
|
|
}
|
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
static int
|
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
|
|
|
ptsdev_ioctl(struct file *fp, u_long cmd, void *data,
|
|
|
|
struct ucred *active_cred, struct thread *td)
|
2006-01-26 01:30:34 +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
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
int error = 0, sig;
|
2006-01-26 01:30:34 +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
|
|
|
switch (cmd) {
|
|
|
|
case FIONBIO:
|
|
|
|
/* This device supports non-blocking operation. */
|
|
|
|
return (0);
|
2008-12-13 07:23:55 +00:00
|
|
|
case FIONREAD:
|
|
|
|
tty_lock(tp);
|
|
|
|
if (psc->pts_flags & PTS_FINISHED) {
|
|
|
|
/* Force read() to be called. */
|
|
|
|
*(int *)data = 1;
|
|
|
|
} else {
|
|
|
|
*(int *)data = ttydisc_getc_poll(tp);
|
|
|
|
}
|
|
|
|
tty_unlock(tp);
|
|
|
|
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
|
|
|
case FIODGNAME: {
|
|
|
|
struct fiodgname_arg *fgn;
|
|
|
|
const char *p;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* Reverse device name lookups, for ptsname() and ttyname(). */
|
|
|
|
fgn = data;
|
2009-02-19 17:54:42 +00:00
|
|
|
p = tty_devname(tp);
|
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
|
|
|
i = strlen(p) + 1;
|
|
|
|
if (i > fgn->len)
|
|
|
|
return (EINVAL);
|
|
|
|
return copyout(p, fgn->buf, i);
|
|
|
|
}
|
2011-06-26 18:26:20 +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
|
|
|
/*
|
|
|
|
* We need to implement TIOCGPGRP and TIOCGSID here again. When
|
|
|
|
* called on the pseudo-terminal master, it should not check if
|
|
|
|
* the terminal is the foreground terminal of the calling
|
|
|
|
* process.
|
|
|
|
*
|
|
|
|
* TIOCGETA is also implemented here. Various Linux PTY routines
|
|
|
|
* often call isatty(), which is implemented by tcgetattr().
|
|
|
|
*/
|
|
|
|
#ifdef PTS_LINUX
|
|
|
|
case TIOCGETA:
|
|
|
|
/* Obtain terminal flags through tcgetattr(). */
|
|
|
|
tty_lock(tp);
|
2009-03-01 09:50:13 +00:00
|
|
|
*(struct termios*)data = tp->t_termios;
|
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(tp);
|
|
|
|
return (0);
|
|
|
|
#endif /* PTS_LINUX */
|
|
|
|
case TIOCSETAF:
|
|
|
|
case TIOCSETAW:
|
|
|
|
/*
|
|
|
|
* We must make sure we turn tcsetattr() calls of TCSAFLUSH and
|
|
|
|
* TCSADRAIN into something different. If an application would
|
|
|
|
* call TCSAFLUSH or TCSADRAIN on the master descriptor, it may
|
|
|
|
* deadlock waiting for all data to be read.
|
|
|
|
*/
|
|
|
|
cmd = TIOCSETA;
|
|
|
|
break;
|
|
|
|
#if defined(PTS_COMPAT) || defined(PTS_LINUX)
|
|
|
|
case TIOCGPTN:
|
|
|
|
/*
|
|
|
|
* Get the device unit number.
|
|
|
|
*/
|
|
|
|
if (psc->pts_unit < 0)
|
|
|
|
return (ENOTTY);
|
|
|
|
*(unsigned int *)data = psc->pts_unit;
|
|
|
|
return (0);
|
|
|
|
#endif /* PTS_COMPAT || PTS_LINUX */
|
|
|
|
case TIOCGPGRP:
|
|
|
|
/* Get the foreground process group ID. */
|
|
|
|
tty_lock(tp);
|
|
|
|
if (tp->t_pgrp != NULL)
|
|
|
|
*(int *)data = tp->t_pgrp->pg_id;
|
|
|
|
else
|
|
|
|
*(int *)data = NO_PID;
|
|
|
|
tty_unlock(tp);
|
|
|
|
return (0);
|
|
|
|
case TIOCGSID:
|
|
|
|
/* Get the session leader process ID. */
|
|
|
|
tty_lock(tp);
|
|
|
|
if (tp->t_session == NULL)
|
|
|
|
error = ENOTTY;
|
|
|
|
else
|
|
|
|
*(int *)data = tp->t_session->s_sid;
|
|
|
|
tty_unlock(tp);
|
|
|
|
return (error);
|
|
|
|
case TIOCPTMASTER:
|
|
|
|
/* Yes, we are a pseudo-terminal master. */
|
|
|
|
return (0);
|
|
|
|
case TIOCSIG:
|
|
|
|
/* Signal the foreground process group. */
|
|
|
|
sig = *(int *)data;
|
|
|
|
if (sig < 1 || sig >= NSIG)
|
|
|
|
return (EINVAL);
|
|
|
|
|
|
|
|
tty_lock(tp);
|
|
|
|
tty_signal_pgrp(tp, sig);
|
|
|
|
tty_unlock(tp);
|
|
|
|
return (0);
|
|
|
|
case TIOCPKT:
|
|
|
|
/* Enable/disable packet mode. */
|
|
|
|
tty_lock(tp);
|
|
|
|
if (*(int *)data)
|
|
|
|
psc->pts_flags |= PTS_PKT;
|
|
|
|
else
|
|
|
|
psc->pts_flags &= ~PTS_PKT;
|
|
|
|
tty_unlock(tp);
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Just redirect this ioctl to the slave device. */
|
|
|
|
tty_lock(tp);
|
2010-01-04 20:59:52 +00:00
|
|
|
error = tty_ioctl(tp, cmd, data, fp->f_flag, td);
|
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(tp);
|
2009-02-19 17:54:42 +00:00
|
|
|
if (error == ENOIOCTL)
|
|
|
|
error = ENOTTY;
|
2006-01-26 01:30:34 +00:00
|
|
|
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
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
|
|
|
ptsdev_poll(struct file *fp, int events, struct ucred *active_cred,
|
|
|
|
struct thread *td)
|
2006-01-26 01:30:34 +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
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
int revents = 0;
|
2006-01-26 01:30:34 +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
|
|
|
tty_lock(tp);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-23 17:12:25 +00:00
|
|
|
if (psc->pts_flags & PTS_FINISHED) {
|
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
|
|
|
/* Slave device is not opened. */
|
|
|
|
tty_unlock(tp);
|
2009-07-08 10:21:52 +00:00
|
|
|
return ((events & (POLLIN|POLLRDNORM)) | POLLHUP);
|
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
|
|
|
}
|
2006-01-26 01:30:34 +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
|
|
|
if (events & (POLLIN|POLLRDNORM)) {
|
|
|
|
/* See if we can getc something. */
|
2008-09-04 16:39:02 +00:00
|
|
|
if (ttydisc_getc_poll(tp) ||
|
|
|
|
(psc->pts_flags & PTS_PKT && psc->pts_pkt))
|
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
|
|
|
revents |= events & (POLLIN|POLLRDNORM);
|
|
|
|
}
|
|
|
|
if (events & (POLLOUT|POLLWRNORM)) {
|
|
|
|
/* See if we can rint something. */
|
|
|
|
if (ttydisc_rint_poll(tp))
|
|
|
|
revents |= events & (POLLOUT|POLLWRNORM);
|
2006-01-26 01:30:34 +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
|
|
|
/*
|
|
|
|
* No need to check for POLLHUP here. This device cannot be used
|
|
|
|
* as a callout device, which means we always have a carrier,
|
|
|
|
* because the master is.
|
|
|
|
*/
|
2006-01-26 01:30:34 +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
|
|
|
if (revents == 0) {
|
|
|
|
/*
|
|
|
|
* This code might look misleading, but the naming of
|
|
|
|
* poll events on this side is the opposite of the slave
|
|
|
|
* device.
|
|
|
|
*/
|
|
|
|
if (events & (POLLIN|POLLRDNORM))
|
|
|
|
selrecord(td, &psc->pts_outpoll);
|
|
|
|
if (events & (POLLOUT|POLLWRNORM))
|
|
|
|
selrecord(td, &psc->pts_inpoll);
|
2006-01-26 01:30:34 +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
|
|
|
|
|
|
|
tty_unlock(tp);
|
|
|
|
|
|
|
|
return (revents);
|
2006-01-26 01:30:34 +00:00
|
|
|
}
|
|
|
|
|
2008-12-11 21:44:02 +00:00
|
|
|
/*
|
|
|
|
* kqueue support.
|
|
|
|
*/
|
|
|
|
|
|
|
|
static void
|
|
|
|
pts_kqops_read_detach(struct knote *kn)
|
|
|
|
{
|
|
|
|
struct file *fp = kn->kn_fp;
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
knlist_remove(&psc->pts_outpoll.si_note, kn, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
pts_kqops_read_event(struct knote *kn, long hint)
|
|
|
|
{
|
|
|
|
struct file *fp = kn->kn_fp;
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
if (psc->pts_flags & PTS_FINISHED) {
|
|
|
|
kn->kn_flags |= EV_EOF;
|
|
|
|
return (1);
|
|
|
|
} else {
|
|
|
|
kn->kn_data = ttydisc_getc_poll(tp);
|
|
|
|
return (kn->kn_data > 0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
pts_kqops_write_detach(struct knote *kn)
|
|
|
|
{
|
|
|
|
struct file *fp = kn->kn_fp;
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
knlist_remove(&psc->pts_inpoll.si_note, kn, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
pts_kqops_write_event(struct knote *kn, long hint)
|
|
|
|
{
|
|
|
|
struct file *fp = kn->kn_fp;
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
if (psc->pts_flags & PTS_FINISHED) {
|
|
|
|
kn->kn_flags |= EV_EOF;
|
|
|
|
return (1);
|
|
|
|
} else {
|
|
|
|
kn->kn_data = ttydisc_rint_poll(tp);
|
|
|
|
return (kn->kn_data > 0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-09-12 20:03:45 +00:00
|
|
|
static struct filterops pts_kqops_read = {
|
|
|
|
.f_isfd = 1,
|
|
|
|
.f_detach = pts_kqops_read_detach,
|
|
|
|
.f_event = pts_kqops_read_event,
|
|
|
|
};
|
|
|
|
static struct filterops pts_kqops_write = {
|
|
|
|
.f_isfd = 1,
|
|
|
|
.f_detach = pts_kqops_write_detach,
|
|
|
|
.f_event = pts_kqops_write_event,
|
|
|
|
};
|
2008-12-11 21:44:02 +00:00
|
|
|
|
|
|
|
static int
|
|
|
|
ptsdev_kqfilter(struct file *fp, struct knote *kn)
|
|
|
|
{
|
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
int error = 0;
|
|
|
|
|
|
|
|
tty_lock(tp);
|
|
|
|
|
|
|
|
switch (kn->kn_filter) {
|
|
|
|
case EVFILT_READ:
|
|
|
|
kn->kn_fop = &pts_kqops_read;
|
|
|
|
knlist_add(&psc->pts_outpoll.si_note, kn, 1);
|
|
|
|
break;
|
|
|
|
case EVFILT_WRITE:
|
|
|
|
kn->kn_fop = &pts_kqops_write;
|
|
|
|
knlist_add(&psc->pts_inpoll.si_note, kn, 1);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
error = EINVAL;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
tty_unlock(tp);
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
static int
|
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
|
|
|
ptsdev_stat(struct file *fp, struct stat *sb, struct ucred *active_cred,
|
|
|
|
struct thread *td)
|
2006-01-26 01:30:34 +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
|
|
|
struct tty *tp = fp->f_data;
|
|
|
|
#ifdef PTS_EXTERNAL
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
#endif /* PTS_EXTERNAL */
|
2008-09-21 19:24:15 +00:00
|
|
|
struct cdev *dev = tp->t_dev;
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2006-09-29 09:53:19 +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
|
|
|
* According to POSIX, we must implement an fstat(). This also
|
|
|
|
* makes this implementation compatible with Linux binaries,
|
|
|
|
* because Linux calls fstat() on the pseudo-terminal master to
|
|
|
|
* obtain st_rdev.
|
|
|
|
*
|
2008-09-21 19:24:15 +00:00
|
|
|
* XXX: POSIX also mentions we must fill in st_dev, but how?
|
2006-09-29 09:53:19 +00:00
|
|
|
*/
|
2006-01-26 01:30:34 +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
|
|
|
bzero(sb, sizeof *sb);
|
|
|
|
#ifdef PTS_EXTERNAL
|
|
|
|
if (psc->pts_cdev != NULL)
|
|
|
|
sb->st_ino = sb->st_rdev = dev2udev(psc->pts_cdev);
|
2006-01-26 20:54:49 +00:00
|
|
|
else
|
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
|
|
|
#endif /* PTS_EXTERNAL */
|
|
|
|
sb->st_ino = sb->st_rdev = tty_udev(tp);
|
2008-09-21 19:24:15 +00:00
|
|
|
|
2010-03-28 13:13:22 +00:00
|
|
|
sb->st_atim = dev->si_atime;
|
|
|
|
sb->st_ctim = dev->si_ctime;
|
|
|
|
sb->st_mtim = dev->si_mtime;
|
2008-09-21 19:24:15 +00:00
|
|
|
sb->st_uid = dev->si_uid;
|
|
|
|
sb->st_gid = dev->si_gid;
|
|
|
|
sb->st_mode = dev->si_mode | S_IFCHR;
|
2011-06-26 18:26:20 +00:00
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
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
|
|
|
ptsdev_close(struct file *fp, struct thread *td)
|
2006-01-26 01:30:34 +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
|
|
|
struct tty *tp = fp->f_data;
|
2006-01-26 01:30:34 +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
|
|
|
/* Deallocate TTY device. */
|
|
|
|
tty_lock(tp);
|
|
|
|
tty_rel_gone(tp);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2010-04-08 08:58:18 +00:00
|
|
|
/*
|
|
|
|
* Open of /dev/ptmx or /dev/ptyXX changes the type of file
|
|
|
|
* from DTYPE_VNODE to DTYPE_PTS. vn_open() increases vnode
|
|
|
|
* use count, we need to decrement it, and possibly do other
|
|
|
|
* required cleanup.
|
|
|
|
*/
|
|
|
|
if (fp->f_vnode != NULL)
|
|
|
|
return (vnops.fo_close(fp, td));
|
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
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
|
|
|
static struct fileops ptsdev_ops = {
|
|
|
|
.fo_read = ptsdev_read,
|
|
|
|
.fo_write = ptsdev_write,
|
2008-12-11 21:44:02 +00:00
|
|
|
.fo_truncate = ptsdev_truncate,
|
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
|
|
|
.fo_ioctl = ptsdev_ioctl,
|
|
|
|
.fo_poll = ptsdev_poll,
|
2008-12-11 21:44:02 +00:00
|
|
|
.fo_kqfilter = ptsdev_kqfilter,
|
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
|
|
|
.fo_stat = ptsdev_stat,
|
|
|
|
.fo_close = ptsdev_close,
|
2011-08-16 20:07:47 +00:00
|
|
|
.fo_chmod = invfo_chmod,
|
|
|
|
.fo_chown = invfo_chown,
|
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
|
|
|
.fo_flags = DFLAG_PASSABLE,
|
|
|
|
};
|
2006-01-26 01:30:34 +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
|
|
|
/*
|
|
|
|
* Driver-side hooks.
|
|
|
|
*/
|
2006-01-26 01:30:34 +00:00
|
|
|
|
|
|
|
static void
|
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
|
|
|
ptsdrv_outwakeup(struct tty *tp)
|
2006-01-26 01:30:34 +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
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
cv_broadcast(&psc->pts_outwait);
|
|
|
|
selwakeup(&psc->pts_outpoll);
|
2008-12-11 21:44:02 +00:00
|
|
|
KNOTE_LOCKED(&psc->pts_outpoll.si_note, 0);
|
2006-01-26 01:30:34 +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
|
|
|
static void
|
|
|
|
ptsdrv_inwakeup(struct tty *tp)
|
2006-01-26 01:30:34 +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
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
2006-01-26 01:30:34 +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
|
|
|
cv_broadcast(&psc->pts_inwait);
|
|
|
|
selwakeup(&psc->pts_inpoll);
|
2008-12-11 21:44:02 +00:00
|
|
|
KNOTE_LOCKED(&psc->pts_inpoll.si_note, 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
|
|
|
}
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-23 17:12:25 +00:00
|
|
|
static int
|
|
|
|
ptsdrv_open(struct tty *tp)
|
|
|
|
{
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
psc->pts_flags &= ~PTS_FINISHED;
|
|
|
|
|
|
|
|
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
|
|
|
static void
|
|
|
|
ptsdrv_close(struct tty *tp)
|
|
|
|
{
|
2008-09-23 17:12:25 +00:00
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
2006-01-26 01:30:34 +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
|
|
|
/* Wake up any blocked readers/writers. */
|
2008-12-21 21:16:57 +00:00
|
|
|
psc->pts_flags |= PTS_FINISHED;
|
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
|
|
|
ptsdrv_outwakeup(tp);
|
|
|
|
ptsdrv_inwakeup(tp);
|
|
|
|
}
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2008-09-04 16:39:02 +00:00
|
|
|
static void
|
|
|
|
ptsdrv_pktnotify(struct tty *tp, char event)
|
|
|
|
{
|
|
|
|
struct pts_softc *psc = tty_softc(tp);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Clear conflicting flags.
|
|
|
|
*/
|
|
|
|
|
|
|
|
switch (event) {
|
|
|
|
case TIOCPKT_STOP:
|
|
|
|
psc->pts_pkt &= ~TIOCPKT_START;
|
|
|
|
break;
|
|
|
|
case TIOCPKT_START:
|
|
|
|
psc->pts_pkt &= ~TIOCPKT_STOP;
|
|
|
|
break;
|
|
|
|
case TIOCPKT_NOSTOP:
|
|
|
|
psc->pts_pkt &= ~TIOCPKT_DOSTOP;
|
|
|
|
break;
|
|
|
|
case TIOCPKT_DOSTOP:
|
|
|
|
psc->pts_pkt &= ~TIOCPKT_NOSTOP;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
psc->pts_pkt |= event;
|
|
|
|
ptsdrv_outwakeup(tp);
|
|
|
|
}
|
|
|
|
|
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
|
|
|
static void
|
|
|
|
ptsdrv_free(void *softc)
|
|
|
|
{
|
|
|
|
struct pts_softc *psc = softc;
|
2006-01-26 01:30:34 +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
|
|
|
/* Make device number available again. */
|
|
|
|
if (psc->pts_unit >= 0)
|
|
|
|
free_unr(pts_pool, psc->pts_unit);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2011-03-03 17:33:22 +00:00
|
|
|
chgptscnt(psc->pts_cred->cr_ruidinfo, -1, 0);
|
2011-04-02 15:02:42 +00:00
|
|
|
racct_sub_cred(psc->pts_cred, RACCT_NPTS, 1);
|
2011-03-03 17:33:22 +00:00
|
|
|
crfree(psc->pts_cred);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
Fix a deficiency in the selinfo interface:
If a selinfo object is recorded (via selrecord()) and then it is
quickly destroyed, with the waiters missing the opportunity to awake,
at the next iteration they will find the selinfo object destroyed,
causing a PF#.
That happens because the selinfo interface has no way to drain the
waiters before to destroy the registered selinfo object. Also this
race is quite rare to get in practice, because it would require a
selrecord(), a poll request by another thread and a quick destruction
of the selrecord()'ed selinfo object.
Fix this by adding the seldrain() routine which should be called
before to destroy the selinfo objects (in order to avoid such case),
and fix the present cases where it might have already been called.
Sometimes, the context is safe enough to prevent this type of race,
like it happens in device drivers which installs selinfo objects on
poll callbacks. There, the destruction of the selinfo object happens
at driver detach time, when all the filedescriptors should be already
closed, thus there cannot be a race.
For this case, mfi(4) device driver can be set as an example, as it
implements a full correct logic for preventing this from happening.
Sponsored by: Sandvine Incorporated
Reported by: rstone
Tested by: pluknet
Reviewed by: jhb, kib
Approved by: re (bz)
MFC after: 3 weeks
2011-08-25 15:51:54 +00:00
|
|
|
seldrain(&psc->pts_inpoll);
|
|
|
|
seldrain(&psc->pts_outpoll);
|
2008-12-11 21:44:02 +00:00
|
|
|
knlist_destroy(&psc->pts_inpoll.si_note);
|
|
|
|
knlist_destroy(&psc->pts_outpoll.si_note);
|
|
|
|
|
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
|
|
|
#ifdef PTS_EXTERNAL
|
|
|
|
/* Destroy master device as well. */
|
|
|
|
if (psc->pts_cdev != NULL)
|
|
|
|
destroy_dev_sched(psc->pts_cdev);
|
|
|
|
#endif /* PTS_EXTERNAL */
|
2006-01-26 01:30:34 +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
|
|
|
free(psc, M_PTS);
|
2006-01-26 01:30:34 +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
|
|
|
static struct ttydevsw pts_class = {
|
|
|
|
.tsw_flags = TF_NOPREFIX,
|
|
|
|
.tsw_outwakeup = ptsdrv_outwakeup,
|
|
|
|
.tsw_inwakeup = ptsdrv_inwakeup,
|
2008-09-23 17:12:25 +00:00
|
|
|
.tsw_open = ptsdrv_open,
|
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
|
|
|
.tsw_close = ptsdrv_close,
|
2008-09-04 16:39:02 +00:00
|
|
|
.tsw_pktnotify = ptsdrv_pktnotify,
|
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
|
|
|
.tsw_free = ptsdrv_free,
|
|
|
|
};
|
|
|
|
|
2009-09-06 10:27:45 +00:00
|
|
|
#ifndef PTS_EXTERNAL
|
|
|
|
static
|
|
|
|
#endif /* !PTS_EXTERNAL */
|
|
|
|
int
|
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
|
|
|
pts_alloc(int fflags, struct thread *td, struct file *fp)
|
2006-01-26 01:30:34 +00:00
|
|
|
{
|
2011-04-02 15:02:42 +00:00
|
|
|
int unit, ok, error;
|
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
|
|
|
struct tty *tp;
|
|
|
|
struct pts_softc *psc;
|
|
|
|
struct proc *p = td->td_proc;
|
2011-03-03 17:33:22 +00:00
|
|
|
struct ucred *cred = td->td_ucred;
|
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
|
|
|
|
|
|
|
/* Resource limiting. */
|
|
|
|
PROC_LOCK(p);
|
2011-04-02 15:02:42 +00:00
|
|
|
error = racct_add(p, RACCT_NPTS, 1);
|
|
|
|
if (error != 0) {
|
|
|
|
PROC_UNLOCK(p);
|
|
|
|
return (EAGAIN);
|
|
|
|
}
|
2011-03-03 17:33:22 +00:00
|
|
|
ok = chgptscnt(cred->cr_ruidinfo, 1, lim_cur(p, RLIMIT_NPTS));
|
2011-04-02 15:02:42 +00:00
|
|
|
if (!ok) {
|
|
|
|
racct_sub(p, RACCT_NPTS, 1);
|
|
|
|
PROC_UNLOCK(p);
|
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
|
|
|
return (EAGAIN);
|
2011-04-02 15:02:42 +00:00
|
|
|
}
|
|
|
|
PROC_UNLOCK(p);
|
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
|
|
|
|
|
|
|
/* Try to allocate a new pts unit number. */
|
|
|
|
unit = alloc_unr(pts_pool);
|
|
|
|
if (unit < 0) {
|
2011-04-02 15:02:42 +00:00
|
|
|
racct_sub(p, RACCT_NPTS, 1);
|
2011-03-03 17:33:22 +00:00
|
|
|
chgptscnt(cred->cr_ruidinfo, -1, 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
|
|
|
return (EAGAIN);
|
2006-01-26 01:30:34 +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
|
|
|
|
|
|
|
/* Allocate TTY and softc. */
|
|
|
|
psc = malloc(sizeof(struct pts_softc), M_PTS, M_WAITOK|M_ZERO);
|
2009-12-18 20:11:29 +00:00
|
|
|
cv_init(&psc->pts_inwait, "ptsin");
|
|
|
|
cv_init(&psc->pts_outwait, "ptsout");
|
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
|
|
|
|
|
|
|
psc->pts_unit = unit;
|
2011-03-03 17:33:22 +00:00
|
|
|
psc->pts_cred = crhold(cred);
|
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
|
|
|
|
2009-05-29 06:41:23 +00:00
|
|
|
tp = tty_alloc(&pts_class, psc);
|
2009-06-10 20:59:32 +00:00
|
|
|
knlist_init_mtx(&psc->pts_inpoll.si_note, tp->t_mtx);
|
|
|
|
knlist_init_mtx(&psc->pts_outpoll.si_note, tp->t_mtx);
|
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
|
|
|
|
|
|
|
/* Expose the slave device as well. */
|
|
|
|
tty_makedev(tp, td->td_ucred, "pts/%u", psc->pts_unit);
|
|
|
|
|
|
|
|
finit(fp, fflags, DTYPE_PTS, tp, &ptsdev_ops);
|
|
|
|
|
2006-01-26 01:30:34 +00:00
|
|
|
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
|
|
|
#ifdef PTS_EXTERNAL
|
|
|
|
int
|
|
|
|
pts_alloc_external(int fflags, struct thread *td, struct file *fp,
|
|
|
|
struct cdev *dev, const char *name)
|
2006-01-26 01:30:34 +00:00
|
|
|
{
|
2011-04-02 15:02:42 +00:00
|
|
|
int ok, error;
|
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
|
|
|
struct tty *tp;
|
|
|
|
struct pts_softc *psc;
|
|
|
|
struct proc *p = td->td_proc;
|
2011-03-03 17:33:22 +00:00
|
|
|
struct ucred *cred = td->td_ucred;
|
2006-01-26 01:30:34 +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
|
|
|
/* Resource limiting. */
|
|
|
|
PROC_LOCK(p);
|
2011-04-02 15:02:42 +00:00
|
|
|
error = racct_add(p, RACCT_NPTS, 1);
|
|
|
|
if (error != 0) {
|
|
|
|
PROC_UNLOCK(p);
|
|
|
|
return (EAGAIN);
|
|
|
|
}
|
2011-03-03 17:33:22 +00:00
|
|
|
ok = chgptscnt(cred->cr_ruidinfo, 1, lim_cur(p, RLIMIT_NPTS));
|
2011-04-02 15:02:42 +00:00
|
|
|
if (!ok) {
|
|
|
|
racct_sub(p, RACCT_NPTS, 1);
|
|
|
|
PROC_UNLOCK(p);
|
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
|
|
|
return (EAGAIN);
|
2011-04-02 15:02:42 +00:00
|
|
|
}
|
|
|
|
PROC_UNLOCK(p);
|
2006-01-26 01:30:34 +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
|
|
|
/* Allocate TTY and softc. */
|
|
|
|
psc = malloc(sizeof(struct pts_softc), M_PTS, M_WAITOK|M_ZERO);
|
2009-12-18 20:11:29 +00:00
|
|
|
cv_init(&psc->pts_inwait, "ptsin");
|
|
|
|
cv_init(&psc->pts_outwait, "ptsout");
|
2006-01-26 01:30:34 +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
|
|
|
psc->pts_unit = -1;
|
|
|
|
psc->pts_cdev = dev;
|
2011-03-03 17:33:22 +00:00
|
|
|
psc->pts_cred = crhold(cred);
|
2006-01-26 01:30:34 +00:00
|
|
|
|
2009-05-29 06:41:23 +00:00
|
|
|
tp = tty_alloc(&pts_class, psc);
|
2009-06-10 20:59:32 +00:00
|
|
|
knlist_init_mtx(&psc->pts_inpoll.si_note, tp->t_mtx);
|
|
|
|
knlist_init_mtx(&psc->pts_outpoll.si_note, tp->t_mtx);
|
2006-04-28 21:39:57 +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
|
|
|
/* Expose the slave device as well. */
|
|
|
|
tty_makedev(tp, td->td_ucred, "%s", name);
|
2006-01-26 01:30:34 +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
|
|
|
finit(fp, fflags, DTYPE_PTS, tp, &ptsdev_ops);
|
2006-01-26 01:30:34 +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
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
#endif /* PTS_EXTERNAL */
|
|
|
|
|
|
|
|
int
|
|
|
|
posix_openpt(struct thread *td, struct posix_openpt_args *uap)
|
|
|
|
{
|
|
|
|
int error, fd;
|
|
|
|
struct file *fp;
|
2006-01-26 01:30:34 +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
|
|
|
* POSIX states it's unspecified when other flags are passed. We
|
|
|
|
* don't allow this.
|
2006-01-26 01:30:34 +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
|
|
|
if (uap->flags & ~(O_RDWR|O_NOCTTY))
|
|
|
|
return (EINVAL);
|
2011-06-26 18:26:20 +00:00
|
|
|
|
2011-04-01 13:28:34 +00:00
|
|
|
error = falloc(td, &fp, &fd, 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
|
|
|
if (error)
|
|
|
|
return (error);
|
|
|
|
|
|
|
|
/* Allocate the actual pseudo-TTY. */
|
|
|
|
error = pts_alloc(FFLAGS(uap->flags & O_ACCMODE), td, fp);
|
|
|
|
if (error != 0) {
|
|
|
|
fdclose(td->td_proc->p_fd, fp, fd, td);
|
|
|
|
return (error);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Pass it back to userspace. */
|
|
|
|
td->td_retval[0] = fd;
|
|
|
|
fdrop(fp, td);
|
|
|
|
|
|
|
|
return (0);
|
|
|
|
}
|
2006-01-26 01:30:34 +00:00
|
|
|
|
|
|
|
static void
|
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
|
|
|
pts_init(void *unused)
|
2006-01-26 01:30:34 +00:00
|
|
|
{
|
|
|
|
|
2009-04-25 10:05:55 +00:00
|
|
|
pts_pool = new_unrhdr(0, INT_MAX, NULL);
|
2006-01-26 01:30:34 +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
|
|
|
SYSINIT(pts, SI_SUB_DRIVERS, SI_ORDER_MIDDLE, pts_init, NULL);
|