Aren/NetBSD
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5ae2aa4a54
NetBSD/lib/libperfuse/perfuse.c

739 lines
19 KiB
C
Raw Normal View History

fix typo
2019-04-17 15:30:51 +03:00
/* $NetBSD: perfuse.c,v 1.42 2019/04/17 12:30:51 maya Exp $ */
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
/*-
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
* Copyright (c) 2010-2011 Emmanuel Dreyfus. All rights reserved.
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>
#include <fcntl.h>
#include <string.h>
#include <errno.h>
#include <puffs.h>
make this compile again, and simplify.
2016-10-18 20:56:31 +03:00
#include <inttypes.h>
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#include <sys/types.h>
Make sure perfused remains locked in memory, otherwise we can get deadlocks in low memory situations, where ioflush waits for perfused to fsync vnodes, and perfused waits for memory to be freed.
2011-09-09 19:35:22 +04:00
#include <sys/mman.h>
perfuse memory usage can grow quite large when using a lot of vnodes, and the amount of data memory involved is not easy to forcast. We therefore raise the limit to the maximum. Patch from Manuel Bouyer. It helps completing a cvs update on a glusterfs colume.
2011-10-23 09:01:00 +04:00
#include <sys/resource.h>
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#include <sys/socket.h>
Add support for extended attributes
2011-06-28 20:19:16 +04:00
#include <sys/extattr.h>
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
#include <sys/hash.h>
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#include <sys/un.h>
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
#include <machine/vmparam.h>
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#define LIBPERFUSE
#include "perfuse.h"
#include "perfuse_if.h"
#include "perfuse_priv.h"
int perfuse_diagflags = 0; /* global used only in DPRINTF/DERR/DWARN */
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
extern char **environ;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
static struct perfuse_state *init_state(void);
static int get_fd(const char *);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
static struct perfuse_state *
init_state(void)
{
struct perfuse_state *ps;
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
size_t len;
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
char opts[1024];
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
int i;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
if ((ps = malloc(sizeof(*ps))) == NULL)
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
DERR(EX_OSERR, "%s:%d malloc failed", __func__, __LINE__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
(void)memset(ps, 0, sizeof(*ps));
ps->ps_max_write = UINT_MAX;
ps->ps_max_readahead = UINT_MAX;
Add a FUSE trace facility, with statistics collection. This should help tracking bugs and performance issues
2011-12-28 21:33:52 +04:00
TAILQ_INIT(&ps->ps_trace);
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
ps->ps_nnidhash = PUFFS_PNODEBUCKETS;
len = sizeof(*ps->ps_nidhash) * ps->ps_nnidhash;
if ((ps->ps_nidhash = malloc(len)) == NULL)
DERR(EX_OSERR, "%s:%d malloc failed", __func__, __LINE__);
for (i = 0; i < ps->ps_nnidhash; i++)
LIST_INIT(&ps->ps_nidhash[i]);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
/*
* Most of the time, access() is broken because the filesystem
* performs the check with root privileges. glusterfs will do that
* if the Linux-specific setfsuid() is missing, for instance.
*/
ps->ps_flags |= PS_NO_ACCESS;
/*
* This is a temporary way to toggle access and creat usage.
* It would be nice if that could be provided as mount options,
* but that will not be obvious to do.
*/
if (getenv_r("PERFUSE_OPTIONS", opts, sizeof(opts)) != -1) {
char *optname;
char *last;
for ((optname = strtok_r(opts, ",", &last));
optname != NULL;
(optname = strtok_r(NULL, ",", &last))) {
if (strcmp(optname, "enable_access") == 0)
ps->ps_flags &= ~PS_NO_ACCESS;
if (strcmp(optname, "disable_access") == 0)
ps->ps_flags |= PS_NO_ACCESS;
if (strcmp(optname, "enable_creat") == 0)
ps->ps_flags &= ~PS_NO_CREAT;
if (strcmp(optname, "disable_creat") == 0)
ps->ps_flags |= PS_NO_CREAT;
}
}
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
return ps;
}
static int
Use C89 function definition
2012-03-21 14:10:36 +04:00
get_fd(const char *data)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
char *string;
const char fdopt[] = "fd=";
char *lastp;
char *opt;
int fd = -1;
if ((string = strdup(data)) == NULL)
return -1;
for (opt = strtok_r(string, ",", &lastp);
opt != NULL;
opt = strtok_r(NULL, ",", &lastp)) {
if (strncmp(opt, fdopt, strlen(fdopt)) == 0) {
fd = atoi(opt + strlen(fdopt));
break;
}
}
/*
* No file descriptor found
*/
if (fd == -1)
errno = EINVAL;
free(string);
return fd;
}
make the env stuff visible.
2016-10-19 04:30:35 +03:00
uint32_t
perfuse_bufvar_from_env(const char *name, uint32_t defval)
Make FUSE socket buffer tunable When dealing with high I/O throughput, we could run out of buffer space if the filesystem was not consuming requests fast enough. Here we slightly raise the buffer size, and we make it tunable through the PERFUSE_BUFSIZE environment variable so that we can cope with higher requirement later. While there, document PERFUSE_OPTIONS environment variable.
2016-10-18 18:06:17 +03:00
{
char valstr[1024];
make this compile again, and simplify.
2016-10-18 20:56:31 +03:00
int e;
uint32_t retval;
if (getenv_r(name, valstr, sizeof(valstr)) == -1)
return defval;
retval = (uint32_t)strtoi(valstr, NULL, 0, 0, UINT32_MAX, &e);
if (!e)
return retval;
Make FUSE socket buffer tunable When dealing with high I/O throughput, we could run out of buffer space if the filesystem was not consuming requests fast enough. Here we slightly raise the buffer size, and we make it tunable through the PERFUSE_BUFSIZE environment variable so that we can cope with higher requirement later. While there, document PERFUSE_OPTIONS environment variable.
2016-10-18 18:06:17 +03:00
make this compile again, and simplify.
2016-10-18 20:56:31 +03:00
DWARNC(e, "conversion from `%s' to uint32_t failed, using %u",
valstr, defval);
return defval;
Make FUSE socket buffer tunable When dealing with high I/O throughput, we could run out of buffer space if the filesystem was not consuming requests fast enough. Here we slightly raise the buffer size, and we make it tunable through the PERFUSE_BUFSIZE environment variable so that we can cope with higher requirement later. While there, document PERFUSE_OPTIONS environment variable.
2016-10-18 18:06:17 +03:00
}
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
int
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_open(const char *path, int flags, mode_t mode)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
int sv[2];
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
struct sockaddr_un sun;
struct sockaddr *sa;
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
char progname[] = _PATH_PERFUSED;
char minus_i[] = "-i";
char fdstr[16];
char *const argv[] = { progname, minus_i, fdstr, NULL};
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
uint32_t opt;
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
uint32_t optlen;
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
int sock_type = SOCK_SEQPACKET;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
if (strcmp(path, _PATH_FUSE) != 0)
return open(path, flags, mode);
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
/*
* Try SOCK_SEQPACKET then SOCK_DGRAM if unavailable
*/
if ((sv[0] = socket(PF_LOCAL, SOCK_SEQPACKET, 0)) == -1) {
sock_type = SOCK_DGRAM;
DWARNX("SEQPACKET local sockets unavailable, using less "
"reliable DGRAM sockets. Expect file operation hangs.");
if ((sv[0] = socket(PF_LOCAL, SOCK_DGRAM, 0)) == -1) {
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#ifdef PERFUSE_DEBUG
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DWARN("%s: %d socket failed", __func__, __LINE__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#endif
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
return -1;
}
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
/*
fix typo
2019-04-17 15:30:51 +03:00
* Set a buffer length large enough so that enough FUSE packets
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
* will fit.
*/
make the env stuff visible.
2016-10-19 04:30:35 +03:00
opt = perfuse_bufvar_from_env("PERFUSE_BUFSIZE",
(uint32_t)(16 * FUSE_BUFSIZE));
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
optlen = sizeof(opt);
if (setsockopt(sv[0], SOL_SOCKET, SO_SNDBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_SNDBUF to %d failed", __func__, opt);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
if (setsockopt(sv[0], SOL_SOCKET, SO_RCVBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_RCVBUF to %d failed", __func__, opt);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
sa = (struct sockaddr *)(void *)&sun;
sun.sun_len = sizeof(sun);
sun.sun_family = AF_LOCAL;
(void)strcpy(sun.sun_path, path);
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
if (connect(sv[0], sa, (socklen_t)sun.sun_len) == 0)
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
return sv[0];
/*
* Attempt to run perfused on our own
* if it does not run yet; In that case
* we will talk using a socketpair
* instead of /dev/fuse.
*/
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
if (socketpair(PF_LOCAL, sock_type, 0, sv) != 0) {
- Do not checkfor peer credentials when perfused is autostarted and therefore runs with filesystem privileges - shut up warnings and debug messages when perfused is autostarted - make perfused patch modifiable with CFLAGS for easier pkgsrc integration - Fix build warnings
2010-09-07 06:11:04 +04:00
DWARN("%s:%d: socketpair failed", __func__, __LINE__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
return -1;
}
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
/*
fix typo
2019-04-17 15:30:51 +03:00
* Set a buffer length large enough so that enough FUSE packets
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
* will fit.
*/
make the env stuff visible.
2016-10-19 04:30:35 +03:00
opt = perfuse_bufvar_from_env("PERFUSE_BUFSIZE",
(uint32_t)(16 * FUSE_BUFSIZE));
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
optlen = sizeof(opt);
if (setsockopt(sv[0], SOL_SOCKET, SO_SNDBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_SNDBUF to %d failed", __func__, opt);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
if (setsockopt(sv[0], SOL_SOCKET, SO_RCVBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_RCVBUF to %d failed", __func__, opt);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
if (setsockopt(sv[1], SOL_SOCKET, SO_SNDBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_SNDBUF to %d failed", __func__, opt);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
if (setsockopt(sv[1], SOL_SOCKET, SO_RCVBUF, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt SO_RCVBUF to %d failed", __func__, opt);
/*
* Request peer credentials. This musr be done before first
* frame is sent.
*/
opt = 1;
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
optlen = sizeof(opt);
if (setsockopt(sv[1], 0, LOCAL_CREDS, &opt, optlen) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s: setsockopt LOCAL_CREDS failed", __func__);
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
(void)sprintf(fdstr, "%d", sv[1]);
switch(fork()) {
case -1:
#ifdef PERFUSE_DEBUG
- Do not checkfor peer credentials when perfused is autostarted and therefore runs with filesystem privileges - shut up warnings and debug messages when perfused is autostarted - make perfused patch modifiable with CFLAGS for easier pkgsrc integration - Fix build warnings
2010-09-07 06:11:04 +04:00
DWARN("%s:%d: fork failed", __func__, __LINE__);
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
#endif
return -1;
/* NOTREACHED */
break;
case 0:
Make sure perfused exit when the filesystem crashed, so that unmount is done. Failure to do so caused deadlocks, with operation that held a lock on the root vnode and got stuck in perfused forever. Approved by releng.
2012-02-03 19:54:15 +04:00
(void)close(sv[0]);
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
(void)execve(argv[0], argv, environ);
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
#ifdef PERFUSE_DEBUG
- Do not checkfor peer credentials when perfused is autostarted and therefore runs with filesystem privileges - shut up warnings and debug messages when perfused is autostarted - make perfused patch modifiable with CFLAGS for easier pkgsrc integration - Fix build warnings
2010-09-07 06:11:04 +04:00
DWARN("%s:%d: execve failed", __func__, __LINE__);
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
#endif
return -1;
/* NOTREACHED */
break;
default:
break;
}
Make sure perfused exit when the filesystem crashed, so that unmount is done. Failure to do so caused deadlocks, with operation that held a lock on the root vnode and got stuck in perfused forever. Approved by releng.
2012-02-03 19:54:15 +04:00
(void)close(sv[1]);
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
return sv[0];
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
int
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_mount(const char *source, const char *target,
const char *filesystemtype, long mountflags, const void *data)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
int s;
size_t len;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
struct perfuse_mount_out *pmo;
struct sockaddr_storage ss;
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
struct sockaddr_un *sun;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
struct sockaddr *sa;
socklen_t sa_len;
size_t sock_len;
char *frame;
char *cp;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#ifdef PERFUSE_DEBUG
- Do not checkfor peer credentials when perfused is autostarted and therefore runs with filesystem privileges - shut up warnings and debug messages when perfused is autostarted - make perfused patch modifiable with CFLAGS for easier pkgsrc integration - Fix build warnings
2010-09-07 06:11:04 +04:00
if (perfuse_diagflags & PDF_MISC)
DPRINTF("%s(\"%s\", \"%s\", \"%s\", 0x%lx, \"%s\")\n",
__func__, source, target, filesystemtype,
mountflags, (const char *)data);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#endif
if ((s = get_fd(data)) == -1)
return -1;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
/*
* If we are connected to /dev/fuse, we need a second
* socket to get replies from perfused.
* XXX This socket is not removed at exit time yet
*/
sock_len = 0;
sa = (struct sockaddr *)(void *)&ss;
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
sun = (struct sockaddr_un *)(void *)&ss;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
sa_len = sizeof(ss);
if ((getpeername(s, sa, &sa_len) == 0) &&
(sa->sa_family = AF_LOCAL) &&
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
(strcmp(sun->sun_path, _PATH_FUSE) == 0)) {
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
sun->sun_len = sizeof(*sun);
sun->sun_family = AF_LOCAL;
(void)sprintf(sun->sun_path, "%s/%s-%d",
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
_PATH_TMP, getprogname(), getpid());
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
if (bind(s, sa, (socklen_t)sa->sa_len) != 0)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DERR(EX_OSERR, "%s:%d bind to \"%s\" failed",
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
__func__, __LINE__, sun->sun_path);
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
sock_len = strlen(sun->sun_path) + 1;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
}
len = sizeof(*pmo);
len += source ? (uint32_t)strlen(source) + 1 : 0;
len += target ? (uint32_t)strlen(target) + 1 : 0;
len += filesystemtype ? (uint32_t)strlen(filesystemtype) + 1 : 0;
len += data ? (uint32_t)strlen(data) + 1 : 0;
len += sock_len;
if ((frame = malloc(len)) == NULL) {
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#ifdef PERFUSE_DEBUG
- Do not checkfor peer credentials when perfused is autostarted and therefore runs with filesystem privileges - shut up warnings and debug messages when perfused is autostarted - make perfused patch modifiable with CFLAGS for easier pkgsrc integration - Fix build warnings
2010-09-07 06:11:04 +04:00
if (perfuse_diagflags & PDF_MISC)
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s:%d malloc failed", __func__, __LINE__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#endif
return -1;
}
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
pmo = (struct perfuse_mount_out *)(void *)frame;
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
pmo->pmo_len = (uint32_t)len;
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
pmo->pmo_error = 0;
pmo->pmo_unique = (uint64_t)-1;
(void)strcpy(pmo->pmo_magic, PERFUSE_MOUNT_MAGIC);
pmo->pmo_source_len = source ? (uint32_t)strlen(source) + 1 : 0;
pmo->pmo_target_len = target ? (uint32_t)strlen(target) + 1: 0;
pmo->pmo_filesystemtype_len =
filesystemtype ? (uint32_t)strlen(filesystemtype) + 1 : 0;
pmo->pmo_mountflags = (uint32_t)mountflags;
pmo->pmo_data_len = data ? (uint32_t)strlen(data) + 1 : 0;
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
pmo->pmo_sock_len = (uint32_t)sock_len;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
cp = (char *)(void *)(pmo + 1);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
if (source) {
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
(void)strcpy(cp, source);
cp += pmo->pmo_source_len;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
if (target) {
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
(void)strcpy(cp, target);
cp += pmo->pmo_target_len;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
if (filesystemtype) {
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
(void)strcpy(cp, filesystemtype);
cp += pmo->pmo_filesystemtype_len;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
if (data) {
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
(void)strcpy(cp, data);
cp += pmo->pmo_data_len;
}
if (sock_len != 0) {
- performance improvement for read, readdir and write. Now we use SOCK_DGRAM, we can send many pages at once without hitting any bug - when creating a file, it is open for FUSE, but not for the kernel. If the kernel does not do a subsequent open, we have a leak. We fight against this by trying to close such file that the kernel left unopen for some time. - some code refactoring to make message exchange debug easier (more to come)
2010-09-20 11:00:21 +04:00
(void)strcpy(cp, sun->sun_path);
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
cp += pmo->pmo_sock_len;
}
- Implement proper unprivilegied user permission verifications Verification is now done in the lookup method, as it is the way to go. Of course there are corner cases, such as the sticky bit which need special handling in the remove method. - Set full fsidx in vftstat method - Do not pass O_APPEND to the filesystem. FUSE always sends the write offset, so setting O_APPEND is useless. If the filesystem uses it in an open(2) system call, it will even cause file corruptions, since offsets given to pwrite(2) will be ignored. This fix allows glusterfs to host a NetBSD ./build.sh -o build - Do not use the FUSE access method, use getattr and check for permission on our own. The problem is that a FUSE filesystem will typically use the Linux-specific setfsuid() to perform access control. If that is missing, any chack is likely to occur on behalf of the user running the filesystem (typically root), causing access method to return wrong information. - When possible, avoid performing a getattr method call and use cached value in puffs_node instead. We still retreive the latest value by calling getattr when performing append write operation, to minimize the chances that another writer appended since the last time we did. - Update puffs_node cached file size in write method - Remove unused argument to perfuse_destroy_pn()
2011-04-25 08:54:53 +04:00
if (send(s, frame, len, MSG_NOSIGNAL) != (ssize_t)len) {
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#ifdef PERFUSE_DEBUG
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
DWARN("%s:%d sendto failed", __func__, __LINE__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
#endif
- Use SOCK_DGRAM instead of SOCK_STREAM, as the filesystem seems to assume datagram semantics: when using SOCK_STREAM, if perfused sends frames faster than the filesystem consumes them, it will grab multiple frames at once and discard anything beyond the first one. For now the code can work both with SOCK_DGRAM and SOCK_STREAM, but SOCK_STREAM support will probably have to be removed for the sake of readability. - Remeber to sync parent directories when moving a node - In debug output, display the requeue type (readdir, write, etc...)
2010-09-15 05:51:43 +04:00
return -1;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
return 0;
}
uint64_t
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_next_unique(struct puffs_usermount *pu)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
return ps->ps_unique++;
}
Deal with limits properly. Don't print strerror() 2ice. XXX: pullup-7
2015-06-19 20:33:20 +03:00
static void
updatelimit(const char *func, int lim, const char *name)
{
struct rlimit rl;
/* Try infinity but that will fail unless we are root */
rl.rlim_cur = RLIM_INFINITY;
rl.rlim_max = RLIM_INFINITY;
if (setrlimit(lim, &rl) != -1)
return;
/* Get and set to the maximum allowed */
if (getrlimit(lim, &rl) == -1)
DERR(EX_OSERR, "%s: getrlimit %s failed", func, name);
rl.rlim_cur = rl.rlim_max;
if (setrlimit(lim, &rl) == -1)
DERR(EX_OSERR, "%s: setrlimit %s to %ju failed", func,
name, (uintmax_t)rl.rlim_cur);
}
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
struct puffs_usermount *
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_init(struct perfuse_callbacks *pc, struct perfuse_mount_info *pmi)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
struct puffs_usermount *pu;
struct puffs_ops *pops;
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
const char *source = _PATH_PUFFS;
char *fstype;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
unsigned int puffs_flags;
struct puffs_node *pn_root;
struct puffs_pathobj *po_root;
perfuse memory usage can grow quite large when using a lot of vnodes, and the amount of data memory involved is not easy to forcast. We therefore raise the limit to the maximum. Patch from Manuel Bouyer. It helps completing a cvs update on a glusterfs colume.
2011-10-23 09:01:00 +04:00
/*
* perfused can grow quite large, let assume there's enough ram ...
*/
Deal with limits properly. Don't print strerror() 2ice. XXX: pullup-7
2015-06-19 20:33:20 +03:00
updatelimit(__func__, RLIMIT_DATA, "RLIMIT_DATA");
updatelimit(__func__, RLIMIT_AS, "RLIMIT_AS");
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps = init_state();
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
ps->ps_owner_uid = pmi->pmi_uid;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
if (pmi->pmi_source) {
if ((ps->ps_source = strdup(pmi->pmi_source)) == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: strdup failed", __func__);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
source = ps->ps_source;
}
if (pmi->pmi_filesystemtype) {
size_t len;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps->ps_filesystemtype = strdup(pmi->pmi_filesystemtype);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
if (ps->ps_filesystemtype == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: strdup failed", __func__);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
len = sizeof("perfuse|") + strlen(ps->ps_filesystemtype) + 1;
if ((fstype = malloc(len)) == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: malloc failed", __func__);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
(void)sprintf(fstype, "perfuse|%s", ps->ps_filesystemtype);
} else {
if ((fstype = strdup("perfuse")) == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: strdup failed", __func__);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
}
if ((ps->ps_target = strdup(pmi->pmi_target)) == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: strdup failed", __func__);
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps->ps_mountflags = pmi->pmi_mountflags;
/*
* Some options are forbidden for non root users
*/
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
if (ps->ps_owner_uid != 0)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps->ps_mountflags |= MNT_NOSUID|MNT_NODEV;
PUFFSOP_INIT(pops);
PUFFSOP_SET(pops, perfuse, fs, unmount);
PUFFSOP_SET(pops, perfuse, fs, statvfs);
PUFFSOP_SET(pops, perfuse, fs, sync);
PUFFSOP_SET(pops, perfuse, node, lookup);
PUFFSOP_SET(pops, perfuse, node, create);
PUFFSOP_SET(pops, perfuse, node, mknod);
PUFFSOP_SET(pops, perfuse, node, open);
PUFFSOP_SET(pops, perfuse, node, close);
PUFFSOP_SET(pops, perfuse, node, access);
PUFFSOP_SET(pops, perfuse, node, getattr);
PUFFSOP_SET(pops, perfuse, node, setattr);
PUFFSOP_SET(pops, perfuse, node, poll);
PUFFSOP_SET(pops, perfuse, node, fsync);
PUFFSOP_SET(pops, perfuse, node, remove);
PUFFSOP_SET(pops, perfuse, node, link);
PUFFSOP_SET(pops, perfuse, node, rename);
PUFFSOP_SET(pops, perfuse, node, mkdir);
PUFFSOP_SET(pops, perfuse, node, rmdir);
PUFFSOP_SET(pops, perfuse, node, symlink);
PUFFSOP_SET(pops, perfuse, node, readdir);
PUFFSOP_SET(pops, perfuse, node, readlink);
PUFFSOP_SET(pops, perfuse, node, reclaim);
Use reclaim2 to fix reclaim/lookup race conditions The PUFFS reclaim operation had a race condition with lookups: we could be asked to lookup a node, then to reclaim it before lookup completion. At lookup completion, we would then create a leaked node. Enter the PUFFS reclaim2 operation, which features a nlookup argument. That let us count how many lookups are pending and avoid the above described scenario. It also makes the codes simplier.
2018-11-16 05:39:02 +03:00
PUFFSOP_SET(pops, perfuse, node, reclaim2);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
PUFFSOP_SET(pops, perfuse, node, inactive);
PUFFSOP_SET(pops, perfuse, node, print);
Improve POSIX compliance of FUSE filesystems through PERUSE - access denied is EPERM and not EACCES - access to file owned by someone else in a sticy-bit directory should be allowed for the sticy-bit directory owner - setting sticky-bit on a non directory should produce EFTYPE - implement PATHCONF method as much as we can.
2014-09-03 20:01:45 +04:00
PUFFSOP_SET(pops, perfuse, node, pathconf);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
PUFFSOP_SET(pops, perfuse, node, advlock);
PUFFSOP_SET(pops, perfuse, node, read);
PUFFSOP_SET(pops, perfuse, node, write);
Add support for extended attributes
2011-06-28 20:19:16 +04:00
#ifdef PUFFS_EXTNAMELEN
PUFFSOP_SET(pops, perfuse, node, getextattr);
PUFFSOP_SET(pops, perfuse, node, setextattr);
PUFFSOP_SET(pops, perfuse, node, listextattr);
PUFFSOP_SET(pops, perfuse, node, deleteextattr);
#endif /* PUFFS_EXTNAMELEN */
- When using PUFFS_KFLAG_CACHE_FS_TTL, do not use puffs_node to carry attribute and TTL fora newly created node. Instead extend puffs_newinfo and add puffs_newinfo_setva() and puffs_newinfo_setttl() - Remove node_mk_common_final in libperfuse. It used to set uid/gid for a newly created vnode but has been made redundant along time ago since uid and gid are properly set in FUSE header. - In libperfuse, check for corner case where opc = 0 on INACTIVE and RECLAIM (how is it possible? Check for it to avoid a crash anyway) - In libperfuse, make sure we unlimit RLIMIT_AS and RLIMIT_DATA so that we do notrun out of memory because the kernel is lazy at reclaiming vnodes. - In libperfuse, cleanup style of perfuse_destroy_pn()
2012-04-18 04:57:21 +04:00
#ifdef PUFFS_KFLAG_CACHE_FS_TTL
PUFFSOP_SET(pops, perfuse, node, getattr_ttl);
PUFFSOP_SET(pops, perfuse, node, setattr_ttl);
#endif /* PUFFS_KFLAG_CACHE_FS_TTL */
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
#ifdef PUFFS_SETATTR_FAF
PUFFSOP_SET(pops, perfuse, node, write2);
#endif /* PUFFS_SETATTR_FAF */
Use just introduced open2 PUFFS method and its PUFFS_OPEN_IO_DIRECT oflag to implement FUSE's OPEN_IO_DIRECT, by which the filesystem tells the kernel that read/write to the file should bypass the page cache. Remove a warning about read beyond EOF which will now normally appear when page cache is bypassed.
2014-08-16 20:28:43 +04:00
#ifdef PUFFS_OPEN_IO_DIRECT
PUFFSOP_SET(pops, perfuse, node, open2);
#endif /* PUFFS_OPEN_IO_DIRECT */
FUSE fallocate support There seems to be no fdiscard FUSE operation at the moment, hence that one is left unused.
2014-10-31 18:12:15 +03:00
#ifdef PUFFS_HAVE_FALLOCATE
PUFFSOP_SET(pops, perfuse, node, fallocate);
#endif /* PUFFS_HAVE_FALLOCATE */
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
Remove PUFFS_KFLAG_WTCACHE, which caused data corruption and slowdown
2011-08-09 10:58:33 +04:00
/*
- Fix the confusion between fileno (opaque FUSE reference) and inode numbers. fileno must be used when exchanging FUSE messages. - Do not use kernel name cache anymore, as it caused modification from other machines to be invisible. - Honour name and attribute cache directive from FUSE filesystem
2011-10-30 09:11:37 +04:00
* PUFFS_KFLAG_NOCACHE_NAME is required so that we can see changes
Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5.
2012-04-08 19:13:06 +04:00
* done by other machines in networked filesystems. In later
* NetBSD releases we use the alternative PUFFS_KFLAG_CACHE_FS_TTL,
* which implement name cache with a filesystem-provided TTL.
Remove PUFFS_KFLAG_WTCACHE, which caused data corruption and slowdown
2011-08-09 10:58:33 +04:00
*/
Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5.
2012-04-08 19:13:06 +04:00
#ifdef PUFFS_KFLAG_CACHE_FS_TTL
puffs_flags = PUFFS_KFLAG_CACHE_FS_TTL;
#else
- Fix the confusion between fileno (opaque FUSE reference) and inode numbers. fileno must be used when exchanging FUSE messages. - Do not use kernel name cache anymore, as it caused modification from other machines to be invisible. - Honour name and attribute cache directive from FUSE filesystem
2011-10-30 09:11:37 +04:00
puffs_flags = PUFFS_KFLAG_NOCACHE_NAME;
Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5.
2012-04-08 19:13:06 +04:00
#endif
Add PUFFS_KFLAG_CACHE_DOTDOT so that vnodes hold a reference on their parent, keeping them active, and allowing to lookup .. without sending a request to the filesystem. Enable the featuure for perfused, as this is how FUSE works.
2012-08-10 20:49:35 +04:00
/*
* Do not lookuo ..
* That means we keep all parent vnode active
*/
#ifdef PUFFS_KFLAG_CACHE_DOTDOT
puffs_flags |= PUFFS_KFLAG_CACHE_DOTDOT;
#endif
Use new PUFFS_KFLAG_CACHE_FS_TTL option to puffs_init(3) so that FUSE TTL on name and attributes are used. This save many PUFFS operations and improves performances. PUFFS_KFLAG_CACHE_FS_TTL is #ifdef'ed in many places for now so that libperfuse can still be used on netbsd-5.
2012-04-08 19:13:06 +04:00
/*
* It would be nice to avoid useless inactive, and only
* get them on file open for writing (PUFFS does
* CLOSE/WRITE/INACTIVE, therefore actual close must be
* done at INACTIVE time). Unfortunatley, puffs_setback
* crashes when called on OPEN, therefore leave it for
* another day.
*/
#ifdef notyet
puffs_flags |= PUFFS_FLAG_IAONDEMAND;
#endif
= Open files = - Restore open on our own in fsycn and readdir, as the node may not already be open, and FUSE really wants it to be. No need to close immediatly, it can be done at inactive time. = Write operations = - fix a nasty bug that corrupted files on write (written added twice) - Keep track of file size in order to honour PUFFS_IO_APPEND = many fixes in rename = - handler overwritten nodes correctly - wait for all operations on the node to drain before doing rename, as filesystems may not cope with operations on a moving file. - setback PUFFS_SETBACK_INACT_N1 cannot be used from rename, we therefore miss the inactive time for an overwritten node. This bounds us to give up PUFFS_KFLAG_IAONDEMAND. = Removed files = - forbid most operations on a removed node, return ENOENT - setback PUFFS_SETBACK_NOREF_N1 at inactive stage to cause removed file reclaim = Misc = - Update outdated ARGSUSED for lint - Fix a memory leak (puffs_pn_remove instead of puffs_pn_put) - Do not use PUFFS_FLAG_BUILDPATH except for debug output. It makes the lookup code much simplier.
2010-09-29 12:01:10 +04:00
Add PUFFS_KFLAG_NOFLUSH_META to prevent sending metadata flush to FUSE FUSE filesystems do not expect to get metadata updates for [amc]time and size, they updates the value on their own after operations. The PUFFS PUFFS_KFLAG_NOFLUSH_META option prevents regular metadata cache flushes to the filesystem , and libperfuse uses it to match Linux FUSE behavior. While there, fix a bug in SETATTR: do not update kernel metadata cache from SETATTR reply when the request is asynchronous, as we do not have the reply yet.
2015-02-15 23:21:29 +03:00
/*
* FUSE filesystem do not expect [amc]time and size
* updates to be sent by the kernel, they do the
* updates on their own after other operations.
*/
#ifdef PUFFS_KFLAG_NOFLUSH_META
puffs_flags |= PUFFS_KFLAG_NOFLUSH_META;
#endif
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
if (perfuse_diagflags & PDF_PUFFS)
puffs_flags |= PUFFS_FLAG_OPDUMP;
Mont FUSE filesystem with proprer source and fstype so that df and mount display something that makes sense
2011-05-12 14:32:41 +04:00
if ((pu = puffs_init(pops, source, fstype, ps, puffs_flags)) == NULL)
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: puffs_init failed", __func__);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
puffs_setncookiehash(pu, PUFFS_PNODEBUCKETS);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps->ps_pu = pu;
/*
* Setup filesystem root
*/
- Correctly handle rename whith overwritten destination - Keep track of file name to avoid lookups when we can. This makes sure we do not have two cookies for the same inode, a situation that cause wreak havoc when we come to remove or rename a node. - Do not use PUFFS_FLAG_BUILDPATH at all, since we now track file names - In open, queue requests after checking for access, as there is no merit to queue a will-be-denied request while we can deny it immediatly - request reclaim of removed nodes at inactive stage
2010-10-03 09:46:47 +04:00
pn_root = perfuse_new_pn(pu, "", NULL);
- Fix the confusion between fileno (opaque FUSE reference) and inode numbers. fileno must be used when exchanging FUSE messages. - Do not use kernel name cache anymore, as it caused modification from other machines to be invisible. - Honour name and attribute cache directive from FUSE filesystem
2011-10-30 09:11:37 +04:00
PERFUSE_NODE_DATA(pn_root)->pnd_nodeid = FUSE_ROOT_ID;
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
PERFUSE_NODE_DATA(pn_root)->pnd_parent_nodeid = FUSE_ROOT_ID;
perfuse_node_cache(ps, pn_root);
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
puffs_setroot(pu, pn_root);
ps->ps_fsid = pn_root->pn_va.va_fsid;
po_root = puffs_getrootpathobj(pu);
if ((po_root->po_path = strdup("/")) == NULL)
DERRX(EX_OSERR, "perfuse_mount_start() failed");
po_root->po_len = 1;
puffs_path_buildhash(pu, po_root);
puffs_vattr_null(&pn_root->pn_va);
pn_root->pn_va.va_type = VDIR;
pn_root->pn_va.va_mode = 0755;
- Fix the confusion between fileno (opaque FUSE reference) and inode numbers. fileno must be used when exchanging FUSE messages. - Do not use kernel name cache anymore, as it caused modification from other machines to be invisible. - Honour name and attribute cache directive from FUSE filesystem
2011-10-30 09:11:37 +04:00
pn_root->pn_va.va_fileid = FUSE_ROOT_ID;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
ps->ps_root = pn_root;
/*
* Callbacks
*/
ps->ps_new_msg = pc->pc_new_msg;
ps->ps_xchg_msg = pc->pc_xchg_msg;
ps->ps_destroy_msg = pc->pc_destroy_msg;
ps->ps_get_inhdr = pc->pc_get_inhdr;
ps->ps_get_inpayload = pc->pc_get_inpayload;
ps->ps_get_outhdr = pc->pc_get_outhdr;
ps->ps_get_outpayload = pc->pc_get_outpayload;
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
ps->ps_umount = pc->pc_umount;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
pc->pc_fsreq = *perfuse_fsreq;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
return pu;
}
void
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_setspecific(struct puffs_usermount *pu, void *priv)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
ps->ps_private = priv;
return;
}
void *
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_getspecific(struct puffs_usermount *pu)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
return ps->ps_private;
}
int
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_inloop(struct puffs_usermount *pu)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
return ps->ps_flags & PS_INLOOP;
}
int
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_mainloop(struct puffs_usermount *pu)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
ps->ps_flags |= PS_INLOOP;
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
if (puffs_mainloop(ps->ps_pu) != 0) {
- fix warn/err confusiog - fix debugging printf - add func arguments to simple formats
2011-08-14 03:12:15 +04:00
DERR(EX_OSERR, "%s: failed", __func__);
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
return -1;
}
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
Use SOCK_SEQPACKET in perfuse if available. This fix file operations hangs where the FUSE filesyste replied to an operation and got an ENOBUFS it did not handle. We now are also able to cleanly unmount
2011-05-30 18:50:08 +04:00
/*
* Normal exit after unmount
*/
return 0;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
/* ARGSUSED0 */
uint64_t
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_get_nodeid(struct puffs_usermount *pu, puffs_cookie_t opc)
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
{
- Fix the confusion between fileno (opaque FUSE reference) and inode numbers. fileno must be used when exchanging FUSE messages. - Do not use kernel name cache anymore, as it caused modification from other machines to be invisible. - Honour name and attribute cache directive from FUSE filesystem
2011-10-30 09:11:37 +04:00
return PERFUSE_NODE_DATA(opc)->pnd_nodeid;
libperfuse(3) is a PUFFS relay to FUSE. In order to use it, FUSE filesystem must be patched to #include <perfuse.h> in the source files that open /dev/fuse and perform the mount(2) system call. The FUSE filesystem must be linked with -lperfuse. libperfuse(3) implements the FUSE kernel interface, on which libfuse or any FUSE filesystem that opens /dev/fuse directly can be used. For now, an external daemon called perfused(8) is used. This may change in the future.
2010-08-25 11:16:00 +04:00
}
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
int
Use C89 function definition
2012-03-21 14:10:36 +04:00
perfuse_unmount(struct puffs_usermount *pu)
- if perfused is not already started (cannot connect to /dev/fuse), FUSE filesystems will attempt to start it on their own, and will communicate using a socketpair - do not advertise NULL file handle as being valid when sending themback to the FUSE filesystem. - unmount if we cannot talk to the FUSE process anymore - set calling process gid properly - debug message cleanup
2010-08-27 13:58:17 +04:00
{
struct perfuse_state *ps;
ps = puffs_getspecific(pu);
return unmount(ps->ps_target, MNT_FORCE);
}
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
void
perfuse_fsreq(struct puffs_usermount *pu, perfuse_msg_t *pm)
{
struct perfuse_state *ps;
struct fuse_out_header *foh;
ps = puffs_getspecific(pu);
foh = GET_OUTHDR(ps, pm);
/*
* There are some operations we may use in a Fire and Forget wey,
* because the kernel does not await a reply, but FUSE still
* sends a reply. This happens for fsyc, setattr (for metadata
* associated with a fsync) and write (for VOP_PUTPAGES). Ignore
* if it was fine, warn or abort otherwise.
*/
switch (foh->error) {
case 0:
break;
case -ENOENT:
/* File disapeared during a FAF operation */
break;
case -ENOTCONN: /* FALLTHROUGH */
case -EAGAIN: /* FALLTHROUGH */
case -EMSGSIZE:
DWARN("operation unique = %"PRId64" failed", foh->unique);
break;
default:
Turn a fatal error into a warning.
2012-09-10 17:56:18 +04:00
DWARNX("Unexpected frame: unique = %"PRId64", error = %d",
foh->unique, foh->error);
- Fix same vnodes associated with multiple cookies The scheme used to retreive known nodes on lookup was flawed, as it only used parent and name. This produced a different cookie for the same file if it was renamed, when looking up ../ or when dealing with multiple files associated with the same name through link(2). We therefore abandon the use of node name and introduce hashed lists of inodes. This causes a huge rewrite of reclaim code, which do not attempt to keep parents allocated until all their children are reclaimed - Fix race conditions in reclaim There are a few situations where we issue multiple FUSE operations for a PUFFS operation. On reclaim, we therefore have to wait for all FUSE operation to complete, not just the current exchanges. We do this by introducing node reference count with node_ref() and node_rele(). - Detect data loss caused by FAF VOP_PUTPAGES causes FAF writes where the kernel does not check the operation result. At least issue a warning on error. - Enjoy FAF shortcut on setattr No need to wait for the result if the kernel does not want it. There is however an exception for setattr that touch the size, we need to wait for completion because we have other operations queued for after the resize. - Fix fchmod() on write-open file fchmod() on a node open with write privilege will send setattr with both mode and size set. This confuses some FUSE filesystem. Therefore we send two FUSE operations, one for mode, and one for size. - Remove node TTL handling for netbsd-5 for simplicity sake. The code still builds on netbsd-5 but does not have the node TTL feature anymore. It works fine with kernel support on netbsd-6.
2012-07-21 09:49:42 +04:00
/* NOTREACHED */
break;
}
ps->ps_destroy_msg(pm);
return;
}
Reference in New Issue Copy Permalink