When system calls indicate an error they return -1, not some arbitrary
value < 0. errno is only updated in this case. Change all (most?)
callers of syscalls to follow this better, and let's see if this strictness
helps us in the future.

Implement cp -a ("archive" mode)

Not standard, but supported at least by GNU cp, IllumOS, NetBSD,
FreeBSD, DragonflyBSD. No need to be gratuitously different. This
should also allow us to drop patches in ~10 ports.

Based on an initial diff by benno@ with input from sthen@ and jmc@
ok benno@ danj@ sthen@ martijn@ deraadt@

Revert earlier revert.
It turned out the issue was a badly applied diff on stsp@'s machine.

OK stsp@

Backout recent cp(1) changes; they broke texlive's mktexlsr(1)

Rename dne in copy_file to exists to be more consistent with the other
copy_* functions.

OK stsp@

Also verify a overwrite for the copy of a fifo, link and device node.

OK stsp@

The combination of -v and -i and the deny of a copy would cause the copy
still to be printed. This fixes that edge-case.

OK stsp@

fix a few lines of unusual length

add a -v verbose flag to cp, mv, and rm. useful for monitoring progress,
and present on several other systems. some ok, some less ok.
from Job Snijders

delete useless setlocale(3);
noticed by Jan Stary <hans at stare dot cz>;
return from main while here;
feedback and OK tb@, OK martijn@

Make cp -i behave as mv -i or rm -i, independently of whether stdin
is a tty or not. From Timo Buhrmester.

ok jca@

Change all tame callers to namechange to pledge(2).

Try again. Both -R and -p prevent use of tame, but other cases can use it.

ah, fchflags. We will come back to this issue later

We cannot tame if -R is specified, because mknod and mkfifo may be called.
But in other cases, we can use tame "stdio rpath wpath cpath fattr", including
for the historical -r option.

Whoops; missed updating a setfile() call.

problem reported dcoppa; ok millert@

Replace <sys/param.h> with <limits.h> and other less dirty headers where
possible. Annotate <sys/param.h> lines with their current reasons. Switch
to PATH_MAX, NGROUPS_MAX, HOST_NAME_MAX+1, LOGIN_NAME_MAX, etc. Change
MIN() and MAX() to local definitions of MINIMUM() and MAXIMUM() where
sensible to avoid pulling in the pollution. These are the files confirmed
through binary verification.
ok guenther, millert, doug (helped with the verification protocol)

Use errc/warnc to simplify code.

ok jca@ krw@

Don't order file and dir creation, no measurable effect, ffs
allocation strategy has changed since a long time and added to that
cp(1) should not try to be smart and guess things it shouldn't know.
ok guenther@ krw@

some small nonfunctional tweaks. ok beck millert

STDIN_FILENO is better than fileno(stdin); zinovik@cs.karelia.ru

various small lint cleanups; ok otto

Call fts_close(); from NetBSD's Coverity analysis; ok jaredy@ millert@

Handle dir not readable case. Problem noted by chefren. ok hshoexer@ jaredy@

Only warn if we try to cp(1) a socket. Consequently, this makes
mv(1)ing trees containing sockets across filesystems much more
smooth. ok millert@ tedu@ fgsch@ pedro@ deraadt@

Fix typo in comment; Bruno Rohee

spacing

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

Undo a misguided attempt to fix speling errors from long ago. UCB
copyrights say "acknowledgement", not "acknowledgment". You cannot
just go changing the text of a license willy-nilly.

Zap unused STRIP_TRAILING_SLASH macro.

Ok millert@.

Use the return value of strcpy() to detect ENAMETOOLONG instead of
doing an extra strlen; deraadt@ OK

lots of sprintf -> snprintf and strcpy -> strlcpy; checked by tedu

Set directory mode/user/times on the post-order pass, not pre-order.
Fixes a problem with "cp -R" (and probably cp -p) with directories
that do not grant the owner write permission. However, we need to
know whether or not we created the directory ourselves so stash
that fact in fts_pointer (treat it as a boolean) but add a macro
to access it for readability. This also allows me to remove the
duplicated file times setting code for "cp -p".

Closes PR 3058; deraadt@ OK

ansi

knf

Part one of userland __P removal. Done with a simple regexp with some minor hand editing to make comments line up correctly. Another pass is forthcoming that handles the cases that could not be done automatically.

explicit braces to avoid ambiguous else

Avoid printing warnings about things like "is a directory" twice.
Found by fgs.

In -p mode, set directory times in post-order pass since otherwise the
times get squashed when we add files to the dir. Problem found by
Christophe.Deroulers@ens.fr, fix by me.

setlocale and get rid of unneeded exit() call. millert@ ok

"cp -f" should unlink the destination before copying. Similar to
a patch from FreeBSD (but this is simpler). Closes PR #821

Fix handling of trailing slashes. Don't ever strip them, because that's
the way the user tells us that he really wants a directory. Also, a
trailing slash does not mean that the last component is null (in fact,
according to POSIX, a component can't be 0 bytes). Rather, the last
component is the one directly preceding the trailing slashes.

spelling

Don't strip trailing '/' in source, fts(3) now does this for us.

i am bored enough to fix terminal space/tab uglies

Strip trailing slashes from source (we already did dest).
Also, use MAXPATHLEN, not MAXPATHLEN+1

tabify

getopt(3) returns -1 when out of args, not EOF, whee!

-Wall'ing.

Don't strip off the last slash in the pathname "/". cp used to
stat the pathname "" in order to decide that the pathname "/" is
a directory. This caused `cp kernel /' to fail if the kernel has
the POSIX behaviour of not allowing the pathname "" to be an alias
for ".". It presumably also caused `cp /etc/motd /' to fail in
the unlikely event that "." is not stat'able.

Be more careful about concatenating pathnames: don't check that
the pathname fits until prefixes have been discarded (the check
was too strict). Print the final pathname in error messages.
Terminate the target directory name properly for error messages.
Don't add a slash between components if there is already a slash.

Fix from FreeBSD

zap getopt() case of -?, come on, it is the default!

update rcsid

branches: 1.1.1;
Initial revision

Implement cp -a ("archive" mode)

Not standard, but supported at least by GNU cp, IllumOS, NetBSD,
FreeBSD, DragonflyBSD. No need to be gratuitously different. This
should also allow us to drop patches in ~10 ports.

Based on an initial diff by benno@ with input from sthen@ and jmc@
ok benno@ danj@ sthen@ martijn@ deraadt@

Revert earlier revert.
It turned out the issue was a badly applied diff on stsp@'s machine.

OK stsp@

Backout recent cp(1) changes; they broke texlive's mktexlsr(1)

Rename dne in copy_file to exists to be more consistent with the other
copy_* functions.

OK stsp@

Also verify a overwrite for the copy of a fifo, link and device node.

OK stsp@

The combination of -v and -i and the deny of a copy would cause the copy
still to be printed. This fixes that edge-case.

OK stsp@

fix a few lines of unusual length

add a -v verbose flag to cp, mv, and rm. useful for monitoring progress,
and present on several other systems. some ok, some less ok.
from Job Snijders

delete useless setlocale(3);
noticed by Jan Stary <hans at stare dot cz>;
return from main while here;
feedback and OK tb@, OK martijn@

Make cp -i behave as mv -i or rm -i, independently of whether stdin
is a tty or not. From Timo Buhrmester.

ok jca@

Change all tame callers to namechange to pledge(2).

Try again. Both -R and -p prevent use of tame, but other cases can use it.

ah, fchflags. We will come back to this issue later

We cannot tame if -R is specified, because mknod and mkfifo may be called.
But in other cases, we can use tame "stdio rpath wpath cpath fattr", including
for the historical -r option.

Whoops; missed updating a setfile() call.

problem reported dcoppa; ok millert@

Replace <sys/param.h> with <limits.h> and other less dirty headers where
possible. Annotate <sys/param.h> lines with their current reasons. Switch
to PATH_MAX, NGROUPS_MAX, HOST_NAME_MAX+1, LOGIN_NAME_MAX, etc. Change
MIN() and MAX() to local definitions of MINIMUM() and MAXIMUM() where
sensible to avoid pulling in the pollution. These are the files confirmed
through binary verification.
ok guenther, millert, doug (helped with the verification protocol)

Use errc/warnc to simplify code.

ok jca@ krw@

Don't order file and dir creation, no measurable effect, ffs
allocation strategy has changed since a long time and added to that
cp(1) should not try to be smart and guess things it shouldn't know.
ok guenther@ krw@

some small nonfunctional tweaks. ok beck millert

STDIN_FILENO is better than fileno(stdin); zinovik@cs.karelia.ru

various small lint cleanups; ok otto

Call fts_close(); from NetBSD's Coverity analysis; ok jaredy@ millert@

Handle dir not readable case. Problem noted by chefren. ok hshoexer@ jaredy@

Only warn if we try to cp(1) a socket. Consequently, this makes
mv(1)ing trees containing sockets across filesystems much more
smooth. ok millert@ tedu@ fgsch@ pedro@ deraadt@

Fix typo in comment; Bruno Rohee

spacing

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

Undo a misguided attempt to fix speling errors from long ago. UCB
copyrights say "acknowledgement", not "acknowledgment". You cannot
just go changing the text of a license willy-nilly.

Zap unused STRIP_TRAILING_SLASH macro.

Ok millert@.

Use the return value of strcpy() to detect ENAMETOOLONG instead of
doing an extra strlen; deraadt@ OK

lots of sprintf -> snprintf and strcpy -> strlcpy; checked by tedu

Set directory mode/user/times on the post-order pass, not pre-order.
Fixes a problem with "cp -R" (and probably cp -p) with directories
that do not grant the owner write permission. However, we need to
know whether or not we created the directory ourselves so stash
that fact in fts_pointer (treat it as a boolean) but add a macro
to access it for readability. This also allows me to remove the
duplicated file times setting code for "cp -p".

Closes PR 3058; deraadt@ OK

ansi

knf

Part one of userland __P removal. Done with a simple regexp with some minor hand editing to make comments line up correctly. Another pass is forthcoming that handles the cases that could not be done automatically.

explicit braces to avoid ambiguous else

Avoid printing warnings about things like "is a directory" twice.
Found by fgs.

In -p mode, set directory times in post-order pass since otherwise the
times get squashed when we add files to the dir. Problem found by
Christophe.Deroulers@ens.fr, fix by me.

setlocale and get rid of unneeded exit() call. millert@ ok

"cp -f" should unlink the destination before copying. Similar to
a patch from FreeBSD (but this is simpler). Closes PR #821

Fix handling of trailing slashes. Don't ever strip them, because that's
the way the user tells us that he really wants a directory. Also, a
trailing slash does not mean that the last component is null (in fact,
according to POSIX, a component can't be 0 bytes). Rather, the last
component is the one directly preceding the trailing slashes.

spelling

Don't strip trailing '/' in source, fts(3) now does this for us.

i am bored enough to fix terminal space/tab uglies

Strip trailing slashes from source (we already did dest).
Also, use MAXPATHLEN, not MAXPATHLEN+1

tabify

getopt(3) returns -1 when out of args, not EOF, whee!

-Wall'ing.

Don't strip off the last slash in the pathname "/". cp used to
stat the pathname "" in order to decide that the pathname "/" is
a directory. This caused `cp kernel /' to fail if the kernel has
the POSIX behaviour of not allowing the pathname "" to be an alias
for ".". It presumably also caused `cp /etc/motd /' to fail in
the unlikely event that "." is not stat'able.

Be more careful about concatenating pathnames: don't check that
the pathname fits until prefixes have been discarded (the check
was too strict). Print the final pathname in error messages.
Terminate the target directory name properly for error messages.
Don't add a slash between components if there is already a slash.

Fix from FreeBSD

zap getopt() case of -?, come on, it is the default!

update rcsid

branches: 1.1.1;
Initial revision

Revert earlier revert.
It turned out the issue was a badly applied diff on stsp@'s machine.

OK stsp@

Backout recent cp(1) changes; they broke texlive's mktexlsr(1)

Rename dne in copy_file to exists to be more consistent with the other
copy_* functions.

OK stsp@

Also verify a overwrite for the copy of a fifo, link and device node.

OK stsp@

The combination of -v and -i and the deny of a copy would cause the copy
still to be printed. This fixes that edge-case.

OK stsp@

fix a few lines of unusual length

add a -v verbose flag to cp, mv, and rm. useful for monitoring progress,
and present on several other systems. some ok, some less ok.
from Job Snijders

delete useless setlocale(3);
noticed by Jan Stary <hans at stare dot cz>;
return from main while here;
feedback and OK tb@, OK martijn@

Make cp -i behave as mv -i or rm -i, independently of whether stdin
is a tty or not. From Timo Buhrmester.

ok jca@

Change all tame callers to namechange to pledge(2).

Try again. Both -R and -p prevent use of tame, but other cases can use it.

ah, fchflags. We will come back to this issue later

We cannot tame if -R is specified, because mknod and mkfifo may be called.
But in other cases, we can use tame "stdio rpath wpath cpath fattr", including
for the historical -r option.

Whoops; missed updating a setfile() call.

problem reported dcoppa; ok millert@

Replace <sys/param.h> with <limits.h> and other less dirty headers where
possible. Annotate <sys/param.h> lines with their current reasons. Switch
to PATH_MAX, NGROUPS_MAX, HOST_NAME_MAX+1, LOGIN_NAME_MAX, etc. Change
MIN() and MAX() to local definitions of MINIMUM() and MAXIMUM() where
sensible to avoid pulling in the pollution. These are the files confirmed
through binary verification.
ok guenther, millert, doug (helped with the verification protocol)

Use errc/warnc to simplify code.

ok jca@ krw@

Don't order file and dir creation, no measurable effect, ffs
allocation strategy has changed since a long time and added to that
cp(1) should not try to be smart and guess things it shouldn't know.
ok guenther@ krw@

some small nonfunctional tweaks. ok beck millert

STDIN_FILENO is better than fileno(stdin); zinovik@cs.karelia.ru

various small lint cleanups; ok otto

Call fts_close(); from NetBSD's Coverity analysis; ok jaredy@ millert@

Handle dir not readable case. Problem noted by chefren. ok hshoexer@ jaredy@

Only warn if we try to cp(1) a socket. Consequently, this makes
mv(1)ing trees containing sockets across filesystems much more
smooth. ok millert@ tedu@ fgsch@ pedro@ deraadt@

Fix typo in comment; Bruno Rohee

spacing

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

Undo a misguided attempt to fix speling errors from long ago. UCB
copyrights say "acknowledgement", not "acknowledgment". You cannot
just go changing the text of a license willy-nilly.

Zap unused STRIP_TRAILING_SLASH macro.

Ok millert@.

Use the return value of strcpy() to detect ENAMETOOLONG instead of
doing an extra strlen; deraadt@ OK

lots of sprintf -> snprintf and strcpy -> strlcpy; checked by tedu

Set directory mode/user/times on the post-order pass, not pre-order.
Fixes a problem with "cp -R" (and probably cp -p) with directories
that do not grant the owner write permission. However, we need to
know whether or not we created the directory ourselves so stash
that fact in fts_pointer (treat it as a boolean) but add a macro
to access it for readability. This also allows me to remove the
duplicated file times setting code for "cp -p".

Closes PR 3058; deraadt@ OK

ansi

knf

Part one of userland __P removal. Done with a simple regexp with some minor hand editing to make comments line up correctly. Another pass is forthcoming that handles the cases that could not be done automatically.

explicit braces to avoid ambiguous else

Avoid printing warnings about things like "is a directory" twice.
Found by fgs.

In -p mode, set directory times in post-order pass since otherwise the
times get squashed when we add files to the dir. Problem found by
Christophe.Deroulers@ens.fr, fix by me.

setlocale and get rid of unneeded exit() call. millert@ ok

"cp -f" should unlink the destination before copying. Similar to
a patch from FreeBSD (but this is simpler). Closes PR #821

Fix handling of trailing slashes. Don't ever strip them, because that's
the way the user tells us that he really wants a directory. Also, a
trailing slash does not mean that the last component is null (in fact,
according to POSIX, a component can't be 0 bytes). Rather, the last
component is the one directly preceding the trailing slashes.

spelling

Don't strip trailing '/' in source, fts(3) now does this for us.

i am bored enough to fix terminal space/tab uglies

Strip trailing slashes from source (we already did dest).
Also, use MAXPATHLEN, not MAXPATHLEN+1

tabify

getopt(3) returns -1 when out of args, not EOF, whee!

-Wall'ing.

Don't strip off the last slash in the pathname "/". cp used to
stat the pathname "" in order to decide that the pathname "/" is
a directory. This caused `cp kernel /' to fail if the kernel has
the POSIX behaviour of not allowing the pathname "" to be an alias
for ".". It presumably also caused `cp /etc/motd /' to fail in
the unlikely event that "." is not stat'able.

Be more careful about concatenating pathnames: don't check that
the pathname fits until prefixes have been discarded (the check
was too strict). Print the final pathname in error messages.
Terminate the target directory name properly for error messages.
Don't add a slash between components if there is already a slash.

Fix from FreeBSD

zap getopt() case of -?, come on, it is the default!

update rcsid

branches: 1.1.1;
Initial revision

fix a few lines of unusual length

add a -v verbose flag to cp, mv, and rm. useful for monitoring progress,
and present on several other systems. some ok, some less ok.
from Job Snijders

delete useless setlocale(3);
noticed by Jan Stary <hans at stare dot cz>;
return from main while here;
feedback and OK tb@, OK martijn@

Make cp -i behave as mv -i or rm -i, independently of whether stdin
is a tty or not. From Timo Buhrmester.

ok jca@

Change all tame callers to namechange to pledge(2).

Try again. Both -R and -p prevent use of tame, but other cases can use it.

ah, fchflags. We will come back to this issue later

We cannot tame if -R is specified, because mknod and mkfifo may be called.
But in other cases, we can use tame "stdio rpath wpath cpath fattr", including
for the historical -r option.

Whoops; missed updating a setfile() call.

problem reported dcoppa; ok millert@

Replace <sys/param.h> with <limits.h> and other less dirty headers where
possible. Annotate <sys/param.h> lines with their current reasons. Switch
to PATH_MAX, NGROUPS_MAX, HOST_NAME_MAX+1, LOGIN_NAME_MAX, etc. Change
MIN() and MAX() to local definitions of MINIMUM() and MAXIMUM() where
sensible to avoid pulling in the pollution. These are the files confirmed
through binary verification.
ok guenther, millert, doug (helped with the verification protocol)

Use errc/warnc to simplify code.

ok jca@ krw@

Don't order file and dir creation, no measurable effect, ffs
allocation strategy has changed since a long time and added to that
cp(1) should not try to be smart and guess things it shouldn't know.
ok guenther@ krw@

some small nonfunctional tweaks. ok beck millert

STDIN_FILENO is better than fileno(stdin); zinovik@cs.karelia.ru

various small lint cleanups; ok otto

Call fts_close(); from NetBSD's Coverity analysis; ok jaredy@ millert@

Handle dir not readable case. Problem noted by chefren. ok hshoexer@ jaredy@

Only warn if we try to cp(1) a socket. Consequently, this makes
mv(1)ing trees containing sockets across filesystems much more
smooth. ok millert@ tedu@ fgsch@ pedro@ deraadt@

Fix typo in comment; Bruno Rohee

spacing

Remove the advertising clause in the UCB license which Berkeley
rescinded 22 July 1999. Proofed by myself and Theo.

Undo a misguided attempt to fix speling errors from long ago. UCB
copyrights say "acknowledgement", not "acknowledgment". You cannot
just go changing the text of a license willy-nilly.

Zap unused STRIP_TRAILING_SLASH macro.

Ok millert@.

Use the return value of strcpy() to detect ENAMETOOLONG instead of
doing an extra strlen; deraadt@ OK

lots of sprintf -> snprintf and strcpy -> strlcpy; checked by tedu

Set directory mode/user/times on the post-order pass, not pre-order.
Fixes a problem with "cp -R" (and probably cp -p) with directories
that do not grant the owner write permission. However, we need to
know whether or not we created the directory ourselves so stash
that fact in fts_pointer (treat it as a boolean) but add a macro
to access it for readability. This also allows me to remove the
duplicated file times setting code for "cp -p".

Closes PR 3058; deraadt@ OK

ansi

knf

Part one of userland __P removal. Done with a simple regexp with some minor hand editing to make comments line up correctly. Another pass is forthcoming that handles the cases that could not be done automatically.

explicit braces to avoid ambiguous else

Avoid printing warnings about things like "is a directory" twice.
Found by fgs.

In -p mode, set directory times in post-order pass since otherwise the
times get squashed when we add files to the dir. Problem found by
Christophe.Deroulers@ens.fr, fix by me.

setlocale and get rid of unneeded exit() call. millert@ ok

"cp -f" should unlink the destination before copying. Similar to
a patch from FreeBSD (but this is simpler). Closes PR #821

Fix handling of trailing slashes. Don't ever strip them, because that's
the way the user tells us that he really wants a directory. Also, a
trailing slash does not mean that the last component is null (in fact,
according to POSIX, a component can't be 0 bytes). Rather, the last
component is the one directly preceding the trailing slashes.

spelling

Don't strip trailing '/' in source, fts(3) now does this for us.

i am bored enough to fix terminal space/tab uglies

Strip trailing slashes from source (we already did dest).
Also, use MAXPATHLEN, not MAXPATHLEN+1

tabify

getopt(3) returns -1 when out of args, not EOF, whee!

-Wall'ing.

Don't strip off the last slash in the pathname "/". cp used to
stat the pathname "" in order to decide that the pathname "/" is
a directory. This caused `cp kernel /' to fail if the kernel has
the POSIX behaviour of not allowing the pathname "" to be an alias
for ".". It presumably also caused `cp /etc/motd /' to fail in
the unlikely event that "." is not stat'able.

Be more careful about concatenating pathnames: don't check that
the pathname fits until prefixes have been discarded (the check
was too strict). Print the final pathname in error messages.
Terminate the target directory name properly for error messages.
Don't add a slash between components if there is already a slash.

Fix from FreeBSD

zap getopt() case of -?, come on, it is the default!

update rcsid

branches: 1.1.1;
Initial revision