Re: [PATCH v3.1 1/5] cc-with-tweaks.sh: Use gdb-add-index.sh

[Pedro Alves <palves at redhat dot com>]

On 07/01/2017 04:23 PM, Jan Kratochvil wrote:
> On Thu, 29 Jun 2017 21:40:32 +0200, Simon Marchi wrote:
> 
>>> +      if [ -f ./contrib/gdb-add-index.sh ]
>>> +      then
>>> +	  GDB_ADD_INDEX="./contrib/gdb-add-index.sh"
>>> +      elif [ -f ../contrib/gdb-add-index.sh ]
>>> +      then
>>> +	  GDB_ADD_INDEX="../contrib/gdb-add-index.sh"
>>> +      elif [ -f ../../contrib/gdb-add-index.sh ]
>>> +      then
>>> +	  GDB_ADD_INDEX="../../contrib/gdb-add-index.sh"
>>> +      else
>>> +	  echo "$myname: unable to find usable contrib/gdb-add-index.sh" >&2
>>> +	  exit 1
>>> +      fi
>>> +  fi
>>> +fi
>>
>> This strategy doesn't work for out of tree builds.
> 
> I use it always with explicit $GDB_ADD_INDEX so this auto-detection does not
> really matter to me:
> 	https://git.jankratochvil.net/?p=nethome.git;a=commitdiff;h=44099fb634eca7837d30ab9b6afde0ec2838f705
> 
> 

Still, it's not OK to break testing with out of tree builds,
which is what most GDB developers use.

>> Perhaps cooking up
>> something based on $0 would be better?  I see that the GDB variable is
>> auto-detected the same way, but the gdb binary is found in the build
>> directory, so it works for the tests, whereas gdb-add-index.sh is in the
>> source directory.
>>
>> I think it would still be nice to improve how GDB is auto-detected (in
>> another patch),
> 
> OK, I believe that is unrelated to this patchset.  I am fine to always specify
> all the tools explicitly.

We must not break out of tree builds testing (with e.g
--target_board=dwarf4-gdb-index) at least.

I couldn't figure out what this hunk in the patch was for:

> --- a/gdb/contrib/cc-with-tweaks.sh
> +++ b/gdb/contrib/cc-with-tweaks.sh
> @@ -74,6 +74,8 @@ DWP=${DWP:-dwp}
>  have_link=unknown
>  next_is_output_file=no
>  output_file=a.out
> +t=/tmp/cc-with-tweaks.$$
> +rm -f $t
>  

so I removed it.


> The problem is when no index is produced whether contrib/cc-with-tweaks.sh
> should fail or not.  As originally contrib/cc-with-tweaks.sh was more quiet
> (=successful) than contrib/gdb-add-index.sh and so after this patch the
> testsuite runs with an index would "regress".  I have tried to keep the
> behavior unchanged.  Some cases still error with:
> 	Ada is not currently supported by the index
> But some cases (such as some trivial gdb.dwarf2/ testcases with no DWARF data
> to index) produce no index while the testcases still PASS now instead of:
> 	-PASS: gdb.arch/i386-bp_permanent.exp: stack pointer value matches
> 	+gdb compile failed, gdb-add-index.sh: No index was created for /quadgdb/testsuite.unix.-m64/outputs/gdb.arch/i386-bp_permanent/i386-bp_permanent
> 	+gdb-add-index.sh: [Was there no debuginfo? Was there already an index?]
> 	+UNTESTED: gdb.arch/i386-bp_permanent.exp: failed to compile

I think we may want to revisit this -- may be better to skip the
tests when testing with an index, using some skip_foo_marker (like
the existing skip_python_tests, etc.), so that testing doesn't spend
time needlessly on those.

But I sympathize with not wanting to change the current behavior.

Thus, here's what I pushed.

>From 6432ec65a8822246db5bcede0c6bd3ed0e464a0b Mon Sep 17 00:00:00 2001
From: Jan Kratochvil <jan.kratochvil@redhat.com>
Date: Fri, 8 Dec 2017 22:44:10 +0000
Subject: [PATCH 1/9] cc-with-tweaks.sh: Use gdb-add-index.sh

With DWARF-5 .debug_names, the commands to add the index to the symbol
file are more complicated, as now also .debug_str needs to be
modified.

Currently, contrib/cc-with-tweaks.sh calls objcopy to handle the '-i'
option instead of using contrib/gdb-add-index.sh which basically does
the same.  To help with .debug_names, this commit makes
contrib/cc-with-tweaks.sh reuse contrib/gdb-add-index.sh instead.

A problem this ran into is whether contrib/cc-with-tweaks.sh should
fail or not when no index is produced.

Currently, contrib/cc-with-tweaks.sh is more quiet (=successful) than
contrib/gdb-add-index.sh and so with no further changes testsuite runs
with an index would "regress".  This commit tries to keep the behavior
unchanged.  Some cases still error with:
	Ada is not currently supported by the index
But some cases (such as some trivial gdb.dwarf2/ testcases with no DWARF data
to index) produce no index while the testcases still PASS now instead of:
	-PASS: gdb.arch/i386-bp_permanent.exp: stack pointer value matches
	+gdb compile failed, gdb-add-index.sh: No index was created for gdb/testsuite.unix.-m64/outputs/gdb.arch/i386-bp_permanent/i386-bp_permanent
	+gdb-add-index.sh: [Was there no debuginfo? Was there already an index?]
	+UNTESTED: gdb.arch/i386-bp_permanent.exp: failed to compile

gdb/ChangeLog
2017-12-08  Jan Kratochvil  <jan.kratochvil@redhat.com>
	    Pedro Alves  <palves@redhat.com>

	* contrib/cc-with-tweaks.sh: Change interpreter to bash, incl. initial
	comment.
	(GDB_ADD_INDEX): New variable.
	<$want_index>: Call $GDB_ADD_INDEX.
---
 gdb/ChangeLog                 |  8 ++++++++
 gdb/contrib/cc-with-tweaks.sh | 40 +++++++++++++++++++++++-----------------
 2 files changed, 31 insertions(+), 17 deletions(-)

diff --git a/gdb/ChangeLog b/gdb/ChangeLog
index 26cf18e..e90a3f0 100644
--- a/gdb/ChangeLog
+++ b/gdb/ChangeLog
@@ -1,3 +1,11 @@
+2017-12-08  Jan Kratochvil  <jan.kratochvil@redhat.com>
+	    Pedro Alves  <palves@redhat.com>
+
+	* contrib/cc-with-tweaks.sh: Change interpreter to bash, incl. initial
+	comment.
+	(GDB_ADD_INDEX): New variable.
+	<$want_index>: Call $GDB_ADD_INDEX.
+
 2017-12-08  Sergio Durigan Junior  <sergiodj@redhat.com>
 
 	* dtrace-probe.c (dtrace_process_dof_probe): Do not declare a new
diff --git a/gdb/contrib/cc-with-tweaks.sh b/gdb/contrib/cc-with-tweaks.sh
index 7d39c00..a65fef8 100755
--- a/gdb/contrib/cc-with-tweaks.sh
+++ b/gdb/contrib/cc-with-tweaks.sh
@@ -1,4 +1,4 @@
-#! /bin/sh
+#!/usr/bin/env bash
 # Wrapper around gcc to tweak the output in various ways when running
 # the testsuite.
 
@@ -27,8 +27,8 @@
 #
 # bash$ cd $objdir/gdb/testsuite
 # bash$ runtest \
-#   CC_FOR_TARGET="/bin/sh $srcdir/gdb/contrib/cc-with-tweaks.sh ARGS gcc" \
-#   CXX_FOR_TARGET="/bin/sh $srcdir/gdb/contrib/cc-with-tweaks.sh ARGS g++"
+#   CC_FOR_TARGET="/bin/bash $srcdir/gdb/contrib/cc-with-tweaks.sh ARGS gcc" \
+#   CXX_FOR_TARGET="/bin/bash $srcdir/gdb/contrib/cc-with-tweaks.sh ARGS g++"
 #
 # For documentation on Fission and dwp files:
 #     http://gcc.gnu.org/wiki/DebugFission
@@ -47,6 +47,7 @@
 # If nothing is given, no changes are made
 
 myname=cc-with-tweaks.sh
+mydir=`dirname "$0"`
 
 if [ -z "$GDB" ]
 then
@@ -93,6 +94,20 @@ while [ $# -gt 0 ]; do
     shift
 done
 
+if [ "$want_index" = true ]
+then
+    if [ -z "$GDB_ADD_INDEX" ]
+    then
+	if [ -f $mydir/gdb-add-index.sh ]
+	then
+	    GDB_ADD_INDEX="$mydir/gdb-add-index.sh"
+	else
+	    echo "$myname: unable to find usable contrib/gdb-add-index.sh" >&2
+	    exit 1
+	fi
+    fi
+fi
+
 for arg in "$@"
 do
     if [ "$next_is_output_file" = "yes" ]
@@ -152,20 +167,11 @@ if [ "$want_objcopy_compress" = true ]; then
 fi
 
 if [ "$want_index" = true ]; then
-    $GDB --batch-silent -nx -ex "set auto-load no" -ex "file $output_file" -ex "save gdb-index $output_dir"
-    rc=$?
-    [ $rc != 0 ] && exit $rc
-
-    # GDB might not always create an index.  Cope.
-    if [ -f "$index_file" ]
-    then
-	$OBJCOPY --add-section .gdb_index="$index_file" \
-	    --set-section-flags .gdb_index=readonly \
-	    "$output_file" "$output_file"
-	rc=$?
-    else
-	rc=0
-    fi
+    # Filter out these messages which would stop dejagnu testcase run:
+    # echo "$myname: No index was created for $file" 1>&2
+    # echo "$myname: [Was there no debuginfo? Was there already an index?]" 1>&2
+    GDB=$GDB $GDB_ADD_INDEX "$output_file" 2>&1|grep -v "^${GDB_ADD_INDEX##*/}: " >&2
+    rc=${PIPESTATUS[0]}
     [ $rc != 0 ] && exit $rc
 fi
 
-- 
2.5.5