[vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[Alexandre Oliva]

[-- Attachment #1: Type: text/plain, Size: 1097 bytes --]

bootstrap-debug is broken in trunk and vta branch on machines that
support assembler-generated CFI and that don't generate CFI by ABI
requirements.  I got bootstrap-debug failures on at least i686 and
ppc64, because compilations with -g0 didn't have an .eh_frame section,
whereas such trivial translation units as ada/final got a simple
.eh_frame section with -g2, because .cfi_startproc and .cfi_endproc
are enough to cause the generation of .eh_frame in addition to debug
info.

I suppose enabling the assembler to tell whether or not we're
interested in .eh_frame, in addition to .debug_frame, would be a win
in terms of stripped binary size, but I wasn't sure it was worth the
effort.  Thoughts?

In the meantime, I needed some way to ignore these differences, so I
came up with this patch, that will discard the .eh_frame section from
one of the object files if the other doesn't have one, and then retry
the object file compare.  This fixed all the boostrap-debug failures I
had.

I'm installing this in the branch.  Is this enough of a regression fix
to make it acceptable for the trunk?



[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: vta-compare-debug-ignore-eh-frame.patch --]
[-- Type: text/x-patch, Size: 2816 bytes --]

for  contrib/ChangeLog.vta
from  Alexandre Oliva  <aoliva@redhat.com>

	* compare-debug: Don't fail just because of .eh_frame differences.

Index: contrib/compare-debug
===================================================================
--- contrib/compare-debug.orig	2008-11-15 21:13:26.000000000 -0200
+++ contrib/compare-debug	2008-11-15 21:14:11.000000000 -0200
@@ -2,7 +2,7 @@
 
 # Compare stripped copies of two given object files.
 
-# Copyright (C) 2007 Free Software Foundation
+# Copyright (C) 2007, 2008 Free Software Foundation
 # Originally by Alexandre Oliva <aoliva@redhat.com>
 
 # This file is part of GCC.
@@ -67,6 +67,82 @@ if cmp "$1.$suf1" "$2.$suf2"; then
   status=0
 else
   status=1
+
+  # Assembler-generated CFI will add an .eh_frame section for -g not
+  # present in -g0.  Try to cope with it by checking that an .eh_frame
+  # section is present in either object file, and then stripping it
+  # off before re-comparing.
+
+  cmd=
+  cmp1=
+  cmp2=
+
+  for t in objdump readelf eu-readelf; do
+    if ($t --help) 2>&1 | grep -e '--\[\?section-\]\?headers' > /dev/null; then
+      cmd=$t
+
+      $cmd --section-headers "$1.$suf1" | grep '\.eh_frame' > /dev/null
+      cmp1=$?
+
+      $cmd --section-headers "$2.$suf2" | grep '\.eh_frame' > /dev/null
+      cmp2=$?
+
+      break
+    fi
+  done
+
+  # If we found .eh_frame in one but not the other, or if we could not
+  # find a command to tell, try to strip off the .eh_frame section
+  # from both.
+  if test "x$cmp1" != "x$cmp2" || test "x$cmd" = "x"; then
+    suf3=$suf1.
+    while test -f "$1.$suf3"; do
+      suf3=$suf3.
+    done
+
+    suf4=$suf2.
+    while test -f "$2.$suf4"; do
+      suf4=$suf4.
+    done
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2" "$1.$suf3" "$2.$suf4"' 0 1 2 15
+
+    if (objcopy -v) 2>&1 | grep -e "--remove-section" > /dev/null; then
+      objcopy --remove-section .eh_frame "$1.$suf1" "$1.$suf3"
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      objcopy --remove-section .eh_frame "$2.$suf2" "$2.$suf4"
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    elif (strip --help) 2>&1 | grep -e --remove-section > /dev/null; then
+      cp "$1.$suf1" "$1.$suf3"
+      strip --remove-section .eh_frame "$1.$suf3"
+
+      cp "$2.$suf2" "$2.$suf4"
+      strip --remove-section .eh_frame "$2.$suf4"
+
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    fi
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2"' 0 1 2 15
+  fi
+
+  if test "x$cmp1" != "x$cmp2" && cmp "$1.$suf1" "$2.$suf2"; then
+    status=0
+  else
+    status=1
+  fi
 fi
 
 $rm "$1.$suf1" "$2.$suf2"

[-- Attachment #3: Type: text/plain, Size: 257 bytes --]


-- 
Alexandre Oliva         http://www.lsd.ic.unicamp.br/~oliva/
Free Software Evangelist  oliva@{lsd.ic.unicamp.br, gnu.org}
FSFLA Board Member       ¡Sé Libre! => http://www.fsfla.org/
Red Hat Compiler Engineer   aoliva@{redhat.com, gcc.gnu.org}

[trunk<-vta] Re: [vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[Alexandre Oliva]

[-- Attachment #1: Type: text/plain, Size: 1202 bytes --]

On Nov 17, 2008, Alexandre Oliva <aoliva@redhat.com> wrote:

> bootstrap-debug is broken in trunk and vta branch on machines that
> support assembler-generated CFI and that don't generate CFI by ABI
> requirements.  I got bootstrap-debug failures on at least i686 and
> ppc64, because compilations with -g0 didn't have an .eh_frame section,
> whereas such trivial translation units as ada/final got a simple
> .eh_frame section with -g2, because .cfi_startproc and .cfi_endproc
> are enough to cause the generation of .eh_frame in addition to debug
> info.

> I suppose enabling the assembler to tell whether or not we're
> interested in .eh_frame, in addition to .debug_frame, would be a win
> in terms of stripped binary size, but I wasn't sure it was worth the
> effort.  Thoughts?

> In the meantime, I needed some way to ignore these differences, so I
> came up with this patch, that will discard the .eh_frame section from
> one of the object files if the other doesn't have one, and then retry
> the object file compare.  This fixed all the boostrap-debug failures I
> had.

> I'm installing this in the branch.  Is this enough of a regression fix
> to make it acceptable for the trunk?

Ping?


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: vta-compare-debug-ignore-eh-frame.patch --]
[-- Type: text/x-patch, Size: 2812 bytes --]

for  contrib/ChangeLog
from  Alexandre Oliva  <aoliva@redhat.com>

	* compare-debug: Don't fail just because of .eh_frame differences.

Index: contrib/compare-debug
===================================================================
--- contrib/compare-debug.orig	2008-11-15 21:13:26.000000000 -0200
+++ contrib/compare-debug	2008-11-15 21:14:11.000000000 -0200
@@ -2,7 +2,7 @@
 
 # Compare stripped copies of two given object files.
 
-# Copyright (C) 2007 Free Software Foundation
+# Copyright (C) 2007, 2008 Free Software Foundation
 # Originally by Alexandre Oliva <aoliva@redhat.com>
 
 # This file is part of GCC.
@@ -67,6 +67,82 @@ if cmp "$1.$suf1" "$2.$suf2"; then
   status=0
 else
   status=1
+
+  # Assembler-generated CFI will add an .eh_frame section for -g not
+  # present in -g0.  Try to cope with it by checking that an .eh_frame
+  # section is present in either object file, and then stripping it
+  # off before re-comparing.
+
+  cmd=
+  cmp1=
+  cmp2=
+
+  for t in objdump readelf eu-readelf; do
+    if ($t --help) 2>&1 | grep -e '--\[\?section-\]\?headers' > /dev/null; then
+      cmd=$t
+
+      $cmd --section-headers "$1.$suf1" | grep '\.eh_frame' > /dev/null
+      cmp1=$?
+
+      $cmd --section-headers "$2.$suf2" | grep '\.eh_frame' > /dev/null
+      cmp2=$?
+
+      break
+    fi
+  done
+
+  # If we found .eh_frame in one but not the other, or if we could not
+  # find a command to tell, try to strip off the .eh_frame section
+  # from both.
+  if test "x$cmp1" != "x$cmp2" || test "x$cmd" = "x"; then
+    suf3=$suf1.
+    while test -f "$1.$suf3"; do
+      suf3=$suf3.
+    done
+
+    suf4=$suf2.
+    while test -f "$2.$suf4"; do
+      suf4=$suf4.
+    done
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2" "$1.$suf3" "$2.$suf4"' 0 1 2 15
+
+    if (objcopy -v) 2>&1 | grep -e "--remove-section" > /dev/null; then
+      objcopy --remove-section .eh_frame "$1.$suf1" "$1.$suf3"
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      objcopy --remove-section .eh_frame "$2.$suf2" "$2.$suf4"
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    elif (strip --help) 2>&1 | grep -e --remove-section > /dev/null; then
+      cp "$1.$suf1" "$1.$suf3"
+      strip --remove-section .eh_frame "$1.$suf3"
+
+      cp "$2.$suf2" "$2.$suf4"
+      strip --remove-section .eh_frame "$2.$suf4"
+
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    fi
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2"' 0 1 2 15
+  fi
+
+  if test "x$cmp1" != "x$cmp2" && cmp "$1.$suf1" "$2.$suf2"; then
+    status=0
+  else
+    status=1
+  fi
 fi
 
 $rm "$1.$suf1" "$2.$suf2"

[-- Attachment #3: Type: text/plain, Size: 257 bytes --]


-- 
Alexandre Oliva, freedom fighter    http://FSFLA.org/~lxoliva/
You must be the change you wish to see in the world. -- Gandhi
Be Free! -- http://FSFLA.org/   FSF Latin America board member
Free Software Evangelist      Red Hat Brazil Compiler Engineer

Re: [trunk<-vta] Re: [vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[Ralf Wildenhues]

Hello Alexandre,

* Alexandre Oliva wrote on Mon, Jun 01, 2009 at 10:16:16AM CEST:
> --- contrib/compare-debug.orig	2008-11-15 21:13:26.000000000 -0200
> +++ contrib/compare-debug	2008-11-15 21:14:11.000000000 -0200

> +  for t in objdump readelf eu-readelf; do
> +    if ($t --help) 2>&1 | grep -e '--\[\?section-\]\?headers' > /dev/null; then

\? is not portable BRE, but \{0,1\} should be.

Cheers,
Ralf

Re: [trunk<-vta] Re: [vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[Richard Henderson]

 > * compare-debug: Don't fail just because of .eh_frame differences.

Ok.


r~

Re: [trunk<-vta] Re: [vta, trunk?] ignore .cfi_*-induced .eh_frame compare-debug differences

[Alexandre Oliva]

[-- Attachment #1: Type: text/plain, Size: 393 bytes --]

On Jun  1, 2009, Ralf Wildenhues <Ralf.Wildenhues@gmx.de> wrote:

> Hello Alexandre,
>> +    if ($t --help) 2>&1 | grep -e '--\[\?section-\]\?headers' > /dev/null; then

> \? is not portable BRE, but \{0,1\} should be.

Uhh.  Indeed.  Thanks for the reminder.  I ended up replacing \? with *,
which works fine for this purpose and is as portable as it gets.

Here's what I've just installed.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: vta-compare-debug-ignore-eh-frame.patch --]
[-- Type: text/x-patch, Size: 2816 bytes --]

for  contrib/ChangeLog
from  Alexandre Oliva  <aoliva@redhat.com>

	* compare-debug: Don't fail just because of .eh_frame differences.

Index: contrib/compare-debug
===================================================================
--- contrib/compare-debug.orig	2008-11-15 21:13:26.000000000 -0200
+++ contrib/compare-debug	2008-11-15 21:14:11.000000000 -0200
@@ -2,7 +2,7 @@
 
 # Compare stripped copies of two given object files.
 
-# Copyright (C) 2007 Free Software Foundation
+# Copyright (C) 2007, 2008, 2009 Free Software Foundation
 # Originally by Alexandre Oliva <aoliva@redhat.com>
 
 # This file is part of GCC.
@@ -67,6 +67,82 @@ if cmp "$1.$suf1" "$2.$suf2"; then
   status=0
 else
   status=1
+
+  # Assembler-generated CFI will add an .eh_frame section for -g not
+  # present in -g0.  Try to cope with it by checking that an .eh_frame
+  # section is present in either object file, and then stripping it
+  # off before re-comparing.
+
+  cmd=
+  cmp1=
+  cmp2=
+
+  for t in objdump readelf eu-readelf; do
+    if ($t --help) 2>&1 | grep -e '--\[*section-\]*headers' > /dev/null; then
+      cmd=$t
+
+      $cmd --section-headers "$1.$suf1" | grep '\.eh_frame' > /dev/null
+      cmp1=$?
+
+      $cmd --section-headers "$2.$suf2" | grep '\.eh_frame' > /dev/null
+      cmp2=$?
+
+      break
+    fi
+  done
+
+  # If we found .eh_frame in one but not the other, or if we could not
+  # find a command to tell, try to strip off the .eh_frame section
+  # from both.
+  if test "x$cmp1" != "x$cmp2" || test "x$cmd" = "x"; then
+    suf3=$suf1.
+    while test -f "$1.$suf3"; do
+      suf3=$suf3.
+    done
+
+    suf4=$suf2.
+    while test -f "$2.$suf4"; do
+      suf4=$suf4.
+    done
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2" "$1.$suf3" "$2.$suf4"' 0 1 2 15
+
+    if (objcopy -v) 2>&1 | grep -e "--remove-section" > /dev/null; then
+      objcopy --remove-section .eh_frame "$1.$suf1" "$1.$suf3"
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      objcopy --remove-section .eh_frame "$2.$suf2" "$2.$suf4"
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    elif (strip --help) 2>&1 | grep -e --remove-section > /dev/null; then
+      cp "$1.$suf1" "$1.$suf3"
+      strip --remove-section .eh_frame "$1.$suf3"
+
+      cp "$2.$suf2" "$2.$suf4"
+      strip --remove-section .eh_frame "$2.$suf4"
+
+      cmp "$1.$suf1" "$1.$suf3" > /dev/null
+      cmp1=$?
+
+      cmp "$2.$suf2" "$2.$suf4" > /dev/null
+      cmp2=$?
+
+      mv "$1.$suf3" "$1.$suf1"
+      mv "$2.$suf4" "$2.$suf2"
+    fi
+
+    trap 'rm -f "$1.$suf1" "$2.$suf2"' 0 1 2 15
+  fi
+
+  if test "x$cmp1" != "x$cmp2" && cmp "$1.$suf1" "$2.$suf2"; then
+    status=0
+  else
+    status=1
+  fi
 fi
 
 $rm "$1.$suf1" "$2.$suf2"

[-- Attachment #3: Type: text/plain, Size: 257 bytes --]


-- 
Alexandre Oliva, freedom fighter    http://FSFLA.org/~lxoliva/
You must be the change you wish to see in the world. -- Gandhi
Be Free! -- http://FSFLA.org/   FSF Latin America board member
Free Software Evangelist      Red Hat Brazil Compiler Engineer