openafs.git
19 years agowindows-afsdb-freelance-notes-20011120
Jeff Riegel [Wed, 21 Nov 2001 18:29:16 +0000]
windows-afsdb-freelance-notes-20011120

notes from Jeff about AFSDB and Freelance clients

19 years agodcache-finegrained-locks-20011121
Nickolai Zeldovich [Wed, 21 Nov 2001 16:01:19 +0000]
dcache-finegrained-locks-20011121

This patch implements more fine-grained locking for dcache entries.
The main advantage is that multiple chunks of the same file can be
fetched at once.  This means that an incorrectly-guessed prefetch
won't block other fetches, prefetches of multiple chunks can occur
in parallel, and multiple processes sharing the same file can read
from different parts of the file at once.

19 years agoquote-modloaddir-when-looking-for-kernel-module-20011120
Yves Schutz [Wed, 21 Nov 2001 15:54:35 +0000]
quote-modloaddir-when-looking-for-kernel-module-20011120

quote variable evaluation

19 years agolinux-rc-make-ifconfig-speak-english-for-parsing-20011120
Thomas Mueller [Wed, 21 Nov 2001 15:41:26 +0000]
linux-rc-make-ifconfig-speak-english-for-parsing-20011120

sadly the script is fluent only in english and can't even pretend it knows
anything else

19 years agobozo-make-logging-more-readable-20011120
Garry Zacheiss [Wed, 21 Nov 2001 15:36:43 +0000]
bozo-make-logging-more-readable-20011120

add missing newlines

19 years agoafs-trace-missing-dependancy-20011120
Nathan Neulinger [Wed, 21 Nov 2001 15:31:24 +0000]
afs-trace-missing-dependancy-20011120

so when afs_trace.et updates the header gets rebuilt

19 years agoktc-dont-prototype-lseek-20011120
Sam Hartman [Wed, 21 Nov 2001 15:29:52 +0000]
ktc-dont-prototype-lseek-20011120

breaks sparc linux 2.2

19 years agowindows-updates-20011121
James Peterson [Wed, 21 Nov 2001 06:44:28 +0000]
windows-updates-20011121

missed updates needed for next release

19 years agoafs-resetvolumes-zero-serverhost-to-avoid-deadlock-20011120
Derrick Brashear [Tue, 20 Nov 2001 20:04:55 +0000]
afs-resetvolumes-zero-serverhost-to-avoid-deadlock-20011120

the equivalent of what's done in InstallUVolumeEntry to avoid a deadlock
but just for this server since it's all we're dealing with

19 years agoviced-getstatus-clear-lengthhi-since-not-now-using-it-20011119
Chas Williams [Mon, 19 Nov 2001 16:59:53 +0000]
viced-getstatus-clear-lengthhi-since-not-now-using-it-20011119

so we don't end up with bogus value here

19 years agostartrxafs-fetchdata64-always-show-result-and-EFBIG-at-7fffffff-20011119
Chas Williams [Mon, 19 Nov 2001 16:58:02 +0000]
startrxafs-fetchdata64-always-show-result-and-EFBIG-at-7fffffff-20011119

show fetch64 errors
also mask high bit for reporting EFBIG

19 years agoviced-fetchdata64-should-call-common-code-in-64-mode-20011119
Chas Williams [Mon, 19 Nov 2001 16:53:46 +0000]
viced-fetchdata64-should-call-common-code-in-64-mode-20011119

both of the stubs shouldn't be using the 32 bit case in common

19 years agorx-sys-fix-broken-install-dependancies-20011119
Chas Williams [Mon, 19 Nov 2001 16:52:02 +0000]
rx-sys-fix-broken-install-dependancies-20011119

this got broken in one of the sets of changes

19 years agoviced-add-missing-comma-in-64bit-detailed-stats-case-20011114
Derrick Brashear [Wed, 14 Nov 2001 23:59:57 +0000]
viced-add-missing-comma-in-64bit-detailed-stats-case-20011114

and due to no prototypes this slipped right on by

19 years agobosserver-log-improve-readability-20011114
Garry Zacheiss [Wed, 14 Nov 2001 17:39:30 +0000]
bosserver-log-improve-readability-20011114

include newlines in boslog output

19 years agorx-fix-resource-starvation-problem-20011113
Nickolai Zeldovich [Wed, 14 Nov 2001 04:02:27 +0000]
rx-fix-resource-starvation-problem-20011113

This patch fixes a resource starvation condition in Rx.  The
  problem arises, for instance, when more than 4 daemons try to
  prefetch chunks of the same file at once.  The fifth daemon is
  stuck in MAKECALL_WAITING state, never getting a chance to run,
  because the other 4 daemons never yield to the scheduler after
  releasing the call, and just grab the call back again.

19 years agoafs-cell-avoid-need-for-new-prototype-20011113
Derrick Brashear [Tue, 13 Nov 2001 20:31:31 +0000]
afs-cell-avoid-need-for-new-prototype-20011113

reogranize so we don't need to add a prototype

based on report from kolya@mit.edu

19 years agohold-afs-xcell-lock-during-setserverprefs-20011113
Derrick Brashear [Tue, 13 Nov 2001 17:24:05 +0000]
hold-afs-xcell-lock-during-setserverprefs-20011113

afs_RemoveCellEntry holds afs_xcell; setserverprefs modified the same
structure but did not which was problematic if something changed out from under
it

19 years ago64bit-client-mods-build-cleanly-20011113
Hartmut Reuter [Tue, 13 Nov 2001 14:47:11 +0000]
64bit-client-mods-build-cleanly-20011113

necessary fixes for aix and solaris, plus fix for linux for when wake_up
doesn't wake the sleeper

19 years agopam-include-limits-h-for-NGROUPS-MAX-20011113
Hartmut Reuter [Tue, 13 Nov 2001 14:40:54 +0000]
pam-include-limits-h-for-NGROUPS-MAX-20011113

pam module needs limits.h to build

19 years agonbobtainwritelock-fix-precedence-problem-20011113
Nickolai Zeldovich [Tue, 13 Nov 2001 13:14:06 +0000]
nbobtainwritelock-fix-precedence-problem-20011113

NBObtainWriteLock always returned 0 due to precedence problem. Also
add NBObtainSharedLock

19 years agorx-compute-ackdatasize-correctly-20011113
Nickolai Zeldovich [Tue, 13 Nov 2001 04:59:22 +0000]
rx-compute-ackdatasize-correctly-20011113

This patch clarifies the implicit 3 reserved bytes in rx_AckDataSize,
  and hopefully makes the code more portable for future platforms.

19 years agoirix-build-cleanup-20011113
Chas Williams [Tue, 13 Nov 2001 04:52:26 +0000]
irix-build-cleanup-20011113

make IP35 port select correctly. make pagsh build again

19 years agoirix-PFLUSHINVAL-should-be-VOP-FLUSHINVAL-PAGES-20011111
Chas Williams [Sun, 11 Nov 2001 17:11:06 +0000]
irix-PFLUSHINVAL-should-be-VOP-FLUSHINVAL-PAGES-20011111

Based on suggestions from gwehrman@sgi.com

19 years agomake-darwin-14-build-with-new-makefile-frag-system-20011110
Chaskiel M Grundman [Sun, 11 Nov 2001 01:57:36 +0000]
make-darwin-14-build-with-new-makefile-frag-system-20011110

head now builds on darwin 1.4; may not build in objdir env yet

19 years agodarwin-14-support-20011110
Chaskiel M Grundman [Sat, 10 Nov 2001 23:20:04 +0000]
darwin-14-support-20011110

base support for darwin 1.4

19 years agomore-64bit-client-fixes-20011110
Hartmut Reuter [Sat, 10 Nov 2001 23:00:55 +0000]
more-64bit-client-fixes-20011110

some afs_offs_t still missing
so obsolete code fixed
and new trace for m.Length change

19 years agouse-NGROUPS-MAX-instead-of-constant-for-curpag-getgroups-call-20011110
Thomas Mueller [Sat, 10 Nov 2001 22:36:45 +0000]
use-NGROUPS-MAX-instead-of-constant-for-curpag-getgroups-call-20011110

NGROUPS_MAX instead of 30 used so we get all groups from getgroups

19 years agowindows-updates-including-dont-lose-tokens-20011110
James Peterson [Sat, 10 Nov 2001 22:31:04 +0000]
windows-updates-including-dont-lose-tokens-20011110

Win2k tokens association problems are fixed. details are in doc/txt/winnotes
in the 1.2.2a updates file

19 years agolinux-afsd-allow-ext3-journals-without-error-20011110
David Thompson [Sat, 10 Nov 2001 18:21:59 +0000]
linux-afsd-allow-ext3-journals-without-error-20011110

an ext3 journal in the vice cache (root of the partition) is allowable

we have no useful way to discern ext2 from ext3 without groveling in fstab
so just allow it

19 years agorx-dont-ackall-a-connection-were-waiting-for-retransmits-on-20011110
Nickolai Zeldovich [Sat, 10 Nov 2001 18:14:30 +0000]
rx-dont-ackall-a-connection-were-waiting-for-retransmits-on-20011110

"My theory of what happened is roughly as follows:

    Process tries to read data from AFS (as part of a page fault);
    issues a new Rx call on an Rx connection to the fileserver.

    The server transmits some data back to the client, but some packet
    is lost.

    Something tries to garbage-collect/destroy the connection; since
    there is an active call, it can't do so, but issues an rx_AckAll
    anyway, which acknowledges all packets transmitted by the server
    as having been received.  Server flushes its retransmit queue.

    Client waits forever for the lost packet to arrive, but since the
    server has already flushed the transmit queue, it cannot possibly
    retransmit it.

All this is happening while the client has read-locked its address
space (since the read is part of a page fault).  /proc accesses that
try to poke into that processes address space hang waiting for said
lock, causing the lossage we actually observed."

19 years agorx-deal-with-missed-keepalives-by-accurately-tracking-call-number-20011110
Garry Zacheiss [Sat, 10 Nov 2001 18:08:40 +0000]
rx-deal-with-missed-keepalives-by-accurately-tracking-call-number-20011110

(as originally discovered by ted@mit.edu)

"This fix deals with the following lose case:
  Client starts a call that, for some reason, takes a long time on the
  server.  While the client waits for the server to finish, client and
  server usually send each other keep alive packets.  If something
  causes those packets to be delayed or dropped, then the client will
  conclude that the call has failed or finished (usually failed), while
  the server is still *busy* doing the call.

  In this circumstance, the client will initiate another call and the
  server will correctly respond that it is busy.  Unfortunately, if the
  callNumber of a received packet doesn't match the callNumber of the
  outstanding call, then the client never sees that the server says it's
  busy.  Instead the server appears as a black hole to the client.

  This fix ensures that the client sees the busy packets when its
  callNumber is reasonably out of sync with the server."

19 years agoirix-memcache-dont-call-osi-vm-storeallsegments-20011110
Chas Williams [Sat, 10 Nov 2001 17:53:04 +0000]
irix-memcache-dont-call-osi-vm-storeallsegments-20011110

this caused a call to pdflush to happen at the wrong time, which should fix
the zero filled files problem, the osi_assert(cred) problem and the
execsorwriters == 0 warnings to go away

19 years agolinux-kdump-also-work-with-early-24-kernels-20011108
Derrick Brashear [Thu, 8 Nov 2001 08:12:59 +0000]
linux-kdump-also-work-with-early-24-kernels-20011108

yup, it had to come from somewhere. now it would compile with early and late 2.4 kernels

19 years agoconfig-makefile-include-xcflags64-20011107
Nickolai Zeldovich [Thu, 8 Nov 2001 03:02:08 +0000]
config-makefile-include-xcflags64-20011107

this was missed during the conversion

19 years agosolaris-vfsck-do-a-better-job-of-putting-on-the-ufs-fsck-face-for-the-world-20011106
Garry Zacheiss [Wed, 7 Nov 2001 00:15:27 +0000]
solaris-vfsck-do-a-better-job-of-putting-on-the-ufs-fsck-face-for-the-world-20011106

if you're not using ufs logging it's ok to replace solaris fsck with vfsck,
except sometimes it exits with 40 and that's not a failure to the solaris
scripts.

make it so for us also

19 years agobutc-allow-comments-in-tapeconfig-20011106
Garry Zacheiss [Wed, 7 Nov 2001 00:11:20 +0000]
butc-allow-comments-in-tapeconfig-20011106

continue to allow extra column for comments as was previously (afs 3.4a) the case

19 years agoafsdb-avoid-dns-case-issues-for-cell-aliases-20011106
Nickolai Zeldovich [Wed, 7 Nov 2001 00:01:52 +0000]
afsdb-avoid-dns-case-issues-for-cell-aliases-20011106

This patch makes sure that in-kernel aliases to non-existant names aren't
  accidentally created due to case mismatch (e.g. "athena" being created as
  a symlink to "athena.MIT.EDU", while "athena.mit.edu" is the real cell
  that already exists).  It also lowercases cell names in AFSDB lookups,
  otherwise the same problem appears in userspace (eg "aklog athena" tries
  to obtain tokens for cell "athena.MIT.EDU").

19 years agolinux-kdump-kernel-also-has-timer-t-sigh-20011105
Derrick Brashear [Mon, 5 Nov 2001 19:43:47 +0000]
linux-kdump-kernel-also-has-timer-t-sigh-20011105

not the cleanest way of dealing, either

19 years agolinux-kdump-timeval-probably-not-needed-20011105
Derrick Brashear [Mon, 5 Nov 2001 19:39:06 +0000]
linux-kdump-timeval-probably-not-needed-20011105

not sure what needs this. if something does we need to figure out what
and only do this for them

19 years agobkg-daemon-dont-break-64bit-pointers-and-handle-requests-in-order-20011102
Nickolai Zeldovich [Mon, 5 Nov 2001 19:34:25 +0000]
bkg-daemon-dont-break-64bit-pointers-and-handle-requests-in-order-20011102

This patch fixes a problem with 64-bit pointers being munged by the
background daemons (by separating sizes and pointers into separate
variables -- this bug was apparently introduced by the 64-bit file
support patch), and makes the background daemons handle requests in
order they came in.  The latter will be mostly just useful for some
prefetching and fine grained dcache-locking patches

19 years agopam-update-for-correct-krb-aware-module-20011102
Thomas Mueller [Mon, 5 Nov 2001 18:40:02 +0000]
pam-update-for-correct-krb-aware-module-20011102

the krb version of the module should be built completely in AFS_KERBEROS_ENV

====================
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.
====================

clean up spacing

19 years agodynamic-root-allocate-space-for-dot-and-dotdot-20011102
Nickolai Zeldovich [Fri, 2 Nov 2001 21:00:41 +0000]
dynamic-root-allocate-space-for-dot-and-dotdot-20011102

no space was being allocated for . and ..; do so

19 years agoafs-checkservers-improve-lock-granularity-20011102
Nickolai Zeldovich [Fri, 2 Nov 2001 08:07:58 +0000]
afs-checkservers-improve-lock-granularity-20011102

work harder to release afs_xserver and afs_xsrvAddr locks in afs_CheckServers
when not needed

19 years agosolaris9-and-makefile-updates-20011102
Nickolai Zeldovich [Fri, 2 Nov 2001 08:03:08 +0000]
solaris9-and-makefile-updates-20011102

updates for solaris 9; also don't have ${XLIBS} in target dependancies
if it's going to include things like -lnsl

19 years agoviced-clear-clientdeleted-when-deleted-clients-are-cleaned-up-20011102
Nickolai Zeldovich [Fri, 2 Nov 2001 07:57:12 +0000]
viced-clear-clientdeleted-when-deleted-clients-are-cleaned-up-20011102

Currently nothing clears the CLIENTDELETED flag in hosts, so once
  a client has been deleted, h_TossStuff_r() will keep getting called
  with every host release.  This patch clears the CLIENTDELETED flag
  every time we take care of deleted clients.

19 years agopam-setcred-dont-ask-for-passwords-20011101
Carsten Jacobi [Thu, 1 Nov 2001 05:45:12 +0000]
pam-setcred-dont-ask-for-passwords-20011101

don't prompt for passwords during setcred step

19 years agodynroot-allow-cell-aliases-20011101
Nickolai Zeldovich [Thu, 1 Nov 2001 05:24:30 +0000]
dynroot-allow-cell-aliases-20011101

allow cell aliases. includes support for non-'V' pioctls

19 years agokdump-display-timeout-for-cell-info-20011101
Nickolai Zeldovich [Thu, 1 Nov 2001 05:18:30 +0000]
kdump-display-timeout-for-cell-info-20011101

added field in kdump output to display afsdb timeout info

19 years agosolaris-rx-kernel-set-listener-pid-from-curproc-20011101
Nickolai Zeldovich [Thu, 1 Nov 2001 05:16:27 +0000]
solaris-rx-kernel-set-listener-pid-from-curproc-20011101

set kernel listener pid from curproc rather than curthread (and clean
up unnecessary break case in code)

19 years agopts-error-message-cleanup-20011031
Nickolai Zeldovich [Thu, 1 Nov 2001 05:11:25 +0000]
pts-error-message-cleanup-20011031

print error in such a way that it makes more sense

19 years agoafsdb-default-in-rc-files-20011031
Nickolai Zeldovich [Thu, 1 Nov 2001 05:10:38 +0000]
afsdb-default-in-rc-files-20011031

enable afsdb using rc files

19 years agoirix-build-ip35-module-only-if-headers-exist-20011031
Derrick Brashear [Thu, 1 Nov 2001 05:05:26 +0000]
irix-build-ip35-module-only-if-headers-exist-20011031

IP35 support requires 6.5.9 or later; only build if we have headers

19 years agobuild-system-specific-makefile-configuration-from-configure-20011031
Nathan Neulinger [Thu, 1 Nov 2001 04:59:26 +0000]
build-system-specific-makefile-configuration-from-configure-20011031

junk all of Makefile.${SYS_NAME} in favor of a common makefile built by
configure, and use it

19 years agomore-64bit-file-size-support-20011031
Hartmut Reuter [Thu, 1 Nov 2001 04:39:08 +0000]
more-64bit-file-size-support-20011031

further 64bit file size support for client

19 years agoclient-64bit-file-size-support-20011031
Hartmut Reuter [Thu, 1 Nov 2001 04:01:22 +0000]
client-64bit-file-size-support-20011031

new RPCs and client-side support for large files

19 years agoaix-namei-fileserver-support-20011031
Hartmut Reuter [Thu, 1 Nov 2001 03:58:36 +0000]
aix-namei-fileserver-support-20011031

necessary for namei support on aix

19 years agofast-restart-exit-if-no-args-20011031
Hartmut Reuter [Thu, 1 Nov 2001 03:54:40 +0000]
fast-restart-exit-if-no-args-20011031

when using fast restart, if salvager took no args, assume it's the fs bnode
and exit

19 years agofreelance-updates-20011031
Jeff Riegel [Thu, 1 Nov 2001 03:40:21 +0000]
freelance-updates-20011031

updates to make freelance patches work, create mountpoints from afsdb on the fly

20 years agoparisc-linux-lwp-updates-20011025
Nathan Neulinger [Fri, 26 Oct 2001 04:50:06 +0000]
parisc-linux-lwp-updates-20011025

update parisc lwp foo

20 years agorx-test-srcdir-support-20011025
Nathan Neulinger [Fri, 26 Oct 2001 04:49:11 +0000]
rx-test-srcdir-support-20011025

continuing srcdir support for objdir builds`

20 years agoexport-dont-include-afsconfig-to-avoid-conflicts-20011024
Hartmut Reuter [Wed, 24 Oct 2001 16:56:00 +0000]
export-dont-include-afsconfig-to-avoid-conflicts-20011024

ncluding afsconfig results in conflicts on some systems due to _KERNEL

20 years agoafsd-make-debug-switch-print-afsops-20011024
Derek Atkins [Wed, 24 Oct 2001 16:25:22 +0000]
afsd-make-debug-switch-print-afsops-20011024

just print one more argument for afs system calls

20 years agobozo-fix-bos-addhost--clone-20011024
Hartmut Reuter [Wed, 24 Oct 2001 16:14:18 +0000]
bozo-fix-bos-addhost--clone-20011024

addhost -clone didn't previously work right. this fixes it

20 years agoundo-makefiles-rewrite-for-objdir-support-20011023
Nathan Neulinger [Wed, 24 Oct 2001 16:03:48 +0000]
undo-makefiles-rewrite-for-objdir-support-20011023

fails certain cases

20 years agolinux-autoconf-reuse-i-truncate-sem-name-for-new-test-20011023
Derrick Brashear [Tue, 23 Oct 2001 23:06:41 +0000]
linux-autoconf-reuse-i-truncate-sem-name-for-new-test-20011023

since we need to initialize a semaphore

20 years agomakefiles-attempt-to-support-gmake-j-20011023
Nathan Neulinger [Tue, 23 Oct 2001 21:57:19 +0000]
makefiles-attempt-to-support-gmake-j-20011023

move toward being able to gmake -j openafs

20 years agovenus-kdump-build-from-srcdir-20011023
Nathan Neulinger [Tue, 23 Oct 2001 21:39:56 +0000]
venus-kdump-build-from-srcdir-20011023

kdump should be built from src in srcdir

20 years agoparisc-linux-lwp-sp-20011023
Nathan Neulinger [Tue, 23 Oct 2001 21:36:05 +0000]
parisc-linux-lwp-sp-20011023

provide stack pointer for parisc linux lwp

20 years agomakefiles-rewrite-for-objdir-support-20011023
Nathan Neulinger [Tue, 23 Oct 2001 19:56:56 +0000]
makefiles-rewrite-for-objdir-support-20011023

rewrite makefiles to support objdir; simplify in the process

20 years agolinux-configure-support-mandrake-kernel-patches-20011023
Derrick Brashear [Tue, 23 Oct 2001 16:27:25 +0000]
linux-configure-support-mandrake-kernel-patches-20011023

init_once in mandrake calls init_rwsem(&inode->i_truncate_sem);

so do we

20 years agodoc-html-index-correct-spelling-20011023
Derrick Brashear [Tue, 23 Oct 2001 16:26:00 +0000]
doc-html-index-correct-spelling-20011023

documentation should be spelled that way

20 years agovenus-install-kdump-scripts-from-srcdir-20011022
Nathan Neulinger [Mon, 22 Oct 2001 17:55:05 +0000]
venus-install-kdump-scripts-from-srcdir-20011022

non-generated scripts come from srcdir

20 years agovos-rename-crypt-to-encrypt-20011020
Derrick Brashear [Sat, 20 Oct 2001 17:27:47 +0000]
vos-rename-crypt-to-encrypt-20011020

so -c is again unique

20 years agocache-server-inlinebulk-support-and-enable-inlinebulk-20011019
Derrick Brashear [Fri, 19 Oct 2001 19:52:28 +0000]
cache-server-inlinebulk-support-and-enable-inlinebulk-20011019

expand struct server's "flags" to unsigned int from char
assume inlinebulk and turn off on first failure

====================
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.
====================

fix precedence error

20 years agoadded button GIF's to the HTML docs
Jeffrey Hutzelman [Fri, 19 Oct 2001 17:22:30 +0000]
added button GIF's to the HTML docs

20 years agodes-use-ar-macro-instead-of-direct-call-like-everyone-else-20011019
Nathan Neulinger [Fri, 19 Oct 2001 16:56:55 +0000]
des-use-ar-macro-instead-of-direct-call-like-everyone-else-20011019

so we can use ar like every other directory

20 years agoirix-ip35-tweak-kernel-cflags-20011019
Derrick Brashear [Fri, 19 Oct 2001 16:48:57 +0000]
irix-ip35-tweak-kernel-cflags-20011019

based on doc provided by gwehrman@sgi.com

20 years agocmd-errors64-is-generated-so-not-from-srcdir-20011019
Nathan Neulinger [Fri, 19 Oct 2001 16:31:04 +0000]
cmd-errors64-is-generated-so-not-from-srcdir-20011019

fix problem with objdir builds

20 years agoirix-first-pass-at-ip35-support-20011017
Chas Williams [Thu, 18 Oct 2001 16:40:08 +0000]
irix-first-pass-at-ip35-support-20011017

i missed this and chas caught it

20 years agosunx86-support-20011017
Eric Lauriault [Thu, 18 Oct 2001 00:36:02 +0000]
sunx86-support-20011017

base sunx86 support for newer than solaris 2.4

20 years agoinitial-objdir-support-20011017
Nathan Neulinger [Thu, 18 Oct 2001 00:06:58 +0000]
initial-objdir-support-20011017

first cut at objdir support

20 years agodevelopment-readme-20011017
Nathan Neulinger [Wed, 17 Oct 2001 23:24:33 +0000]
development-readme-20011017

hints for developers

20 years agolinux-dont-require-osi-vfs-to-be-installed-20011017
Derrick Brashear [Wed, 17 Oct 2001 23:17:53 +0000]
linux-dont-require-osi-vfs-to-be-installed-20011017

based on reports from hartmans@mit.edu and jhutz@cmu.edu

20 years agoirix-first-pass-at-ip35-support-20011017
Derrick Brashear [Wed, 17 Oct 2001 22:53:48 +0000]
irix-first-pass-at-ip35-support-20011017

based on extensive feedback from Forest Godfrey and Michael Raymond
of SGI

20 years agosgi-ip20-work-around-cvt-bug-20011017
Chas Williams [Wed, 17 Oct 2001 22:07:49 +0000]
sgi-ip20-work-around-cvt-bug-20011017

"apparently the rev 1 r5000 chips implement 'cvt' incorrectly.  the irix
kernel works around this problem by checking each text page mapped into
memory and doing a fixup on the cvt instructions.  it tries to maintain
a hash of these pages using fid2() or fid() if fid2() returns ENOSYS.
afs, in an effort to prevent people from doing checkpoints on an afs
filesystem, makes fid2() return EINVAL.  this also keeps the kernel from
mapping executables that are in afs space on the broken r5000's.

this is the patch i have been using for the past couple years while
waiting for an official fix.  it makes fid2() return ENOSYS, so you
now need to have to have v_ckpt.  however i disabled the rest of the
CKPT code since i have no idea how well that code actually works.
additionally, this behavior is only functional on machines with the
'broken' r5000 h/w.  i cant think of a better way to fix this problem
since i cant change the irix kernel."

20 years agotop-makefile-make-installs-not-depend-on-extra-building-20011017
Derek Atkins [Wed, 17 Oct 2001 22:03:45 +0000]
top-makefile-make-installs-not-depend-on-extra-building-20011017

kernel module should not make usermode stuff and vice versa

20 years agobozo-getcellhost-takes-uint32-not-int32-20011017
Nickolai Zeldovich [Wed, 17 Oct 2001 21:46:18 +0000]
bozo-getcellhost-takes-uint32-not-int32-20011017

pass in a uint32 rather than an int32

20 years agocmdebug-addrs-option-20011017
Nickolai Zeldovich [Wed, 17 Oct 2001 17:43:17 +0000]
cmdebug-addrs-option-20011017

This patch adds an "-addrs" option to cmdebug, which lists
the interfaces of a remote cache manager, via the WhoAreYou
AFSCB RPC.

20 years agocvsignore-updates-20011017
Derrick Brashear [Wed, 17 Oct 2001 17:36:07 +0000]
cvsignore-updates-20011017

based on comment from nneul@umr.edu

20 years agojunk-redhat-sh-since-configure-tests-cover-us-20011017
Derrick Brashear [Wed, 17 Oct 2001 16:12:00 +0000]
junk-redhat-sh-since-configure-tests-cover-us-20011017

we don't need this anymore

20 years agodux-kernel-lacks-mem-functions-20011016
Derrick Brashear [Wed, 17 Oct 2001 04:06:22 +0000]
dux-kernel-lacks-mem-functions-20011016

from report from jimmy@e.kth.se

you can tell how many people are using openafs on digital unix/tru64. sigh.

20 years agodisable-client-inlinebulk-20011013
Derrick Brashear [Sat, 13 Oct 2001 06:28:07 +0000]
disable-client-inlinebulk-20011013

until we can remember which servers support it

20 years agoupdate-news-file-20011013
Derrick Brashear [Sat, 13 Oct 2001 06:27:19 +0000]
update-news-file-20011013

update news file for new advances

20 years agolinux-no-umount-begin-20011012
Derrick Brashear [Fri, 12 Oct 2001 18:55:33 +0000]
linux-no-umount-begin-20011012

to fix problem reported by rtm@cert.org

this code needs to be reworked before it can be reactivated

20 years agofbsd-makefile-typo-20011012
Ed Symanzik [Fri, 12 Oct 2001 14:55:03 +0000]
fbsd-makefile-typo-20011012

add missing =

20 years agomake-logging-enhancements-aix-friendly-20011011
Derrick Brashear [Fri, 12 Oct 2001 01:24:22 +0000]
make-logging-enhancements-aix-friendly-20011011

AIX lacks WCOREDUMP so we work around it

20 years agofix-attach-directories-code-to-properly-use-vgetpartition-20011011
Derrick Brashear [Thu, 11 Oct 2001 22:51:13 +0000]
fix-attach-directories-code-to-properly-use-vgetpartition-20011011

irix hates us when we misuse functions

20 years agoirix-adapt-mtu-correctly-in-multi-interface-case-20011011
Nickolai Zeldovich [Thu, 11 Oct 2001 22:41:45 +0000]
irix-adapt-mtu-correctly-in-multi-interface-case-20011011

let hash_enum iterate so we get the right interface

====================
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.
====================

undo chas's earlier patch
fix another flub

20 years agologging-enhancements-need-ctime-decl-20011011
Derrick Brashear [Thu, 11 Oct 2001 22:33:05 +0000]
logging-enhancements-need-ctime-decl-20011011

make irix happy