Avoid condition where we could read past the end of lrealm (unlikely
to occur in practice and not remotely exploitable; cell and lrealm are
both MAXKTCREALMLEN and lrealm is configured locally and not read from the
network)
"Here are the symptoms. I have 2 OpenAFS Linux 2.4 machines, one SMP and
one not.
--With both of them I access a series of directories and files within
(/afs/w/x/y/z/).
--I then move to the root directory.
--From a third machine, I remove the directories x, y, and z. Then re-add
them (same names) with some other files.
--From the SMP box, I do an ls in /afs/w and see x. But x is inaccessible.
--From the UP box I can see and access x.
"
====================
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.
====================
Missed earlier: not part of the IBM delta.
====================
Part of the alpha patch slipped into the delta; Revert to previous
revision
"Simple patch to add "-syslog" option to bosserver. Causes it to direct all
logging activity to syslogd instead of to logs/BosLog"
currently unimplemented for NT
====================
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.
====================
Don't use a static outside its scope
"This prevents pts from crashing like this:
pepsi-one~> pts createg a:b -id -1234
/bin/athena/pts: Permission denied ; unable to create group a:b with id -1234
owned by 'Segmentation fault
"
" I have found problem with file locking and 64bit linux (tested on
ultrasparc). When issued fcntl(fd, f_SETLK, ..) with l_start = 0 and l_len
= 0, it sets l_len to LONG_MAX, so it is needed to test this value"
Make check the return code from EvalMountPoint()
====================
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.
====================
maintain old behavior consistent with checking eval return
for new pioctl calls for mrafs
====================
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.
====================
for new pioctl calls for mrafs; convert spaces to tab
Dirpath additions for MR-AFS files
====================
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.
====================
Dirpath additions for MR-AFS files; remove extra newline
Changes needed to support ubik (non-voting) clones.
====================
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.
====================
Needed kaserver support for ubik clones
====================
Needed support for ubik clones for the ptserver
====================
Support for ubik clones in vlserver
====================
auth layer support for ubik readonly clones
====================
ubik library support for readonly clones
export some previously internal routines, and rename to not conflict
with libkrb;
with NO_AFS_CLIENT you can use the ktc routines also for architectures
which not yet have kernel support for AFS tokens (Cray T3E, NEC SX5).
Necessary additions to bos for MR-AFS support
====================
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.
====================
FS bnode changes for MR-AFS
====================
allow hex numbers for residencies, used by bos
Required to placate AIX compiler
====================
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.
====================
