[PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Yao Qi]

I see the following two fails on arm-none-eabi target, because argv[0]
isn't available.

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept file symbolic link name

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept directory symbolic link name

My first thought is to check [target_info exists noargs], and skip the
test if it returns true.  However, noargs is set in gdbserver board
files, so argv0-symlink.exp will be skipped on gdbserver board file.
The change is too aggressive.

When the program is running with gdbserver, argv[1] to argv[N] aren't
available, but argv[0] is.  Fortunately, argv0-symlink.exp only
requires argv[0].  argv0-symlink.exp can be run with gdbserver board
file, as what we do now.

What we need to check is whether argv[0] is available, so I add a new
proc gdb_has_argv0 to do so by starting a program, and check
argc/argv[0] to see whether argv[0] is available.

Dan fixed the similar problem by checking noargs, which is too strong.
https://sourceware.org/ml/gdb-patches/2010-02/msg00398.html as a
result, the test is skipped on gdbserver.  This patch fixed it too.

Re-run argv0-symlink.exp, scm-value.exp, and py-value.exp on x86-linux
(both native and gdbserver) and arm-none-eabi.  No fail on x86-linux,
and argv0-symlink.exp is skipped on arm-none-eabi target.

gdb/testsuite:

2014-10-16  Yao Qi  <yao@codesourcery.com>

	* gdb.base/argv0-symlink.exp: Skip it if gdb_has_argv0 return
	false.
	* gdb.guile/scm-value.exp (test_value_in_inferior): Don't
	check [target_info exists noargs], check [gdb_has_argv0]
	instead.
	* gdb.python/py-value.exp (test_value_in_inferior): Likewise.
	* lib/gdb.exp (gdb_has_argv0, gdb_has_argv0_1): New
	procedures.
---
 gdb/testsuite/gdb.base/argv0-symlink.exp |  5 ++
 gdb/testsuite/gdb.guile/scm-value.exp    |  5 +-
 gdb/testsuite/gdb.python/py-value.exp    |  5 +-
 gdb/testsuite/lib/gdb.exp                | 79 ++++++++++++++++++++++++++++++++
 4 files changed, 92 insertions(+), 2 deletions(-)

diff --git a/gdb/testsuite/gdb.base/argv0-symlink.exp b/gdb/testsuite/gdb.base/argv0-symlink.exp
index d849b4c..183fd70 100644
--- a/gdb/testsuite/gdb.base/argv0-symlink.exp
+++ b/gdb/testsuite/gdb.base/argv0-symlink.exp
@@ -13,6 +13,11 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
+if { ![gdb_has_argv0] } {
+    unsupported "Can't get argv\[0\]."
+    return
+}
+
 standard_testfile
 
 if { [build_executable ${testfile}.exp ${testfile} ${srcfile}] == -1 } {
diff --git a/gdb/testsuite/gdb.guile/scm-value.exp b/gdb/testsuite/gdb.guile/scm-value.exp
index ae80d1b..2b49134 100644
--- a/gdb/testsuite/gdb.guile/scm-value.exp
+++ b/gdb/testsuite/gdb.guile/scm-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-guile.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -86,7 +88,8 @@ proc test_value_in_inferior {} {
 	"set arg0"
 
     # Check that the dereferenced value is sane.
-    if { ! [target_info exists noargs] } {
+    global has_argv0
+    if { $has_argv0 } {
 	gdb_test "gu (print arg0)" \
 	    "0x.*$testfile\"" "verify dereferenced value"
     }
diff --git a/gdb/testsuite/gdb.python/py-value.exp b/gdb/testsuite/gdb.python/py-value.exp
index 0d18bef..afa55a5 100644
--- a/gdb/testsuite/gdb.python/py-value.exp
+++ b/gdb/testsuite/gdb.python/py-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-python.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -221,7 +223,8 @@ proc test_value_in_inferior {} {
   gdb_py_test_silent_cmd "python arg0 = argv.dereference ()" "dereference value" 1
 
   # Check that the dereferenced value is sane
-  if { ! [target_info exists noargs] } {
+  global has_argv0
+  if { $has_argv0 } {
     gdb_test "python print (arg0)" "0x.*$testfile\"" "verify dereferenced value"
   }
 
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 7b2a402..dd525dc 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -4227,6 +4227,85 @@ gdb_caching_proc gdb_skip_xml_test {
     return $xml_missing
 }
 
+# Return true if argv[0] is available.
+
+gdb_caching_proc gdb_has_argv0 {
+    set result 0
+
+    # Set up, compile, and execute a test program to check whether
+    # argv[0] is available.
+    set src [standard_temp_file has_argv0[pid].c]
+    set exe [standard_temp_file has_argv0[pid].x]
+
+    gdb_produce_source $src {
+	int main (int argc, char **argv) {
+	    return 0;
+	}
+    }
+
+    gdb_compile $src $exe executable {debug}
+
+    # Helper proc.
+    proc gdb_has_argv0_1 { exe } {
+	global srcdir subdir
+	global gdb_prompt hex decimal
+
+	gdb_exit
+	gdb_start
+	gdb_reinitialize_dir $srcdir/$subdir
+	gdb_load "$exe"
+
+	# Set breakpoint on main.
+	send_gdb "break main\n"
+	gdb_expect {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Run to main.
+	gdb_run_cmd
+	gdb_expect {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Check whether argc is 1.
+	send_gdb "p argc\n"
+	gdb_expect {
+	    -re " = 1\r\n${gdb_prompt} $" {
+
+		send_gdb "p argv\[0\]\n"
+		gdb_expect {
+		    -re " = $hex \".*$exe\"\r\n${gdb_prompt} $" {
+			return 1
+		    }
+		    -re "${gdb_prompt} $" {
+			return 0
+		    }
+		}
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+	return 0
+    }
+
+    set result [gdb_has_argv0_1 $exe]
+
+    gdb_exit
+    file delete $src
+    file delete $exe
+
+    return $result
+}
+
 # Note: the procedure gdb_gnu_strip_debug will produce an executable called
 # ${binfile}.dbglnk, which is just like the executable ($binfile) but without
 # the debuginfo. Instead $binfile has a .gnu_debuglink section which contains
-- 
1.9.3

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Pedro Alves]

On 10/16/2014 07:25 AM, Yao Qi wrote:

> 
> When the program is running with gdbserver, argv[1] to argv[N] aren't
> available, but argv[0] is.  Fortunately, argv0-symlink.exp only
> requires argv[0].  argv0-symlink.exp can be run with gdbserver board
> file, as what we do now.
> 
> What we need to check is whether argv[0] is available, so I add a new
> proc gdb_has_argv0 to do so by starting a program, and check
> argc/argv[0] to see whether argv[0] is available.

Sounds good.

> --- a/gdb/testsuite/gdb.base/argv0-symlink.exp
> +++ b/gdb/testsuite/gdb.base/argv0-symlink.exp
> @@ -13,6 +13,11 @@
>  # You should have received a copy of the GNU General Public License
>  # along with this program.  If not, see <http://www.gnu.org/licenses/>.
>  
> +if { ![gdb_has_argv0] } {
> +    unsupported "Can't get argv\[0\]."
> +    return
> +}
> +

I think we shouldn't skip the whole test though; only the
printing argv[0] test.  The test file also makes sure
that "info inferiors" shows the symlink name, not the target
of the symlink, and that is host-dependent, not target
dependent.  See git 4856b6bc.

>  
>    # Check that the dereferenced value is sane
> -  if { ! [target_info exists noargs] } {
> +  global has_argv0
> +  if { $has_argv0 } {
>      gdb_test "python print (arg0)" "0x.*$testfile\"" "verify dereferenced value"
>    }
>  
> diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
> index 7b2a402..dd525dc 100644
> --- a/gdb/testsuite/lib/gdb.exp
> +++ b/gdb/testsuite/lib/gdb.exp
> @@ -4227,6 +4227,85 @@ gdb_caching_proc gdb_skip_xml_test {
>      return $xml_missing
>  }
>  
> +# Return true if argv[0] is available.
> +
> +gdb_caching_proc gdb_has_argv0 {
> +    set result 0
> +
> +    # Set up, compile, and execute a test program to check whether
> +    # argv[0] is available.
> +    set src [standard_temp_file has_argv0[pid].c]
> +    set exe [standard_temp_file has_argv0[pid].x]
> +
> +    gdb_produce_source $src {
> +	int main (int argc, char **argv) {
> +	    return 0;
> +	}
> +    }
> +
> +    gdb_compile $src $exe executable {debug}
> +
> +    # Helper proc.
> +    proc gdb_has_argv0_1 { exe } {
> +	global srcdir subdir
> +	global gdb_prompt hex decimal
> +
> +	gdb_exit
> +	gdb_start
> +	gdb_reinitialize_dir $srcdir/$subdir
> +	gdb_load "$exe"
> +
> +	# Set breakpoint on main.
> +	send_gdb "break main\n"
> +	gdb_expect {
> +	    -re "Breakpoint.*${gdb_prompt} $" {
> +	    }
> +	    -re "${gdb_prompt} $" {
> +		return 0
> +	    }
> +	}
> +
> +	# Run to main.
> +	gdb_run_cmd
> +	gdb_expect {
> +	    -re "Breakpoint.*${gdb_prompt} $" {
> +	    }
> +	    -re "${gdb_prompt} $" {
> +		return 0
> +	    }
> +	}
> +
> +	# Check whether argc is 1.
> +	send_gdb "p argc\n"
> +	gdb_expect {
> +	    -re " = 1\r\n${gdb_prompt} $" {
> +
> +		send_gdb "p argv\[0\]\n"
> +		gdb_expect {
> +		    -re " = $hex \".*$exe\"\r\n${gdb_prompt} $" {

I suspect this may break if remote (host|target) testing,
and not sharing the filesystem between build/host/target.
Isn't $exe here a full path on the build?

> +			return 1
> +		    }
> +		    -re "${gdb_prompt} $" {
> +			return 0
> +		    }
> +		}
> +	    }
> +	    -re "${gdb_prompt} $" {
> +		return 0
> +	    }
> +	}
> +	return 0
> +    }
> +

I think these gdb_expect's should be gdb_test_multiple's instead.

Thanks,
Pedro Alves

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Yao Qi]

Pedro Alves <palves@redhat.com> writes:

> I think we shouldn't skip the whole test though; only the
> printing argv[0] test.  The test file also makes sure
> that "info inferiors" shows the symlink name, not the target
> of the symlink, and that is host-dependent, not target
> dependent.  See git 4856b6bc.

Yeah, we can wrap the test on argv[0] with if { $has_argv0 } {...}.  See
below in the updated patch.

>> +
>> +		send_gdb "p argv\[0\]\n"
>> +		gdb_expect {
>> +		    -re " = $hex \".*$exe\"\r\n${gdb_prompt} $" {
>
> I suspect this may break if remote (host|target) testing,
> and not sharing the filesystem between build/host/target.
> Isn't $exe here a full path on the build?
>

$exe can be a full path, and can be a base name too.  It depends on proc
standard_temp_file.  You are right that it is a mistake to use the build
file path on host.  I change it to 

  ".*[file tail $exe]\"\r\n${gdb_prompt} $"

>> +			return 1
>> +		    }
>> +		    -re "${gdb_prompt} $" {
>> +			return 0
>> +		    }
>> +		}
>> +	    }
>> +	    -re "${gdb_prompt} $" {
>> +		return 0
>> +	    }
>> +	}
>> +	return 0
>> +    }
>> +
>
> I think these gdb_expect's should be gdb_test_multiple's instead.

Fixed.

-- 
Yao (齐尧)

Subject: [PATCH] Skip testing argv[0] on target argv[0] isn't available

I see the following two fails on arm-none-eabi target, because argv[0]
isn't available.

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept file symbolic link name

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept directory symbolic link name

My first thought is to check [target_info exists noargs], and skip the
test if it returns true.  However, noargs is set in gdbserver board
files, so argv0-symlink.exp will be skipped on gdbserver board file.
The change is too aggressive.

When the program is running with gdbserver, argv[1] to argv[N] aren't
available, but argv[0] is.  Fortunately, argv0-symlink.exp only
requires argv[0].  argv0-symlink.exp can be run with gdbserver board
file, as what we do now.

What we need to check is whether argv[0] is available, so I add a new
proc gdb_has_argv0 to do so by starting a program, and check
argc/argv[0] to see whether argv[0] is available.

Dan fixed the similar problem by checking noargs, which is too strong.
https://sourceware.org/ml/gdb-patches/2010-02/msg00398.html as a
result, the test is skipped on gdbserver.  This patch fixed it too.

gdb/testsuite:

2014-10-17  Yao Qi  <yao@codesourcery.com>

	* gdb.base/argv0-symlink.exp: Check argv[0] value if
	gdb_has_argv0 return true.
	* gdb.guile/scm-value.exp (test_value_in_inferior): Don't
	check [target_info exists noargs], check [gdb_has_argv0]
	instead.
	* gdb.python/py-value.exp (test_value_in_inferior): Likewise.
	* lib/gdb.exp (gdb_has_argv0, gdb_has_argv0_1): New
	procedures.

diff --git a/gdb/testsuite/gdb.base/argv0-symlink.exp b/gdb/testsuite/gdb.base/argv0-symlink.exp
index d849b4c..cda2128 100644
--- a/gdb/testsuite/gdb.base/argv0-symlink.exp
+++ b/gdb/testsuite/gdb.base/argv0-symlink.exp
@@ -15,6 +15,8 @@
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 if { [build_executable ${testfile}.exp ${testfile} ${srcfile}] == -1 } {
     return -1
 }
@@ -39,7 +41,9 @@ if ![runto_main] {
 gdb_test_no_output "set print repeats 10000"
 gdb_test_no_output "set print elements 10000"
 
-gdb_test {print argv[0]} "/$filelink\"" $test
+if { $has_argv0 } {
+    gdb_test {print argv[0]} "/$filelink\"" $test
+}
 
 # For a link named /PATH/TO/DIR/LINK, we want to check the output
 # against "/DIR/LINK", but computed in a way that doesn't make
@@ -73,9 +77,12 @@ if ![runto_main] {
 gdb_test_no_output "set print repeats 10000"
 gdb_test_no_output "set print elements 10000"
 
-# gdbserver does not have this issue.
-if ![is_remote target] {
-    setup_kfail "*-*-*" gdb/15934
+if { $has_argv0 } {
+    # gdbserver does not have this issue.
+    if ![is_remote target] {
+	setup_kfail "*-*-*" gdb/15934
+    }
+    gdb_test {print argv[0]} "/$dirlink/$filelink\"" $test
 }
-gdb_test {print argv[0]} "/$dirlink/$filelink\"" $test
+
 gdb_test "info inferiors" "/$lastdir/$filelink *" "$test for info inferiors"
diff --git a/gdb/testsuite/gdb.guile/scm-value.exp b/gdb/testsuite/gdb.guile/scm-value.exp
index ae80d1b..2b49134 100644
--- a/gdb/testsuite/gdb.guile/scm-value.exp
+++ b/gdb/testsuite/gdb.guile/scm-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-guile.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -86,7 +88,8 @@ proc test_value_in_inferior {} {
 	"set arg0"
 
     # Check that the dereferenced value is sane.
-    if { ! [target_info exists noargs] } {
+    global has_argv0
+    if { $has_argv0 } {
 	gdb_test "gu (print arg0)" \
 	    "0x.*$testfile\"" "verify dereferenced value"
     }
diff --git a/gdb/testsuite/gdb.python/py-value.exp b/gdb/testsuite/gdb.python/py-value.exp
index 0d18bef..afa55a5 100644
--- a/gdb/testsuite/gdb.python/py-value.exp
+++ b/gdb/testsuite/gdb.python/py-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-python.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -221,7 +223,8 @@ proc test_value_in_inferior {} {
   gdb_py_test_silent_cmd "python arg0 = argv.dereference ()" "dereference value" 1
 
   # Check that the dereferenced value is sane
-  if { ! [target_info exists noargs] } {
+  global has_argv0
+  if { $has_argv0 } {
     gdb_test "python print (arg0)" "0x.*$testfile\"" "verify dereferenced value"
   }
 
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 7b2a402..53a7b1f 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -4227,6 +4227,82 @@ gdb_caching_proc gdb_skip_xml_test {
     return $xml_missing
 }
 
+# Return true if argv[0] is available.
+
+gdb_caching_proc gdb_has_argv0 {
+    set result 0
+
+    # Set up, compile, and execute a test program to check whether
+    # argv[0] is available.
+    set src [standard_temp_file has_argv0[pid].c]
+    set exe [standard_temp_file has_argv0[pid].x]
+
+    gdb_produce_source $src {
+	int main (int argc, char **argv) {
+	    return 0;
+	}
+    }
+
+    gdb_compile $src $exe executable {debug}
+
+    # Helper proc.
+    proc gdb_has_argv0_1 { exe } {
+	global srcdir subdir
+	global gdb_prompt hex decimal
+
+	gdb_exit
+	gdb_start
+	gdb_reinitialize_dir $srcdir/$subdir
+	gdb_load "$exe"
+
+	# Set breakpoint on main.
+	gdb_test_multiple "break main" "break main" {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Run to main.
+	gdb_run_cmd
+	gdb_test_multiple "" "run to main" {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Check whether argc is 1.
+	gdb_test_multiple "p argc" "p argc" {
+	    -re " = 1\r\n${gdb_prompt} $" {
+
+		gdb_test_multiple "p argv\[0\]" "p argv\[0\]" {
+		    -re " = $hex \".*[file tail $exe]\"\r\n${gdb_prompt} $" {
+			return 1
+		    }
+		    -re "${gdb_prompt} $" {
+			return 0
+		    }
+		}
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+	return 0
+    }
+
+    set result [gdb_has_argv0_1 $exe]
+
+    gdb_exit
+    file delete $src
+    file delete $exe
+
+    return $result
+}
+
 # Note: the procedure gdb_gnu_strip_debug will produce an executable called
 # ${binfile}.dbglnk, which is just like the executable ($binfile) but without
 # the debuginfo. Instead $binfile has a .gnu_debuglink section which contains

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Pedro Alves]

On 10/17/2014 08:20 AM, Yao Qi wrote:

>>> >> +
>>> >> +		send_gdb "p argv\[0\]\n"
>>> >> +		gdb_expect {
>>> >> +		    -re " = $hex \".*$exe\"\r\n${gdb_prompt} $" {
>> >
>> > I suspect this may break if remote (host|target) testing,
>> > and not sharing the filesystem between build/host/target.
>> > Isn't $exe here a full path on the build?
>> >
> $exe can be a full path, and can be a base name too.  It depends on proc
> standard_temp_file.  You are right that it is a mistake to use the build
> file path on host.  I change it to 
> 
>   ".*[file tail $exe]\"\r\n${gdb_prompt} $"

OK.  So I'm somewhat worried about silently reducing coverage due to
some problem here, either now, or in the future.

So how about instead of silently skipping the tests, call
unsupported, like:

  global has_argv0
  set test "verify dereferenced value"
  if { $has_argv0 } {
     gdb_test "python print (arg0)" "0x.*$testfile\"" $test
  } else {
     unsupported $test
  }

etc.

> +gdb_caching_proc gdb_has_argv0 {
> +    set result 0
...
> +    # Helper proc.
> +    proc gdb_has_argv0_1 { exe } {
...
> +    set result [gdb_has_argv0_1 $exe]

And then in addition, add an assert here for the main OSs
we support, like so:

 if { !$result
      && ([istarget *-*-linux*]
         || [istarget *-*-gnu*]
         || [istarget *-*-cygwin*]
         || [istarget *-*-mingw*]
         || [istarget *-*-darwin*]
         || [istarget *-*-bsd*]
         || [istarget *-*-solaris*]
         || [istarget *-*-msdosdjgpp*]
         || [istarget *-*-go32*]
	 || [istarget *-*-aix*]
         || [istarget *-*-hpux*]) } {
    fail "argv[0] should be available on this target"
 }

> +    # Helper proc.
> +    proc gdb_has_argv0_1 { exe } {
> +	global srcdir subdir
> +	global gdb_prompt hex decimal
> +
> +	gdb_exit
> +	gdb_start
> +	gdb_reinitialize_dir $srcdir/$subdir
> +	gdb_load "$exe"

BTW, use clean_restart here?  (if so, I think 'global srcdir subdir'
becomes unnecessary too)

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Yao Qi]

Pedro Alves <palves@redhat.com> writes:

> So how about instead of silently skipping the tests, call
> unsupported, like:
>
>   global has_argv0
>   set test "verify dereferenced value"
>   if { $has_argv0 } {
>      gdb_test "python print (arg0)" "0x.*$testfile\"" $test
>   } else {
>      unsupported $test
>   }
>

OK, fixed in the updated patch.

> etc.
>
>> +gdb_caching_proc gdb_has_argv0 {
>> +    set result 0
> ...
>> +    # Helper proc.
>> +    proc gdb_has_argv0_1 { exe } {
> ...
>> +    set result [gdb_has_argv0_1 $exe]
>
> And then in addition, add an assert here for the main OSs
> we support, like so:
>
>  if { !$result
>       && ([istarget *-*-linux*]
>          || [istarget *-*-gnu*]
>          || [istarget *-*-cygwin*]
>          || [istarget *-*-mingw*]
>          || [istarget *-*-darwin*]
>          || [istarget *-*-bsd*]
>          || [istarget *-*-solaris*]
>          || [istarget *-*-msdosdjgpp*]
>          || [istarget *-*-go32*]
> 	 || [istarget *-*-aix*]
>          || [istarget *-*-hpux*]) } {
>     fail "argv[0] should be available on this target"
>  }
>

OK, I list all OSes GDB supports (extracted from configure.tgt and configure.ac).

>> +    # Helper proc.
>> +    proc gdb_has_argv0_1 { exe } {
>> +	global srcdir subdir
>> +	global gdb_prompt hex decimal
>> +
>> +	gdb_exit
>> +	gdb_start
>> +	gdb_reinitialize_dir $srcdir/$subdir
>> +	gdb_load "$exe"
>
> BTW, use clean_restart here?  (if so, I think 'global srcdir subdir'
> becomes unnecessary too)

Unfortunately we can't.  clean_restart loads file from $objdir/$subdir,
but $exe is in $objdir or $objdir/temp.

-- 
Yao (齐尧)

Subject: [PATCH] Skip testing argv[0] on target argv[0] isn't available

I see the following two fails on arm-none-eabi target, because argv[0]
isn't available.

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept file symbolic link name

print argv[0]^M
$1 = 0x1f78 "/dev/null"^M
(gdb) FAIL: gdb.base/argv0-symlink.exp: kept directory symbolic link name

My first thought is to check [target_info exists noargs], and skip the
test if it returns true.  However, noargs is set in gdbserver board
files, so argv0-symlink.exp will be skipped on gdbserver board file.
The change is too aggressive.

When the program is running with gdbserver, argv[1] to argv[N] aren't
available, but argv[0] is.  Fortunately, argv0-symlink.exp only
requires argv[0].  argv0-symlink.exp can be run with gdbserver board
file, as what we do now.

What we need to check is whether argv[0] is available, so I add a new
proc gdb_has_argv0 to do so by starting a program, and check
argc/argv[0] to see whether argv[0] is available.

Dan fixed the similar problem by checking noargs, which is too strong.
https://sourceware.org/ml/gdb-patches/2010-02/msg00398.html as a
result, the test is skipped on gdbserver.  This patch fixed it too.

gdb/testsuite:

2014-10-17  Yao Qi  <yao@codesourcery.com>

	* gdb.base/argv0-symlink.exp: Check argv[0] value if
	gdb_has_argv0 return true.
	* gdb.guile/scm-value.exp (test_value_in_inferior): Don't
	check [target_info exists noargs], check [gdb_has_argv0]
	instead.
	* gdb.python/py-value.exp (test_value_in_inferior): Likewise.
	* lib/gdb.exp (gdb_has_argv0, gdb_has_argv0_1): New
	procedures.

diff --git a/gdb/testsuite/gdb.base/argv0-symlink.exp b/gdb/testsuite/gdb.base/argv0-symlink.exp
index d849b4c..cd459bb 100644
--- a/gdb/testsuite/gdb.base/argv0-symlink.exp
+++ b/gdb/testsuite/gdb.base/argv0-symlink.exp
@@ -15,6 +15,8 @@
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 if { [build_executable ${testfile}.exp ${testfile} ${srcfile}] == -1 } {
     return -1
 }
@@ -39,7 +41,11 @@ if ![runto_main] {
 gdb_test_no_output "set print repeats 10000"
 gdb_test_no_output "set print elements 10000"
 
-gdb_test {print argv[0]} "/$filelink\"" $test
+if { $has_argv0 } {
+    gdb_test {print argv[0]} "/$filelink\"" $test
+} else {
+    unsupported $test
+}
 
 # For a link named /PATH/TO/DIR/LINK, we want to check the output
 # against "/DIR/LINK", but computed in a way that doesn't make
@@ -73,9 +79,14 @@ if ![runto_main] {
 gdb_test_no_output "set print repeats 10000"
 gdb_test_no_output "set print elements 10000"
 
-# gdbserver does not have this issue.
-if ![is_remote target] {
-    setup_kfail "*-*-*" gdb/15934
+if { $has_argv0 } {
+    # gdbserver does not have this issue.
+    if ![is_remote target] {
+	setup_kfail "*-*-*" gdb/15934
+    }
+    gdb_test {print argv[0]} "/$dirlink/$filelink\"" $test
+} else {
+    unsupported $test
 }
-gdb_test {print argv[0]} "/$dirlink/$filelink\"" $test
+
 gdb_test "info inferiors" "/$lastdir/$filelink *" "$test for info inferiors"
diff --git a/gdb/testsuite/gdb.guile/scm-value.exp b/gdb/testsuite/gdb.guile/scm-value.exp
index ae80d1b..033633c 100644
--- a/gdb/testsuite/gdb.guile/scm-value.exp
+++ b/gdb/testsuite/gdb.guile/scm-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-guile.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -86,9 +88,12 @@ proc test_value_in_inferior {} {
 	"set arg0"
 
     # Check that the dereferenced value is sane.
-    if { ! [target_info exists noargs] } {
-	gdb_test "gu (print arg0)" \
-	    "0x.*$testfile\"" "verify dereferenced value"
+    global has_argv0
+    set test "verify dereferenced value"
+    if { $has_argv0 } {
+	gdb_test "gu (print arg0)" "0x.*$testfile\"" $test
+    } else {
+	unsupported $test
     }
 
     # Smoke-test value-optimized-out?.
diff --git a/gdb/testsuite/gdb.python/py-value.exp b/gdb/testsuite/gdb.python/py-value.exp
index 0d18bef..0e1534a 100644
--- a/gdb/testsuite/gdb.python/py-value.exp
+++ b/gdb/testsuite/gdb.python/py-value.exp
@@ -20,6 +20,8 @@ load_lib gdb-python.exp
 
 standard_testfile
 
+set has_argv0 [gdb_has_argv0]
+
 # Build inferior to language specification.
 # LANG is one of "c" or "c++".
 proc build_inferior {exefile lang} {
@@ -221,8 +223,12 @@ proc test_value_in_inferior {} {
   gdb_py_test_silent_cmd "python arg0 = argv.dereference ()" "dereference value" 1
 
   # Check that the dereferenced value is sane
-  if { ! [target_info exists noargs] } {
-    gdb_test "python print (arg0)" "0x.*$testfile\"" "verify dereferenced value"
+  global has_argv0
+  set test "verify dereferenced value"
+  if { $has_argv0 } {
+    gdb_test "python print (arg0)" "0x.*$testfile\"" $test
+  } else {
+    unsupported $test
   }
 
   # Smoke-test is_optimized_out attribute
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 7b2a402..37e701c 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -4227,6 +4227,104 @@ gdb_caching_proc gdb_skip_xml_test {
     return $xml_missing
 }
 
+# Return true if argv[0] is available.
+
+gdb_caching_proc gdb_has_argv0 {
+    set result 0
+
+    # Set up, compile, and execute a test program to check whether
+    # argv[0] is available.
+    set src [standard_temp_file has_argv0[pid].c]
+    set exe [standard_temp_file has_argv0[pid].x]
+
+    gdb_produce_source $src {
+	int main (int argc, char **argv) {
+	    return 0;
+	}
+    }
+
+    gdb_compile $src $exe executable {debug}
+
+    # Helper proc.
+    proc gdb_has_argv0_1 { exe } {
+	global srcdir subdir
+	global gdb_prompt hex
+
+	gdb_exit
+	gdb_start
+	gdb_reinitialize_dir $srcdir/$subdir
+	gdb_load "$exe"
+
+	# Set breakpoint on main.
+	gdb_test_multiple "break main" "break main" {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Run to main.
+	gdb_run_cmd
+	gdb_test_multiple "" "run to main" {
+	    -re "Breakpoint.*${gdb_prompt} $" {
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+
+	# Check whether argc is 1.
+	gdb_test_multiple "p argc" "p argc" {
+	    -re " = 1\r\n${gdb_prompt} $" {
+
+		gdb_test_multiple "p argv\[0\]" "p argv\[0\]" {
+		    -re " = $hex \".*[file tail $exe]\"\r\n${gdb_prompt} $" {
+			return 1
+		    }
+		    -re "${gdb_prompt} $" {
+			return 0
+		    }
+		}
+	    }
+	    -re "${gdb_prompt} $" {
+		return 0
+	    }
+	}
+	return 0
+    }
+
+    set result [gdb_has_argv0_1 $exe]
+
+    gdb_exit
+    file delete $src
+    file delete $exe
+
+    if { !$result
+      && ([istarget *-*-linux*]
+	  || [istarget *-*-freebsd*] || [istarget *-*-kfreebsd*]
+	  || [istarget *-*-netbsd*] || [istarget *-*-knetbsd*]
+	  || [istarget *-*-openbsd*]
+	  || [istarget *-*-darwin*]
+	  || [istarget *-*-solaris*]
+	  || [istarget *-*-aix*]
+	  || [istarget *-*-gnu*]
+	  || [istarget *-*-cygwin*] || [istarget *-*-mingw32*]
+	  || [istarget *-*-*djgpp*] || [istarget *-*-go32*]
+	  || [istarget *-wince-pe] || [istarget *-*-mingw32ce*]
+	  || [istarget *-*-symbianelf*]
+	  || [istarget *-*-osf*]
+	  || [istarget *-*-hpux*]
+	  || [istarget *-*-dicos*]
+	  || [istarget *-*-nto*]
+	  || [istarget *-*-*vms*]
+	  || [istarget *-*-lynx*178]) } {
+	fail "argv\[0\] should be available on this target"
+    }
+
+    return $result
+}
+
 # Note: the procedure gdb_gnu_strip_debug will produce an executable called
 # ${binfile}.dbglnk, which is just like the executable ($binfile) but without
 # the debuginfo. Instead $binfile has a .gnu_debuglink section which contains

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Pedro Alves]

On 10/17/2014 02:16 PM, Yao Qi wrote:
> Pedro Alves <palves@redhat.com> writes:
> 
>> So how about instead of silently skipping the tests, call
>> unsupported, like:
>>
>>   global has_argv0
>>   set test "verify dereferenced value"
>>   if { $has_argv0 } {
>>      gdb_test "python print (arg0)" "0x.*$testfile\"" $test
>>   } else {
>>      unsupported $test
>>   }
>>
> 
> OK, fixed in the updated patch.
> 
>> etc.
>>
>>> +gdb_caching_proc gdb_has_argv0 {
>>> +    set result 0
>> ...
>>> +    # Helper proc.
>>> +    proc gdb_has_argv0_1 { exe } {
>> ...
>>> +    set result [gdb_has_argv0_1 $exe]
>>
>> And then in addition, add an assert here for the main OSs
>> we support, like so:
>>
>>  if { !$result
>>       && ([istarget *-*-linux*]
>>          || [istarget *-*-gnu*]
>>          || [istarget *-*-cygwin*]
>>          || [istarget *-*-mingw*]
>>          || [istarget *-*-darwin*]
>>          || [istarget *-*-bsd*]
>>          || [istarget *-*-solaris*]
>>          || [istarget *-*-msdosdjgpp*]
>>          || [istarget *-*-go32*]
>> 	 || [istarget *-*-aix*]
>>          || [istarget *-*-hpux*]) } {
>>     fail "argv[0] should be available on this target"
>>  }
>>
> 
> OK, I list all OSes GDB supports (extracted from configure.tgt and configure.ac).
> 
>>> +    # Helper proc.
>>> +    proc gdb_has_argv0_1 { exe } {
>>> +	global srcdir subdir
>>> +	global gdb_prompt hex decimal
>>> +
>>> +	gdb_exit
>>> +	gdb_start
>>> +	gdb_reinitialize_dir $srcdir/$subdir
>>> +	gdb_load "$exe"
>>
>> BTW, use clean_restart here?  (if so, I think 'global srcdir subdir'
>> becomes unnecessary too)
> 
> Unfortunately we can't.  clean_restart loads file from $objdir/$subdir,
> but $exe is in $objdir or $objdir/temp.
> 

OK.  This version looks good to me.

Thanks,
Pedro Alves

Re: [PATCH] Skip argv0-symlink.exp on target argv[0] isn't available

[Yao Qi]

Pedro Alves <palves@redhat.com> writes:

> OK.  This version looks good to me.

Thanks, patch is pushed in.

-- 
Yao (齐尧)