- The exportfs code is updated for the new export ops. The changes are made
conditional on a new configure test that detects the new ops. fh_to_dentry()
basically replaces decode_fh and uses our own get_dentry function instead of the
now defunct find_exported_dentry.
- A check for fh_len=4 is removed - in testing this value is always 6, possibly
because of changes in the kernel code.
- The check for authtab in osi_nfssrv.c assumes that an undefined weak symbol is
0. On my system, an unresolved weak symbol in a loaded module gets the value
0xfffffffe (-2 or -ENOENT) - again, probably a change on the kernel side. Check
that the pointer is not an error constant using IS_ERR().
- In osi_vfsops.c, only use the export_ops bits if building the translator
afspag: the problem here was dealing with unresolved symbols
- afs_showflags is redefined in afs_pag_call.c so it's available for the afspag
module
- A new source file afs_warn.c gets the afs_warn* functions from afs_util.c.
This allows the afspag module to get the afs_warnuser function without dragging
in too many symbols. The new file is attached separately.
other:
- d_path() now takes a struct path argument - includes a new configure test
- osi_vfsmnt -> osi_vfsmount in osi_misc.c Looked like a typo?
- reorder the remove_proc_entry() calls - we need to remove the children before
the parent (introduced with my earlier patch)
When probing in configure for cpp, use the user's PATH first and then
the /lib and /usr/ccs/lib directories. Set PATH_CPP instead of CPP,
since the latter is just overwritten by Autoconf. Use PATH_CPP by
preference in rxgen and when pre-processing assembly, since otherwise
newer gcc will treat the input file as linker input and not do anything
with it. Don't assume that cpp writes to the second argument; gcc
apparently doesn't always do this. Instead, redirect standard output.
Fix a disconnected-mode merge conflict with an older import of disconnected
to HEAD that caused a duplicate definition of PDiscon. I believe this
problem was only on HEAD.
Use the correct subcommand name (fs getacl instead of fs_getacl) on the
HTML index page, even though we have to use the underscore in the NAME
section of the actual POD documentation.
Use undef in the arch-specific param file to override UCONTEXT rather
than adding a conditional in a generic file, making arm work the same
as 32-bit SPARC. Disable UCONTEXT on parisc as well; glibc doesn't
provide the ucontext functions on that architecture.
Current Linux kernels always append a period and the PID to "core" when
creating core files for multithreaded processes. Teach bosserver to scan
the directory into which core files are deposited looking for such core
files and treat them like any other core file if the PID matches the
process that just died.
This patch has been in the Debian OpenAFS packages for some time without
problems.
Add additional missing commands to the to-do list. Add a section on man
page section numbers and their rationale. Document that embedding a
license in the man page isn't required if it's one of the licenses in our
LICENSE file.
When an SMB virtual circuit is prematurely closed due to an error
condition by one thread it is possible that there are still several
other threads still using the VC which will determine that the VC
is dead shortly. As a result we maintain a dead vc list to ensure
that dead vcs are not destroyed while still in use.
With the addition of locking though this causes a problem because the
locks allocated to the virtual circuit are not freed until the vc is
destroyed.
This patch reworks the cleanup routine to make the freeing of resources
the responsibility of the release routine when the refcount hits 0
and the VC is marked dead.
Fix several issues with the afs(1) man page. Based on the patch by Jason,
but I also documented /vicepiv as the maximum, and /vicepiu as the
maximum recommended, partition.
This patch is the second stage of the Unicode conversion. In this stage
the cache manager has been converted from using 8-bit UTF8 C strings to
16-bit UTF16 C strings in the SMB, the DNLC, the B+ tree, and other
directly related modules. The cm_cell, cm_volume, cm_scache, cm_buf,
cm_dcache, and cm_dir modules are left 8-bit because their data is all
8-bit UTF8 since they work only on file server strings.
The SMB layer accepts 16-bit UTF16, the B+ tree and DNLC use normalized
strings as the key, and everything uses UTF8. Efforts have been made
to minimize the number of transitions from UTF8 to UTF16 and back.
For the most part strings are tagged with clientchar_t and normchar_t
and fschar_t types in order to distinguish between the various types of
strings that are in use.
Refactor the pioctl handling to provide a layer of abstraction
between the smb server and the actual pioctl data processing.
This will permit a redirector pioctl layer to be implemented
on top of the core cm_ioctl interfaces.
The general approach is that there is a new cm_ioctl_t object
which is embedded withing a higher level abstraction object.
This object maintains all of the pioctl state processing and
removes any notion of file descriptors or other communication
channel parameters.
The cm_ioctl module maintains just about all of the core
functional processing except for SetToken which needs further
abstraction. Path processing is now performed at the higher
layer and cm_ioctl functions accept cm_user_t, cm_scache_t,
and cm_req_t objects from the higher layer.
The smb_ioctl module performs all of the path translation
using smb server knowledge and passes the necessary objects
to the cm_ioctl module for processing.
Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
Correct an abstraction layer violation. cm_scache_t objects should
be marked deleted in cm_Unlink() and cm_RemoveDir() and not in
smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
performed in cm_ReleaseSCache() when the reference count hits zero.
Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
Prevent corruption of the cm_data.scacheHashTablep lists when removing
and then re-adding entries to the hash table. The cm_scache_t nextp
fields were not being reset to NULL which permitted loops to be created.
Add documentation of foreign realm user registration and cross-realm PTS
groups. Add documentation of missing ptserver flags. Add some additional
to-do entries for the man pages.
It is frequently the case that rx_getAllAddr() is called before
rx_Init() or rx_InitHost(). rx_getAllAddr() obtains the list of
interfaces by using rx_GetIFInfo() which in turn computes and
allocates the number of addition rx packets. Unfortunately,
rxi_MorePackets() relies on the existence of an initialized mutex
and the mutex is not initialized (on Windows) until the rx_InitHost()
call. Therefore, we must delay the rxi_MorePackets() call until
after rx_InitHost() if rx_getAllAddr() is called previously.
in rxi_MorePackets() if osi_Alloc() fails to return the requested memory
call osi_Assert() instead of permitting subsequent null pointer dereferences
to corrupt the stack.
Define new flag CM_BUF_REDIR for use with the redirector.
When set, the flag indicates that the buffer is being held by the redirector
and it should be left untouched by the cache manager.
An incremental commit. This patch adds support for normalization of Unicode
but we have concluded that the normalization rules are incorrect. Normalized
strings should not be written to the file server or returned to the application.
if we can't use hlist_unhashed (which is dumb) or we don't have it (less dumb) work around it.
====================
This delta was composed from multiple commits as part of the CVS->Git migration.
The checkin message with each commit was inconsistent.
The following are the additional commit messages.
====================
LICENSE IPL10
FIXES 93871
if we can't use hlist_unhashed (which is dumb) or we don't have it (less dumb) w
ork around it.
This delta provides a fairly complete implementation of Unicode character
set support for the Windows Cache Manager and supporting tools including
fs.exe, symlink.exe, the pioctl library, and the explorer shell extension.
New Build requirements:
In order to build the Microsoft IDN Mitigation APIs SDK 1.1 is now required.
This SDK provides the normalization.h header and the redistribution
install packages required to install normaliz.dll. The AFSDEV_INCLUDE
path must be modified to point to the directory containing normalization.h.
There are no lib files for this package.
Cache Manager:
The CM SMB server has been modified to negotiate the use of UNICODE.
By default it is on. The "fs smbunicode" command can be used to disable
UNICODE at runtime.
cm_utils now contains UNICODE normalization and conversion routines built
on top of the IDN Mitigation APIs.
All input strings are normalized with UNICODE Normalization Form C.
The pioctl interface now supports UTF8 strings in addition to ANSI.
UTF8 strings are prefixed with <ESC> % 8 as per the ISO 2022 extension.
Pioctl Library:
New apps should use the pioctl_utf8() function instead of pioctl().
pioctl() is for ANSI strings and provides backward compatibility
with third party apps.
fs.exe and symlink.exe:
Converted to Unicode applications. All strings are now communicated
using UTF-8.
Explorer Shell Extension:
Converted to a Unicode DLL. All strings are now communicated using
UTF-8. Fonts on U.S. Windows distributions are not fully populated.
As a result not all of the characters can be displayed in all of the
dialog boxes.
Still to do:
1. Add Unicode normalization to AFS directory entries and mountpoint
and symlink target strings.
2. Use the Unicode version of GetCurrentDirectory in fs_utils.c
3. Update the installers to install the IDN Mitigation APIs on XP
and 2003. Vista already has them and they are not supported on
pre XP SP2 releases.
if we return leaving a non HOSTDELETED host while it does not have ALTADDR set,
a loop can happen. avoid it by simply setting ALTADDR again when we're done,
if we believe some addresses are still good this is fine.
Once we know that an attempt to stablize a buffer has failed due to
VNOVNODE, stop trying to Stablize future buffers and just invalidate
all of them belonging to the same FID.
in the event a server is "melting down" and not responding with data to RPCs
this adds the ability to time out the calls rather than letting keepalives
effectively hang the client forever. does not mark the server down since...
it's not down.
Side effect: adds similar tracking for authentication errors so tokens need
not be discarded if one server's clock has drifted.
The pthread butc crashes with SIGSEGV, when I restore to a volume
which already exists. This does happen with OpenAFS 1.4.7pre4 on
i386_linux26.
Looking into the code I think butc is using the wrong function
Lp_GetRwIndex (at volser/vsprocs.c:4394). It uses the one in the
library lib/libvosadmin.a, but should use the function defined
in volser/lockprocs.c.
Add "IdleDeadTimeout" registry value and call rx_SetConnIdleDeadTime()
with it when creating new rx connections. This forces the Rx stack
to return RX_CALL_TIMEOUT if the file server has returned VBUSY for
longer than the IdleDeadTimeout.
When RX_CALL_TIMEOUT is received, the cache manager will now attempt to
find another server for the object that might be less busy.
projects / packages / o / openafs.git / log