Commit Graph

65 Commits

Author SHA1 Message Date
John Birrell
3c526fbfd3 CSU source for Alpha obtained from NetBSD. The makefile will require
more work when we get a half-way usable libc (which is next).
1998-01-11 03:30:39 +00:00
Brian Somers
9822c98d98 const correctness for dl*() 1997-11-22 03:34:46 +00:00
Satoshi Asami
fc702c5bc8 Make this file p-make clean. (Use "ld -O foo" instead of "ld; mv
a.out foo".)

Reviewed by:	bde (actually more like "Suggested by")
1997-10-11 02:37:42 +00:00
Wolfram Schneider
bf5cbf3551 Sort cross refereces in section SEE ALSO. 1997-09-29 19:11:55 +00:00
John Polstra
7e7344e2f4 Implement dlsym(RTLD_NEXT, symbol). 1997-08-02 04:56:44 +00:00
Mike Smith
dff682856f Use our copy of dlfcn.h, not the version in /usr/include, which may not
be up-to-date when we are building.
Submitted by:	Terry Lambert <terry@lambert.org>
1997-04-30 03:12:09 +00:00
John Polstra
f3112eb32a Declare the constructor/destructor linker sets as extern rather than
common.  Add one do-nothing element to each set.  This ensures that
the linker realizes that they are linker sets rather than simple
commons, and makes it possible to link c++rt0.o into every shared
library regardless of whether it is a C++ library or not.  Without
this change, the constructors and destructors in the main program
could be executed multiple times.

This change is going to make it possible to get rid of the
CPLUSPLUSLIB makefile variable once and for all.  It is a piece of
the solution to PR gnu/3505 (gcc -shared).  Finally, it fixes a
heretofore unreported bug:  If CPLUSPLUSLIB was set in a makefile
for a C++ shared library that had no static constructors or
destructors in it, then the main program's constructors and
destructors would be executed multiple times.
1997-04-09 19:14:31 +00:00
John Polstra
bb789b9ea2 Fix an error in the previous revision that caused make world breakage. 1997-04-02 16:49:18 +00:00
Bruce Evans
67fd3b9895 Simplified install rule. 1997-04-01 13:53:16 +00:00
Peter Wemm
7e546392b5 Revert $FreeBSD$ to $Id$ 1997-02-22 15:12:41 +00:00
Jordan K. Hubbard
1130b656e5 Make the long-awaited change from $Id$ to $FreeBSD$
This will make a number of things easier in the future, as well as (finally!)
avoiding the Id-smashing problem which has plagued developers for so long.

Boy, I'm glad we're not using sup anymore.  This update would have been
insane otherwise.
1997-01-14 07:20:47 +00:00
John Polstra
f08cd1247b Convert to mdoc format.
Add descriptions of RTLD_LAZY and RTLD_NOW.
Correct the synopsis to agree with the actual function prototypes.
Add clarifications of a few things.
Clean up the wording in a few places.
1997-01-12 19:55:49 +00:00
John Polstra
8a8c5a99ee Add definitions of RTLD_LAZY and RTLD_NOW. 1997-01-12 18:37:46 +00:00
John Polstra
65eabc3424 Set the "crt_ldso" member of the crt-to-ld.so interface structure. This
was apparently overlooked at the time the member was added.  Its absence
causes some error messages from the dynamic linker to begin with
"(null):" instead of with the pathname of the dynamic linker as they
should.

I am also adding a work-around to the dynamic linker, to cope with
legacy binaries that were built with older versions of crt0.
1997-01-11 23:59:34 +00:00
Bruce Evans
fd0a86436e Use ${COPY} instead of -C for installing non-source files. crt*.o
should be installed using the same flag as libraries, but ${COPY}
is currently used for libraries.
1997-01-01 04:52:55 +00:00
Peter Wemm
fbde87863d When linking with no rtld support, provide stub dl*() functions that
just return errors.  This removes the need for awful hacks like that in
our build of libtcl which would get link errors when linked static.

John Polstra once mentioned that this was on his "todo" list.

Note that one can use:
  cc -Wl,-Bstatic -o foo foo.o
and get an executable that has it's libraries statically linked, but has
a fully functional runtime linker so the executable can call dlopen() and
have it work.  (I've tested this)
1996-12-28 17:10:52 +00:00
Marc G. Fournier
d453b6a53d Fixed prototyping of dlopen/dlsym in dlfcn.h, to match how crt0.c defines
it and link.h prototypes it

Error of my ways pointed out by Peter
1996-10-08 01:41:51 +00:00
Steve Price
3d78e32756 Remove garbage initcode reference so that 'gcc -Dlint ...'
will compile without error.
1996-10-06 03:19:26 +00:00
Peter Wemm
4e89f352f0 Support crt0 <-> ld.so interface version 4. This should be both backwards
and forwards compatable with version 3.  This is needed to enable storing
a run-time library path in the dynamic linking headers.  The crt startup
tries version 4 first, and falls back to version 3, so an executable that
is linked on -current will work with the ld.so on 2.1.x and less.

Reviewed by: nate, jdp
Obtained from: NetBSD
1996-10-01 00:54:48 +00:00
Peter Wemm
dce96329a9 cmp -s || install -c --> install -C 1996-08-30 01:41:52 +00:00
Wolfram Schneider
9fb933075e `mv'' -> `mv -f''
``rm'' -> ``rm -f''
so mv/rm may not ask for confirmation if you are not root
1996-05-07 23:19:49 +00:00
John Polstra
89370a31f2 Changed the dimensions of __CTOR_LIST__ and __DTOR_LIST__ from 0
to 2.  This makes them agree with the declarations in libgcc, and
clears the way once again for linking c++rt0.o into all libraries,
and eliminating CPLUSPLUSLIB from <bsd.lib.mk>.  (I have not made
that change yet, because there is still a bootstrapping problem
for "make world".)

Also, removed a check which ensured that the constructor count in
the first word of __CTOR_LIST__ was greater than zero before
traversing the list.  I had added that check earlier, but it is no
longer necessary, now that there is guaranteed to be at least 2
words in __CTOR_LIST__.
1996-02-20 04:07:26 +00:00
Nate Williams
98c4ae7b2b Back out the thread_init code in order to allow -current to bootstrap
from -stable, until a better solution is found.

Submitted by:	Consensus of mailing list and the almighty Jordan :)
1996-01-30 05:55:20 +00:00
Julian Elischer
f70177e76e Reviewed by: julian and (hsu?)
Submitted by:	 John Birrel(L?)

changes for threadsafe operations
1996-01-22 00:02:33 +00:00
John Polstra
d5489b0d8f Check the count in the first word of __CTOR_LIST__ before executing
the loop that invokes the static constructors.  That makes it safe
to link c++rt0.o into any shared library, even one that does not
have any static constructors.  Formerly, doing that would cause a
bus error.  If the library has no static constructors, __CTOR_LIST__
comes out as a simple 4-byte COMMON region, and it does not have
the usual NULL word that terminates the list of constructors.  This
caused the old code to "call" a garbage address via the non-existent
entry __CTOR_LIST__[1].

The analogous code that invokes the static destructors was already safe.

This change is fully backward-compatible.

Reviewed by:	dfr@render.com (Doug Rabson)
1996-01-15 17:53:25 +00:00
Peter Wemm
6065a0be11 This commit was generated by cvs2svn to compensate for changes in r13122,
which included commits to RCS files with non-trunk default branches.
1995-12-30 19:02:48 +00:00
Peter Wemm
a5b996a7ec recording cvs-1.6 file death 1995-12-30 19:02:48 +00:00
Andrey A. Chernov
3e912c9dee Remove my locale hack. Sigh. 1995-11-02 12:42:42 +00:00
Poul-Henning Kamp
05fc2de514 Clean up and make code (more) readable. 1995-10-29 09:49:21 +00:00
Bruce Evans
ccbc94648a Fixed dependencies for scrt0.o.
Build a static gcrt0.o (sgcrt0.o) too.  Currently only the dynamic gcrt0.o
is used, although -pg forces -static.
Sorted the .o targets.
1995-10-22 18:36:47 +00:00
Poul-Henning Kamp
6b44b1a3fa -fomit-frame-pointer is becomming an increasingly popular optimization,
so before somebody screws up royally, make sure this always works by
adding a -fno-omit-frame-pointer here.
1995-10-20 20:05:15 +00:00
Poul-Henning Kamp
ee61b5a715 put the _getenv and _strncmp under #ifdef DEBUG, which is the only time
they are used.  Saves a few bytes here and there, nothing major.
1995-10-18 15:56:55 +00:00
David Greenman
14c175f7a9 Create a scrt0.o file that specifically excludes the shared-lib support.
This will be used for -static programs.
1995-10-18 04:19:00 +00:00
Nate Williams
4d2131fbc9 Fixup the "ld.so failed" message for the case when ld.so finds undefined
symbols.

An easy example to see this is to develop an X program which links
against Xt, but doesn't add -lX11 to the link line.  It will link fine,
but cause run-time errors by ld.so because of missing symbols used by Xt
defined in X11.  This patch makes the errors more readable.

Submitted by:	jdp@polstra.com (John Polstra)
1995-09-27 23:13:33 +00:00
Bruce Evans
97cefc5891 Install source files with the -c flag, not with the optional flag ${COPY}. 1995-08-06 12:37:41 +00:00
Bruce Evans
48cfb668fc Change install' to ${INSTALL}' so that default install flags can be
specified in the top level Makefiles.

Previously I missed dozens of Makefiles that skip the install after
using `cmp -s' to decide that the install isn't necessary.
1995-08-06 12:24:38 +00:00
Garrett Wollman
4fa53f8ca6 Numerous Makefile fixes:
1) Do dependencies.
2) Install all appropriate links to manual pages.
3) Install header file in `beforeinstall' like all the rest.
4) Install header file only if changed.
5) Install object files only if changed.
1995-06-30 15:30:35 +00:00
Doug Rabson
82aaeb09ad Change ld.so to correctly load dependant libraries for dlopen and unload them
on dlclose.  Also correctly call constructors and destructors for libraries
linked with /usr/lib/c++rt0.o.
Change interpretation of dlopen manpage to call _init() rather than init()
for dlopened objects.
Change c++rt0.o to avoid using atexit to call destructors, allowing dlclose to
call destructors when an object is unloaded.
Change interface between crt0 and ld.so to allow crt0 to call a function on
exit to call destructors for shared libraries explicitly.

These changes are backwards compatible.  Old binaries will work with the new
ld.so and new binaries will work with the old ld.so.  A version number has
been introduced in the crt0-ld.so interface to allow for future changes.

Reviewed by:	GAWollman, Craig Struble <cstruble@singularity.bevc.blacksburg.va.us>
1995-06-27 09:53:27 +00:00
Rodney W. Grimes
6c06b4e2aa Remove trailing whitespace. 1995-05-30 05:51:47 +00:00
Poul-Henning Kamp
139762f2be Install the dlopen.3 manpage. 1995-03-30 06:40:31 +00:00
Poul-Henning Kamp
e9434bccf7 Remove an unused variable. 1995-02-24 07:51:13 +00:00
Jordan K. Hubbard
f2c81e4dd7 Install the header files. 1995-02-09 08:00:20 +00:00
Jordan K. Hubbard
2a947a4bbf Add two parts of Mark's/Gary's dlopen() changes that I missed before. 1995-02-08 17:56:27 +00:00
Jordan K. Hubbard
38f4fb1525 Support for more Sun compatible dlopen() and friends. Also added proper error
handling.
Reviewed by:    gj
Submitted by:   Mark Diekhans <markd@grizzly.com>
1995-02-07 13:27:29 +00:00
Andrey A. Chernov
662bfb6660 Call reduced (8-bit only) startup_setlocale() 1994-09-24 16:01:30 +00:00
Andrey A. Chernov
54335ac349 Change level of setlocale hack enabling from compile option STARTUP_LOCALE
to check (via getenv) environment variable "ENABLE_STARTUP_LOCALE" at
runtime.
Submitted by: me per Bruce suggestion
1994-09-19 02:00:21 +00:00
Andrey A. Chernov
f32a347542 Add (#ifdef'ed by STARTUP_LOCALE) following line to crt0.c
(void) setlocale(LC_ALL, "");

It will be easiest way now to make national chars available
for all ctype-oriented programs at once by simple:

setenv LANG Your_National_Charset

Default case (without "LANG" environment
variable) will be fully ANSI compatible (got "C" locale).

If "LANG" variable present, extention becomes active.

Effect of this extention is great: in one time all ctype
oriented programs can accept/print national characters
without any touching source/binary code, it is big win, IMHO.

This method is fully compatible with ISO8859-* and russian koi8-r
too (in general -- with all 8-bit character sets). I think
it is very useful.
I got this idea from Xenix locale implementation.

This extention is even never compiled in, unless you set
	setenv STARTUP_LOCALE
before rebuilding crt0.c or corresponding variable in /etc/make.conf
1994-09-18 22:21:04 +00:00
Bruce Evans
459e429cab Remove dead code. gmon stuff is now done better in libc/gmon an
<machine/profile.h>.  The old version was writing an incomplete
header without the profrate field that is necessary to handle the
current faster profiling clock.  The counters that are where the
the profrate should be are usually 0 and gprof converts a profrate
of 0 to hz so the old version gave times too large by a factor of
profhz/hz = 10.24.
1994-08-22 15:13:41 +00:00
Garrett Wollman
b4162996d5 Make it work with new mmap syscall. 1994-08-05 17:53:07 +00:00
Rodney W. Grimes
f46145aa78 MAP_FILE is the default on mmap now, and is no longer defined just
like on a sun, so #define it to be 0 if we are running BSD >=199306.
1994-05-28 13:59:16 +00:00