2005-11-09 05:37:07 +00:00
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
em28xx-core.c - driver for Empia EM2800/EM2820/2840 USB video capture devices
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2005-11-09 05:38:25 +00:00
|
|
|
Copyright (C) 2005 Ludovico Cavedon <cavedon@sssup.it>
|
|
|
|
Markus Rechberger <mrechberger@gmail.com>
|
2006-04-03 10:53:40 +00:00
|
|
|
Mauro Carvalho Chehab <mchehab@infradead.org>
|
2005-11-09 05:38:25 +00:00
|
|
|
Sascha Sommer <saschasommer@freenet.de>
|
2005-11-09 05:37:07 +00:00
|
|
|
|
|
|
|
This program is free software; you can redistribute it and/or modify
|
|
|
|
it under the terms of the GNU General Public License as published by
|
|
|
|
the Free Software Foundation; either version 2 of the License, or
|
|
|
|
(at your option) any later version.
|
|
|
|
|
|
|
|
This program is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
GNU General Public License for more details.
|
|
|
|
|
|
|
|
You should have received a copy of the GNU General Public License
|
|
|
|
along with this program; if not, write to the Free Software
|
|
|
|
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <linux/init.h>
|
|
|
|
#include <linux/list.h>
|
|
|
|
#include <linux/module.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
|
|
|
#include <linux/slab.h>
|
2005-11-09 05:37:07 +00:00
|
|
|
#include <linux/usb.h>
|
|
|
|
#include <linux/vmalloc.h>
|
2008-12-29 01:18:14 +00:00
|
|
|
#include <media/v4l2-common.h>
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2005-11-09 05:38:25 +00:00
|
|
|
#include "em28xx.h"
|
2005-11-09 05:37:07 +00:00
|
|
|
|
|
|
|
/* #define ENABLE_DEBUG_ISOC_FRAMES */
|
|
|
|
|
2008-04-22 17:41:48 +00:00
|
|
|
static unsigned int core_debug;
|
2009-02-11 02:28:24 +00:00
|
|
|
module_param(core_debug, int, 0644);
|
|
|
|
MODULE_PARM_DESC(core_debug, "enable debug messages [core]");
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
#define em28xx_coredbg(fmt, arg...) do {\
|
2005-11-09 05:37:43 +00:00
|
|
|
if (core_debug) \
|
|
|
|
printk(KERN_INFO "%s %s :"fmt, \
|
2008-04-09 02:20:00 +00:00
|
|
|
dev->name, __func__ , ##arg); } while (0)
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-04-22 17:41:48 +00:00
|
|
|
static unsigned int reg_debug;
|
2009-02-11 02:28:24 +00:00
|
|
|
module_param(reg_debug, int, 0644);
|
|
|
|
MODULE_PARM_DESC(reg_debug, "enable debug messages [URB reg]");
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
#define em28xx_regdbg(fmt, arg...) do {\
|
2005-11-09 05:37:43 +00:00
|
|
|
if (reg_debug) \
|
|
|
|
printk(KERN_INFO "%s %s :"fmt, \
|
2008-04-09 02:20:00 +00:00
|
|
|
dev->name, __func__ , ##arg); } while (0)
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2009-11-07 17:51:01 +00:00
|
|
|
static int alt;
|
2005-11-09 05:37:07 +00:00
|
|
|
module_param(alt, int, 0644);
|
|
|
|
MODULE_PARM_DESC(alt, "alternate setting to use for video endpoint");
|
|
|
|
|
2009-09-01 04:19:46 +00:00
|
|
|
static unsigned int disable_vbi;
|
|
|
|
module_param(disable_vbi, int, 0644);
|
|
|
|
MODULE_PARM_DESC(disable_vbi, "disable vbi support");
|
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
/* FIXME */
|
|
|
|
#define em28xx_isocdbg(fmt, arg...) do {\
|
|
|
|
if (core_debug) \
|
|
|
|
printk(KERN_INFO "%s %s :"fmt, \
|
|
|
|
dev->name, __func__ , ##arg); } while (0)
|
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
* em28xx_read_reg_req()
|
2005-11-09 05:37:07 +00:00
|
|
|
* reads data from the usb device specifying bRequest
|
|
|
|
*/
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_read_reg_req_len(struct em28xx *dev, u8 req, u16 reg,
|
2005-11-09 05:37:07 +00:00
|
|
|
char *buf, int len)
|
|
|
|
{
|
2008-11-26 12:58:48 +00:00
|
|
|
int ret;
|
|
|
|
int pipe = usb_rcvctrlpipe(dev->udev, 0);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2006-02-07 08:49:11 +00:00
|
|
|
if (dev->state & DEV_DISCONNECTED)
|
2008-11-18 17:51:08 +00:00
|
|
|
return -ENODEV;
|
|
|
|
|
|
|
|
if (len > URB_MAX_CTRL_SIZE)
|
|
|
|
return -EINVAL;
|
2006-02-07 08:49:11 +00:00
|
|
|
|
2008-11-26 12:58:48 +00:00
|
|
|
if (reg_debug) {
|
2009-02-11 02:28:24 +00:00
|
|
|
printk(KERN_DEBUG "(pipe 0x%08x): "
|
2008-11-26 12:58:48 +00:00
|
|
|
"IN: %02x %02x %02x %02x %02x %02x %02x %02x ",
|
|
|
|
pipe,
|
|
|
|
USB_DIR_IN | USB_TYPE_VENDOR | USB_RECIP_DEVICE,
|
|
|
|
req, 0, 0,
|
|
|
|
reg & 0xff, reg >> 8,
|
|
|
|
len & 0xff, len >> 8);
|
|
|
|
}
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-11-19 09:17:44 +00:00
|
|
|
mutex_lock(&dev->ctrl_urb_lock);
|
2008-11-26 12:58:48 +00:00
|
|
|
ret = usb_control_msg(dev->udev, pipe, req,
|
2005-11-09 05:37:07 +00:00
|
|
|
USB_DIR_IN | USB_TYPE_VENDOR | USB_RECIP_DEVICE,
|
2008-11-18 17:51:08 +00:00
|
|
|
0x0000, reg, dev->urb_buf, len, HZ);
|
|
|
|
if (ret < 0) {
|
|
|
|
if (reg_debug)
|
|
|
|
printk(" failed!\n");
|
2008-11-19 09:17:44 +00:00
|
|
|
mutex_unlock(&dev->ctrl_urb_lock);
|
2008-11-18 17:51:08 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (len)
|
|
|
|
memcpy(buf, dev->urb_buf, len);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-11-19 09:17:44 +00:00
|
|
|
mutex_unlock(&dev->ctrl_urb_lock);
|
|
|
|
|
2008-04-18 00:41:10 +00:00
|
|
|
if (reg_debug) {
|
2008-11-26 12:58:48 +00:00
|
|
|
int byte;
|
|
|
|
|
|
|
|
printk("<<<");
|
2008-04-18 00:41:10 +00:00
|
|
|
for (byte = 0; byte < len; byte++)
|
2008-04-18 00:46:14 +00:00
|
|
|
printk(" %02x", (unsigned char)buf[byte]);
|
|
|
|
printk("\n");
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
* em28xx_read_reg_req()
|
2005-11-09 05:37:07 +00:00
|
|
|
* reads data from the usb device specifying bRequest
|
|
|
|
*/
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_read_reg_req(struct em28xx *dev, u8 req, u16 reg)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
|
|
|
int ret;
|
2008-11-26 12:58:48 +00:00
|
|
|
u8 val;
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-11-26 12:58:48 +00:00
|
|
|
ret = em28xx_read_reg_req_len(dev, req, reg, &val, 1);
|
|
|
|
if (ret < 0)
|
2008-11-18 17:51:08 +00:00
|
|
|
return ret;
|
2005-11-09 05:37:07 +00:00
|
|
|
|
|
|
|
return val;
|
|
|
|
}
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_read_reg(struct em28xx *dev, u16 reg)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2005-11-09 05:38:27 +00:00
|
|
|
return em28xx_read_reg_req(dev, USB_REQ_GET_STATUS, reg);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
* em28xx_write_regs_req()
|
2005-11-09 05:37:07 +00:00
|
|
|
* sends data to the usb device, specifying bRequest
|
|
|
|
*/
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_write_regs_req(struct em28xx *dev, u8 req, u16 reg, char *buf,
|
2005-11-09 05:37:07 +00:00
|
|
|
int len)
|
|
|
|
{
|
|
|
|
int ret;
|
2008-11-26 12:58:48 +00:00
|
|
|
int pipe = usb_sndctrlpipe(dev->udev, 0);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2006-02-07 08:49:11 +00:00
|
|
|
if (dev->state & DEV_DISCONNECTED)
|
2008-04-18 00:48:00 +00:00
|
|
|
return -ENODEV;
|
|
|
|
|
2008-11-18 17:51:08 +00:00
|
|
|
if ((len < 1) || (len > URB_MAX_CTRL_SIZE))
|
2008-04-18 00:48:00 +00:00
|
|
|
return -EINVAL;
|
2006-02-07 08:49:11 +00:00
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
if (reg_debug) {
|
2008-11-26 12:58:48 +00:00
|
|
|
int byte;
|
|
|
|
|
2009-02-11 02:28:24 +00:00
|
|
|
printk(KERN_DEBUG "(pipe 0x%08x): "
|
2008-11-26 12:58:48 +00:00
|
|
|
"OUT: %02x %02x %02x %02x %02x %02x %02x %02x >>>",
|
|
|
|
pipe,
|
|
|
|
USB_DIR_OUT | USB_TYPE_VENDOR | USB_RECIP_DEVICE,
|
|
|
|
req, 0, 0,
|
|
|
|
reg & 0xff, reg >> 8,
|
|
|
|
len & 0xff, len >> 8);
|
|
|
|
|
|
|
|
for (byte = 0; byte < len; byte++)
|
|
|
|
printk(" %02x", (unsigned char)buf[byte]);
|
2008-04-18 00:46:14 +00:00
|
|
|
printk("\n");
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2008-11-19 09:17:44 +00:00
|
|
|
mutex_lock(&dev->ctrl_urb_lock);
|
2008-11-18 17:51:08 +00:00
|
|
|
memcpy(dev->urb_buf, buf, len);
|
2008-11-26 12:58:48 +00:00
|
|
|
ret = usb_control_msg(dev->udev, pipe, req,
|
2005-11-09 05:37:07 +00:00
|
|
|
USB_DIR_OUT | USB_TYPE_VENDOR | USB_RECIP_DEVICE,
|
2008-11-18 17:51:08 +00:00
|
|
|
0x0000, reg, dev->urb_buf, len, HZ);
|
2008-11-19 09:17:44 +00:00
|
|
|
mutex_unlock(&dev->ctrl_urb_lock);
|
2008-11-18 17:51:08 +00:00
|
|
|
|
2008-04-18 00:42:14 +00:00
|
|
|
if (dev->wait_after_write)
|
|
|
|
msleep(dev->wait_after_write);
|
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_write_regs(struct em28xx *dev, u16 reg, char *buf, int len)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-04-18 00:48:00 +00:00
|
|
|
int rc;
|
|
|
|
|
|
|
|
rc = em28xx_write_regs_req(dev, USB_REQ_GET_STATUS, reg, buf, len);
|
|
|
|
|
|
|
|
/* Stores GPO/GPIO values at the cache, if changed
|
|
|
|
Only write values should be stored, since input on a GPIO
|
|
|
|
register will return the input bits.
|
|
|
|
Not sure what happens on reading GPO register.
|
|
|
|
*/
|
|
|
|
if (rc >= 0) {
|
2008-11-12 05:05:06 +00:00
|
|
|
if (reg == dev->reg_gpo_num)
|
2008-04-18 00:48:00 +00:00
|
|
|
dev->reg_gpo = buf[0];
|
2008-11-12 05:05:06 +00:00
|
|
|
else if (reg == dev->reg_gpio_num)
|
2008-04-18 00:48:00 +00:00
|
|
|
dev->reg_gpio = buf[0];
|
|
|
|
}
|
|
|
|
|
|
|
|
return rc;
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2008-11-20 12:52:20 +00:00
|
|
|
/* Write a single register */
|
|
|
|
int em28xx_write_reg(struct em28xx *dev, u16 reg, u8 val)
|
|
|
|
{
|
|
|
|
return em28xx_write_regs(dev, reg, &val, 1);
|
|
|
|
}
|
2011-07-04 00:05:06 +00:00
|
|
|
EXPORT_SYMBOL_GPL(em28xx_write_reg);
|
2008-11-20 12:52:20 +00:00
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
* em28xx_write_reg_bits()
|
2005-11-09 05:37:07 +00:00
|
|
|
* sets only some bits (specified by bitmask) of a register, by first reading
|
|
|
|
* the actual value
|
|
|
|
*/
|
2009-12-05 11:27:49 +00:00
|
|
|
int em28xx_write_reg_bits(struct em28xx *dev, u16 reg, u8 val,
|
2005-11-09 05:37:07 +00:00
|
|
|
u8 bitmask)
|
|
|
|
{
|
|
|
|
int oldval;
|
|
|
|
u8 newval;
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2008-04-18 00:48:00 +00:00
|
|
|
/* Uses cache for gpo/gpio registers */
|
2008-11-12 05:05:06 +00:00
|
|
|
if (reg == dev->reg_gpo_num)
|
2008-04-18 00:48:00 +00:00
|
|
|
oldval = dev->reg_gpo;
|
2008-11-12 05:05:06 +00:00
|
|
|
else if (reg == dev->reg_gpio_num)
|
2008-04-18 00:48:00 +00:00
|
|
|
oldval = dev->reg_gpio;
|
|
|
|
else
|
|
|
|
oldval = em28xx_read_reg(dev, reg);
|
2008-04-18 00:41:10 +00:00
|
|
|
|
|
|
|
if (oldval < 0)
|
2005-11-09 05:37:07 +00:00
|
|
|
return oldval;
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
newval = (((u8) oldval) & ~bitmask) | (val & bitmask);
|
2008-04-18 00:48:00 +00:00
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
return em28xx_write_regs(dev, reg, &newval, 1);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
/*
|
|
|
|
* em28xx_is_ac97_ready()
|
|
|
|
* Checks if ac97 is ready
|
|
|
|
*/
|
|
|
|
static int em28xx_is_ac97_ready(struct em28xx *dev)
|
|
|
|
{
|
|
|
|
int ret, i;
|
|
|
|
|
|
|
|
/* Wait up to 50 ms for AC97 command to complete */
|
|
|
|
for (i = 0; i < 10; i++, msleep(5)) {
|
|
|
|
ret = em28xx_read_reg(dev, EM28XX_R43_AC97BUSY);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
if (!(ret & 0x01))
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
em28xx_warn("AC97 command still being executed: not handled properly!\n");
|
|
|
|
return -EBUSY;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* em28xx_read_ac97()
|
|
|
|
* write a 16 bit value to the specified AC97 address (LSB first!)
|
|
|
|
*/
|
2008-12-22 16:18:27 +00:00
|
|
|
int em28xx_read_ac97(struct em28xx *dev, u8 reg)
|
2008-11-19 15:01:33 +00:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
u8 addr = (reg & 0x7f) | 0x80;
|
|
|
|
u16 val;
|
|
|
|
|
|
|
|
ret = em28xx_is_ac97_ready(dev);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ret = em28xx_write_regs(dev, EM28XX_R42_AC97ADDR, &addr, 1);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
ret = dev->em28xx_read_reg_req_len(dev, 0, EM28XX_R40_AC97LSB,
|
|
|
|
(u8 *)&val, sizeof(val));
|
|
|
|
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
return le16_to_cpu(val);
|
|
|
|
}
|
2011-06-19 16:06:40 +00:00
|
|
|
EXPORT_SYMBOL_GPL(em28xx_read_ac97);
|
2008-11-19 15:01:33 +00:00
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
/*
|
2005-11-09 05:38:27 +00:00
|
|
|
* em28xx_write_ac97()
|
2005-11-09 05:37:07 +00:00
|
|
|
* write a 16 bit value to the specified AC97 address (LSB first!)
|
|
|
|
*/
|
2008-12-22 16:18:27 +00:00
|
|
|
int em28xx_write_ac97(struct em28xx *dev, u8 reg, u16 val)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-11-19 15:01:33 +00:00
|
|
|
int ret;
|
2005-11-09 05:37:07 +00:00
|
|
|
u8 addr = reg & 0x7f;
|
2008-11-19 15:01:33 +00:00
|
|
|
__le16 value;
|
|
|
|
|
|
|
|
value = cpu_to_le16(val);
|
|
|
|
|
|
|
|
ret = em28xx_is_ac97_ready(dev);
|
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
ret = em28xx_write_regs(dev, EM28XX_R40_AC97LSB, (u8 *) &value, 2);
|
2008-04-18 00:41:10 +00:00
|
|
|
if (ret < 0)
|
2005-11-09 05:37:07 +00:00
|
|
|
return ret;
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2008-04-18 00:44:58 +00:00
|
|
|
ret = em28xx_write_regs(dev, EM28XX_R42_AC97ADDR, &addr, 1);
|
2008-04-18 00:41:10 +00:00
|
|
|
if (ret < 0)
|
2005-11-09 05:37:07 +00:00
|
|
|
return ret;
|
2008-02-06 21:34:13 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2011-06-19 16:06:40 +00:00
|
|
|
EXPORT_SYMBOL_GPL(em28xx_write_ac97);
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2011-06-18 10:02:49 +00:00
|
|
|
struct em28xx_vol_itable {
|
2008-11-20 16:39:39 +00:00
|
|
|
enum em28xx_amux mux;
|
2008-11-20 12:06:09 +00:00
|
|
|
u8 reg;
|
|
|
|
};
|
|
|
|
|
2011-06-18 10:02:49 +00:00
|
|
|
static struct em28xx_vol_itable inputs[] = {
|
2008-11-20 12:06:09 +00:00
|
|
|
{ EM28XX_AMUX_VIDEO, AC97_VIDEO_VOL },
|
|
|
|
{ EM28XX_AMUX_LINE_IN, AC97_LINEIN_VOL },
|
|
|
|
{ EM28XX_AMUX_PHONE, AC97_PHONE_VOL },
|
|
|
|
{ EM28XX_AMUX_MIC, AC97_MIC_VOL },
|
|
|
|
{ EM28XX_AMUX_CD, AC97_CD_VOL },
|
|
|
|
{ EM28XX_AMUX_AUX, AC97_AUX_VOL },
|
|
|
|
{ EM28XX_AMUX_PCM_OUT, AC97_PCM_OUT_VOL },
|
|
|
|
};
|
|
|
|
|
|
|
|
static int set_ac97_input(struct em28xx *dev)
|
2008-11-19 15:01:33 +00:00
|
|
|
{
|
2008-11-20 12:06:09 +00:00
|
|
|
int ret, i;
|
|
|
|
enum em28xx_amux amux = dev->ctl_ainput;
|
2008-11-19 15:01:33 +00:00
|
|
|
|
2008-11-20 12:06:09 +00:00
|
|
|
/* EM28XX_AMUX_VIDEO2 is a special case used to indicate that
|
|
|
|
em28xx should point to LINE IN, while AC97 should use VIDEO
|
|
|
|
*/
|
|
|
|
if (amux == EM28XX_AMUX_VIDEO2)
|
2008-11-20 15:10:44 +00:00
|
|
|
amux = EM28XX_AMUX_VIDEO;
|
2008-11-19 15:01:33 +00:00
|
|
|
|
2008-11-20 12:06:09 +00:00
|
|
|
/* Mute all entres but the one that were selected */
|
|
|
|
for (i = 0; i < ARRAY_SIZE(inputs); i++) {
|
2008-11-20 16:39:39 +00:00
|
|
|
if (amux == inputs[i].mux)
|
2008-11-20 12:06:09 +00:00
|
|
|
ret = em28xx_write_ac97(dev, inputs[i].reg, 0x0808);
|
|
|
|
else
|
|
|
|
ret = em28xx_write_ac97(dev, inputs[i].reg, 0x8000);
|
2008-11-19 15:01:33 +00:00
|
|
|
|
2008-11-20 12:06:09 +00:00
|
|
|
if (ret < 0)
|
|
|
|
em28xx_warn("couldn't setup AC97 register %d\n",
|
|
|
|
inputs[i].reg);
|
|
|
|
}
|
|
|
|
return 0;
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2008-02-06 21:34:13 +00:00
|
|
|
static int em28xx_set_audio_source(struct em28xx *dev)
|
2008-01-05 12:53:54 +00:00
|
|
|
{
|
2008-02-05 10:37:21 +00:00
|
|
|
int ret;
|
2008-01-05 12:53:54 +00:00
|
|
|
u8 input;
|
|
|
|
|
2008-11-25 12:39:50 +00:00
|
|
|
if (dev->board.is_em2800) {
|
2008-11-20 12:06:09 +00:00
|
|
|
if (dev->ctl_ainput == EM28XX_AMUX_VIDEO)
|
2008-01-05 12:53:54 +00:00
|
|
|
input = EM2800_AUDIO_SRC_TUNER;
|
2008-11-20 12:06:09 +00:00
|
|
|
else
|
|
|
|
input = EM2800_AUDIO_SRC_LINE;
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-04-18 00:44:58 +00:00
|
|
|
ret = em28xx_write_regs(dev, EM2800_R08_AUDIOSRC, &input, 1);
|
2008-01-05 12:53:54 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2008-11-25 12:39:50 +00:00
|
|
|
if (dev->board.has_msp34xx)
|
2008-01-05 12:53:54 +00:00
|
|
|
input = EM28XX_AUDIO_SRC_TUNER;
|
|
|
|
else {
|
|
|
|
switch (dev->ctl_ainput) {
|
|
|
|
case EM28XX_AMUX_VIDEO:
|
|
|
|
input = EM28XX_AUDIO_SRC_TUNER;
|
|
|
|
break;
|
2008-11-19 15:01:33 +00:00
|
|
|
default:
|
2008-01-05 12:53:54 +00:00
|
|
|
input = EM28XX_AUDIO_SRC_LINE;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2009-03-04 11:27:52 +00:00
|
|
|
if (dev->board.mute_gpio && dev->mute)
|
|
|
|
em28xx_gpio_set(dev, dev->board.mute_gpio);
|
|
|
|
else
|
|
|
|
em28xx_gpio_set(dev, INPUT(dev->ctl_input)->gpio);
|
|
|
|
|
2008-04-18 00:44:58 +00:00
|
|
|
ret = em28xx_write_reg_bits(dev, EM28XX_R0E_AUDIOSRC, input, 0xc0);
|
2008-01-05 12:53:54 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
2008-02-06 21:34:13 +00:00
|
|
|
msleep(5);
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
switch (dev->audio_mode.ac97) {
|
|
|
|
case EM28XX_NO_AC97:
|
|
|
|
break;
|
2008-11-20 12:06:09 +00:00
|
|
|
default:
|
|
|
|
ret = set_ac97_input(dev);
|
2008-11-19 15:01:33 +00:00
|
|
|
}
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-11-20 12:06:09 +00:00
|
|
|
return ret;
|
2008-01-05 12:53:54 +00:00
|
|
|
}
|
|
|
|
|
2011-06-18 10:02:49 +00:00
|
|
|
struct em28xx_vol_otable {
|
|
|
|
enum em28xx_aout mux;
|
|
|
|
u8 reg;
|
|
|
|
};
|
|
|
|
|
|
|
|
static const struct em28xx_vol_otable outputs[] = {
|
2008-11-20 16:39:39 +00:00
|
|
|
{ EM28XX_AOUT_MASTER, AC97_MASTER_VOL },
|
|
|
|
{ EM28XX_AOUT_LINE, AC97_LINE_LEVEL_VOL },
|
|
|
|
{ EM28XX_AOUT_MONO, AC97_MASTER_MONO_VOL },
|
|
|
|
{ EM28XX_AOUT_LFE, AC97_LFE_MASTER_VOL },
|
|
|
|
{ EM28XX_AOUT_SURR, AC97_SURR_MASTER_VOL },
|
2008-11-20 15:40:51 +00:00
|
|
|
};
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_audio_analog_set(struct em28xx *dev)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-11-20 15:40:51 +00:00
|
|
|
int ret, i;
|
2008-11-25 10:05:06 +00:00
|
|
|
u8 xclk;
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
if (!dev->audio_mode.has_audio)
|
|
|
|
return 0;
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-11-20 12:06:09 +00:00
|
|
|
/* It is assumed that all devices use master volume for output.
|
|
|
|
It would be possible to use also line output.
|
|
|
|
*/
|
2008-11-19 15:01:33 +00:00
|
|
|
if (dev->audio_mode.ac97 != EM28XX_NO_AC97) {
|
2008-11-20 15:40:51 +00:00
|
|
|
/* Mute all outputs */
|
|
|
|
for (i = 0; i < ARRAY_SIZE(outputs); i++) {
|
2008-11-20 16:39:39 +00:00
|
|
|
ret = em28xx_write_ac97(dev, outputs[i].reg, 0x8000);
|
2008-11-20 15:40:51 +00:00
|
|
|
if (ret < 0)
|
|
|
|
em28xx_warn("couldn't setup AC97 register %d\n",
|
2008-11-20 16:39:39 +00:00
|
|
|
outputs[i].reg);
|
2008-11-20 15:40:51 +00:00
|
|
|
}
|
2008-11-19 15:01:33 +00:00
|
|
|
}
|
2008-01-05 12:53:54 +00:00
|
|
|
|
2008-11-25 12:39:50 +00:00
|
|
|
xclk = dev->board.xclk & 0x7f;
|
2008-01-05 20:01:41 +00:00
|
|
|
if (!dev->mute)
|
2009-01-19 00:44:46 +00:00
|
|
|
xclk |= EM28XX_XCLK_AUDIO_UNMUTE;
|
2008-01-05 20:01:41 +00:00
|
|
|
|
2008-11-25 10:05:06 +00:00
|
|
|
ret = em28xx_write_reg(dev, EM28XX_R0F_XCLK, xclk);
|
2008-01-05 12:53:54 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
2008-01-05 20:01:41 +00:00
|
|
|
msleep(10);
|
2008-01-05 12:53:54 +00:00
|
|
|
|
|
|
|
/* Selects the proper audio input */
|
|
|
|
ret = em28xx_set_audio_source(dev);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
/* Sets volume */
|
|
|
|
if (dev->audio_mode.ac97 != EM28XX_NO_AC97) {
|
|
|
|
int vol;
|
|
|
|
|
2009-01-19 00:45:28 +00:00
|
|
|
em28xx_write_ac97(dev, AC97_POWER_DOWN_CTRL, 0x4200);
|
|
|
|
em28xx_write_ac97(dev, AC97_EXT_AUD_CTRL, 0x0031);
|
|
|
|
em28xx_write_ac97(dev, AC97_PCM_IN_SRATE, 0xbb80);
|
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
/* LSB: left channel - both channels with the same level */
|
|
|
|
vol = (0x1f - dev->volume) | ((0x1f - dev->volume) << 8);
|
|
|
|
|
|
|
|
/* Mute device, if needed */
|
|
|
|
if (dev->mute)
|
|
|
|
vol |= 0x8000;
|
|
|
|
|
|
|
|
/* Sets volume */
|
2008-11-20 16:39:39 +00:00
|
|
|
for (i = 0; i < ARRAY_SIZE(outputs); i++) {
|
|
|
|
if (dev->ctl_aoutput & outputs[i].mux)
|
|
|
|
ret = em28xx_write_ac97(dev, outputs[i].reg,
|
|
|
|
vol);
|
|
|
|
if (ret < 0)
|
|
|
|
em28xx_warn("couldn't setup AC97 register %d\n",
|
|
|
|
outputs[i].reg);
|
|
|
|
}
|
2009-01-13 00:50:52 +00:00
|
|
|
|
|
|
|
if (dev->ctl_aoutput & EM28XX_AOUT_PCM_IN) {
|
|
|
|
int sel = ac97_return_record_select(dev->ctl_aoutput);
|
|
|
|
|
2009-02-11 02:28:24 +00:00
|
|
|
/* Use the same input for both left and right
|
|
|
|
channels */
|
2009-01-13 00:50:52 +00:00
|
|
|
sel |= (sel << 8);
|
|
|
|
|
|
|
|
em28xx_write_ac97(dev, AC97_RECORD_SELECT, sel);
|
|
|
|
}
|
2008-11-19 15:01:33 +00:00
|
|
|
}
|
2008-02-06 21:34:13 +00:00
|
|
|
|
2008-01-05 12:53:54 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_audio_analog_set);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-11-19 15:01:33 +00:00
|
|
|
int em28xx_audio_setup(struct em28xx *dev)
|
|
|
|
{
|
|
|
|
int vid1, vid2, feat, cfg;
|
2008-11-20 11:56:19 +00:00
|
|
|
u32 vid;
|
2008-11-19 15:01:33 +00:00
|
|
|
|
2011-05-02 21:19:14 +00:00
|
|
|
if (dev->chip_id == CHIP_ID_EM2870 || dev->chip_id == CHIP_ID_EM2874
|
|
|
|
|| dev->chip_id == CHIP_ID_EM28174) {
|
2008-11-19 15:01:33 +00:00
|
|
|
/* Digital only device - don't load any alsa module */
|
2011-06-17 18:15:12 +00:00
|
|
|
dev->audio_mode.has_audio = false;
|
|
|
|
dev->has_audio_class = false;
|
|
|
|
dev->has_alsa_audio = false;
|
2008-11-19 15:01:33 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2011-06-17 18:15:12 +00:00
|
|
|
dev->audio_mode.has_audio = true;
|
2008-11-19 15:01:33 +00:00
|
|
|
|
|
|
|
/* See how this device is configured */
|
|
|
|
cfg = em28xx_read_reg(dev, EM28XX_R00_CHIPCFG);
|
2009-06-10 02:40:39 +00:00
|
|
|
em28xx_info("Config register raw data: 0x%02x\n", cfg);
|
|
|
|
if (cfg < 0) {
|
|
|
|
/* Register read error? */
|
2008-11-19 15:01:33 +00:00
|
|
|
cfg = EM28XX_CHIPCFG_AC97; /* Be conservative */
|
2009-06-10 02:40:39 +00:00
|
|
|
} else if ((cfg & EM28XX_CHIPCFG_AUDIOMASK) == 0x00) {
|
|
|
|
/* The device doesn't have vendor audio at all */
|
2011-06-17 18:15:12 +00:00
|
|
|
dev->has_alsa_audio = false;
|
|
|
|
dev->audio_mode.has_audio = false;
|
2009-06-10 02:40:39 +00:00
|
|
|
return 0;
|
|
|
|
} else if ((cfg & EM28XX_CHIPCFG_AUDIOMASK) ==
|
|
|
|
EM28XX_CHIPCFG_I2S_3_SAMPRATES) {
|
2008-11-19 15:01:33 +00:00
|
|
|
em28xx_info("I2S Audio (3 sample rates)\n");
|
|
|
|
dev->audio_mode.i2s_3rates = 1;
|
2009-06-10 02:40:39 +00:00
|
|
|
} else if ((cfg & EM28XX_CHIPCFG_AUDIOMASK) ==
|
|
|
|
EM28XX_CHIPCFG_I2S_5_SAMPRATES) {
|
2008-11-19 15:01:33 +00:00
|
|
|
em28xx_info("I2S Audio (5 sample rates)\n");
|
|
|
|
dev->audio_mode.i2s_5rates = 1;
|
|
|
|
}
|
|
|
|
|
2008-12-30 03:17:09 +00:00
|
|
|
if ((cfg & EM28XX_CHIPCFG_AUDIOMASK) != EM28XX_CHIPCFG_AC97) {
|
|
|
|
/* Skip the code that does AC97 vendor detection */
|
2008-11-19 15:01:33 +00:00
|
|
|
dev->audio_mode.ac97 = EM28XX_NO_AC97;
|
|
|
|
goto init_audio;
|
|
|
|
}
|
|
|
|
|
|
|
|
dev->audio_mode.ac97 = EM28XX_AC97_OTHER;
|
|
|
|
|
|
|
|
vid1 = em28xx_read_ac97(dev, AC97_VENDOR_ID1);
|
|
|
|
if (vid1 < 0) {
|
2009-11-27 15:52:54 +00:00
|
|
|
/*
|
|
|
|
* Device likely doesn't support AC97
|
|
|
|
* Note: (some) em2800 devices without eeprom reports 0x91 on
|
|
|
|
* CHIPCFG register, even not having an AC97 chip
|
|
|
|
*/
|
2008-11-19 15:01:33 +00:00
|
|
|
em28xx_warn("AC97 chip type couldn't be determined\n");
|
2009-11-27 15:52:54 +00:00
|
|
|
dev->audio_mode.ac97 = EM28XX_NO_AC97;
|
2011-06-17 18:15:12 +00:00
|
|
|
dev->has_alsa_audio = false;
|
|
|
|
dev->audio_mode.has_audio = false;
|
2008-11-19 15:01:33 +00:00
|
|
|
goto init_audio;
|
|
|
|
}
|
|
|
|
|
|
|
|
vid2 = em28xx_read_ac97(dev, AC97_VENDOR_ID2);
|
|
|
|
if (vid2 < 0)
|
|
|
|
goto init_audio;
|
|
|
|
|
2008-11-20 11:56:19 +00:00
|
|
|
vid = vid1 << 16 | vid2;
|
|
|
|
|
|
|
|
dev->audio_mode.ac97_vendor_id = vid;
|
|
|
|
em28xx_warn("AC97 vendor ID = 0x%08x\n", vid);
|
2008-11-19 15:01:33 +00:00
|
|
|
|
|
|
|
feat = em28xx_read_ac97(dev, AC97_RESET);
|
|
|
|
if (feat < 0)
|
|
|
|
goto init_audio;
|
|
|
|
|
|
|
|
dev->audio_mode.ac97_feat = feat;
|
|
|
|
em28xx_warn("AC97 features = 0x%04x\n", feat);
|
|
|
|
|
2008-11-20 11:56:19 +00:00
|
|
|
/* Try to identify what audio processor we have */
|
2011-12-20 20:45:18 +00:00
|
|
|
if (((vid == 0xffffffff) || (vid == 0x83847650)) && (feat == 0x6a90))
|
2008-11-19 15:01:33 +00:00
|
|
|
dev->audio_mode.ac97 = EM28XX_AC97_EM202;
|
2008-11-20 15:49:33 +00:00
|
|
|
else if ((vid >> 8) == 0x838476)
|
|
|
|
dev->audio_mode.ac97 = EM28XX_AC97_SIGMATEL;
|
2008-11-19 15:01:33 +00:00
|
|
|
|
|
|
|
init_audio:
|
|
|
|
/* Reports detected AC97 processor */
|
|
|
|
switch (dev->audio_mode.ac97) {
|
|
|
|
case EM28XX_NO_AC97:
|
|
|
|
em28xx_info("No AC97 audio processor\n");
|
|
|
|
break;
|
|
|
|
case EM28XX_AC97_EM202:
|
|
|
|
em28xx_info("Empia 202 AC97 audio processor detected\n");
|
|
|
|
break;
|
2008-11-20 15:49:33 +00:00
|
|
|
case EM28XX_AC97_SIGMATEL:
|
|
|
|
em28xx_info("Sigmatel audio processor detected(stac 97%02x)\n",
|
|
|
|
dev->audio_mode.ac97_vendor_id & 0xff);
|
|
|
|
break;
|
2008-11-19 15:01:33 +00:00
|
|
|
case EM28XX_AC97_OTHER:
|
|
|
|
em28xx_warn("Unknown AC97 audio processor detected!\n");
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return em28xx_audio_analog_set(dev);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_audio_setup);
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_colorlevels_set_default(struct em28xx *dev)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
V4L/DVB (9758): em28xx: replace some magic by register descriptions where known
Replaces all occurrences of em28xx_write_regs_req() and em28xx_write_reg()
used to setup register names by em28xx_write_reg().
Also, documents the register names that are known.
This patch were generated by this small perl script:
my %reg_map = (
# Register table - the same as defined on parse_em28xx.pl script
);
while (<>) {
if (m/(.*)em28xx_write_regs_req\(dev\,\s*0x00\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} elsif (m/(.*)em28xx_write_regs\(dev\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} else {
print $_;
}
}
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2008-11-26 19:28:42 +00:00
|
|
|
em28xx_write_reg(dev, EM28XX_R20_YGAIN, 0x10); /* contrast */
|
|
|
|
em28xx_write_reg(dev, EM28XX_R21_YOFFSET, 0x00); /* brightness */
|
|
|
|
em28xx_write_reg(dev, EM28XX_R22_UVGAIN, 0x10); /* saturation */
|
|
|
|
em28xx_write_reg(dev, EM28XX_R23_UOFFSET, 0x00);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R24_VOFFSET, 0x00);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R25_SHARPNESS, 0x00);
|
|
|
|
|
|
|
|
em28xx_write_reg(dev, EM28XX_R14_GAMMA, 0x20);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R15_RGAIN, 0x20);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R16_GGAIN, 0x20);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R17_BGAIN, 0x20);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R18_ROFFSET, 0x00);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R19_GOFFSET, 0x00);
|
|
|
|
return em28xx_write_reg(dev, EM28XX_R1A_BOFFSET, 0x00);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_capture_start(struct em28xx *dev, int start)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-04-18 00:37:31 +00:00
|
|
|
int rc;
|
2008-11-12 05:05:24 +00:00
|
|
|
|
2011-07-04 00:05:06 +00:00
|
|
|
if (dev->chip_id == CHIP_ID_EM2874 ||
|
|
|
|
dev->chip_id == CHIP_ID_EM2884 ||
|
|
|
|
dev->chip_id == CHIP_ID_EM28174) {
|
2008-11-12 05:05:24 +00:00
|
|
|
/* The Transport Stream Enable Register moved in em2874 */
|
|
|
|
if (!start) {
|
|
|
|
rc = em28xx_write_reg_bits(dev, EM2874_R5F_TS_ENABLE,
|
|
|
|
0x00,
|
|
|
|
EM2874_TS1_CAPTURE_ENABLE);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Enable Transport Stream */
|
|
|
|
rc = em28xx_write_reg_bits(dev, EM2874_R5F_TS_ENABLE,
|
|
|
|
EM2874_TS1_CAPTURE_ENABLE,
|
|
|
|
EM2874_TS1_CAPTURE_ENABLE);
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
/* FIXME: which is the best order? */
|
|
|
|
/* video registers are sampled by VREF */
|
2008-04-18 00:44:58 +00:00
|
|
|
rc = em28xx_write_reg_bits(dev, EM28XX_R0C_USBSUSP,
|
2008-04-18 00:37:31 +00:00
|
|
|
start ? 0x10 : 0x00, 0x10);
|
|
|
|
if (rc < 0)
|
|
|
|
return rc;
|
|
|
|
|
|
|
|
if (!start) {
|
|
|
|
/* disable video capture */
|
V4L/DVB (9758): em28xx: replace some magic by register descriptions where known
Replaces all occurrences of em28xx_write_regs_req() and em28xx_write_reg()
used to setup register names by em28xx_write_reg().
Also, documents the register names that are known.
This patch were generated by this small perl script:
my %reg_map = (
# Register table - the same as defined on parse_em28xx.pl script
);
while (<>) {
if (m/(.*)em28xx_write_regs_req\(dev\,\s*0x00\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} elsif (m/(.*)em28xx_write_regs\(dev\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} else {
print $_;
}
}
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2008-11-26 19:28:42 +00:00
|
|
|
rc = em28xx_write_reg(dev, EM28XX_R12_VINENABLE, 0x27);
|
2008-04-18 00:40:45 +00:00
|
|
|
return rc;
|
2008-04-18 00:37:31 +00:00
|
|
|
}
|
|
|
|
|
2009-08-07 21:43:00 +00:00
|
|
|
if (dev->board.is_webcam)
|
|
|
|
rc = em28xx_write_reg(dev, 0x13, 0x0c);
|
|
|
|
|
2005-11-09 05:37:07 +00:00
|
|
|
/* enable video capture */
|
V4L/DVB (9758): em28xx: replace some magic by register descriptions where known
Replaces all occurrences of em28xx_write_regs_req() and em28xx_write_reg()
used to setup register names by em28xx_write_reg().
Also, documents the register names that are known.
This patch were generated by this small perl script:
my %reg_map = (
# Register table - the same as defined on parse_em28xx.pl script
);
while (<>) {
if (m/(.*)em28xx_write_regs_req\(dev\,\s*0x00\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} elsif (m/(.*)em28xx_write_regs\(dev\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} else {
print $_;
}
}
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2008-11-26 19:28:42 +00:00
|
|
|
rc = em28xx_write_reg(dev, 0x48, 0x00);
|
2008-04-18 00:40:45 +00:00
|
|
|
|
2008-04-18 00:37:31 +00:00
|
|
|
if (dev->mode == EM28XX_ANALOG_MODE)
|
V4L/DVB (9758): em28xx: replace some magic by register descriptions where known
Replaces all occurrences of em28xx_write_regs_req() and em28xx_write_reg()
used to setup register names by em28xx_write_reg().
Also, documents the register names that are known.
This patch were generated by this small perl script:
my %reg_map = (
# Register table - the same as defined on parse_em28xx.pl script
);
while (<>) {
if (m/(.*)em28xx_write_regs_req\(dev\,\s*0x00\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} elsif (m/(.*)em28xx_write_regs\(dev\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} else {
print $_;
}
}
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2008-11-26 19:28:42 +00:00
|
|
|
rc = em28xx_write_reg(dev, EM28XX_R12_VINENABLE, 0x67);
|
2008-04-18 00:37:31 +00:00
|
|
|
else
|
V4L/DVB (9758): em28xx: replace some magic by register descriptions where known
Replaces all occurrences of em28xx_write_regs_req() and em28xx_write_reg()
used to setup register names by em28xx_write_reg().
Also, documents the register names that are known.
This patch were generated by this small perl script:
my %reg_map = (
# Register table - the same as defined on parse_em28xx.pl script
);
while (<>) {
if (m/(.*)em28xx_write_regs_req\(dev\,\s*0x00\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} elsif (m/(.*)em28xx_write_regs\(dev\,\s*(.*)\,\s*\"\\x(..)\",\s*1\)\;(.*)/) {
my $reg = $2;
my $val = $3;
$val =~ tr/A-f/a-f/;
$reg = $reg_map{$reg} if defined($reg_map{$reg});
printf "$1em28xx_write_reg(dev, %s, 0x%s);$4\n", $reg, $val;
} else {
print $_;
}
}
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2008-11-26 19:28:42 +00:00
|
|
|
rc = em28xx_write_reg(dev, EM28XX_R12_VINENABLE, 0x37);
|
2008-04-18 00:37:31 +00:00
|
|
|
|
2008-04-18 00:41:10 +00:00
|
|
|
msleep(6);
|
2008-04-18 00:37:31 +00:00
|
|
|
|
|
|
|
return rc;
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2009-09-01 04:19:46 +00:00
|
|
|
int em28xx_vbi_supported(struct em28xx *dev)
|
|
|
|
{
|
|
|
|
/* Modprobe option to manually disable */
|
|
|
|
if (disable_vbi == 1)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (dev->chip_id == CHIP_ID_EM2860 ||
|
|
|
|
dev->chip_id == CHIP_ID_EM2883)
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
/* Version of em28xx that does not support VBI */
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-12-20 12:06:37 +00:00
|
|
|
int em28xx_set_outfmt(struct em28xx *dev)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-12-20 12:06:37 +00:00
|
|
|
int ret;
|
2009-09-01 04:19:46 +00:00
|
|
|
u8 vinctrl;
|
2008-12-20 12:06:37 +00:00
|
|
|
|
|
|
|
ret = em28xx_write_reg_bits(dev, EM28XX_R27_OUTFMT,
|
2009-07-14 15:33:25 +00:00
|
|
|
dev->format->reg | 0x20, 0xff);
|
2008-12-20 12:06:37 +00:00
|
|
|
if (ret < 0)
|
2009-06-29 14:35:05 +00:00
|
|
|
return ret;
|
2008-12-20 12:06:37 +00:00
|
|
|
|
2009-07-14 15:33:25 +00:00
|
|
|
ret = em28xx_write_reg(dev, EM28XX_R10_VINMODE, dev->vinmode);
|
2008-12-20 12:06:37 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
2009-09-01 04:19:46 +00:00
|
|
|
vinctrl = dev->vinctl;
|
|
|
|
if (em28xx_vbi_supported(dev) == 1) {
|
|
|
|
vinctrl |= EM28XX_VINCTRL_VBI_RAW;
|
|
|
|
em28xx_write_reg(dev, EM28XX_R34_VBI_START_H, 0x00);
|
2009-11-25 02:17:25 +00:00
|
|
|
em28xx_write_reg(dev, EM28XX_R36_VBI_WIDTH, dev->vbi_width/4);
|
|
|
|
em28xx_write_reg(dev, EM28XX_R37_VBI_HEIGHT, dev->vbi_height);
|
|
|
|
if (dev->norm & V4L2_STD_525_60) {
|
|
|
|
/* NTSC */
|
|
|
|
em28xx_write_reg(dev, EM28XX_R35_VBI_START_V, 0x09);
|
|
|
|
} else if (dev->norm & V4L2_STD_625_50) {
|
|
|
|
/* PAL */
|
|
|
|
em28xx_write_reg(dev, EM28XX_R35_VBI_START_V, 0x07);
|
|
|
|
}
|
2009-09-01 04:19:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return em28xx_write_reg(dev, EM28XX_R11_VINCTRL, vinctrl);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2006-05-22 13:31:42 +00:00
|
|
|
static int em28xx_accumulator_set(struct em28xx *dev, u8 xmin, u8 xmax,
|
|
|
|
u8 ymin, u8 ymax)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2008-04-18 00:41:10 +00:00
|
|
|
em28xx_coredbg("em28xx Scale: (%d,%d)-(%d,%d)\n",
|
|
|
|
xmin, ymin, xmax, ymax);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-04-18 00:44:58 +00:00
|
|
|
em28xx_write_regs(dev, EM28XX_R28_XMIN, &xmin, 1);
|
|
|
|
em28xx_write_regs(dev, EM28XX_R29_XMAX, &xmax, 1);
|
|
|
|
em28xx_write_regs(dev, EM28XX_R2A_YMIN, &ymin, 1);
|
|
|
|
return em28xx_write_regs(dev, EM28XX_R2B_YMAX, &ymax, 1);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2006-05-22 13:31:42 +00:00
|
|
|
static int em28xx_capture_area_set(struct em28xx *dev, u8 hstart, u8 vstart,
|
2005-11-09 05:37:07 +00:00
|
|
|
u16 width, u16 height)
|
|
|
|
{
|
|
|
|
u8 cwidth = width;
|
|
|
|
u8 cheight = height;
|
|
|
|
u8 overflow = (height >> 7 & 0x02) | (width >> 8 & 0x01);
|
|
|
|
|
2008-04-18 00:41:10 +00:00
|
|
|
em28xx_coredbg("em28xx Area Set: (%d,%d)\n",
|
|
|
|
(width | (overflow & 2) << 7),
|
2005-11-09 05:37:07 +00:00
|
|
|
(height | (overflow & 1) << 8));
|
|
|
|
|
2008-04-18 00:44:58 +00:00
|
|
|
em28xx_write_regs(dev, EM28XX_R1C_HSTART, &hstart, 1);
|
|
|
|
em28xx_write_regs(dev, EM28XX_R1D_VSTART, &vstart, 1);
|
|
|
|
em28xx_write_regs(dev, EM28XX_R1E_CWIDTH, &cwidth, 1);
|
|
|
|
em28xx_write_regs(dev, EM28XX_R1F_CHEIGHT, &cheight, 1);
|
|
|
|
return em28xx_write_regs(dev, EM28XX_R1B_OFLOW, &overflow, 1);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2006-05-22 13:31:42 +00:00
|
|
|
static int em28xx_scaler_set(struct em28xx *dev, u16 h, u16 v)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
2005-11-09 05:38:10 +00:00
|
|
|
u8 mode;
|
|
|
|
/* the em2800 scaler only supports scaling down to 50% */
|
2009-06-29 14:35:05 +00:00
|
|
|
|
2009-07-13 23:15:02 +00:00
|
|
|
if (dev->board.is_em2800) {
|
2005-11-09 05:38:10 +00:00
|
|
|
mode = (v ? 0x20 : 0x00) | (h ? 0x10 : 0x00);
|
2009-06-29 14:35:05 +00:00
|
|
|
} else {
|
2005-11-09 05:38:10 +00:00
|
|
|
u8 buf[2];
|
2009-06-29 14:35:05 +00:00
|
|
|
|
2005-11-09 05:38:10 +00:00
|
|
|
buf[0] = h;
|
|
|
|
buf[1] = h >> 8;
|
2008-04-18 00:44:58 +00:00
|
|
|
em28xx_write_regs(dev, EM28XX_R30_HSCALELOW, (char *)buf, 2);
|
2009-06-29 14:35:05 +00:00
|
|
|
|
2005-11-09 05:38:10 +00:00
|
|
|
buf[0] = v;
|
|
|
|
buf[1] = v >> 8;
|
2008-04-18 00:44:58 +00:00
|
|
|
em28xx_write_regs(dev, EM28XX_R32_VSCALELOW, (char *)buf, 2);
|
2008-04-18 00:41:10 +00:00
|
|
|
/* it seems that both H and V scalers must be active
|
|
|
|
to work correctly */
|
2009-02-11 02:28:24 +00:00
|
|
|
mode = (h || v) ? 0x30 : 0x00;
|
2005-11-09 05:37:33 +00:00
|
|
|
}
|
2008-04-18 00:44:58 +00:00
|
|
|
return em28xx_write_reg_bits(dev, EM28XX_R26_COMPR, mode, 0x30);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* FIXME: this only function read values from dev */
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_resolution_set(struct em28xx *dev)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
|
|
|
int width, height;
|
|
|
|
width = norm_maxw(dev);
|
2009-08-08 06:14:55 +00:00
|
|
|
height = norm_maxh(dev);
|
|
|
|
|
2009-11-25 02:17:25 +00:00
|
|
|
/* Properly setup VBI */
|
|
|
|
dev->vbi_width = 720;
|
|
|
|
if (dev->norm & V4L2_STD_525_60)
|
|
|
|
dev->vbi_height = 12;
|
|
|
|
else
|
|
|
|
dev->vbi_height = 18;
|
|
|
|
|
2009-08-08 06:14:55 +00:00
|
|
|
if (!dev->progressive)
|
|
|
|
height >>= norm_maxh(dev);
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2008-12-20 12:06:37 +00:00
|
|
|
em28xx_set_outfmt(dev);
|
2009-06-29 14:35:05 +00:00
|
|
|
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
em28xx_accumulator_set(dev, 1, (width - 4) >> 2, 1, (height - 4) >> 2);
|
2009-09-01 04:19:46 +00:00
|
|
|
|
2010-01-22 05:05:24 +00:00
|
|
|
/* If we don't set the start position to 2 in VBI mode, we end up
|
|
|
|
with line 20/21 being YUYV encoded instead of being in 8-bit
|
|
|
|
greyscale. The core of the issue is that line 21 (and line 23 for
|
|
|
|
PAL WSS) are inside of active video region, and as a result they
|
|
|
|
get the pixelformatting associated with that area. So by cropping
|
|
|
|
it out, we end up with the same format as the rest of the VBI
|
|
|
|
region */
|
2009-09-01 04:19:46 +00:00
|
|
|
if (em28xx_vbi_supported(dev) == 1)
|
2010-01-22 05:05:24 +00:00
|
|
|
em28xx_capture_area_set(dev, 0, 2, width >> 2, height >> 2);
|
2009-09-01 04:19:46 +00:00
|
|
|
else
|
|
|
|
em28xx_capture_area_set(dev, 0, 0, width >> 2, height >> 2);
|
2009-06-29 14:35:05 +00:00
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
return em28xx_scaler_set(dev, dev->hscale, dev->vscale);
|
2005-11-09 05:37:07 +00:00
|
|
|
}
|
|
|
|
|
2005-11-09 05:38:27 +00:00
|
|
|
int em28xx_set_alternate(struct em28xx *dev)
|
2005-11-09 05:37:07 +00:00
|
|
|
{
|
|
|
|
int errCode, prev_alt = dev->alt;
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
int i;
|
2008-04-13 18:11:08 +00:00
|
|
|
unsigned int min_pkt_size = dev->width * 2 + 4;
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
|
2009-11-07 17:51:01 +00:00
|
|
|
/*
|
|
|
|
* alt = 0 is used only for control messages, so, only values
|
|
|
|
* greater than 0 can be used for streaming.
|
|
|
|
*/
|
|
|
|
if (alt && alt < dev->num_alt) {
|
|
|
|
em28xx_coredbg("alternate forced to %d\n", dev->alt);
|
|
|
|
dev->alt = alt;
|
|
|
|
goto set_alt;
|
|
|
|
}
|
|
|
|
|
2008-04-22 17:45:47 +00:00
|
|
|
/* When image size is bigger than a certain value,
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
the frame size should be increased, otherwise, only
|
|
|
|
green screen will be received.
|
|
|
|
*/
|
2008-04-13 18:11:08 +00:00
|
|
|
if (dev->width * 2 * dev->height > 720 * 240 * 2)
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
min_pkt_size *= 2;
|
|
|
|
|
2008-04-22 17:45:47 +00:00
|
|
|
for (i = 0; i < dev->num_alt; i++) {
|
|
|
|
/* stop when the selected alt setting offers enough bandwidth */
|
|
|
|
if (dev->alt_max_pkt_size[i] >= min_pkt_size) {
|
|
|
|
dev->alt = i;
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
break;
|
2008-04-22 17:45:47 +00:00
|
|
|
/* otherwise make sure that we end up with the maximum bandwidth
|
|
|
|
because the min_pkt_size equation might be wrong...
|
|
|
|
*/
|
|
|
|
} else if (dev->alt_max_pkt_size[i] >
|
|
|
|
dev->alt_max_pkt_size[dev->alt])
|
|
|
|
dev->alt = i;
|
|
|
|
}
|
2005-11-09 05:37:07 +00:00
|
|
|
|
2009-11-07 17:51:01 +00:00
|
|
|
set_alt:
|
2005-11-09 05:37:07 +00:00
|
|
|
if (dev->alt != prev_alt) {
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
em28xx_coredbg("minimum isoc packet size: %u (alt=%d)\n",
|
|
|
|
min_pkt_size, dev->alt);
|
2005-11-09 05:37:07 +00:00
|
|
|
dev->max_pkt_size = dev->alt_max_pkt_size[dev->alt];
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
em28xx_coredbg("setting alternate %d with wMaxPacketSize=%u\n",
|
|
|
|
dev->alt, dev->max_pkt_size);
|
2005-11-09 05:37:07 +00:00
|
|
|
errCode = usb_set_interface(dev->udev, 0, dev->alt);
|
|
|
|
if (errCode < 0) {
|
2008-04-18 00:41:10 +00:00
|
|
|
em28xx_errdev("cannot change alternate number to %d (error=%i)\n",
|
V4L/DVB (7179): Allow more than one em28xx board
em28xx driver is capable of handling more than one usb device. However, isoc
transfers require a large amount of data to be transfered.
Before this patch, just one em28xx board were enough to allocate more than 50%
URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=480/800 us (60%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, only one board could use an USB host at the same time. After the patch, it
is possible to use more than one em28xx at the same time, on the same usb host,
if the image size is slower or equal to 345600, since those images will
require about 30% of the URBs:
T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=480 MxCh= 8
B: Alloc=232/800 us (29%), #Int= 0, #Iso= 2
D: Ver= 2.00 Cls=09(hub ) Sub=00 Prot=01 MxPS=64 #Cfgs= 1
So, in thesis, after the patch, it would be possible to use up to 3 boards by
each usb host, if the devices are generating small images.
Signed-off-by: Mauro Carvalho Chehab <mchehab@infradead.org>
2008-02-08 18:44:25 +00:00
|
|
|
dev->alt, errCode);
|
2005-11-09 05:37:07 +00:00
|
|
|
return errCode;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
2008-04-18 00:40:16 +00:00
|
|
|
|
2008-04-18 00:48:00 +00:00
|
|
|
int em28xx_gpio_set(struct em28xx *dev, struct em28xx_reg_seq *gpio)
|
|
|
|
{
|
|
|
|
int rc = 0;
|
|
|
|
|
|
|
|
if (!gpio)
|
|
|
|
return rc;
|
|
|
|
|
2008-11-27 12:10:40 +00:00
|
|
|
if (dev->mode != EM28XX_SUSPEND) {
|
|
|
|
em28xx_write_reg(dev, 0x48, 0x00);
|
|
|
|
if (dev->mode == EM28XX_ANALOG_MODE)
|
|
|
|
em28xx_write_reg(dev, EM28XX_R12_VINENABLE, 0x67);
|
|
|
|
else
|
|
|
|
em28xx_write_reg(dev, EM28XX_R12_VINENABLE, 0x37);
|
|
|
|
msleep(6);
|
|
|
|
}
|
2008-04-18 00:48:00 +00:00
|
|
|
|
|
|
|
/* Send GPIO reset sequences specified at board entry */
|
|
|
|
while (gpio->sleep >= 0) {
|
|
|
|
if (gpio->reg >= 0) {
|
|
|
|
rc = em28xx_write_reg_bits(dev,
|
|
|
|
gpio->reg,
|
|
|
|
gpio->val,
|
|
|
|
gpio->mask);
|
|
|
|
if (rc < 0)
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
if (gpio->sleep > 0)
|
|
|
|
msleep(gpio->sleep);
|
|
|
|
|
|
|
|
gpio++;
|
|
|
|
}
|
|
|
|
return rc;
|
|
|
|
}
|
2011-07-04 00:05:06 +00:00
|
|
|
EXPORT_SYMBOL_GPL(em28xx_gpio_set);
|
2008-04-18 00:48:00 +00:00
|
|
|
|
|
|
|
int em28xx_set_mode(struct em28xx *dev, enum em28xx_mode set_mode)
|
|
|
|
{
|
|
|
|
if (dev->mode == set_mode)
|
|
|
|
return 0;
|
|
|
|
|
2008-11-27 12:10:40 +00:00
|
|
|
if (set_mode == EM28XX_SUSPEND) {
|
2008-04-18 00:48:00 +00:00
|
|
|
dev->mode = set_mode;
|
2008-11-27 12:10:40 +00:00
|
|
|
|
|
|
|
/* FIXME: add suspend support for ac97 */
|
|
|
|
|
|
|
|
return em28xx_gpio_set(dev, dev->board.suspend_gpio);
|
2008-04-18 00:48:00 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
dev->mode = set_mode;
|
|
|
|
|
|
|
|
if (dev->mode == EM28XX_DIGITAL_MODE)
|
2008-11-27 12:00:00 +00:00
|
|
|
return em28xx_gpio_set(dev, dev->board.dvb_gpio);
|
2008-04-18 00:48:00 +00:00
|
|
|
else
|
2008-11-27 12:00:00 +00:00
|
|
|
return em28xx_gpio_set(dev, INPUT(dev->ctl_input)->gpio);
|
2008-04-18 00:48:00 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_set_mode);
|
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
/* ------------------------------------------------------------------
|
|
|
|
URB control
|
|
|
|
------------------------------------------------------------------*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* IRQ callback, called by URB callback
|
|
|
|
*/
|
|
|
|
static void em28xx_irq_callback(struct urb *urb)
|
|
|
|
{
|
2009-09-01 04:19:46 +00:00
|
|
|
struct em28xx *dev = urb->context;
|
2011-06-17 23:28:51 +00:00
|
|
|
int i;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
2009-02-08 16:09:11 +00:00
|
|
|
switch (urb->status) {
|
|
|
|
case 0: /* success */
|
|
|
|
case -ETIMEDOUT: /* NAK */
|
|
|
|
break;
|
|
|
|
case -ECONNRESET: /* kill */
|
|
|
|
case -ENOENT:
|
|
|
|
case -ESHUTDOWN:
|
|
|
|
return;
|
|
|
|
default: /* error */
|
|
|
|
em28xx_isocdbg("urb completition error %d.\n", urb->status);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
/* Copy data from URB */
|
|
|
|
spin_lock(&dev->slock);
|
2011-06-17 23:28:51 +00:00
|
|
|
dev->isoc_ctl.isoc_copy(dev, urb);
|
2008-04-18 00:40:16 +00:00
|
|
|
spin_unlock(&dev->slock);
|
|
|
|
|
|
|
|
/* Reset urb buffers */
|
|
|
|
for (i = 0; i < urb->number_of_packets; i++) {
|
|
|
|
urb->iso_frame_desc[i].status = 0;
|
|
|
|
urb->iso_frame_desc[i].actual_length = 0;
|
|
|
|
}
|
|
|
|
urb->status = 0;
|
|
|
|
|
|
|
|
urb->status = usb_submit_urb(urb, GFP_ATOMIC);
|
|
|
|
if (urb->status) {
|
2008-04-21 10:01:09 +00:00
|
|
|
em28xx_isocdbg("urb resubmit failed (error=%i)\n",
|
|
|
|
urb->status);
|
2008-04-18 00:40:16 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Stop and Deallocate URBs
|
|
|
|
*/
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
void em28xx_uninit_isoc(struct em28xx *dev, enum em28xx_mode mode)
|
2008-04-18 00:40:16 +00:00
|
|
|
{
|
|
|
|
struct urb *urb;
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
struct em28xx_usb_isoc_bufs *isoc_bufs;
|
2008-04-18 00:40:16 +00:00
|
|
|
int i;
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_isocdbg("em28xx: called em28xx_uninit_isoc in mode %d\n", mode);
|
|
|
|
|
|
|
|
if (mode == EM28XX_DIGITAL_MODE)
|
|
|
|
isoc_bufs = &dev->isoc_ctl.digital_bufs;
|
|
|
|
else
|
|
|
|
isoc_bufs = &dev->isoc_ctl.analog_bufs;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
for (i = 0; i < isoc_bufs->num_bufs; i++) {
|
|
|
|
urb = isoc_bufs->urb[i];
|
2008-04-18 00:40:16 +00:00
|
|
|
if (urb) {
|
2009-01-25 16:08:07 +00:00
|
|
|
if (!irqs_disabled())
|
|
|
|
usb_kill_urb(urb);
|
|
|
|
else
|
|
|
|
usb_unlink_urb(urb);
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
if (isoc_bufs->transfer_buffer[i]) {
|
2010-04-12 11:17:25 +00:00
|
|
|
usb_free_coherent(dev->udev,
|
2008-04-18 00:41:10 +00:00
|
|
|
urb->transfer_buffer_length,
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->transfer_buffer[i],
|
2008-04-18 00:41:10 +00:00
|
|
|
urb->transfer_dma);
|
2008-04-18 00:40:16 +00:00
|
|
|
}
|
|
|
|
usb_free_urb(urb);
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->urb[i] = NULL;
|
2008-04-18 00:40:16 +00:00
|
|
|
}
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->transfer_buffer[i] = NULL;
|
2008-04-18 00:40:16 +00:00
|
|
|
}
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
kfree(isoc_bufs->urb);
|
|
|
|
kfree(isoc_bufs->transfer_buffer);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->urb = NULL;
|
|
|
|
isoc_bufs->transfer_buffer = NULL;
|
|
|
|
isoc_bufs->num_bufs = 0;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
em28xx_capture_start(dev, 0);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_uninit_isoc);
|
|
|
|
|
2012-03-22 11:48:17 +00:00
|
|
|
/*
|
|
|
|
* Stop URBs
|
|
|
|
*/
|
|
|
|
void em28xx_stop_urbs(struct em28xx *dev)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct urb *urb;
|
|
|
|
struct em28xx_usb_isoc_bufs *isoc_bufs = &dev->isoc_ctl.digital_bufs;
|
|
|
|
|
|
|
|
em28xx_isocdbg("em28xx: called em28xx_stop_urbs\n");
|
|
|
|
|
|
|
|
for (i = 0; i < isoc_bufs->num_bufs; i++) {
|
|
|
|
urb = isoc_bufs->urb[i];
|
|
|
|
if (urb) {
|
|
|
|
if (!irqs_disabled())
|
|
|
|
usb_kill_urb(urb);
|
|
|
|
else
|
|
|
|
usb_unlink_urb(urb);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
em28xx_capture_start(dev, 0);
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_stop_urbs);
|
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
/*
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
* Allocate URBs
|
2008-04-18 00:40:16 +00:00
|
|
|
*/
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
int em28xx_alloc_isoc(struct em28xx *dev, enum em28xx_mode mode,
|
|
|
|
int max_packets, int num_bufs, int max_pkt_size)
|
2008-04-18 00:40:16 +00:00
|
|
|
{
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
struct em28xx_usb_isoc_bufs *isoc_bufs;
|
2008-04-18 00:40:16 +00:00
|
|
|
int i;
|
|
|
|
int sb_size, pipe;
|
|
|
|
struct urb *urb;
|
|
|
|
int j, k;
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_isocdbg("em28xx: called em28xx_alloc_isoc in mode %d\n", mode);
|
|
|
|
|
|
|
|
if (mode == EM28XX_DIGITAL_MODE)
|
|
|
|
isoc_bufs = &dev->isoc_ctl.digital_bufs;
|
|
|
|
else
|
|
|
|
isoc_bufs = &dev->isoc_ctl.analog_bufs;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
/* De-allocates all pending stuff */
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_uninit_isoc(dev, mode);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->num_bufs = num_bufs;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->urb = kzalloc(sizeof(void *)*num_bufs, GFP_KERNEL);
|
|
|
|
if (!isoc_bufs->urb) {
|
2008-04-18 00:40:16 +00:00
|
|
|
em28xx_errdev("cannot alloc memory for usb buffers\n");
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->transfer_buffer = kzalloc(sizeof(void *)*num_bufs,
|
|
|
|
GFP_KERNEL);
|
|
|
|
if (!isoc_bufs->transfer_buffer) {
|
2009-04-09 21:27:12 +00:00
|
|
|
em28xx_errdev("cannot allocate memory for usb transfer\n");
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
kfree(isoc_bufs->urb);
|
2008-04-18 00:40:16 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->max_pkt_size = max_pkt_size;
|
|
|
|
isoc_bufs->num_packets = max_packets;
|
2009-09-01 04:54:54 +00:00
|
|
|
dev->isoc_ctl.vid_buf = NULL;
|
|
|
|
dev->isoc_ctl.vbi_buf = NULL;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
sb_size = isoc_bufs->num_packets * isoc_bufs->max_pkt_size;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
/* allocate urbs and transfer buffers */
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
for (i = 0; i < isoc_bufs->num_bufs; i++) {
|
|
|
|
urb = usb_alloc_urb(isoc_bufs->num_packets, GFP_KERNEL);
|
2008-04-18 00:40:16 +00:00
|
|
|
if (!urb) {
|
|
|
|
em28xx_err("cannot alloc isoc_ctl.urb %i\n", i);
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_uninit_isoc(dev, mode);
|
2008-04-18 00:40:16 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->urb[i] = urb;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->transfer_buffer[i] = usb_alloc_coherent(dev->udev,
|
2008-04-18 00:40:16 +00:00
|
|
|
sb_size, GFP_KERNEL, &urb->transfer_dma);
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
if (!isoc_bufs->transfer_buffer[i]) {
|
2008-04-18 00:40:16 +00:00
|
|
|
em28xx_err("unable to allocate %i bytes for transfer"
|
|
|
|
" buffer %i%s\n",
|
|
|
|
sb_size, i,
|
2009-02-11 02:28:24 +00:00
|
|
|
in_interrupt() ? " while in int" : "");
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_uninit_isoc(dev, mode);
|
2008-04-18 00:40:16 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
memset(isoc_bufs->transfer_buffer[i], 0, sb_size);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
/* FIXME: this is a hack - should be
|
|
|
|
'desc.bEndpointAddress & USB_ENDPOINT_NUMBER_MASK'
|
|
|
|
should also be using 'desc.bInterval'
|
|
|
|
*/
|
2008-04-18 00:41:10 +00:00
|
|
|
pipe = usb_rcvisocpipe(dev->udev,
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
mode == EM28XX_ANALOG_MODE ?
|
2011-12-28 21:55:41 +00:00
|
|
|
EM28XX_EP_ANALOG : EM28XX_EP_DIGITAL);
|
2008-04-18 00:41:10 +00:00
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
usb_fill_int_urb(urb, dev->udev, pipe,
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->transfer_buffer[i], sb_size,
|
2009-09-01 04:19:46 +00:00
|
|
|
em28xx_irq_callback, dev, 1);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
urb->number_of_packets = isoc_bufs->num_packets;
|
2009-03-23 02:41:28 +00:00
|
|
|
urb->transfer_flags = URB_ISO_ASAP | URB_NO_TRANSFER_DMA_MAP;
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
k = 0;
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
for (j = 0; j < isoc_bufs->num_packets; j++) {
|
2008-04-18 00:40:16 +00:00
|
|
|
urb->iso_frame_desc[j].offset = k;
|
|
|
|
urb->iso_frame_desc[j].length =
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
isoc_bufs->max_pkt_size;
|
|
|
|
k += isoc_bufs->max_pkt_size;
|
2008-04-18 00:40:16 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_alloc_isoc);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Allocate URBs and start IRQ
|
|
|
|
*/
|
|
|
|
int em28xx_init_isoc(struct em28xx *dev, enum em28xx_mode mode,
|
|
|
|
int max_packets, int num_bufs, int max_pkt_size,
|
|
|
|
int (*isoc_copy) (struct em28xx *dev, struct urb *urb))
|
|
|
|
{
|
|
|
|
struct em28xx_dmaqueue *dma_q = &dev->vidq;
|
|
|
|
struct em28xx_dmaqueue *vbi_dma_q = &dev->vbiq;
|
|
|
|
struct em28xx_usb_isoc_bufs *isoc_bufs;
|
|
|
|
int i;
|
|
|
|
int rc;
|
|
|
|
int alloc;
|
|
|
|
|
|
|
|
em28xx_isocdbg("em28xx: called em28xx_init_isoc in mode %d\n", mode);
|
|
|
|
|
|
|
|
dev->isoc_ctl.isoc_copy = isoc_copy;
|
|
|
|
|
|
|
|
if (mode == EM28XX_DIGITAL_MODE) {
|
|
|
|
isoc_bufs = &dev->isoc_ctl.digital_bufs;
|
|
|
|
/* no need to free/alloc isoc buffers in digital mode */
|
|
|
|
alloc = 0;
|
|
|
|
} else {
|
|
|
|
isoc_bufs = &dev->isoc_ctl.analog_bufs;
|
|
|
|
alloc = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (alloc) {
|
|
|
|
rc = em28xx_alloc_isoc(dev, mode, max_packets,
|
|
|
|
num_bufs, max_pkt_size);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2008-04-18 00:40:16 +00:00
|
|
|
init_waitqueue_head(&dma_q->wq);
|
2009-09-01 04:54:54 +00:00
|
|
|
init_waitqueue_head(&vbi_dma_q->wq);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
2008-04-18 00:48:00 +00:00
|
|
|
em28xx_capture_start(dev, 1);
|
2008-04-18 00:40:16 +00:00
|
|
|
|
|
|
|
/* submit urbs and enables IRQ */
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
for (i = 0; i < isoc_bufs->num_bufs; i++) {
|
|
|
|
rc = usb_submit_urb(isoc_bufs->urb[i], GFP_ATOMIC);
|
2008-04-18 00:40:16 +00:00
|
|
|
if (rc) {
|
|
|
|
em28xx_err("submit of urb %i failed (error=%i)\n", i,
|
|
|
|
rc);
|
[media] em28xx: pre-allocate DVB isoc transfer buffers
On MIPS/ARM set-top-boxes, as well as old x86 PCs, memory allocation failures
in the em28xx driver are common, due to memory fragmentation over time, that
makes impossible to allocate large chunks of coherent memory.
A typical system with 256/512 MB of RAM fails after just 1 day of uptime (see
the old thread for detailed reports and crashlogs).
In fact, the em28xx driver allocates memory for USB isoc transfers at runtime,
as opposite to the dvb-usb drivers that allocates the USB buffers when the
device is initialized, and frees them when the device is disconnected.
Moreover, in digital mode the USB isoc transfer buffers are freed, allocated
and cleared every time the user selects a new channel, wasting time and
resources.
This patch solves both problems by allocating DVB isoc transfer buffers in
em28xx_usb_probe(), and freeing them in em28xx_usb_disconnect().
In fact, the buffers size and number depend only on the max USB packet size
that is parsed from the USB descriptors in em28xx_usb_probe(), so it can
never change for a given device.
This approach makes no sense in analog mode (as the buffer size depends on
the alternate mode selected at runtime), the patch creates two separate sets
of buffers for digital and analog modes.
For digital-only devices, USB buffers are created when the device is probed
and freed when the device is disconnected.
For analog-only devices, nothing changes: isoc buffers are created at runtime.
For hybrid devices, two sets of buffers are maintained: the digital-mode
buffers are created when the device is probed, and freed when the device is
disconnected; analog-mode buffers are created/destroyed at runtime as before.
So, in analog mode, digital and analog buffers coexists at the same time: this
can be justified by the fact that digital mode is by far more commonly used
nowadays, so it makes sense to optimize the driver for this use case scenario.
The patch has been tested in the last few days on a x86 PC and a MIPS
set-top-box, with the PCTV 290e (digital only) and the Terratec Hybrid XS
(hybrid device). With the latter, I switched several times between analog and
digital mode (Kaffeine/TvTime) with no issue at all.
I unplugged/plugged the devices several times with no problem.
Also, after over 3 days of normal usage in the MPIS set-top-box, the PCTV 290e
was still up and running.
Signed-off-by: Gianluca Gennari <gennarone@gmail.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
2012-02-13 16:59:22 +00:00
|
|
|
em28xx_uninit_isoc(dev, mode);
|
2008-04-18 00:40:16 +00:00
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(em28xx_init_isoc);
|
2008-12-29 01:18:14 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* em28xx_wake_i2c()
|
|
|
|
* configure i2c attached devices
|
|
|
|
*/
|
|
|
|
void em28xx_wake_i2c(struct em28xx *dev)
|
|
|
|
{
|
2009-04-02 14:26:22 +00:00
|
|
|
v4l2_device_call_all(&dev->v4l2_dev, 0, core, reset, 0);
|
|
|
|
v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_routing,
|
|
|
|
INPUT(dev->ctl_input)->vmux, 0, 0);
|
2009-03-31 20:10:58 +00:00
|
|
|
v4l2_device_call_all(&dev->v4l2_dev, 0, video, s_stream, 0);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Device control list
|
|
|
|
*/
|
|
|
|
|
|
|
|
static LIST_HEAD(em28xx_devlist);
|
|
|
|
static DEFINE_MUTEX(em28xx_devlist_mutex);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Extension interface
|
|
|
|
*/
|
|
|
|
|
|
|
|
static LIST_HEAD(em28xx_extension_devlist);
|
|
|
|
|
|
|
|
int em28xx_register_extension(struct em28xx_ops *ops)
|
|
|
|
{
|
|
|
|
struct em28xx *dev = NULL;
|
|
|
|
|
|
|
|
mutex_lock(&em28xx_devlist_mutex);
|
|
|
|
list_add_tail(&ops->next, &em28xx_extension_devlist);
|
|
|
|
list_for_each_entry(dev, &em28xx_devlist, devlist) {
|
2010-05-22 19:53:27 +00:00
|
|
|
ops->init(dev);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
|
|
|
mutex_unlock(&em28xx_devlist_mutex);
|
2011-08-20 11:31:05 +00:00
|
|
|
printk(KERN_INFO "Em28xx: Initialized (%s) extension\n", ops->name);
|
2008-12-29 01:18:14 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(em28xx_register_extension);
|
|
|
|
|
|
|
|
void em28xx_unregister_extension(struct em28xx_ops *ops)
|
|
|
|
{
|
|
|
|
struct em28xx *dev = NULL;
|
|
|
|
|
|
|
|
mutex_lock(&em28xx_devlist_mutex);
|
|
|
|
list_for_each_entry(dev, &em28xx_devlist, devlist) {
|
2010-05-22 19:53:27 +00:00
|
|
|
ops->fini(dev);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
|
|
|
list_del(&ops->next);
|
|
|
|
mutex_unlock(&em28xx_devlist_mutex);
|
2011-08-20 11:31:05 +00:00
|
|
|
printk(KERN_INFO "Em28xx: Removed (%s) extension\n", ops->name);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(em28xx_unregister_extension);
|
|
|
|
|
|
|
|
void em28xx_init_extension(struct em28xx *dev)
|
|
|
|
{
|
2011-08-20 11:28:17 +00:00
|
|
|
const struct em28xx_ops *ops = NULL;
|
2008-12-29 01:18:14 +00:00
|
|
|
|
2010-04-07 20:07:58 +00:00
|
|
|
mutex_lock(&em28xx_devlist_mutex);
|
2011-08-20 11:28:17 +00:00
|
|
|
list_add_tail(&dev->devlist, &em28xx_devlist);
|
|
|
|
list_for_each_entry(ops, &em28xx_extension_devlist, next) {
|
|
|
|
if (ops->init)
|
|
|
|
ops->init(dev);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
2010-04-07 20:07:58 +00:00
|
|
|
mutex_unlock(&em28xx_devlist_mutex);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void em28xx_close_extension(struct em28xx *dev)
|
|
|
|
{
|
2011-09-24 14:02:32 +00:00
|
|
|
const struct em28xx_ops *ops = NULL;
|
2008-12-29 01:18:14 +00:00
|
|
|
|
2010-04-07 20:07:58 +00:00
|
|
|
mutex_lock(&em28xx_devlist_mutex);
|
2011-09-24 14:02:32 +00:00
|
|
|
list_for_each_entry(ops, &em28xx_extension_devlist, next) {
|
|
|
|
if (ops->fini)
|
|
|
|
ops->fini(dev);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|
2011-09-24 14:02:32 +00:00
|
|
|
list_del(&dev->devlist);
|
2010-04-07 20:07:58 +00:00
|
|
|
mutex_unlock(&em28xx_devlist_mutex);
|
2008-12-29 01:18:14 +00:00
|
|
|
}
|