Michael Meffie [Wed, 29 Apr 2015 16:00:24 +0000 (12:00 -0400)]
afs: add afsd -inumcalc option
This commit adds the afsd -inumcalc command line switch to specify the
inode number calculation method in a platform neutral way.
Inode numbers reported for files within the AFS filesystem are generated
by the cache manager using a calculation which derives a number from a
FID. Long ago, a new type of calculation was added which generates inode
numbers using a MD5 message digest of the FID. The MD5 inode number
calculation variant is computationally more expensive but greatly
reduces the chances for inode number collisions.
The MD5 calculation can be enabled on the Linux cache manager using the
Linux sysctl interface. Other than the sysctl method of selecting the
inode calculation type, the MD5 inode number calculation method is not
specific to Linux.
This change introduces a command-line option which accepts a value to
indicate the calculation method, instead of a simple flag to enable MD5
inode numbers. This should allow for new inode calculation methods
in the future without the need for additional afsd command-line flags.
Two values are currently accepted for -inumcalc. The value of 'compat'
specifies the legacy inode number calculation. The value 'md5' indicates
that the new MD5 calculation is to be used.
Reviewed-on: https://gerrit.openafs.org/11855 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 0028ea92ad3e7aac6a4c51f63703a4d9d7b9dcd6)
Change-Id: I9021eea9f64c754157061d039f63b6f744ec2ec5
Reviewed-on: https://gerrit.openafs.org/12608 Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de> Tested-by: Stephan Wiesand <stephan.wiesand@desy.de>
when processing "fs sysname" on a client, a rmtsys-related
checks are executed by default. These prevent a user with gid
2748 and 2750 (0xabc and 0xabe) from executing this command.
Add a new flag inside the cachemanager for the rmtsys-
functionality. This flag is set through a new ioctl by the afsd
on startup.
When the OpenAFS client is unmounted on DARWIN, the blocks of packets
allocated by RX are released. Historically, the memory used by those
packets was never properly released.
As we can see, ‘rx_mallocedP’ is a global pointer that stores the
first address of the last allocated block of packets. As a result, when
‘rxi_FreeAllPackets’ is called, only the last block is released.
However, 230dcebcd61064cc9aab6d20d34ff866a5c575ea moved the global
pointer in question to the end of the last block. As a result, when the
OpenAFS client is unmounted on DARWIN, the ‘rxi_FreeAllPackets’
function releases the wrong block of memory. This problem was exposed
on OS X 10.12 Sierra where the system crashes when the OpenAFS client
is unmounted.
To fix this problem, store the address of every single block of packets
in a queue and release one by one when the OpenAFS client is unmounted.
Reviewed-on: https://gerrit.openafs.org/12427 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 5b28061fb593f5f48df549b07f0ccd848348b93c)
Change-Id: Id8606b1c1444861df69ed4af8169e343964a691d
Reviewed-on: https://gerrit.openafs.org/12602 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Marcio Barbosa [Tue, 31 Jan 2017 14:43:18 +0000 (11:43 -0300)]
vol: detach offline volumes on dafs
Taking a volume offline always clears the inService bit. Taking a
volume out of service also takes it offline. Therefore, if the
inService flag is false, the volume in question should be offline.
On dafs, an offline volume should be unattached.
The attach2() function does not change the state of the volume received
as an argument to unattached when the inService flag is false. Instead,
this function changes the state of the volume in question to
pre-attached and returns VNOVOL to the client. As result, subsequent
accesses to this volume will make the server try and fail to attach
this offline volume over and over again, writing to the FileLog each
time.
To fix this problem, detach the volume received as an argument if the
inService flag is false. Since the new state of this volume will be
unattached, subsequent accesses will not hit attach2().
This situation where a volume is not offline but is also not in service
can occur if a volume is taken offline with vos offline and some time
later the DAFS fileserver is shutdown and restarted; the volume is
placed into the preattach state by default when the server restarts.
Each access to the volume by clients then causes the fileserver to
attempt to attach the volume, which fails, since the in-service flag in
the volume header is false from the previous vos offline. The
fileserver will log a warning to the FileLog on each attempt to attach
the volume, and this will fill the FileLog with duplicate messages
corresponding to the number of attempted accesses.
Reviewed-on: https://gerrit.openafs.org/12515 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 2421da2bf327525216ec7e79b9aa81fa2c4f77d5)
Change-Id: I95cffb6a91797341d9202cbbef3b205c11348d5e
Reviewed-on: https://gerrit.openafs.org/12569 Reviewed-by: Marcio Brito Barbosa <mbarbosa@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Tested-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Mark Vitale [Tue, 21 Feb 2017 01:16:47 +0000 (20:16 -0500)]
DAFS: do not save or restore host state if CPS in progress
If a fileserver is shutdown while one or more PR_GetHostCPS calls
are in progress, this state is saved in the fsstate.dat file as
hostFlags HCPS_WAITING, HCPS_INPROGRESS. Other hosts that are
merely waiting will have HCPS_WAITING recorded.
However, it makes no sense to restore host structs in this state,
because the GetCPS calls will no longer be in progress. Once these
hosts become active, they will block server threads and quickly cause
all server threads to be exhausted as other CPS requests are blocked
behind them.
Instead, exclude these states from both save and restore.
Reviewed-on: https://gerrit.openafs.org/12561 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com>
(cherry picked from commit 38a3f51fb8b3910ecdd7cacb06f35ec681990aea)
Change-Id: I0e02543fd2e547fcc9f95db0973f09e5951a1da1
Reviewed-on: https://gerrit.openafs.org/12568 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Michael Meffie [Fri, 7 Apr 2017 02:50:41 +0000 (22:50 -0400)]
redhat: update rpm spec file
Update the spec file to keep up with accumulated changes.
* Correct installation location of db check programs.
* Install afsd to the legacy location to avoid breaking
init scrips and systemd configs.
* Exclude yet another duplicated copy of kpwvalid.
* libubik_pthread.a is gone.
* Install the kpwvalid man page.
* Continue to remove the obsolete kdb program.
* Update the names of the pam_afs symlinks.
* Add libkopenafs to authlibs.
* Package dafssync-debug man pages.
* Package opr/queue.h in devel.
* Package akeyconvert and man page.
* Do not package fuse version of afsd. A separate sub-package
for afsd.fuse is warrented, since it adds new libfuse
dependencies.
* Package new server man pages, including dafsssync-* pages.
* Package libafsrfc3961.a as a devel lib.
* Continue to package kauth programs.
Change-Id: I875c3b8dee53abbc67b0f05f8b291bb58abf41a5
Reviewed-on: https://gerrit.openafs.org/12595 Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
Tim Creech [Sun, 5 Mar 2017 23:13:45 +0000 (18:13 -0500)]
FBSD: build fix for FreeBSD 11
r285819 eliminated b_saveaddr from struct buf, while r292373 changed the
arguments to VOP_GETPAGES. The approach used by this patch to address
these changes was inspired by FreeBSD's nfs and samba clients.
Michael Meffie [Wed, 5 Apr 2017 20:48:36 +0000 (16:48 -0400)]
redhat: convert rpm spec file to make install
Convert the build and install from the deprecated 'make dest' to the
modern 'make install' method.
* Clarify the install section by unrolling the shell scripts,
reorganizing, and improving the comments.
* Remove the gzip glob of the man pages; rpmbuild automatically
compresses the man pages and will handle symlinks correctly.
* Remove the generated temporary list file and specify files directly.
* Remove the extra tar commands to install the man pages out of the doc
directory; 'make install..' installs the man pagess.
* Remove code in the install section which determines the sysname. This is
no longer needed during the install.
* Update the kernel module install commands to accommodate the
conversion from 'make dest'.
Change-Id: I97ec80185a2b11704b27ea74941b50ff4a5aca8c
Reviewed-on: https://gerrit.openafs.org/12594 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
Mark Vitale [Tue, 12 Jan 2016 23:06:51 +0000 (18:06 -0500)]
afs: fs getcacheparms miscounts dcaches for large files
fs getcacheparms issued with the -excessive option tabulates in-memory
dcaches ("DCentries") by size. However, any dcache with validPos > 2^31
is miscounted in the 4k-16k bucket. This is caused by a type mismatch
between 'validPos' (afs_size_t) and 'size' (int) which leads to a
negative value for size by sign-extension. The size comparison "sieve"
fails for negative numbers; it skips the first bucket (0-4K) and dumps
them in the second one (4k-16k).
Move the declaration of 'size' closer to its use, and declare it with
the same type as 'validPos' (afs_size_t) so the comparison sieve
correctly places these dcaches in the last (>=1M) bucket.
Reviewed-on: https://gerrit.openafs.org/12347 Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit b5e4e8c14130f601bbf43dee5927222ebf7613fa)
Change-Id: I659fd86f05b29c1eac1a262d340bcc1ce2640797
Reviewed-on: https://gerrit.openafs.org/12605 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Mark Vitale [Sat, 7 Jan 2017 11:22:47 +0000 (06:22 -0500)]
LINUX: eliminate unused variable warning
Commit c3bbf0b4444db88192eea4580ac9e9ca3de0d286 added routine
osi_TryEvictDentries and included new logic for D_INVALIDATE_IS_VOID.
Unfortunately, this new code path no longer uses dentry; it also should
have been made conditional at that time.
Wrap the declaration of dentry in #ifndef D_INVALIDATE_IS_VOID to
eliminate the unused variable warning.
Reviewed-on: https://gerrit.openafs.org/12505 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 19599b5ef5f7dff2741e13974692fe4a84721b59)
Change-Id: Ic15df733fcbccfaf9870ecd335bb2d549ab0d43d
Reviewed-on: https://gerrit.openafs.org/12513 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Michael Meffie [Thu, 27 Aug 2015 17:06:05 +0000 (13:06 -0400)]
afs: shake harder in shake-loose-vcaches
Linux based cache managers will allocate vcaches on demand and
deallocate batches of vcaches in the background. This feature is called
dynamic vcaches.
Vcaches to be deallocated are found by traversing the vcache LRU list
(VLRU) from the oldest vcache to the newest. Up to a target number of
vcaches are attempted to be evicted. The afs_xvcache lock protecting
the VLRU may be dropped and re-acquired while attempting to evict a
vcache. When this happens, it is possible the VLRU may have changed, so
the traversal of the VLRU is restarted. This restarting of the VLRU
transversal is limited to 100 iterations to avoid looping indefinitely.
Vcaches which are busy cannot be evicted and remain in the VLRU. When a
busy cache was not evicted and the afs_xvache lock was dropped, the VLRU
traversal is restarted from the end of the VLRU. When the busy vcache is
encountered on the retry, it will trigger additional retries until the
loop limit is reached, at which point the target number of vcaches will
not be deallocated.
This can leave a very large number of unbusy vcaches which are never
deallocated. On a busy machine, tens of millions of unused vcaches can
remain in memory. When the busy vcache at the end of the VLRU is finally
evicted, the log jam is broken, and the background deamon will hold the
afs_xvcache lock for an excessively long time, hanging the system.
Fix this by moving busy vcaches to the head of the VLRU before
restarting the VLRU traversal. These busy vcaches will be skipped when
retrying the VLRU traversal, allowing the cache manager to make progress
deallocating vcaches down to the target level.
This was already done on the mac osx platform while attempting to evict
vcaches. Move the code to move busy vcaches to the head of the VLRU up
the the platform agnostic caller.
Thanks to Andrew Deason for the initial version of this patch.
Reviewed-on: https://gerrit.openafs.org/11654 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Andrew Deason <adeason@dson.org> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 5c136c7d93ed97166f39bf716cc7f5d579b70677)
Change-Id: If60b1889d012a739aa5b43e842abb80a6ebfdb6a
Reviewed-on: https://gerrit.openafs.org/12451 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Mark Vitale [Thu, 4 Aug 2016 22:42:27 +0000 (18:42 -0400)]
LINUX: do not use d_invalidate to evict dentries
When working within the AFS filespace, commands which access large
numbers of OpenAFS files (e.g., git operations and builds) may result in
active files (e.g., the current working directory) being evicted from the
dentry cache. One symptom of this is the following message upon return
to the shell prompt:
"fatal: unable to get current working directory: No such file or
directory"
Starting with Linux 3.18, d_invalidate returns void because it always
succeeds. Commit a42f01d5ebb13da575b3123800ee6990743155ab adapted
OpenAFS to cope with the new return type, but not with the changed
semantics of d_invalidate. Because d_invalidate can no longer fail with
-EBUSY when invoked on an in-use dentry. OpenAFS must no longer trust it
to preserve in-use dentries.
Modify the dentry eviction code to use a method (d_prune_aliases) that
does not evict in-use dentries.
Reviewed-on: https://gerrit.openafs.org/12363 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit c3bbf0b4444db88192eea4580ac9e9ca3de0d286)
Change-Id: Ic72a280f136cc414b54d4b8ec280f225290df122
Reviewed-on: https://gerrit.openafs.org/12450 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Mark Vitale [Thu, 4 Aug 2016 22:18:15 +0000 (18:18 -0400)]
LINUX: split dentry eviction from osi_TryEvictVCache
To make osi_TryEvictVCache clearer, and to prepare for a future change
in dentry eviction, split the dentry eviction logic into its own routine
osi_TryEvictDentries.
No functional difference should be incurred by this commit.
Reviewed-on: https://gerrit.openafs.org/12362 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Joe Gorse <jhgorse@gmail.com>
(cherry picked from commit 742643e306929ac979ab69515a33ee2a3f2fa3fa)
Change-Id: I750fc7606ca56e784a60bdbc13a32d21fe307429
Reviewed-on: https://gerrit.openafs.org/12448 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Modify the helpfile to describe the actual restrictions imposed by
-restricted mode.
Reviewed-on: https://gerrit.openafs.org/12454 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 3af0460a4a6d7bf22e1789fd9e375659e20c3a55)
Change-Id: Ifa544c322e67da712a0bc96b3797e51786e4d399
Reviewed-on: https://gerrit.openafs.org/12476 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Stephan Wiesand [Tue, 11 Apr 2017 09:58:55 +0000 (11:58 +0200)]
Linux: only include cred.h if it exists
Commit c89fd17df1032ec2eacc0d0c9b73e19c5e8db7d2 introduced an explicit
include of linux/cred.h since the latest kernel no longer includes it
implicitly in sched.h. Alas, older kernels (like 2.6.18) don't have this
file. Add a configure test for the existence of cred.h and only include
it if actually present.
Reviewed-on: https://gerrit.openafs.org/12593 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com>
(cherry picked from commit 6b7b4239ab22fbb301e3b50e2ca4072445ba4e9e)
Change-Id: I64970ba471180d32fa5af5445e7604bbe8511b32
Reviewed-on: https://gerrit.openafs.org/12598 Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Several components of libafs which require cred.h were picking it by
including sched.h.
Instead, explicitly add an include for cred.h. cred.h begins with a
customary one-shot to prevent multiple loads:
#ifndef _LINUX_CRED_H
#define _LINUX_CRED_H
Therefore we don't need a new autoconf test or preprocessor conditional
to prevent redundant includes on older Linux releases.
Reviewed-on: https://gerrit.openafs.org/12574 Tested-by: Mark Vitale <mvitale@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Joe Gorse <jhgorse@gmail.com> Tested-by: Joe Gorse <jhgorse@gmail.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
(cherry picked from commit c89fd17df1032ec2eacc0d0c9b73e19c5e8db7d2)
Change-Id: I235a6272c55a8f734be07b578bbb1a324cf34e2e
Reviewed-on: https://gerrit.openafs.org/12590 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
This breaks existing OpenAFS autoconf tests for recalc_sigpending() and
task_struct.signal->rlim, so that the OpenAFS kernel module can no
longer build.
Modify OpenAFS autoconfig tests to cope.
Reviewed-on: https://gerrit.openafs.org/12573 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Joe Gorse <jhgorse@gmail.com> Tested-by: Joe Gorse <jhgorse@gmail.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
(cherry picked from commit ad001550949b612ff6b4899fa8da50ee58f87533)
Change-Id: I491208d77e45d45cc0089b8033892a6408da431c
Reviewed-on: https://gerrit.openafs.org/12589 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
statx: Add a system call to make enhanced file info available
The Linux getattr inode operation is altered to take two additional
arguments: a u32 request_mask and an unsigned int flags that indicate
the synchronisation mode. This change is propagated to the
vfs_getattr*() function.
- int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
+ int (*getattr) (const struct path *, struct kstat *,
+ u32 request_mask, unsigned int sync_mode);
The first argument, request_mask, indicates which fields of the statx
structure are of interest to the userland call. The second argument,
flags, currently may take the values defined in
include/uapi/linux/fcntl.h and are optionally used for cache coherence:
(1) AT_STATX_SYNC_AS_STAT tells statx() to behave as stat() does.
(2) AT_STATX_FORCE_SYNC will require a network filesystem to
synchronise its attributes with the server - which might require
data writeback to occur to get the timestamps correct.
(3) AT_STATX_DONT_SYNC will suppress synchronisation with the server in
a network filesystem. The resulting values should be considered
approximate.
This patch provides a new autoconf test and conditional compilation to
cope with the changes in our getattr implementation.
Reviewed-on: https://gerrit.openafs.org/12572 Reviewed-by: Joe Gorse <jhgorse@gmail.com> Tested-by: Joe Gorse <jhgorse@gmail.com> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit de5ee1a67d1c3284d65dc69bbbf89664af70b357)
Change-Id: I41ff134e1e71944f0629c9837d38cfbc495264c8
Reviewed-on: https://gerrit.openafs.org/12588 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Stephan Wiesand [Tue, 11 Apr 2017 09:58:55 +0000 (11:58 +0200)]
Linux: only include cred.h if it exists
Commit c89fd17df1032ec2eacc0d0c9b73e19c5e8db7d2 introduced an explicit
include of linux/cred.h since the latest kernel no longer includes it
implicitly in sched.h. Alas, older kernels (like 2.6.18) don't have this
file. Add a configure test for the existence of cred.h and only include
it if actually present.
Jonathon Weiss [Thu, 10 Nov 2016 22:06:18 +0000 (17:06 -0500)]
Prevent double-starting client on RHEL7
On RHEL7 if the AFS client is stopped with 'service openafs-client
stop', but that fails for some reason (most commonly because some
process has a file or directory in AFS open) systemd will decide that
the openafs-client is in a failed state when it is actually running.
If one then runs 'service openafs-client start' systemd will start a
new AFS client. At this point AFS access will continue to work until
the functional AFS client is (successfully) stopped, at which point a
reboot is required to recover.
Have systemd check the status of 'fs sysname' before starting the
AFS client, so we avoid getting into a state that requires a reboot.
Reviewed-on: https://gerrit.openafs.org/12443 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit c666bfee8848183ccbc566c9e0fa019088e56505)
This breaks existing OpenAFS autoconf tests for recalc_sigpending() and
task_struct.signal->rlim, so that the OpenAFS kernel module can no
longer build.
Modify OpenAFS autoconfig tests to cope.
Change-Id: Ic9f174b92704eabcbd374feffe5fbeb92c8987ce
Reviewed-on: https://gerrit.openafs.org/12573 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Joe Gorse <jhgorse@gmail.com> Tested-by: Joe Gorse <jhgorse@gmail.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
statx: Add a system call to make enhanced file info available
The Linux getattr inode operation is altered to take two additional
arguments: a u32 request_mask and an unsigned int flags that indicate
the synchronisation mode. This change is propagated to the
vfs_getattr*() function.
- int (*getattr) (struct vfsmount *, struct dentry *, struct kstat *);
+ int (*getattr) (const struct path *, struct kstat *,
+ u32 request_mask, unsigned int sync_mode);
The first argument, request_mask, indicates which fields of the statx
structure are of interest to the userland call. The second argument,
flags, currently may take the values defined in
include/uapi/linux/fcntl.h and are optionally used for cache coherence:
(1) AT_STATX_SYNC_AS_STAT tells statx() to behave as stat() does.
(2) AT_STATX_FORCE_SYNC will require a network filesystem to
synchronise its attributes with the server - which might require
data writeback to occur to get the timestamps correct.
(3) AT_STATX_DONT_SYNC will suppress synchronisation with the server in
a network filesystem. The resulting values should be considered
approximate.
This patch provides a new autoconf test and conditional compilation to
cope with the changes in our getattr implementation.
Change-Id: Ie4206140ae249c00a8906331c57da359c4a372c4
Reviewed-on: https://gerrit.openafs.org/12572 Reviewed-by: Joe Gorse <jhgorse@gmail.com> Tested-by: Joe Gorse <jhgorse@gmail.com> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Jonathon Weiss [Thu, 10 Nov 2016 22:06:18 +0000 (17:06 -0500)]
Prevent double-starting client on RHEL7
On RHEL7 if the AFS client is stopped with 'service openafs-client
stop', but that fails for some reason (most commonly because some
process has a file or directory in AFS open) systemd will decide that
the openafs-client is in a failed state when it is actually running.
If one then runs 'service openafs-client start' systemd will start a
new AFS client. At this point AFS access will continue to work until
the functional AFS client is (successfully) stopped, at which point a
reboot is required to recover.
Have systemd check the status of 'fs sysname' before starting the
AFS client, so we avoid getting into a state that requires a reboot.
Change-Id: I28a5cca746823d69183ea5ce65c10e1725009c5c
Reviewed-on: https://gerrit.openafs.org/12443 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Stephan Wiesand [Wed, 1 Mar 2017 12:57:41 +0000 (13:57 +0100)]
Make OpenAFS 1.6.20.2
Update configure version strings for 1.6.20.2. Note that macOS kext
can be of form XXXX.YY[.ZZ[(d|a|b|fc)NNN]] where d dev, a alpha,
b beta, f final candidate so we have no way to represent 1.6.20.2.
Switch to 1.6.21 dev 2 for macOS.
Change-Id: Iccc613ea6b7d1194e7a1b20fa38c54b192c3c7b4
Reviewed-on: https://gerrit.openafs.org/12532 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Benjamin Kaduk [Tue, 21 Feb 2017 04:18:09 +0000 (22:18 -0600)]
XBSD: do not claim AFS_VM_RDWR_ENV
The AFS_VM_RDWR_ENV configuration parameter (defined or not defined
in each platform's param.h) is undocumented, but appears to be an
indication of a property of the platform OS's VFS layer, or perhaps just
of the complexity of the read/write vnops that we implement for it.
That is, AFS_VM_RDWR_ENV is defined when the read/write vnops implement
partial write logic (and presumably when they interact with the OS VM
layer in ways not expressed by the afs_write() abstraction); systems
that do not define AFS_VM_RDWR_ENV can use the afs_write() function
fairly directly as the vnode operation. Use of AFS_VM_RDWR_ENV
evolved over time, with the original (AFS 3.2/3.3-era) code using a
simple scheme that handled partial writes directly in afs_write()
and avoided complexity in callers. In AFS 3.4, sunos and solaris
gained a more complicated read/write vnop that incorporated the
afs_DoPartialWrite() call itself, and eventually in 3.6 we see the
behavior established at the original IBM import, with all the (Unix)
OSes supported at that time defining AFS_VM_RDWR_ENV.
When DARWIN support was brought in in commit a41175cfbbf4d06ccfe14ae54bef8b7464ecd80b, its param.h properly did
not define AFS_VM_RDWR_ENV, as OS X uses a VFS interface that shares
some level of abstraction with the traditional BSD VFS and its
read/write/getpages/putpages operations, so the afs_write() behavior
was natural and no extra complications needed for integration with the
VM layer or other optimizations.
However, when the initial FreeBSD support came in a few months later,
it seems to have taken inspiration from the OSes that were supported
in the initial IBM import, and kept the AFS_VM_RDWR_ENV definition.
This was then propagated to all the later BSDs as they were added.
Perhaps the most noticeable consequence of this definition is that
the calls to afs_DoPartialWrite() from afs_write() are bypassed, with
a comment that "[i]f write is implemented via VM, afs_DoPartialWrite()
is called from the high-level write op" (and calls to afs_FakeOpen()
and afs_FakeClose() are similarly skipped). This means that attempting
to write a file larger than the local cache will hang waiting for
more space to be freed, which will never happen as the vcache remains
locked and will not be written out in the background.
In the absence of a documented meaning for AFS_VM_RDWR_ENV, this
also gives us a proxy that can be used to indicate whether a given
OS's support intended to claim the AFS_VM_RDWR_ENV -- such platforms
will actually contain the call to afs_DoPartialWrite() in the
appropriate vnode operation. This can be used to sanity-check the
places where AFS_VM_RDWR_ENV is removed by this commit. Interestingly,
HP-UX does not call afs_DoPartialWrite() but also is clearly in a VFS
that uses a rdwr()-based approach, as the corresponding vnode operation
is implemented by mp_afs_rdwr(), so leave it unchanged for now.
Tim Creech is responsible for noting the lack of calls to
afs_DoPartialWrite() on FreeBSD, and Chaskiel Grundman for the
historical research into pre-OpenAFS AFS behavior.
Designing and implementing more complicated BSD read/write vnops that
incorporate afs_DoPartialWrite() and other improvements is left for
future work.
Change-Id: I8e89855ac31303934f97d0753b64899fb7e3867c
Reviewed-on: https://gerrit.openafs.org/12520 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Antoine Verheijen <apv@ualberta.ca> Reviewed-by: Tim Creech <tcreech@tcreech.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Mark Vitale [Tue, 28 Feb 2017 23:02:39 +0000 (18:02 -0500)]
SOLARIS: prevent BAD TRAP panic with Studio 12.5
Starting with Solaris Studio 12.3, it is documented that Solaris kernel
modules (such as libafs) must not use any floating point, vector, or
SIMD/SSE instructions on x86 hardware. However, each new Studio
compiler release (12.4 and especially 12.5) is more likely to use these
types of instructions by default.
If the libafs kernel module includes any forbidden kernel instructions,
Solaris will panic the system with:
BAD TRAP: type=7 (#nm Device not available)
Provide a new autoconfig test to specify the required compiler options
(-xvector=%none -xregs=no%float) when building the OpenAFS kernel module
for Solaris, so that no invalid x86 instructions are used.
In addition, reinstate default kernel module optimization for Solaris.
It had been disabled in commit 80592c53cbb0bce782eb39a5e64860786654be9f
to address this same issue in Studio 12.3 and 12.4. However, Studio
12.5 started using some SSE instructions even with no optimization.
This commit has been tested with OpenAFS master and Studio 12.5 at all
optimization levels (none, -xO1 through -xO5) and verified to contain no
XMM register instructions via the following command:
$ gobjdump -dlr libafs64.o | grep xmm | wc -l
[wiesand: limit change to solaris 5.11 for stable branch]
Reviewed-on: https://gerrit.openafs.org/12558 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 22d841a45fff7026318b529a41dd957ce8bb0ddf)
Change-Id: I2e87f26dbac47289694346639b396dfc556368f4
Reviewed-on: https://gerrit.openafs.org/12567 Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Michael Meffie [Wed, 8 Mar 2017 16:48:14 +0000 (11:48 -0500)]
linux24: remove unused NUMPAGGROUPS define
Remove the unused NUMPAGGROUPS define in the pag group handling
implementation for linux24. PAGS always take two group ids in linux24,
so the NUMPAGGROUPS define was not used in linux24. Remove the unused
constant.
This is a 1.6.x only change, since linux24 support has been removed on
the master branch.
Change-Id: I10d4d5744420b075a10deecf052d2ecc128ad8fe
Reviewed-on: https://gerrit.openafs.org/12563 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Andrew Deason [Sat, 8 Aug 2015 21:49:50 +0000 (16:49 -0500)]
SOLARIS: Use AFS_PAG_ONEGROUP_ENV for Solaris 11
On Solaris 11 (specifically, Solaris 11.1+), the supplemental group
list for a process is supposed to be sorted. Starting with Solaris
11.2, more authorization checks are done that assume the list is
sorted (e.g., to do a binary search), so having them out of order
can cause incorrect behavior. For example:
$ echo foo > /tmp/testfile
$ chmod 660 /tmp/testfile
$ sudo chown root:daemon /tmp/testfile
$ cat /tmp/testfile
foo
$ id -a
uid=100(adeason) gid=10(staff) groups=10(staff),12(daemon),20(games),21(ftp),50(gdm),60(xvm),90(postgres)
$ pagsh
$ cat /tmp/testfile
cat: cannot open /tmp/testfile: Permission denied
$ id -a
uid=100(adeason) gid=10(staff) groups=33536,32514,10(staff),12(daemon),20(games),21(ftp),50(gdm),60(xvm),90(postgres)
Solaris sorts the groups given to crsetgroups() on versions which
required the group ids to be sorted, but we currently manually put our
PAG groups in our own order in afs_setgroups(). This is currently
required, since various places in the code assume that PAG groups are
the first two groups in a process's group list.
To get around this, do not require the PAG gids to be the first two
gids anymore. To more easily identify PAG gids in group processes, use
a single gid instead of two gids to identify a PAG, like modern Linux
currently uses (under the AFS_PAG_ONEGROUP_ENV). High-numbered groups
have been possible for quite a long time on Solaris, allegedly further
back than Solaris 8. Only do this for Solaris 11, though, to reduce
the platforms we affect.
[mmeffie@sinenomine.net: Define AFS_PAG_ONEGROUP_ENV in param.h.]
Reviewed-on: https://gerrit.openafs.org/11979 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com>
(cherry picked from commit aab1e71628e6a4ce68c5e59e2f815867438280d1)
Change-Id: I54c1f4c1be4eed1804293aebae795b165954a3a4
Reviewed-on: https://gerrit.openafs.org/12526 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Andrew Deason [Sat, 8 Aug 2015 21:13:54 +0000 (16:13 -0500)]
afs: Make ONEGROUP_ENV not Linux-specific
The functionality in AFS_LINUX26_ONEGROUP_ENV does not really need to
be Linux-specific (it's just only implemented for Linux right now).
Rename it to AFS_PAG_ONEGROUP_ENV, and remove some Linux-specific
checks when checking for "onegroup" PAG GIDs.
[mmeffie@sinenomine.net: Move AFS_PAG_ONEGROUP_ENV to param.h]
Reviewed-on: https://gerrit.openafs.org/11978 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com>
(cherry picked from commit ee08dbe37d9db4fe314bd88b9280bf73c92c37bd)
Change-Id: Ifef8f833599eca4241b41035142e74f32e6efa99
Reviewed-on: https://gerrit.openafs.org/12525 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Marcio Barbosa [Tue, 31 Jan 2017 14:43:18 +0000 (11:43 -0300)]
vol: detach offline volumes on dafs
Taking a volume offline always clears the inService bit. Taking a
volume out of service also takes it offline. Therefore, if the
inService flag is false, the volume in question should be offline.
On dafs, an offline volume should be unattached.
The attach2() function does not change the state of the volume received
as an argument to unattached when the inService flag is false. Instead,
this function changes the state of the volume in question to
pre-attached and returns VNOVOL to the client. As result, subsequent
accesses to this volume will make the server try and fail to attach
this offline volume over and over again, writing to the FileLog each
time.
To fix this problem, detach the volume received as an argument if the
inService flag is false. Since the new state of this volume will be
unattached, subsequent accesses will not hit attach2().
This situation where a volume is not offline but is also not in service
can occur if a volume is taken offline with vos offline and some time
later the DAFS fileserver is shutdown and restarted; the volume is
placed into the preattach state by default when the server restarts.
Each access to the volume by clients then causes the fileserver to
attempt to attach the volume, which fails, since the in-service flag in
the volume header is false from the previous vos offline. The
fileserver will log a warning to the FileLog on each attempt to attach
the volume, and this will fill the FileLog with duplicate messages
corresponding to the number of attempted accesses.
Mark Vitale [Tue, 28 Feb 2017 23:02:39 +0000 (18:02 -0500)]
SOLARIS: prevent BAD TRAP panic with Studio 12.5
Starting with Solaris Studio 12.3, it is documented that Solaris kernel
modules (such as libafs) must not use any floating point, vector, or
SIMD/SSE instructions on x86 hardware. However, each new Studio
compiler release (12.4 and especially 12.5) is more likely to use these
types of instructions by default.
If the libafs kernel module includes any forbidden kernel instructions,
Solaris will panic the system with:
BAD TRAP: type=7 (#nm Device not available)
Provide a new autoconfig test to specify the required compiler options
(-xvector=%none -xregs=no%float) when building the OpenAFS kernel module
for Solaris, so that no invalid x86 instructions are used.
In addition, reinstate default kernel module optimization for Solaris.
It had been disabled in commit 80592c53cbb0bce782eb39a5e64860786654be9f
to address this same issue in Studio 12.3 and 12.4. However, Studio
12.5 started using some SSE instructions even with no optimization.
This commit has been tested with OpenAFS master and Studio 12.5 at all
optimization levels (none, -xO1 through -xO5) and verified to contain no
XMM register instructions via the following command:
$ gobjdump -dlr libafs64.o | grep xmm | wc -l
Mark Vitale [Tue, 21 Feb 2017 01:16:47 +0000 (20:16 -0500)]
DAFS: do not save or restore host state if CPS in progress
If a fileserver is shutdown while one or more PR_GetHostCPS calls
are in progress, this state is saved in the fsstate.dat file as
hostFlags HCPS_WAITING, HCPS_INPROGRESS. Other hosts that are
merely waiting will have HCPS_WAITING recorded.
However, it makes no sense to restore host structs in this state,
because the GetCPS calls will no longer be in progress. Once these
hosts become active, they will block server threads and quickly cause
all server threads to be exhausted as other CPS requests are blocked
behind them.
Instead, exclude these states from both save and restore.
Joe Gorse [Thu, 16 Feb 2017 23:01:50 +0000 (18:01 -0500)]
LINUX: Bring debug symbols back to the Linux kernel module.
Starting with 4.8 Linux kernels our existing build script
generator, make_kbuild_makefile.pl, does not pass the debugging
symbols CFLAGS that were present when building for previous kernels.
This fix appends the $(KERN_DBG) variable which will only be defined
when the configuration includes the --enable-debug-kernel option.
Reviewed-on: https://gerrit.openafs.org/12519 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 961cee00b8f5c302de5f66beb81caa33242c7971)
Change-Id: I1d16382c4a744d4624cac9a9ba2810fa664abe93
Reviewed-on: https://gerrit.openafs.org/12534 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Sergio Gelato [Wed, 22 Feb 2017 21:55:33 +0000 (13:55 -0800)]
LINUX: Debian/Ubuntu build regression on kernel 3.16.39
Now that kernel 4.9 has hit jessie-backports, it becomes desirable to
also backport the associated openafs patches.
Unfortunately, Linux-4.9-inode_change_ok-becomes-setattr_prepare.patch
causes a build failure against jessie's current default kernel,
3.16.39-1, due to the fact that setattr_prepare() is available (it was
cherrypicked to address CVE-2015-1350) but file_dentry() is not (it was
introduced in kernel 4.6).
This makes it difficult to have a version of openafs for jessie that
supports both kernels.
To deal with this, follow the implementation of file_dentry() in 4.6,
and simplify it to account for the lack of d_real() support in older
kernels.
Note that inode_change_ok() has been added back to 3.16.39-1 to avoid
ABI changes. That means the current openafs packages in jessie continue
to work with kernel 3.16.39-1 since they do not include
Linux-4.9-inode_change_ok-becomes-setattr_prepare.patch.
Originally reported at
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=855366
FIXES RT134158
Reviewed-on: https://gerrit.openafs.org/12523 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 6ea6c182c7fb6c22dafbbf203abcc23726e06cba)
Change-Id: I06951dacef3f7639f749e82439df89ec3d78b592
Reviewed-on: https://gerrit.openafs.org/12535 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Marcio Barbosa [Thu, 2 Mar 2017 21:01:48 +0000 (18:01 -0300)]
osx: build afscell only for active architecture
The InstallerPlugins framework provided by the MacOSX10.12.sdk does not
define symbols for architecture i386. As a result, the OpenAFS code
cannot be built on OS X 10.12.
To fix this problem, build the afscell xcode project only for active
architecture.
Reviewed-on: https://gerrit.openafs.org/12531 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com>
(cherry picked from commit d39e7c7af77b4e1b043611e1a6e78267f5f956ef)
Marcio Barbosa [Thu, 2 Mar 2017 21:01:48 +0000 (18:01 -0300)]
osx: build afscell only for active architecture
The InstallerPlugins framework provided by the MacOSX10.12.sdk does not
define symbols for architecture i386. As a result, the OpenAFS code
cannot be built on OS X 10.12.
To fix this problem, build the afscell xcode project only for active
architecture.
Michael Meffie [Sat, 5 Nov 2016 16:42:19 +0000 (12:42 -0400)]
SOLARIS: convert from ancient _depends_on to ELF dependencies
The ancient way of declaring module dependencies with _depends_on has
been deprecated since SunOS 2.6 (circa 1996). The presence of the old
_depends_on symbol triggers a warning message on the console starting
with Solaris 12, and the kernel runtime loader (krtld) feature of using
the _depends_on symbol to load dependencies may be removed in a future
version of Solaris.
Convert the kernel module from the ancient _depends_on method to modern
ELF dependencies. Remove the old _depends_on symbol and specify the -dy
and -N <name> linker options to set the ELF dependencies at link time,
as recommended in the Solaris device driver developer guidelines [1].
This commit does not change the declared dependencies, which may be
vestiges of ancient afs versions.
Michael Meffie [Wed, 21 Jan 2015 19:58:35 +0000 (14:58 -0500)]
bozo: do not exit when the client config already exists
The bosserver creates symlinks for the client CSDB and ThisCell config
files during initialization. Avoid exiting if the client CSDB or
ThisCell configuration already exists, otherwise the bosserver cannot be
restarted with bos restart.
This change fixes numerous places where the return values of various
system calls and standard library routines are not checked. In
particular, this fixes occurrances called out when building on Ubuntu
12.10, with gcc 4.7.2 and eglibc 2.15-0ubuntu20.1, when the possible
failure is one we actually do (or should) care about. This change
does not consider calls where the failure is one we deliberately
choose to ignore.
Mark Vitale [Wed, 7 Dec 2016 16:11:45 +0000 (11:11 -0500)]
Linux 4.10: have_submounts is gone
Linux commit f74e7b33c37e vfs: remove unused have_submounts() function
(v4.10-rc2) removes have_submounts from the tree after providing a
replacement (path_has_submounts) for its last in-tree caller, autofs.
However, it turns out that OpenAFS is better off not using the new
path_has_submounts. Instead, OpenAFS could/should have stopped using
have_submounts() much earlier, back in Linux v3.18 when d_invalidate
became void. At that time, most in-tree callers of have_submounts had
already been converted to use check_submounts_and_drop back in v3.12.
At v3.18, a series of commits modified check_submounts_and_drop to
automatically remove child submounts (instead of returning -EBUSY if a
submount was detected), then subsumed it into d_invalidate. The end
result was that VFS now implicitly handles much of the housekeeping
previously called explicitly by the various filesystem d_revalidate
routines:
- shrink_dcache_parent
- check_submounts_and_drop
- d_drop
- d_invalidate
All in-tree filesystem d_revalidate routines were updated to take
advantage of this new VFS support.
Modify afs_linux_dentry_revalidate to no longer perform any special
handling for invalid dentries when D_INVALIDATE_IS_VOID. Instead, allow
our VFS caller to properly clean up any invalid dentry when we return 0.
Reviewed-on: https://gerrit.openafs.org/12506 Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 789319bf0f2b26ad67995f8cbe88cee87a1bbdc0)
Change-Id: I7ed22338e7896f69a204be78ed0a4f6136a3dab8
Reviewed-on: https://gerrit.openafs.org/12530 Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Michael Meffie [Thu, 11 Jun 2015 17:14:27 +0000 (13:14 -0400)]
libafs: vldb cache timeout option (-volume-ttl)
The unix cache manager caches VLDB information for read-only volumes as
long as a volume callback is held for a read-only volume. The volume
callback may be held as long as files in the read-only volume are being
accessed. The cache manager caches VLDB information for read/write
volumes as long as volume level errors (such as VMOVED) are not returned
by a fileserver while accessing files within the volume.
Add a new option to set the maximum amount of time VLDB information will
be cached, even if a callback is still held for a read-only volume, or
no volume errors have been encounted while accessing files in read/write
volumes.
This avoids situations where the vldb information is cached indefinitely
for read-only and read/write volumes. Instead, the VL servers will be
periodically probed for volume information.
Change-Id: I5f2a57cdaf5cbe7b1bc0440ed6408226cc988fed
Reviewed-on: https://gerrit.openafs.org/11898 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Neale Ferguson [Thu, 8 Dec 2016 16:47:09 +0000 (11:47 -0500)]
s390: desupport 32-bit Linux kernels on s390/s390x
Remove the obsolete and custom lwp assembler for the s390 and s390x
architectures. That assembler is no longer needed since 32-bit
mainframe Linux distributions are no longer supported and are very
unlikely to be in use.
The generic process.default.s is sufficient for modern 64-bit Linux
distributions on s390/s390x.
[mmeffie@sinenomine.net: commit message wording]
Reviewed-on: https://gerrit.openafs.org/12475 Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit 1d8cb56999a4ab25ae4cbc8e8a688b8100aedd3b)
Change-Id: Iee572ef3a86f5502e37ddc0775da13b874add669
Reviewed-on: https://gerrit.openafs.org/12499 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Marcio Barbosa [Wed, 11 Jan 2017 14:05:04 +0000 (06:05 -0800)]
osx: let prefpane knows where binaries can be found
Starting from OS X 10.11, the OpenAFS binaries were moved to the
following directories: /opt/openafs/bin and /opt/openafs/sbin. However,
the OpenAFS prefpane is not aware of the change mentioned above. As a
result, some functionalities provided by the OpenAFS prefpane are not
working properly.
To fix this problem, add the new paths to the proper environment
variable.
Change-Id: Idaa2f0329af2092cf9ad1d63f1a01300b150227a
Reviewed-on: https://gerrit.openafs.org/12507 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
(cherry picked from commit a92a3a0675d941536103b60d708a6b3305b9b8fa)
Reviewed-on: https://gerrit.openafs.org/12512 Reviewed-by: Marcio Brito Barbosa <mbarbosa@sinenomine.net> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
Sergio Gelato [Wed, 22 Feb 2017 21:55:33 +0000 (13:55 -0800)]
LINUX: Debian/Ubuntu build regression on kernel 3.16.39
Now that kernel 4.9 has hit jessie-backports, it becomes desirable to
also backport the associated openafs patches.
Unfortunately, Linux-4.9-inode_change_ok-becomes-setattr_prepare.patch
causes a build failure against jessie's current default kernel,
3.16.39-1, due to the fact that setattr_prepare() is available (it was
cherrypicked to address CVE-2015-1350) but file_dentry() is not (it was
introduced in kernel 4.6).
This makes it difficult to have a version of openafs for jessie that
supports both kernels.
To deal with this, follow the implementation of file_dentry() in 4.6,
and simplify it to account for the lack of d_real() support in older
kernels.
Note that inode_change_ok() has been added back to 3.16.39-1 to avoid
ABI changes. That means the current openafs packages in jessie continue
to work with kernel 3.16.39-1 since they do not include
Linux-4.9-inode_change_ok-becomes-setattr_prepare.patch.
Originally reported at
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=855366
FIXES RT134158
Change-Id: I157aa2ff25945c1c6e3b8e4a600557125711a681
Reviewed-on: https://gerrit.openafs.org/12523 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Mark Vitale [Wed, 7 Dec 2016 16:11:45 +0000 (11:11 -0500)]
Linux 4.10: have_submounts is gone
Linux commit f74e7b33c37e vfs: remove unused have_submounts() function
(v4.10-rc2) removes have_submounts from the tree after providing a
replacement (path_has_submounts) for its last in-tree caller, autofs.
However, it turns out that OpenAFS is better off not using the new
path_has_submounts. Instead, OpenAFS could/should have stopped using
have_submounts() much earlier, back in Linux v3.18 when d_invalidate
became void. At that time, most in-tree callers of have_submounts had
already been converted to use check_submounts_and_drop back in v3.12.
At v3.18, a series of commits modified check_submounts_and_drop to
automatically remove child submounts (instead of returning -EBUSY if a
submount was detected), then subsumed it into d_invalidate. The end
result was that VFS now implicitly handles much of the housekeeping
previously called explicitly by the various filesystem d_revalidate
routines:
- shrink_dcache_parent
- check_submounts_and_drop
- d_drop
- d_invalidate
All in-tree filesystem d_revalidate routines were updated to take
advantage of this new VFS support.
Modify afs_linux_dentry_revalidate to no longer perform any special
handling for invalid dentries when D_INVALIDATE_IS_VOID. Instead, allow
our VFS caller to properly clean up any invalid dentry when we return 0.
Change-Id: I0c4d777e6d445857c395a7b5f9a43c9024b098e9
Reviewed-on: https://gerrit.openafs.org/12506 Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Joe Gorse [Thu, 16 Feb 2017 23:01:50 +0000 (18:01 -0500)]
LINUX: Bring debug symbols back to the Linux kernel module.
Starting with 4.8 Linux kernels our existing build script
generator, make_kbuild_makefile.pl, does not pass the debugging
symbols CFLAGS that were present when building for previous kernels.
This fix appends the $(KERN_DBG) variable which will only be defined
when the configuration includes the --enable-debug-kernel option.
Change-Id: I9a85dc0311a3a706239bc9e471b2d7197ebe1946
Reviewed-on: https://gerrit.openafs.org/12519 Reviewed-by: Benjamin Kaduk <kaduk@mit.edu> Tested-by: Benjamin Kaduk <kaduk@mit.edu>
Michael Meffie [Fri, 10 Feb 2017 15:39:09 +0000 (10:39 -0500)]
build: add --without-swig to override swig check
Add the --without-swig option to disable the automatic swig detection
and disable the optional features which depend on swig. This allows
builders to avoid swig even if present on the build system.
Also, add the --with-swig option to force the check and fail if not
detected. This allows builders to declare the swig features are
mandatory.
The default continues to be to check for swig, and if present, build the
optional features which require swig.
To disable the automatic check for swig and disable the features which
depend on swig:
./configure --without-swig # or --with-swig=no
To force the check and fail if swig is not present on the system:
./configure --with-swig # or --with-swig=yes
If --with-swig is given and swig is not detected, then configure will
fail with the message:
configure: error: swig requested but not found
The Perl 5 bindings for libuafs is the only feature which requires swig
at this time.
Andrew Deason [Fri, 10 Feb 2017 07:29:28 +0000 (01:29 -0600)]
PERLUAFS: Modernize lang-specific swig typemaps
Currently, our swig bindings for PERLUAFS define a couple of typemaps
like so:
%typemap(in, numinputs=1, perl5) (char *READBUF, int LENGTH) {
[...]
}
Embedding the target language name in the typemap arguments is a very
old way of specifying what language the typemap is for; they were
removed after swig 1.1. With swig 3.0.x releases (and possibly
others), the specific combination of this deprecated syntax and some
other features we're using causes a segfault. That's clearly a bug in
swig, but we shouldn't be using the deprecated syntax anyway.
Update this to instead use preprocessor symbols to specify
language-specific typemaps (#ifdef SWIGPERL). We only actually define
these for perl right now, so make sure to throw an error if we're not
running for perl.
FIXES 134103
Change-Id: I14264a2dfada53d99413808ed5d60b79b1ee44f3
Reviewed-on: https://gerrit.openafs.org/12517 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Anders Kaseorg [Tue, 6 Dec 2016 15:48:31 +0000 (10:48 -0500)]
AFS_component_version_number.c: Respect SOURCE_DATE_EPOCH if set
To improve build reproducibility, if the SOURCE_DATE_EPOCH environment
variable is set, use it to deterministically replace the embedded build
date, and do not include the username or hostname in this case.
Marcio Barbosa [Wed, 11 Jan 2017 14:05:04 +0000 (06:05 -0800)]
osx: let prefpane knows where binaries can be found
Starting from OS X 10.11, the OpenAFS binaries were moved to the
following directories: /opt/openafs/bin and /opt/openafs/sbin. However,
the OpenAFS prefpane is not aware of the change mentioned above. As a
result, some functionalities provided by the OpenAFS prefpane are not
working properly.
To fix this problem, add the new paths to the proper environment
variable.
Mark Vitale [Sat, 7 Jan 2017 11:22:47 +0000 (06:22 -0500)]
LINUX: eliminate unused variable warning
Commit c3bbf0b4444db88192eea4580ac9e9ca3de0d286 added routine
osi_TryEvictDentries and included new logic for D_INVALIDATE_IS_VOID.
Unfortunately, this new code path no longer uses dentry; it also should
have been made conditional at that time.
Wrap the declaration of dentry in #ifndef D_INVALIDATE_IS_VOID to
eliminate the unused variable warning.
Andrew Deason [Sat, 8 Aug 2015 21:49:50 +0000 (16:49 -0500)]
SOLARIS: Use AFS_PAG_ONEGROUP_ENV for Solaris 11
On Solaris 11 (specifically, Solaris 11.1+), the supplemental group
list for a process is supposed to be sorted. Starting with Solaris
11.2, more authorization checks are done that assume the list is
sorted (e.g., to do a binary search), so having them out of order
can cause incorrect behavior. For example:
$ echo foo > /tmp/testfile
$ chmod 660 /tmp/testfile
$ sudo chown root:daemon /tmp/testfile
$ cat /tmp/testfile
foo
$ id -a
uid=100(adeason) gid=10(staff) groups=10(staff),12(daemon),20(games),21(ftp),50(gdm),60(xvm),90(postgres)
$ pagsh
$ cat /tmp/testfile
cat: cannot open /tmp/testfile: Permission denied
$ id -a
uid=100(adeason) gid=10(staff) groups=33536,32514,10(staff),12(daemon),20(games),21(ftp),50(gdm),60(xvm),90(postgres)
Solaris sorts the groups given to crsetgroups() on versions which
required the group ids to be sorted, but we currently manually put our
PAG groups in our own order in afs_setgroups(). This is currently
required, since various places in the code assume that PAG groups are
the first two groups in a process's group list.
To get around this, do not require the PAG gids to be the first two
gids anymore. To more easily identify PAG gids in group processes, use
a single gid instead of two gids to identify a PAG, like modern Linux
currently uses (under the AFS_PAG_ONEGROUP_ENV). High-numbered groups
have been possible for quite a long time on Solaris, allegedly further
back than Solaris 8. Only do this for Solaris 11, though, to reduce
the platforms we affect.
[mmeffie@sinenomine.net: Define AFS_PAG_ONEGROUP_ENV in param.h.]
Benjamin Kaduk [Mon, 26 Dec 2016 18:15:35 +0000 (12:15 -0600)]
afsd_kernel: remove gratuitous OS dependence
Commit 94c15f62 in 2010 gave NetBSD and only NetBSD the debug
printing of errno and the strerror() output, with no justification
in the commit message. In the interest of unifying behavior and
avoiding unnecessary OS dependence, give all platforms the errno
and strerror() behavior.
Michael Meffie [Tue, 13 Sep 2016 02:21:59 +0000 (22:21 -0400)]
afsd: print syscalls on separate lines with afsd -debug
afsd prints information to standard out for testing and debugging when the
-debug option is given. However, syscall tracing is emitted without trailing
newlines on all platforms except netbsd, creating an unreadable wall of text.
Michael Meffie [Mon, 26 Sep 2016 15:19:13 +0000 (11:19 -0400)]
vol: convert vnode macros to inline functions
Convert the vnode macros to inline functions to fix integer overflows
for very large vnode numbers (and generally improve the code robustness
and readability).
The macro version of vnodeIndexOffset() will evaluate to an incorrect
offset for very large vnode numbers due to 32-bit integer overflow. The
vnode index file will then be corrupted when writing to the incorrect
offset.
In code paths where the vnode number incorrectly defined as a signed
32-bit integer, this change prevents vnodeIndexOffset() from evaluating
to a negative result when a vnode number is larger than 2^31.
Thanks to Mark Vitale for reporting and providing analysis.
Benjamin Kaduk [Thu, 22 Dec 2016 19:32:22 +0000 (14:32 -0500)]
Add versioned build-depends on heimdal-multidev
The base64 en/decode functions got renamed with a rk_ prefix to
avoid conflicts, and we have a patch to adopt to those new symbol
names, so require the corresponding header/library version.
$ make check
…
volser/vos..............FAILED 6
…
$ cd tests
$ ./libwrap ../lib ./runtests -o volser/vos
1..6
ok 1 - Successfully got security class
ok 2 - Successfully built ubik client structure
ok 3 - First address registration succeeds
ok 4 - Second address registration succeeds
ok 5 - vos output matches
Server exited with code 15
# wanted: 0
# seen: -1
not ok 6 - Server exited cleanly
# Looks like you failed 1 test of 6
afstest_StopServer has a check for the process terminating with signal
15 (SIGTERM), but not for the process exiting with code 15.
Change-Id: I022965ea2b5440486ea1cf562551d3bbd0516104
Reviewed-on: https://gerrit.openafs.org/12489 Tested-by: Anders Kaseorg <andersk@mit.edu> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Anders Kaseorg [Fri, 16 Dec 2016 05:29:21 +0000 (00:29 -0500)]
doc/man-pages/Makefile.in: mkdir man[158] in case we did regen.sh -q
Fixes this error:
$ git clean -xdf
$ ./regen.sh -q
$ ./configure
$ make
[…]
make[3]: Entering directory '/…/openafs/doc/man-pages'
rm -f man*/*.noinstall
if [ "no" = "no" ] ; then \
for M in man1/klog.1 man1/knfs.1 […] man8/kpwvalid.8 man1/klog.krb.1; do \
touch $M.noinstall; \
done; \
fi
touch: cannot touch 'man1/klog.1.noinstall': No such file or directory
touch: cannot touch 'man1/knfs.1.noinstall': No such file or directory
[…]
touch: cannot touch 'man8/kpwvalid.8.noinstall': No such file or directory
touch: cannot touch 'man1/klog.krb.1.noinstall': No such file or directory
Makefile:34: recipe for target 'prep-noinstall' failed
make[3]: *** [prep-noinstall] Error 1
make[3]: Leaving directory '/…/openafs/doc/man-pages'
Change-Id: I95098fb2b27f1d87fc9769497b225e9f91f72266
Reviewed-on: https://gerrit.openafs.org/12492 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Michael Meffie <mmeffie@sinenomine.net> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Anders Kaseorg [Wed, 14 Dec 2016 20:47:21 +0000 (15:47 -0500)]
tests/opr/softsig-t: Avoid hanging due to intermediate sh -c
If the build directory happened to contain shell metacharacters, like
the ~ in /build/openafs-vb8tid/openafs-1.8.0~pre1 used by the Debian
builders, Perl was running softsig-helper via an intermediate sh -c,
which would then intercept the signals we tried to send to
softsig-helper. Use the list syntax to avoid this sh -c.
Benjamin Kaduk [Fri, 16 Dec 2016 04:12:01 +0000 (22:12 -0600)]
tests: use exec to call libwrap'd executables
No need to leave the shell process hanging around.
In particular, if we are manually running softsig-helper under
libwrap to debug test failures, the child process of the shell is
another shell, which interprets some signals that we wanted to
be passed through, like SIGTERM. On the other hand, once the
softsig-helper is exec()'d, you basically need another shell to
terminate it, which is a different problem....
Change-Id: Iff7c519886a018cb68e692746d40c427b6299457
Reviewed-on: https://gerrit.openafs.org/12490 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Anders Kaseorg <andersk@mit.edu> Tested-by: Anders Kaseorg <andersk@mit.edu> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Michael Meffie [Tue, 16 Aug 2016 16:56:47 +0000 (12:56 -0400)]
tests: fix signo to signame lookup in opr/softsig tests
Fix the loop condition when scanning the signal number to name table to
convert a signal number to a name. Instead of looping sizeof(size_t)
times, loop for the number of elements in the table.
This bug was masked on 64 bit-platforms, since the signal number to name
table table currently has 8 elements, which is coincidently the same as
sizeof(size_t) on 64-bit platforms. The bug becomes apparent on 32-bit
systems; only the first 4 elements of the table are checked.
Example error output before this fix:
$ cd tests
$ ./libwrap ../lib ./runtests -o opr/softsig
1..11
ok 1
ok 2
ok 3
ok 4
ok 5
not ok 6
# Failed test in ./opr/softsig-t at line 57.
# got: 'Received UNK
# '
# expected: 'Received TERM
# '
not ok 7
# Failed test in ./opr/softsig-t at line 60.
# got: 'Received UNK
# '
# expected: 'Received USR1
# '
not ok 8
# Failed test in ./opr/softsig-t at line 63.
# got: 'Received UNK
# '
# expected: 'Received USR2
# '
ok 9 - Helper exited on KILL signal.
ok 10 - Helper exited on SEGV signal.
ok 11 # skip Skipping buserror test; SIGBUS constant is not defined.
# Looks like you failed 3 tests of 11.
Benjamin Kaduk [Fri, 16 Dec 2016 05:08:27 +0000 (00:08 -0500)]
Work on fixing test failures on !amd64
One is reproducible on i386, and the result of an extra sizeof()
in a loop boundary condition, so it stops at sizeof(size_t) rather
than the size of the array (8). The other only appears on the
buildds for some more exotic architectures, so run that test
verbosely and see if we can get enough information to track
things down.
Neale Ferguson [Thu, 8 Dec 2016 16:47:09 +0000 (11:47 -0500)]
s390: desupport 32-bit Linux kernels on s390/s390x
Remove the obsolete and custom lwp assembler for the s390 and s390x
architectures. That assembler is no longer needed since 32-bit
mainframe Linux distributions are no longer supported and are very
unlikely to be in use.
The generic process.default.s is sufficient for modern 64-bit Linux
distributions on s390/s390x.
[mmeffie@sinenomine.net: commit message wording]
Change-Id: I654b10dfc257e7de90c9a50048982427276f4d61
Reviewed-on: https://gerrit.openafs.org/12475 Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Stephan Wiesand [Mon, 12 Dec 2016 14:09:50 +0000 (15:09 +0100)]
Make OpenAFS 1.6.20.1
Update configure version strings for 1.6.20.1. Note that macos kext
can be of form XXXX.YY[.ZZ[(d|a|b|fc)NNN]] where d dev, a alpha,
b beta, f final candidate so we have no way to represent 1.6.20.1.
Switch to 1.6.21 dev 1 for macOs.
Change-Id: If9a54680d6807687136f6149ca48ad8c33db32f7
Reviewed-on: https://gerrit.openafs.org/12485 Tested-by: BuildBot <buildbot@rampaginggeek.com> Reviewed-by: Mark Vitale <mvitale@sinenomine.net> Reviewed-by: Stephan Wiesand <stephan.wiesand@desy.de>
projects / packages / o / openafs.git / log