1 .. SPDX-License-Identifier: GPL-2.0
2
3 ===========================
4 Coda Kernel-Venus Interface
5 ===========================
6
7 .. Note::
8
9 This is one of the technical documents desc
10 Coda -- this document describes the client
11
12 For more information:
13
14 http://www.coda.cs.cmu.edu
15
16 For user level software needed to run Coda:
17
18 ftp://ftp.coda.cs.cmu.edu
19
20 To run Coda you need to get a user level cache
21 named Venus, as well as tools to manipulate AC
22 client needs to have the Coda filesystem selec
23 configuration.
24
25 The server needs a user level server and at pr
26 kernel support.
27
28 The Venus kernel interface
29
30 Peter J. Braam
31
32 v1.0, Nov 9, 1997
33
34 This document describes the communication be
35 level filesystem code needed for the operati
36 tem. This document version is meant to desc
37 (version 1.0) as well as improvements we env
38
39 .. Table of Contents
40
41 1. Introduction
42
43 2. Servicing Coda filesystem calls
44
45 3. The message layer
46
47 3.1 Implementation details
48
49 4. The interface at the call level
50
51 4.1 Data structures shared by the kernel
52 4.2 The pioctl interface
53 4.3 root
54 4.4 lookup
55 4.5 getattr
56 4.6 setattr
57 4.7 access
58 4.8 create
59 4.9 mkdir
60 4.10 link
61 4.11 symlink
62 4.12 remove
63 4.13 rmdir
64 4.14 readlink
65 4.15 open
66 4.16 close
67 4.17 ioctl
68 4.18 rename
69 4.19 readdir
70 4.20 vget
71 4.21 fsync
72 4.22 inactive
73 4.23 rdwr
74 4.24 odymount
75 4.25 ody_lookup
76 4.26 ody_expand
77 4.27 prefetch
78 4.28 signal
79
80 5. The minicache and downcalls
81
82 5.1 INVALIDATE
83 5.2 FLUSH
84 5.3 PURGEUSER
85 5.4 ZAPFILE
86 5.5 ZAPDIR
87 5.6 ZAPVNODE
88 5.7 PURGEFID
89 5.8 REPLACE
90
91 6. Initialization and cleanup
92
93 6.1 Requirements
94
95 1. Introduction
96 ===============
97
98 A key component in the Coda Distributed File
99 manager, Venus.
100
101 When processes on a Coda enabled system acce
102 filesystem, requests are directed at the fil
103 operating system. The operating system will
104 service the request for the process. Venus
105 client cache and makes remote procedure call
106 related servers (such as authentication serv
107 requests it receives from the operating syst
108 serviced a request it replies to the operati
109 return codes, and other data related to the
110 kernel support for Coda may maintain a minic
111 requests to limit the number of interactions
112 possesses the facility to inform the kernel
113 minicache are no longer valid.
114
115 This document describes precisely this commu
116 kernel and Venus. The definitions of so cal
117 will be given with the format of the data th
118 describe the semantic invariants resulting f
119
120 Historically Coda was implemented in a BSD f
121 The interface between the kernel and Venus i
122 VFS interface. Similar functionality is pro
123 the parameters and returned data is very sim
124 leads to an almost natural environment for i
125 filesystem driver for Coda in a BSD system.
126 systems such as Linux and Windows 95 and NT
127 with different interfaces.
128
129 To implement Coda on these systems some reve
130 Venus/Kernel protocol is necessary. Also it
131 systems could profit significantly from cert
132 and modifications to the protocol. To facili
133 to make future ports easier, communication b
134 kernel should be documented in great detail.
135 document.
136
137 2. Servicing Coda filesystem calls
138 ===================================
139
140 The service of a request for a Coda file sys
141 a process P which accessing a Coda file. It
142 traps to the OS kernel. Examples of such cal
143 are ``read``, ``write``, ``open``, ``close``
144 ``rmdir``, ``chmod`` in a Unix ontext. Simi
145 environment, and are named ``CreateFile``.
146
147 Generally the operating system handles the r
148 filesystem (VFS) layer, which is named I/O M
149 manager in Windows 95. The VFS is responsib
150 of the request and for locating the specific
151 service parts of the request. Usually the i
152 assists in locating the correct FS drivers.
153 pre-processing, the VFS starts invoking expo
154 driver. This is the point where the FS spec
155 request starts, and here the Coda specific k
156 play.
157
158 The FS layer for Coda must expose and implem
159 First and foremost the VFS must be able to m
160 the Coda FS layer, so the Coda FS driver mus
161 as applicable in the operating system. These
162 among operating systems, but share features
163 read/write and create and remove objects. T
164 such VFS requests by invoking one or more we
165 offered by the cache manager Venus. When th
166 come back to the FS driver, servicing of the
167 finishes with a reply to the kernel's VFS. F
168 returns to the process.
169
170 As a result of this design a basic interface
171 must allow Venus to manage message traffic.
172 be able to retrieve and place messages and t
173 arrival of a new message. The notification m
174 which does not block Venus since Venus must
175 when no messages are waiting or being proces
176
177 **Interfaces of the Coda FS Driver**
178
179 Furthermore the FS layer provides for a spec
180 between a user process and Venus, called the
181 pioctl interface is used for Coda specific s
182 requesting detailed information about the pe
183 Venus. Here the involvement of the kernel is
184 the calling process and passes the informati
185 Venus replies the response is passed back to
186 form.
187
188 Finally Venus allows the kernel FS driver to
189 certain services. This is done to avoid exc
190 and results in an efficient system. However
191 information, for example from the network wh
192 information must be flushed or replaced. Ven
193 to the Coda FS layer to request flushes or u
194 kernel FS driver handles such requests synch
195
196 Among these interfaces the VFS interface and
197 receive and be notified of messages are plat
198 not go into the calls exported to the VFS la
199 requirements of the message exchange mechani
200
201
202 3. The message layer
203 =====================
204
205 At the lowest level the communication betwee
206 proceeds through messages. The synchronizat
207 requesting Coda file service and Venus relie
208 up processes. The Coda FS driver processes
209 on behalf of a process P, creates messages f
210 and finally returns to the caller. The impl
211 of messages is platform specific, but the se
212 appeared to be generally applicable. Data b
213 FS Driver in kernel memory on behalf of P an
214 Venus.
215
216 The FS Driver while servicing P makes upcall
217 upcall is dispatched to Venus by creating a
218 structure contains the identification of P,
219 number, the size of the request and a pointe
220 memory for the request. Since the data buff
221 reply from Venus, there is a field for the s
222 field is used in the message to precisely re
223 message. Additional platform dependent stru
224 determine the position of the message on que
225 synchronization objects. In the upcall rout
226 is filled in, flags are set to 0, and it is
227 queue. The routine calling upcall is respon
228 data buffer; its structure will be described
229
230 A facility must exist to notify Venus that t
231 created, and implemented using available syn
232 the OS. This notification is done in the upc
233 P. When the message is on the pending queue,
234 in upcall. The (kernel mode) processing of
235 request routine must be suspended until Venu
236 the calling thread in P is blocked in upcall
237 message structure will locate the synchroniz
238 sleeping.
239
240 Venus detects the notification that a messag
241 driver allow Venus to retrieve the message w
242 call. This action finishes in the kernel by
243 queue of processing messages and setting fla
244 passed the contents of the data buffer. The
245 now returns and Venus processes the request.
246
247 At some later point the FS driver receives a
248 namely when Venus calls sendmsg_to_kernel.
249 driver looks at the contents of the message
250
251
252 * the message is a reply for a suspended th
253 the message from the processing queue and
254 WRITTEN. Finally, the FS driver unblocks
255 mode context of Venus) and the sendmsg_to
256 Venus. The process P will be scheduled a
257 processing its upcall with the data buffe
258 from Venus.
259
260 * The message is a ``downcall``. A downcal
261 the FS Driver. The FS driver processes th
262 (usually a cache eviction or replacement)
263 sendmsg_to_kernel returns.
264
265 Now P awakes and continues processing upcall
266 subtleties to take account of. First P will
267 up in upcall by a signal from some other sou
268 attempt to terminate P) or as is normally th
269 sendmsg_to_kernel call. In the normal case,
270 deallocate the message structure and return.
271 with its processing.
272
273
274 **Sleeping and IPC arrangements**
275
276 In case P is woken up by a signal and not by
277 at the flags field. If the message is not y
278 handle its signal without notifying Venus.
279 the request should not be processed, P can s
280 to indicate that it should disregard the pre
281 signals are put in the queue at the head, an
282 the message is already marked as WRITTEN it
283 processing. The VFS routine will now contin
284 involves more than one upcall, this can lead
285 extra field "handle_signals" could be added
286 to indicate points of no return have been pa
287
288
289
290 3.1. Implementation details
291 ----------------------------
292
293 The Unix implementation of this mechanism ha
294 implementation of a character device associa
295 retrieves messages by doing a read on the de
296 with a write and notification is through the
297 file descriptor for the device. The process
298 interruptible wait queue object.
299
300 In Windows NT and the DPMI Windows 95 implem
301 call is used. The DeviceIoControl call is d
302 from user memory to kernel memory with OPCOD
303 is issued as a synchronous call, while the g
304 asynchronous. Windows EventObjects are used
305 message arrival. The process P is kept wait
306 object in NT and a semaphore in Windows 95.
307
308
309 4. The interface at the call level
310 ===================================
311
312
313 This section describes the upcalls a Coda FS
314 Each of these upcalls make use of two struct
315 outputArgs. In pseudo BNF form the structu
316 form::
317
318
319 struct inputArgs {
320 u_long opcode;
321 u_long unique; /* Keep multipl
322 u_short pid; /* Co
323 u_short pgid; /* Co
324 struct CodaCred cred; /* Co
325
326 <union "in" of call dependent part
327 };
328
329 struct outputArgs {
330 u_long opcode;
331 u_long unique; /* Keep multi
332 u_long result;
333
334 <union "out" of call dependent par
335 };
336
337
338
339 Before going on let us elucidate the role of
340 inputArgs start with the opcode which define
341 requested from Venus. There are approximatel
342 which we will discuss. The unique field la
343 unique number which will identify the messag
344 process group id are passed. Finally the cr
345 are included.
346
347 Before delving into the specific calls we ne
348 data structures shared by the kernel and Ven
349
350
351
352
353 4.1. Data structures shared by the kernel and
354 ----------------------------------------------
355
356
357 The CodaCred structure defines a variety of
358 they are set for the calling process. The vu
359 unsigned integers. It also defines group me
360 Unix the CodaCred has proven sufficient to i
361 semantics for Coda but the structure may hav
362 for the Windows environment when these matur
363
364 struct CodaCred {
365 vuid_t cr_uid, cr_euid, cr_suid, c
366 vgid_t cr_gid, cr_egid, cr_sgid, c
367 vgid_t cr_groups[NGROUPS];
368 };
369
370
371 .. Note::
372
373 It is questionable if we need CodaCreds i
374 doesn't know about groups, although it do
375 default uid/gid. Perhaps the list of gro
376
377
378 The next item is the fundamental identifier
379 files, the ViceFid. A fid of a file uniquel
380 directory in the Coda filesystem within a ce
381
382 typedef struct ViceFid {
383 VolumeId Volume;
384 VnodeId Vnode;
385 Unique_t Unique;
386 } ViceFid;
387
388 .. [1] A cell is agroup of Coda servers acti
389 system control machine or SCM. See th
390 for a detailed description of the rol
391
392 Each of the constituent fields: VolumeId, Vn
393 unsigned 32 bit integers. We envisage that
394 to be prefixed to identify the Coda cell; th
395 form of a Ipv6 size IP address naming the Co
396
397 The next important structure shared between
398 the attributes of the file. The following s
399 exchange information. It has room for futur
400 support for device files (currently not pres
401
402
403 struct coda_timespec {
404 int64_t tv_sec;
405 long tv_nsec;
406 };
407
408 struct coda_vattr {
409 enum coda_vtype va_type;
410 u_short va_mode;
411 short va_nlink;
412 vuid_t va_uid;
413 vgid_t va_gid;
414 long va_fsid;
415 long va_fileid;
416 u_quad_t va_size;
417 long va_blocksize;
418 struct coda_timespec va_atime;
419 struct coda_timespec va_mtime;
420 struct coda_timespec va_ctime;
421 u_long va_gen;
422 u_long va_flags;
423 dev_t va_rdev;
424 u_quad_t va_bytes;
425 u_quad_t va_filerev;
426 u_int va_vaflags;
427 long va_spare;
428 };
429
430
431 4.2. The pioctl interface
432 --------------------------
433
434
435 Coda specific requests can be made by applic
436 interface. The pioctl is implemented as an o
437 fictitious file /coda/.CONTROL. The pioctl
438 a file handle and makes the ioctl call. Fina
439
440 The kernel involvement in this is limited to
441 open and close and pass the ioctl message an
442 the pioctl data buffers is a file in a Coda
443
444 The kernel is handed a data packet of the fo
445
446 struct {
447 const char *path;
448 struct ViceIoctl vidata;
449 int follow;
450 } data;
451
452
453
454 where::
455
456
457 struct ViceIoctl {
458 caddr_t in, out; /* Dat
459 short in_size; /* Siz
460 short out_size; /* Max
461 };
462
463
464
465 The path must be a Coda file, otherwise the
466 made.
467
468 .. Note:: The data structures and code are a
469
470
471 **We now proceed to document the individual ca
472
473
474 4.3. root
475 ----------
476
477
478 Arguments
479 in
480
481 empty
482
483 out::
484
485 struct cfs_root_out {
486 ViceFid VFid;
487 } cfs_root;
488
489
490
491 Description
492 This call is made to Venus during the init
493 the Coda filesystem. If the result is zero
494 contains the ViceFid of the root of the Co
495 result is generated, its value is a platfo
496 indicating the difficulty Venus encountere
497 the Coda filesystem.
498
499 4.4. lookup
500 ------------
501
502
503 Summary
504 Find the ViceFid and type of an object in
505
506 Arguments
507 in::
508
509 struct cfs_lookup_in {
510 ViceFid VFid;
511 char *name;
512 } cfs_lookup;
513
514
515
516 out::
517
518 struct cfs_lookup_out {
519 ViceFid VFid;
520 int vtype;
521 } cfs_lookup;
522
523
524
525 Description
526 This call is made to determine the ViceFid
527 a directory entry. The directory entry re
528 and Venus will search the directory identi
529 The result may indicate that the name does
530 difficulty was encountered in finding it (
531 If the result is zero, the field cfs_looku
532 targets ViceFid and cfs_lookup_out.vtype t
533 type of object the name designates.
534
535 The name of the object is an 8 bit character
536 CFS_MAXNAMLEN, currently set to 256 (includi
537
538 It is extremely important to realize that Ve
539 cfs_lookup.vtype with CFS_NOCACHE to indicat
540 not be put in the kernel name cache.
541
542 .. Note::
543
544 The type of the vtype is currently wrong.
545 coda_vtype. Linux does not take note of C
546
547
548 4.5. getattr
549 -------------
550
551
552 Summary Get the attributes of a file.
553
554 Arguments
555 in::
556
557 struct cfs_getattr_in {
558 ViceFid VFid;
559 struct coda_vattr attr; /*
560 } cfs_getattr;
561
562
563
564 out::
565
566 struct cfs_getattr_out {
567 struct coda_vattr attr;
568 } cfs_getattr;
569
570
571
572 Description
573 This call returns the attributes of the fi
574
575 Errors
576 Errors can occur if the object with fid do
577 unaccessible or if the caller does not hav
578 attributes.
579
580 .. Note::
581
582 Many kernel FS drivers (Linux, NT and Win
583 the attributes as well as the Fid for the
584 "inode" or "FileHandle". A significant i
585 such systems could be made by combining t
586 both at the Venus/kernel interaction leve
587
588 The vattr structure included in the input ar
589 should be removed.
590
591
592 4.6. setattr
593 -------------
594
595
596 Summary
597 Set the attributes of a file.
598
599 Arguments
600 in::
601
602 struct cfs_setattr_in {
603 ViceFid VFid;
604 struct coda_vattr attr;
605 } cfs_setattr;
606
607
608
609
610 out
611
612 empty
613
614 Description
615 The structure attr is filled with attribut
616 in BSD style. Attributes not to be change
617 vtype which is set to VNON. Other are set
618 The only attributes which the FS driver ma
619 mode, owner, groupid, atime, mtime and cti
620 indicates success or failure.
621
622 Errors
623 A variety of errors can occur. The object
624 be inaccessible, or permission may not be
625
626
627 4.7. access
628 ------------
629
630
631 Arguments
632 in::
633
634 struct cfs_access_in {
635 ViceFid VFid;
636 int flags;
637 } cfs_access;
638
639
640
641 out
642
643 empty
644
645 Description
646 Verify if access to the object identified
647 operations described by flags is permitted
648 access will be granted. It is important t
649 ACLs to enforce protection and that ultima
650 clients enforce the security of the system
651 will depend on whether a token is held by
652
653 Errors
654 The object may not exist, or the ACL descr
655 may not be accessible.
656
657
658 4.8. create
659 ------------
660
661
662 Summary
663 Invoked to create a file
664
665 Arguments
666 in::
667
668 struct cfs_create_in {
669 ViceFid VFid;
670 struct coda_vattr attr;
671 int excl;
672 int mode;
673 char *name;
674 } cfs_create;
675
676
677
678
679 out::
680
681 struct cfs_create_out {
682 ViceFid VFid;
683 struct coda_vattr attr;
684 } cfs_create;
685
686
687
688 Description
689 This upcall is invoked to request creation
690 The file will be created in the directory
691 will be name, and the mode will be mode.
692 be returned if the file already exists. I
693 set to zero the file will be truncated. T
694 are set by converting the CodaCred to a ui
695 (this macro is platform dependent). Upon
696 attributes of the file are returned. The
697 instantiate a vnode, inode or file handle
698 object.
699
700
701 Errors
702 A variety of errors can occur. Permissions
703 If the object exists and is not a file the
704 under Unix.
705
706 .. Note::
707
708 The packing of parameters is very ineffic
709 indicate confusion between the system cal
710 create. The VFS operation create is only
711 This create call differs from the Unix on
712 to return a file descriptor. The truncate
713 together with the mode, could simply be p
714 under Unix. There should be no flags arg
715 (2) to return a file descriptor for READ
716
717 The attributes of the directory should be re
718 and mtime changed.
719
720
721 4.9. mkdir
722 -----------
723
724
725 Summary
726 Create a new directory.
727
728 Arguments
729 in::
730
731 struct cfs_mkdir_in {
732 ViceFid VFid;
733 struct coda_vattr attr;
734 char *name;
735 } cfs_mkdir;
736
737
738
739 out::
740
741 struct cfs_mkdir_out {
742 ViceFid VFid;
743 struct coda_vattr attr;
744 } cfs_mkdir;
745
746
747
748
749 Description
750 This call is similar to create but creates
751 Only the mode field in the input parameter
752 Upon successful creation, the attr returne
753 the new directory.
754
755 Errors
756 As for create.
757
758 .. Note::
759
760 The input parameter should be changed to
761 attributes.
762
763 The attributes of the parent should be retur
764 mtime changes.
765
766
767 4.10. link
768 -----------
769
770
771 Summary
772 Create a link to an existing file.
773
774 Arguments
775 in::
776
777 struct cfs_link_in {
778 ViceFid sourceFid;
779 ViceFid destFid;
780 char *tname;
781 } cfs_link;
782
783
784
785 out
786
787 empty
788
789 Description
790 This call creates a link to the sourceFid
791 identified by destFid with name tname. Th
792 target's parent, i.e. the source must be h
793 does not support cross directory hard link
794 relevant. It indicates success or the typ
795
796 Errors
797 The usual errors can occur.
798
799
800 4.11. symlink
801 --------------
802
803
804 Summary
805 create a symbolic link
806
807 Arguments
808 in::
809
810 struct cfs_symlink_in {
811 ViceFid VFid;
812 char *srcname;
813 struct coda_vattr attr;
814 char *tname;
815 } cfs_symlink;
816
817
818
819 out
820
821 none
822
823 Description
824 Create a symbolic link. The link is to be
825 directory identified by VFid and named tna
826 pathname srcname. The attributes of the n
827 be set to attr.
828
829 .. Note::
830
831 The attributes of the target directory sh
832 its size changed.
833
834
835 4.12. remove
836 -------------
837
838
839 Summary
840 Remove a file
841
842 Arguments
843 in::
844
845 struct cfs_remove_in {
846 ViceFid VFid;
847 char *name;
848 } cfs_remove;
849
850
851
852 out
853
854 none
855
856 Description
857 Remove file named cfs_remove_in.name in di
858 identified by VFid.
859
860
861 .. Note::
862
863 The attributes of the directory should be
864 mtime and size may change.
865
866
867 4.13. rmdir
868 ------------
869
870
871 Summary
872 Remove a directory
873
874 Arguments
875 in::
876
877 struct cfs_rmdir_in {
878 ViceFid VFid;
879 char *name;
880 } cfs_rmdir;
881
882
883
884 out
885
886 none
887
888 Description
889 Remove the directory with name 'name' from
890 identified by VFid.
891
892 .. Note:: The attributes of the parent direc
893 its mtime and size may change.
894
895
896 4.14. readlink
897 ---------------
898
899
900 Summary
901 Read the value of a symbolic link.
902
903 Arguments
904 in::
905
906 struct cfs_readlink_in {
907 ViceFid VFid;
908 } cfs_readlink;
909
910
911
912 out::
913
914 struct cfs_readlink_out {
915 int count;
916 caddr_t data;
917 } cfs_readlink;
918
919
920
921 Description
922 This routine reads the contents of symboli
923 identified by VFid into the buffer data.
924 to hold any name up to CFS_MAXNAMLEN (PATH
925
926 Errors
927 No unusual errors.
928
929
930 4.15. open
931 -----------
932
933
934 Summary
935 Open a file.
936
937 Arguments
938 in::
939
940 struct cfs_open_in {
941 ViceFid VFid;
942 int flags;
943 } cfs_open;
944
945
946
947 out::
948
949 struct cfs_open_out {
950 dev_t dev;
951 ino_t inode;
952 } cfs_open;
953
954
955
956 Description
957 This request asks Venus to place the file
958 VFid in its cache and to note that the cal
959 it with flags as in open(2). The return v
960 for Unix and Windows systems. For Unix sy
961 informed of the device and inode number of
962 fields dev and inode. For Windows the pat
963 returned to the kernel.
964
965
966 .. Note::
967
968 Currently the cfs_open_out structure is n
969 deal with the Windows case. It might be
970 upcalls, one to open aiming at a containe
971 container file inode.
972
973
974 4.16. close
975 ------------
976
977
978 Summary
979 Close a file, update it on the servers.
980
981 Arguments
982 in::
983
984 struct cfs_close_in {
985 ViceFid VFid;
986 int flags;
987 } cfs_close;
988
989
990
991 out
992
993 none
994
995 Description
996 Close the file identified by VFid.
997
998 .. Note::
999
1000 The flags argument is bogus and not used
1001 has room to deal with an execp input fie
1002 be used to inform Venus that the file wa
1003 mapped for execution. There are comment
1004 fetching the data in Venus vproc_vfscall
1005 file is being closed, the data in the co
1006 data. Here again the execp flag might b
1007 currently Venus might think a file can b
1008 it is still memory mapped. This needs t
1009
1010
1011 4.17. ioctl
1012 ------------
1013
1014
1015 Summary
1016 Do an ioctl on a file. This includes the
1017
1018 Arguments
1019 in::
1020
1021 struct cfs_ioctl_in {
1022 ViceFid VFid;
1023 int cmd;
1024 int len;
1025 int rwflag;
1026 char *data;
1027 } cfs_ioctl;
1028
1029
1030
1031 out::
1032
1033
1034 struct cfs_ioctl_out {
1035 int len;
1036 caddr_t data;
1037 } cfs_ioctl;
1038
1039
1040
1041 Description
1042 Do an ioctl operation on a file. The com
1043 data arguments are filled as usual. flag
1044
1045 .. Note::
1046
1047 Another bogus parameter. flags is not u
1048 business about PREFETCHING in the Venus
1049
1050
1051
1052 4.18. rename
1053 -------------
1054
1055
1056 Summary
1057 Rename a fid.
1058
1059 Arguments
1060 in::
1061
1062 struct cfs_rename_in {
1063 ViceFid sourceFid;
1064 char *srcname;
1065 ViceFid destFid;
1066 char *destname;
1067 } cfs_rename;
1068
1069
1070
1071 out
1072
1073 none
1074
1075 Description
1076 Rename the object with name srcname in di
1077 sourceFid to destname in destFid. It is
1078 srcname and destname are 0 terminated str
1079 kernels are not always null terminated.
1080
1081
1082 4.19. readdir
1083 --------------
1084
1085
1086 Summary
1087 Read directory entries.
1088
1089 Arguments
1090 in::
1091
1092 struct cfs_readdir_in {
1093 ViceFid VFid;
1094 int count;
1095 int offset;
1096 } cfs_readdir;
1097
1098
1099
1100
1101 out::
1102
1103 struct cfs_readdir_out {
1104 int size;
1105 caddr_t data;
1106 } cfs_readdir;
1107
1108
1109
1110 Description
1111 Read directory entries from VFid starting
1112 read at most count bytes. Returns the da
1113 the size in size.
1114
1115
1116 .. Note::
1117
1118 This call is not used. Readdir operatio
1119 files. We will re-evaluate this during
1120 about to take place.
1121
1122
1123 4.20. vget
1124 -----------
1125
1126
1127 Summary
1128 instructs Venus to do an FSDB->Get.
1129
1130 Arguments
1131 in::
1132
1133 struct cfs_vget_in {
1134 ViceFid VFid;
1135 } cfs_vget;
1136
1137
1138
1139 out::
1140
1141 struct cfs_vget_out {
1142 ViceFid VFid;
1143 int vtype;
1144 } cfs_vget;
1145
1146
1147
1148 Description
1149 This upcall asks Venus to do a get operat
1150 labelled by VFid.
1151
1152 .. Note::
1153
1154 This operation is not used. However, it
1155 since it can be used to deal with read/w
1156 These can be "pinned" in the Venus cache
1157 inactive.
1158
1159
1160 4.21. fsync
1161 ------------
1162
1163
1164 Summary
1165 Tell Venus to update the RVM attributes o
1166
1167 Arguments
1168 in::
1169
1170 struct cfs_fsync_in {
1171 ViceFid VFid;
1172 } cfs_fsync;
1173
1174
1175
1176 out
1177
1178 none
1179
1180 Description
1181 Ask Venus to update RVM attributes of obj
1182 should be called as part of kernel level
1183 result indicates if the syncing was succe
1184
1185 .. Note:: Linux does not implement this cal
1186
1187
1188 4.22. inactive
1189 ---------------
1190
1191
1192 Summary
1193 Tell Venus a vnode is no longer in use.
1194
1195 Arguments
1196 in::
1197
1198 struct cfs_inactive_in {
1199 ViceFid VFid;
1200 } cfs_inactive;
1201
1202
1203
1204 out
1205
1206 none
1207
1208 Description
1209 This operation returns EOPNOTSUPP.
1210
1211 .. Note:: This should perhaps be removed.
1212
1213
1214 4.23. rdwr
1215 -----------
1216
1217
1218 Summary
1219 Read or write from a file
1220
1221 Arguments
1222 in::
1223
1224 struct cfs_rdwr_in {
1225 ViceFid VFid;
1226 int rwflag;
1227 int count;
1228 int offset;
1229 int ioflag;
1230 caddr_t data;
1231 } cfs_rdwr;
1232
1233
1234
1235
1236 out::
1237
1238 struct cfs_rdwr_out {
1239 int rwflag;
1240 int count;
1241 caddr_t data; /* Pl
1242 } cfs_rdwr;
1243
1244
1245
1246 Description
1247 This upcall asks Venus to read or write f
1248
1249
1250 .. Note::
1251
1252 It should be removed since it is against
1253 read/write operations never reach Venus.
1254 operation does not work. It is not curre
1255
1256
1257
1258 4.24. odymount
1259 ---------------
1260
1261
1262 Summary
1263 Allows mounting multiple Coda "filesystem
1264
1265 Arguments
1266 in::
1267
1268 struct ody_mount_in {
1269 char *name;
1270 } ody_mount;
1271
1272
1273
1274 out::
1275
1276 struct ody_mount_out {
1277 ViceFid VFid;
1278 } ody_mount;
1279
1280
1281
1282 Description
1283 Asks Venus to return the rootfid of a Cod
1284 name. The fid is returned in VFid.
1285
1286 .. Note::
1287
1288 This call was used by David for dynamic
1289 removed since it causes a jungle of poin
1290 It is not used by Coda proper. Call is
1291
1292
1293 4.25. ody_lookup
1294 -----------------
1295
1296
1297 Summary
1298 Looks up something.
1299
1300 Arguments
1301 in
1302
1303 irrelevant
1304
1305
1306 out
1307
1308 irrelevant
1309
1310
1311 .. Note:: Gut it. Call is not implemented b
1312
1313
1314 4.26. ody_expand
1315 -----------------
1316
1317
1318 Summary
1319 expands something in a dynamic set.
1320
1321 Arguments
1322 in
1323
1324 irrelevant
1325
1326 out
1327
1328 irrelevant
1329
1330 .. Note:: Gut it. Call is not implemented b
1331
1332
1333 4.27. prefetch
1334 ---------------
1335
1336
1337 Summary
1338 Prefetch a dynamic set.
1339
1340 Arguments
1341
1342 in
1343
1344 Not documented.
1345
1346 out
1347
1348 Not documented.
1349
1350 Description
1351 Venus worker.cc has support for this call
1352 noted that it doesn't work. Not surprisi
1353 have support for it. (ODY_PREFETCH is not
1354
1355
1356 .. Note:: Gut it. It isn't working and isn'
1357
1358
1359
1360 4.28. signal
1361 -------------
1362
1363
1364 Summary
1365 Send Venus a signal about an upcall.
1366
1367 Arguments
1368 in
1369
1370 none
1371
1372 out
1373
1374 not applicable.
1375
1376 Description
1377 This is an out-of-band upcall to Venus to
1378 that the calling process received a signa
1379 message from the input queue. Venus is s
1380 operation.
1381
1382 Errors
1383 No reply is given.
1384
1385 .. Note::
1386
1387 We need to better understand what Venus
1388 it is doing this correctly. Also we nee
1389 per system call situations correctly. I
1390 what state changes in Venus take place a
1391 kernel is responsible for notifying Venu
1392 definitely is such a state change, but m
1393
1394
1395 5. The minicache and downcalls
1396 ===============================
1397
1398
1399 The Coda FS Driver can cache results of loo
1400 limit the frequency of upcalls. Upcalls ca
1401 context switch needs to take place. The co
1402 information is that Venus will notify the F
1403 entries must be flushed or renamed.
1404
1405 The kernel code generally has to maintain a
1406 internal file handles (called vnodes in BSD
1407 FileHandles in Windows) with the ViceFid's
1408 reason is that frequent translations back a
1409 order to make upcalls and use the results o
1410 objects are called cnodes.
1411
1412 The current minicache implementations have
1413 the following:
1414
1415 1. the name of the file
1416
1417 2. the cnode of the directory containing th
1418
1419 3. a list of CodaCred's for which the looku
1420
1421 4. the cnode of the object
1422
1423 The lookup call in the Coda FS Driver may r
1424 desired object from the cache, by passing i
1425 CodaCred's of the caller. The cache will r
1426 that it cannot be found. The Coda FS Drive
1427 invalidate cache entries when it modifies o
1428
1429 When Venus obtains information that indicat
1430 no longer valid, it will make a downcall to
1431 intercepted by the Coda FS Driver and lead
1432 the kind described below. The Coda FS Driv
1433 unless the downcall data could not be read
1434
1435
1436 5.1. INVALIDATE
1437 ----------------
1438
1439
1440 No information is available on this call.
1441
1442
1443 5.2. FLUSH
1444 -----------
1445
1446
1447
1448 Arguments
1449 None
1450
1451 Summary
1452 Flush the name cache entirely.
1453
1454 Description
1455 Venus issues this call upon startup and w
1456 is to prevent stale cache information bei
1457 systems allow the kernel name cache to be
1458 When this is done, this downcall is made.
1459
1460
1461 5.3. PURGEUSER
1462 ---------------
1463
1464
1465 Arguments
1466 ::
1467
1468 struct cfs_purgeuser_out {/* CFS_PU
1469 struct CodaCred cred;
1470 } cfs_purgeuser;
1471
1472
1473
1474 Description
1475 Remove all entries in the cache carrying
1476 call is issued when tokens for a user exp
1477
1478
1479 5.4. ZAPFILE
1480 -------------
1481
1482
1483 Arguments
1484 ::
1485
1486 struct cfs_zapfile_out { /* CFS_ZA
1487 ViceFid CodaFid;
1488 } cfs_zapfile;
1489
1490
1491
1492 Description
1493 Remove all entries which have the (dir vn
1494 This is issued as a result of an invalida
1495 a vnode.
1496
1497 .. Note::
1498
1499 Call is not named correctly in NetBSD an
1500 zapfile routine takes different argument
1501 the invalidation of attributes correctly
1502
1503
1504
1505 5.5. ZAPDIR
1506 ------------
1507
1508
1509 Arguments
1510 ::
1511
1512 struct cfs_zapdir_out { /* CFS_ZA
1513 ViceFid CodaFid;
1514 } cfs_zapdir;
1515
1516
1517
1518 Description
1519 Remove all entries in the cache lying in
1520 CodaFid, and all children of this directo
1521 Venus receives a callback on the director
1522
1523
1524 5.6. ZAPVNODE
1525 --------------
1526
1527
1528
1529 Arguments
1530 ::
1531
1532 struct cfs_zapvnode_out { /* CFS_ZA
1533 struct CodaCred cred;
1534 ViceFid VFid;
1535 } cfs_zapvnode;
1536
1537
1538
1539 Description
1540 Remove all entries in the cache carrying
1541 as in the arguments. This downcall is pro
1542
1543
1544 5.7. PURGEFID
1545 --------------
1546
1547
1548 Arguments
1549 ::
1550
1551 struct cfs_purgefid_out { /* CFS_PU
1552 ViceFid CodaFid;
1553 } cfs_purgefid;
1554
1555
1556
1557 Description
1558 Flush the attribute for the file. If it i
1559 vnode), purge its children from the namec
1560 namecache.
1561
1562
1563
1564 5.8. REPLACE
1565 -------------
1566
1567
1568 Summary
1569 Replace the Fid's for a collection of nam
1570
1571 Arguments
1572 ::
1573
1574 struct cfs_replace_out { /* cfs_rep
1575 ViceFid NewFid;
1576 ViceFid OldFid;
1577 } cfs_replace;
1578
1579
1580
1581 Description
1582 This routine replaces a ViceFid in the na
1583 another. It is added to allow Venus duri
1584 locally allocated temp fids while disconn
1585 when the reference counts on those fids a
1586
1587
1588 6. Initialization and cleanup
1589 ==============================
1590
1591
1592 This section gives brief hints as to desira
1593 FS Driver at startup and upon shutdown or V
1594 entering the discussion it is useful to rep
1595 maintains the following data:
1596
1597
1598 1. message queues
1599
1600 2. cnodes
1601
1602 3. name cache entries
1603
1604 The name cache entries are entirely priv
1605 can easily be manipulated. The message
1606 clear points of initialization and destr
1607 much more delicate. User processes hold
1608 filesystems and it can be difficult to c
1609
1610 It can expect requests through:
1611
1612 1. the message subsystem
1613
1614 2. the VFS layer
1615
1616 3. pioctl interface
1617
1618 Currently the pioctl passes through the
1619 treat these similarly.
1620
1621
1622 6.1. Requirements
1623 ------------------
1624
1625
1626 The following requirements should be accomm
1627
1628 1. The message queues should have open and
1629 the opening of the character devices are
1630
1631 - Before opening, no messages can be pla
1632
1633 - Opening will remove any old messages s
1634
1635 - Close will notify any sleeping process
1636 be completed.
1637
1638 - Close will free all memory allocated b
1639
1640
1641 2. At open the namecache shall be initializ
1642
1643 3. Before the message queues are open, all
1644 Fortunately this can be achieved by maki
1645 Coda filesystem cannot succeed before op
1646
1647 4. After closing of the queues, no VFS oper
1648 one needs to be careful, since a few ope
1649 read/write, readdir) can proceed without
1650 explicitly blocked.
1651
1652 5. Upon closing the namecache shall be flus
1653
1654 6. All memory held by cnodes can be freed w
1655
1656 7. Unmounting the file system can be done w
1657
1658 8. Mounting the Coda filesystem should fail
1659 get the rootfid or the attributes of the
1660 best implemented by Venus fetching these
1661 to mount.
1662
1663 .. Note::
1664
1665 NetBSD in particular but also Linux have
1666 above requirements fully. For smooth op
1667 corrected.
1668
1669
1670
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.