Lines Matching +full:not +full:- +full:swapped

4 .\" Copyright (C) Caldera International Inc. 2001-2002.  All rights reserved.
29 .\" IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
33 .\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
40 .EH 'PSD:2-%''UNIX Implementation'
41 .OH 'UNIX Implementation''PSD:2-%'
55 \&\\$3\s-1\\$1\\s0\&\\$2
69 .AU "MH 2C-523" 2394
75 This paper describes in high-level terms the
101 functions not possible in C.
116 This does not mean to allow the user
120 but have that way be the least-common divisor
123 What is or is not implemented in the kernel
125 It is a soap-box platform on
159 from a read-only text segment,
165 from shared-text segments.
169 that there is no need to swap read-only
173 programs that tend to be swapped while
179 from the same copy of a read-only segment,
185 is not common.
194 All current read-only text segments in the
206 When a process first executes a shared-text segment,
214 read-write data
218 the system does not use the user's
233 Also associated and swapped with
234 a process is a small fixed-size
248 The system data segment is not
258 .IT not
296 if the parent process was executing from a read-only
309 (usually non-identical)
326 .IT not
366 SL-5.
377 are swapped to and from secondary
382 (When low-latency devices, such as bubbles,
389 by the same simple first-fit algorithm.
395 If there is not enough primary memory,
397 The process is swapped out onto the
399 ready to be swapped in with
408 that is swapped out and is
418 that can be swapped out.
428 are swapped out is to be swapped in?
431 The one with the longest time out is swapped in first.
434 are loaded is to be swapped out?
436 (i.e., not currently running or waiting for
443 but are not taken out unless they are
453 This is not bad in itself, because
454 the swapping does not impact the
526 to the event-wait mechanism.
536 is adapted to multiple-processor configurations.
541 The event-wait code in the kernel
542 is like a co-routine linkage.
544 all but one of the processes has called event-wait.
546 When it calls event-wait,
549 returns from its call to event-wait.
559 and time-of-day events are very low.
563 All user-process priorities are lower than the
565 User-process priorities are assigned
570 compute time in the last real-time
582 The compute-to-real-time ratio is updated
587 scheduled round-robin with a
588 1-second quantum.
589 A high-priority process waking up will
590 preempt a running, low-priority process.
596 At the same time, if a low-priority
665 If the block is not in the cache,
703 but does not solve, the problem.
712 On non-random devices,
722 devices that do not fall into the block I/O model.
727 they are not used in a stereotyped way,
728 for example, 80-byte physical records on tape
729 and track-at-a-time disk copies.
758 that the user is not swapped during this
764 The only really disk-specific code in normal
765 disk drivers is the pre-sort of transactions to
771 Real character-oriented devices may
787 A typical character-output device
825 to insert real-time delay after certain control characters.
830 Some device-dependent code conversion and
849 but do not fit the disk I/O mold.
861 a file is a (one-dimensional) array of bytes.
885 512-byte blocks.
887 four self-identifying regions.
892 contains the so-called ``super-block.''
897 Next comes the i-list,
900 a 64-byte structure, called an i-node.
901 The offset of a particular i-node
902 within the i-list is called its i-number.
904 (major and minor numbers) and i-number
906 After the i-list,
922 Since all allocation is in fixed-size
932 An i-node contains 13 disk addresses.
952 It contains 16-byte entries consisting of
953 a 14-byte name and an i-number.
954 The root of the hierarchy is at a known i-number
990 there are 25,000 files containing 130M bytes of data-file content.
991 The overhead (i-node, indirect blocks, and last block breakage)
1004 Because the i-node defines a file,
1006 around access to the i-node.
1008 i-nodes.
1010 the system locates the corresponding i-node,
1011 allocates an i-node table entry, and reads
1012 the i-node into primary memory.
1015 version of the i-node.
1016 Modifications to the i-node are made to
1018 When the last access to the i-node goes
1021 secondary store i-list and the table entry is freed.
1033 with the aid of the corresponding i-node table entry.
1036 The user is not aware of i-nodes and i-numbers.
1039 Converting a path name into an i-node table entry
1041 Starting at some known i-node
1045 This gives an i-number and an implied device
1047 Thus the next i-node table entry can be accessed.
1049 then this i-node is the result.
1050 If not,
1051 this i-node is the directory needed to look up
1069 corresponding i-node table entries.
1094 in the i-node table nor can
1107 only share the i-node table entry,
1112 converts a file system path name into an i-node
1114 A pointer to the i-node table entry is placed in a
1119 first creates a new i-node entry,
1120 writes the i-number into a directory, and
1126 just access the i-node entry as described above.
1136 the i-node table entries to free these structures after
1140 number of directories pointing at the given i-node.
1141 When the last reference to an i-node table entry
1143 if the i-node has no directories pointing to it,
1144 then the file is removed and the i-node is freed.
1165 first-in-first-out.
1193 pairs of designated leaf i-nodes and
1195 When converting a path name into an i-node,
1196 a check is made to see if the new i-node is a
1199 the i-node of the root
1206 mounted devices does not have a common pool of
1216 does for the user\-a
1220 Most of these things are not very
1223 does not need them.
1225 applications, for example, better inter-process communication.
1241 kernel does not support
1255 log-in,
1256 or log-out.