Brian Behlendorf [Tue, 19 Apr 2011 17:02:21 +0000 (10:02 -0700)]
Fix gcc configure warnings
Newer versions of gcc are getting smart enough to detect the sloppy
syntax used for the autoconf tests. It is now generating warnings
for unused/undeclared variables. Newer version of gcc even have
the -Wunused-but-set-variable option set by default. This isn't a
problem except when -Werror is set and they get promoted to an error.
In this case the autoconf test will return an incorrect result which
will result in a build failure latter on.
To handle this I'm tightening up many of the autoconf tests to
explicitly mark variables as unused to suppress the gcc warning.
Remember, all of the autoconf code can never actually be run we
just want to get a clean build error to detect which APIs are
available. Never using a variable is absolutely fine for this.
Brian Behlendorf [Mon, 18 Apr 2011 23:44:22 +0000 (16:44 -0700)]
Fix gcc compiler warning, parse_option()
When compiling ZFS in user space gcc-4.6.0 correctly identifies
the variable 'value' as being set but never used. This generates a
warning and a build failure when using --enable-debug. Once again
this is correct but I'm reluctant to remove 'value' because we are
breaking the string in to name/value pairs. While it is not used
now there's a good chance it will be soon and I'd rather not have
to reinvent this. To suppress the warning with just as a VERIFY().
This was observed under Fedora 15.
cmd/mount_zfs/mount_zfs.c: In function ‘parse_option’:
cmd/mount_zfs/mount_zfs.c:112:21: error: variable ‘value’ set but not
used [-Werror=unused-but-set-variable]
Brian Behlendorf [Mon, 18 Apr 2011 23:27:45 +0000 (16:27 -0700)]
Fix gcc compiler warning, dsl_pool_create()
When compiling ZFS in user space gcc-4.6.0 correctly identifies
the variable 'os' as being set but never used. This generates a
warning and a build failure when using --enable-debug. However,
the code is correct we only want to use 'os' for the kernel space
builds. To suppress the warning the call was wrapped with a
VERIFY() which has the nice side effect of ensuring the 'os'
actually never is NULL. This was observed under Fedora 15.
module/zfs/dsl_pool.c: In function ‘dsl_pool_create’:
module/zfs/dsl_pool.c:229:12: error: variable ‘os’ set but not used
[-Werror=unused-but-set-variable]
Brian Behlendorf [Mon, 18 Apr 2011 21:12:28 +0000 (14:12 -0700)]
Linux 2.6.39 compat, invalidate_inodes()
Update code to use the spl_invalidate_inodes() wrapper. This hides
some of the complexity of determining if invalidate_inodes() was
exported, and if so what is its prototype. The second argument
of spl_invalidate_inodes() determined the behavior of how dirty
inodes are handled. By passing a zero we are indicated that we
want those inodes to be treated as busy and skipped.
Brian Behlendorf [Mon, 18 Apr 2011 19:48:44 +0000 (12:48 -0700)]
Autogen refresh for kernel-insert-inode-locked.m4
Several Makefile.in's were accidentally not updated when the
kernel-insert-inode-locked.m4 check was added. This change simply
refreshes the missed files.
When rebuilding the source RPM under el5 you need to append the
target_cpu. However, under el6/ch5 things are packaged correctly
and the arch is already part of kver. For this reason it also
needs to be stripped from kver when setting kverpkg.
The .sync_fs fix as applied did not use the updated SPL credential
API. This broke builds on Debian Lenny, this change applies the
needed fix to use the portable API. The original credential changes
are part of commit 81e97e21872a9c38ad66c37fafe1436ee25abee3.
Apply all of Rudd-O's changes for the Fedora init script. The
initial init script was one I threw together based on Rudd-O's
original work. It worked for me but it has some flaws.
Rudd-O has invested considerable time updating it to be significantly
smarter. It now handles using ZFS as your root filesystem plus
various other quirks. Since he is familiar with the right
way to do things on Fedora and has tested this init script we
are integrating all of his changes.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Disable the normal reclaim path for the txg_sync thread. This
ensures the thread will never enter dmu_tx_assign() which can
otherwise occur due to direct reclaim. If this is allowed to
happen the system can deadlock. Direct reclaim call path:
Brian Behlendorf [Wed, 30 Mar 2011 01:08:59 +0000 (18:08 -0700)]
Add direct+indirect ARC reclaim
Under OpenSolaris all memory reclaim is done asyncronously. Under
Linux memory reclaim is done asynchronously _and_ synchronously.
When a process allocates memory with GFP_KERNEL it explicitly allows
the kernel to do reclaim on its behalf to satify the allocation.
If that GFP_KERNEL allocation fails the kernel may take more drastic
measures to reclaim the memory such as killing user space processes.
This was observed to happen with ZFS because the ARC could consume
a large fraction of the system memory but no synchronous reclaim
could be performed on it. The result was GFP_KERNEL allocations
could fail resulting in OOM events, and only moments latter the
arc_reclaim thread would free unused memory from the ARC.
This change leaves the arc_thread in place to manage the fundamental
ARC behavior. But it adds a synchronous (direct) reclaim path for
the ARC which can be called when memory is badly needed. It also
adds an asynchronous (indirect) reclaim path which is called
much more frequently to prune the ARC slab caches.
Brian Behlendorf [Wed, 30 Mar 2011 06:04:39 +0000 (23:04 -0700)]
Call d_instantiate before unlocking inode
Under Linux a dentry referencing an inode must be instantiated before
the inode is unlocked. To accomplish this without overly modifing
the core ZFS code the dentry it passed via the vattr_t. There are
cases such as replay when a dentry is not available. In which case
it is obviously not initialized at inode creation time, if a dentry
is needed it will be spliced as when required via d_lookup().
Fix `make distclean` for `./configure --with-config=user
Making distclean in module
make[1]: Entering directory `/zfs/module'
make -C SUBDIRS=`pwd` clean
make: Entering an unknown directory
make: *** SUBDIRS=/zfs/module: No such file or directory. Stop.
When using --with-config=user the 'distclean' target would fail
because it assumes the kernel configuration infrastrure is set up.
This is not the case, nor does it need to be, because the
'--with-config=user' option will prune the entire ./module subtree
from SUBDIRS. This prevents most build rules from operating in the
./module directory.
However, the 'dist*' rules will still traverse this directory
because it is listed in DIST_SUBDIRS. This is correct because we
need to ensure the dist rules package the directory contents
regardless of the configuration for the 'dist' rule. The correct
way to handle this is to only invoke the kernel build system as
part of the 'clean' rule when CONFIG_KERNEL_TRUE is set.
Initial fix provided by Darik Horn <dajhorn@vanadac.com>.
This commit is a slightly refined form of the original.
Ned Bass [Fri, 1 Apr 2011 16:47:05 +0000 (09:47 -0700)]
Call udevadm trigger more safely
Some udev hooks are not designed to be idempotent, so calling udevadm
trigger outside of the distribution's initialization scripts can have
unexpected (and potentially dangerous) side effects. For example, the
system time may change or devices may appear multiple times. See Ubuntu
launchpad bug 320200 and this mailing list post for more details:
To avoid these problems we call udevadm trigger with --action=change
--subsystem-match=block. The first argument tells udev just to refresh
devices, and make sure everything's as it should be. The second
argument limits the scope to block devices, so devices belonging to
other subsystems cannot be affected.
This doesn't fix the problem on older udev implementations that don't
provide udevadm but instead have udevtrigger as a standalone program.
In this case the above options aren't available so there's no way to
call call udevtrigger safely. But we can live with that since this
issue only exists in optional test and helper scripts, and most
zfs-on-linux users are running newer systems anyways.
Brian Behlendorf [Thu, 31 Mar 2011 20:43:49 +0000 (13:43 -0700)]
Update CHAOS 5 Packaging
The CHAOS 5 kernels are now packaged identially to the RHEL6 kernels.
Therefore we can simply use the RHEL6 rules in the spec file when
building packages.
Brian Behlendorf [Thu, 31 Mar 2011 19:16:24 +0000 (12:16 -0700)]
Fix libzpool cv_* build error
This build failure was accidentally introduced by previous commit bfd214a which fixed the load average. Unfortunately, the wrapper
for cv_wait_interruptible was not available in the zfs_context.h
user compatibility code. I failed to notice this because I didn't
rebuild everything cleanly before committing.
undefined reference to `cv_wait_interruptible'
collect2: ld returned 1 exit status
Kernel threads which sleep uninterruptibly on Linux are marked in the (D)
state. These threads are usually in the process of performing IO and are
thus counted against the load average. The txg_quiesce and txg_sync threads
were always sleeping uninterruptibly and thus inflating the load average.
This change makes them sleep interruptibly. Some care is required however
because these threads may now be woken early by signals. In this case the
callers are all careful to check that the required conditions are met after
waking up. If we're woken early due to a signal they will simply go back
to sleep. In this case these changes are safe.
Fajar A. Nugraha [Fri, 25 Mar 2011 17:01:28 +0000 (10:01 -0700)]
Spec file compat, %{datadir}
The dracut change caused an error during "make rpm". The cause
is simple, RHEL5 does not recognize the %{datarootdir} macro in
zfs.spec. It was changed to %{datadir} which fixes the build.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Thu, 24 Mar 2011 18:34:41 +0000 (11:34 -0700)]
Set cmd paths in udev rules using --prefix
The udev/rules.d scripts must use absolute paths to their support
binaries. However, where those binaries get installed depends
on what --prefix was set to when the package was configured.
This change makes the udev/rules.d helpers to *.in files which
are processed by configure. This allows them to be dynamically
updated to include the specified --prefix.
Additionally, this change updates 60-zvol.rules to handle both
the 'add' and 'change' actions. This ensures that that all
valid zvol devices are correctly linked.
Fajar A. Nugraha [Thu, 24 Mar 2011 08:22:52 +0000 (15:22 +0700)]
Fixes to enable zvol symlink creation
This commit fixes issue on
https://github.com/behlendorf/zfs/issues/#issue/172
Changes:
- update BLKZNAME to use _IOR instead of _IO. Kernel 2.6.32 allows
read parameters (copy_to_user) with _IO, while newer kernels (tested
Archlinux's 2.6.37 kernel) enforces _IOR (which is correct)
- fix return code and message on error
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Tue, 22 Mar 2011 18:22:49 +0000 (11:22 -0700)]
Linux 2.6.29 compat, .freeze_fs/.unfreeze_fs
The .freeze_fs/.unfreeze_fs hooks were not added until Linux 2.6.29
Since these hooks are currently unused they are being removed to
allow support of older kernels.
Brian Behlendorf [Tue, 22 Mar 2011 18:13:41 +0000 (11:13 -0700)]
Linux 2.6.29 compat, credentials
As of Linux 2.6.29 a clean credential API was added to the Linux kernel.
Previously the credential was embedded in the task_struct. Because the
SPL already has considerable support for handling this API change the
ZPL code has been updated to use the Solaris credential API.
Brian Behlendorf [Tue, 22 Mar 2011 16:55:09 +0000 (09:55 -0700)]
Linux 2.6.28 compat, insert_inode_locked()
Added insert_inode_locked() helper function, prior to this most callers
used insert_inode_hash(). The older method doesn't check for collisions
in the inode_hashtable but it still acceptible for use. Fallback to
using insert_inode_hash() when insert_inode_locked() is unavailable.
Brian Behlendorf [Tue, 22 Mar 2011 16:26:38 +0000 (09:26 -0700)]
Linux 2.6.27 compat, blk_queue_stackable()
The blk_queue_stackable() queue flag was added in 2.6.27 to handle dm
stacking drivers. Prior to this request stacking drivers were detected
by checking (q->request_fn == NULL), for earlier kernels we revert to
this legacy behavior.
Brian Behlendorf [Mon, 21 Mar 2011 23:54:59 +0000 (16:54 -0700)]
Linux compat, umount2(2) flags
Older glibc <sys/mount.h> headers did not define all the available
umount2(2) flags. Both MNT_FORCE and MNT_DETACH are supported in the
kernel back to 2.4.11 so we define them correctly if they are missing.
Brian Behlendorf [Mon, 21 Mar 2011 17:19:30 +0000 (10:19 -0700)]
Fix evict() deadlock
Now that KM_SLEEP is not defined as GFP_NOFS there is the possibility
of synchronous reclaim deadlocks. These deadlocks never existed in the
original OpenSolaris code because all memory reclaim on Solaris is done
asyncronously. Linux does both synchronous (direct) and asynchronous
(indirect) reclaim.
This commit addresses a deadlock caused by inode eviction. A KM_SLEEP
allocation may trigger direct memory reclaim and shrink the inode cache.
This can occur while a mutex in the array of ZFS_OBJ_HOLD mutexes is
held. Through the ->shrink_icache_memory()->evict()->zfs_inactive()->
zfs_zinactive() call path the same mutex may be reacquired resulting
in a deadlock. To avoid this deadlock the process must not reacquire
the mutex when it is already holding it.
This is a reasonable fix for now but longer term the ZFS_OBJ_HOLD
mutex locking should be reevaluated. This infrastructure already
prevents us from ever using the Linux lock dependency analysis tools,
and it may limit scalability.
Brian Behlendorf [Sat, 19 Mar 2011 21:34:30 +0000 (14:34 -0700)]
Use KM_PUSHPAGE instead of KM_SLEEP
It used to be the case that all KM_SLEEP allocations were GFS_NOFS.
Unfortunately this often resulted in the kernel being unable to
reclaim the ARC, inode, and dentry caches in a timely manor.
The fix was to make KM_SLEEP a GFP_KERNEL allocation in the SPL.
However, this increases the posibility of deadlocking the system
on a zfs write thread. If a zfs write thread attempts to perform
an allocation it may trigger synchronous reclaim. This reclaim
may attempt to flush dirty data/inode to disk to free memory.
Unforunately, this write cannot finish because the write thread
which would handle it is holding the previous transaction open.
Deadlock.
To avoid this all allocations in the zfs write thread path must
use KM_PUSHPAGE which prohibits synchronous reclaim for that
thread. In this way forward progress in ensured. The risk
with this change is I missed updating an allocation for the
write threads leaving an increased posibility of deadlock. If
any deadlocks remain they will be unlikely but we'll have to
make sure they all get fixed.
Brian Behlendorf [Fri, 18 Mar 2011 21:47:19 +0000 (14:47 -0700)]
Fix 'LDFLAGS=-Wl,--as-needed' build error
Compiling with 'LDFLAGS=-Wl,--as-needed' exposed the fact that
there were some library linking problems introduced by mount_zfs.
In particular, the libzfs library does use nvpair symbols, and
mount_zfs contains no dependencies on libzpool.
Brian Behlendorf [Fri, 18 Mar 2011 20:54:27 +0000 (13:54 -0700)]
Fix getcwd() warning
New versions glibc declare getcwd() with the warn_unused_result attribute.
This results in a warning because the updated mount helper was not
checking this return value. This issue was fixed by checking the return
type and in the case of an error simply returning the passed dataset.
One possible, but unlikely, error would be having your cwd directory
unlinked while the mount command was running.
cmd/mount_zfs/mount_zfs.c: In function ‘parse_dataset’:
cmd/mount_zfs/mount_zfs.c:223:2: error: ignoring return value of
‘getcwd’, declared with attribute warn_unused_result
To simplify the process of using zfs as your root filesystem a
zfs-drucat sub-package has been added. This sub-package adds a zfs
dracut module which allows your initramfs to be rebuilt with zfs
support. The process for doing this is still complicated but there
is clearly interest from the community about getting this working
well and documented. This should help lay some of the groundwork.
Longer term these changes should be pushed in the upstream dracut
package. Once that occurs this subpackage will no longer be
required for new systems, however we may want to conditionally
build this package in the future for systems running older
dracut versions.
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Thu, 17 Mar 2011 22:02:28 +0000 (15:02 -0700)]
Add init scripts
To support automatically mounting your zfs on filesystem on boot
a basic init script is needed. Unfortunately, every distribution
has their own idea of the _right_ way to do things. Rather than
write one very complicated portable init script, which would be
invariably replaced by the distributions own anyway. I have
instead added support to provide multiple distribution specific
init scripts.
The correct init script for your distribution will be selected
by ZFS_AC_DEFAULT_PACKAGE which will set DEFAULT_INIT_SCRIPT.
During 'make install' the correct script for your system will
be installed from zfs/etc/init.d/zfs.DEFAULT_INIT_SCRIPT to the
usual /etc/init.d/zfs location.
Currently, there is zfs.fedora and a more generic zfs.lsb init
script. Hopefully, the distribution maintainers who know best
how they want their init scripts to function will feedback their
approved versions to be included in the project.
This change does not consider upstart jobs but I'm not at all
opposed to add that sort of thing.
Brian Behlendorf [Tue, 15 Mar 2011 19:41:19 +0000 (12:41 -0700)]
Register .remount_fs handler
Register the missing .remount_fs handler. This handler isn't strictly
required because the VFS does a pretty good job updating most of the
MS_* flags. However, there's no harm in using the hook to call the
registered zpl callback for various MS_* flags. Additionaly, this
allows us to lay the ground work for more complicated argument parsing
in the future.
Brian Behlendorf [Tue, 15 Mar 2011 19:03:42 +0000 (12:03 -0700)]
Register .sync_fs handler
Register the missing .sync_fs handler. This is a noop in most cases
because the usual requirement is that sync just be initiated. As part
of the DMU's normal transaction processing txgs will be frequently
synced. However, when the 'wait' flag is set the requirement is that
.sync_fs must not return until the data is safe on disk. With the
addition of the .sync_fs handler this is now properly implemented.
Brian Behlendorf [Tue, 15 Mar 2011 18:17:33 +0000 (11:17 -0700)]
Strip 'zfsutil,remount' from /etc/mtab
When updating /etc/mtab we should be careful and strip certain
options. In particular, we need to strip 'zfsutil' because if
we don't the mount utility will helpfull provide it to the
mount helper when we issue mount(8) again. This subverts the
check that the caller is zfs(8) and not mount(8).
Brian Behlendorf [Tue, 15 Mar 2011 16:34:56 +0000 (09:34 -0700)]
Always allow '-o remount,ro'
Allow the mount(8) utility to always operate on all datasets when
remounting them read-only. This critical for rc.sysinit/umountroot
which remounts the root filesystem read-only during shutdown to
ensure everything is correctly flushed to disk.
Fix minor typo, the check to set zfsutil should use the bitwise
'&'. I must have accidentally hit the adjacent '*' and obviously
neither the compiler or my code review caught this. Fix it now.
Brian Behlendorf [Thu, 10 Mar 2011 21:34:17 +0000 (13:34 -0800)]
Don't set I/O Scheduler for Partitions
ZFS should only change the i/o scheduler for a disk when it has
ownership of the whole disk. This is basically the same logic as
adjusting the write cache behavior on a disk. This change updates
the vdev disk code to skip partitions when setting the i/o scheduler.
Brian Behlendorf [Thu, 10 Mar 2011 20:58:44 +0000 (12:58 -0800)]
Check for trailing '/' in mount.zfs
When run with a root '/' cwd the mount.zfs helper would strip not
only the '/' but also the next character from the dataset name.
For example, '/tank' was changed to 'ank' instead of just 'tank'.
Originally, this was done for the '/tmp' cwd case where we needed
to strip the '/' following the cwd. For example '/tmp/tank' needed
to remove the '/tmp' cwd plus 1 character for the '/'.
This change fixes the problem by checking the cwd and if it ends in
a '/' it does not strip and extra character. Otherwise it will strip
the next character. I believe this should only ever be true for the
root directory.
Because we are dependent of the system mount/umount utilities to
ensure correct mtab locking, we should not suppress their error
output. During a successful mount/umount they will be silent,
but during a failure the error message they print is the only sure
way to know why a mount failed. This is because the (u)mount(8)
return code does not contain the result of the system call issued.
The only way to clearly idenify why thing failed is to rely on
the error message printed by the tool.
Longer term once libmount is available we can issue the mount/umount
system calls within the tool and still be ensured correct mtab locking.
Several issues related to strange mount/umount behavior were reported
and this commit should address most of them. The original idea was
to put in place a zfs mount helper (mount.zfs). This helper is used
to enforce 'legacy' mount behavior, and perform any extra mount argument
processing (selinux, zfsutil, etc). This helper wasn't ready for the
0.6.0-rc1 release but with this change it's functional but needs to
extensively tested.
This change addresses the following open issues.
Closes #101
Closes #107
Closes #113
Closes #115
Closes #119
Due to an uninitialized variable files opened with O_APPEND may
overwrite the start of the file rather than append to it. This
was introduced accidentally when I removed the Solaris vnodes.
The zfs_range_lock_writer() function used to key off zf->z_vnode
to determine if a znode_t was for a zvol of zpl object. With
the removal of vnodes this was replaced by the flag zp->z_is_zvol.
This flag was used to control the append behavior for range locks.
Unfortunately, this value was never properly initialized after
the vnode removal. However, because most of memory is usually
zeros it happened to be set correctly most of the time making
the bug appear racy. Properly initializing zp->z_is_zvol to
zero completely resolves the problem with O_APPEND.
Move 'bulk' and 'xattr_bulk' from the stack to the heap to minimize
stack space usage. These two arrays consumed 448 bytes on the stack
and have been replaced by two 8 byte points for a total stack space
saving of 432 bytes. The zfs_setattr() path had been previously
observed to overrun the stack in certain circumstances.
The original range lock implementation had to be modified by commit 8926ab7 because it was unsafe on Linux. In particular, calling
cv_destroy() immediately after cv_broadcast() is dangerous because
the waiters may still be asleep. Thus the following cv_destroy()
will free memory which may still be in use.
This was fixed by updating cv_destroy() to block on waiters but
this in turn introduced a deadlock. The deadlock was resolved
with the use of a taskq to move the offending free outside the
range lock. This worked well but using the taskq for the free
resulted in a serious performace hit. This is somewhat ironic
because at the time I felt using the taskq might improve things
by making the free asynchronous.
This patch refines the original fix and moves the free from the
taskq to a private free list. Then items which must be free'd
are simply inserted in to the list. When the range lock is dropped
it's safe to free the items. The list is walked and all rl_t
entries are freed.
This change improves small cached read performance by 26x. This
was expected because for small reads the number of locking calls
goes up significantly. More surprisingly this change significantly
improves large cache read performance. This probably attributable
to better cpu/memory locality. Very likely the same processor
which allocated the memory is now freeing it.
In the original implementation the zfs_open()/zfs_close() hooks
were dropped for simplicity. This was functional but not 100%
correct with the expected ZFS sematics. Updating and re-adding the
zfs_open()/zfs_close() hooks resolves the following issues.
1) The ZFS_APPENDONLY file attribute is once again honored. While
there are still no Linux tools to set/clear these attributes once
there are it should behave correctly.
2) Minimal virus scan file attribute hooks were added. Once again
this support in disabled but the infrastructure is back in place.
3) Most importantly correctly handle assigning files which were
opened syncronously to the intent log. Without this change O_SYNC
modifications could be lost during a system crash even though they
were marked synchronous.
Filesystems like ZFS must use what the kernel calls an anonymous super
block. Basically, this is just a filesystem which is not backed by a
single block device. Normally this block device's dev_t is stored in
the super block. For anonymous super blocks a unique reserved dev_t
is assigned as part of get_sb().
This sb->s_dev must then be set in the returned stat structures as
stat->st_dev. This allows userspace utilities to easily detect the
boundries of a specific filesystem. Tools such as 'du' depend on this
for proper accounting.
Additionally, under OpenSolaris the statfs->f_fsid is set to the device
id. To preserve consistency with OpenSolaris we also set the fsid to
the device id. Other Linux filesystem (ext) set the fsid to a unique
value determined by the filesystems uuid. This value is unique but
maintains no relationship to the device id. This may be desirable
when exporting NFS filesystem because it minimizes to chance of a
client observing the same fsid from two different servers.
Detect early on in configure if the Modules.symvers file is missing.
Without this file there will be build failures later and it's best
to catch this early and provide a useful error. In this case the
most likely problem is the kernel-devel packages are not installed.
It may also be possible that they are using an unbuilt custom kernel
in which case they must build the kernel first.
Until support is added for preemptible kernels detect this at
configure time and make it fatal. Otherwise, it is possible to
have a successful build and kernel modules with flakey behavior.
The libspl and libzpool libraries were missing from the libzfs
Makefile.am. They should be explicitly listed to avoid build
issues when compiling static libraries and binaries.
Additionally, ensure libzpool is built before libzfs because
libzfs is dependent on libzpool. This was also exposed as an
issue when forcing static linking.
The AT_ versions of these macros are used on Solaris and while they
map to their Linux equivilants the code has been updated to use the
ATTR_ versions.
Move 'tmpxvattr' from the stack to the heap to minimize stack
space usage. This is enough to get us below the 1024 byte stack
frame warning. That however is still a large stack frame and it
should be further reduced by moving the 'bulk' and 'xattr_bulk'
sa_bulk_attr_t variables to the heap in a future patch.
When I began work on the Posix layer it immediately became clear to
me that to integrate cleanly with the Linux VFS certain Solaris
specific things would have to go. One of these things was to elimate
as many Solaris specific types from the ZPL layer as possible. They
would be replaced with their Linux equivalents. This would not only
be good for performance, but for the general readability and health of
the code. The Solaris and Linux VFS are different beasts and should
be treated as such. Most of the code remains common for constructing
transactions and such, but there are subtle and important differenced
which need to be repsected.
This policy went quite for for certain types such as the vnode_t,
and it initially seemed to be working out well for the vattr_t. There
was a relatively small amount of related xvattr_t code I was forced to
comment out with HAVE_XVATTR. But it didn't look that hard to come
back soon and replace it all with a native Linux type.
However, after going doing this path with xvattr some distance it
clear that this code was woven in the ZPL more deeply than I thought.
In particular its hooks went very deep in to the ZPL replay code
and replacing it would not be as easy as I originally thought.
Rather than continue persuing replacing and removing this code I've
taken a step back and reevaluted things. This commit reverts many of
my previous commits which removed xvattr related code. It restores
much of the code to its original upstream state and now relies on
improved xvattr_t support in the zfs package itself.
The result of this is that much of the code which I had commented
out, which accidentally broke things like replay, is now back in
place and working. However, there may be a small performance
impact for getattr/setattr operations because they now require
a translation from native Linux to Solaris types. For now that's
a price I'm willing to pay. Once everything is completely functional
we can revisting the issue of removing the vattr_t/xvattr_t types.
With the removal of the minimal xvattr support from the spl this
support needs to be replaced in the zfs package. This is fairly
easily accomplished by directly adding portions of the sys/vnode.h
header from OpenSolaris. These xvattr additions have been placed
in the sys/xvattr.h header file and included as needed where simply
a sys/vnode.h was included before.
In additon to the xvattr types and helper macros two functions
were also included. The xva_init() and xva_getxoptattr() functions
were included as static inline functions in xvattr.h. They are
simple enough and it was simpler to place them here rather than
in their own .c file.
Darik Horn [Mon, 28 Feb 2011 16:15:05 +0000 (10:15 -0600)]
Add the zpool and filesystem versions
Print the supported zpool and filesystem versions at module load
time. This change removes an ambiguity and adds information that
system administrators care about. The phrase "ZFS pool version %s"
is the same as zpool upgrade -v so that the operator is familiar
with the message.
ZFS: Loaded module v0.6.0, ZFS pool version 28, ZFS filesystem version 5
ZFS: Unloaded module v0.6.0
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Fri, 25 Feb 2011 19:26:41 +0000 (11:26 -0800)]
Fix set block scheduler warnings
There were two cases when attempting to set the vdev block device
scheduler which would causes console warnings.
The first case was when the vdev used a loop, ram, dm, or other
such device which doesn't support a configurable scheduler. In
these cases attempting to set a scheduler is pointless and can
be safely skipped.
The secord case is slightly more troubling. We were seeing
transient cases where setting the elevator would return -EFAULT.
On retry everything is fine so there appears to be a small window
where this is possible. To handle that case we silently retry
up to three times before reporting the warning.
In all of the above cases the warning is harmless and at worse you
may see slightly different performance characteristics from one
or more of your vdevs.
Fajar A. Nugraha [Tue, 22 Feb 2011 10:58:44 +0000 (17:58 +0700)]
Use udev to create /dev/zvol/[dataset_name] links
This commit allows zvols with names longer than 32 characters, which
fixes issue on https://github.com/behlendorf/zfs/issues/#issue/102.
Changes include:
- use /dev/zd* device names for zvol, where * is the device minor
(include/sys/fs/zfs.h, module/zfs/zvol.c).
- add BLKZNAME ioctl to get dataset name from userland
(include/sys/fs/zfs.h, module/zfs/zvol.c, cmd/zvol_id).
- add udev rule to create /dev/zvol/[dataset_name] and the legacy
/dev/[dataset_name] symlink. For partitions on zvol, it will create
/dev/zvol/[dataset_name]-part* (etc/udev/rules.d/60-zvol.rules,
cmd/zvol_id).
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Brian Behlendorf [Wed, 23 Feb 2011 23:13:03 +0000 (15:13 -0800)]
Remove rdev packing
Remove custom code to pack/unpack dev_t's. Under Linux all dev_t's
are an unsigned 32-bit value even on 64-bit platforms. The lower
20 bits are used for the minor number and the upper 12 for the major
number.
This means if your importing a pool from Solaris you may get strange
major/minor numbers. But it doesn't really matter because even if
we add compatibility code to translate the encoded Solaris major/minor
they won't do you any good under Linux. You will still need to
recreate the dev_t with a major/minor which maps to reserved major
numbers used under Linux.
Dropping this code also resolves 32-bit builds by removing the
offending 32-bit compatibility code.
Brian Behlendorf [Wed, 23 Feb 2011 23:03:30 +0000 (15:03 -0800)]
Use correct ASSERT3* variant
ASSERT3P should be used instead of ASSERT3U when comparing
pointers. Using ASSERT3U with the cast causes a compiler
warning for 32-bit builds which is fatal with --enable-debug.
Brian Behlendorf [Wed, 23 Feb 2011 21:57:50 +0000 (13:57 -0800)]
Increase fragment size to block size
The underlying storage pool actually uses multiple block
size. Under Solaris frsize (fragment size) is reported as
the smallest block size we support, and bsize (block size)
as the filesystem's maximum block size. Unfortunately,
under Linux the fragment size and block size are often used
interchangeably. Thus we are forced to report both of them
as the filesystem's maximum block size.
Brian Behlendorf [Wed, 23 Feb 2011 21:03:19 +0000 (13:03 -0800)]
Fix 'statement with no effect' warning
Because the secpolicy_* macros are all currently defined to (0).
And because the caller of this function does not check the return
code. The compiler complains that this statement has no effect
which is correct and OK. To suppress the warning explictly cast
the result to (void).
Brian Behlendorf [Wed, 23 Feb 2011 20:50:05 +0000 (12:50 -0800)]
Fix enum compiler warning
Generally it's a good idea to use enums for switch statements,
but in this case it causes warning because the enum is really a
set of flags. These flags are OR'ed together in some cases
resulting in values which are not part of the original enum.
This causes compiler warning such as this about invalid cases.
error: case value ‘33’ not in enumerated type ‘zprop_source_t’
To handle this we simply case the enum to an int for the switch
statement. This leaves all other enum type checking in place
and effectively disabled these warnings.
Brian Behlendorf [Tue, 22 Feb 2011 22:55:35 +0000 (14:55 -0800)]
Linux 2.6.38 compat, blkdev_get_by_path()
The open_bdev_exclusive() function has been replaced (again) by the
more generic blkdev_get_by_path() function. Additionally, the
counterpart function close_bdev_exclusive() has been replaced by
blkdev_put(). Because these functions are more generic versions
of the functions they replaced the compatibility macro must add
the FMODE_EXCL mask to ensure they are exclusive.
Brian Behlendorf [Tue, 22 Feb 2011 20:15:13 +0000 (12:15 -0800)]
Linux 2.6.x compat, blkdev_compat.h
For legacy reasons the zvol.c and vdev_disk.c Linux compatibility
code ended up in sys/blkdev.h and sys/vdev_disk.h headers. While
there are worse places for this code to live it should be in a
linux/blkdev_compat.h header. This change moves this block device
Linux compatibility code in to the linux/blkdev_compat.h header
and updates all the correct #include locations. This is not a
functional change or bug fix, it is just code cleanup.
Brian Behlendorf [Fri, 18 Feb 2011 17:33:12 +0000 (09:33 -0800)]
Prep zfs-0.6.0-rc1 tag
Create the first 0.6.0 release candidate tag (rc1). The Posix
layer is now functional and passes fstest and several other
test suites cleanly. We now need this release candidate tag
to broaden the test coverage before we can release the official
zfs-0.6.0.
Brian Behlendorf [Thu, 17 Feb 2011 22:17:44 +0000 (14:17 -0800)]
Use provided uid/gid for setattr
When changing the uid/gid of a file via zfs_setattr() use the
Posix id passed in iattr->ia_uid/gid. While the zfs_fuid_create()
code already had the fuid support disabled for Linux it was
returning the uid/gid from the credential. With this change
the 'chown' command which relies on setxattr is now working
properly.
Also remove a little stray white space which was in front of
zfs_update_inode() call and the end of zfs_setattr().
Brian Behlendorf [Thu, 17 Feb 2011 19:26:51 +0000 (11:26 -0800)]
Fix symlink(2) inode reference count
Under Linux sys_symlink(2) should result in a inode being created
with one reference for the inode itself, and a second reference on
the inode which is held by the new dentry. Under Solaris this
appears not to be the case. Their zfs_symlink() handler drops
the inode reference before returning.
The result of this under Linux is that the reference count for
symlinks is always one smaller than it should have been. This
results in a BUG() when the symlink is unlinked. To handle this
the Linux port now keeps the inode reference which differs from
the Solaris behavior. This results in correct reference counts.
Brian Behlendorf [Thu, 17 Feb 2011 17:48:06 +0000 (09:48 -0800)]
Use -zfs_readlink() error
The zfs_readlink() function returns a Solaris positive error value
and that needs to be converted to a Linux negative error value.
While in this case nothing would actually go wrong, it's still
incorrect and should be fixed if for no other reason than clarity.
Brian Behlendorf [Thu, 17 Feb 2011 17:21:39 +0000 (09:21 -0800)]
Improve 'zpool import' safety
There are three improvements here to 'zpool import' proposed by Fajar
in Github issue #98. They are all good so I'm commiting all three.
1) Add descriptions for "hpet" and "core" blacklist entries.
2) Add "core" to the blacklist, as described in the issue accessing
this device will crash Xen dom0.
3) Refine probing behavior to use fstatat64(). This allows us to
determine if a device is a block device or a regular file without
having to open it. This is the safest appraoch when probing /dev/
because the simple act of opening a device may have unexpected
consequences.
Brian Behlendorf [Wed, 16 Feb 2011 23:54:55 +0000 (15:54 -0800)]
Fix readlink(2)
This patch addresses three issues related to symlinks.
1) Revert the zfs_follow_link() function to a modified version
of the original zfs_readlink(). The only changes from the
original OpenSolaris version relate to using Linux types.
For the moment this means no vnode's and no zfsvfs_t. The
caller zpl_follow_link() was also updated accordingly. This
change was reverted because it was slightly gratuitious.
2) Update zpl_follow_link() to use local variables for the
link buffer. I'd forgotten that iov.iov_base is updated by
uiomove() so after the call to zfs_readlink() it can not longer
be used. We need our own private copy of the link pointer.
3) Allocate MAXPATHLEN instead of MAXPATHLEN+1. By default
MAXPATHLEN is 4096 bytes which is a full page, adding one to
it pushes it slightly over a page. That means you'll likely
end up allocating 2 pages which is wasteful of memory and
possibly slightly slower.
This adds an API to wait for pending commit callbacks of already-synced
transactions to finish processing. This is needed by the DMU-OSD in
Lustre during device finalization when some callbacks may still not be
called, this leads to non-zero reference count errors. See lustre.org
bug 23931.
Brian Behlendorf [Wed, 16 Feb 2011 19:10:31 +0000 (11:10 -0800)]
Update 'zfs.sh -u' to umount all zfs filesystems
Before it is safe to unload the zfs module stack all mounted
zfs filesystems must be unmounted. If they are not unmounted,
there will be references held on the modules and the stack cannot
be removed. To handle this have 'zfs.sh -u' which is used by all
of the test scripts umount all zfs filesystem before attempting
to unload the module stack.
Brian Behlendorf [Wed, 16 Feb 2011 19:05:55 +0000 (11:05 -0800)]
Suppress share error on mount
Until code is added to support automatically sharing datasets
we should return success instead of failure. This prevents the
command line tools from returning a non-zero error code. While
a user likely won't notice this, test scripts like zconfig.sh
do and correctly fail because of it.
Brian Behlendorf [Wed, 16 Feb 2011 17:55:53 +0000 (09:55 -0800)]
Add get/setattr, get/setxattr hooks
While the attr/xattr hooks were already in place for regular
files this hooks can also apply to directories and special files.
While they aren't typically used in this way, it should be
supported. This patch registers these additional callbacks
for both directory and special inode types.
Brian Behlendorf [Wed, 16 Feb 2011 17:51:44 +0000 (09:51 -0800)]
Fix FIFO and socket handling
Under Linux when creating a fifo or socket type device in the ZFS
filesystem it's critical that the rdev is stored in a SA. This
was already being correctly done for character and block devices,
but that logic needed to be extended to include FIFOs and sockets.
This patch takes care of device creation but a follow on patch
may still be required to verify that the dev_t is being correctly
packed/unpacked from the SA.
Brian Behlendorf [Wed, 16 Feb 2011 17:40:29 +0000 (09:40 -0800)]
Create minors for all zvols
It was noticed that when you have zvols in multiple datasets
not all of the zvol devices are created at module load time.
Fajarnugraha did the leg work to identify that the root cause of
this bug is a non-zero return value from zvol_create_minors_cb().
Returning a non-zero value from the dmu_objset_find_spa() callback
function results in aborting processing the remaining children in
a dataset. Since we want to ensure that the callback in run on
all children regardless of error simply unconditionally return
zero from the zvol_create_minors_cb(). This callback function
is solely used for this purpose so surpressing the error is safe.
Brian Behlendorf [Fri, 11 Feb 2011 21:46:10 +0000 (13:46 -0800)]
Linux 2.6.36 compat, sops->evict_inode()
The new prefered inteface for evicting an inode from the inode cache
is the ->evict_inode() callback. It replaces both the ->delete_inode()
and ->clear_inode() callbacks which were previously used for this.
Brian Behlendorf [Fri, 11 Feb 2011 18:33:01 +0000 (10:33 -0800)]
Linux 2.6.33 compat, get/set xattr callbacks
The xattr handler prototypes were sanitized with the idea being that
the same handlers could be used for multiple methods. The result of
this was the inode type was changes to a dentry, and both the get()
and set() hooks had a handler_flags argument added. The list()
callback was similiarly effected but no autoconf check was added
because we do not use the list() callback.
Brian Behlendorf [Fri, 11 Feb 2011 16:58:55 +0000 (08:58 -0800)]
Linux 2.6.35 compat, fops->fsync()
The fsync() callback in the file_operations structure used to take
3 arguments. The callback now only takes 2 arguments because the
dentry argument was determined to be unused by all consumers. To
handle this a compatibility prototype was added to ensure the right
prototype is used. Our implementation never used the dentry argument
either so it's just a matter of using the right prototype.
Brian Behlendorf [Fri, 11 Feb 2011 00:16:52 +0000 (16:16 -0800)]
Linux 2.6.35 compat, const struct xattr_handler
The const keyword was added to the 'struct xattr_handler' in the
generic Linux super_block structure. To handle this we define an
appropriate xattr_handler_t typedef which can be used. This was
the preferred solution because it keeps the code clean and readable.
Brian Behlendorf [Thu, 10 Feb 2011 22:54:33 +0000 (14:54 -0800)]
Prefer /lib/modules/$(uname -r)/ links
Preferentially use the /lib/modules/$(uname -r)/source and
/lib/modules/$(uname -r)/build links. Only if neither of these
links exist fallback to alternate methods for deducing which
kernel to build with. This resolves the need to manually
specify --with-linux= and --with-linux-obj= on Debian systems.
Brian Behlendorf [Thu, 10 Feb 2011 20:11:05 +0000 (12:11 -0800)]
MS_DIRSYNC and MS_REC compat
It turns out that older versions of the glibc headers do not
properly define MS_DIRSYNC despite it being explicitly mentioned
in the man pages. They instead call it S_WRITE, so for system
where this is not correct defined map MS_DIRSYNC to S_WRITE.
At the time of this commit both Ubuntu Lucid, and Debian Squeeze
both use the out of date glibc headers.
As for MS_REC this field is also not available in the older headers.
Since there is no obvious mapping in this case we simply disable
the recursive mount option which used it.
Brian Behlendorf [Thu, 10 Feb 2011 18:57:29 +0000 (10:57 -0800)]
Add missing -ldl linker option
The inclusion on dlsym(), dlopen(), and dlclose() symbols require
us to link against the dl library. Be careful to add the flag to
both the libzfs library and the commands which depend on the library.
This file has gotten stale and needed to be updated. There are
individuals who deserve to be recognized for their contributions
to the project. I've done my best to assemble names from the
commit logs of those who have submitted patches. This list may
not be comprehensive, if you feel I've overlooked your contribution
please let me know and we can get your name added.
Initial testing has shown the the right IO scheduler to use under Linux
is noop. This strikes the ideal balance by allowing the zfs elevator
to do all request ordering and prioritization. While allowing the
Linux elevator to do the maximum front/back merging allowed by the
physical device. This yields the largest possible requests for the
device with the lowest total overhead.
While 'noop' should be right for your system you can choose a different
IO scheduler with the 'zfs_vdev_scheduler' option. You may set this
value to any of the standard Linux schedulers: noop, cfq, deadline,
anticipatory. In addition, if you choose 'none' zfs will not attempt
to change the IO scheduler for the block device.
The following warning was observed under normal operation. It's
not fatal but it's something to be addressed long term. Flag the
offending allocation with KM_NODEBUG to suppress the warning and
flag the call site.
When performing a 'zfs rollback' it's critical to invalidate
the previous dcache and inode cache. If we don't there will
stale cache entries which when accessed will result in EIOs.
These two warnings in libefi serve no real purpose. When running
without DEBUG they are already supressed, and even when DEBUG is
enabled all they indicate is the device doesn't already have an
EFI label. For a Linux machine this is probably the common case.
With the recent SPL change (d599e4fa) that forces cv_destroy()
to block until all waiters have been woken. It is now unsafe
to call cv_destroy() under the zp->z_range_lock() because it
is used as the condition variable mutex. If there are waiters
cv_destroy() will block until they wake up and aquire the mutex.
However, they will never aquire the mutex because cv_destroy()
will not return allowing it's caller to drop the lock. Deadlock.
To avoid this cv_destroy() is now run asynchronously in a taskq.
This solves two problems:
1) It is no longer run under the zp->z_range_lock so no deadlock.
2) Since cv_destroy() may now block we don't want this slowing
down zfs_range_unlock() and throttling the system.
This was not as much of an issue under OpenSolaris because their
cv_destroy() implementation does not do anything. They do however
risk a bad paging request if cv_destroy() returns, the memory holding
the condition variable is free'd, and then the waiters wake up and
try to reference it. It's a very small unlikely race, but it is
possible.
It's worth taking a moment to describe how mmap is implemented
for zfs because it differs considerably from other Linux filesystems.
However, this issue is handled the same way under OpenSolaris.
The issue is that by design zfs bypasses the Linux page cache and
leaves all caching up to the ARC. This has been shown to work
well for the common read(2)/write(2) case. However, mmap(2)
is problem because it relies on being tightly integrated with the
page cache. To handle this we cache mmap'ed files twice, once in
the ARC and a second time in the page cache. The code is careful
to keep both copies synchronized.
When a file with an mmap'ed region is written to using write(2)
both the data in the ARC and existing pages in the page cache
are updated. For a read(2) data will be read first from the page
cache then the ARC if needed. Neither a write(2) or read(2) will
will ever result in new pages being added to the page cache.
New pages are added to the page cache only via .readpage() which
is called when the vfs needs to read a page off disk to back the
virtual memory region. These pages may be modified without
notifying the ARC and will be written out periodically via
.writepage(). This will occur due to either a sync or the usual
page aging behavior. Note because a read(2) of a mmap'ed file
will always check the page cache first even when the ARC is out
of date correct data will still be returned.
While this implementation ensures correct behavior it does have
have some drawbacks. The most obvious of which is that it
increases the required memory footprint when access mmap'ed
files. It also adds additional complexity to the code keeping
both caches synchronized.
Longer term it may be possible to cleanly resolve this wart by
mapping page cache pages directly on to the ARC buffers. The
Linux address space operations are flexible enough to allow
selection of which pages back a particular index. The trick
would be working out the details of which subsystem is in
charge, the ARC, the page cache, or both. It may also prove
helpful to move the ARC buffers to a scatter-gather lists
rather than a vmalloc'ed region.
Additionally, zfs_write/read_common() were used in the readpage
and writepage hooks because it was fairly easy. However, it
would be better to update zfs_fillpage and zfs_putapage to be
Linux friendly and use them instead.