* PATCH: Make ld testsuite usable outside build directory
@ 2005-03-24 14:45 Mark Mitchell
2005-03-24 14:52 ` Ian Lance Taylor
2005-03-24 14:54 ` Daniel Jacobowitz
0 siblings, 2 replies; 13+ messages in thread
From: Mark Mitchell @ 2005-03-24 14:45 UTC (permalink / raw)
To: binutils
At present, the ld testsuite does not work on an installed linker due
to two minor problems:
1. The testsuite wants to read tmpdir/libpath.exp, which is generated
by the Makefiles, but obviously not available when testing outside
of the build directory. Since the point of testing an installed
toolchain is to see if it works by itself, this list of explict -L
options is unncessary in that situation.
2. Several test scripts check [which $CC] to see if there's a compiler
available. That depends on the fact that "make check" passes in
"CC=$(CC_FOR_TARGET)". Users who run with "runtest" directly won't
have CC set; this should be treated just like the case where there
is no available compiler.
(I think a further improvement would be to actually look for a
compiler, using [transform gcc]. If people agree that this is a
good idea, I can add that.)
Tested with an arm-linux toolchain. OK?
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
2005-03-23 Mark Mitchell <mark@codesourcery.com>
* config/default.exp: Do not load libpath.exp if it does not
exist.
* ld-elfcomm/elfcomm.exp: Do not use CC if it does not exist.
* ld-scripts/crossref.exp: Likewise.
* ld-srec/srec.exp: Likewise.
* ld-undefined/undefined.exp: Likewise.
Index: config/default.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/config/default.exp,v
retrieving revision 1.1.5.1
diff -c -5 -p -r1.1.5.1 default.exp
*** config/default.exp 26 Nov 2003 04:18:08 -0000 1.1.5.1
--- config/default.exp 24 Mar 2005 03:36:57 -0000
*************** if {![file isdirectory tmpdir/ld]} then
*** 68,81 ****
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
--- 68,83 ----
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! if {[file exists tmpdir/libpath.exp]} {
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
! }
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
Index: ld-elfcomm/elfcomm.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/ld-elfcomm/elfcomm.exp,v
retrieving revision 1.1.5.1
diff -c -5 -p -r1.1.5.1 elfcomm.exp
*** ld-elfcomm/elfcomm.exp 26 Nov 2003 04:18:09 -0000 1.1.5.1
--- ld-elfcomm/elfcomm.exp 24 Mar 2005 03:36:57 -0000
*************** set test1 "size/aligment change of commo
*** 29,39 ****
set test1w1 "$test1 (warning 1)"
set test1w2 "$test1 (warning 2)"
set test1c1 "$test1 (change 1)"
set test1c2 "$test1 (change 2)"
! if { [which $CC] == 0 } {
untested $test1w1
untested $test1w2
untested $test1c1
untested $test1c2
return
--- 29,39 ----
set test1w1 "$test1 (warning 1)"
set test1w2 "$test1 (warning 2)"
set test1c1 "$test1 (change 1)"
set test1c2 "$test1 (change 2)"
! if { ![info exists CC] || [which $CC] == 0 } {
untested $test1w1
untested $test1w2
untested $test1c1
untested $test1c2
return
Index: ld-scripts/crossref.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/ld-scripts/crossref.exp,v
retrieving revision 1.1.5.2
diff -c -5 -p -r1.1.5.2 crossref.exp
*** ld-scripts/crossref.exp 17 Jan 2005 03:28:41 -0000 1.1.5.2
--- ld-scripts/crossref.exp 24 Mar 2005 03:36:57 -0000
***************
*** 18,28 ****
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
set test1 "NOCROSSREFS 1"
set test2 "NOCROSSREFS 2"
! if { [which $CC] == 0 } {
untested $test1
untested $test2
return
}
--- 18,28 ----
# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
set test1 "NOCROSSREFS 1"
set test2 "NOCROSSREFS 2"
! if { ![info exists CC] || [which $CC] == 0 } {
untested $test1
untested $test2
return
}
Index: ld-srec/srec.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/ld-srec/srec.exp,v
retrieving revision 1.1.5.2
diff -c -5 -p -r1.1.5.2 srec.exp
*** ld-srec/srec.exp 17 Jan 2005 03:28:41 -0000 1.1.5.2
--- ld-srec/srec.exp 24 Mar 2005 03:36:57 -0000
*************** set sizeof_headers [string match "*SIZEO
*** 332,342 ****
# First test linking a C program. We don't require any libraries. We
# link it normally, and objcopy to the S-record format, and then link
# directly to the S-record format, and require that the two files
# contain the same data.
! if { [which $CC] == 0 } {
untested $test1
untested $test2
return
}
--- 332,342 ----
# First test linking a C program. We don't require any libraries. We
# link it normally, and objcopy to the S-record format, and then link
# directly to the S-record format, and require that the two files
# contain the same data.
! if { ![info exists CC] || [which $CC] == 0 } {
untested $test1
untested $test2
return
}
Index: ld-undefined/undefined.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/ld-undefined/undefined.exp,v
retrieving revision 1.1.5.2
diff -c -5 -p -r1.1.5.2 undefined.exp
*** ld-undefined/undefined.exp 17 Jan 2005 03:28:41 -0000 1.1.5.2
--- ld-undefined/undefined.exp 24 Mar 2005 03:36:57 -0000
***************
*** 20,30 ****
set testund "undefined"
set testfn "undefined function"
set testline "undefined line"
! if { [which $CC] == 0 } {
verbose "Could not find C compiler!" 1
untested $testund
untested $testfn
untested $testline
return
--- 20,30 ----
set testund "undefined"
set testfn "undefined function"
set testline "undefined line"
! if { ![info exists CC] || [which $CC] == 0 } {
verbose "Could not find C compiler!" 1
untested $testund
untested $testfn
untested $testline
return
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 14:45 PATCH: Make ld testsuite usable outside build directory Mark Mitchell
@ 2005-03-24 14:52 ` Ian Lance Taylor
2005-03-24 14:56 ` Mark Mitchell
2005-03-24 14:54 ` Daniel Jacobowitz
1 sibling, 1 reply; 13+ messages in thread
From: Ian Lance Taylor @ 2005-03-24 14:52 UTC (permalink / raw)
To: mark; +Cc: binutils
Mark Mitchell <mark@codesourcery.com> writes:
> (I think a further improvement would be to actually look for a
> compiler, using [transform gcc]. If people agree that this is a
> good idea, I can add that.)
Seems appropriate.
> 2005-03-23 Mark Mitchell <mark@codesourcery.com>
>
> * config/default.exp: Do not load libpath.exp if it does not
> exist.
> * ld-elfcomm/elfcomm.exp: Do not use CC if it does not exist.
> * ld-scripts/crossref.exp: Likewise.
> * ld-srec/srec.exp: Likewise.
> * ld-undefined/undefined.exp: Likewise.
OK.
Ian
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 14:45 PATCH: Make ld testsuite usable outside build directory Mark Mitchell
2005-03-24 14:52 ` Ian Lance Taylor
@ 2005-03-24 14:54 ` Daniel Jacobowitz
1 sibling, 0 replies; 13+ messages in thread
From: Daniel Jacobowitz @ 2005-03-24 14:54 UTC (permalink / raw)
To: binutils
On Wed, Mar 23, 2005 at 07:44:19PM -0800, Mark Mitchell wrote:
> (I think a further improvement would be to actually look for a
> compiler, using [transform gcc]. If people agree that this is a
> good idea, I can add that.)
[find_gcc]?
--
Daniel Jacobowitz
CodeSourcery, LLC
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 14:52 ` Ian Lance Taylor
@ 2005-03-24 14:56 ` Mark Mitchell
2005-03-24 19:34 ` Daniel Jacobowitz
0 siblings, 1 reply; 13+ messages in thread
From: Mark Mitchell @ 2005-03-24 14:56 UTC (permalink / raw)
To: Ian Lance Taylor; +Cc: binutils
[-- Attachment #1: Type: text/plain, Size: 750 bytes --]
Ian Lance Taylor wrote:
> Mark Mitchell <mark@codesourcery.com> writes:
>
>
>> (I think a further improvement would be to actually look for a
>> compiler, using [transform gcc]. If people agree that this is a
>> good idea, I can add that.)
>
>
> Seems appropriate.
Thanks for the speedy review!
In that case, we can go with this simpler patch, which I tested by
running both in and out of the build directory for an arm-linux
toolchain, with extra debugging output that verified that CC was getting
the value provided by "make check" when in the build directory.
If this is approved, I'll not check in the other patch, as this one
makes it unnecessary.
OK?
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
[-- Attachment #2: v3.patch --]
[-- Type: text/plain, Size: 3461 bytes --]
2005-03-23 Mark Mitchell <mark@codesourcery.com>
* testsuite/libstdc++-dg/normal.exp: Read testsuite_files, if it
exists.
Index: libstdc++-dg/normal.exp
===================================================================
RCS file: /cvs/gcc/gcc/libstdc++-v3/testsuite/libstdc++-dg/normal.exp,v
retrieving revision 1.6
diff -c -5 -p -r1.6 normal.exp
*** libstdc++-dg/normal.exp 23 Mar 2005 21:46:07 -0000 1.6
--- libstdc++-dg/normal.exp 24 Mar 2005 03:22:06 -0000
***************
*** 22,61 ****
dg-init
# Build the support objects.
v3-build_support
- # Find directories that might have tests.
- set subdirs [glob "$srcdir/\[0-9\]\[0-9\]*"]
- foreach d [glob "$srcdir/\[a-z\]*"] {
- if {[file isdirectory $d]} {
- lappend subdirs $d
- }
- }
-
- # Find all the tests.
set tests [list]
! foreach s $subdirs {
! set subdir_tests [find $s *.cc]
! # Filter out tests that should not be run.
! foreach t $subdir_tests {
! # The DejaGNU "find" procedure sometimes returns a list
! # containing an empty string, when it should realy return
! # an empty list.
! if { $t == "" } {
! continue
}
! # Filter out:
! # 1. interactive tests.
! # 2. performance tests.
! # 3. wchar_t tests, if not supported.
! # 4. thread tests, if not supported.
! if { [string first _xin $t] == -1
! && [string first performance $t] == -1
! && (${v3-wchar_t} || [string first wchar_t $t] == -1)
! && (${v3-threads} || [string first thread $t] == -1) } {
! lappend tests $t
}
}
}
set tests [lsort $tests]
--- 22,87 ----
dg-init
# Build the support objects.
v3-build_support
set tests [list]
!
! # If there is a "testsuite_files" file, use it.
! #
! # This is a workaround for problems reported with using:
! #
! # runtest normal.exp="`cat testsuite_files`"
! #
! # See:
! # http://gcc.gnu.org/ml/libstdc++/2005-03/msg00278.html
! # for discussion of the problem.
! #
! # If that worked consistently, we could modify "make check" to
! # pass that option, and then remove this code.
! if {[info exists blddir]} {
! set tests_file "${blddir}/testsuite/testsuite_files"
! }
! if {[info exists tests_file] && [file exists $tests_file]} {
! set f [open $tests_file]
! while { ! [eof $f] } {
! set t [gets $f]
! if { [string length "$t"] != 0 } {
! lappend tests ${srcdir}/${t}
! }
! }
! close $f
! } else {
! # Find directories that might have tests.
! set subdirs [glob "$srcdir/\[0-9\]\[0-9\]*"]
! foreach d [glob "$srcdir/\[a-z\]*"] {
! if {[file isdirectory $d]} {
! lappend subdirs $d
}
! }
! # Find all the tests.
! foreach s $subdirs {
! set subdir_tests [find $s *.cc]
! # Filter out tests that should not be run.
! foreach t $subdir_tests {
! # The DejaGNU "find" procedure sometimes returns a list
! # containing an empty string, when it should really return
! # an empty list.
! if { $t == "" } {
! continue
! }
! # Filter out:
! # 1. interactive tests.
! # 2. performance tests.
! # 3. wchar_t tests, if not supported.
! # 4. thread tests, if not supported.
! if { [string first _xin $t] == -1
! && [string first performance $t] == -1
! && (${v3-wchar_t} || [string first wchar_t $t] == -1)
! && (${v3-threads} || [string first thread $t] == -1) } {
! lappend tests $t
! }
}
}
}
set tests [lsort $tests]
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 14:56 ` Mark Mitchell
@ 2005-03-24 19:34 ` Daniel Jacobowitz
2005-03-24 20:39 ` Mark Mitchell
0 siblings, 1 reply; 13+ messages in thread
From: Daniel Jacobowitz @ 2005-03-24 19:34 UTC (permalink / raw)
To: Mark Mitchell; +Cc: Ian Lance Taylor, binutils
On Wed, Mar 23, 2005 at 10:39:45PM -0800, Mark Mitchell wrote:
> Ian Lance Taylor wrote:
> >Mark Mitchell <mark@codesourcery.com> writes:
> >
> >
> >> (I think a further improvement would be to actually look for a
> >> compiler, using [transform gcc]. If people agree that this is a
> >> good idea, I can add that.)
> >
> >
> >Seems appropriate.
>
> Thanks for the speedy review!
>
> In that case, we can go with this simpler patch, which I tested by
> running both in and out of the build directory for an arm-linux
> toolchain, with extra debugging output that verified that CC was getting
> the value provided by "make check" when in the build directory.
>
> If this is approved, I'll not check in the other patch, as this one
> makes it unnecessary.
>
> OK?
> 2005-03-23 Mark Mitchell <mark@codesourcery.com>
>
> * testsuite/libstdc++-dg/normal.exp: Read testsuite_files, if it
> exists.
Wrong patch.
--
Daniel Jacobowitz
CodeSourcery, LLC
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 19:34 ` Daniel Jacobowitz
@ 2005-03-24 20:39 ` Mark Mitchell
2005-03-24 20:51 ` Daniel Jacobowitz
0 siblings, 1 reply; 13+ messages in thread
From: Mark Mitchell @ 2005-03-24 20:39 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: Ian Lance Taylor, binutils
[-- Attachment #1: Type: text/plain, Size: 715 bytes --]
Daniel Jacobowitz wrote:
>>Thanks for the speedy review!
>>
>>In that case, we can go with this simpler patch, which I tested by
>>running both in and out of the build directory for an arm-linux
>>toolchain, with extra debugging output that verified that CC was getting
>>the value provided by "make check" when in the build directory.
>>
>>If this is approved, I'll not check in the other patch, as this one
>>makes it unnecessary.
>>
>>OK?
>
>
>>2005-03-23 Mark Mitchell <mark@codesourcery.com>
>>
>> * testsuite/libstdc++-dg/normal.exp: Read testsuite_files, if it
>> exists.
>
>
> Wrong patch.
Gah! Corrected form attached.
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
[-- Attachment #2: ld.patch --]
[-- Type: text/plain, Size: 2022 bytes --]
2005-02-23 Mark Mitchell <mark@codesourcery.com>
* config/default.exp: Do not load libpath.exp if it does not
exist.
(CC): Provide fallback definition.
(CFLAGS): Likewise.
(CXX): Likewise.
(CXXFLAGS): Likewise.
Index: config/default.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/config/default.exp,v
retrieving revision 1.1.5.1.2.2
diff -c -5 -p -r1.1.5.1.2.2 default.exp
*** config/default.exp 24 Mar 2005 05:36:30 -0000 1.1.5.1.2.2
--- config/default.exp 24 Mar 2005 06:35:18 -0000
*************** if {![file isdirectory tmpdir/ld]} then
*** 68,81 ****
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
--- 68,99 ----
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! if {[file exists tmpdir/libpath.exp]} {
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
! }
! }
!
! # "CC=$(CC_FOR_TARGET)", and other similar parameters. But, if the
! # user invokes runtest directly (as when testing an installed linker),
! # these flags may not b set.
! if {![info exists CC]} {
! set CC [transform gcc]
! }
! if {![info exists CFLAGS]} {
! set CFLAGS "-g -O2"
! }
! if {![info exists CXX]} {
! set CXX [transform g++]
! }
! if {![info exists CXXFLAGS]} {
! set CXXFLAGS ""
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 20:39 ` Mark Mitchell
@ 2005-03-24 20:51 ` Daniel Jacobowitz
2005-03-24 20:59 ` Mark Mitchell
2005-03-24 21:15 ` Mark Mitchell
0 siblings, 2 replies; 13+ messages in thread
From: Daniel Jacobowitz @ 2005-03-24 20:51 UTC (permalink / raw)
To: Mark Mitchell; +Cc: Ian Lance Taylor, binutils
On Thu, Mar 24, 2005 at 06:52:36AM -0800, Mark Mitchell wrote:
> Daniel Jacobowitz wrote:
>
> >>Thanks for the speedy review!
> >>
> >>In that case, we can go with this simpler patch, which I tested by
> >>running both in and out of the build directory for an arm-linux
> >>toolchain, with extra debugging output that verified that CC was getting
> >>the value provided by "make check" when in the build directory.
> >>
> >>If this is approved, I'll not check in the other patch, as this one
> >>makes it unnecessary.
> >>
> >>OK?
> >
> >
> >>2005-03-23 Mark Mitchell <mark@codesourcery.com>
> >>
> >> * testsuite/libstdc++-dg/normal.exp: Read testsuite_files, if it
> >> exists.
> >
> >
> >Wrong patch.
>
> Gah! Corrected form attached.
Did you see my mention of [find_gcc]? Does that work for you? It will
have the added benefit of using the in-tree compiler when CC is unset
in a combined-tree build.
--
Daniel Jacobowitz
CodeSourcery, LLC
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 20:51 ` Daniel Jacobowitz
@ 2005-03-24 20:59 ` Mark Mitchell
2005-03-24 21:15 ` Mark Mitchell
1 sibling, 0 replies; 13+ messages in thread
From: Mark Mitchell @ 2005-03-24 20:59 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: Ian Lance Taylor, binutils
Daniel Jacobowitz wrote:
> Did you see my mention of [find_gcc]? Does that work for you? It will
> have the added benefit of using the in-tree compiler when CC is unset
> in a combined-tree build.
No, I didn't see that, but I'll give it a spin, and report back. Thanks!
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 20:51 ` Daniel Jacobowitz
2005-03-24 20:59 ` Mark Mitchell
@ 2005-03-24 21:15 ` Mark Mitchell
2005-03-24 21:25 ` Ian Lance Taylor
1 sibling, 1 reply; 13+ messages in thread
From: Mark Mitchell @ 2005-03-24 21:15 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: Ian Lance Taylor, binutils
[-- Attachment #1: Type: text/plain, Size: 632 bytes --]
Daniel Jacobowitz wrote:
> Did you see my mention of [find_gcc]? Does that work for you? It will
> have the added benefit of using the in-tree compiler when CC is unset
> in a combined-tree build.
Yes, that seems to work too. Reading the DejaGNU code for find_gcc does
definitely suggest that it is the right thing to use.
I tested an installed compiler, and a just-binutils build directory
(with "make check"), and verified that the right CC still got set in
both cases. Full disclosure: I did not actually test a combined tree.
This version OK?
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
[-- Attachment #2: ld.patch --]
[-- Type: text/plain, Size: 1792 bytes --]
Index: config/default.exp
===================================================================
RCS file: /home/cvs/Repository/binutils/ld/testsuite/config/default.exp,v
retrieving revision 1.1.5.1.2.2
diff -c -5 -p -r1.1.5.1.2.2 default.exp
*** config/default.exp 24 Mar 2005 05:36:30 -0000 1.1.5.1.2.2
--- config/default.exp 24 Mar 2005 15:12:37 -0000
*************** if {![file isdirectory tmpdir/ld]} then
*** 68,81 ****
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
--- 68,99 ----
catch "exec ln -s ../../ld-new tmpdir/ld/ld" status
}
set gcc_ld_flag "-B[pwd]/tmpdir/ld/"
# load the linker path
! if {[file exists tmpdir/libpath.exp]} {
! load_lib tmpdir/libpath.exp
! foreach dir $libpath {
! set gcc_ld_flag "$gcc_ld_flag -L$dir"
! }
! }
!
! # "CC=$(CC_FOR_TARGET)", and other similar parameters. But, if the
! # user invokes runtest directly (as when testing an installed linker),
! # these flags may not be set.
! if {![info exists CC]} {
! set CC [find_gcc]
! }
! if {![info exists CFLAGS]} {
! set CFLAGS "-g -O2"
! }
! if {![info exists CXX]} {
! set CXX [find_g++]
! }
! if {![info exists CXXFLAGS]} {
! set CXXFLAGS ""
}
# The mips64-*-linux-gnu compiler defaults to the N32 ABI after
# installed, but to the O32 ABI in the build tree, because of some
# specs-file hacks. Make sure we use an ABI that is compatible with
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 21:15 ` Mark Mitchell
@ 2005-03-24 21:25 ` Ian Lance Taylor
2005-03-25 9:31 ` Mark Mitchell
0 siblings, 1 reply; 13+ messages in thread
From: Ian Lance Taylor @ 2005-03-24 21:25 UTC (permalink / raw)
To: Mark Mitchell; +Cc: Daniel Jacobowitz, binutils
Mark Mitchell <mark@codesourcery.com> writes:
> Daniel Jacobowitz wrote:
>
> > Did you see my mention of [find_gcc]? Does that work for you? It will
> > have the added benefit of using the in-tree compiler when CC is unset
> > in a combined-tree build.
>
> Yes, that seems to work too. Reading the DejaGNU code for find_gcc
> does definitely suggest that it is the right thing to use.
>
> I tested an installed compiler, and a just-binutils build directory
> (with "make check"), and verified that the right CC still got set in
> both cases. Full disclosure: I did not actually test a combined tree.
>
> This version OK?
This is OK.
Ian
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-24 21:25 ` Ian Lance Taylor
@ 2005-03-25 9:31 ` Mark Mitchell
2005-03-25 15:23 ` Daniel Jacobowitz
0 siblings, 1 reply; 13+ messages in thread
From: Mark Mitchell @ 2005-03-25 9:31 UTC (permalink / raw)
To: Ian Lance Taylor; +Cc: Daniel Jacobowitz, binutils
Ian Lance Taylor wrote:
>>I tested an installed compiler, and a just-binutils build directory
>>(with "make check"), and verified that the right CC still got set in
>>both cases. Full disclosure: I did not actually test a combined tree.
>>
>>This version OK?
>
>
> This is OK.
Thanks, installed.
Shall I put it on the 2.16 release branch, too?
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-25 9:31 ` Mark Mitchell
@ 2005-03-25 15:23 ` Daniel Jacobowitz
2005-03-25 16:00 ` Mark Mitchell
0 siblings, 1 reply; 13+ messages in thread
From: Daniel Jacobowitz @ 2005-03-25 15:23 UTC (permalink / raw)
To: Mark Mitchell; +Cc: Ian Lance Taylor, binutils
On Thu, Mar 24, 2005 at 03:36:54PM -0800, Mark Mitchell wrote:
> Ian Lance Taylor wrote:
>
> >>I tested an installed compiler, and a just-binutils build directory
> >>(with "make check"), and verified that the right CC still got set in
> >>both cases. Full disclosure: I did not actually test a combined tree.
> >>
> >>This version OK?
> >
> >
> >This is OK.
>
> Thanks, installed.
>
> Shall I put it on the 2.16 release branch, too?
Go ahead.
--
Daniel Jacobowitz
CodeSourcery, LLC
^ permalink raw reply [flat|nested] 13+ messages in thread
* Re: PATCH: Make ld testsuite usable outside build directory
2005-03-25 15:23 ` Daniel Jacobowitz
@ 2005-03-25 16:00 ` Mark Mitchell
0 siblings, 0 replies; 13+ messages in thread
From: Mark Mitchell @ 2005-03-25 16:00 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: Ian Lance Taylor, binutils
Daniel Jacobowitz wrote:
> On Thu, Mar 24, 2005 at 03:36:54PM -0800, Mark Mitchell wrote:
>
>>Ian Lance Taylor wrote:
>>
>>
>>>>I tested an installed compiler, and a just-binutils build directory
>>>>(with "make check"), and verified that the right CC still got set in
>>>>both cases. Full disclosure: I did not actually test a combined tree.
>>>>
>>>>This version OK?
>>>
>>>
>>>This is OK.
>>
>>Thanks, installed.
>>
>>Shall I put it on the 2.16 release branch, too?
>
>
> Go ahead.
Committed, thanks.
--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com
(916) 791-8304
^ permalink raw reply [flat|nested] 13+ messages in thread
end of thread, other threads:[~2005-03-25 2:22 UTC | newest]
Thread overview: 13+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2005-03-24 14:45 PATCH: Make ld testsuite usable outside build directory Mark Mitchell
2005-03-24 14:52 ` Ian Lance Taylor
2005-03-24 14:56 ` Mark Mitchell
2005-03-24 19:34 ` Daniel Jacobowitz
2005-03-24 20:39 ` Mark Mitchell
2005-03-24 20:51 ` Daniel Jacobowitz
2005-03-24 20:59 ` Mark Mitchell
2005-03-24 21:15 ` Mark Mitchell
2005-03-24 21:25 ` Ian Lance Taylor
2005-03-25 9:31 ` Mark Mitchell
2005-03-25 15:23 ` Daniel Jacobowitz
2005-03-25 16:00 ` Mark Mitchell
2005-03-24 14:54 ` Daniel Jacobowitz
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).