linux/drivers/firewire/core-device.c

1329 lines
34 KiB
C
Raw Normal View History

/*
* Device probing and sysfs code.
*
* Copyright (C) 2005-2006 Kristian Hoegsberg <krh@bitplanet.net>
*
* 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., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*/
#include <linux/bug.h>
#include <linux/ctype.h>
#include <linux/delay.h>
#include <linux/device.h>
#include <linux/errno.h>
#include <linux/firewire.h>
#include <linux/firewire-constants.h>
#include <linux/idr.h>
#include <linux/jiffies.h>
#include <linux/kobject.h>
#include <linux/list.h>
#include <linux/mod_devicetable.h>
#include <linux/module.h>
#include <linux/mutex.h>
#include <linux/random.h>
#include <linux/rwsem.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>
#include <linux/spinlock.h>
#include <linux/string.h>
#include <linux/workqueue.h>
#include <linux/atomic.h>
#include <asm/byteorder.h>
#include "core.h"
void fw_csr_iterator_init(struct fw_csr_iterator *ci, const u32 *p)
{
ci->p = p + 1;
ci->end = ci->p + (p[0] >> 16);
}
EXPORT_SYMBOL(fw_csr_iterator_init);
int fw_csr_iterator_next(struct fw_csr_iterator *ci, int *key, int *value)
{
*key = *ci->p >> 24;
*value = *ci->p & 0xffffff;
return ci->p++ < ci->end;
}
EXPORT_SYMBOL(fw_csr_iterator_next);
static const u32 *search_leaf(const u32 *directory, int search_key)
{
struct fw_csr_iterator ci;
int last_key = 0, key, value;
fw_csr_iterator_init(&ci, directory);
while (fw_csr_iterator_next(&ci, &key, &value)) {
if (last_key == search_key &&
key == (CSR_DESCRIPTOR | CSR_LEAF))
return ci.p - 1 + value;
last_key = key;
}
return NULL;
}
static int textual_leaf_to_string(const u32 *block, char *buf, size_t size)
{
unsigned int quadlets, i;
char c;
if (!size || !buf)
return -EINVAL;
quadlets = min(block[0] >> 16, 256U);
if (quadlets < 2)
return -ENODATA;
if (block[1] != 0 || block[2] != 0)
/* unknown language/character set */
return -ENODATA;
block += 3;
quadlets -= 2;
for (i = 0; i < quadlets * 4 && i < size - 1; i++) {
c = block[i / 4] >> (24 - 8 * (i % 4));
if (c == '\0')
break;
buf[i] = c;
}
buf[i] = '\0';
return i;
}
/**
* fw_csr_string() - reads a string from the configuration ROM
* @directory: e.g. root directory or unit directory
* @key: the key of the preceding directory entry
* @buf: where to put the string
* @size: size of @buf, in bytes
*
* The string is taken from a minimal ASCII text descriptor leaf after
* the immediate entry with @key. The string is zero-terminated.
* An overlong string is silently truncated such that it and the
* zero byte fit into @size.
*
* Returns strlen(buf) or a negative error code.
*/
int fw_csr_string(const u32 *directory, int key, char *buf, size_t size)
{
const u32 *leaf = search_leaf(directory, key);
if (!leaf)
return -ENOENT;
return textual_leaf_to_string(leaf, buf, size);
}
EXPORT_SYMBOL(fw_csr_string);
static void get_ids(const u32 *directory, int *id)
{
struct fw_csr_iterator ci;
int key, value;
fw_csr_iterator_init(&ci, directory);
while (fw_csr_iterator_next(&ci, &key, &value)) {
switch (key) {
case CSR_VENDOR: id[0] = value; break;
case CSR_MODEL: id[1] = value; break;
case CSR_SPECIFIER_ID: id[2] = value; break;
case CSR_VERSION: id[3] = value; break;
}
}
}
static void get_modalias_ids(struct fw_unit *unit, int *id)
{
get_ids(&fw_parent_device(unit)->config_rom[5], id);
get_ids(unit->directory, id);
}
static bool match_ids(const struct ieee1394_device_id *id_table, int *id)
{
int match = 0;
if (id[0] == id_table->vendor_id)
match |= IEEE1394_MATCH_VENDOR_ID;
if (id[1] == id_table->model_id)
match |= IEEE1394_MATCH_MODEL_ID;
if (id[2] == id_table->specifier_id)
match |= IEEE1394_MATCH_SPECIFIER_ID;
if (id[3] == id_table->version)
match |= IEEE1394_MATCH_VERSION;
return (match & id_table->match_flags) == id_table->match_flags;
}
static const struct ieee1394_device_id *unit_match(struct device *dev,
struct device_driver *drv)
{
const struct ieee1394_device_id *id_table =
container_of(drv, struct fw_driver, driver)->id_table;
int id[] = {0, 0, 0, 0};
get_modalias_ids(fw_unit(dev), id);
for (; id_table->match_flags != 0; id_table++)
if (match_ids(id_table, id))
return id_table;
return NULL;
}
static bool is_fw_unit(struct device *dev);
static int fw_unit_match(struct device *dev, struct device_driver *drv)
{
/* We only allow binding to fw_units. */
return is_fw_unit(dev) && unit_match(dev, drv) != NULL;
}
static int fw_unit_probe(struct device *dev)
{
struct fw_driver *driver =
container_of(dev->driver, struct fw_driver, driver);
return driver->probe(fw_unit(dev), unit_match(dev, dev->driver));
}
static int fw_unit_remove(struct device *dev)
{
struct fw_driver *driver =
container_of(dev->driver, struct fw_driver, driver);
return driver->remove(fw_unit(dev)), 0;
}
static int get_modalias(struct fw_unit *unit, char *buffer, size_t buffer_size)
{
firewire: core: fix Model_ID in modalias The modalias string of devices that represent units on a FireWire node did not show Module_ID entries within unit directories. This was because firewire-core searched only the root directory of the configuration ROM for a Model_ID entry. We now search first the root directory, then the unit directory. IOW honor a unit directory's Model_ID if present, otherwise fall back to the root directory's model ID (if present). Furthermore, apply the same change to Vendor_ID. This had the same issue but it was less apparent because most devices provide Vendor_ID only in the root directory. And finally, also use this strategy for the remaining two IDs in the modalias, Specifier_ID and Version. It does not actually make sense to look for them elsewhere than in the unit directory because they are mandatory there. However, a uniform search order simplifies the implementation and has no adverse affect in practice. Side notes: - The older counterpart of this, nodemgr.c of ieee1394, looked for Vendor_ID first in the root directory, then in the unit directory, and for Model_ID only in the unit directory. - There is a single mainline driver which requires Vendor_ID and Model_ID --- the firedtv driver. This one worked because FireDTVs provide Vendor_ID in the root directory and Model_ID identically in root directory and unit directory. - Apart from firedtv, there are currently no drivers known to me (including userspace drivers) that look at the Vendor_ID or Model_ID of the modalias. Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2010-03-18 23:38:29 +00:00
int id[] = {0, 0, 0, 0};
get_modalias_ids(unit, id);
return snprintf(buffer, buffer_size,
"ieee1394:ven%08Xmo%08Xsp%08Xver%08X",
firewire: core: fix Model_ID in modalias The modalias string of devices that represent units on a FireWire node did not show Module_ID entries within unit directories. This was because firewire-core searched only the root directory of the configuration ROM for a Model_ID entry. We now search first the root directory, then the unit directory. IOW honor a unit directory's Model_ID if present, otherwise fall back to the root directory's model ID (if present). Furthermore, apply the same change to Vendor_ID. This had the same issue but it was less apparent because most devices provide Vendor_ID only in the root directory. And finally, also use this strategy for the remaining two IDs in the modalias, Specifier_ID and Version. It does not actually make sense to look for them elsewhere than in the unit directory because they are mandatory there. However, a uniform search order simplifies the implementation and has no adverse affect in practice. Side notes: - The older counterpart of this, nodemgr.c of ieee1394, looked for Vendor_ID first in the root directory, then in the unit directory, and for Model_ID only in the unit directory. - There is a single mainline driver which requires Vendor_ID and Model_ID --- the firedtv driver. This one worked because FireDTVs provide Vendor_ID in the root directory and Model_ID identically in root directory and unit directory. - Apart from firedtv, there are currently no drivers known to me (including userspace drivers) that look at the Vendor_ID or Model_ID of the modalias. Reported-by: Maciej Żenczykowski <zenczykowski@gmail.com> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2010-03-18 23:38:29 +00:00
id[0], id[1], id[2], id[3]);
}
static int fw_unit_uevent(struct device *dev, struct kobj_uevent_env *env)
{
struct fw_unit *unit = fw_unit(dev);
char modalias[64];
get_modalias(unit, modalias, sizeof(modalias));
if (add_uevent_var(env, "MODALIAS=%s", modalias))
return -ENOMEM;
return 0;
}
struct bus_type fw_bus_type = {
.name = "firewire",
.match = fw_unit_match,
.probe = fw_unit_probe,
.remove = fw_unit_remove,
};
EXPORT_SYMBOL(fw_bus_type);
int fw_device_enable_phys_dma(struct fw_device *device)
{
int generation = device->generation;
/* device->node_id, accessed below, must not be older than generation */
smp_rmb();
return device->card->driver->enable_phys_dma(device->card,
device->node_id,
generation);
}
EXPORT_SYMBOL(fw_device_enable_phys_dma);
struct config_rom_attribute {
struct device_attribute attr;
u32 key;
};
static ssize_t show_immediate(struct device *dev,
struct device_attribute *dattr, char *buf)
{
struct config_rom_attribute *attr =
container_of(dattr, struct config_rom_attribute, attr);
struct fw_csr_iterator ci;
const u32 *dir;
int key, value, ret = -ENOENT;
down_read(&fw_device_rwsem);
if (is_fw_unit(dev))
dir = fw_unit(dev)->directory;
else
dir = fw_device(dev)->config_rom + 5;
fw_csr_iterator_init(&ci, dir);
while (fw_csr_iterator_next(&ci, &key, &value))
if (attr->key == key) {
ret = snprintf(buf, buf ? PAGE_SIZE : 0,
"0x%06x\n", value);
break;
}
up_read(&fw_device_rwsem);
return ret;
}
#define IMMEDIATE_ATTR(name, key) \
{ __ATTR(name, S_IRUGO, show_immediate, NULL), key }
static ssize_t show_text_leaf(struct device *dev,
struct device_attribute *dattr, char *buf)
{
struct config_rom_attribute *attr =
container_of(dattr, struct config_rom_attribute, attr);
const u32 *dir;
size_t bufsize;
char dummy_buf[2];
int ret;
down_read(&fw_device_rwsem);
if (is_fw_unit(dev))
dir = fw_unit(dev)->directory;
else
dir = fw_device(dev)->config_rom + 5;
if (buf) {
bufsize = PAGE_SIZE - 1;
} else {
buf = dummy_buf;
bufsize = 1;
}
ret = fw_csr_string(dir, attr->key, buf, bufsize);
if (ret >= 0) {
/* Strip trailing whitespace and add newline. */
while (ret > 0 && isspace(buf[ret - 1]))
ret--;
strcpy(buf + ret, "\n");
ret++;
}
up_read(&fw_device_rwsem);
return ret;
}
#define TEXT_LEAF_ATTR(name, key) \
{ __ATTR(name, S_IRUGO, show_text_leaf, NULL), key }
static struct config_rom_attribute config_rom_attributes[] = {
IMMEDIATE_ATTR(vendor, CSR_VENDOR),
IMMEDIATE_ATTR(hardware_version, CSR_HARDWARE_VERSION),
IMMEDIATE_ATTR(specifier_id, CSR_SPECIFIER_ID),
IMMEDIATE_ATTR(version, CSR_VERSION),
IMMEDIATE_ATTR(model, CSR_MODEL),
TEXT_LEAF_ATTR(vendor_name, CSR_VENDOR),
TEXT_LEAF_ATTR(model_name, CSR_MODEL),
TEXT_LEAF_ATTR(hardware_version_name, CSR_HARDWARE_VERSION),
};
static void init_fw_attribute_group(struct device *dev,
struct device_attribute *attrs,
struct fw_attribute_group *group)
{
struct device_attribute *attr;
int i, j;
for (j = 0; attrs[j].attr.name != NULL; j++)
group->attrs[j] = &attrs[j].attr;
for (i = 0; i < ARRAY_SIZE(config_rom_attributes); i++) {
attr = &config_rom_attributes[i].attr;
if (attr->show(dev, attr, NULL) < 0)
continue;
group->attrs[j++] = &attr->attr;
}
group->attrs[j] = NULL;
group->groups[0] = &group->group;
group->groups[1] = NULL;
group->group.attrs = group->attrs;
dev->groups = (const struct attribute_group **) group->groups;
}
static ssize_t modalias_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_unit *unit = fw_unit(dev);
int length;
length = get_modalias(unit, buf, PAGE_SIZE);
strcpy(buf + length, "\n");
return length + 1;
}
static ssize_t rom_index_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_device *device = fw_device(dev->parent);
struct fw_unit *unit = fw_unit(dev);
return snprintf(buf, PAGE_SIZE, "%d\n",
(int)(unit->directory - device->config_rom));
}
static struct device_attribute fw_unit_attributes[] = {
__ATTR_RO(modalias),
__ATTR_RO(rom_index),
__ATTR_NULL,
};
static ssize_t config_rom_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_device *device = fw_device(dev);
size_t length;
down_read(&fw_device_rwsem);
length = device->config_rom_length * 4;
memcpy(buf, device->config_rom, length);
up_read(&fw_device_rwsem);
return length;
}
static ssize_t guid_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_device *device = fw_device(dev);
int ret;
down_read(&fw_device_rwsem);
ret = snprintf(buf, PAGE_SIZE, "0x%08x%08x\n",
device->config_rom[3], device->config_rom[4]);
up_read(&fw_device_rwsem);
return ret;
}
static ssize_t is_local_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_device *device = fw_device(dev);
return sprintf(buf, "%u\n", device->is_local);
}
static int units_sprintf(char *buf, const u32 *directory)
firewire: core: add sysfs attribute for easier udev rules This adds the attribute /sys/bus/firewire/devices/fw[0-9]+/units. It can be used in udev rules like the following ones: # IIDC devices: industrial cameras and some webcams SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x00010?*", GROUP="video" # AV/C devices: camcorders, set-top boxes, TV sets, audio devices, ... SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x010001*", GROUP="video" Background: firewire-core manages two device types: - fw_device is a FireWire node. A character device file is associated with it. - fw_unit is a unit directory on a node. Each fw_device may have 0..n children of type fw_unit. The units tell us what kinds of protocols a node implements. We want to set ownership or ACLs or permissions of the character device file of an fw_device, or/and create symlinks to it, based on available protocols. Until now udev rules had to look at the fw_unit devices and then modify their parent's character device file accordingly. This is problematic for two reasons: 1) It happens sometime after the creation of the fw_device, 2) an access policy may require that information from all children is evaluated before a decision about the parent is made. Problem 1) can ultimately not be avoided since this is the nature of FireWire nodes: They may add or remove unit directories at any point in time. However, we can still help userland a lot by providing the protocol type information of all units in a summary sysfs attribute directly at the fw_device. This way, - the information is immediately available at the affected device when userspace goes about to handle an ADD or CHANGE event of the fw_device, - with most policies, it won't be necessary anymore to dig through child attributes. The new attribute is called "units". It contains space-separated tuples of specifier_id and version of each present unit. The delimiter within tuples is a colon. Specifier_id and version are printed as 0x%06x. Here is an example of a node which implements an IPv4 unit and an IPv6 unit: $ cat /sys/bus/firewire/devices/fw2/units 0x00005e:0x000001 0x00005e:0x000002 Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2009-05-22 22:03:29 +00:00
{
struct fw_csr_iterator ci;
int key, value;
int specifier_id = 0;
int version = 0;
fw_csr_iterator_init(&ci, directory);
while (fw_csr_iterator_next(&ci, &key, &value)) {
switch (key) {
case CSR_SPECIFIER_ID:
specifier_id = value;
break;
case CSR_VERSION:
version = value;
break;
}
}
return sprintf(buf, "0x%06x:0x%06x ", specifier_id, version);
}
static ssize_t units_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct fw_device *device = fw_device(dev);
struct fw_csr_iterator ci;
int key, value, i = 0;
down_read(&fw_device_rwsem);
fw_csr_iterator_init(&ci, &device->config_rom[5]);
while (fw_csr_iterator_next(&ci, &key, &value)) {
if (key != (CSR_UNIT | CSR_DIRECTORY))
continue;
i += units_sprintf(&buf[i], ci.p + value - 1);
if (i >= PAGE_SIZE - (8 + 1 + 8 + 1))
break;
}
up_read(&fw_device_rwsem);
if (i)
buf[i - 1] = '\n';
return i;
}
static struct device_attribute fw_device_attributes[] = {
__ATTR_RO(config_rom),
__ATTR_RO(guid),
__ATTR_RO(is_local),
firewire: core: add sysfs attribute for easier udev rules This adds the attribute /sys/bus/firewire/devices/fw[0-9]+/units. It can be used in udev rules like the following ones: # IIDC devices: industrial cameras and some webcams SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x00010?*", GROUP="video" # AV/C devices: camcorders, set-top boxes, TV sets, audio devices, ... SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x010001*", GROUP="video" Background: firewire-core manages two device types: - fw_device is a FireWire node. A character device file is associated with it. - fw_unit is a unit directory on a node. Each fw_device may have 0..n children of type fw_unit. The units tell us what kinds of protocols a node implements. We want to set ownership or ACLs or permissions of the character device file of an fw_device, or/and create symlinks to it, based on available protocols. Until now udev rules had to look at the fw_unit devices and then modify their parent's character device file accordingly. This is problematic for two reasons: 1) It happens sometime after the creation of the fw_device, 2) an access policy may require that information from all children is evaluated before a decision about the parent is made. Problem 1) can ultimately not be avoided since this is the nature of FireWire nodes: They may add or remove unit directories at any point in time. However, we can still help userland a lot by providing the protocol type information of all units in a summary sysfs attribute directly at the fw_device. This way, - the information is immediately available at the affected device when userspace goes about to handle an ADD or CHANGE event of the fw_device, - with most policies, it won't be necessary anymore to dig through child attributes. The new attribute is called "units". It contains space-separated tuples of specifier_id and version of each present unit. The delimiter within tuples is a colon. Specifier_id and version are printed as 0x%06x. Here is an example of a node which implements an IPv4 unit and an IPv6 unit: $ cat /sys/bus/firewire/devices/fw2/units 0x00005e:0x000001 0x00005e:0x000002 Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2009-05-22 22:03:29 +00:00
__ATTR_RO(units),
__ATTR_NULL,
};
static int read_rom(struct fw_device *device,
int generation, int index, u32 *data)
{
firewire: core: handle ack_busy when fetching the Config ROM Some older Panasonic made camcorders (Panasonic AG-EZ30 and NV-DX110, Grundig Scenos DLC 2000) reject requests with ack_busy_X if a request is sent immediately after they sent a response to a prior transaction. This causes firewire-core to fail probing of the camcorder with "giving up on config rom for node id ...". Consequently, programs like kino or dvgrab are unaware of the presence of a camcorder. Such transaction failures happen also with the ieee1394 driver stack (of the 2.4...2.6 kernel series until 2.6.36 inclusive) but with a lower likelihood, such that kino or dvgrab are generally able to use these camcorders via the older driver stack. The cause for firewire-ohci's or firewire-core's worse behavior is not yet known. Gap count optimization in firewire-core is not the cause. Perhaps the slightly higher latency of transaction completion in the older stack plays a role. (ieee1394: AR-resp DMA context tasklet -> packet completion ktread -> user process; firewire-core: tasklet -> user process.) This change introduces retries and delays after ack_busy_X into firewire-core's Config ROM reader, such that at least firewire-core's probing and /dev/fw* creation are successful. This still leaves the problem that userland processes are facing transaction failures. gscanbus's built-in retry routines deal with them successfully, but neither kino's nor dvgrab's do ever succeed. But at least DV capture with "dvgrab -noavc -card 0" works now. Live video preview in kino works too, but not actual capture. One way to prevent Configuration ROM reading failures in application programs is to modify libraw1394 to synthesize read responses by means of firewire-core's Configuration ROM cache. This would only leave CMP and FCP transaction failures as a potential problem source for applications. Reported-and-tested-by: Thomas Seilund <tps@netmaster.dk> Reported-and-tested-by: René Fritz <rene@colorcube.de> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2011-08-07 13:20:18 +00:00
u64 offset = (CSR_REGISTER_BASE | CSR_CONFIG_ROM) + index * 4;
int i, rcode;
/* device->node_id, accessed below, must not be older than generation */
smp_rmb();
firewire: core: handle ack_busy when fetching the Config ROM Some older Panasonic made camcorders (Panasonic AG-EZ30 and NV-DX110, Grundig Scenos DLC 2000) reject requests with ack_busy_X if a request is sent immediately after they sent a response to a prior transaction. This causes firewire-core to fail probing of the camcorder with "giving up on config rom for node id ...". Consequently, programs like kino or dvgrab are unaware of the presence of a camcorder. Such transaction failures happen also with the ieee1394 driver stack (of the 2.4...2.6 kernel series until 2.6.36 inclusive) but with a lower likelihood, such that kino or dvgrab are generally able to use these camcorders via the older driver stack. The cause for firewire-ohci's or firewire-core's worse behavior is not yet known. Gap count optimization in firewire-core is not the cause. Perhaps the slightly higher latency of transaction completion in the older stack plays a role. (ieee1394: AR-resp DMA context tasklet -> packet completion ktread -> user process; firewire-core: tasklet -> user process.) This change introduces retries and delays after ack_busy_X into firewire-core's Config ROM reader, such that at least firewire-core's probing and /dev/fw* creation are successful. This still leaves the problem that userland processes are facing transaction failures. gscanbus's built-in retry routines deal with them successfully, but neither kino's nor dvgrab's do ever succeed. But at least DV capture with "dvgrab -noavc -card 0" works now. Live video preview in kino works too, but not actual capture. One way to prevent Configuration ROM reading failures in application programs is to modify libraw1394 to synthesize read responses by means of firewire-core's Configuration ROM cache. This would only leave CMP and FCP transaction failures as a potential problem source for applications. Reported-and-tested-by: Thomas Seilund <tps@netmaster.dk> Reported-and-tested-by: René Fritz <rene@colorcube.de> Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2011-08-07 13:20:18 +00:00
for (i = 10; i < 100; i += 10) {
rcode = fw_run_transaction(device->card,
TCODE_READ_QUADLET_REQUEST, device->node_id,
generation, device->max_speed, offset, data, 4);
if (rcode != RCODE_BUSY)
break;
msleep(i);
}
be32_to_cpus(data);
return rcode;
}
#define MAX_CONFIG_ROM_SIZE 256
/*
* Read the bus info block, perform a speed probe, and read all of the rest of
* the config ROM. We do all this with a cached bus generation. If the bus
* generation changes under us, read_config_rom will fail and get retried.
* It's better to start all over in this case because the node from which we
* are reading the ROM may have changed the ROM during the reset.
* Returns either a result code or a negative error code.
*/
static int read_config_rom(struct fw_device *device, int generation)
{
struct fw_card *card = device->card;
const u32 *old_rom, *new_rom;
u32 *rom, *stack;
u32 sp, key;
int i, end, length, ret;
rom = kmalloc(sizeof(*rom) * MAX_CONFIG_ROM_SIZE +
sizeof(*stack) * MAX_CONFIG_ROM_SIZE, GFP_KERNEL);
if (rom == NULL)
return -ENOMEM;
stack = &rom[MAX_CONFIG_ROM_SIZE];
memset(rom, 0, sizeof(*rom) * MAX_CONFIG_ROM_SIZE);
device->max_speed = SCODE_100;
/* First read the bus info block. */
for (i = 0; i < 5; i++) {
ret = read_rom(device, generation, i, &rom[i]);
if (ret != RCODE_COMPLETE)
goto out;
/*
* As per IEEE1212 7.2, during initialization, devices can
* reply with a 0 for the first quadlet of the config
* rom to indicate that they are booting (for example,
* if the firmware is on the disk of a external
* harddisk). In that case we just fail, and the
* retry mechanism will try again later.
*/
if (i == 0 && rom[i] == 0) {
ret = RCODE_BUSY;
goto out;
}
}
device->max_speed = device->node->max_speed;
/*
* Determine the speed of
* - devices with link speed less than PHY speed,
* - devices with 1394b PHY (unless only connected to 1394a PHYs),
* - all devices if there are 1394b repeaters.
* Note, we cannot use the bus info block's link_spd as starting point
* because some buggy firmwares set it lower than necessary and because
* 1394-1995 nodes do not have the field.
*/
if ((rom[2] & 0x7) < device->max_speed ||
device->max_speed == SCODE_BETA ||
card->beta_repeaters_present) {
u32 dummy;
/* for S1600 and S3200 */
if (device->max_speed == SCODE_BETA)
device->max_speed = card->link_speed;
while (device->max_speed > SCODE_100) {
if (read_rom(device, generation, 0, &dummy) ==
RCODE_COMPLETE)
break;
device->max_speed--;
}
}
/*
* Now parse the config rom. The config rom is a recursive
* directory structure so we parse it using a stack of
* references to the blocks that make up the structure. We
* push a reference to the root directory on the stack to
* start things off.
*/
length = i;
sp = 0;
stack[sp++] = 0xc0000005;
while (sp > 0) {
/*
* Pop the next block reference of the stack. The
* lower 24 bits is the offset into the config rom,
* the upper 8 bits are the type of the reference the
* block.
*/
key = stack[--sp];
i = key & 0xffffff;
if (WARN_ON(i >= MAX_CONFIG_ROM_SIZE)) {
ret = -ENXIO;
goto out;
}
/* Read header quadlet for the block to get the length. */
ret = read_rom(device, generation, i, &rom[i]);
if (ret != RCODE_COMPLETE)
goto out;
end = i + (rom[i] >> 16) + 1;
if (end > MAX_CONFIG_ROM_SIZE) {
/*
* This block extends outside the config ROM which is
* a firmware bug. Ignore this whole block, i.e.
* simply set a fake block length of 0.
*/
fw_err(card, "skipped invalid ROM block %x at %llx\n",
rom[i],
i * 4 | CSR_REGISTER_BASE | CSR_CONFIG_ROM);
rom[i] = 0;
end = i;
}
i++;
/*
* Now read in the block. If this is a directory
* block, check the entries as we read them to see if
* it references another block, and push it in that case.
*/
for (; i < end; i++) {
ret = read_rom(device, generation, i, &rom[i]);
if (ret != RCODE_COMPLETE)
goto out;
if ((key >> 30) != 3 || (rom[i] >> 30) < 2)
continue;
/*
* Offset points outside the ROM. May be a firmware
* bug or an Extended ROM entry (IEEE 1212-2001 clause
* 7.7.18). Simply overwrite this pointer here by a
* fake immediate entry so that later iterators over
* the ROM don't have to check offsets all the time.
*/
if (i + (rom[i] & 0xffffff) >= MAX_CONFIG_ROM_SIZE) {
fw_err(card,
"skipped unsupported ROM entry %x at %llx\n",
rom[i],
i * 4 | CSR_REGISTER_BASE | CSR_CONFIG_ROM);
rom[i] = 0;
continue;
}
stack[sp++] = i + rom[i];
}
if (length < i)
length = i;
}
old_rom = device->config_rom;
new_rom = kmemdup(rom, length * 4, GFP_KERNEL);
if (new_rom == NULL) {
ret = -ENOMEM;
goto out;
}
down_write(&fw_device_rwsem);
device->config_rom = new_rom;
device->config_rom_length = length;
up_write(&fw_device_rwsem);
kfree(old_rom);
ret = RCODE_COMPLETE;
device->max_rec = rom[2] >> 12 & 0xf;
device->cmc = rom[2] >> 30 & 1;
device->irmc = rom[2] >> 31 & 1;
out:
kfree(rom);
return ret;
}
static void fw_unit_release(struct device *dev)
{
struct fw_unit *unit = fw_unit(dev);
fw_device_put(fw_parent_device(unit));
kfree(unit);
}
static struct device_type fw_unit_type = {
.uevent = fw_unit_uevent,
.release = fw_unit_release,
};
static bool is_fw_unit(struct device *dev)
{
return dev->type == &fw_unit_type;
}
static void create_units(struct fw_device *device)
{
struct fw_csr_iterator ci;
struct fw_unit *unit;
int key, value, i;
i = 0;
fw_csr_iterator_init(&ci, &device->config_rom[5]);
while (fw_csr_iterator_next(&ci, &key, &value)) {
if (key != (CSR_UNIT | CSR_DIRECTORY))
continue;
/*
* Get the address of the unit directory and try to
* match the drivers id_tables against it.
*/
unit = kzalloc(sizeof(*unit), GFP_KERNEL);
if (unit == NULL)
continue;
unit->directory = ci.p + value - 1;
unit->device.bus = &fw_bus_type;
unit->device.type = &fw_unit_type;
unit->device.parent = &device->device;
dev_set_name(&unit->device, "%s.%d", dev_name(&device->device), i++);
BUILD_BUG_ON(ARRAY_SIZE(unit->attribute_group.attrs) <
ARRAY_SIZE(fw_unit_attributes) +
ARRAY_SIZE(config_rom_attributes));
init_fw_attribute_group(&unit->device,
fw_unit_attributes,
&unit->attribute_group);
if (device_register(&unit->device) < 0)
goto skip_unit;
fw_device_get(device);
continue;
skip_unit:
kfree(unit);
}
}
static int shutdown_unit(struct device *device, void *data)
{
device_unregister(device);
return 0;
}
/*
* fw_device_rwsem acts as dual purpose mutex:
* - serializes accesses to fw_device_idr,
* - serializes accesses to fw_device.config_rom/.config_rom_length and
* fw_unit.directory, unless those accesses happen at safe occasions
*/
DECLARE_RWSEM(fw_device_rwsem);
DEFINE_IDR(fw_device_idr);
int fw_cdev_major;
struct fw_device *fw_device_get_by_devt(dev_t devt)
{
struct fw_device *device;
down_read(&fw_device_rwsem);
device = idr_find(&fw_device_idr, MINOR(devt));
if (device)
fw_device_get(device);
up_read(&fw_device_rwsem);
return device;
}
struct workqueue_struct *fw_workqueue;
EXPORT_SYMBOL(fw_workqueue);
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
static void fw_schedule_device_work(struct fw_device *device,
unsigned long delay)
{
queue_delayed_work(fw_workqueue, &device->work, delay);
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
}
/*
* These defines control the retry behavior for reading the config
* rom. It shouldn't be necessary to tweak these; if the device
* doesn't respond to a config rom read within 10 seconds, it's not
* going to respond at all. As for the initial delay, a lot of
* devices will be able to respond within half a second after bus
* reset. On the other hand, it's not really worth being more
* aggressive than that, since it scales pretty well; if 10 devices
* are plugged in, they're all getting read within one second.
*/
#define MAX_RETRIES 10
#define RETRY_DELAY (3 * HZ)
#define INITIAL_DELAY (HZ / 2)
#define SHUTDOWN_DELAY (2 * HZ)
static void fw_device_shutdown(struct work_struct *work)
{
struct fw_device *device =
container_of(work, struct fw_device, work.work);
int minor = MINOR(device->device.devt);
if (time_before64(get_jiffies_64(),
device->card->reset_jiffies + SHUTDOWN_DELAY)
&& !list_empty(&device->card->link)) {
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, SHUTDOWN_DELAY);
return;
}
if (atomic_cmpxchg(&device->state,
FW_DEVICE_GONE,
FW_DEVICE_SHUTDOWN) != FW_DEVICE_GONE)
return;
fw_device_cdev_remove(device);
device_for_each_child(&device->device, NULL, shutdown_unit);
device_unregister(&device->device);
down_write(&fw_device_rwsem);
idr_remove(&fw_device_idr, minor);
up_write(&fw_device_rwsem);
fw_device_put(device);
}
static void fw_device_release(struct device *dev)
{
struct fw_device *device = fw_device(dev);
struct fw_card *card = device->card;
unsigned long flags;
/*
* Take the card lock so we don't set this to NULL while a
* FW_NODE_UPDATED callback is being handled or while the
* bus manager work looks at this node.
*/
spin_lock_irqsave(&card->lock, flags);
device->node->data = NULL;
spin_unlock_irqrestore(&card->lock, flags);
fw_node_put(device->node);
kfree(device->config_rom);
kfree(device);
fw_card_put(card);
}
static struct device_type fw_device_type = {
.release = fw_device_release,
};
static bool is_fw_device(struct device *dev)
{
return dev->type == &fw_device_type;
}
static int update_unit(struct device *dev, void *data)
{
struct fw_unit *unit = fw_unit(dev);
struct fw_driver *driver = (struct fw_driver *)dev->driver;
if (is_fw_unit(dev) && driver != NULL && driver->update != NULL) {
Driver core: create lock/unlock functions for struct device In the future, we are going to be changing the lock type for struct device (once we get the lockdep infrastructure properly worked out) To make that changeover easier, and to possibly burry the lock in a different part of struct device, let's create some functions to lock and unlock a device so that no out-of-core code needs to be changed in the future. This patch creates the device_lock/unlock/trylock() functions, and converts all in-tree users to them. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jean Delvare <khali@linux-fr.org> Cc: Dave Young <hidave.darkstar@gmail.com> Cc: Ming Lei <tom.leiming@gmail.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Phil Carmody <ext-phil.2.carmody@nokia.com> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: Cornelia Huck <cornelia.huck@de.ibm.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Cc: Pavel Machek <pavel@ucw.cz> Cc: Len Brown <len.brown@intel.com> Cc: Magnus Damm <damm@igel.co.jp> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: Randy Dunlap <randy.dunlap@oracle.com> Cc: Stefan Richter <stefanr@s5r6.in-berlin.de> Cc: David Brownell <dbrownell@users.sourceforge.net> Cc: Vegard Nossum <vegard.nossum@gmail.com> Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Cc: Alex Chiang <achiang@hp.com> Cc: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andrew Patterson <andrew.patterson@hp.com> Cc: Yu Zhao <yu.zhao@intel.com> Cc: Dominik Brodowski <linux@dominikbrodowski.net> Cc: Samuel Ortiz <sameo@linux.intel.com> Cc: Wolfram Sang <w.sang@pengutronix.de> Cc: CHENG Renquan <rqcheng@smu.edu.sg> Cc: Oliver Neukum <oliver@neukum.org> Cc: Frans Pop <elendil@planet.nl> Cc: David Vrabel <david.vrabel@csr.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Cc: Sarah Sharp <sarah.a.sharp@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-17 18:57:05 +00:00
device_lock(dev);
driver->update(unit);
Driver core: create lock/unlock functions for struct device In the future, we are going to be changing the lock type for struct device (once we get the lockdep infrastructure properly worked out) To make that changeover easier, and to possibly burry the lock in a different part of struct device, let's create some functions to lock and unlock a device so that no out-of-core code needs to be changed in the future. This patch creates the device_lock/unlock/trylock() functions, and converts all in-tree users to them. Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Jean Delvare <khali@linux-fr.org> Cc: Dave Young <hidave.darkstar@gmail.com> Cc: Ming Lei <tom.leiming@gmail.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Phil Carmody <ext-phil.2.carmody@nokia.com> Cc: Arjan van de Ven <arjan@linux.intel.com> Cc: Cornelia Huck <cornelia.huck@de.ibm.com> Cc: Rafael J. Wysocki <rjw@sisk.pl> Cc: Pavel Machek <pavel@ucw.cz> Cc: Len Brown <len.brown@intel.com> Cc: Magnus Damm <damm@igel.co.jp> Cc: Alan Stern <stern@rowland.harvard.edu> Cc: Randy Dunlap <randy.dunlap@oracle.com> Cc: Stefan Richter <stefanr@s5r6.in-berlin.de> Cc: David Brownell <dbrownell@users.sourceforge.net> Cc: Vegard Nossum <vegard.nossum@gmail.com> Cc: Jesse Barnes <jbarnes@virtuousgeek.org> Cc: Alex Chiang <achiang@hp.com> Cc: Kenji Kaneshige <kaneshige.kenji@jp.fujitsu.com> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Andrew Patterson <andrew.patterson@hp.com> Cc: Yu Zhao <yu.zhao@intel.com> Cc: Dominik Brodowski <linux@dominikbrodowski.net> Cc: Samuel Ortiz <sameo@linux.intel.com> Cc: Wolfram Sang <w.sang@pengutronix.de> Cc: CHENG Renquan <rqcheng@smu.edu.sg> Cc: Oliver Neukum <oliver@neukum.org> Cc: Frans Pop <elendil@planet.nl> Cc: David Vrabel <david.vrabel@csr.com> Cc: Kay Sievers <kay.sievers@vrfy.org> Cc: Sarah Sharp <sarah.a.sharp@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2010-02-17 18:57:05 +00:00
device_unlock(dev);
}
return 0;
}
static void fw_device_update(struct work_struct *work)
{
struct fw_device *device =
container_of(work, struct fw_device, work.work);
fw_device_cdev_update(device);
device_for_each_child(&device->device, NULL, update_unit);
}
/*
* If a device was pending for deletion because its node went away but its
* bus info block and root directory header matches that of a newly discovered
* device, revive the existing fw_device.
* The newly allocated fw_device becomes obsolete instead.
*/
static int lookup_existing_device(struct device *dev, void *data)
{
struct fw_device *old = fw_device(dev);
struct fw_device *new = data;
struct fw_card *card = new->card;
int match = 0;
if (!is_fw_device(dev))
return 0;
down_read(&fw_device_rwsem); /* serialize config_rom access */
spin_lock_irq(&card->lock); /* serialize node access */
if (memcmp(old->config_rom, new->config_rom, 6 * 4) == 0 &&
atomic_cmpxchg(&old->state,
FW_DEVICE_GONE,
FW_DEVICE_RUNNING) == FW_DEVICE_GONE) {
struct fw_node *current_node = new->node;
struct fw_node *obsolete_node = old->node;
new->node = obsolete_node;
new->node->data = new;
old->node = current_node;
old->node->data = old;
old->max_speed = new->max_speed;
old->node_id = current_node->node_id;
smp_wmb(); /* update node_id before generation */
old->generation = card->generation;
old->config_rom_retries = 0;
fw_notice(card, "rediscovered device %s\n", dev_name(dev));
old->workfn = fw_device_update;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(old, 0);
if (current_node == card->root_node)
fw_schedule_bm_work(card, 0);
match = 1;
}
spin_unlock_irq(&card->lock);
up_read(&fw_device_rwsem);
return match;
}
enum { BC_UNKNOWN = 0, BC_UNIMPLEMENTED, BC_IMPLEMENTED, };
static void set_broadcast_channel(struct fw_device *device, int generation)
{
struct fw_card *card = device->card;
__be32 data;
int rcode;
if (!card->broadcast_channel_allocated)
return;
/*
* The Broadcast_Channel Valid bit is required by nodes which want to
* transmit on this channel. Such transmissions are practically
* exclusive to IP over 1394 (RFC 2734). IP capable nodes are required
* to be IRM capable and have a max_rec of 8 or more. We use this fact
* to narrow down to which nodes we send Broadcast_Channel updates.
*/
if (!device->irmc || device->max_rec < 8)
return;
/*
* Some 1394-1995 nodes crash if this 1394a-2000 register is written.
* Perform a read test first.
*/
if (device->bc_implemented == BC_UNKNOWN) {
rcode = fw_run_transaction(card, TCODE_READ_QUADLET_REQUEST,
device->node_id, generation, device->max_speed,
CSR_REGISTER_BASE + CSR_BROADCAST_CHANNEL,
&data, 4);
switch (rcode) {
case RCODE_COMPLETE:
if (data & cpu_to_be32(1 << 31)) {
device->bc_implemented = BC_IMPLEMENTED;
break;
}
/* else fall through to case address error */
case RCODE_ADDRESS_ERROR:
device->bc_implemented = BC_UNIMPLEMENTED;
}
}
if (device->bc_implemented == BC_IMPLEMENTED) {
data = cpu_to_be32(BROADCAST_CHANNEL_INITIAL |
BROADCAST_CHANNEL_VALID);
fw_run_transaction(card, TCODE_WRITE_QUADLET_REQUEST,
device->node_id, generation, device->max_speed,
CSR_REGISTER_BASE + CSR_BROADCAST_CHANNEL,
&data, 4);
}
}
int fw_device_set_broadcast_channel(struct device *dev, void *gen)
{
if (is_fw_device(dev))
set_broadcast_channel(fw_device(dev), (long)gen);
return 0;
}
static void fw_device_init(struct work_struct *work)
{
struct fw_device *device =
container_of(work, struct fw_device, work.work);
struct fw_card *card = device->card;
struct device *revived_dev;
int minor, ret;
/*
* All failure paths here set node->data to NULL, so that we
* don't try to do device_for_each_child() on a kfree()'d
* device.
*/
ret = read_config_rom(device, device->generation);
if (ret != RCODE_COMPLETE) {
firewire: fix crash in automatic module unloading "modprobe firewire-ohci; sleep .1; modprobe -r firewire-ohci" used to result in crashes like this: BUG: unable to handle kernel paging request at ffffffff8807b455 IP: [<ffffffff8807b455>] PGD 203067 PUD 207063 PMD 7c170067 PTE 0 Oops: 0010 [1] PREEMPT SMP CPU 0 Modules linked in: i915 drm cpufreq_ondemand acpi_cpufreq freq_table applesmc input_polldev led_class coretemp hwmon eeprom snd_seq_oss snd_seq_midi_event snd_seq snd_seq_device snd_pcm_oss snd_mixer_oss button thermal processor sg snd_hda_intel snd_pcm snd_timer snd snd_page_alloc sky2 i2c_i801 rtc [last unloaded: crc_itu_t] Pid: 9, comm: events/0 Not tainted 2.6.25-rc2 #3 RIP: 0010:[<ffffffff8807b455>] [<ffffffff8807b455>] RSP: 0018:ffff81007dcdde88 EFLAGS: 00010246 RAX: ffff81007dc95040 RBX: ffff81007dee5390 RCX: 0000000000005e13 RDX: 0000000000008c8b RSI: 0000000000000001 RDI: ffff81007dee5388 RBP: ffff81007dc5eb40 R08: 0000000000000002 R09: ffffffff8022d05c R10: ffffffff8023b34c R11: ffffffff8041a353 R12: ffff81007dee5388 R13: ffffffff8807b455 R14: ffffffff80593bc0 R15: 0000000000000000 FS: 0000000000000000(0000) GS:ffffffff8055a000(0000) knlGS:0000000000000000 CS: 0010 DS: 0018 ES: 0018 CR0: 000000008005003b CR2: ffffffff8807b455 CR3: 0000000000201000 CR4: 00000000000006e0 DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400 Process events/0 (pid: 9, threadinfo ffff81007dcdc000, task ffff81007dc95040) Stack: ffffffff8023b396 ffffffff88082524 0000000000000000 ffffffff8807d9ae ffff81007dc5eb40 ffff81007dc9dce0 ffff81007dc5eb40 ffff81007dc5eb80 ffff81007dc9dce0 ffffffffffffffff ffffffff8023be87 0000000000000000 Call Trace: [<ffffffff8023b396>] ? run_workqueue+0xdf/0x1df [<ffffffff8023be87>] ? worker_thread+0xd8/0xe3 [<ffffffff8023e917>] ? autoremove_wake_function+0x0/0x2e [<ffffffff8023bdaf>] ? worker_thread+0x0/0xe3 [<ffffffff8023e813>] ? kthread+0x47/0x74 [<ffffffff804198e0>] ? trace_hardirqs_on_thunk+0x35/0x3a [<ffffffff8020c008>] ? child_rip+0xa/0x12 [<ffffffff8020b6e3>] ? restore_args+0x0/0x3d [<ffffffff8023e68a>] ? kthreadd+0x14c/0x171 [<ffffffff8023e68a>] ? kthreadd+0x14c/0x171 [<ffffffff8023e7cc>] ? kthread+0x0/0x74 [<ffffffff8020bffe>] ? child_rip+0x0/0x12 Code: Bad RIP value. RIP [<ffffffff8807b455>] RSP <ffff81007dcdde88> CR2: ffffffff8807b455 ---[ end trace c7366c6657fe5bed ]--- Note that this crash happened _after_ firewire-core was unloaded. The shared workqueue tried to run firewire-core's device initialization jobs or similar jobs. The fix makes sure that firewire-ohci and hence firewire-core is not unloaded before all device shutdown jobs have been completed. This is determined by the count of device initializations minus device releases. Also skip useless retries in the node initialization job if the node is to be shut down. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Signed-off-by: Jarod Wilson <jwilson@redhat.com>
2008-02-27 21:14:27 +00:00
if (device->config_rom_retries < MAX_RETRIES &&
atomic_read(&device->state) == FW_DEVICE_INITIALIZING) {
device->config_rom_retries++;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, RETRY_DELAY);
} else {
if (device->node->link_on)
fw_notice(card, "giving up on node %x: reading config rom failed: %s\n",
device->node_id,
fw_rcode_string(ret));
if (device->node == card->root_node)
fw_schedule_bm_work(card, 0);
fw_device_release(&device->device);
}
return;
}
revived_dev = device_find_child(card->device,
device, lookup_existing_device);
if (revived_dev) {
put_device(revived_dev);
fw_device_release(&device->device);
return;
}
device_initialize(&device->device);
fw_device_get(device);
down_write(&fw_device_rwsem);
minor = idr_alloc(&fw_device_idr, device, 0, 1 << MINORBITS,
GFP_KERNEL);
up_write(&fw_device_rwsem);
if (minor < 0)
goto error;
device->device.bus = &fw_bus_type;
device->device.type = &fw_device_type;
device->device.parent = card->device;
device->device.devt = MKDEV(fw_cdev_major, minor);
dev_set_name(&device->device, "fw%d", minor);
BUILD_BUG_ON(ARRAY_SIZE(device->attribute_group.attrs) <
ARRAY_SIZE(fw_device_attributes) +
ARRAY_SIZE(config_rom_attributes));
init_fw_attribute_group(&device->device,
fw_device_attributes,
&device->attribute_group);
if (device_add(&device->device)) {
fw_err(card, "failed to add device\n");
goto error_with_cdev;
}
create_units(device);
/*
* Transition the device to running state. If it got pulled
* out from under us while we did the intialization work, we
* have to shut down the device again here. Normally, though,
* fw_node_event will be responsible for shutting it down when
* necessary. We have to use the atomic cmpxchg here to avoid
* racing with the FW_NODE_DESTROYED case in
* fw_node_event().
*/
if (atomic_cmpxchg(&device->state,
FW_DEVICE_INITIALIZING,
FW_DEVICE_RUNNING) == FW_DEVICE_GONE) {
device->workfn = fw_device_shutdown;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, SHUTDOWN_DELAY);
} else {
fw_notice(card, "created device %s: GUID %08x%08x, S%d00\n",
dev_name(&device->device),
device->config_rom[3], device->config_rom[4],
1 << device->max_speed);
device->config_rom_retries = 0;
set_broadcast_channel(device, device->generation);
add_device_randomness(&device->config_rom[3], 8);
}
/*
* Reschedule the IRM work if we just finished reading the
* root node config rom. If this races with a bus reset we
* just end up running the IRM work a couple of extra times -
* pretty harmless.
*/
if (device->node == card->root_node)
fw_schedule_bm_work(card, 0);
return;
error_with_cdev:
down_write(&fw_device_rwsem);
idr_remove(&fw_device_idr, minor);
up_write(&fw_device_rwsem);
error:
fw_device_put(device); /* fw_device_idr's reference */
put_device(&device->device); /* our reference */
}
/* Reread and compare bus info block and header of root directory */
static int reread_config_rom(struct fw_device *device, int generation,
bool *changed)
{
u32 q;
int i, rcode;
for (i = 0; i < 6; i++) {
rcode = read_rom(device, generation, i, &q);
if (rcode != RCODE_COMPLETE)
return rcode;
if (i == 0 && q == 0)
/* inaccessible (see read_config_rom); retry later */
return RCODE_BUSY;
if (q != device->config_rom[i]) {
*changed = true;
return RCODE_COMPLETE;
}
}
*changed = false;
return RCODE_COMPLETE;
}
static void fw_device_refresh(struct work_struct *work)
{
struct fw_device *device =
container_of(work, struct fw_device, work.work);
struct fw_card *card = device->card;
int ret, node_id = device->node_id;
bool changed;
ret = reread_config_rom(device, device->generation, &changed);
if (ret != RCODE_COMPLETE)
goto failed_config_rom;
if (!changed) {
if (atomic_cmpxchg(&device->state,
FW_DEVICE_INITIALIZING,
FW_DEVICE_RUNNING) == FW_DEVICE_GONE)
goto gone;
fw_device_update(work);
device->config_rom_retries = 0;
goto out;
}
/*
* Something changed. We keep things simple and don't investigate
* further. We just destroy all previous units and create new ones.
*/
device_for_each_child(&device->device, NULL, shutdown_unit);
ret = read_config_rom(device, device->generation);
if (ret != RCODE_COMPLETE)
goto failed_config_rom;
fw_device_cdev_update(device);
create_units(device);
firewire: core: add sysfs attribute for easier udev rules This adds the attribute /sys/bus/firewire/devices/fw[0-9]+/units. It can be used in udev rules like the following ones: # IIDC devices: industrial cameras and some webcams SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x00010?*", GROUP="video" # AV/C devices: camcorders, set-top boxes, TV sets, audio devices, ... SUBSYSTEM=="firewire", ATTR{units}=="*0x00a02d:0x010001*", GROUP="video" Background: firewire-core manages two device types: - fw_device is a FireWire node. A character device file is associated with it. - fw_unit is a unit directory on a node. Each fw_device may have 0..n children of type fw_unit. The units tell us what kinds of protocols a node implements. We want to set ownership or ACLs or permissions of the character device file of an fw_device, or/and create symlinks to it, based on available protocols. Until now udev rules had to look at the fw_unit devices and then modify their parent's character device file accordingly. This is problematic for two reasons: 1) It happens sometime after the creation of the fw_device, 2) an access policy may require that information from all children is evaluated before a decision about the parent is made. Problem 1) can ultimately not be avoided since this is the nature of FireWire nodes: They may add or remove unit directories at any point in time. However, we can still help userland a lot by providing the protocol type information of all units in a summary sysfs attribute directly at the fw_device. This way, - the information is immediately available at the affected device when userspace goes about to handle an ADD or CHANGE event of the fw_device, - with most policies, it won't be necessary anymore to dig through child attributes. The new attribute is called "units". It contains space-separated tuples of specifier_id and version of each present unit. The delimiter within tuples is a colon. Specifier_id and version are printed as 0x%06x. Here is an example of a node which implements an IPv4 unit and an IPv6 unit: $ cat /sys/bus/firewire/devices/fw2/units 0x00005e:0x000001 0x00005e:0x000002 Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
2009-05-22 22:03:29 +00:00
/* Userspace may want to re-read attributes. */
kobject_uevent(&device->device.kobj, KOBJ_CHANGE);
if (atomic_cmpxchg(&device->state,
FW_DEVICE_INITIALIZING,
FW_DEVICE_RUNNING) == FW_DEVICE_GONE)
goto gone;
fw_notice(card, "refreshed device %s\n", dev_name(&device->device));
device->config_rom_retries = 0;
goto out;
failed_config_rom:
if (device->config_rom_retries < MAX_RETRIES &&
atomic_read(&device->state) == FW_DEVICE_INITIALIZING) {
device->config_rom_retries++;
fw_schedule_device_work(device, RETRY_DELAY);
return;
}
fw_notice(card, "giving up on refresh of device %s: %s\n",
dev_name(&device->device), fw_rcode_string(ret));
gone:
atomic_set(&device->state, FW_DEVICE_GONE);
device->workfn = fw_device_shutdown;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, SHUTDOWN_DELAY);
out:
if (node_id == card->root_node->node_id)
fw_schedule_bm_work(card, 0);
}
static void fw_device_workfn(struct work_struct *work)
{
struct fw_device *device = container_of(to_delayed_work(work),
struct fw_device, work);
device->workfn(work);
}
void fw_node_event(struct fw_card *card, struct fw_node *node, int event)
{
struct fw_device *device;
switch (event) {
case FW_NODE_CREATED:
/*
* Attempt to scan the node, regardless whether its self ID has
* the L (link active) flag set or not. Some broken devices
* send L=0 but have an up-and-running link; others send L=1
* without actually having a link.
*/
create:
device = kzalloc(sizeof(*device), GFP_ATOMIC);
if (device == NULL)
break;
/*
* Do minimal intialization of the device here, the
* rest will happen in fw_device_init().
*
* Attention: A lot of things, even fw_device_get(),
* cannot be done before fw_device_init() finished!
* You can basically just check device->state and
* schedule work until then, but only while holding
* card->lock.
*/
atomic_set(&device->state, FW_DEVICE_INITIALIZING);
device->card = fw_card_get(card);
device->node = fw_node_get(node);
device->node_id = node->node_id;
device->generation = card->generation;
device->is_local = node == card->local_node;
mutex_init(&device->client_list_mutex);
INIT_LIST_HEAD(&device->client_list);
/*
* Set the node data to point back to this device so
* FW_NODE_UPDATED callbacks can update the node_id
* and generation for the device.
*/
node->data = device;
/*
* Many devices are slow to respond after bus resets,
* especially if they are bus powered and go through
* power-up after getting plugged in. We schedule the
* first config rom scan half a second after bus reset.
*/
device->workfn = fw_device_init;
INIT_DELAYED_WORK(&device->work, fw_device_workfn);
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, INITIAL_DELAY);
break;
case FW_NODE_INITIATED_RESET:
case FW_NODE_LINK_ON:
device = node->data;
if (device == NULL)
goto create;
device->node_id = node->node_id;
smp_wmb(); /* update node_id before generation */
device->generation = card->generation;
if (atomic_cmpxchg(&device->state,
FW_DEVICE_RUNNING,
FW_DEVICE_INITIALIZING) == FW_DEVICE_RUNNING) {
device->workfn = fw_device_refresh;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device,
device->is_local ? 0 : INITIAL_DELAY);
}
break;
case FW_NODE_UPDATED:
device = node->data;
if (device == NULL)
break;
device->node_id = node->node_id;
smp_wmb(); /* update node_id before generation */
device->generation = card->generation;
if (atomic_read(&device->state) == FW_DEVICE_RUNNING) {
device->workfn = fw_device_update;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device, 0);
}
break;
case FW_NODE_DESTROYED:
case FW_NODE_LINK_OFF:
if (!node->data)
break;
/*
* Destroy the device associated with the node. There
* are two cases here: either the device is fully
* initialized (FW_DEVICE_RUNNING) or we're in the
* process of reading its config rom
* (FW_DEVICE_INITIALIZING). If it is fully
* initialized we can reuse device->work to schedule a
* full fw_device_shutdown(). If not, there's work
* scheduled to read it's config rom, and we just put
* the device in shutdown state to have that code fail
* to create the device.
*/
device = node->data;
if (atomic_xchg(&device->state,
FW_DEVICE_GONE) == FW_DEVICE_RUNNING) {
device->workfn = fw_device_shutdown;
firewire: core: use non-reentrant workqueue with rescuer firewire-core manages the following types of work items: fw_card.br_work: - resets the bus on a card and possibly sends a PHY packet before that - does not sleep for long or not at all - is scheduled via fw_schedule_bus_reset() by - firewire-ohci's pci_probe method - firewire-ohci's set_config_rom method, called by kernelspace protocol drivers and userspace drivers which add/remove Configuration ROM descriptors - userspace drivers which use the bus reset ioctl - itself if the last reset happened less than 2 seconds ago fw_card.bm_work: - performs bus management duties - usually does not (but may in corner cases) sleep for long - is scheduled via fw_schedule_bm_work() by - firewire-ohci's self-ID-complete IRQ handler tasklet - firewire-core's fw_device.work instances whenever the root node device was (successfully or unsuccessfully) discovered, refreshed, or rediscovered - itself in case of resource allocation failures or in order to obey the 125ms bus manager arbitration interval fw_device.work: - performs node probe, update, shutdown, revival, removal; including kernel driver probe, update, shutdown and bus reset notification to userspace drivers - usually sleeps moderately long, in corner cases very long - is scheduled by - firewire-ohci's self-ID-complete IRQ handler tasklet via the core's fw_node_event - firewire-ohci's pci_remove method via core's fw_destroy_nodes/ fw_node_event - itself during retries, e.g. while a node is powering up iso_resource.work: - accesses registers at the Isochronous Resource Manager node - usually does not (but may in corner cases) sleep for long - is scheduled via schedule_iso_resource() by - the owning userspace driver at addition and removal of the resource - firewire-core's fw_device.work instances after bus reset - itself in case of resource allocation if necessary to obey the 1000ms reallocation period after bus reset fw_card.br_work instances should not, and instances of the others must not, be executed in parallel by multiple CPUs -- but were not protected against that. Hence allocate a non-reentrant workqueue for them. fw_device.work may be used in the memory reclaim path in case of SBP-2 device updates. Hence we need a workqueue with rescuer and cannot use system_nrt_wq. Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de> Reviewed-by: Tejun Heo <tj@kernel.org>
2010-10-13 11:39:46 +00:00
fw_schedule_device_work(device,
list_empty(&card->link) ? 0 : SHUTDOWN_DELAY);
}
break;
}
}