There are no races. Racism does exist though.
Change-Id: I0a4cde55a5f470649eb99c5d7f30c9cec86d9baa
Reviewed-on: https://gerrit.openafs.org/14320
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>
Currently, 'linktest' in libuafs is built with a weird custom rule
that specifies several various CFLAGS and LDFLAGS, etc. One
side-effect of this is that linktest is built without specifying -O,
even if optimization is otherwise enabled.
Normally nobody would care about the optimization of linktest, since
it's never supposed to be run, but this can cause an error when
building with -D_FORTIFY_SOURCE=1 on some systems (such as RHEL7):
In file included from /usr/include/sys/types.h:25:0,
from /.../src/config/afsconfig.h:1485,
from /.../src/libuafs/linktest.c:15:
/usr/include/features.h:330:4: error: #warning _FORTIFY_SOURCE requires compiling with optimization (-O) [-Werror=cpp]
# warning _FORTIFY_SOURCE requires compiling with optimization (-O)
^
cc1: all warnings being treated as errors
make[3]: *** [linktest] Error 1
For now, to fix this just include $(COMMON_CFLAGS) in the flags we
give for linktest, so $(OPTMZ) also gets pulled in, and building
linktest gets a little closer to a normal compilation step.
Change-Id: I3362dcfe8407825ab88854ae59da4188ed16be9d
Reviewed-on: https://gerrit.openafs.org/14324
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
It looks like a call to ubik_SetLock(.. LOCKREAD) was left in
place in listSuperGroups after locking was moved to ReadPreamble
in commit a6d64d70 (ptserver: Refactor per-call ubik initialisation)
When compiled with 'supergroups', and once contacted by
"pts mem -expandgroups ..", ptserver will therefore abort() with
Ubik: Internal Error: attempted to take lock twice
This patch removes the superfluous ubik_SetLock.
FIXES 135147
Change-Id: I8779710a6d68e4126fc482123b576690d86e4225
Reviewed-on: https://gerrit.openafs.org/14338
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The change associated with gerrit #14300 removed support for older
Linux kernels (2.6.10 and earlier).
The commit 'Import of code from autoconf-archive' (d8205bbb4) introduced
a check for Autoconf 2.64. Autoconf 2.64 was released in 2009.
The commit 'regen.sh: Use libtoolize -i, and .gitignore generated
build-tools' (a7cc505d3) introduced a dependency on libtool's '-i'
option. Libtool supported the '-i' option with libtool 1.9b in 2004.
Update the INSTALL instructions to document a minimum Linux kernel
level and the minimum levels for autoconf and libtool.
Notes: RHEL4 (EOL in 2017) had a 2.6.9 kernel and RHEL5 has a 2.6.18
kernel. RHEL5 has libtool 1.5.22 and autoconf 2.59, RHEL6 has libtool
2.2.6 and autoconf 2.63, and RHEL7 has libtool 2.4.2 and autoconf 2.69.
Change-Id: I235eeffa4adb152e05aab7aca839700816e62c83
Reviewed-on: https://gerrit.openafs.org/14305
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Inside release_conns_user_server, connection vector is traversed and after
destroying a connection new eligible connection is found on which NatPing
event will be set. Ideally there should be only one connection on which
NatPing should be set but in current code while traversing all connection
of server a NatPing event is set on all connections to that server. In
cases where we have large number of connection to a server this can lead
to huge number of “RX_PACKET_TYPE_VERSION” packets sent to a server.
Since this happen during Garbage collection of user structs, to simulate
this issue below steps were tried
- had one script which “cd” to a volume mount and then script sleeps for
large time.
- Ran one infinite while loop where above script was called using PAG
based tokens (As new connection will be created for each PAG)
- Instrumented the code, so that we hit above code segment where NatPing
event is set. Mainly reduced NOTOKTIMEOUT to 60 sec.
To fix this issue set NatPing on one connection and once it is set break
from “for” loop traversing the server connection.
Change-Id: Ia38cec0403fde76cdd59aa664bd261481e2edee6
Reviewed-on: https://gerrit.openafs.org/14312
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reported symptoms:
If a 'vos rename' is interrupted after it has locked the volume and
replaced the VLDB entry, but before it has unlocked the volume, the
volume will remain locked. However, the locked volume will NOT be
listed as locked in any vos commands that display locked status (see
below for details).
Background:
Most vos write operations lock the VLDB volume entry before proceeding,
then release the volume lock when finished. This is accomplished via
VL_SetLock and VL_ReleaseLock, respectively.
VL_SetLock always sets these members in the VLDB volume entry:
- flags is modified to set the required VLOP_* code bit as specified
- LockAFSid is set to 0 (never implemented)
- LockTimestamp is set to the current time
VL_ReleaseLock always sets them as follows:
- flags is cleared of any VLOP_* code bit
- LockAFSid is set to 0 (never implemented)
- LockTimestamp is set to 0
VL_ReplaceEntry(N) may also optionally clear each of these members:
- flags operation bits may be explicitly cleared via LOCKREL_OPCODE
- LockAFSid may be explicitly cleared via LOCKREL_AFSID
- LockTimestamp may be explicitly cleared via LOCKREL_TIMESTAMP
When all 3 options are specified, VL_ReplaceEntry also does the
functional equivalent of a VL_ReleaseLock. Most vos operations use this
method. However, when no lock release options are specified on
VL_ReplaceEntry(N), the VLDB entry is simply replaced with the supplied
entry. This includes whatever flags values are specified in the
supplied entry; therefore, this amounts to an additional, implicit way
to set or modify the flags.
Root cause:
'vos rename' (UV_RenameVolume) is the only vos operation that does all
of the following things:
- accepts a replacement volume entry that was obtained before VL_SetLock
(and thus does NOT have any lock flags set)
- issues VL_SetLock (which sets the lock flag in the VLDB)
- issues VL_ReplaceEntry(N) with the original unlocked entry, and with
no lock release options (thus with explicit intent to leave the lock
flag unchanged, but inadvertently doing an implicit clear of the lock
flag in the VLDB)
- (performs some additional volserver work)
- issues VL_ReleaseLock to release the volume lock
Therefore, if 'vos rename' is cancelled or killed before reaching the
final VL_ReleaseLock step, the VLDB entry is left with the lock flags
cleared but the LockTimestamp still set. As we will see below, this
'half-locked' state produces confusing results from other vos commands.
Detection of locked state:
The 'vos lock' command (and all other vos commands that issue
VL_SetLock) use the lock timestamp to determine if a volume is locked.
However, several other vos commands ('vos listvldb <vol>', 'vos examine
<vol>', 'vos listvldb -locked') use the VLDB entry's lock flags (not the
lock timestamp) to determine if the volume is locked. Therefore, if the
lock flags have been cleared but the lock timestamp is still set, these
commands fail to detect that the volume is still locked. Yet an
administrator's 'vos lock <volume>' will still fail with:
Could not lock VLDB entry for volume <volume>
VLDB: vldb entry is already locked
This is the external manifestation of the 'half-locked' state.
Workaround and fix:
This scenario has a simple workaround: 'vos unlock <volume>'. However,
to avoid this confusing outcome in the first place, modify the 'vos
rename' logic so that the lock flags are no longer inadvertently
cleared. Now, if the 'vos rename' is interrupted before the volume is
unlocked, it will still appear locked in normal vos command output.
Change-Id: I6cc16d20c4487de4e9a866c6f0c89d950efd2f7d
Reviewed-on: https://gerrit.openafs.org/14157
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Since the original IBM code import, hndle_param_tail has been dead code.
It was later ifdef'd out in commit 8f2df21ffe
'pull-prototypes-to-head-20020821'
Remove the dead code from the tree.
No functional change is incurred by this commit.
Change-Id: I29128eecc93a5871f5bb9369c3983baf5b537beb
Reviewed-on: https://gerrit.openafs.org/14322
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Commit d008089a7 (Add interface to select client security objects)
consolidated the code that selects the client security objects into a
set of new interfaces. Before this commit, the "bos: running
unauthenticated" message, which warns the user when an unauthenticated
connection is established, used to be suppressed if the -noauth flag was
specified.
Similarly to commit b3c16324e (ubik: Make ugen_ClientInit honor
noAuthFlag), recover the original behavior avoiding warn messages about
unauthenticated connections if the -noauth flag is provided.
Change-Id: Iaf0ac6bd91ea160256823512f060afc94b5926bf
Reviewed-on: https://gerrit.openafs.org/14306
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The ListAttributesN2() RPC can fail to list read-only entries under
certain circumstances. This RPC is used by the `vos listvldb` command to
retrieve vldb entries (unless the -name option is given). The `vos
listvldb` command fails to list volume entries when run with the
'-server' option for volumes that have read-only replicas, but have not
been released.
Consider the following example volume:
$ vos create fs1.example.com a test
$ vos addsite fs1.example.com a test
$ vos addsite fs2.example.com a test
$ vos listvldb
...
test
RWrite: 536870921
number of sites -> 3
server fs1.example.com partition /vicepa RW Site
server fs1.example.com partition /vicepa RO Site -- Not released
server fs2.example.com partition /vicepa RO Site -- Not released
`vos listvldb` fails to find the volume when the search is limited to
server 'fs2':
$ vos listvldb -server fs2.example.com
VLDB entries for server fs2.example.com
Total entries: 0
Instead of the expected results:
$ vos listvldb -server fs2.example.com
test
RWrite: 536870921
number of sites -> 3
server fs1.example.com partition /vicepa RW Site
server fs1.example.com partition /vicepa RO Site -- Not released
server fs2.example.com partition /vicepa RO Site -- Not released
This situation makes it difficult to remove old server addresses from
the vldb. In this situation, 'vos remaddrs' and 'vos changeaddr
-remove' commands will complain the server addresses are still in use by
volume entries, however running 'vos listvldb -server' will not show
which volumes entries are in use.
The entries are not listed for unreleased volumes because the
ListAttributesN2() RPC is currently checking the volume VLF_ROEXISTS
flag, instead of the server site flags (serverFlags) to determine when
the entry is a read-only site. The volume VLF_ROEXISTS flag is set when
a volume is released.
To fix this, make ListAttributesN2 check for the VLSF_ROVOL site flag,
instead of the VLF_ROEXISTS entry flag.
Change-Id: Ib636fbe016d1d2f5b117624d9930dba83ebcef8a
Reviewed-on: https://gerrit.openafs.org/14154
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Linux-5.9-rc1 commit 'fs: remove the HAVE_UNLOCKED_IOCTL and
HAVE_COMPAT_IOCTL defines' (4e24566a) removed the two referenced macros
from the kernel.
The support for unlocked_ioctl and compat_ioctl were introduced in
Linux 2.6.11.
Remove references to HAVE_UNLOCKED_IOCTL and HAVE_COMPAT_IOCTL using
the assumption that they were always defined.
Notes:
With this change, building against kernels 2.6.10 and older will fail.
RHEL4 (EOL in March 2017) used a 2.6.9 kernel. RHEL5 uses a 2.6.18
kernel.
In linux-2.6.33-rc1 the commit messages for "staging: comedi:
Remove check for HAVE_UNLOCKED_IOCTL" (00a1855c) and "Staging: comedi:
remove check for HAVE_COMPAT_IOCTL" (5d7ae225) both state that all new
kernels have support for unlocked_ioctl/compat_ioctl so the checks can
be removed along with removing support for older kernels.
Change-Id: Idd2716f3573ea455f8a5e1535bca584af0787717
Reviewed-on: https://gerrit.openafs.org/14300
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, the UV_RestoreVolume2 function always attempts to create a
new volume, even when doing a incremental restore over an existing
volume. When the volume already exists, the volume creation operation
fails on the volume server with a VVOLEXISTS error. The client will then
attempt to obtain a transaction on the existing volume. If a transaction
is obtained, the incremental restore operation will proceed. If a full
restore is being done, the existing volume is removed and a new empty
volume is created.
Unfortunately, the failed volume creation is logged to by the volume
server, and so litters the log file with:
Volser: CreateVolume: Unable to create the volume; aborted, error code 104
To avoid polluting the volume server log with these messages, reverse
the logic in UV_RestoreVolume2. Assume the volume already exists and try
to get the transaction first when doing an incremental restore. Create a
new volume if the transaction cannot be obtained because the volume is
not present. When doing a full restore, remove the existing volume, if
one exists, and then create a new empty volume.
Change-Id: I8bdc13130d12c81cd2cd18a9484852708cac64d7
Reviewed-on: https://gerrit.openafs.org/14208
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Marcio Brito Barbosa <marciobritobarbosa@gmail.com>
Tested-by: Marcio Brito Barbosa <marciobritobarbosa@gmail.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The SOURCE and BUILD environment variables have been changed to
C_TAP_SOURCE and C_TAP_BUILD in the new version of c-tap-harness. The
runtests command syntax has changed as well.
Convert all of the old SOURCE and BUILD environment variables to the new
C_TAP_SOURCE and C_TAP_BUILD names.
Add the required -l command line option to specify the test list.
Add the new runtests -v option to run the tests in verbose mode to make
it easier to see which tests failed.
Change-Id: I209a6dc13d6cd1507519234fce1564fc4641e70b
Reviewed-on: https://gerrit.openafs.org/14295
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
This commit updates the code imported from c-tap-harness to
abdb66561ffd4d2f238fdb06f448ccf09d80c059 (release/4.7)
Upstream changes are:
Daniel Collins (1):
Add is_blob() test function.
Daniel Kahn Gillmor (1):
LICENSE: use https for all URLs
Daria Brashear (1):
Add verbose mode environment variable to runtests
Julien ÉLIE (2):
Document -v in usage and comments of runtests
Avoid realloc of zero length in tests/runtests.c
Marc Dionne (1):
Add test_cleanup_register_with_data
Russ Allbery (115):
clang --analyze cleanups for runtests
Modernize POD tests
Update README to my current layout
Explicitly note that test programs must be executable
Fix comment typo in tests/runtests.c
Switch to a copyright-format 1.0 LICENSE file
Flush harness output after each line
Show the test count as ? when the plan is deferred
More correctly backspace over test counts when aborting
Refactor test list handling
Allow passing tests on the runtests command line
Don't allow command-line arguments if a list was given
Search for tests under the name given as well
Release 2.0
Fix backward incompatibility when searching for tests
Document decision to ignore TAP version directives
Release 2.1
Document different runtests behavior in bail handling
Change exit status of bail to 255
Release 2.2
Add a new test_cleanup_register C API
Add warn_unused_result attributes
Add portability for warn_unsed_result attributes to tap/macros.h
Minor coding style fix (spacing) in runtests.c
Split the runtests usage string for ISO C90 string limits
Include stddef.h
Diagnose failure to register the exit handler
Use diag internally in the basic C TAP library
Some additional comments about cleanup functions
Move repetitive printing code in the C TAP library to a macro
Set a flag when bailing for more correct cleanup
Change my email address to eagle@eyrie.org
Release 2.3
Add diag_file_add and diag_file_remove functions
Don't die for unknown files passed to diag_file_remove
Release 2.4
Update comment about AIX and WCOREDUMP
Don't test for NULL before calling free
Be more careful about file descriptors in child processes
Run cleanup functions in non-primary processes as well
Release 3.0
Update collective package copyright notices at start of LICENSE
Check integer overflows on memory allocation, fix string creation
Switch POD spelling test to use Lancaster consensus variable
Add new bnrealloc API for brealloc with checked multiplication
Rename nrealloc to reallocarray
Return the test status from test functions
Fix the overflow check for breallocarray
Fix the overflow check for xreallocarray in runtests
Restructure test result reallocation in runtests
Change diag and sysdiag to always return true
Release 3.1
Fix typos in basic.c and basic.h
Fix usage message when running runtests with no arguments
Update introductory runtests comments for current syntax
Add the -l flag to suggested runtests invocation in README
Support comments and blank lines in test lists
Release 3.2
Update licensing information
Various improvements to verbose support
Compile warning-free with Clang, check Autoconf macros
Release 3.3
Remove unnecessary assert.h include in tap/basic.c
Fix some additional -v documentation issues
Rebalance usage to avoid too-long strings
Fix segfault in runtests with empty test list
Release 3.4
Document running autogen if starting from Git
Rename autogen to bootstrap
Support and prefer C_TAP_SOURCE and C_TAP_BUILD
Fix comment typo in tests/runtests.c
Add missing va_end to is_double
Release 4.0
Fix all non-https www.eyrie.org URLs
Add is_bool C test function
Add DocKnot metadata and a Markdown README file
Update documentation for new DocKnot standards
Release 4.1
Use more defaults from DocKnot templates
Fix new fall-through warning in GCC 7
Use compiler warnings from rra-c-util, fix issues
Merge pull request #4 from solemnwarning/master
Coding style fixes and NEWS for is_blob
Re-enable -Wunknown-pragmas for GCC
Avoid zero-length realloc allocations in breallocarray
Update copyright date on tests/runtests.c
Release 4.2
Add SPDX-License-Identifier headers to source files
Add and run new check-cppcheck target
Fix instructions for running one test
Identify values as left and right
Fix is_string comparisons with NULL pointers
Add support for running tests under valgrind
Replace putc with fprintf
Update shared files from rra-c-util
Release 4.3
Update NEWS date for 4.3 release
Collapse some copyright dates
NEWS and coding style for test_cleanup_register_with_data
Remove unused variables caught by Clang scan-build
Update to rra-c-util 8.0
Fix error checking in bstrndup
Release 4.4
Add support for C++
Document that C TAP Harness can be built as C++
Release 4.5
Regenerate README files
Reformat using clang-format 10
Update to rra-c-util 8.1
Release 4.6
Fix spelling errors caught by codespell
Protect the test suite against C_TAP_VERBOSE
Switch to GitHub Actions for CI
Add NEWS entry for GCC 10 warning fixes
Release 4.7
Change-Id: I5a78215bf99b53bd848f0fa6bb9092deab38f24e
Reviewed-on: https://gerrit.openafs.org/14294
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Since OpenAFS 1.0, osi_GetTime has taken a timeval-like pointer, which
contains 32-bit fields (the actual type has been called either
osi_timeval_t or osi_timeval32_t over time). For platforms that have a
native timeval-like type with 32-bit fields, we just define
osi_timeval32_t to that type, and elsewhere we define our own struct
to be osi_timeval32_t. For platforms that use the native timeval, we
can then define osi_GetTime() to just be, e.g., microtime().
This approach is difficult to maintain, though, because we must keep
track of whether 'struct timeval' contains 32-bit fields on each
platform, which can depend on many factors. It's easy to make mistakes
(the current tree already contains mistakes), and there's not much
benefit.
To avoid all of this, just always define osi_timeval32_t to be our own
struct with afs_int32 fields, and provide definitions for osi_GetTime
that convert from the native time struct to our osi_timeval32_t. This
does mean that for some platforms we do an unnecessary type
conversion, but this is a small price to pay for more straightforward
and maintainable code.
To be a little more sure that our types are correct, change
osi_GetTime to be defined as an inline function instead of a macro.
At the same time, do a similar conversion for the KERNEL
implementation of the rx clock_GetTime function. Get rid of
platform-specific mess, and do a straightforward type conversion
between osi_timeval32_t and struct clock in an inline function.
Change-Id: I18819acb556a2a7f1b6da6994db9783c48108934
Reviewed-on: https://gerrit.openafs.org/14238
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Most platforms currently #define osi_GetTime in their param.h. This is
really redundant, since the definition of osi_GetTime almost never
changes for a given platform, so we end up with many copies of the
same osi_GetTime definition for a given platform.
Move osi_GetTime out of param.h for these platforms, and define it in
osi_machdep.h instead, which is where most platform-specific
definitions go.
For DFBSD, we don't have an osi_machdep.h at all yet, so create a new
one to contain the osi_GetTime definition. Currently we don't build
libafs at all on DFBSD, but do this anyway so we don't lose the
existing osi_GetTime definition.
For NBSD, we were providing (conflicting!) definitions for osi_GetTime
in param.h and in osi_machdep.h. Just remove the definitions in
param.h, since those should have been getting overridden by the
osi_machdep.h definition.
Change-Id: I7097d9fe2fcd38c06ecc275e8fe3a2c69c9d0436
Reviewed-on: https://gerrit.openafs.org/14237
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Building with clang-10 produces the warning/error message
warning: converting the result of '<<' to a boolean always evaluates
to true [-Wtautological-constant-compare]
for the expression
abp->f_fsid = (AFS_VFSMAGIC << 16) || AFS_VFSFSID;
The message is because a logical OR '||' is used instead of a bitwise
OR '|'. The result of this expression will always set the f_fsid
member to a 1 and not the intended value of AFS_VFSMAGIC combined with
AFS_VFSFSID.
Update the expression to use a bitwise OR instead of the logical OR.
Note: This will change value stored in the f_fsid that is returned from
statfs.
Using a logical OR has existed since OpenAFS 1.0 for hpux/solaris and in
UKERNEL since OpenAFS 1.5 with the commit 'UKERNEL: add uafs_statvfs'
b822971a.
Change-Id: I3e85ba48058ac68e3e3ac7f277623f660187926c
Reviewed-on: https://gerrit.openafs.org/14292
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently when UKERNEL is defined, AFS_VFSFSID is always set to
AFS_MOUNT_AFS, which is a string for many platforms for UKERNEL.
Update src/afs/afs.h to insure that the define for AFS_VFSFSID is a
numeric value when building UKERNEL.
Clean up the preprocessor indentation in src/afs/afs.h in the area
around the AFS_VFSFSID defines.
Thanks to adeason@sinenomine.net for pointing out a much easier solution
for resolving this problem.
Change-Id: I618fc4c89029a6cca2ca6f530b8f65399299a9d1
Reviewed-on: https://gerrit.openafs.org/14279
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Clang-10 will not recognize '/* fall through */' as an indicator to
turn off the fallthrough warning due to the lack of a 'break' in a case
statement.
Code generated by flex uses the '/* fall through */' comments to turn
off compiler warnings for fallthroughs in case statements.
For code generated by flex, ignore the implicit-fallthrough via pragma
or disable the warning via a compile time flag.
Add new env variable "CFLAGS_NOIMPLICIT_FALLTHROUGH" to selectively
disable the compile check in Makefiles when checking is enabled.
Change-Id: I4c054defda03daa2aeb645ae2271dfa0cb54925f
Reviewed-on: https://gerrit.openafs.org/14275
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Clang-10 will not recognize '/* fallthrough */' as an indicator to
turn off the fallthrough diagnostic due to the lack of a 'break' in a
case statement. Clang-10 requires the '__attribute__((fallthrough))'
statement to disable the diagnostic.
In addition clang-10 is finding additional locations where fall throughs
occur.
Determine if the compiler supports '__attribute__((fallthrough))' to
disable the implicit fallthrough diagnostic.
Define a new macro 'AFS_FALLTHROUGH' that will disable the fallthrough
diagnostic. Set it as a wrapper for the Linux kernel's 'fallthrough'
macro if available, otherwise set it as a wrapper macro for
'__attribute__((fallthrough))' if the compiler supports it.
Update CODING to document the use of AFS_FALLTHROUGH when needing to
fallthrough between case statements.
Replace the '/* fallthrough */' comments with AFS_FALLTHROUGH, and add
AFS_FALLTHROUGH as needed.
Replace some fallthroughs with a break (or goto) if the flow was was
just to a break (or goto).
e.g. case x: case x:
somestmt; somestmt;
break;
case y: case y:
break; break;
Correct a mis-indented brace '}' in src/WINNT/afsd/smb3.c
Note, the clang maintainers have rejected the use of comments as a flag
to turn off the fall through warnings.
Change-Id: Ia5da10fc14fc1874baca035a3cf471e618e0d5f5
Reviewed-on: https://gerrit.openafs.org/14274
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
If `make` is not installed before dkms-openafs, the OpenAFS kernel
module is not built during the dkms-openafs package installation.
The failure happens in the "checking if linux kernel module build works"
configure step, which invokes `make` to check the linux buildsystem.
configure fails when `make` is not available, and gives the unhelpful
suggestion (in this case) of configuring with --disable-kernel module.
Running the configure.log in the dkms build directory shows:
configure:7739: checking if linux kernel module build works
make -C /lib/modules/4.18.0-193.6.3.el8_2.x86_64/build M=/var/lib/dkms/openafs/...
./configure: line 7771: make: command not found
configure: failed using Makefile:
Avoid this build failure by adding `make` to the list of dkms-openafs
package pre-requirements.
Change-Id: I98b3508341eea1df4fa7b6f43e88add1bda9ee2c
Reviewed-on: https://gerrit.openafs.org/14266
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Instead of needing to set every single field in the 'opts' structure
individually, blank the whole thing to make sure the entire struct is
initialized. Remove the now-redundant lines that initialize various
items to 0.
Change-Id: I799cdb55becd66a8f3d6ec2f81338843038d0abd
Reviewed-on: https://gerrit.openafs.org/14280
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Kailas Zadbuke <kailashsz@in.ibm.com>
Reviewed-by: Yadavendra Yadav <yadayada@in.ibm.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, the volserver SAFSVolDump RPC and the 'voldump' utility
handle short reads from pread() for vnode payloads by padding the
missing data with NUL bytes. That is, if we request 4k of data for our
pread() call, and we only get back 1k of data, we'll write 1k of data
to the volume dump stream followed by 3k of NUL bytes, and log
messages like this:
1 Volser: DumpFile: Error reading inode 1234 for vnode 5678
1 Volser: DumpFile: Null padding file: 3072 bytes at offset 40960
This can happen if we hit EOF on the underlying file sooner than
expected, or if the OS just responds with fewer bytes than requested
for any reason.
The same code path tries to do the same NUL-padding if pread() returns
an error (for example, EIO), padding the entire e.g. 4k block with
NULs. However, in this case, the "padding" code often doesn't work as
intended, because we compare 'n' (set to -1) with 'howMany' (set to 4k
in this example), like so:
if (n < howMany)
Here, 'n' is signed (ssize_t), and 'howMany' is unsigned (size_t), and
so compilers will promote 'n' to the unsigned type, causing this
conditional to fail when n is -1. As a result, all of the relevant log
messages are skipped, and the data in the dumpstream gets corrupted
(we skip a block of data, and our 'howFar' offset goes back by 1). So
this can result in rare silent data corruption in volume dumps, which
can occur during volume releases, moves, etc.
To fix all of this, remove this bizarre NUL-padding behavior in the
volserver. Instead:
- For actual errors from pread(), return an error, like we do for I/O
errors in most other code paths.
- For short reads, just write out the amount of data we actually read,
and keep going.
- For premature EOF, treat it like a pread() error, but log a slightly
different message.
For the 'voldump' utility, the padding behavior can make sense if a
user is trying to recover volume data offline in a disaster recovery
scenario. So for voldump, add a new switch (-pad-errors) to enable the
padding behavior, but change the default behavior to bail out on
errors.
Change-Id: Ibd6e76c5ea0dea95e3354d9b34536296f81b4f67
Reviewed-on: https://gerrit.openafs.org/14255
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Building with clang-10 results in 2 warnings/errors associated with
with trying to convert 0x7fffffff to a floating point value.
tcmain.c:240:18: error: implicit conversion from 'int' to 'float'
changes value from 2147483647 to 2147483648 [-Werror,
-Wimplicit-int-float-conversion]
if ((total > 0x7fffffff) || (total < 0)) /* Don't go over 2G */
and the same conversion warning on the statement on the following line:
total = 0x7fffffff;
Use floating point and decimal constants instead of the hex constants.
For the test, use 2147483648.0 which is cleanly represented by a float.
Change the comparison in the test from '>' to '>='.
If the total value exceeds 2G, just assign the max value directly to the
return variable.
Change-Id: I79b2afa006496a756bd7b50976050c24827aa027
Reviewed-on: https://gerrit.openafs.org/14277
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Due to bug <https://savannah.gnu.org/patch/?9949>,
ax_gcc_func_attribute.m4 fails to properly detect __attribute__((fallthrough))
in clang. Until this is fixed in autoconf-archive upstream, fix our
local copy of ax_gcc_func_attribute.m4, so we can detect
__attribute__((fallthrough)) to make --enable-checking work with clang.
Change-Id: I80a4557384f8e1438344e48bfe722e20c8773882
Reviewed-on: https://gerrit.openafs.org/14273
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Make a local copy of ax_gcc_func_attribute from autoconf-archive. This
is needed in order to fix a bug in the detection of the fallthrough
attribute.
Remove ax_gcc_func_attribute.m4 from src/external/autoconf-archive/m4.
Update LICENSE file to point to the local copy in src/cf.
Change-Id: I6c4244d2cd4edab4262c1820435c00419d85303b
Reviewed-on: https://gerrit.openafs.org/14272
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The rxi_connectionCache (AFS_PTHREAD_ENV only) allows applications to
reuse rx_connection structs. Cached rx_connections are obtained via
rx_GetCachedConnection and released via rx_ReleaseCachedConnection.
This feature is used most heavily by libadmin and kauth, but there are
other users in the tree as well.
For instance, ubikclient routines ubik_ClientInit and ubik_ClientDestroy
call rx_ReleaseCachedConnections (if AFS_PTHREAD_ENV) when disposing of
their rx_connections. Unfortunately, in many cases these rx_connections
were obtained via rx_NewConnection, _not_ from the cache via
rx_GetCachedConnection. In those cases, rx_ReleaseCachedConnection will
not find the rx_connection in the rxi_connectionCache, and thus it
returns without doing anything.
Therefore, when ubik_ClientInit is passed an existing ubik_client (for
re-initialization) that contains rx_connections NOT allocated via
rx_GetCachedConnection, those connections are not destroyed, but will be
silently leaked. Similarly, ubik_ClientDestroy will leak its
rx_connections when it frees the ubik_client struct.
For example, the fileserver host package calls ubik_ClientInit (via
hpr_Initialize) and ubik_ClientDestroy (via hpr_End) to manage
connections to the ptserver. However, these connections were obtained
via rx_NewConnection, not rx_GetCachedConnection. If the fileserver has
a failed call to the ptserver that sets prfail=1, the next RPC scheduled
for that client (in CallPreamble) will refresh the thread's ubik_client
(viced_uclient_key) by calling hprEnd -> ubik_ClientDestroy ->
rx_ReleaseCachedConnection. The "released" connections will be leaked.
This problem exists in all versions of OpenAFS going back to IBM 1.0.
Starting with 1.8.x, many components that were formerly LWP-only are now
pthreaded and thus susceptible to this leak.
It seems difficult and error-prone to identify all possible code paths
that may pass a non-cached rx_connection to rx_ReleaseCachedConnection,
and convert them to obtain connections via rx_GetCachedConnection.
Instead, prevent all existing and future leaks by modifying the connection
cache to:
- flag all rx_connections it allocates
- correctly release any rx_connection it is passed, whether they came
from the cache or not.
Change-Id: Ibe164ccd30a8ddd799438c28fd6e1d8a0a9040dd
Reviewed-on: https://gerrit.openafs.org/13042
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Commit 496fb87372 ("rx: avoid nat ping
until connection is attached") introduced functionality to defer turning
on NAT ping for server connections until after reachability had been
established for the client.
Unfortunately, this feature could never work correctly because it
assigned an out-of-range flag value of 256 (0x100) for the u_char flags
field. Instead of calling this out as an error, both gcc and Solaris cc
elide this flag so that it is never set in
rx_SetConnSecondsUntilNatPing(), Furthermore, the test in
rxi_ConnClearAttachWait() will always fail; therefore
rxi_ScheduleNatKeepAliveEvent is never called after attach wait has
ended.
Fortunately, this bug is currently moot - not actually exposed in
OpenAFS. (It was discovered by inspection). This is because there are
currently no rx_connection objects in the tree that have both NAT ping
and checkReach (rx_SetCheckReach) enabled. I also searched git history
and found no time when this bug could ever have been exposed. This does
raise the question of why the original commit was needed; but instead of
reverting the original commit, this commit attempts to fix it.
To prevent problems if NAT ping and checkReach are ever both enabled for
an rx_connection, enlarge the rx_connection flags member so that the
RX_CONN_NAT_PING value is no longer out of range.
Change-Id: Ib667ece632f66fa5c63a76398acb3153fed6f9c3
Reviewed-on: https://gerrit.openafs.org/13041
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Since commit 35777145 (solaris-fopen-sucks-20060916), cellconfig.c has
redirected fopen, fclose, and fgets to local functions on
non-64bit-sparc Solaris, in order to work around that platform's stdio
limitations.
Commit 7c431f7571 (auth: retire writeconfig.c) moved the contents of
writeconfig.c into cellconfig.c. The previous writeconfig.c contained
some calls to stdio, including calling fprintf() on a pointer returned
by fopen() in that file.
Because fopen() was redirected to our local version, this means that
afsconf_SetExtendedCellInfo() calls fopen() to get an
afsconf_iobuffer*, and passes that pointer to the real system
fprintf() later on (instead of a native FILE*). The compiler does warn
about this, but this only happens on Solaris, where --enable-checking
is not implemented, so the build never fails.
To avoid this, remove the #defines for fopen, fgets, and fclose.
Instead, change all of the old cellconfig.c callers to explicitly call
afsconf_fopen, afsconf_fgets, and afsconf_fclose. On the affected
Solaris platforms, we keep our local definitions, and for other
platforms, we just make those functions call their system stdio
equivalents. For the code that was pulled in from writeconfig.c,
callers will just call the system fopen, fprintf, and fclose.
We still keep our local afsconf_FILE* definition on all platforms, so
the compiler will still do typechecking for our local afsconf_f*
functions on all platforms. So now if we make a mistake, it should be
a mistake on all platforms, so platforms with --enable-checking should
flag the error.
Change-Id: I4064d7f5ee82d5acab04a33b01c0603564a391e8
Reviewed-on: https://gerrit.openafs.org/14214
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Mark Vitale <mvitale@sinenomine.net>
Tested-by: Mark Vitale <mvitale@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, when afs_ShakeLooseVCaches runs osi_TryEvictVCache, we
check if osi_TryEvictVCache slept (i.e. dropped afs_xvcache/GLOCK). If
we sleep over 100 times, then we stop trying to evict vcaches and
return.
If we have recently accessed a lot of AFS files, this limitation can
severely reduce our ability to keep our number of vcaches limited to a
reasonable size. For example:
Say a Linux client runs a process that quickly accesses 1 million
files (a simple 'find' command) and then does nothing else. A few
minutes later, afs_ShakeLooseVCaches is run, but since all of the
newly accessed vcaches have dentries attached to them, we will sleep
on each one in order to try to prune the attached dentries. This means
that afs_ShakeLooseVCaches will evict 100 vcaches, and then return,
leaving us with still almost 1 million vcaches. This will happen
repeatedly until afs_ShakeLooseVCaches finally works its way through
all of the vcaches (which takes quite a while, if we only clear 100 at
once), or the dentries get pruned by other means (such as, if Linux
evicts them due to memory pressure).
The limit of 100 sleeps was originally added in commit 29277d96
(newvcache-dont-spin-20060128), but the current effect of it was
largely introduced in commit 9be76c0d (Refactor afs_NewVCache). It
exists to ensure that afs_ShakeLooseVCaches doesn't take forever to
run, but the limit of 100 sleeps may seem quite low, especially if
those 100 sleeps run very quickly.
To avoid the situation described above, instead of limiting
afs_ShakeLooseVCaches based on a fixed number of sleeps, limit it
based on how long we've been running, and set an arbitrary limit of
roughly 3 seconds. Only check how long we've been running after 100
sleeps like before, so we're not constantly checking the time while
running.
Log a new warning if we exit afs_ShakeLooseVCaches prematurely if
we've been running for too long, to help indicate what is going on.
Change-Id: I65729ace748e8507cc0d5c26dec39e74d7bff5d2
Reviewed-on: https://gerrit.openafs.org/14254
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, afs_lookup() will try to prefetch dir entries for normal
dirs via bulkstat whenever multiple pids are reading that dir.
However, if we already have a lot of vcaches, ShakeLooseVCaches may be
struggling to limit the vcaches we already have. Entering
afs_DoBulkStat can make this worse, since we grab afs_xvcache
repeatedly, we may kick out other vcaches, and we'll possibly create
30 new vcaches that may not even be used before they're evicted.
To try to avoid this, skip running afs_DoBulkStat if it looks like the
stat cache is really full.
Change-Id: I1634530170a189f32cb962dd7df28f88bc758b71
Reviewed-on: https://gerrit.openafs.org/13256
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, afs_ShakeLooseVCaches has a kind of warning that is logged
when we fail to free up any vcaches. This information can be useful to
know, since it may be a sign that users are trying to access way more
files than our configured vcache limit, hindering performance as we
constantly try to evict and re-create vcaches for files.
However, the current warning is not clear at all to non-expert users,
and it can only occur for non-dynamic vcaches (which is uncommon these
days).
To improve this, try to make a general determination if it looks like
the stat cache is "stressed", and log a message if so after
afs_ShakeLooseVCaches runs (for all platforms, regardless of dynamic
vcaches). Also try to make the message a little more user-friendly,
and only log it (at most) once per 4 hours.
Determining whether the stat cache looks stressed or not is difficult
and arguably subjective (especially for dynamic vcaches). This commit
draws a few arbitrary lines in the sand to make the decision, so at
least something will be logged in the cases where users are constantly
accessing way more files than our configured vcache limit.
Change-Id: I022478dc8abb7fdef24ccc06d477b349cca759ac
Reviewed-on: https://gerrit.openafs.org/13255
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The fileserver's FiveMinuteCheckLWP periodically calls
CleanupTimedOutCallBacks, and logs an informational messages if the
return code indicates that any callbacks were discarded.
However, since the original IBM code import, CleanupTimedOutCallBacks
has 1) ignored the return value from CleanupTimedOutCallBacks_r and 2)
unconditionally returned 0. This makes the informational message
essentially dead code.
Instead, check the code from CleanupTimedOutCallBacks_r and pass it back
to the caller.
Change-Id: I631831c398e43431b79f4a3a0c6f01307ac0c05e
Reviewed-on: https://gerrit.openafs.org/14256
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
In afs_linux_readpage_fastpath, if we discover that our disk cache fs
has no ->readpage function, we'll 'goto out', but we never close our
cacheFp. To make sure we close it, add a filp_close() call to the
'goto out' cleanup code.
Change-Id: I371c1d7ec51b03447fbcbe58fb89be7be0235022
Reviewed-on: https://gerrit.openafs.org/14252
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Commit 'LINUX: Return NULL for afs_linux_raw_open error' (f6af4a155)
updated afs_linux_raw_open to return NULL on some errors, but still
panics if obtaining the dentry fails.
Commit 'afs: Verify osi_UFSOpen worked' (c6b61a451) updated callers of
osi_UFSOpen to verify whether or not the open was successful. This
meant osi_UFSOpen (and routines it calls) could pass back an error
indication rather than panic when an error is encountered.
Update afs_linux_raw_open to return a failure instead of panic if unable
to obtain a dentry.
Update osi_UFSOpen to return a NULL instead of panic if unable to obtain
memory or fails to open the file. All callers of osi_UFSOpen handle a
fail return, though some will still issue a panic.
Update afs_linux_readpage_fastpath and afs_linux_readpages to not panic
if afs_linux_raw_open fails. Instead of panic, return an error.
For testing, an error can be forced by removing a file from the
cache directory.
Note this work is based on a commit by pruiter@sinenomine.net
Change-Id: Ic47e4868b4f81d99fbe3b2e4958778508ae4851f
Reviewed-on: https://gerrit.openafs.org/14242
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Mark Vitale <mvitale@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
afs_CFileOpen is a macro that invokes the open "method" of the
afs_cacheOps structure, and for disk caches the osi_UFSOpen function is
used.
Currently osi_UFSOpen will panic if there is an error encountered while
opening a file.
Prepare to handle osi_UFSOpen function returning a NULL instead of
issuing a panic (future commit).
Update callers of afs_CFileOpen to test for an error and to return an
error instead of issuing a panic.
While this commit eliminates some panics, it does not address some of the
more complex cases associated with errors from afs_CFileOpen.
Change-Id: I2bdd525633dd44ebf8e26fcfd7059dfdfffb6142
Reviewed-on: https://gerrit.openafs.org/14241
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Mark Vitale <mvitale@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
With Linux-5.8-rc1 commit 'mm: fold and remove lru_cache_add_anon() and
lru_cache_add_file()' (6058eaec), the lru_cache_add_file function is
removed since it was functionally equivalent to lru_cache_add.
Replace lru_cache_add_file with lru_cache_add.
Introduce a new autoconf test to determine if lru_cache_add is present
For reference, the Linux changes associated with the lru caches:
__pagevec_lru_add introduced before v2.6.12-rc2
lru_cache_add_file introduced in v2.6.28-rc1
__pagevec_lru_add_file replaces __pagevec_lru_add in v2.6.28-rc1
vmscan: split LRU lists into anon & file sets (4f98a2fee)
__pagevec_lru_add removed in v5.7 with a note to use lru_cache_add_file
mm/swap.c: not necessary to export __pagevec_lru_add() (bde07cfc6)
lru_cache_add_file removed in v5.8
mm: fold and remove lru_cache_add_anon() and lru_cache_add_file()
(6058eaec)
lru_cache_add exported
mm: fold and remove lru_cache_add_anon() and lru_cache_add_file()
(6058eaec)
Openafs will use:
lru_cache_add on 5.8 kernels
lru_cache_add_file from 2.6.28 through 5.7 kernels
__pagevec_lru_add/__pagevec_lru_add_file on pre 2.6.28 kernels
Change-Id: I79ebe4a81425bf8a8a327ddf2d3474aff9df039d
Reviewed-on: https://gerrit.openafs.org/14249
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Yadavendra Yadav <yadayada@in.ibm.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Reported by Debian's lintian(1).
The CellServDB, as an externally maintained file, is left unchanged.
Change-Id: I3bf241b924cb8cd7799a4c3e799f6acd375b2e8a
Reviewed-on: https://gerrit.openafs.org/14265
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Michael Meffie <mmeffie@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Currently, afs_DoBulkStat will scan the entire directory blob, looking
for entries to stat. If all or almost all entries are already stat'd,
we'll scan through the entire directory, doing nontrivial work on
each entry (we grab afs_xvcache, at least). All of this work is pretty
pointless, since the entries are already cached and so we won't do
anything. If many processes are trying to acquire afs_xvcache, this
can contribute to performance issues.
To avoid this, provide a constant bound on the number of entries we'll
search through: nentries * 4. The current arbitrary limits cap
nentries at 30, so this means we're capping the afs_DoBulkStat search
to 120 entries.
Change-Id: I66e9af5b27844ddf6cf37c8286fcc65f8e0d3f96
Reviewed-on: https://gerrit.openafs.org/13253
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
The callers of afs_FindVCache must hold at least a read lock on
afs_xvcache; some hold a shared or write lock (and set IS_SLOCK or
IS_WLOCK in the given flags). Two callers (afs_EvalFakeStat_int and
afs_DoBulkStat) currently hold a write lock, but neither of them need
to.
In the optimal case, where afs_FindVCache finds the given vcache, this
means that we unnecessarily hold a write lock on afs_xvcache. This can
impact performance, since afs_xvcache can be a very frequently
accessed lock (a simple operation like afs_PutVCache briefly holds a
read lock, for example).
To avoid this, have afs_DoBulkStat hold a shared lock on afs_xvcache,
upgrading to a write lock when needed. afs_EvalFakeStat_int doesn't
ever need a write lock at all, so just convert it to a read lock.
Change-Id: I5bd58b9e3a577c9e1ebf1bc3719e65a6c0af5cb8
Reviewed-on: https://gerrit.openafs.org/12656
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
If a process forks when another thread has serverLogMutex locked, the
child process inherits the locked serverLogMutex. This causes a deadlock
when code in the child process tries to lock serverLogMutex, since we
can never unlock serverLogMutex because the locking thread no longer
exists. This can happen in the salvageserver, since the salvageserver
locks serverLogMutex in different threads, and forks to handle salvage
jobs.
To avoid this deadlock, we register handlers using pthread_atfork()
so that the serverLogMutex will be held during the fork. The fork will
be blocked until the worker thread releases the serverLogMutex. Hence the
serverLogMutex will be held until the fork is complete and it will be
released in the parent and child threads.
Thanks to Yadavendra Yadav(yadayada@in.ibm.com) for working with me
on this issue.
Change-Id: I191c8272825c1667bb2150146e04b1dfe36a54e4
Reviewed-on: https://gerrit.openafs.org/14239
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Our current if() statement for determining whether we should run
afs_DoBulkStat to prefetch dir entries is a bit large, and grows over
time. Split this logic out into a separate function to make it easier
to maintain, and add some comments to help explain each condition.
This commit should have no visible effects; it's just code
reorganization.
Change-Id: I0086189308d2f5e4b321c63f24110d74cda6433c
Reviewed-on: https://gerrit.openafs.org/13254
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
afs_VerifyVCache is a macro that (on most platforms) effectively
expands to:
if ((avc->f.states & CStatd)) {
return 0;
} else {
return afs_VerifyVCache2(...);
}
Some callers call afs_VerifyVCache2 directly, since they already check
for CStatd for other reasons. A few callers currently call
afs_VerifyVCache2, but without guaranteeing that CStatd is not set.
Specifically, in afs_getattr and afs_linux_VerifyVCache, CStatd could
be set while afs_CreateReq drops GLOCK. And in afs_linux_readdir,
CStatd could be cleared at multiple different points before the
VerifyVCache call.
This can result in afs_VerifyVCache2 acquiring a write-lock on the
vcache, even when CStatd is already set, which is an unnecessary
performance hit.
To avoid this, change these call sites to use afs_VerifyVCache instead
of calling afs_VerifyVCache2 directly, which skips the write lock when
CStatd is already set.
Change-Id: I7b75c9755af147b42a48160fa90c9849f2f03ddb
Reviewed-on: https://gerrit.openafs.org/12655
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
If osi_linux_free fails, it printf's an error message, then calls BUG().
This is the sole open-coded call to BUG() in OpenAFS; all other calls
to BUG() are indirect via osi_Panic().
For consistency, eliminate this direct BUG() call by replacing the
printf and BUG() with an equivalent osi_Panic(). This also ensures that
the error messsage is logged as critical, and prefixed with "openafs:".
Change-Id: Id319dffa859308528a66991bbbc522ca49552d51
Reviewed-on: https://gerrit.openafs.org/14250
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: Benjamin Kaduk <kaduk@mit.edu>
Linux-5.8-rc1 commit 'bdi: remove the name field in struct
backing_dev_info' (1cd925d5838)
Do not set the name field in the backing_dev_info structure if it is
not available. Uses an existing config test
'STRUCT_BACKING_DEV_INFO_HAS_NAME'
Note the name field in the backing_dev_info structure was added in
Linux-2.6.32
Change-Id: I20b80e49e8a15a2949003101f24d9ce39f63b59b
Reviewed-on: https://gerrit.openafs.org/14248
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Linux 5.8-rc1 commit 'net: remove kernel_setsockopt' (5a892ff2facb)
retires the kernel_setsockopt function. In prior kernel commits new
functions (ip_sock_set_*) were added to replace the specific functions
performed by kernel_setsockopt.
Define new config test 'HAVE_IP_SOCK_SET' if the 'ip_sock_set' functions
are available. The config define 'HAVE_KERNEL_SETSOCKOPT' is no longer
set in Linux 5.8.
Create wrapper functions that replace the kernel_setsockopt calls with
calls to the appropriate Linux kernel function(s) (depending on what
functions the kernel supports).
Remove the unused 'kernel_getsockopt' function (used for building with
pre 2.6.19 kernels).
For reference
Linux 2.6.19 introduced kernel_setsockopt
Linux 5.8 removed kernel_setsockopt and replaced the functionality
with a set of new functions (ip_sock_set_*)
Change-Id: I517b674303c5decc19313d9de51d04ddef36b421
Reviewed-on: https://gerrit.openafs.org/14247
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
tests/auth/writekeyfile.c contains some code used to generate
tests/auth/KeyFile, which is used to test code interpreting the
old-style KeyFile format. This code currently has a few problems:
- We don't check the results of afstest_mkdtemp, which could allow
symlink attacks from other users on the system.
- We duplicate some logic from afstest_BuildTestConfig, in order to
build a temporary config dir.
- writekeyfile isn't built or run by default (it only exists to
generate KeyFile, so it's almost never run), so eventual bitrot is
quite likely, and the existing code already generates warnings.
To avoid this, change writekeyfile.c to use the existing
afstest_BuildTestConfig to generate a local config dir. To ensure we
avoid bitrot, build writekeyfile by default, and create a test to run
it, to make sure it can generate a KeyFile as expected.
Note that the KeyFile.short we test against is different than the
KeyFile currently in the tree. The existing KeyFile was generated from
an older OpenAFS release, which always generated 100-byte KeyFiles,
even if we only have a few keys. The current codebase only writes out
as much key data as needed, so the generated KeyFiles are shorter (but
still understandable by older OpenAFS releases).
Keep the old 100-byte KeyFile around, since that's what older OpenAFS
would generate, and create a new KeyFile.short to test against, to
make sure our code for generating KeyFiles doesn't change any further.
Change-Id: Ibe9246c6dd808ed2b2225dd7be2b27bbdee072fd
Reviewed-on: https://gerrit.openafs.org/14246
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Commit "Build tests by default" 68f406436c
changes the build so tests are always built.
On Solaris 10 the build fails because nanosleep is in librt, which we do
not link against.
Replace nanosleep with usleep. This avoids introducing extra configure
tests just for Solaris 10.
Note that with Solaris 11 nanosleep was moved from librt to libc, the
standard C library.
Change-Id: I6639f32bb8c8ace438e0092a866f06561dad54f1
Reviewed-on: https://gerrit.openafs.org/14244
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Commit "Build tests by default" 68f406436c
changes the build so tests are always built.
On Solaris 10 Update 10 and earlier the build fails because the mkdtemp
function is not available.
Introduce a wrapper 'afstest_mkdtemp' that uses mkdtemp if available,
otherwise uses mktemp/mkdir.
Change-Id: I0118f838ed9a89927e2ddac4cad822574601558a
Reviewed-on: https://gerrit.openafs.org/14243
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Run git describe once at the beginning of make-release to find the
version information used to derive the tarball file names and saved in
the .version file.
This is a cleanup and refactoring change to prepare for a future commit.
Change-Id: I0debeeffa5d2c63ab1498588766cb36424d15cd5
Reviewed-on: https://gerrit.openafs.org/14150
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
Automatically create the --dir directory if it does not already exist,
which makes this script slightly easier to use. Remove the now
uneeded mkdir from the top-level makefile.
Change-Id: I1f4561120a70263b0b2b194e65fec55fb5666f40
Reviewed-on: https://gerrit.openafs.org/14115
Reviewed-by: Cheyenne Wills <cwills@sinenomine.net>
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Benjamin Kaduk <kaduk@mit.edu>
