Discussion:
Unable to build aufs-util on aufs4.1 branch due to recent Makefile change
OmegaPhil
2016-10-02 20:14:55 UTC
Permalink
I'm currently updating aufs-util in the usual way (looks like aufs4.1 is
the latest branch), building succeeds but make install fails with the
following:

===================================================================

make[1]: Entering directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
Makefile:61: LibAuDir=
install -dv
install: missing file operand
Try 'install --help' for more information.
Makefile:61: recipe for target 'install_ulib' failed
make[1]: *** [install_ulib] Error 1
make[1]: Leaving directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
Makefile:187: recipe for target 'install_ulib' failed
make: *** [install_ulib] Error 2

===================================================================

Looking at the main directory Makefile, the problem appears to have been
introduced in the recent C library building changes - if I hardcode
LibAuDir beneath the 'ifneq (${CC},${HOSTCC})' block:

====================

LibAuDir = /usr/lib

====================

It works. I don't have much makefile experience so I haven't gone
further. For reference the make install command used:

===================================================================

sudo env CPPFLAGS="-I /usr/include -I
/home/omega/files/source_code/kernel/aufs/aufs4-standalone/usr/include"
make install

===================================================================

I had to include CPPFLAGS this time since the install step actually does
a compilation (??).

Building against kernel v4.7.5-1 on a Devuan Testing (very similar to
Debian Testing) machine.
s***@users.sourceforge.net
2016-10-02 20:36:18 UTC
Permalink
Hello OmegaPhil,
Thanx for reporting.
Post by OmegaPhil
I'm currently updating aufs-util in the usual way (looks like aufs4.1 is
the latest branch), building succeeds but make install fails with the
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
make[1]: Entering directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
Makefile:61: LibAuDir=3D
:::

Hmm, what is your 'make' command line?
Did you set CC?

aufs-util.git expects LibAuDir to be specified whenever CC is given.
For example,

$ make -s BuildFHSM=yes CPPFLAGS=-I.../aufs4-linux/usr/include all
or
$ make -s -C compat32 BuildFHSM=yes CC="cc -m32" LibAuDir=/usr/lib/i386-linux-gnu CPPFLAGS=-I.../aufs4-linux/usr/include all

If you specify CC but LibAuDir, then such bug may appear I am afraid.


J. R. Okajima
OmegaPhil
2016-10-04 20:28:52 UTC
Permalink
Post by s***@users.sourceforge.net
Hello OmegaPhil,
Thanx for reporting.
Post by OmegaPhil
I'm currently updating aufs-util in the usual way (looks like aufs4.1 is
the latest branch), building succeeds but make install fails with the
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
make[1]: Entering directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
Makefile:61: LibAuDir=3D
Hmm, what is your 'make' command line?
Did you set CC?
CC is unset, here is the original make command:

==================================================================

CPPFLAGS="-I /usr/include -I
/home/omega/files/source_code/kernel/aufs/aufs4-standalone/usr/include"
make -j 6

==================================================================
Post by s***@users.sourceforge.net
aufs-util.git expects LibAuDir to be specified whenever CC is given.
For example,
$ make -s BuildFHSM=yes CPPFLAGS=-I.../aufs4-linux/usr/include all
or
$ make -s -C compat32 BuildFHSM=yes CC="cc -m32" LibAuDir=/usr/lib/i386-linux-gnu CPPFLAGS=-I.../aufs4-linux/usr/include all
If you specify CC but LibAuDir, then such bug may appear I am afraid.
J. R. Okajima
I can confirm adding 'LibAuDir=/usr/lib' to the make invocation (e.g.
before -j) as a variable assignment allows the build to proceed without
hacking the Makefile.

Note that currently even 'make clean' causes compilation to occur...
s***@users.sourceforge.net
2016-10-05 04:24:06 UTC
Permalink
Ok, thanx.
I will apply this and release next Monday.

$ .../aufs-util.git$aufs4.1$ git diff -w
diff --git a/Makefile b/Makefile
index 72b68ee..313b62b 100644
--- a/Makefile
+++ b/Makefile
@@ -65,9 +65,7 @@ endef
$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})

-ifneq (${CC},${HOSTCC})
ifeq (${LibAuDir},)
-$(warning Warning: CC is set, but LibAuDir.)
LibAuDir = $(shell ldconfig -p | \
fgrep libc. | \
head -n 1 | \
@@ -77,7 +75,6 @@ $(warning Warning: CC is set, but LibAuDir.)
LibAuDir = /usr/lib
endif
endif
-endif

ExtlibPath = extlib/glibc
ExtlibObj = au_nftw.o


Hmm, I amy remove "LibAuDir = $(shell ldconfig -p ...)" line and set
"/usr/lib" unconditionally after a little thinking.
Post by OmegaPhil
Note that currently even 'make clean' causes compilation to occur...
That is really really strange.
Would you post more details?


J. R. Okajima
OmegaPhil
2016-10-05 16:09:09 UTC
Permalink
Post by s***@users.sourceforge.net
Ok, thanx.
I will apply this and release next Monday.
$ .../aufs-util.git$aufs4.1$ git diff -w
diff --git a/Makefile b/Makefile
index 72b68ee..313b62b 100644
--- a/Makefile
+++ b/Makefile
@@ -65,9 +65,7 @@ endef
$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
-ifneq (${CC},${HOSTCC})
ifeq (${LibAuDir},)
-$(warning Warning: CC is set, but LibAuDir.)
LibAuDir = $(shell ldconfig -p | \
fgrep libc. | \
head -n 1 | \
@@ -77,7 +75,6 @@ $(warning Warning: CC is set, but LibAuDir.)
LibAuDir = /usr/lib
endif
endif
-endif
ExtlibPath = extlib/glibc
ExtlibObj = au_nftw.o
Hmm, I amy remove "LibAuDir = $(shell ldconfig -p ...)" line and set
"/usr/lib" unconditionally after a little thinking.
Post by OmegaPhil
Note that currently even 'make clean' causes compilation to occur...
That is really really strange.
Would you post more details?
J. R. Okajima
Here is the situation on a dumb 'make clean' after a successful build
(i.e. I'm not configuring an appropriate environment for any aufs stuff
to build by not hacking the includes etc etc):

====================================================================

$ make clean

ver.c:19:29: fatal error: linux/aufs_type.h: No such file or directory
#include <linux/aufs_type.h>
^
compilation terminated.
rm -f aufs.5 auibusy aumvdown auplink mount.aufs umount.aufs
etc_default_aufs libautil.a libau.so* *~
rm -f auibusy.o aumvdown.o auplink.o mount.aufs.o umount.aufs.o perror.o
proc_mnt.o br.o plink.o mtab.o au_nftw.o au_decode_mntpnt.o error_at_line.o
for i in au_nftw.c au_decode_mntpnt.c error_at_line.c; \
do test -L ${i} && rm -f ${i} || :; \
done
make CC="cc" CPPFLAGS="-D_GNU_SOURCE -I./libau
-DAUFHSM_CMD=\"/usr/bin/aufhsm\" -DMOUNT_CMD=\"/bin/mount\"
-DUMOUNT_CMD=\"/bin/umount\"
-I/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/extlib/non-glibc
-DAUPLINK_FTW_CMD=\"/sbin/auplink_ftw\"" CFLAGS="-O -Wall"
INSTALL="install" Install="install -o root -g root -p"
ManDir="/usr/share/man"
TopDir="/home/omega/files/source_code/kernel/aufs/aufs-aufs-util"
LibUtilHdr="au_util.h extlib/non-glibc/error_at_line.h" Glibc="no"
LibAuDir="" ExtlibPath="extlib/non-glibc" -C libau clean
make[1]: Entering directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
rm -f libau.so libau.so.* libau.o rdu_lib.o rdu.o pathconf.o
error_at_line.o *~
for i in rdu64.c error_at_line.[ch]; \
do test -L ${i} && rm -f ${i} || :; \
done
make[1]: Leaving directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
# empty

====================================================================

Reading the top of the main Makefile, I'm assuming that test_glibc is
doing a compilation that always happens regardless of the make target
requested.
s***@users.sourceforge.net
2016-10-05 16:52:55 UTC
Permalink
Post by OmegaPhil
Reading the top of the main Makefile, I'm assuming that test_glibc is
doing a compilation that always happens regardless of the make target
requested.
You are right.
Although I don't think it a big problem, here is a patch for you.


J. R. Okajima

commit 61ba8e01f1960b43e8ff3b7f0b7ef894562b7703
Author: J. R. Okajima <***@gmail.com>
Date: Thu Oct 6 01:50:14 2016 +0900

bugfix, stop evaluating test_glibc when 'make clean'

Reported-by: OmegaPhil <***@startmail.com>
See-also: http://www.mail-archive.com/aufs-***@lists.sourceforge.net/msg05400.html
Signed-off-by: J. R. Okajima <***@gmail.com>

diff --git a/Makefile b/Makefile
index fa96b6c..f624d3c 100644
--- a/Makefile
+++ b/Makefile
@@ -62,7 +62,8 @@ define test_glibc
echo yes || \
echo no)
endef
-$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
+$(filter-out clean, ${MAKECMDGOALS}): \
+ $(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})

LibAuDir ?= /usr/lib
OmegaPhil
2016-10-05 18:28:43 UTC
Permalink
Post by s***@users.sourceforge.net
Post by OmegaPhil
Reading the top of the main Makefile, I'm assuming that test_glibc is
doing a compilation that always happens regardless of the make target
requested.
You are right.
Although I don't think it a big problem, here is a patch for you.
J. R. Okajima
commit 61ba8e01f1960b43e8ff3b7f0b7ef894562b7703
Date: Thu Oct 6 01:50:14 2016 +0900
bugfix, stop evaluating test_glibc when 'make clean'
diff --git a/Makefile b/Makefile
index fa96b6c..f624d3c 100644
--- a/Makefile
+++ b/Makefile
@@ -62,7 +62,8 @@ define test_glibc
echo yes || \
echo no)
endef
-$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
+$(filter-out clean, ${MAKECMDGOALS}): \
+ $(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
LibAuDir ?= /usr/lib
Cheers - as you're changing that bit now, would also make sense to not
compile during the install target as well (can't think of any other
targets a normal user would use).
OmegaPhil
2016-10-05 18:33:24 UTC
Permalink
Post by OmegaPhil
Post by s***@users.sourceforge.net
Post by OmegaPhil
Reading the top of the main Makefile, I'm assuming that test_glibc is
doing a compilation that always happens regardless of the make target
requested.
You are right.
Although I don't think it a big problem, here is a patch for you.
J. R. Okajima
commit 61ba8e01f1960b43e8ff3b7f0b7ef894562b7703
Date: Thu Oct 6 01:50:14 2016 +0900
bugfix, stop evaluating test_glibc when 'make clean'
diff --git a/Makefile b/Makefile
index fa96b6c..f624d3c 100644
--- a/Makefile
+++ b/Makefile
@@ -62,7 +62,8 @@ define test_glibc
echo yes || \
echo no)
endef
-$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
+$(filter-out clean, ${MAKECMDGOALS}): \
+ $(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
LibAuDir ?= /usr/lib
Cheers - as you're changing that bit now, would also make sense to not
compile during the install target as well (can't think of any other
targets a normal user would use).
Ah sorry, this is precisely when you want it... sorry I'm used to
autotools systems and this kind of thing being ./configured before
everything.
s***@users.sourceforge.net
2016-10-05 19:01:06 UTC
Permalink
Post by OmegaPhil
Post by OmegaPhil
Cheers - as you're changing that bit now, would also make sense to not
compile during the install target as well (can't think of any other
targets a normal user would use).
Ah sorry, this is precisely when you want it... sorry I'm used to
autotools systems and this kind of thing being ./configured before
everything.
I agree that ./configure is the best solution for such case. Why didn't
I do that? It's just because I'm lazy. :-)

About filtering-out 'make install', I won't do that since people may run
'make install' without running 'make all' (I've just add 'all' to the
dependency of 'install').

And here is the updated commit.


J. R. Okajima


commit 3557ce3589462600d8f6888160f735673ff8a8af
Author: J. R. Okajima <***@gmail.com>
Date: Thu Oct 6 01:50:14 2016 +0900

bugfix, stop evaluating test_glibc when 'make clean'

Reported-by: OmegaPhil <***@startmail.com>
See-also: http://www.mail-archive.com/aufs-***@lists.sourceforge.net/msg05400.html
Signed-off-by: J. R. Okajima <***@gmail.com>

diff --git a/Makefile b/Makefile
index fa96b6c..f90540e 100644
--- a/Makefile
+++ b/Makefile
@@ -54,23 +54,30 @@ LibUtilObj += perror.o proc_mnt.o br.o plink.o mtab.o
LibUtilHdr = au_util.h

TopDir = ${CURDIR}
+LastTestGlibc = test_glibc
# don't use -q for fgrep here since it exits when the string is found,
# and it causes the broken pipe error.
define test_glibc
- $(shell ${1} ${CPPFLAGS} -I ${TopDir}/extlib/non-glibc -E -P -dM ${2} |\
+ $(shell cat ${LastTestGlibc} 2> /dev/null || \
+ { ${1} ${CPPFLAGS} -I ${TopDir}/extlib/non-glibc -E -P -dM ${2} |\
fgrep -w __GNU_LIBRARY__ > /dev/null && \
echo yes || \
- echo no)
+ echo no; } |\
+ tee ${LastTestGlibc})
endef
-$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
+$(filter-out clean, ${MAKECMDGOALS}): \
+ $(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})

LibAuDir ?= /usr/lib
+ExtlibGlibcObj = au_nftw.o
+ExtlibNonGlibcObj = ${ExtlibGlibcObj} au_decode_mntpnt.o error_at_line.o
+
ExtlibPath = extlib/glibc
-ExtlibObj = au_nftw.o
+ExtlibObj = ${ExtlibGlibcObj}
ifeq (${Glibc},no)
ExtlibPath = extlib/non-glibc
-ExtlibObj += au_decode_mntpnt.o error_at_line.o
+ExtlibObj = ${ExtlibNonGlibcObj}
LibUtilHdr += ${ExtlibPath}/error_at_line.h
override CPPFLAGS += -I${CURDIR}/${ExtlibPath}
endif
@@ -99,9 +106,9 @@ all: ver_test ${Man} ${Bin} ${Etc}
$(call MakeFHSM, $@)

clean:
- ${RM} ${Man} ${Bin} ${Etc} ${LibUtil} libau.so* *~
+ ${RM} ${Man} ${Bin} ${Etc} ${LibUtil} libau.so* ${LastTestGlibc} *~
${RM} ${BinObj} ${LibUtilObj}
- for i in ${ExtlibSrc}; \
+ for i in $(patsubst %.o,%.c, ${ExtlibGlibcObj} ${ExtlibNonGlibcObj}); \
do test -L $${i} && ${RM} $${i} || :; \
done
${MAKE} -C libau $@
OmegaPhil
2016-10-05 20:44:07 UTC
Permalink
Post by s***@users.sourceforge.net
Post by OmegaPhil
Post by OmegaPhil
Cheers - as you're changing that bit now, would also make sense to not
compile during the install target as well (can't think of any other
targets a normal user would use).
Ah sorry, this is precisely when you want it... sorry I'm used to
autotools systems and this kind of thing being ./configured before
everything.
I agree that ./configure is the best solution for such case. Why didn't
I do that? It's just because I'm lazy. :-)
About filtering-out 'make install', I won't do that since people may run
'make install' without running 'make all' (I've just add 'all' to the
dependency of 'install').
And here is the updated commit.
J. R. Okajima
commit 3557ce3589462600d8f6888160f735673ff8a8af
Date: Thu Oct 6 01:50:14 2016 +0900
bugfix, stop evaluating test_glibc when 'make clean'
diff --git a/Makefile b/Makefile
index fa96b6c..f90540e 100644
--- a/Makefile
+++ b/Makefile
@@ -54,23 +54,30 @@ LibUtilObj += perror.o proc_mnt.o br.o plink.o mtab.o
LibUtilHdr = au_util.h
TopDir = ${CURDIR}
+LastTestGlibc = test_glibc
# don't use -q for fgrep here since it exits when the string is found,
# and it causes the broken pipe error.
define test_glibc
- $(shell ${1} ${CPPFLAGS} -I ${TopDir}/extlib/non-glibc -E -P -dM ${2} |\
+ $(shell cat ${LastTestGlibc} 2> /dev/null || \
+ { ${1} ${CPPFLAGS} -I ${TopDir}/extlib/non-glibc -E -P -dM ${2} |\
fgrep -w __GNU_LIBRARY__ > /dev/null && \
echo yes || \
- echo no)
+ echo no; } |\
+ tee ${LastTestGlibc})
endef
-$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
+$(filter-out clean, ${MAKECMDGOALS}): \
+ $(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
LibAuDir ?= /usr/lib
+ExtlibGlibcObj = au_nftw.o
+ExtlibNonGlibcObj = ${ExtlibGlibcObj} au_decode_mntpnt.o error_at_line.o
+
ExtlibPath = extlib/glibc
-ExtlibObj = au_nftw.o
+ExtlibObj = ${ExtlibGlibcObj}
ifeq (${Glibc},no)
ExtlibPath = extlib/non-glibc
-ExtlibObj += au_decode_mntpnt.o error_at_line.o
+ExtlibObj = ${ExtlibNonGlibcObj}
LibUtilHdr += ${ExtlibPath}/error_at_line.h
override CPPFLAGS += -I${CURDIR}/${ExtlibPath}
endif
@@ -99,9 +106,9 @@ all: ver_test ${Man} ${Bin} ${Etc}
- ${RM} ${Man} ${Bin} ${Etc} ${LibUtil} libau.so* *~
+ ${RM} ${Man} ${Bin} ${Etc} ${LibUtil} libau.so* ${LastTestGlibc} *~
${RM} ${BinObj} ${LibUtilObj}
- for i in ${ExtlibSrc}; \
+ for i in $(patsubst %.o,%.c, ${ExtlibGlibcObj} ${ExtlibNonGlibcObj}); \
do test -L $${i} && ${RM} $${i} || :; \
done
Sorry, I've put this email through patch, and then manually copied out
the text and tried again, but the patch doesn't apply. Is this supposed
to go on top of aufs4.1?

I already have a buildable aufs-util environment so I'm happy anyway,
but happy to test all the same.
s***@users.sourceforge.net
2016-10-05 20:56:01 UTC
Permalink
Post by OmegaPhil
Sorry, I've put this email through patch, and then manually copied out
the text and tried again, but the patch doesn't apply. Is this supposed
to go on top of aufs4.1?
Sorry. There is another patch before that.
My current aufs4.1 looks like this.

95f0af1 2016-10-06 make sure all are built before installing
a74dda0 2016-10-06 tiny, silence a garbage
3557ce3 2016-10-06 bugfix, stop evaluating test_glibc when 'make clean'
4647760 2016-10-06 bugfix, simplify LibAuDir variable
1bbc953 2016-10-06 version aufs4.1
92d857b 2016-10-01 aufs-util 20161003
4d295ae 2016-10-01 possible bugfix, auplink_ftw, prohibit 'touch' to create a new file
:::

The in my previous mail is in above list.
3557ce3 2016-10-06 bugfix, stop evaluating test_glibc when 'make clean'

Just before that, there is
4647760 2016-10-06 bugfix, simplify LibAuDir variable

And without 4647760, 3557ce3 won't be applied, sorry.
Here is 4647760.


J. R. Okajima

commit 4647760ff509de9d85c492f38492e8984770c2f7
Author: J. R. Okajima <***@gmail.com>
Date: Wed Oct 5 20:58:37 2016 +0900

bugfix, simplify LibAuDir variable

Simply set the default dir.

Reported-by: OmegaPhil <***@startmail.com>
See-also: http://www.mail-archive.com/aufs-***@lists.sourceforge.net/msg05395.html
Signed-off-by: J. R. Okajima <***@gmail.com>

diff --git a/Makefile b/Makefile
index 72b68ee..fa96b6c 100644
--- a/Makefile
+++ b/Makefile
@@ -65,20 +65,7 @@ endef
$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})

-ifneq (${CC},${HOSTCC})
- ifeq (${LibAuDir},)
-$(warning Warning: CC is set, but LibAuDir.)
- LibAuDir = $(shell ldconfig -p | \
- fgrep libc. | \
- head -n 1 | \
- cut -f2 -d'>' | \
- xargs -r dirname)
- ifeq (${LibAuDir},)
- LibAuDir = /usr/lib
- endif
- endif
-endif
-
+LibAuDir ?= /usr/lib
ExtlibPath = extlib/glibc
ExtlibObj = au_nftw.o
ifeq (${Glibc},no)
diff --git a/README b/README
index 50b44f8..0f533ae 100644
--- a/README
+++ b/README
@@ -66,10 +66,9 @@ Makefile in this tree has some customizable make-variables.

- LibAuDir
specify the directory path to install the shared object library
- libau.so. The default is taken from the output of ldconfig -p. If it
- fails, then set to /usr/lib.
- This make variable will be useful for the system who have both of
- glibc and musl-libc.
+ libau.so. The default is /usr/lib.
+ This make variable will be useful for the system who has both of
+ glibc and musl-libc, or the system who has x86_64 and i686 libraries.

- MountCmd, UmountCmd
specify mount(8) and umount(8) in full path. By default, they are
OmegaPhil
2016-10-06 15:17:35 UTC
Permalink
Post by s***@users.sourceforge.net
Post by OmegaPhil
Sorry, I've put this email through patch, and then manually copied out
the text and tried again, but the patch doesn't apply. Is this supposed
to go on top of aufs4.1?
Sorry. There is another patch before that.
My current aufs4.1 looks like this.
95f0af1 2016-10-06 make sure all are built before installing
a74dda0 2016-10-06 tiny, silence a garbage
3557ce3 2016-10-06 bugfix, stop evaluating test_glibc when 'make clean'
4647760 2016-10-06 bugfix, simplify LibAuDir variable
1bbc953 2016-10-06 version aufs4.1
92d857b 2016-10-01 aufs-util 20161003
4d295ae 2016-10-01 possible bugfix, auplink_ftw, prohibit 'touch' to create a new file
The in my previous mail is in above list.
3557ce3 2016-10-06 bugfix, stop evaluating test_glibc when 'make clean'
Just before that, there is
4647760 2016-10-06 bugfix, simplify LibAuDir variable
And without 4647760, 3557ce3 won't be applied, sorry.
Here is 4647760.
J. R. Okajima
OK, applying 4647760 then 3557ce3 on top of 92d857b results in the
following build failure (from the usual make all call from the top
directory):

========================================================================

make CC="cc" CPPFLAGS="-I /usr/include -I
/home/omega/files/source_code/kernel/aufs/aufs4-standalone/usr/include
-D_GNU_SOURCE -I./libau -DAUFHSM_CMD=\"/usr/bin/aufhsm\"
-DMOUNT_CMD=\"/bin/mount\" -DUMOUNT_CMD=\"/bin/umount\"" CFLAGS="-O
-Wall" INSTALL="install" Install="install -o root -g root -p"
ManDir="/usr/share/man"
TopDir="/home/omega/files/source_code/kernel/aufs/aufs-aufs-util"
LibUtilHdr="au_util.h" Glibc="" LibAuDir="/usr/lib"
ExtlibPath="extlib/glibc" -C libau all
make[1]: Entering directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
make[1]: *** No rule to make target '../extlib/glibc/error_at_line.h',
needed by 'libau.o'. Stop.
make[1]: Leaving directory
'/home/omega/files/source_code/kernel/aufs/aufs-aufs-util/libau'
Makefile:104: recipe for target 'all' failed
make: *** [all] Error 2
rm ver c2tmac c2sh au_nftw.c

========================================================================

Commits 1bbc953, a74dda0 and 95f0af1 aren't available currently.
s***@users.sourceforge.net
2016-10-06 15:31:19 UTC
Permalink
Post by OmegaPhil
OK, applying 4647760 then 3557ce3 on top of 92d857b results in the
following build failure (from the usual make all call from the top
Thanx for testing.
Would you test this patch?

J. R. Okajima

diff --git a/Makefile b/Makefile
index 298ae05..3838f0e 100644
--- a/Makefile
+++ b/Makefile
@@ -66,7 +66,7 @@ define test_glibc
echo no; } |\
tee ${LastTestGlibc})
endef
-$(filter-out clean, ${MAKECMDGOALS}): \
+$(filter-out clean, ${MAKECMDGOALS} all): \
$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
OmegaPhil
2016-10-06 15:46:37 UTC
Permalink
Post by s***@users.sourceforge.net
Post by OmegaPhil
OK, applying 4647760 then 3557ce3 on top of 92d857b results in the
following build failure (from the usual make all call from the top
Thanx for testing.
Would you test this patch?
J. R. Okajima
diff --git a/Makefile b/Makefile
index 298ae05..3838f0e 100644
--- a/Makefile
+++ b/Makefile
@@ -66,7 +66,7 @@ define test_glibc
echo no; } |\
tee ${LastTestGlibc})
endef
-$(filter-out clean, ${MAKECMDGOALS}): \
+$(filter-out clean, ${MAKECMDGOALS} all): \
$(eval Glibc=$(call test_glibc, ${CC}, ver.c))
#$(warning Glibc=${Glibc})
Great, make all works, and make install triggers no compilation (so the
non-hacked checkinstall command works again).
s***@users.sourceforge.net
2016-10-06 15:52:08 UTC
Permalink
Post by OmegaPhil
Great, make all works, and make install triggers no compilation (so the
non-hacked checkinstall command works again).
Glad to hear that!
Thanx again for your repeated tests.
Those patches will be released on next Monday.


J. R. Okajima

Loading...