[CFR] OpenBSD install(1) fixes: atomic install, etc.
Hi!
The attached patch incorporates most of OpenBSD fixes to install(1).
It does not include manpage update. Most significant changes are:
o All TODOs are acted upon.
o New flags: -b and -B
: -bBackup any existing files before overwriting them by
: renaming them to file.old. See -B for specifying a
: different backup suffix.
:
: -B suffix
: Use suffix as the backup suffix if -b is given.
o New flag: -S (atomic install)
: -SSafe copy. Normally, install unlinks an existing target before
: installing the new file. With the -S flag a temporary file is
: used and then renamed to be the target. The reason this is safer
: is that if the copy or rename fails, the existing target is left
: untouched.
o The -c flag is now the default, and only provided for backwards
compatibility. We now never remove the original file.
o Flags -v and -D were withdrawn.
o strip(1) failure is not considered fatal.
Please review.
Thanks,
--
Ruslan Ermilov Oracle Developer/DBA,
[EMAIL PROTECTED] Sunbay Software AG,
[EMAIL PROTECTED] FreeBSD committer,
+380.652.512.251Simferopol, Ukraine
http://www.FreeBSD.org The Power To Serve
http://www.oracle.com Enabling The Information Age
Index: xinstall.c
===
RCS file: /home/ncvs/src/usr.bin/xinstall/xinstall.c,v
retrieving revision 1.40
diff -u -p -r1.40 xinstall.c
--- xinstall.c 2000/10/08 09:17:56 1.40
+++ xinstall.c 2001/04/20 15:42:20
@@ -45,21 +45,6 @@ static const char rcsid[] =
"$FreeBSD: src/usr.bin/xinstall/xinstall.c,v 1.40 2000/10/08 09:17:56 bde Exp $";
#endif /* not lint */
-/*-
- * Todo:
- * o for -C, compare original files except in -s case.
- * o for -C, don't change anything if nothing needs be changed. In
- * particular, don't toggle the immutable flags just to allow null
- * attribute changes and don't clear the dump flag. (I think inode
- * ctimes are not updated for null attribute changes, but this is a
- * bug.)
- * o independent of -C, if a copy must be made, then copy to a tmpfile,
- * set all attributes except the immutable flags, then rename, then
- * set the immutable flags. It's annoying that the immutable flags
- * defeat the atomicicity of rename - it seems that there must be
- * a window where the target is not immutable.
- */
-
#include sys/param.h
#include sys/wait.h
#include sys/mman.h
@@ -82,45 +67,29 @@ static const char rcsid[] =
#include "pathnames.h"
-/* Bootstrap aid - this doesn't exist in most older releases */
-#ifndef MAP_FAILED
-#define MAP_FAILED ((void *)-1)/* from sys/mman.h */
-#endif
-
-int debug, docompare, docopy, dodir, dopreserve, dostrip, nommap, verbose;
-int mode = S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
-char *group, *owner, pathbuf[MAXPATHLEN];
-char pathbuf2[MAXPATHLEN];
-
#defineDIRECTORY 0x01/* Tell install it's a directory. */
#defineSETFLAGS0x02/* Tell install to set flags. */
#defineNOCHANGEBITS(UF_IMMUTABLE | UF_APPEND | SF_IMMUTABLE | SF_APPEND)
+#defineBACKUP_SUFFIX ".old"
+struct passwd *pp;
+struct group *gp;
+int dobackup, docompare, dodir, dopreserve, dostrip, nommap, safecopy;
+int mode = S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH;
+char pathbuf[MAXPATHLEN], tempfile[MAXPATHLEN];
+char *suffix = BACKUP_SUFFIX;
+uid_t uid;
+gid_t gid;
+
void copy __P((int, char *, int, char *, off_t));
-intcompare __P((int, const char *, int, const char *,
-const struct stat *, const struct stat *));
+intcompare __P((int, const char *, size_t, int, const char *, size_t));
+intcreate_newfile __P((char *, struct stat *));
+intcreate_tempfile __P((char *, char *, size_t));
void install __P((char *, char *, u_long, u_int));
void install_dir __P((char *));
void strip __P((char *));
-void usage __P((void));
inttrymmap __P((int));
-
-#define ALLOW_NUMERIC_IDS 1
-#ifdef ALLOW_NUMERIC_IDS
-
-uid_t uid = -1;
-gid_t gid = -1;
-
-uid_t resolve_uid __P((char *));
-gid_t resolve_gid __P((char *));
-u_long numeric_id __P((char *, char *));
-
-#else
-
-struct passwd *pp;
-struct group *gp;
-
-#endif /* ALLOW_NUMERIC_IDS */
+void usage __P((void));
int
main(argc, argv)
@@ -132,20 +101,23 @@ main(argc, argv)
u_long fset;
u_int iflags;
int ch, no_target;
- char *flags, *to_name;
+ char *flags, *to_name, *group = NULL, *owner = NULL;
iflags = 0;
- while ((ch = getopt(argc, argv, "CcdDf:g:m:Mo:psv")) != -1)
+ while ((ch = getopt(argc, argv, "BbCcdf:g:Mm:o:pSs")) != -1)
switch((char)ch) {
+ case 'B':
+ suffix = optarg;
+ /* FALLTHROUGH */
+ case 'b':
+ dobackup = 1;
+ break;
case 'C':
-
Re: [CFR] OpenBSD install(1) fixes: atomic install, etc.
Hi, Ruslan Ermilov wrote: The attached patch incorporates most of OpenBSD fixes to install(1). It does not include manpage update. Most significant changes are: o New flag: -S (atomic install) : -SSafe copy. Normally, install unlinks an existing target before : installing the new file. With the -S flag a temporary file is : used and then renamed to be the target. The reason this is safer : is that if the copy or rename fails, the existing target is left : untouched. Just curious: why not make this way of doing install default (i.e. always use it)? Regards, Konstantin. -- * *Konstantin Chuguev - Application Engineer * * Francis House, 112 Hills Road * Cambridge CB2 1PQ, United Kingdom D A N T E WWW:http://www.dante.net To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: [CFR] OpenBSD install(1) fixes: atomic install, etc.
Konstantin Chuguev wrote: Hi, Ruslan Ermilov wrote: The attached patch incorporates most of OpenBSD fixes to install(1). It does not include manpage update. Most significant changes are: o New flag: -S (atomic install) : -SSafe copy. Normally, install unlinks an existing target before : installing the new file. With the -S flag a temporary file is : used and then renamed to be the target. The reason this is safer : is that if the copy or rename fails, the existing target is left : untouched. Just curious: why not make this way of doing install default (i.e. always use it)? It may effectively doubles disk space requirements during copy (when destination file is not on a sofdep-enabled partition and is not open at the moment when install(8) unlinks it). For small files it doesn't matter, but for a big ones it could lead to a problem. -Maxim To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: [CFR] OpenBSD install(1) fixes: atomic install, etc.
On Fri, Apr 20, 2001 at 07:41:36PM +0300, Maxim Sobolev wrote: Konstantin Chuguev wrote: Hi, Ruslan Ermilov wrote: The attached patch incorporates most of OpenBSD fixes to install(1). It does not include manpage update. Most significant changes are: o New flag: -S (atomic install) : -SSafe copy. Normally, install unlinks an existing target before : installing the new file. With the -S flag a temporary file is : used and then renamed to be the target. The reason this is safer : is that if the copy or rename fails, the existing target is left : untouched. Just curious: why not make this way of doing install default (i.e. always use it)? It may effectively doubles disk space requirements during copy (when destination file is not on a sofdep-enabled partition and is not open at the moment when install(8) unlinks it). For small files it doesn't matter, but for a big ones it could lead to a problem. I think -S should be made the default for `installworld' (this was the main reason that triggered this work). But I don't think this should be turned `on' by default in install(1). BTW, this binary just completed the `installworld' test on -STABLE. Cheers, -- Ruslan Ermilov Oracle Developer/DBA, [EMAIL PROTECTED] Sunbay Software AG, [EMAIL PROTECTED] FreeBSD committer, +380.652.512.251Simferopol, Ukraine http://www.FreeBSD.org The Power To Serve http://www.oracle.com Enabling The Information Age To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
Re: [CFR] OpenBSD install(1) fixes: atomic install, etc.
On Fri, Apr 20, 2001 at 08:03:20PM +0300, Ruslan Ermilov wrote: On Fri, Apr 20, 2001 at 07:41:36PM +0300, Maxim Sobolev wrote: Konstantin Chuguev wrote: Hi, Ruslan Ermilov wrote: The attached patch incorporates most of OpenBSD fixes to install(1). It does not include manpage update. Most significant changes are: o New flag: -S (atomic install) : -SSafe copy. Normally, install unlinks an existing target before : installing the new file. With the -S flag a temporary file is : used and then renamed to be the target. The reason this is safer : is that if the copy or rename fails, the existing target is left : untouched. Just curious: why not make this way of doing install default (i.e. always use it)? It may effectively doubles disk space requirements during copy (when destination file is not on a sofdep-enabled partition and is not open at the moment when install(8) unlinks it). For small files it doesn't matter, but for a big ones it could lead to a problem. I think -S should be made the default for `installworld' (this was the main reason that triggered this work). But I don't think this should be turned `on' by default in install(1). BTW, this binary just completed the `installworld' test on -STABLE. Just tried -j32 installworld with modified install(1) binary for which -S is the default, and it still choked on lib/libncurses. The problem is in bsd.man.mk: we should separate installing of manpages and their MLINKS, otherwise links could be attempted before their originals are installed. Continuing with -DNOMAN now... Cheers, -- Ruslan Ermilov Oracle Developer/DBA, [EMAIL PROTECTED] Sunbay Software AG, [EMAIL PROTECTED] FreeBSD committer, +380.652.512.251Simferopol, Ukraine http://www.FreeBSD.org The Power To Serve http://www.oracle.com Enabling The Information Age To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
