CVS log for pvfs2/configure
	Help

Request diff between arbitrary revisions / Display revisions graphically

Merged robust security code into Orange-Branch.

fix configure

Fixing scandir and dlfcn issues for SL

OK trying to get this scandir thing right.  This should do it.

fixed scandir def to compile with both new and old versions
added step in iocommon_readorwrite to generate PVFS_Request

still not quite right ...

added configure items for acl includes

initial working versionof usrint code
auto configures usrint on an ucache off (not working yet)
new acl repimplemented - old one has #def
some minor format edits

backout super.c change, looks like a back configure script

added support for usrint and usrint-cache to build system

turned out to be C_INCLUDE_PATH environment issue, remove change

add back optimization flag to configure checks this time only when include of linux/fs.h breaks without -Os

fix for strange include path issue with opensuse 11.2

remove -O2 for now, breaks configure checks in kernels before 2.6.40

typo

add include path of the full kernel source (kernel_source/source) if a SuSE-style obj directory is given as the kernel path

configure change to support 3.X kernel series

updates to support kernels newer than 2.6.38. Changes include use of DEFINE_SPINLOCK, change file_system_type struct to use mount instead of get_sb, and add a flag to dirty_inode member of super_operations. In all cases, leave previous behavior in place when newer and older options both exist.

merge with Orange-Branch

small build change for libuuid and printing of temp change for handle printing in pvfs2-db-display

uuid converstation, PVFS_x_position changes, server/client side first pass done

changes to PVFS_handle_X

update configure checks and kernel code for 2.6.38

merge with Orange-Branch

regenerated configure that was broken by the previous change to it

Added unused attribute to several variables and functions to prevent the compiling
from complaining about it.

update kernel check for inode_setattr in configure

first pass of changes to support kernel changes in 2.6.36 and removal of the BKL. note: no locking has been added to the ioctls so it may not yet be safe

Provide warning during configure if version of berkeley db is not at least 4.8.30

enforce checks for binaries required in the PATH in configure

update version string in branch, update changelog

version string refactoring

Changes for release

release changes

merge with Orange-Branch

additional kernel module changes around xattr_handler and fixing of checks from HANLDER to HANDLER. Also add check for fsync in newer kernels that don't take second dentry parameter

add true/false enum if stddef.h doesn't have it defined as is the case in early 2.6 kernels

don't error out on lack of utsrelease.h file, just error out if we can't find UTS_RELEASE

merge with Orange-Branch

regenerated configure to include recent kernel.m4 changes to support newer kernels. kernel.m4 also finds autoconf.h in the new include/generated/, adds check for struct ctl_table changes, and adds check for struct xattr_handler changes.

update configure.in for new utsrelease.h location in include/generated/ and bump version in branch to 2.8.3

merge with Orange-Branch

Updated configure script

Updated configure script

Modified configure.in for release 2.8.3, updated pvfs2-config.h.in, and reran
prepare to generage configure.

Modified configure.in with PVFS2_VERSION_SUB=3 and added PVFS2_VERSION_RELEASE
to pvfs2-config.h.in and then regenerated configure

JNI patch applied.

Temporarily changed configure to set the cflags with -O0 instead of -O2

orangefs experimental 20101012 release (distributed directories)

Merged in changes from Orange-Branch. There were a number of bugs fixed there since this branch was created.

Fixed warnings when compiling on 64 bit

merge with Orange-Branch

orangefs 20100920 release

change configure back to non-release

release 20100920

merge with Orange-Branch

update version string in Orange-Branch

merging Orange Branch changes in

initial merge with Orange-Branch. much will be broken

kmod builds for 2.6.33 are busted but this will at least get configure running
again

added stubbed security functions so security can be disabled

removed old enable/disable security options

regenerated configure...forgot the prepare statement!

regenerated configure with current version of configure.in using autoconf 2.63

revert cu-security-branch to before the attempted merge with Orange-Branch

Addition of src/apps/devel directory and pvfs2-display-db to give a dump of berkeley dbs in text form to assist in development/debugging. Also made configure and make changes with new build target develtools

merge of B2O-Blue-Sync,HEAD,and latest of Orange.

Added --with-cuda and --with-cuda-sdk optional compilation to the configure script.

forcing new version of configure into CVS.  I regenerated it from configure.in.

cvs test

Added a conditional compilation check for enabling/disabling CUDA, --with-cuda=<dir>
Modified the kmeans.sm to make it compile depending on --with-cuda option.

add 2.8.2 changes to ChangeLog.  Update configure for 2.8.2 version.

merge 2.6.32 kernel updates from trunk to pvfs-2-8-branch

fixes for 2.6.32:
- configure test for backing_dev_info.name field
- configure test for bdi_init()  (first appeared several versions ago)
- trust kernel to handle dirty inode list automatically on unmount

fix for sb->s_dirty kernel config check.

fix for sb->s_dirty kernel config check.

updated kernel.m4 needs new configure.

fixes for 2.6.32.

This includes primarily two build fixes:

* sysctl proc handlers >2.6.31 don't have a file pointer parameter.  Added configure check and proper #ifdefs.

* The writeback bdi changes introduced in 2.6.32 move parameters from the super_block struct to the bdi_writeback struct in backing-dev.h.

The remaining pieces of BMI ZOID support.

Regenerate configure script for pvfs-2-8-branch to sync up with m4 files.
Compiles cleanly against latest 2.6.31.4 kernel now.

*** empty log message ***

revert earlier commit re: driver class support; seg faults on some kernels

Modified version of patch from Nick Mills; detect different variations on
device class api, but disable class functions entirely for RHEL4 kernels.

Updated configure.in and configure to recognize src/apps/user/module.mk.in

re-generated configure script

Merged NWU's CUDA code.
Modified configure script and Makefile to enable CUDA compilation.

Merge with work from this summer.  Metadata and data storage spaces can now be located on different disks.  Fixed a small bug in string encoding.  Updated test code to reflect new Trove dual-storage space API.  Genconfig updated as well as affected admin apps. In the correct branch this time.

merged in changes from latest Orange-Branch

merged in changes from summer at LANL

patch from Bart Taylor; moves permission checking code back to common
directory in order to add safety checks for truncate operation permission

Merged Phil's tcp-peer-comm-branch that allows simultaneous server to server communications.
Added basic allreduce module:
  - implemented a recursive distance doubling algorithm.
  - works only when the # of servers are power of 2.
  - when activated, servers simply exchange their id to each other.

TODO
  - handle arbitrary sizes of buffers of various data types.

Setting up Orange Branch

initial support for solaris.  doesn't include /dev/epoll support.  See trac ticket #97.

fix typos in new arch detection logic, reported by Maxence Dunnewind on IRC

Experimental fix to recent compile problems; copy the kernel makefile's
logic for detecting host arch and use it to add extra include paths for
configure tests.  Skip if user manually specifies ARCH for cross compile.

This needs testing.

configure test to handle the fact that 2.6.29 no longer has fsuid/fsgid
fields in the task_struct structure

minor progress towards 2.6.29 build; at least get include path right for
configure if you manually specify ARCH=x86 (or similar)

version set for pending 2.8.1 release.

merge changes from 2.8 branch to HEAD.

tagged release point, update version.

set version for release.

fixes for 2.6.28 kernel.

merging BUILD_KERNEL updates and sys-io.sm logging updates to 2-8 branch

updates to BUILD_KERNEL flag: move to positive section of --with-kernel
test in configure.in, add to --with-kernel24 test, and add Makefile.in
flag to pick up at AC_SUBST time.

trac ticket #67.

merge of Sumit's pvfs2fuse code to HEAD.

disable events by default

srcdir only gets output by configure (only available to files it generates).

fixup the compat header to be sensible.  Also bump the minor version in lieu of a pending release.

applied bmidist patch from Dries.  Also added a number of cleanups of my own.

Make pvfs2.h include the version numbers to be exported.  This allows consumers of the sysint calls to determine which parameters to include.

merging hints/events code into trunk.

bmi changes to allow separate bmi packaging

pull in direct IO changes from HEAD.

add tau formatting code from Aroon.

merge of direct IO code to PVFS HEAD.

patch from Sumit Narayan to fix remaining compilation problems with 2.6.27
kernel related to permission function changes

2.6.27 adds a 5th way to lock address_space mappings

New configure test to fix one of the compile problems with 2.6.27
kernels.  The kmem_cache interface change that we detected in 2.6.24 has
changed yet again.

reverse merge of changes to HEAD since branch.  Includes small file changes.

merge trunk updates down to small-file-branch.  Passes basic tests but needs
some double checking of pint-cached-config and sys-create conflicts.

Modified version of patch provided by Sumit Narayan to fix bug reported by
Ryuta Suzuki.  Allows pvfs2 kernel module to build against 2.6.26.

updated configure.

adding threaded odirect impl

Added configure flag "--disable-security"
Removed configure flag "--enable-security=[none]"

Reverse merged and ported to HEAD.

Added fsx-mpi.c - tests PVFS2 file system semantics using MPI.

reverse merge of trunk changes since branch.

reverse merge latest trunk changes to small-file-branch and update msgpair
usage where needed

Merge from HEAD.  Fleshed out credential.  Encode/decode added for credential and cleaned up for capability.  Added capability/credential into request/response structures.  Fleshed out comments in security module.

configure tests for noatime mnt flags (needed for RHEL4 at least)

Patch from Murali to implement file_aio_write_iovec as well as export
operations for NFS export on newer kernels.

Still need to test on older kernels to see if additional configure tests
are needed.

marked the following configure options as experimental:
--enable-mmap-racache
--enable-trusted-connections
--enable-kernel-sendfile
--enable-threaded-kmod-helper
--disable-aio-threaded-callbacks

Added --enable-security=[none, rsa, dsa] into configure.in to enable toggling of security features for export/import.

reverse merge of latest changes from HEAD to he branch.

Updated branch with code from HEAD

updated build system for security module

Remove deprecated --disable-disk-io option to configure; this hasn't
actually been supported in the code in a long time.  There is now a
runtime selectable trove method (null-aio) to accomplish the same thing.

modified version of configure test from Scott Atchley to make configure fail
if mx_decompose_endpoint_addr2() is not found.

merge i_sem/i_mutex configure check from trunk to pvfs-2-7-branch

configure test to account for 2.6.15->2.6.16 change from using inode->i_sem
to inode->i_mutex for inode locking

one more minor configure test merge from trunk

merge a collection of configure test updates from trunk to pvfs-2-7-branch;
this fixes the kmem_cache tests for platforms like rhel4

in preparazione per 2.7.1.

syncing small-file-branch back up with trunk at small-file-branch-point2 tag
(reverse merge)

merge a bunch of fixes made to HEAD since 2.7.0.

merge kmem cache create two param changes to 2.7 branch.

library versioning

This patch adds a version code to the installed shared libraries,
libpvfs2.so and possibly libpvfs2-threaded.so.  Library versioning
is considered a good thing, in that apps can know what version of the
library they need.  Use the actual PVFS version numbers to build
the library version string.  This is not always the right thing to do,
but works for us because we pay attention to the meaning of "major",
"minor" and "release" changes.  Adapted from a patch by Justin Bronder.

crypto not crypt

Long-standing bug pulls in unneeded libcrypt.  We want libcrypto
however, which was pulled in by libssl via gssapi, but make it
explicit as pvfs calls it directly.

kmod: unused extra gcc flags

This should have gone away as part of "kmod: remove Werror tests".
Now that the tests do not rely on -Werror to get the right result,
we don't particularly care if the kernel throws extra (spurious)
warnings.

build: disable static server fix

The configure option --disable-static-server was not doing the
right thing.  Patch by Justin Bronder.

build: use LIBS not LDFLAGS

To test if server needs -lrt, and in general, everyplace, we
should test-alter LIBS.  From a patch by Justin Bronder.

kmod: remove Werror tests

A couple of kernel configure tests that looked at linux/fs.h were
failing due to spurious warnings from elsewhere in the kernel, at
least on a recent rhel5 gcc and 2.6.24-rc1.  Rewrite these to avoid
using -Werror by identifying more easily testable features and
testing for those instead.  Easy means generates an error, not just
a warning, and thus is easier to confirm in the presence of other
unrelated warnings.

merging device class configure test updates from trunk

updated device class configure test to make sure we don't give a positive
result on RHEL4 kernels

kmod: read inode

From Sam.  2.6.25-rc1 removes the .read_inode member in
struct super_operations.  We can continue to use the same
function, just not include it in the structure.  Add a
configure test for this too.

merging ARCH configure time detection from trunk

merging ARCH env variable handling from trunk

make configure detect ARCH environment variable and adjust include paths to
match what the kernel module will get at build time

threaded kmod helper

Add --enable-threaded-kmod-helper option to build a threaded
pvfs2-client-core, that is, an executable linked against
pvfs2-threaded.so.  Take care to make sure only what libraries
are needed based on --enable-static, --enable-shared, and this
new option.

Remove the "--threaded" argument from pvfs2-client.  Whatever version
of client-core that was configured is all that is available at runtime.
It will always be installed as pvfs2-client-core.

Sam's prelude and scheduler updates

gm-mx-autoconf

Convert GM and MX to use autoconf tests to find includes and libs.  Move
these tests into new config files to keep the top-level Makefile clean.
This fixes a configuration problem with GM on x86_64 where libs are
in lib64/, and generalizes both GM and MX to be a bit more clever at
autodetection.

Signed-off-by: Pete Wyckoff <pw@osc.edu>
---
 Makefile.in                    |   32 +++-----
 configure                      |  167 ++++++++++++++++++++++++++++++++--------
 configure.in                   |   49 ++----------
 maint/config/gm.m4             |   72 +++++++++++++++++
 maint/config/mx.m4             |   73 +++++++++++++++++
 src/apps/admin/pvfs2-config.in |    8 +-
 src/io/bmi/bmi_gm/module.mk.in |   45 ++++++++----
 src/io/bmi/bmi_mx/module.mk.in |   40 +++++++---
 8 files changed, 363 insertions(+), 123 deletions(-)
 create mode 100644 maint/config/gm.m4
 create mode 100644 maint/config/mx.m4

Avoid accumulating tested libraries in LIBS; it breaks future tests.

fixes for segv backtrace checking.

fix borked configure tests for aiocb return and error_code fields.

Following a patch by murali, fix the changed signature on kmem_cache_create
constructor functions in the kernel.  Also add missing include for struct
kmem_cache declaration needed on new kernels.  Please test if you have
something old.

Fix struct kmem_cache test, as much as it can be, as it was indeed previously
broken for old kernels that did not have this type---the test always succeeded.
Define the result in preparation for fixing kmem_cache ctor changes in 2.6.24.

[murali] Make checking message more descriptive.

Accommodate new rwlock type in struct address_space.  From murali with
fixes.

make segv-backtrace enabled by default

revert commit of repair stranded util from Phil.  We need to figure out the right way to do shared state actions first (probably nested sm).

allow kmod to build against kernel with rt patches

get the versioning right for MAIN.

preparing for 2.7.0.

from HEAD: fix the exportfs.h header check

ok, i think i got this configure test correct at last

from HEAD: cherry-pick the "make it build with 2.6.23" fixes

linux 2.6.23 introduces exportfs.h

fix kmod compile with kernels earlier than 2.6.23

Include location bug for Portals.

generated file

it's certainly confusing if we print "yes" for both success and failure

dare i dream that this is the last feature test needed for linux-2.6.6 kernels?

fix test

somewhere between 2.6.6 and 2.6.20 the 'struct page' changed the 'count' member
to '_count'

add a feature test for the FS_IOC_* flags.  These were turned into generic
versions sometime around 2.6.20 and don't exist on 2.6.6 (where I happen to be
compiling pvfs today...)

Warn people about the current polling situation in multi-BMI-device setups.
Users keep discovering this, and until we implement a separate thread per
device in BMI, PVFS will always perform poorly on both devices.

use perl's Math::BigInt module instead of assuming perl has 64bit int support.  Check for BigInt in configure.

remove mistaken add of kzalloc check.

Move a big chunk of configure out to a separate M4 include:  --with-ib.

Hide PVFS2_SEGV_BACKTRACE option down in server code where it applies.
Fix little gossip module.mk.in thinko.

Always define gossip_backtrace(), just as a noop if it cannot be built.  This
simplifies some ifdef nests in a few spots.  No behavior change.

Keep bmi-tcp epoll issue out of global cflags.

Move gcc options for --enable-strict from configure to Makefile, so they
are near the other build flags and so non-GCC compilers can be handled
differently.  Also redefine --enable-strict to apply only warning options,
not to turn on -g too.  And the option works with CFLAGS= override on the
command line too, unlike --enable-fast.  Finally, set default CFLAGS to
"-g -O2", adding the -g in hopes we will get better tracebacks on errors.
This will increase code size somewhat.  Builds that use --enable-fast or
custom CFLAGS are not affected.

generated files

generated files

Generated file

Generated files (big, sorry)

mx fixes

generate huge configure with my ancient fc6 autoconf

Merge HEAD changes to TAS-branch.

typo in the gcc version logic

merge lk v21 fixes to branch

fixes for linux kernel version 2.6.21

forgot a space between these args

regenerate configure script

check in generated file

the -Wno-pointer-sign compiler option doesn't work on BGL, and is no longer
needed on FC5

update to Changelog and increased major version in anticipation of _major_ changes :-)

fixes to build on darwin

walt3 reverse merge from head includes merges of pw's sm changes (no state declarations), cleanup of state machine code, and other general merging/fixes.

fixup berkeley db invocations to be backward compatible with older versions of bdb

fix kzalloc checking in branch

Upgrade to current CVS version..

updates to changelog, up-ed version to 2.6.3, regenerated configure.

forgot to check in generated file

fix bad kzalloc check.

get the f[gs]etxattr checks right so that older platforms build (sles8)

Update migration branch to current CVS version

Sync hint-branch to current CVS version

Initial commit of BMI MX implementation provided by Scott and Myricom.  Thanks Scott!

this should get the branch building against the recently released 2.6.20 kernel

kzalloc checks

commit checks for kzalloc that older kernels don't support

merge 2.6 branch changes to head

regenerated build files

update configure to go with change to bdb.m4

merge a bunch of minor fixes pete made to head recently.  See the changelogs from Dec 15th to Jan 15th for more info (esp. Dec 19th).  Also updating the changelog and minor release version number for impending release (2.6.2).

Synchronization with HEAD

revert changes that duplicate pete's fixes for 2.6.19

commit Murali's fixes for 2.6.19 and the fs function pointers:

http://www.beowulf-underground.org/pipermail/pvfs2-developers/2006-November/002900.html

generated files

avoid false positive on ioctl32_conversion

set version for 2.6.1

merge pw's fixes to the 2.6 branch.

generated files

Upgrade to current CVS version

fix problems on ppc64:

* make all segment count variables unsigned long
* cleanup warnings caused by config checks failing due to warnings
* cleanup warnings caused by invalid llu macros defined from configure

from HEAD: 2.6.18.1 no longer provides linux/config.h

merge latest changes in 2.6 to head.

update version

commit of new configure

from HEAD: deal with the 2.6.18.1 removal of i_blksize (turns out this approach
is fine after all)

Trove TAS modul added.

commit generated files

use an otherwise unused variable to get another configure test passing that was
failing for the wrong reasons.

do something in config test to silence spurious warning treated at error.

in 2.6.18.1 kernels, linux/config.h now complains whenever included.

Ok, this gets us compiling again, but is almost assuredly not the correct way
to deal with the removal of i_blksize

reverse merge of HEAD to WALT3 branch.

autogen file

[phil]: added ACCESS_DEBUG option to print hostnames instead of IP addresses in debug output.
[slang]: fix to previous commit of debug statement in remove.sm

merge from 2.6 branch

changed version to 2.6.0.pre...  Also changed tarball and dir names for packages to pvfs-...

More cleanups of the non contig vfs interfaces.
- Consolidate readx and writex paths similar to the earlier consolidation for regular read and write
- Remove the conditional HAVE_FILE_READX and HAVE_FILE_WRITEX and redefine xtvec structure so that
  we could reuse this for doing large I/O through the page-cache with non-contig file offsets
- Added configure time checks for xtvec structure to determine whether to redefine xtvec or not..

bugfix: adding pthread library for mpe

- Support for immutable files
- Cleanups in the pvfs2-bufmap interfaces (replace int with size_t)
- Utility for displaying extended attributes and setting them
- Support for disabling atime updates at the object level, file system level (mount time)
  and directories.

backmerging of HEAD to branch...

Patch addresses the long standing issue of handling 64 bit handles in the kmod
without collissions. This basically involves migration from an iget() interface
to locate inodes to the iget(45)_locked() interface. The basic problem boils down to
not depending on inode->i_ino member field for the handle.

Replace all occurrences of inode->i_ino to the appropriate function for extracting the
actual 64 bit handle.

Patch also fixes a bug in lookup by replacing a return NULL with a return to ERR_PTR(-ENOMEM)

Fairly well tested on 2.6 kernels on x86 and x86_64 machines. Minimal testing on 2.4 kernels.

correct the openssl LIBS issue

fix for configure failures if ssl libs don't exist.

Merge HEAD into Walt's branch.
Rework new state machines to the new cleanups introduced by Walt.

updates to openssl checking.

Added a configure option --disable-openssl like Pete suggested both for toplevel
and test/configure that will prevent us from checking for the openssl libs/headers
even if they exist in the default paths..
Also fixed posix/sha1.c test program from doing any crypto stuff if configure says so.

Merge posix-extensions-branch to HEAD
This branch implemented patches to the 2.6.16 kernel for the proposed POSIX I/O extensions and those patches are under
the patches subdirectory.
It also implements the PVFS2 specific hooks for these system calls.

Tools that may be of immediate use to the pvfs2 general audience is the pvfs2-lsplus utility in src/apps/admin
that should be noticeably faster than the pvfs2-ls utility if there are a lot of objects in a single directory.
Other features are left out by configure and are not even built if the kernels do not support those features
and callbacks.

openssl checks performed even when --with-openssl is not specified.
Associated configure time changes.
Fix test builds when mpich2 paths are not specified.

fix error message typo

Merge from HEAD to posix-extensions-branch...
This should facilitate merging the posix branch to HEAD...

Added crypto digests utility functions and configure time checks for openssl libs and includes
so that pvfs2-ping utility can now optionally check whether fs.conf files are the same on
all servers..
i.e. pvfs2-ping -m /pvfs -c will now print whether fs.conf integrity checks fail or pass.
Some of the infrastructure used here will be useful for doing SIGHUP handling at the server
so that fs.conf or server.conf files may be refetched etc..

Alternative Trove implementation branch with Trove transaction support for db >= 4.3.

Modification for migration added, some functions still need to be adapted after moving to a new branch...

PVFS2-hint support + minor changes (verbose request scheduler)

change pre version back after tagging

configure changes for bgl version

Added support for the getdents_plus_lite and getdents64_plus_lite system calls.
and the pvfs2 hooks for it.
Turned out to be fairly simple..
Also added modifications to the getdents program under test/posix.
It now accepts a -l command line parameter to use the new system call when -p is also specified.

Reverse merges from HEAD..

build fixes on Chiba.

configure check for an openib API change

Backmerge with HEAD

Added PVFS-hint, changed tests for pwrite/odirect/transactions/aio dbpf
 support to allow users to explicitly enable or disable the functionality.
In the aio implementation is a bug since somebody else modified it.

Reverse merge changes from HEAD to posix-extensions-branch.

Fixes for the upcoming 2.6.18 kernel. pointed by RobL.
->get_sb() now takes a vfsmount structure and returns an integer instead of the sb encoded
as an error. We can now reference sb by peeking into the vfsmount structure.
->statfs() now takes struct dentry instead of a struct sb*. We can of course deref dentry->d_sb
to get to the sb.
configure changes to support the argument changes.

sigh.. at this rate it might be better to use fuse or merge pvfs2 in kernel :)

remove refs to ready queue

more of the same

Change default to aio, bugfix aio, I broke it for non-threaded, change
names to use underscores...

a first stab at 2.6.18 compatiblity.  not sufficient yet, but gets configure
detecting enough

Patch fixes the atime, mtime and ctime problems that people have
reported with PVFS2.
Patch changes the notion of who sets these times. Earlier behavior was the
clients providing the times to the server. New behavior is to have the
server set these times except in  the implementation of utimes() system
call whereby clients explicitly set the time (atime and mtime).
NOTE: We dont allow the clients to dictate the ctime setting at all since
that is consistent with all file systems.
So what we have is:
read, readdir -> atime update required
write, truncate -> mtime update required
ctime is updated on every setattr implicitly on the server.

Since there is no need for parent_*time fields in the server request for chdirent,
crdirent and rmdirent, the next step is to remove those fields, fix the endecode
macros and bump up the major version of the protocol.

Patch also implements these times lazily by flushing them only at the time
of a close rather than after every read/write which could be pretty
expensive.

Changed bzero to memset in trove, added O_DIRECT and PWRITE tests for
autoconf. (O_DIR test is not used in trove right now).

configure option --enable-aio determines if aio or dbpf-threaded (new)
is used... dbpf-bstream-threaded.c contains threaded I/O version.
This file need some revision... It will be used for testing O_DIRECT.

Checkin of previous patches

merge of fixes for db->set_errcall from 1.5 branch.  Also incremented the release version.

upgrade minor version

merge of murali's changes from trunk.

generated file

reverse merge of recent bug fixes from trunk

*sigh* the 2.6.17-rc6 kernel changes the return type of invalidatepdage.
Harmless to leave this as-is, but does take care of a warning.

Reverse merge from trunk to lockd-branch...

reverse merge of branch

turn on epoll by default (but do a check to make sure its there).  Add --[en/dis]able-epoll options to prevent the check.

reverse merge of trunk to branch

merges fixes that went into head for consolidating I/O paths (read, write) and (readv, writev)

Also contains the hooks for implementing readx, writex natively in PVFS2. (the vfs patch is part of the patches
sub directory). This involved adding the following infrastructures
a) Have a way of sending a variable length trailer as part of the upcalling mechanism. Currently the readx, writex
implementation upcalls the file offsets, length pairs as this variable length trailer.
b) Client-core has a new upcall request (called FILE_IOX: very unimaginative name :))
which now does a isys_io with file_req set to req_hindexed as opposed to contiguous for FILE_IO upcalls.
c) Client-core does a read of the upcall from the device, if it finds that trailer is present, it does another read
from the device.
d) kmod which maintains the pvfs2_kernel_op_t structures in the request list needs to keep the
structure around in the list (and not insert in the in-progress hash table) if the op has an upcall trailer.
kmod has hooks for feeding either the op or the trailer depending on the which read call comes into the device.
It also has hooks for delaying adding the op to the hash table until the trailer is picked off the device.
e) Added a bunch of test programs for the new readx/writex system calls.. Seems to pass tests on an x86 laptop!

Also contains fixes and cleanups of the readdir/readdirplus implementattion which required a variable length trailer as
part of the downcall. These cleanups enable a very symmetrical implementation of variable length upcall
and downcalls!

generated file

reverting patchset 5116 (make epoll the default) because the configure test
passes ernoneously on linux-2.4

let's make epoll the default where available

looks like autoconf-2.13 generates broken configure scripts now.

use the formerly unused variable so we pass or fail this test correctly

Reverse fixes from trunk for build fixes with 64 bit compat platforms on 2.4 kernels.

Lee Whatley reported that the compat 32 patches did not build/work correctly on old(er) 2.4 kernels.
Fixes to address that. Basically 2.4 kernels did not have many of the functions that 2.6 has to allocate
user space addresses on the caller stack (perhaps, the ABI itself has changed or something, dunno).
Anyways, work-around is to invoke the handler directly instead of calling sys_ioctl..

Critical bugs exposed by Opteron experiments.
includes premature freeing of the state machine message pair array in readdirplus which were somehow not revealed
on x86!?
Added lib64 to openssl.m4 so that we also look for these libraries in the 64 bit path prior to the 32 bit paths.
If openssl was not present or not configured, make fskey_get return a NULL key and 0 keylength instead
of a random pointer that will cause us to crash with a seg fault..
All test module.mk had to be augmented with -lssl -lcrypto to get them to build on opterons.
Maximum number of directory entries (requested at 1 time) cannot be larger than maximum number of handles
that we can handle per server list attr request. Dont know how I can make this a compile-time test...
Fixed a few warnings in the test programs for Opterons.
Upshot is that with this patch, all getdents/getdents64/getdents_plus/getdents64_plus work correctly on Opterons to
the best of my knowledge on the sample workloads ran!!

Includes reverse merges from trunk
in addition, it includes fixes for readdir and readdirplus to have a variable length
number of directory entries passed in from client-core using a trailer page (currently).
However, there is nothing that prevents us from sharing those pages with the kernel using vmap
and friends which is the next logical step.
So client-core now does a readdir/readdirplus and writev's the trailer page (which is essentially
an encoded version of the readdir/readdirplus response) to the kmod.
The kernel module decodes that and copies it to the user-space app (which issued the getdents/getdents_plus
system call)
Added/edited the getdents.c test program in the test/posix sub directory to issue
getdents/getdents64/getdents_plus/getdents64_plus system calls
Also fixed a critical bug that was somehow never triggered in readdir if the buffer size
provided by glibc/user was not sufficient enough and we advance the f_pos token beyond where we stopped.

avoid -Werror but still test for function signatures inside structures by looking at related functions that were changed at the same time

do not always define _GNU_SOURCE if strnlen prototype is found

add explicit test for strerror_r; _GNU_SOURCE was not determining the proper version correctly

First cut of the sys_readdirplus implementation where clients can issue a single
system interface call to get the directory entries as well as the attribute information
of the directory entries.
What this means is that we do not need to call a sys_getattr of the individual directory entries
anymore.
(The kernel version is still untested and won't work until the VFS patch is in place! :))

On the server, there is now a new request/state machine called listattr(poor name, i know!)
which essentially does a vectored getattr for a list of handles as opposed to a getattr
which only does for 1. Perhaps, we need to consolidate them into 1 at a later point of time after
others have given this code a look-see!
On the client, there is a new system interface, and a corresponding statemachine implementation
called readdirplus which does the following algorithm
a) Step 1 - Do a regular readdir
b) Step 2 - Collate all meta handles (obtained from the readdir) and send out 1 listattr
to every server that owns atleast 1 handle from the set.
c) Step 3 - Handles that refer to metafiles (determined by the attributes of the response)
need an additional step to retrieve their file sizes. So we collate all data file handles
of all metafiles, and send out 1 listattr to every server that owns atleast 1 handle from
the set.
d) Step 4 - Copy object attributes obtained for all objects into the sys response structure and
also compute file sizes of regular files since we know the individual dfile handles' file sizes
from Step 3.

Clearly, this is a huge win in terms of performance of listing directories which have large number of directory
entries.
Next step is to hammer out the VFS patches and coreutils patch for ls to get the same benefit.
Also added a new utility called pvfs2-lsplus in src/apps/admin that uses this API.

Added hooks for vfs getattr_lite() callback to implement statlite() and friends of the POSIX
I/O system call extensions in PVFS2
and the necessary VFS infrastructure and system call stubs.

make AC_TRY_COMPILE fail if test routines call an implicitly defined routine.
this makes touch_atime test fail (like it should) on RHEL 3.

get branch to build

Added a few more missing files and fixes from HEAD to lockd-branch.
lockd-branch should build cleanly now.

Added synch. server files to the branch

PVFS2 specific hooks for implementing openg and openfh.
This includes callbacks for
a) fill_handle as part of the inode operations structure.
b) find_inode_handle as part of the super operations
c) get_fs_key as part of the super operations
and statfs_lite() also as part of the super_operations structure.
Credit to Sam for implementing a secure key per fsid that is used for the crypto
verifications of the handle buffer.

Timing information is also added so that every time critical operation
is logged for openfh/openg.

Patch disables sendfile system call support by default from PVFS2 since it has negative fallouts
on NFSD performance.
However, this also means Apache would not work off PVFS2 unless an EnableSendfile option
is turned off for Apache.
However users are still given a configure time option to --enable-kernel-sendfile
if they don't plan on re-exporting PVFS2 through NFS.
Updated FAQ to reflect this change.

commit regenerated (not stale) config file

finally get that damn touch_atime configure test correct

Regenerated configure script

onekeyval merge

reverse merge from trunk.  see if sio tests pass now as well.

fixes for darwin with latest AIO_THREADED support turned on.  Use union sigval instead of sigval_t.  Check for extra args to fgetxattr.

take a BIG RED HAMMER to the "test for working aio callbacks" code.  We weren't
testing for working aio callbacks.  we were testing for known-good versions of
glibc.  The approach was unable to handle glibc-2.4 and Darwin's glibc.  Even
Debian Stable ships with a new-enough glibc that aio callbacks work.
--disable-aio-callbacks is there for the poor folks on ancient systems.

Feel free to commit a better approach.

shovel all the kernel featuers (almost 8k worth of text!) into a separate
function.

2.6.16 (and the FC5 2.6.15-1.2054_FC5 kernel) no longer provide update_atime.
Use touch_atime, which appears to be in all 2.6 kernels.  Add a configure-time
test so we can fall back to the older update_atime method.

look in /lib64 and elsewhere for libc too

test for fgetxattr prototype. turns out it could be in <sys/xattr.h> on some
older systems.

fix compile error and merge from trunk

mostly working onekey impl

from HEAD: (minus typo) deal with really old systems that might not have
            fgetxattr(2)

fgetxattr doesn't exist on some really old systems.  Since those old systems
won't have xattributes anyway,  let pvfs2-viewdist fail gracefully

bump the version number

merging with pvfs2-1.4.0 branch now that it's released

1.4.0 release is imminent!

ib: separate config options for includes and libraries, optionally

guh! -Wdeclaration-after-statement is only available in gcc-4.0

even though declaration-after-statement is valid C99, it causes problems when
the inevitable gcc-2.95 user tries to build pvfs2

merging phil's patches into 1-4-0-branch

commit of phil's dirent remove patches.  See:
http://www.beowulf-underground.org/pipermail/pvfs2-developers/2006-January/001778.html.

*** empty log message ***

bumping version number

added --enable-static-server to config

merging changes from HEAD to dbpf branch

work around some kernel api changes

add configure check to detect read_descriptor_t changes

add config test for HAVE_PROC_HANDLER_SIX_ARG missed from previous patch merge.  Add enabled declaration missed in acache.c as well.

regenerated configure

Some systems (Debian, Ubuntu) don't have the attr/xattr.h header file, causing
a build failure.  On these systems, the fgetxattr symbol is in libc.  Murali,
if there's a better way to fix this, feel free

Added a sendfile callback to the pvfs2 file_operations structure so that Apache
can work off PVFS2 volumes.
This callback does not make use of the page-cache to avoid any of the associated
consistency problems.

merging in 1.3.2 changes from branch

three questions in 24 hours... let's just have a new release and be done with it

we need a way to disable kernel aio on some broken kernels (SLES 9)

commit generated files

commit generated stuff

stuff from trunk.

merging in fixes to berkeley db checking from 1.3.1 branch

more fixes to berkeley db checking.  This fix adds the correct -I<path> to CFLAGS if the --with-db option is specified at configure and another db.h exists in the standard include path.  Also, adds -lpthread to link tests (in some libdb installs its required).

fixed a bug in flow caused by outsize being updated correctly in trove now.  Also, s/transfered/transferred/.

commit fixes to 1.3.1 branch as well

fix for berkdb checking when --with-db is used

fix to not set NEEDS_LIBRT when --disable-server is present

fix --disable-server bug introduced with NEEDS_LIBRT

get the version string correct on the branch

make changes to 1.3.1 as well

fix LP64 issue with Lu/Ld/Lx that I borked with the darwin patch.

bump things for future development

just about to tag release

* malloc.h is obsolete in freebsd, check for it
* No error.h in freebsd
* linux allows Lu,Lx, and Ld to be interchangeable with llu,llx, and
lld.  Other platforms (solaris, bsd) don't.  Replaced all the L*
with ll*.
* bsd doesn't have mntent.h.  use fstab.h and our own funcs.
* bsd doesn't know about ssize_t.  Use size_t instead.
* bsd doesn't have asm errnos.  Check for each of those
errno macros and if not defined define to -1
* check for db3/db.h db4/db.h header if db path is specified using
--with-db
* darwin has aio support builtin, doesn't need -lrt
* remove kernel stuff in pvfs2-types.h that sets WORDSIZE.  gcc
defines __LP64__ on archs with 64bit longs so we use that instead.
* darwin's open doesn't accept O_LARGEFILE.  off_t is an int64 by
default.  define O_LARGEFILE if it hasn't been.
* need to run ranlib on darwin after libpvfs2.a gets installed to regenerate
table of contents

fix spelling error that broke my build

regenerated config scripts after moving berkeley db checks

make this test not always fail

Attached patches implements trusted ports and networks similar to
what PVFS1 implemented.
It adds a new <Security> </Security> tag to the fs.conf config file
and parse utilities accordingly. We can edit these later if there
is a strong need.

Within these tags, you have to specify TrustedPorts <Range of ports> and
TrustedNetwork tcp://<network> tcp://<mask>.
I dont know how to support this cleanly on  non tcp/ip bmi
architectures, so currently this only enables it for the bmi_tcp module
alone. (i guess we could do it for bmi_ib module as well but not done
until someone says that is the right thing to do...)

Also adds support for a --trusted argument to the pvfs2-genconfig script
that will prompt for the trusted port ranges and the network address/masks and write it out
to the fs.conf and server.conf file.

Configure now supports a --enable-trusted to enable all these features.
(NOTE that the fs.conf tags (or lack of one) still allow you to not
select these features even if configure was asked to compile with this
option).

this patch also cleans up all the warnings that arises with gcc
(3.4) and --enable-strict build mode.

Added configure options to feature detect aio support in the kernel headers
We dont enable aio for 2.4 kernels and for 2.6 kernels we disable aio support
for early 2.6.x kernels (where x < 8, I think)
	The configure test checks whether the kiocb structure provides a ki_dtor()
	method to cleanup operations such as aio_cancel and for freeing allocated
	memory. We use this method to determine whether to disable or enable.
Hopefully, this should get us to build on older 2.6 kernels..

bump version for further devel

- list our changes
- bump version number

bump the version number

[pcarns]: configure time test for prototype of setxattr: sometimes it has a
const argument, sometimes it doesn't.

needed to redo the include logic for kernel feature tests a bit.  When testing
for certain linux kernel headers, jump through hoops to avoid bringing in
system headers (e.g. xattr_acl.h can be both a kernel header and in
/usr/include/linux)

*sigh* ... linux/xattr_acl.h was removed in 2.6.13, so now we have to
AC_CHECK_HEADERS it

added murali's xattr patches.

feature-test for extended attributes. thanks murali

more work to get feature testing working on more platforms

merge changes from trunk since branch point into branch.  Effectively moves the branchpoint to the latest trunk.

[pcarns]: add a new bmi method that can resolve connections from unexpected
requests into hostnames

bump for more development

arbitrarily pick 1.2.0 for version number.. i *think* the wire protocol is the
same, but bump minor to reflect significant development

Added link check for valid mpe libs

*sigh* don't forget to check in generated configure

fix AC_TRY_COMPILE failures that check args for db->stat so that warnings cause failure.  --enable-strict should work now.  (no, really)