Skip to contentSkip to conversation

RFC - Debian patches

PATCH New Fabio M. Di Nitto 4 participants 57 messages Oct 8, 2009 – Nov 10, 2009

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hi guys,

I was asked to look into the patch set Debian carries around to see what we can merge upstream and what not, to try to reduce the delta between distributions and upstream.

This morning I had a very pleasent talk with Bastian Blank (Debian maintainer for lvm2) and this is the full patch queue rebased on latest CVS.

Debian at this point is at 2.02.53 from released tarball.

Bastian gave me explanations on all the patches.

This patch set is neutral (it means I am not asking inclusion of any patch right away) but only review of ideas and possible bug fixes behind the patches.

Once we settled on what patches could be included and what changes need to be done to merge them upstream, we will take appropriate steps, with repost and extra review, etc.

Thanks
Fabio

#1 lore raw

PATCH 1/11 debian: fix library search path in example.conf

this patch is debian specific and sets the library search path to /lib/lvm2 to avoid cluttering /lib

---
:100644 100644 1067665... 46c4dba... M doc/example.conf

 doc/example.conf |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
doc/example.conf +1 -1
index 1067665..46c4dba 100644
@@ -312,7 +312,7 @@ global {
     # Full pathnames can be given.
 
     # Search this directory first for shared libraries.
-    #   library_dir = "/lib"
+    #   library_dir = "/lib/lvm2"
 
     # The external locking library to load if locking_type is set to 2.
     #   locking_library = "liblvm2clusterlock.so"
Signature · 1 line
1.5.4.3
#2 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 2 lines · Fabio M. Di Nitto

this patch is debian specific and sets the library search path to /lib/lvm2 to avoid cluttering /lib

Add configure parm for this and generate example.conf from template?

In fact, it would be useful to know the current 'configure' line that Debian uses.

Alasdair

#3 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 8 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

this patch is debian specific and sets the library search path to /lib/lvm2 to avoid cluttering /lib

Add configure parm for this and generate example.conf from template?

In fact, it would be useful to know the current 'configure' line that Debian uses.

I think this patch is bogus even in Debian.

After some investigation "/lib/lvm2/" used to be the default for Debian up to version 2.00.32-1 and then it was reverted to upstream default to fix bug http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=287042

Other than the example config file there are no traces of /lib/lvm2 anywhere in the package.

I attached debian/rules anyway as it could be helpful for other patches.

Fabio
-------------- next part --------------
A non-text attachment was scrubbed...
Name: rules
Type: text/x-makefile
Size: 9017 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/lvm-devel/attachments/20091011/542b5fd9/attachment.bin>

#4 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 2 lines · Fabio M. Di Nitto

Other than the example config file there are no traces of /lib/lvm2 anywhere in the package.

Its currently unused. The cluster locking lib was the only used external lib.

Bastian

Signature · 2 lines
The sight of death frightens them [Earthers].
		-- Kras the Klingon, "Friday's Child", stardate 3497.2
#5 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 6 lines · Bastian Blank
Quoted 2 lines · Fabio M. Di Nitto

Other than the example config file there are no traces of /lib/lvm2 anywhere in the package.

Its currently unused. The cluster locking lib was the only used external lib.

Alasdair, given that cluster locking lib is gone, do we have a use case to keep this option configurable? If so I'll spend sometime to make it a build option, otherwise I would suggest to simply drop it from the example.conf.

Fabio

#6 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 6 lines · Fabio M. Di Nitto
Quoted 2 lines · Bastian Blank

Its currently unused. The cluster locking lib was the only used external lib.

Alasdair, given that cluster locking lib is gone, do we have a use case to keep this option configurable?

It's used for all lvm2 external libraries and needs to stay. The commented-out example is only worth making configurable if people are actually installing libraries into private directories (and arguably it should then be hard-coded into the source too).

We only stopped building the cluster locking library externally on Red Hat distributions because of some complexities of the Red Hat build system. If they get resolved (as I hope they will be with RHEL6) then we will probably reinstate it.

Alasdair

#7 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 13 lines · Alasdair G Kergon
Quoted 6 lines · Fabio M. Di Nitto
Quoted 2 lines · Bastian Blank

Its currently unused. The cluster locking lib was the only used external lib.

Alasdair, given that cluster locking lib is gone, do we have a use case to keep this option configurable?

It's used for all lvm2 external libraries and needs to stay. The commented-out example is only worth making configurable if people are actually installing libraries into private directories (and arguably it should then be hard-coded into the source too).

Quoted 5 lines

We only stopped building the cluster locking library externally on Red Hat distributions because of some complexities of the Red Hat build system. If they get resolved (as I hope they will be with RHEL6) then we will probably reinstate it.

The patch in attachment:
- allows lvm2libdir to be configurable
- dynamically update docs and man pages
- changes get_shared_library_path to understand LVM2_LIB_DIR
- make get_shared_library_path returns error (void to int)
- update get_shared_library_path users

Fabio

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: lvm2libdir.diff
URL: <http://listman.redhat.com/archives/lvm-devel/attachments/20091109/2204a72c/attachment.ksh>

#8 lore raw

PATCH 1/11 debian: fix library search path in example.conf

Quoted 19 lines · Fabio M. Di Nitto
Quoted 12 lines · Alasdair G Kergon
Quoted 6 lines · Fabio M. Di Nitto
Quoted 2 lines · Bastian Blank

Its currently unused. The cluster locking lib was the only used external lib.

Alasdair, given that cluster locking lib is gone, do we have a use case to keep this option configurable?

It's used for all lvm2 external libraries and needs to stay. The commented-out example is only worth making configurable if people are actually installing libraries into private directories (and arguably it should then be hard-coded into the source too).

Quoted 5 lines

We only stopped building the cluster locking library externally on Red Hat distributions because of some complexities of the Red Hat build system. If they get resolved (as I hope they will be with RHEL6) then we will probably reinstate it.

Quoted 10 lines

The patch in attachment:
- allows lvm2libdir to be configurable
- dynamically update docs and man pages
- changes get_shared_library_path to understand LVM2_LIB_DIR
- make get_shared_library_path returns error (void to int)
- update get_shared_library_path users

Fabio

I just realized that the patch doesn?t show properly the rename of doc/example.conf to doc/example.conf.in. Just a minor detail.

Fabio

#9 lore raw

RFC - Debian patches (resend - one more time)

Sorry for the repost.. there has been some technical issues with SMTP server

#10 lore raw

PATCH 2/11 debian: drop realtime

the realtime support is completely unused and just includes another library dependency

---
:100644 100644 b4e7f28... 45dc2c9... M configure.in
:100644 100644 2bd6fba... d4e24d4... M lib/Makefile.in
:100644 000000 62e1be8... 0000000... D lib/misc/timestamp.c
:100644 000000 50e2a85... 0000000... D lib/misc/timestamp.h

 configure.in         |   23 ---------
 lib/Makefile.in      |    1 -
 lib/misc/timestamp.c |  130 --------------------------------------------------
 lib/misc/timestamp.h |   33 -------------
 4 files changed, 0 insertions(+), 187 deletions(-)
configure.in +0 -23
index b4e7f28..45dc2c9 100644
@@ -40,7 +40,6 @@ case "$host_os" in
 		ODIRECT=yes
 		DM_IOCTLS=yes
 		SELINUX=yes
-		REALTIME=yes
 		CLUSTER=internal
 		FSADM=yes
 		;;
@@ -55,7 +54,6 @@ case "$host_os" in
 		ODIRECT=no
 		DM_IOCTLS=no
 		SELINUX=no
-		REALTIME=no
 		CLUSTER=none
 		FSADM=no
 		;;
@@ -316,13 +314,6 @@ AC_ARG_ENABLE([readline],
 AC_MSG_RESULT($READLINE)
 
 ################################################################################
-dnl -- Disable realtime clock support
-AC_MSG_CHECKING(whether to enable realtime support)
-AC_ARG_ENABLE(realtime, [  --disable-realtime      Disable realtime clock support],
-REALTIME=$enableval)
-AC_MSG_RESULT($REALTIME)
-
-################################################################################
 dnl -- Init pkg-config with dummy invokation:
 dnl -- this is required because PKG_CHECK_MODULES macro is expanded
 dnl -- to initialize the pkg-config environment only at the first invokation,
@@ -868,19 +859,6 @@ if test x$SELINUX = xyes; then
 fi
 
 ################################################################################
-dnl -- Check for realtime clock support
-if test x$REALTIME = xyes; then
-	AC_CHECK_LIB(rt, clock_gettime, HAVE_REALTIME=yes, HAVE_REALTIME=no)
-
-	if test x$HAVE_REALTIME = xyes; then
-		AC_DEFINE([HAVE_REALTIME], 1, [Define to 1 to include support for realtime clock.])
-		LIBS="-lrt $LIBS"
-	else
-		AC_MSG_WARN(Disabling realtime clock)
-	fi
-fi
-
-################################################################################
 dnl -- Check for getopt
 AC_CHECK_HEADERS(getopt.h, AC_DEFINE([HAVE_GETOPTLONG], 1, [Define to 1 if getopt_long is available.]))
 
@@ -1098,7 +1076,6 @@ AC_SUBST(GROUP)
 AC_SUBST(GULM_CFLAGS)
 AC_SUBST(GULM_LIBS)
 AC_SUBST(HAVE_LIBDL)
-AC_SUBST(HAVE_REALTIME)
 AC_SUBST(HAVE_SELINUX)
 AC_SUBST(INTL)
 AC_SUBST(INTL_PACKAGE)
lib/Makefile.in +0 -1
index 2bd6fba..d4e24d4 100644
@@ -82,7 +82,6 @@ SOURCES =\
 	misc/lvm-globals.c \
 	misc/lvm-string.c \
 	misc/lvm-wrappers.c \
-	misc/timestamp.c \
 	misc/util.c \
 	mm/memlock.c \
 	report/report.c \
lib/misc/timestamp.c deleted +0 -130

130 lines not on this page · raw message

lib/misc/timestamp.h deleted +0 -33
deleted file mode 100644
index 50e2a85..0000000
@@ -1,33 +0,0 @@
-/*
- * Copyright (C) 2006 Rackable Systems All rights reserved.  
- *
- * This file is part of LVM2.
- *
- * This copyrighted material is made available to anyone wishing to use,
- * modify, copy, or redistribute it subject to the terms and conditions
- * of the GNU Lesser General Public License v.2.1.
- *
- * You should have received a copy of the GNU Lesser General Public License
- * along with this program; if not, write to the Free Software Foundation,
- * Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
- */
-
-#ifndef _LVM_TIMESTAMP_H
-#define _LVM_TIMESTAMP_H
-
-struct timestamp;
-
-struct timestamp *get_timestamp(void);
-
-/* cmp_timestamp: Compare two timestamps
- * 
- * Return: -1 if t1 is less than t2
- *  	    0 if t1 is equal to t2
- *          1 if t1 is greater than t2
- */
-int cmp_timestamp(struct timestamp *t1, struct timestamp *t2);
-
-void destroy_timestamp(struct timestamp *t);
-
-#endif /* _LVM_TIMESTAMP_H */
-
Signature · 1 line
1.5.4.3
#11 lore raw

PATCH 2/11 debian: drop realtime

Quoted 2 lines · Fabio M. Di Nitto

the realtime support is completely unused and just includes another library dependency

Just fix the build so REALTIME=no really does exclude it completely.

Alasdair

#12 lore raw

PATCH 2/11 debian: drop realtime

Quoted 4 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

the realtime support is completely unused and just includes another library dependency

Just fix the build so REALTIME=no really does exclude it completely.

The code is not used. Why is it there?

Bastian

Signature · 1 line
Warp 7 -- It's a law we can live with.
#13 lore raw

PATCH 2/11 debian: drop realtime

Quoted 1 line · Bastian Blank

The code is not used. Why is it there?

This is an upstream development tree and instead of having multiple branches, we keep the code in one place but exclude or disable code that is not used. The code that needed to use that particular feature has not been finished.

Alasdair

#14 lore raw

PATCH 2/11 debian: drop realtime

Quoted 5 lines · Alasdair G Kergon
Quoted 1 line · Bastian Blank

The code is not used. Why is it there?

This is an upstream development tree and instead of having multiple branches, we keep the code in one place but exclude or disable code that is not used.

The code
* is enabled by default,
* is unused in the published code since a long time and
* is not marked in any way.
=> Can be removed.

Quoted 2 lines

The code that needed to use that particular feature has
not been finished.

Then it can be handled as part of that particular feature.

Bastian

Signature · 1 line
Well, Jim, I'm not much of an actor either.
#15 lore raw

PATCH 2/11 debian: drop realtime

Quoted 13 lines · Bastian Blank
Quoted 5 lines · Alasdair G Kergon
Quoted 1 line · Bastian Blank

The code is not used. Why is it there?

This is an upstream development tree and instead of having multiple branches, we keep the code in one place but exclude or disable code that is not used.

The code
* is enabled by default,
* is unused in the published code since a long time and
* is not marked in any way.
=> Can be removed.

Quoted 7 lines
Quoted 2 lines

The code that needed to use that particular feature has
not been finished.

Then it can be handled as part of that particular feature.

Bastian

this patch should make everybody happy.

Sets realitime support off by default and still allow to have it configurable for developers.

Fabio
-------------- next part --------------
A non-text attachment was scrubbed...
Name: realtime.diff
Type: text/x-patch
Size: 1313 bytes
Desc: not available
URL: <http://listman.redhat.com/archives/lvm-devel/attachments/20091011/6d7b145c/attachment.bin>

#16 lore raw

PATCH 2/11 debian: drop realtime

Quoted 4 lines · Fabio M. Di Nitto

this patch should make everybody happy.

Sets realitime support off by default and still allow to have it configurable for developers.

Committed to cvs with agk ACK.

Fabio

#18 lore raw

PATCH 3/11 debian: force modprobe

From: Fabio M. Di Nitto <[email protected]>

well, a linux system without modprobe can't use any modules. so this just assumes that it is available in a real system

---
:100644 100644 45dc2c9... 06759b6... M configure.in

 configure.in |    6 +-----
 1 files changed, 1 insertions(+), 5 deletions(-)
configure.in +1 -5
index 45dc2c9..06759b6 100644
@@ -974,11 +974,7 @@ if test x$UDEV_SYNC = xyes; then
 fi
 
 ################################################################################
-AC_PATH_PROG(MODPROBE_CMD, modprobe)
-
-if test x$MODPROBE_CMD != x; then
-	AC_DEFINE_UNQUOTED([MODPROBE_CMD], ["$MODPROBE_CMD"], [The path to 'modprobe', if available.])
-fi
+AC_DEFINE_UNQUOTED([MODPROBE_CMD], ["/sbin/modprobe"], [The path to 'modprobe', if available.])
 
 ################################################################################
 dnl -- dmeventd pidfile and executable path
Signature · 1 line
1.5.4.3
#19 lore raw

PATCH 3/11 debian: force modprobe

Quoted 2 lines · Fabio M. Di Nitto

well, a linux system without modprobe can't use any modules. so this just assumes that it is available in a real system

I see no need for this one upstream.

(Things like this are there so the code can be used on some BSD-based systems.)

Alasdair

#20 lore raw

PATCH 3/11 debian: force modprobe

Quoted 5 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

well, a linux system without modprobe can't use any modules. so this just assumes that it is available in a real system

I see no need for this one upstream.
(Things like this are there so the code can be used on some BSD-based systems.)

modprobe is Linux-only, so this useless.

Bastian

Signature · 2 lines
Earth -- mother of the most beautiful women in the universe.
		-- Apollo, "Who Mourns for Adonais?" stardate 3468.1
#21 lore raw

PATCH 4/11 debian: fix implicit pointer

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

not sure why the ifdef does not fire, but this is a readline function and must not be redefined

---
:100644 100644 e1f7900... 257cc40... M tools/lvm.c

 tools/lvm.c |    3 ---
 1 files changed, 0 insertions(+), 3 deletions(-)
tools/lvm.c +0 -3
index e1f7900..257cc40 100644
@@ -25,9 +25,6 @@ int main(int argc, char **argv)
 
 #  include <readline/readline.h>
 #  include <readline/history.h>
-#  ifndef HAVE_RL_COMPLETION_MATCHES
-#    define rl_completion_matches(a, b) completion_matches((char *)a, b)
-#  endif
 
 static struct cmdline_context *_cmdline;
 
Signature · 1 line
1.5.4.3
#22 lore raw

PATCH 4/11 debian: fix implicit pointer

Quoted 2 lines · Fabio M. Di Nitto

not sure why the ifdef does not fire, but this is a readline function and must not be redefined

This hack was needed for compilation on some systems (functions renamed in library).

If there are other systems where it causes problems, then please send details.

Alasdair

#23 lore raw

PATCH 4/11 debian: fix implicit pointer

Quoted 9 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

not sure why the ifdef does not fire, but this is a readline function and must not be redefined

This hack was needed for compilation on some systems (functions renamed in library).

If there are other systems where it causes problems, then please send details.

This might very well be a bug in Debian toolchain/autotools/readline because I can't trigger it on any of my systems.

Bastian, I'd like to investigate this further. Can you please me provide me with details of your test machine? Debian SID? x86? x86_64? tool chain version? etc.

Thanks
Fabio

#24 lore raw

PATCH 4/11 debian: fix implicit pointer

Quoted 3 lines · Fabio M. Di Nitto

Bastian, I'd like to investigate this further. Can you please me provide me with details of your test machine? Debian SID? x86? x86_64? tool chain version? etc.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=516867
https://buildd.debian.org/fetch.cgi?&pkg=lvm2&ver=2.02.44-1&arch=ia64&stamp=1235444991&file=log

Actually configure was not able to find rl_completion_matches but it was usable.

Bastian

Signature · 1 line
Warp 7 -- It's a law we can live with.
#25 lore raw

PATCH 4/11 debian: fix implicit pointer

Quoted 10 lines · Bastian Blank
Quoted 3 lines · Fabio M. Di Nitto

Bastian, I'd like to investigate this further. Can you please me provide me with details of your test machine? Debian SID? x86? x86_64? tool chain version? etc.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=516867
https://buildd.debian.org/fetch.cgi?&pkg=lvm2&ver=2.02.44-1&arch=ia64&stamp=1235444991&file=log

Actually configure was not able to find rl_completion_matches but it was usable.

This seems to be a bug introduced by Debian autoreconf.

Using upstream configure:

configure:13440: checking for rl_completion_matches
configure:13496: gcc -o conftest -Wl,--export-dynamic conftest.c
-lreadline -lrt -lselinux -lsepol -ldl -lncurses >&5 configure:13503: $? = 0
configure:13525: result: yes

Using Debian regenerated configure:

configure:8220: checking for rl_completion_matches
configure:8220: gcc -o conftest -g -O2 conftest.c -ldl >&5
/tmp/ccUvvocx.o: In function `main':
/root/lvm2-2.02.53/debian/build/build-deb/conftest.c:142: undefined reference to
`rl_completion_matches'
collect2: ld returned 1 exit status
configure:8220: $? = 1

Maybe it's a regression introduced by autoconf 2.64 that's not available in Fedora yet.

Fabio

#26 lore raw

PATCH 4/11 debian: fix implicit pointer

In reply to Bastian Blank
Quoted 10 lines · Bastian Blank
Quoted 3 lines · Fabio M. Di Nitto

Bastian, I'd like to investigate this further. Can you please me provide me with details of your test machine? Debian SID? x86? x86_64? tool chain version? etc.

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=516867
https://buildd.debian.org/fetch.cgi?&pkg=lvm2&ver=2.02.44-1&arch=ia64&stamp=1235444991&file=log

Actually configure was not able to find rl_completion_matches but it was usable.

Found it...

The regression comes from libs-cleanup.patch:

- AC_CHECK_LIB([readline], [readline], , [rl_found=no])
+ AC_CHECK_LIB([readline], [readline], [LIBS_READLINE="-lreadline
$LIBS_READLINE"], [rl_found=no])

by overriding default [ACTION-IF-FOUND], -lreadline is not part of LIBS= that's used later on by AC_CHECK_FUNC and the test fails.

Fabio

#27 lore raw

PATCH 5/11 debian: keep libdm-abi consistent

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

don't change the ABI of libdm on selinux enable, just make the selinux function a NOP

Also noted by Alasdair that some other feature set affects the libdm ABI.

---
:100644 100644 e412cff... 6e4f91a... M lib/activate/fs.c
:100644 100644 175d8d1... 248d3ac... M libdm/libdm-common.c

 lib/activate/fs.c    |    2 --
 libdm/libdm-common.c |    6 ++----
 2 files changed, 2 insertions(+), 6 deletions(-)
lib/activate/fs.c +0 -2
index e412cff..6e4f91a 100644
@@ -201,10 +201,8 @@ static int _mk_link(const char *dev_dir, const char *vg_name,
 		return 0;
 	}
 
-#ifdef HAVE_SELINUX
 	if (!dm_set_selinux_context(lv_path, S_IFLNK))
 		return_0;
-#endif
 
 	return 1;
 }
libdm/libdm-common.c +2 -4
index 175d8d1..248d3ac 100644
@@ -373,9 +373,9 @@ int dm_task_add_target(struct dm_task *dmt, uint64_t start, uint64_t size,
 	return 1;
 }
 
-#ifdef HAVE_SELINUX
 int dm_set_selinux_context(const char *path, mode_t mode)
 {
+#ifdef HAVE_SELINUX
 	security_context_t scontext;
 
 	if (is_selinux_enabled() <= 0)
@@ -396,9 +396,9 @@ int dm_set_selinux_context(const char *path, mode_t mode)
 	}
 
 	freecon(scontext);
+#endif
 	return 1;
 }
-#endif
 
 static int _add_dev_node(const char *dev_name, uint32_t major, uint32_t minor,
 			 uid_t uid, gid_t gid, mode_t mode)
@@ -445,10 +445,8 @@ static int _add_dev_node(const char *dev_name, uint32_t major, uint32_t minor,
 
 	log_debug("Created %s", path);
 
-#ifdef HAVE_SELINUX
 	if (!dm_set_selinux_context(path, S_IFBLK))
 		return 0;
-#endif
 
 	return 1;
 }
Signature · 1 line
1.5.4.3
#28 lore raw

PATCH 5/11 debian: keep libdm-abi consistent

Quoted 2 lines · Fabio M. Di Nitto

don't change the ABI of libdm on selinux enable, just make the selinux function a NOP

Clearly this needs fixing upstream.

Alasdair

#29 lore raw

PATCH 5/11 debian: keep libdm-abi consistent

Quoted 5 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

don't change the ABI of libdm on selinux enable, just make the selinux function a NOP

Clearly this needs fixing upstream.

I have merged this one in CVS.

It appears to be the only bit missing from libdm to retain ABI consistency.

Fabio

#30 lore raw

PATCH 6/11 debian/ubuntu: make the device node creation more atomic

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=455745

---
:100644 100644 248d3ac... 90e0907... M libdm/libdm-common.c

 libdm/libdm-common.c |   32 ++++++++++++++++++++++++--------
 1 files changed, 24 insertions(+), 8 deletions(-)
libdm/libdm-common.c +24 -8
index 248d3ac..90e0907 100644
@@ -403,12 +403,19 @@ int dm_set_selinux_context(const char *path, mode_t mode)
 static int _add_dev_node(const char *dev_name, uint32_t major, uint32_t minor,
 			 uid_t uid, gid_t gid, mode_t mode)
 {
-	char path[PATH_MAX];
+	char path[PATH_MAX], tmppath[PATH_MAX];
 	struct stat info;
 	dev_t dev = MKDEV(major, minor);
 	mode_t old_mask;
 
 	_build_dev_path(path, sizeof(path), dev_name);
+	_build_dev_path(tmppath, sizeof(tmppath), ".XXXXXX");
+
+	mktemp(tmppath);
+	if (strlen(tmppath) == 0) {
+		log_error("Unable to create name of temporary file");
+		return 0;
+	}
 
 	if (stat(path, &info) >= 0) {
 		if (!S_ISBLK(info.st_mode)) {
@@ -431,24 +438,33 @@ static int _add_dev_node(const char *dev_name, uint32_t major, uint32_t minor,
 			 "node creation.", path);
 
 	old_mask = umask(0);
-	if (mknod(path, S_IFBLK | mode, dev) < 0) {
+	if (mknod(tmppath, S_IFBLK | mode, dev) < 0) {
 		umask(old_mask);
 		log_error("Unable to make device node for '%s'", dev_name);
-		return 0;
+		goto error;
 	}
 	umask(old_mask);
 
-	if (chown(path, uid, gid) < 0) {
+	if (chown(tmppath, uid, gid) < 0) {
 		log_sys_error("chown", path);
-		return 0;
+		goto error;
 	}
 
-	log_debug("Created %s", path);
+	if (!dm_set_selinux_context(tmppath, S_IFBLK))
+		goto error;
 
-	if (!dm_set_selinux_context(path, S_IFBLK))
-		return 0;
+	if (rename(tmppath, path) < 0) {
+		log_error("Unable to replace device node for '%s'", dev_name);
+		goto error;
+	}
+
+	log_debug("Created %s", path);
 
 	return 1;
+
+error:
+	unlink(tmppath);
+	return 0;
 }
 
 static int _rm_dev_node(const char *dev_name)
Signature · 1 line
1.5.4.3
#31 lore raw

PATCH 6/11 debian/ubuntu: make the device node creation more atomic

This one needs a fresh review. (I think I bounced it for further changes when it was originally submitted, but don't recall seeing a revised version.)

Alasdair

#32 lore raw

PATCH 6/11 debian/ubuntu: make the device node creation more atomic

Quoted 2 lines · Alasdair G Kergon

This one needs a fresh review. (I think I bounced it for further changes when it was originally submitted, but don't recall seeing a revised version.)

Do you have a reference to the original submission? I can't find it on either lvm-devel or linux-lvm mailing lists.

Thanks
Fabio

#33 lore raw

PATCH 6/11 debian/ubuntu: make the device node creation more atomic

Quoted 2 lines · Fabio M. Di Nitto

Do you have a reference to the original submission? I can't find it on either lvm-devel or linux-lvm mailing lists.

Forwarded. Debian bug 402132.

Alasdair

#34 lore raw

PATCH 6/11 debian/ubuntu: make the device node creation more atomic

Quoted 5 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

Do you have a reference to the original submission? I can't find it on either lvm-devel or linux-lvm mailing lists.

Forwarded. Debian bug 402132.

The old patch had hardcoded temporary file names.

This new one address the problem by using mktemp(3) and also the only comment you had on the Debian bug.

I think it's safe to pull it in.

Fabio

#35 lore raw

PATCH 7/11 debian: workaround for broken kernel headers

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

this patch might be debian specific and needs more investigation.

---
:100644 100644 5ba63ce... a8edf55... M libdm/ioctl/libdm-compat.h
:100644 100644 5fcda74... ec990d7... M libdm/misc/kdev_t.h

 libdm/ioctl/libdm-compat.h |    5 +++--
 libdm/misc/kdev_t.h        |    4 ++++
 2 files changed, 7 insertions(+), 2 deletions(-)
libdm/ioctl/libdm-compat.h +3 -2
index 5ba63ce..a8edf55 100644
@@ -19,6 +19,7 @@
 #include "kdev_t.h"
 #include "dm-ioctl.h"
 #include <inttypes.h>
+#include <linux/types.h>
 #include <sys/ioctl.h>
 
 struct dm_task;
@@ -40,7 +41,7 @@ struct dm_ioctl_v1 {
 	int32_t open_count;	/* out */
 	uint32_t flags;		/* in/out */
 
-	__kernel_dev_t dev;	/* in/out */
+	devmapper_kernel_dev_t dev;	/* in/out */
 
 	char name[DM_NAME_LEN];	/* device name */
 	char uuid[DM_UUID_LEN];	/* unique identifier for
@@ -60,7 +61,7 @@ struct dm_target_spec_v1 {
 struct dm_target_deps_v1 {
 	uint32_t count;
 
-	__kernel_dev_t dev[0];	/* out */
+	devmapper_kernel_dev_t dev[0];	/* out */
 };
 
 enum {
libdm/misc/kdev_t.h +4 -0
index 5fcda74..ec990d7 100644
@@ -15,8 +15,12 @@
 #ifndef _LIBDM_KDEV_H
 #define _LIBDM_KDEV_H
 
+#include <linux/types.h>
+
 #define MAJOR(dev)      ((dev & 0xfff00) >> 8)
 #define MINOR(dev)      ((dev & 0xff) | ((dev >> 12) & 0xfff00))
 #define MKDEV(ma,mi)    ((mi & 0xff) | (ma << 8) | ((mi & ~0xff) << 12))
 
+typedef __u32 devmapper_kernel_dev_t;
+
 #endif
Signature · 1 line
1.5.4.3
#36 lore raw

PATCH 7/11 debian: workaround for broken kernel headers

Quoted 1 line · Fabio M. Di Nitto

this patch might be debian specific and needs more investigation.

Yes, I'm not sure whether or not it's something we can deal with upstream somehow.

Alasdair

#38 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

use selinux pkgconfig file instead of hardcoding the contents

---
:100644 100644 b657090... d4ca036... M libdm/libdevmapper.pc.in

 libdm/libdevmapper.pc.in |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
libdm/libdevmapper.pc.in +1 -1
index b657090..d4ca036 100644
@@ -8,4 +8,4 @@ Description: device-mapper library
 Version: @DM_LIB_PATCHLEVEL@
 Cflags: -I${includedir} 
 Libs: -L${libdir} -ldevmapper
-Libs.private: @SELINUX_LIBS@
+Requires.private: libselinux
Signature · 1 line
1.5.4.3
#39 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

Quoted 1 line · Fabio M. Di Nitto

use selinux pkgconfig file instead of hardcoding the contents

I don't think this patch goes far enough...
The build should be using the pc file too (if present), as we now do for clvmd!

Alasdair

#40 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

Quoted 3 lines · Alasdair G Kergon
Quoted 1 line · Fabio M. Di Nitto

use selinux pkgconfig file instead of hardcoding the contents

I don't think this patch goes far enough...

This is the external view.

Quoted 1 line

The build should be using the pc file too (if present), as we now do for clvmd!

This is the internal view. This are different parts.

Bastian

Signature · 2 lines
Those who hate and fight must stop themselves -- otherwise it is not stopped.
		-- Spock, "Day of the Dove", stardate unknown
#41 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

In reply to Alasdair G Kergon
Quoted 5 lines · Alasdair G Kergon
Quoted 1 line · Fabio M. Di Nitto

use selinux pkgconfig file instead of hardcoding the contents

I don't think this patch goes far enough...
The build should be using the pc file too (if present), as we now do for clvmd!

We need to differentiate as Bastian said.

For the internal view, we can switch to use PKG_CONFIG inside our configure.in, but that means:

1) it will be a hard dependency (and we recently agreed to avoid that).

2) we assume that lvm2 developers and packagers will build/ship with recent enough build dependencies to guarantee working pkg-config files.

While I am generally ok with this approach as it will make our configure.in a lot simpler, it will kill the option to build on older distro (I believe we only lose RHEL4 at this point, but it would require cross-checking).

For the external view, as it is now, we can create the correct .pc file. If we detect that selinux has pkg-config support, then we use the "Requires.private: libselinux" that will automatically expand via pkg-config. If we don't have selinux pkg-config, then we can hardcode as we do now.

Fabio

#42 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

Quoted 5 lines · Fabio M. Di Nitto

For the internal view, we can switch to use PKG_CONFIG inside our configure.in, but that means:
1) it will be a hard dependency (and we recently agreed to avoid that).
2) we assume that lvm2 developers and packagers will build/ship with recent enough build dependencies to guarantee working pkg-config files.

So any switch would still need a fallback for when pkgconfig is not available. (The real question is whether pkgconfig was generally available some time before selinux or not. If it was, then you can argue that everyone with access to selinux also has access to pkgconfig.)

Quoted 5 lines

For the external view, as it is now, we can create the correct .pc file. If we detect that selinux has pkg-config support, then we use the "Requires.private: libselinux" that will automatically expand via pkg-config. If we don't have selinux pkg-config, then we can hardcode as we do now.

Yes.

Alasdair

#43 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

Quoted 4 lines · Alasdair G Kergon

So any switch would still need a fallback for when pkgconfig is not available. (The real question is whether pkgconfig was generally available some time before selinux or not. If it was, then you can argue that everyone with access to selinux also has access to pkgconfig.)

According to the Debian changelog, pkg-config was first usable in 2001, long before selinux was even considered a good idea.

Quoted 6 lines
Quoted 5 lines

For the external view, as it is now, we can create the correct .pc file. If we detect that selinux has pkg-config support, then we use the "Requires.private: libselinux" that will automatically expand via pkg-config. If we don't have selinux pkg-config, then we can hardcode as we do now.

Yes.

Well, libselinux at least ship the file since 1.28.

Bastian

Signature · 2 lines
No one may kill a man.  Not for any purpose.  It cannot be condoned.
		-- Kirk, "Spock's Brain", stardate 5431.6
#44 lore raw

PATCH 8/11 debian: use selinux pkgconfig info instead of hardcoding

In reply to Alasdair G Kergon
Quoted 11 lines · Alasdair G Kergon
Quoted 5 lines · Fabio M. Di Nitto

For the internal view, we can switch to use PKG_CONFIG inside our configure.in, but that means:
1) it will be a hard dependency (and we recently agreed to avoid that).
2) we assume that lvm2 developers and packagers will build/ship with recent enough build dependencies to guarantee working pkg-config files.

So any switch would still need a fallback for when pkgconfig is not available. (The real question is whether pkgconfig was generally available some time before selinux or not. If it was, then you can argue that everyone with access to selinux also has access to pkgconfig.)

Ok, this will take some more time to be done properly across the board. I don't want to special case every library/pkgconfig in a different way.

Fabio

#45 lore raw

PATCH 9/11 debian: libs cleanup

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

patch with several changes on how libs are found and used (selinux and readline)

patch is not clean or ready for submission

---
:100644 100644 06759b6... ad871fd... M configure.in
:100644 100644 207c0e2... 2a14ba6... M daemons/dmeventd/Makefile.in
:100644 100644 4273eeb... f20ad5d... M lib/misc/configure.h.in
:100644 100644 6dd25ff... 191e2ae... M libdm/Makefile.in
:100644 100644 6d12920... 5464188... M make.tmpl.in
:100644 100644 5702f4f... 4f8be22... M tools/Makefile.in

 configure.in                 |   57 +++++------------------------------------
 daemons/dmeventd/Makefile.in |    2 +-
 lib/misc/configure.h.in      |   14 +++-------
 libdm/Makefile.in            |    4 +-
 make.tmpl.in                 |    5 +--
 tools/Makefile.in            |   14 +++++-----
 6 files changed, 23 insertions(+), 73 deletions(-)
configure.in +7 -50
index 06759b6..ad871fd 100644
@@ -34,7 +34,6 @@ case "$host_os" in
 		CLDWHOLEARCHIVE="-Wl,-whole-archive"
 		CLDNOWHOLEARCHIVE="-Wl,-no-whole-archive"
 		LDDEPS="$LDDEPS .export.sym"
-		LDFLAGS="$LDFLAGS -Wl,--export-dynamic"
 		LIB_SUFFIX=so
 		DEVMAPPER=yes
 		ODIRECT=yes
@@ -779,24 +778,6 @@ if [[ "x$exec_prefix" = xNONE -a "x$prefix" = xNONE ]];
 fi;
 
 ################################################################################
-dnl -- Check for termcap (Shamelessly copied from parted 1.4.17)
-if test x$READLINE != xno; then
-	AC_SEARCH_LIBS([tgetent], [tinfo ncurses curses termcap termlib],
-	  [tg_found=yes], [tg_found=no])
-	test x$READLINE:$tg_found = xyes:no &&
-	  AC_MSG_ERROR(
-termcap could not be found which is required for the
---enable-readline option (which is enabled by default).  Either disable readline
-support with --disable-readline or download and install termcap from:
-	ftp.gnu.org/gnu/termcap
-Note: if you are using precompiled packages you will also need the development
-  package as well (which may be called termcap-devel or something similar).
-Note: (n)curses also seems to work as a substitute for termcap.  This was
-  not found either - but you could try installing that as well.
-)
-fi
-
-################################################################################
 dnl -- Check for dlopen
 AC_CHECK_LIB(dl, dlopen, HAVE_LIBDL=yes, HAVE_LIBDL=no)
 
@@ -827,35 +808,9 @@ AC_MSG_RESULT($SELINUX)
 ################################################################################
 dnl -- Check for selinux
 if test x$SELINUX = xyes; then
-	AC_CHECK_LIB(sepol, sepol_check_context, HAVE_SEPOL=yes, HAVE_SEPOL=no)
-
-	if test x$HAVE_SEPOL = xyes; then
-		AC_DEFINE([HAVE_SEPOL], 1,
-		  [Define to 1 if sepol_check_context is available.])
-		SELINUX_LIBS="-lsepol $SELINUX_LIBS"
-	fi
-
-	AC_CHECK_LIB(selinux, is_selinux_enabled, HAVE_SELINUX=yes, HAVE_SELINUX=no)
-
-	if test x$HAVE_SELINUX = xyes; then
-		AC_DEFINE([HAVE_SELINUX], 1, [Define to 1 to include support for selinux.])
-		SELINUX_LIBS="-lselinux $SELINUX_LIBS"
-	else
-		AC_MSG_WARN(Disabling selinux)
-	fi
-	LIBS="$SELINUX_LIBS $LIBS"
-
-	# With --enable-static_link and selinux enabled, linking
-	# fails on at least Debian unstable due to unsatisfied references
-	# to pthread_mutex_lock and _unlock.  See if we need -lpthread.
-	if test "$STATIC_LINK-$HAVE_SELINUX" = yes-yes; then
-		lvm_saved_libs=$LIBS
-		LIBS="$LIBS -static"
-		AC_SEARCH_LIBS([pthread_mutex_lock], [pthread],
-		  [test "$ac_cv_search_pthread_mutex_lock" = "none required" ||
-				LIB_PTHREAD=-lpthread])
-		LIBS=$lvm_saved_libs
-	fi
+	AC_DEFINE([HAVE_SELINUX], 1, [Define to 1 to include support for selinux.])
+	LIBS_SELINUX="$(pkg-config --libs libselinux)"
+	LIBS_SELINUX_STATIC="$(pkg-config --libs --static libselinux)"
 fi
 
 ################################################################################
@@ -866,7 +821,7 @@ AC_CHECK_HEADERS(getopt.h, AC_DEFINE([HAVE_GETOPTLONG], 1, [Define to 1 if getop
 dnl -- Check for readline (Shamelessly copied from parted 1.4.17)
 if test x$READLINE != xno; then
 	rl_found=yes
-	AC_CHECK_LIB([readline], [readline], , [rl_found=no])
+	AC_CHECK_LIB([readline], [readline], [LIBS_READLINE="-lreadline $LIBS_READLINE"], [rl_found=no])
 	test x$READLINE:$rl_found = xyes:no &&
 	  AC_MSG_ERROR(
 GNU Readline could not be found which is required for the
@@ -1078,6 +1033,9 @@ AC_SUBST(INTL_PACKAGE)
 AC_SUBST(JOBS)
 AC_SUBST(LDDEPS)
 AC_SUBST(LIBS)
+AC_SUBST(LIBS_READLINE)
+AC_SUBST(LIBS_SELINUX)
+AC_SUBST(LIBS_SELINUX_STATIC)
 AC_SUBST(LIB_SUFFIX)
 AC_SUBST(LOCALEDIR)
 AC_SUBST(LVM1)
@@ -1105,7 +1063,6 @@ AC_SUBST(SNAPSHOTS)
 AC_SUBST(STATICDIR)
 AC_SUBST(STATIC_LINK)
 AC_SUBST(UDEV_RULES)
-AC_SUBST([LIB_PTHREAD])
 AC_SUBST(interface)
 AC_SUBST(kerneldir)
 AC_SUBST(missingkernel)
daemons/dmeventd/Makefile.in +1 -1
index 207c0e2..2a14ba6 100644
@@ -46,7 +46,7 @@ LIBS += -ldl -ldevmapper -lpthread
 
 dmeventd: $(LIB_SHARED) $(VERSIONED_SHLIB) dmeventd.o
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ dmeventd.o \
-	-L. -ldevmapper-event $(LIBS) -rdynamic
+	-L. -ldevmapper-event
 
 .PHONY: install_dynamic install_static install_include \
 	install_pkgconfig install_dmeventd
lib/misc/configure.h.in +4 -10
index 4273eeb..f20ad5d 100644
@@ -120,9 +120,6 @@
 /* Define to 1 if you have the <libintl.h> header file. */
 #undef HAVE_LIBINTL_H
 
-/* Define to 1 if you have the `readline' library (-lreadline). */
-#undef HAVE_LIBREADLINE
-
 /* Define to 1 if you have the <limits.h> header file. */
 #undef HAVE_LIMITS_H
 
@@ -185,9 +182,6 @@
 /* Define to 1 if you have the <readline/readline.h> header file. */
 #undef HAVE_READLINE_READLINE_H
 
-/* Define to 1 to include support for realtime clock. */
-#undef HAVE_REALTIME
-
 /* Define to 1 if you have the `rl_completion_matches' function. */
 #undef HAVE_RL_COMPLETION_MATCHES
 
@@ -206,9 +200,6 @@
 /* Define to 1 if you have the <selinux/selinux.h> header file. */
 #undef HAVE_SELINUX_SELINUX_H
 
-/* Define to 1 if sepol_check_context is available. */
-#undef HAVE_SEPOL
-
 /* Define to 1 if you have the `setlocale' function. */
 #undef HAVE_SETLOCALE
 
@@ -273,7 +264,7 @@
 /* Define to 1 if you have the `strtoul' function. */
 #undef HAVE_STRTOUL
 
-/* Define to 1 if `st_rdev' is member of `struct stat'. */
+/* Define to 1 if `struct stat' is a member of `st_rdev'. */
 #undef HAVE_STRUCT_STAT_ST_RDEV
 
 /* Define to 1 if you have the <syslog.h> header file. */
@@ -410,6 +401,9 @@
 /* Define to the one symbol short name of this package. */
 #undef PACKAGE_TARNAME
 
+/* Define to the home page for this package. */
+#undef PACKAGE_URL
+
 /* Define to the version of this package. */
 #undef PACKAGE_VERSION
 
libdm/Makefile.in +2 -2
index 6dd25ff..191e2ae 100644
@@ -35,9 +35,7 @@ SOURCES =\
 
 INCLUDES = -I$(srcdir)/$(interface) -I$(srcdir)
 
-ifeq ("@STATIC_LINK@", "yes")
 LIB_STATIC = $(interface)/libdevmapper.a
-endif
 
 ifeq ("@LIB_SUFFIX@","dylib")
   LIB_SHARED = $(interface)/libdevmapper.dylib
@@ -47,6 +45,8 @@ endif
 VERSIONED_SHLIB = libdevmapper.$(LIB_SUFFIX).$(LIB_VERSION_DM)
 LIB_VERSION = $(LIB_VERSION_DM)
 
+LDFLAGS += @LIBS_SELINUX@
+
 DEFS += -DDM_DEVICE_UID=@DM_DEVICE_UID@ -DDM_DEVICE_GID=@DM_DEVICE_GID@ \
 	-DDM_DEVICE_MODE=@DM_DEVICE_MODE@
 
make.tmpl.in +2 -3
index 6d12920..5464188 100644
@@ -105,7 +105,6 @@ endif
 #DEFS += -DBOUNDS_CHECK
 
 #CFLAGS += -pg
-#LDFLAGS += -pg
 
 STRIP=
 #STRIP = -s
@@ -209,13 +208,13 @@ $(TARGETS): $(OBJECTS)
 ifeq ("@LIB_SUFFIX@","so")
 $(LIB_SHARED): $(OBJECTS) $(LDDEPS)
 	$(CC) -shared -Wl,-soname,$(notdir $@).$(LIB_VERSION) \
-	$(CFLAGS) $(CLDFLAGS) $(OBJECTS) $(LIBS) -o $@
+	$(CFLAGS) $(LDFLAGS) $(CLDFLAGS) $(OBJECTS) -o $@
 endif
 
 ifeq ("@LIB_SUFFIX@","dylib")
 $(LIB_SHARED): $(OBJECTS) $(LDDEPS)
 	$(CC) -dynamiclib -dylib_current_version,$(LIB_VERSION) \
-	$(CFLAGS) $(CLDFLAGS) $(OBJECTS) $(LIBS) -o $@
+	$(CFLAGS) $(LDFLAGS) $(CLDFLAGS) $(OBJECTS) -o $@
 endif
 
 %.so: %.a
tools/Makefile.in +7 -7
index 5702f4f..4f8be22 100644
@@ -17,8 +17,6 @@ top_srcdir = @top_srcdir@
 top_builddir = @top_builddir@
 VPATH = @srcdir@
 
-LIB_PTHREAD = @LIB_PTHREAD@
-
 SOURCES =\
 	dumpconfig.c \
 	formats.c \
@@ -105,6 +103,9 @@ endif
 
 LVMLIBS += -ldevmapper
 
+LDFLAGS_DMSETUP_STATIC = @LIBS_SELINUX_STATIC@
+LDFLAGS_LVM = @LIBS_READLINE@
+
 DEFS += -DLVM_SHARED_PATH=\"$(exec_prefix)/sbin/lvm\"
 
 include ../make.tmpl
@@ -113,22 +114,21 @@ device-mapper: $(TARGETS_DM)
 
 dmsetup: dmsetup.o $(top_builddir)/libdm/libdevmapper.$(LIB_SUFFIX)
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ dmsetup.o \
-	      -L$(top_builddir)/libdm -ldevmapper $(LIBS) $(LIB_PTHREAD)
+	      -L$(top_builddir)/libdm -ldevmapper $(LIBS)
 
 dmsetup.static: dmsetup.o $(interfacedir)/libdevmapper.a
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ dmsetup.o -static \
-	      -L$(interfacedir) -ldevmapper $(LIBS) \
-	      $(LIB_PTHREAD)
+	      -L$(interfacedir) -ldevmapper $(LIBS) $(LDFLAGS_DMSETUP_STATIC)
 
 all: device-mapper
 
 lvm: $(OBJECTS) lvm.o $(top_builddir)/lib/liblvm-internal.a
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) lvm.o \
-		$(LVMLIBS) $(LIBS) $(LIB_PTHREAD) -rdynamic
+		$(LDFLAGS) $(LVMLIBS) $(LIBS) $(LDFLAGS_LVM) -rdynamic
 
 lvm.static: $(OBJECTS) lvm-static.o $(top_builddir)/lib/liblvm-internal.a  $(interfacedir)/libdevmapper.a
 	$(CC) $(CFLAGS) $(LDFLAGS) -o $@ $(OBJECTS) lvm-static.o -static \
-		-L$(interfacedir) $(LVMLIBS) $(LIBS) $(LIB_PTHREAD) -rdynamic
+		-L$(interfacedir) $(LVMLIBS) $(LIBS) -rdynamic
 
 liblvm2cmd.a: $(top_builddir)/lib/liblvm-internal.a $(OBJECTS) lvmcmdlib.o lvm2cmd.o
 	cat $(top_builddir)/lib/liblvm-internal.a > $@
Signature · 1 line
1.5.4.3
#46 lore raw

PATCH 9/11 debian: libs cleanup

Quoted 3 lines · Fabio M. Di Nitto

patch with several changes on how libs are found and used (selinux and readline)
patch is not clean or ready for submission

Yes, this one needs picking through carefully to see what needs changing and then to decide how.

Alasdair

#47 lore raw

PATCH 10/11 debian: make the subdir recursion -j capable

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

but it removes some dependencies on the way

---
:100644 100644 556d1ff... 2e8a010... M Makefile.in
:100644 100644 5464188... 56dfa8b... M make.tmpl.in
:100644 100644 f0c98fd... f0acb84... M man/Makefile.in

 Makefile.in     |   27 +----------------------
 make.tmpl.in    |   64 +++++++++++-------------------------------------------
 man/Makefile.in |    4 +--
 3 files changed, 15 insertions(+), 80 deletions(-)
Makefile.in +1 -26
index 556d1ff..2e8a010 100644
@@ -27,7 +27,7 @@ ifeq ("@INTL@", "yes")
   SUBDIRS += po
 endif
 
-SUBDIRS += lib tools daemons libdm
+SUBDIRS += libdm lib tools daemons
 
 ifeq ("@APPLIB@", "yes")
   SUBDIRS += liblvm
@@ -54,31 +54,6 @@ endif
 
 include make.tmpl
 
-libdm: include
-lib: libdm
-liblvm: lib
-daemons: lib tools
-tools: lib device-mapper
-po: tools daemons
-
-libdm.device-mapper: include.device-mapper
-daemons.device-mapper: libdm.device-mapper
-tools.device-mapper: libdm.device-mapper
-device-mapper: tools.device-mapper daemons.device-mapper man.device-mapper
-
-ifeq ("@INTL@", "yes")
-lib.pofile: include.pofile
-tools.pofile: lib.pofile
-daemons.pofile: lib.pofile
-po.pofile: tools.pofile daemons.pofile
-pofile: po.pofile
-endif
-
-ifneq ("@CFLOW_CMD@", "")
-tools.cflow: lib.cflow
-cflow: tools.cflow
-endif
-
 ifneq ("@CSCOPE_CMD@", "")
 cscope.out: tools
 	@CSCOPE_CMD@ -b -R
make.tmpl.in +13 -51

90 lines not on this page · raw message

man/Makefile.in +1 -3
index f0c98fd..f0acb84 100644
@@ -49,9 +49,7 @@ all: man
 
 .PHONY: man
 
-device-mapper: $(MAN8DM)
-
-man: $(MAN5) $(MAN8) $(MAN8CLUSTER)
+man: $(MAN5) $(MAN8) $(MAN8CLUSTER) $(MAN8DM)
 
 $(MAN5) $(MAN8) $(MAN8CLUSTER):	Makefile
 
Signature · 1 line
1.5.4.3
#48 lore raw

PATCH 10/11 debian: make the subdir recursion -j capable

Quoted 1 line · Fabio M. Di Nitto

but it removes some dependencies on the way

The tree isn't constructed for a basic recursion mechanism to work properly at the moment, and the inter-dependencies all play a role in coping with whatever configuration and make targets are thrown at it. (Nor does the current setup work perfectly today.)

So let's park this one for now, then include the possibility of moving some files/directories/targets around so a simple recursion mechanism could perhaps be made to cope.

Alasdair

#49 lore raw

PATCH 10/11 debian: make the subdir recursion -j capable

Quoted 4 lines · Alasdair G Kergon

The tree isn't constructed for a basic recursion mechanism to work properly at the moment, and the inter-dependencies all play a role in coping with whatever configuration and make targets are thrown at it. (Nor does the current setup work perfectly today.)

Even the dependencies reduces it to a simple recursion model. make is simply not able to provide something else if you don't use one large makefile. And it works quite well.

Bastian

Signature · 2 lines
Killing is wrong.
		-- Losira, "That Which Survives", stardate unknown
#50 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

In reply to Fabio M. Di Nitto

From: Fabio M. Di Nitto <[email protected]>

remove owner and group definitions and install anything writable as mendated by the debian policy

---
:100644 100644 ad871fd... 6b93b77... M configure.in
:100644 100644 dad568e... a6ecfe9... M daemons/clvmd/Makefile.in
:100644 100644 debab27... fa08f03... M daemons/cmirrord/Makefile.in
:100644 100644 2a14ba6... bf71fdb... M daemons/dmeventd/Makefile.in
:100644 100644 581e5ec... b9f6974... M daemons/dmeventd/plugins/mirror/Makefile.in
:100644 100644 823ca3f... a07558e... M daemons/dmeventd/plugins/snapshot/Makefile.in
:100644 100644 b4689af... f589489... M doc/Makefile.in
:100644 100644 e12cc6f... f70adf7... M lib/format1/Makefile.in
:100644 100644 d67c982... 36230f4... M lib/format_pool/Makefile.in
:100644 100644 584c8d6... 8fa8eb6... M lib/locking/Makefile.in
:100644 100644 d5b678b... b7076f7... M lib/mirror/Makefile.in
:100644 100644 3f960fb... 3de0373... M lib/snapshot/Makefile.in
:100644 100644 191e2ae... dea2771... M libdm/Makefile.in
:100644 100644 d9afc40... 5b37426... M liblvm/Makefile.in
:100644 100644 56dfa8b... e4acf88... M make.tmpl.in
:100644 100644 f0acb84... 647488d... M man/Makefile.in
:100644 100644 c9f8400... 12d4489... M po/Makefile.in
:100644 100644 33a10c1... 677f965... M scripts/Makefile.in
:100644 100644 4f8be22... bd834b9... M tools/Makefile.in
:100644 100644 341479f... 8c71c2b... M udev/Makefile.in

 configure.in                                  |   29 -------------------------
 daemons/clvmd/Makefile.in                     |    4 +-
 daemons/cmirrord/Makefile.in                  |    2 +-
 daemons/dmeventd/Makefile.in                  |   10 ++++----
 daemons/dmeventd/plugins/mirror/Makefile.in   |    2 +-
 daemons/dmeventd/plugins/snapshot/Makefile.in |    2 +-
 doc/Makefile.in                               |    2 +-
 lib/format1/Makefile.in                       |    3 +-
 lib/format_pool/Makefile.in                   |    3 +-
 lib/locking/Makefile.in                       |    3 +-
 lib/mirror/Makefile.in                        |    3 +-
 lib/snapshot/Makefile.in                      |    3 +-
 libdm/Makefile.in                             |   16 ++++---------
 liblvm/Makefile.in                            |    8 +++---
 make.tmpl.in                                  |    9 +-------
 man/Makefile.in                               |    8 +++---
 po/Makefile.in                                |    4 +-
 scripts/Makefile.in                           |    6 ++--
 tools/Makefile.in                             |   16 +++++++-------
 udev/Makefile.in                              |    4 +-
 20 files changed, 45 insertions(+), 92 deletions(-)
configure.in +0 -29
index ad871fd..6b93b77 100644
@@ -141,33 +141,6 @@ dnl -- Prefix is /usr by default, the exec_prefix default is setup later
 AC_PREFIX_DEFAULT(/usr)
 
 ################################################################################
-dnl -- Setup the ownership of the files
-AC_MSG_CHECKING(file owner)
-OWNER="root"
-
-AC_ARG_WITH(user,
-  [  --with-user=USER        Set the owner of installed files [[USER=root]] ],
-  [ OWNER="$withval" ])
-AC_MSG_RESULT($OWNER)
-
-if test x$OWNER != x; then
-	OWNER="-o $OWNER"
-fi
-
-################################################################################
-dnl -- Setup the group ownership of the files
-AC_MSG_CHECKING(group owner)
-GROUP="root"
-AC_ARG_WITH(group,
-  [  --with-group=GROUP      Set the group owner of installed files [[GROUP=root]] ],
-  [ GROUP="$withval" ])
-AC_MSG_RESULT($GROUP)
-
-if test x$GROUP != x; then
-	GROUP="-g $GROUP"
-fi
-
-################################################################################
 dnl -- Setup device node ownership
 AC_MSG_CHECKING(device node uid)
 
@@ -1023,7 +996,6 @@ AC_SUBST(DM_IOCTLS)
 AC_SUBST(DM_LIB_VERSION)
 AC_SUBST(DM_LIB_PATCHLEVEL)
 AC_SUBST(FSADM)
-AC_SUBST(GROUP)
 AC_SUBST(GULM_CFLAGS)
 AC_SUBST(GULM_LIBS)
 AC_SUBST(HAVE_LIBDL)
@@ -1049,7 +1021,6 @@ AC_SUBST(LVM_RELEASE)
 AC_SUBST(LVM_RELEASE_DATE)
 AC_SUBST(MIRRORS)
 AC_SUBST(MSGFMT)
-AC_SUBST(OWNER)
 AC_SUBST(PKGCONFIG)
 AC_SUBST(POOL)
 AC_SUBST(QUORUM_CFLAGS)
daemons/clvmd/Makefile.in +2 -2
index dad568e..a6ecfe9 100644
@@ -98,8 +98,8 @@ clvmd: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
 .PHONY: install_clvmd
 
 install_clvmd: $(TARGETS)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) clvmd \
-		$(usrsbindir)/clvmd
+	$(INSTALL) -D clvmd \
+		$(sbindir)/clvmd
 
 install: $(INSTALL_TARGETS)
 
daemons/cmirrord/Makefile.in +1 -1
index debab27..fa08f03 100644
@@ -36,5 +36,5 @@ cmirrord: $(OBJECTS) $(top_builddir)/lib/liblvm-internal.a
 		$(LVMLIBS) $(LMLIBS) $(LIBS)
 
 install: $(TARGETS)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) cmirrord \
+	$(INSTALL) -D cmirrord \
 		$(usrsbindir)/cmirrord
daemons/dmeventd/Makefile.in +5 -5
index 2a14ba6..bf71fdb 100644
@@ -66,24 +66,24 @@ install: $(INSTALL_TYPE) install_include install_dmeventd
 install_device-mapper: $(INSTALL_TYPE) install_include install_dmeventd
 
 install_include:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/libdevmapper-event.h \
+	$(INSTALL) -D -m 644 $(srcdir)/libdevmapper-event.h \
 		$(includedir)/libdevmapper-event.h
 
 install_dynamic: libdevmapper-event.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION)
 	$(LN_S) -f libdevmapper-event.$(LIB_SUFFIX).$(LIB_VERSION) \
 		$(libdir)/libdevmapper-event.$(LIB_SUFFIX)
 
 install_dmeventd: dmeventd
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+	$(INSTALL) -D $< $(sbindir)/$<
 
 install_pkgconfig:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper-event.pc \
+	$(INSTALL) -D -m 644 libdevmapper-event.pc \
 		$(usrlibdir)/pkgconfig/devmapper-event.pc
 
 install_static: libdevmapper-event.a
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/libdevmapper-event.a.$(LIB_VERSION)
 	$(LN_S) -f libdevmapper-event.a.$(LIB_VERSION) $(libdir)/libdevmapper-event.a
 
daemons/dmeventd/plugins/mirror/Makefile.in +1 -1
index 581e5ec..b9f6974 100644
@@ -34,7 +34,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include $(top_builddir)/make.tmpl
 
 install_lvm2: libdevmapper-event-lvm2mirror.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/$<.$(LIB_VERSION)
 	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
 
daemons/dmeventd/plugins/snapshot/Makefile.in +1 -1
index 823ca3f..a07558e 100644
@@ -34,7 +34,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include $(top_builddir)/make.tmpl
 
 install_lvm2: libdevmapper-event-lvm2snapshot.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/$<.$(LIB_VERSION)
 	$(LN_S) -f $<.$(LIB_VERSION) $(libdir)/$<
 
doc/Makefile.in +1 -1
index b4689af..f589489 100644
@@ -24,7 +24,7 @@ include ../make.tmpl
 install_lvm2: $(CONFSRC)
 	@if [ ! -e $(confdir)/$(CONFDEST) ]; then \
 		echo "Installing $(CONFSRC) as $(confdir)/$(CONFDEST)"; \
-		@INSTALL@ -D $(OWNER) $(GROUP) -m 644 $< \
+		@INSTALL@ -D $< \
 			$(confdir)/$(CONFDEST); \
 	fi
 
lib/format1/Makefile.in +1 -2
index e12cc6f..f70adf7 100644
@@ -32,7 +32,6 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include ../../make.tmpl
 
 install: liblvm2format1.so
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/liblvm2format1.so.$(LIB_VERSION)
+	$(INSTALL) -D $< $(libdir)/liblvm2format1.so.$(LIB_VERSION)
 	$(LN_S) -f liblvm2format1.so.$(LIB_VERSION) $(libdir)/liblvm2format1.so
 
lib/format_pool/Makefile.in +1 -2
index d67c982..36230f4 100644
@@ -29,8 +29,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include ../../make.tmpl
 
 install: liblvm2formatpool.so
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/liblvm2formatpool.so.$(LIB_VERSION)
+	$(INSTALL) -D $< $(libdir)/liblvm2formatpool.so.$(LIB_VERSION)
 	$(LN_S) -f liblvm2formatpool.so.$(LIB_VERSION) \
 		$(libdir)/liblvm2formatpool.so
 
lib/locking/Makefile.in +1 -2
index 584c8d6..8fa8eb6 100644
@@ -25,8 +25,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include ../../make.tmpl
 
 install install_cluster: liblvm2clusterlock.so
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/liblvm2clusterlock.so.$(LIB_VERSION)
+	$(INSTALL) -D $< $(libdir)/liblvm2clusterlock.so.$(LIB_VERSION)
 	$(LN_S) -f liblvm2clusterlock.so.$(LIB_VERSION) \
 		$(libdir)/liblvm2clusterlock.so
 
lib/mirror/Makefile.in +1 -2
index d5b678b..b7076f7 100644
@@ -25,7 +25,6 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include ../../make.tmpl
 
 install: liblvm2mirror.so
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/liblvm2mirror.so.$(LIB_VERSION)
+	$(INSTALL) -D $< $(libdir)/liblvm2mirror.so.$(LIB_VERSION)
 	$(LN_S) -f liblvm2mirror.so.$(LIB_VERSION) $(libdir)/liblvm2mirror.so
 
lib/snapshot/Makefile.in +1 -2
index 3f960fb..3de0373 100644
@@ -25,8 +25,7 @@ LIB_VERSION = $(LIB_VERSION_LVM)
 include ../../make.tmpl
 
 install: liblvm2snapshot.so
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/liblvm2snapshot.so.$(LIB_VERSION)
+	$(INSTALL) -D $< $(libdir)/liblvm2snapshot.so.$(LIB_VERSION)
 	$(LN_S) -f liblvm2snapshot.so.$(LIB_VERSION) \
 		$(libdir)/liblvm2snapshot.so
 
libdm/Makefile.in +5 -11
index 191e2ae..dea2771 100644
@@ -58,11 +58,7 @@ device-mapper: all
 	install_ioctl install_ioctl_static \
 	install_pkgconfig
 
-INSTALL_TYPE = install_dynamic
-
-ifeq ("@STATIC_LINK@", "yes")
-  INSTALL_TYPE += install_static
-endif
+INSTALL_TYPE = install_dynamic install_static
 
 ifeq ("@PKGCONFIG@", "yes")
   INSTALL_TYPE += install_pkgconfig
@@ -73,7 +69,7 @@ install: $(INSTALL_TYPE) install_include
 install_device-mapper: install
 
 install_include:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/libdevmapper.h \
+	$(INSTALL_DATA) -D $(srcdir)/libdevmapper.h \
 		$(includedir)/libdevmapper.h
 
 install_dynamic: install_ at interface@
@@ -81,19 +77,17 @@ install_dynamic: install_ at interface@
 		$(libdir)/libdevmapper.$(LIB_SUFFIX)
 
 install_static: install_@interface@_static
-	$(LN_S) -f libdevmapper.a.$(LIB_VERSION_DM) $(libdir)/libdevmapper.a
 
 install_ioctl: ioctl/libdevmapper.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/libdevmapper.$(LIB_SUFFIX).$(LIB_VERSION_DM)
 
 install_pkgconfig:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 libdevmapper.pc \
+	$(INSTALL_DATA) -D libdevmapper.pc \
 		$(usrlibdir)/pkgconfig/devmapper.pc
 
 install_ioctl_static: ioctl/libdevmapper.a
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
-		$(libdir)/libdevmapper.a.$(LIB_VERSION_DM)
+	$(INSTALL) -D $< $(usrlibdir)/libdevmapper.a
 
 $(VERSIONED_SHLIB): %.$(LIB_SUFFIX).$(LIB_VERSION_DM): $(interface)/%.$(LIB_SUFFIX)
 	rm -f $@
liblvm/Makefile.in +4 -4
index d9afc40..5b37426 100644
@@ -67,21 +67,21 @@ endif
 install: $(INSTALL_TYPE) install_include
 
 install_include:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/lvm2app.h $(includedir)/lvm2app.h
+	$(INSTALL_DATA) -D $(srcdir)/lvm2app.h $(includedir)/lvm2app.h
 
 install_dynamic: $(LIB_SHARED)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/$(LIB_SHARED).$(LIB_VERSION_APP)
 	$(LN_S) -f $(LIB_SHARED).$(LIB_VERSION_APP) \
 		$(libdir)/$(LIB_SHARED)
 
 install_static: $(LIB_STATIC)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/$(LIB_STATIC).$(LIB_VERSION_APP)
 	$(LN_S) -f $(LIB_STATIC).$(LIB_VERSION_APP) $(libdir)/$(LIB_STATIC)
 
 install_pkgconfig:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(LIB_NAME).pc \
+	$(INSTALL_DATA) -D $(LIB_NAME).pc \
 		$(usrlibdir)/pkgconfig/lvm2app.pc
 
 liblvm.cflow: $(SOURCES)
make.tmpl.in +1 -8
index 56dfa8b..e4acf88 100644
@@ -21,6 +21,7 @@ CC = @CC@
 RANLIB = @RANLIB@
 SHELL = /bin/sh
 INSTALL = @INSTALL@
+INSTALL_DATA = $(INSTALL) -m 644
 MKDIR_P = @MKDIR_P@
 MSGFMT = @MSGFMT@
 LCOV = @LCOV@
@@ -56,11 +57,6 @@ udevdir = $(DESTDIR)@udevdir@
 interface = @interface@
 interfacedir = $(top_srcdir)/libdm/$(interface)
 
-# setup misc variables
-# define the ownership variables for the binaries and man pages
-OWNER = @OWNER@
-GROUP = @GROUP@
-
 # The number of jobs to run, if blank, defaults to the make standard
 ifndef MAKEFLAGS
 MAKEFLAGS = @JOBS@
@@ -106,9 +102,6 @@ endif
 
 #CFLAGS += -pg
 
-STRIP=
-#STRIP = -s
-
 LVM_VERSION := $(shell cat $(top_srcdir)/VERSION)
 
 LIB_VERSION_LVM := $(shell cat $(top_srcdir)/VERSION | \
man/Makefile.in +4 -4
index f0acb84..647488d 100644
@@ -61,14 +61,14 @@ install_lvm2:
 	@for f in $(MAN8); \
 	do \
 	   $(RM) $(MAN8DIR)/$$f; \
-	   @INSTALL@ -D $(OWNER) $(GROUP) -m 444 $$f $(MAN8DIR)/$$f; \
+	   $(INSTALL_DATA) -D $$f $(MAN8DIR)/$$f; \
 	done
 
 	@echo "Installing $(MAN5) in $(MAN5DIR)"
 	@for f in $(MAN5); \
 	do \
 	   $(RM) $(MAN5DIR)/$$f; \
-	   @INSTALL@ -D $(OWNER) $(GROUP) -m 444 $$f $(MAN5DIR)/$$f; \
+	   $(INSTALL_DATA) -D $$f $(MAN5DIR)/$$f; \
 	done
 
 install_cluster:
@@ -76,7 +76,7 @@ install_cluster:
 	@for f in $(MAN8CLUSTER); \
 	do \
 	   $(RM) $(MAN8DIR)/$$f; \
-	   @INSTALL@ -D $(OWNER) $(GROUP) -m 444 $$f $(MAN8DIR)/$$f; \
+	   $(INSTALL_DATA) -D $$f $(MAN8DIR)/$$f; \
 	done
 
 install_device-mapper:
@@ -84,7 +84,7 @@ install_device-mapper:
 	@for f in $(MAN8DM); \
 	do \
 	   $(RM) $(MAN8DIR)/$$f; \
-	   @INSTALL@ -D $(OWNER) $(GROUP) -m 444 $$f $(MAN8DIR)/$$f; \
+	   $(INSTALL_DATA) -D $$f $(MAN8DIR)/$$f; \
 	done
 
 install: install_lvm2 install_device-mapper
po/Makefile.in +2 -2
index c9f8400..12d4489 100644
@@ -54,13 +54,13 @@ install: $(TARGETS)
 	@echo Installing translation files in $(localedir)
 	@( \
 		for lang in $(LANGS); do \
-			$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $$lang.mo \
+			$(INSTALL_DATA) -D $$lang.mo \
 			    $(localedir)/$$lang/LC_MESSAGES/lvm2.mo;\
 	  done; \
 	)
 	@( \
 		for lang in $(LANGS); do \
-			$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $$lang.mo \
+			$(INSTALL_DATA) -D $$lang.mo \
 			    $(localedir)/$$lang/LC_MESSAGES/device-mapper.mo;\
 	  done; \
 	)
scripts/Makefile.in +3 -3
index 33a10c1..677f965 100644
@@ -19,12 +19,12 @@ VPATH = @srcdir@
 include ../make.tmpl
 
 install:
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(srcdir)/lvm_dump.sh \
+	$(INSTALL) -D $(srcdir)/lvm_dump.sh \
 		$(sbindir)/lvmdump
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(srcdir)/vgimportclone.sh \
+	$(INSTALL) -D $(srcdir)/vgimportclone.sh \
 		$(sbindir)/vgimportclone
 ifeq ("@FSADM@", "yes")
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(srcdir)/fsadm.sh \
+	$(INSTALL) -D $(srcdir)/fsadm.sh \
 		$(sbindir)/fsadm
 endif
 
tools/Makefile.in +8 -8
index 4f8be22..bd834b9 100644
@@ -165,21 +165,21 @@ cflow: lvm.cflow lvm.xref lvm.tree lvm.rtree lvm.rxref
 endif
 
 install_cmdlib_dynamic: liblvm2cmd.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/liblvm2cmd.$(LIB_SUFFIX).$(LIB_VERSION)
 	$(LN_S) -f liblvm2cmd.$(LIB_SUFFIX).$(LIB_VERSION) $(libdir)/liblvm2cmd.$(LIB_SUFFIX)
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/lvm2cmd.h \
+	$(INSTALL_DATA) -D $(srcdir)/lvm2cmd.h \
 		$(includedir)/lvm2cmd.h
 
 install_cmdlib_static: liblvm2cmd-static.a
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(libdir)/liblvm2cmd.a.$(LIB_VERSION)
 	$(LN_S) -f liblvm2cmd.a.$(LIB_VERSION) $(libdir)/liblvm2cmd.a
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 444 $(srcdir)/lvm2cmd.h \
+	$(INSTALL_DATA) -D $(srcdir)/lvm2cmd.h \
 		$(includedir)/lvm2cmd.h
 
 install_tools_dynamic: lvm .commands
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) lvm \
+	$(INSTALL) -D lvm \
 		$(sbindir)/lvm
 	@echo Creating symbolic links for individual commands in $(sbindir)
 	@( \
@@ -190,14 +190,14 @@ install_tools_dynamic: lvm .commands
 	)
 
 install_tools_static: lvm.static
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< \
+	$(INSTALL) -D $< \
 		$(staticdir)/lvm.static
 
 install_dmsetup_dynamic: dmsetup
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+	$(INSTALL) -D $< $(sbindir)/$<
 
 install_dmsetup_static: dmsetup.static
-	$(INSTALL) -D $(OWNER) $(GROUP) -m 555 $(STRIP) $< $(sbindir)/$<
+	$(INSTALL) -D $< $(sbindir)/$<
 
 install_device-mapper: $(INSTALL_DMSETUP_TARGETS)
 
udev/Makefile.in +2 -2
index 341479f..8c71c2b 100644
@@ -32,7 +32,7 @@ install_lvm2: $(LVM_RULES)
 	@for f in $(LVM_RULES); \
 	do \
 	   $(RM) $(udevdir)/$$f; \
-	   $(INSTALL) -c -D $(OWNER) $(GROUP) -m 644 $(srcdir)/$$f $(udevdir)/$$f; \
+	   $(INSTALL_DATA) -c -D  $(srcdir)/$$f $(udevdir)/$$f; \
 	done
 
 install_device-mapper: $(DM_RULES)
@@ -40,7 +40,7 @@ install_device-mapper: $(DM_RULES)
 	@for f in $(DM_RULES); \
 	do \
 	   $(RM) $(udevdir)/$$f; \
-	   $(INSTALL) -c -D $(OWNER) $(GROUP) -m 644 $(srcdir)/$$f $(udevdir)/$$f; \
+	   $(INSTALL_DATA) -c -D $(srcdir)/$$f $(udevdir)/$$f; \
 	done
 
 install: install_lvm2 install_device-mapper
Signature · 1 line
1.5.4.3
#51 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 2 lines · Fabio M. Di Nitto

remove owner and group definitions and install anything writable as mendated by the debian policy

--with-user= and --with-group= should already handle the owner and group (i.e. set them to blank like we do in RHEL/Fedora)

The hard-coded perms in the makefiles probably need treating in a similar fashion.

Alasdair

#52 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 5 lines · Alasdair G Kergon
Quoted 2 lines · Fabio M. Di Nitto

remove owner and group definitions and install anything writable as mendated by the debian policy

--with-user= and --with-group= should already handle the owner and group (i.e. set them to blank like we do in RHEL/Fedora)

So noone uses that functionality. Why is it there and why is it used by default? Every system where the defaults used by install (maybe *BSD) have there own mechanism to solve this.

Bastian

Signature · 2 lines
Conquest is easy. Control is not.
		-- Kirk, "Mirror, Mirror", stardate unknown
#53 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 2 lines · Bastian Blank

why is it used by
default?

I'm open to changing that default.

Alasdair

#54 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 5 lines · Alasdair G Kergon
Quoted 2 lines · Bastian Blank

why is it used by
default?

I'm open to changing that default.

Let's split this patch in 2:

1) OWNER/GROUP defaults, I'd say we should change the default to "".

Both Fedora and Debian drops it, all packaging systems I have seen around expect to build/install as user and then convert to proper owner/group as mandate by distribution policies.
I don't see a real good reason to force a default, and as upstream we would still allow override via configure options.

2) install permissions.

I have more or less the same comment as #1. All distribution packaging do a permission cleanup after "make install" by forcing distribution policies.

Unless there is a real security concern behind it, I think we can delegate such tasks to install and packagers.

The Debian defaults seem sane enough to me, but I am open to suggestions.

Making it a configurable set of options seems a bit overkilling.

Fabio

#55 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 1 line · Fabio M. Di Nitto

1) OWNER/GROUP defaults, I'd say we should change the default to "".

Yes.

Quoted 10 lines

2) install permissions.

I have more or less the same comment as #1. All distribution packaging do a permission cleanup after "make install" by forcing distribution policies.

Unless there is a real security concern behind it, I think we can delegate such tasks to install and packagers.

Making it a configurable set of options seems a bit overkilling.

If I run this outside a packaging framework I still need the permissions to be set.

So then leave this unchanged, and there is no point in Debian or any other packaging system removing the existing permission setting because the packaging system is going to override it afterwards anyway.

Alasdair

#56 lore raw

PATCH 11/11 debian: drop concept of OWNER/GROUP and let install do it's job for permissions

Quoted 2 lines · Alasdair G Kergon

If I run this outside a packaging framework I still need the permissions to be set.

Hu? All it changes is to make them writable for the user. Having unwritable files does not really make fun.

Quoted 3 lines

So then leave this unchanged, and there is no point in Debian or any other packaging system removing the existing permission setting because the packaging system is going to override it afterwards anyway.

It still is, because it breaks my assumptions of the development process to have unwritable files en masse.

Bastian

Signature · 3 lines
Each kiss is as the first.
		-- Miramanee, Kirk's wife, "The Paradise Syndrome",
		   stardate 4842.6