[PATCH v2] Don't run personality syscall at configure time; don't check it at all

[PATCH v2] Don't run personality syscall at configure time; don't check it at all

[Pedro Alves]

v2: whoops, fixed commit log.

Currently, in order to tell whether support for disabling address
space randomization on Linux is available, GDB checks if the
personality syscall works, at configure time.  I.e., it does a run
test, instead of a compile/link test:

  AC_RUN_IFELSE([PERSONALITY_TEST],
		[have_personality=true],
		[have_personality=false],

This is a bit bogus, because the machine the build is done on may not
(and is when you consider distro gdbs) be the machine that eventually
runs gdb.  It would be better if this were a compile/link test
instead, and then at runtime, GDB coped with the personality syscall
failing.  Actually, GDB already copes.

One environment where this is problematic is building GDB in a Docker
container -- by default, Docker runs the container with seccomp, with
a profile that disables the personality syscall.  You can tell Docker
to use a less restricted seccomp profile, but I think we should just
fix it in GDB.

"man 2 personality" says:

       This system call first appeared in Linux 1.1.20 (and thus first
       in a stable kernel release with Linux 1.2.0); library support
       was added in glibc 2.3.

...

       ADDR_NO_RANDOMIZE (since Linux 2.6.12)
              With this flag set, disable address-space-layout randomization.

glibc 2.3 was released in 2002.
Linux 2.6.12 was released in 2005.

The original patch that added the configure checks was submitted in
2008.  The first version of the patch that was submitted to the list
called personality from common code:

 https://sourceware.org/pipermail/gdb-patches/2008-June/058204.html

and then was moved to Linux-specific code:

 https://sourceware.org/pipermail/gdb-patches/2008-June/058209.html

Since HAVE_PERSONALITY is only checked in Linux code, and
ADDR_NO_RANDOMIZE exists for over 15 years, I propose just completely
removing the configure checks.

If for some odd reason, some remotely modern system still needs a
configure check, then we can revert this commit but drop the
AC_RUN_IFELSE in favor of always doing the AC_LINK_IFELSE
cross-compile fallback.

gdb/ChangeLog:

	* linux-nat.c (linux_nat_target::supports_disable_randomization):
	Remove references to HAVE_PERSONALITY.
	* nat/linux-personality.c: Remove references to HAVE_PERSONALITY.
	(maybe_disable_address_space_randomization)
	(~maybe_disable_address_space_randomizatio): Remove references to
	HAVE_PERSONALITY.
	* config.in, configure: Regenerate.

gdbserver/ChangeLog:

	* linux-low.cc: Remove references to HAVE_PERSONALITY.
	(linux_process_target::supports_disable_randomization): Remove
	reference to HAVE_PERSONALITY.
	* config.in, configure: Regenerate.

gdbsupport/ChangeLog:

	* common.m4 (personality test): Remove.
---
 gdb/config.in               |  7 ----
 gdb/configure               | 74 -------------------------------------
 gdb/linux-nat.c             |  4 --
 gdb/nat/linux-personality.c | 12 +-----
 gdbserver/config.in         |  7 ----
 gdbserver/configure         | 74 -------------------------------------
 gdbserver/linux-low.cc      | 11 +-----
 gdbsupport/common.m4        | 23 ------------
 8 files changed, 2 insertions(+), 210 deletions(-)

diff --git a/gdb/config.in b/gdb/config.in
index 5d10377b736..99c924f9ba0 100644
--- a/gdb/config.in
+++ b/gdb/config.in
@@ -96,10 +96,6 @@
 /* define if the compiler supports basic C++11 syntax */
 #undef HAVE_CXX11
 
-/* Define to 1 if you have the declaration of `ADDR_NO_RANDOMIZE', and to 0 if
-   you don't. */
-#undef HAVE_DECL_ADDR_NO_RANDOMIZE
-
 /* Define to 1 if you have the declaration of `asprintf', and to 0 if you
    don't. */
 #undef HAVE_DECL_ASPRINTF
@@ -301,9 +297,6 @@
 /* Define to 1 if you have the <nlist.h> header file. */
 #undef HAVE_NLIST_H
 
-/* Define if you support the personality syscall. */
-#undef HAVE_PERSONALITY
-
 /* Define to 1 if you have the `pipe' function. */
 #undef HAVE_PIPE
 
diff --git a/gdb/configure b/gdb/configure
index b47de77bf64..cdc112e10dc 100755
--- a/gdb/configure
+++ b/gdb/configure
@@ -13819,80 +13819,6 @@ fi
 done
 
 
-      ac_fn_c_check_decl "$LINENO" "ADDR_NO_RANDOMIZE" "ac_cv_have_decl_ADDR_NO_RANDOMIZE" "#include <sys/personality.h>
-"
-if test "x$ac_cv_have_decl_ADDR_NO_RANDOMIZE" = xyes; then :
-  ac_have_decl=1
-else
-  ac_have_decl=0
-fi
-
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_ADDR_NO_RANDOMIZE $ac_have_decl
-_ACEOF
-
-
-  if test "$cross_compiling" = yes; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <sys/personality.h>
-int
-main ()
-{
-
-  #      if !HAVE_DECL_ADDR_NO_RANDOMIZE
-  #       define ADDR_NO_RANDOMIZE 0x0040000
-  #      endif
-	 /* Test the flag could be set and stays set.  */
-	 personality (personality (0xffffffff) | ADDR_NO_RANDOMIZE);
-	 if (!(personality (personality (0xffffffff)) & ADDR_NO_RANDOMIZE))
-	     return 1
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  have_personality=true
-else
-  have_personality=false
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <sys/personality.h>
-int
-main ()
-{
-
-  #      if !HAVE_DECL_ADDR_NO_RANDOMIZE
-  #       define ADDR_NO_RANDOMIZE 0x0040000
-  #      endif
-	 /* Test the flag could be set and stays set.  */
-	 personality (personality (0xffffffff) | ADDR_NO_RANDOMIZE);
-	 if (!(personality (personality (0xffffffff)) & ADDR_NO_RANDOMIZE))
-	     return 1
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  have_personality=true
-else
-  have_personality=false
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-  if $have_personality
-  then
-
-$as_echo "#define HAVE_PERSONALITY 1" >>confdefs.h
-
-  fi
-
   ac_fn_c_check_decl "$LINENO" "strstr" "ac_cv_have_decl_strstr" "$ac_includes_default"
 if test "x$ac_cv_have_decl_strstr" = xyes; then :
   ac_have_decl=1
diff --git a/gdb/linux-nat.c b/gdb/linux-nat.c
index c45e335a762..e595c71c3d3 100644
--- a/gdb/linux-nat.c
+++ b/gdb/linux-nat.c
@@ -4022,11 +4022,7 @@ linux_nat_target::supports_multi_process ()
 bool
 linux_nat_target::supports_disable_randomization ()
 {
-#ifdef HAVE_PERSONALITY
   return true;
-#else
-  return false;
-#endif
 }
 
 /* SIGCHLD handler that serves two purposes: In non-stop/async mode,
diff --git a/gdb/nat/linux-personality.c b/gdb/nat/linux-personality.c
index 87d19218626..9ce345bc6de 100644
--- a/gdb/nat/linux-personality.c
+++ b/gdb/nat/linux-personality.c
@@ -20,12 +20,7 @@
 #include "gdbsupport/common-defs.h"
 #include "nat/linux-personality.h"
 
-#ifdef HAVE_PERSONALITY
-# include <sys/personality.h>
-# if !HAVE_DECL_ADDR_NO_RANDOMIZE
-#  define ADDR_NO_RANDOMIZE 0x0040000
-# endif /* ! HAVE_DECL_ADDR_NO_RANDOMIZE */
-#endif /* HAVE_PERSONALITY */
+#include <sys/personality.h>
 
 /* See comment on nat/linux-personality.h.  */
 
@@ -34,7 +29,6 @@ maybe_disable_address_space_randomization (int disable_randomization)
   : m_personality_set (false),
     m_personality_orig (0)
 {
-#ifdef HAVE_PERSONALITY
   if (disable_randomization)
     {
       errno = 0;
@@ -49,14 +43,11 @@ maybe_disable_address_space_randomization (int disable_randomization)
 	warning (_("Error disabling address space randomization: %s"),
 		 safe_strerror (errno));
     }
-#endif /* HAVE_PERSONALITY */
 }
 
 maybe_disable_address_space_randomization::
 ~maybe_disable_address_space_randomization ()
 {
-#ifdef HAVE_PERSONALITY
-
   if (m_personality_set)
     {
       errno = 0;
@@ -65,5 +56,4 @@ maybe_disable_address_space_randomization::
 	warning (_("Error restoring address space randomization: %s"),
 		 safe_strerror (errno));
     }
-#endif /* HAVE_PERSONALITY */
 }
diff --git a/gdbserver/config.in b/gdbserver/config.in
index 99391b4acff..611bfd7aa76 100644
--- a/gdbserver/config.in
+++ b/gdbserver/config.in
@@ -31,10 +31,6 @@
 /* define if the compiler supports basic C++11 syntax */
 #undef HAVE_CXX11
 
-/* Define to 1 if you have the declaration of `ADDR_NO_RANDOMIZE', and to 0 if
-   you don't. */
-#undef HAVE_DECL_ADDR_NO_RANDOMIZE
-
 /* Define to 1 if you have the declaration of `asprintf', and to 0 if you
    don't. */
 #undef HAVE_DECL_ASPRINTF
@@ -182,9 +178,6 @@
 /* Define to 1 if you have the <netinet/tcp.h> header file. */
 #undef HAVE_NETINET_TCP_H
 
-/* Define if you support the personality syscall. */
-#undef HAVE_PERSONALITY
-
 /* Define to 1 if you have the `pipe' function. */
 #undef HAVE_PIPE
 
diff --git a/gdbserver/configure b/gdbserver/configure
index 032b4ae65bb..aab72c0b8c5 100755
--- a/gdbserver/configure
+++ b/gdbserver/configure
@@ -7130,80 +7130,6 @@ fi
 done
 
 
-      ac_fn_c_check_decl "$LINENO" "ADDR_NO_RANDOMIZE" "ac_cv_have_decl_ADDR_NO_RANDOMIZE" "#include <sys/personality.h>
-"
-if test "x$ac_cv_have_decl_ADDR_NO_RANDOMIZE" = xyes; then :
-  ac_have_decl=1
-else
-  ac_have_decl=0
-fi
-
-cat >>confdefs.h <<_ACEOF
-#define HAVE_DECL_ADDR_NO_RANDOMIZE $ac_have_decl
-_ACEOF
-
-
-  if test "$cross_compiling" = yes; then :
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <sys/personality.h>
-int
-main ()
-{
-
-  #      if !HAVE_DECL_ADDR_NO_RANDOMIZE
-  #       define ADDR_NO_RANDOMIZE 0x0040000
-  #      endif
-	 /* Test the flag could be set and stays set.  */
-	 personality (personality (0xffffffff) | ADDR_NO_RANDOMIZE);
-	 if (!(personality (personality (0xffffffff)) & ADDR_NO_RANDOMIZE))
-	     return 1
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_link "$LINENO"; then :
-  have_personality=true
-else
-  have_personality=false
-fi
-rm -f core conftest.err conftest.$ac_objext \
-    conftest$ac_exeext conftest.$ac_ext
-else
-  cat confdefs.h - <<_ACEOF >conftest.$ac_ext
-/* end confdefs.h.  */
-#include <sys/personality.h>
-int
-main ()
-{
-
-  #      if !HAVE_DECL_ADDR_NO_RANDOMIZE
-  #       define ADDR_NO_RANDOMIZE 0x0040000
-  #      endif
-	 /* Test the flag could be set and stays set.  */
-	 personality (personality (0xffffffff) | ADDR_NO_RANDOMIZE);
-	 if (!(personality (personality (0xffffffff)) & ADDR_NO_RANDOMIZE))
-	     return 1
-  ;
-  return 0;
-}
-_ACEOF
-if ac_fn_c_try_run "$LINENO"; then :
-  have_personality=true
-else
-  have_personality=false
-fi
-rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
-  conftest.$ac_objext conftest.beam conftest.$ac_ext
-fi
-
-  if $have_personality
-  then
-
-$as_echo "#define HAVE_PERSONALITY 1" >>confdefs.h
-
-  fi
-
   ac_fn_c_check_decl "$LINENO" "strstr" "ac_cv_have_decl_strstr" "$ac_includes_default"
 if test "x$ac_cv_have_decl_strstr" = xyes; then :
   ac_have_decl=1
diff --git a/gdbserver/linux-low.cc b/gdbserver/linux-low.cc
index 251a54f5c4a..27556ffb88a 100644
--- a/gdbserver/linux-low.cc
+++ b/gdbserver/linux-low.cc
@@ -60,12 +60,7 @@
 #endif
 #include "nat/linux-namespaces.h"
 
-#ifdef HAVE_PERSONALITY
-# include <sys/personality.h>
-# if !HAVE_DECL_ADDR_NO_RANDOMIZE
-#  define ADDR_NO_RANDOMIZE 0x0040000
-# endif
-#endif
+#include <sys/personality.h>
 
 #ifndef O_LARGEFILE
 #define O_LARGEFILE 0
@@ -6231,11 +6226,7 @@ linux_process_target::core_of_thread (ptid_t ptid)
 bool
 linux_process_target::supports_disable_randomization ()
 {
-#ifdef HAVE_PERSONALITY
   return true;
-#else
-  return false;
-#endif
 }
 
 bool
diff --git a/gdbsupport/common.m4 b/gdbsupport/common.m4
index 3ed95dabc85..2e709dbbdbb 100644
--- a/gdbsupport/common.m4
+++ b/gdbsupport/common.m4
@@ -55,29 +55,6 @@ AC_DEFUN([GDB_AC_COMMON], [
 		  ptrace64 sbrk setns sigaltstack sigprocmask \
 		  setpgid setpgrp getrusage getauxval])
 
-  dnl Check if we can disable the virtual address space randomization.
-  dnl The functionality of setarch -R.
-  AC_CHECK_DECLS([ADDR_NO_RANDOMIZE],,, [#include <sys/personality.h>])
-  define([PERSONALITY_TEST], [AC_LANG_PROGRAM([#include <sys/personality.h>], [
-  #      if !HAVE_DECL_ADDR_NO_RANDOMIZE
-  #       define ADDR_NO_RANDOMIZE 0x0040000
-  #      endif
-	 /* Test the flag could be set and stays set.  */
-	 personality (personality (0xffffffff) | ADDR_NO_RANDOMIZE);
-	 if (!(personality (personality (0xffffffff)) & ADDR_NO_RANDOMIZE))
-	     return 1])])
-  AC_RUN_IFELSE([PERSONALITY_TEST],
-		[have_personality=true],
-		[have_personality=false],
-		[AC_LINK_IFELSE([PERSONALITY_TEST],
-				[have_personality=true],
-				[have_personality=false])])
-  if $have_personality
-  then
-      AC_DEFINE([HAVE_PERSONALITY], 1,
-		[Define if you support the personality syscall.])
-  fi
-
   AC_CHECK_DECLS([strstr])
 
   # ----------------------- #

base-commit: ec66d6ea54679d00c1f0d52a5a9fa0cbb8b86d47
-- 
2.26.2

Re: [PATCH v2] Don't run personality syscall at configure time; don't check it at all

[Simon Marchi]

On 2021-05-06 6:47 p.m., Pedro Alves wrote:
> v2: whoops, fixed commit log.
> 
> Currently, in order to tell whether support for disabling address
> space randomization on Linux is available, GDB checks if the
> personality syscall works, at configure time.  I.e., it does a run
> test, instead of a compile/link test:
> 
>   AC_RUN_IFELSE([PERSONALITY_TEST],
> 		[have_personality=true],
> 		[have_personality=false],
> 
> This is a bit bogus, because the machine the build is done on may not
> (and is when you consider distro gdbs) be the machine that eventually
> runs gdb.  It would be better if this were a compile/link test
> instead, and then at runtime, GDB coped with the personality syscall
> failing.  Actually, GDB already copes.
> 
> One environment where this is problematic is building GDB in a Docker
> container -- by default, Docker runs the container with seccomp, with
> a profile that disables the personality syscall.  You can tell Docker
> to use a less restricted seccomp profile, but I think we should just
> fix it in GDB.
> 
> "man 2 personality" says:
> 
>        This system call first appeared in Linux 1.1.20 (and thus first
>        in a stable kernel release with Linux 1.2.0); library support
>        was added in glibc 2.3.
> 
> ...
> 
>        ADDR_NO_RANDOMIZE (since Linux 2.6.12)
>               With this flag set, disable address-space-layout randomization.
> 
> glibc 2.3 was released in 2002.
> Linux 2.6.12 was released in 2005.
> 
> The original patch that added the configure checks was submitted in
> 2008.  The first version of the patch that was submitted to the list
> called personality from common code:
> 
>  https://sourceware.org/pipermail/gdb-patches/2008-June/058204.html
> 
> and then was moved to Linux-specific code:
> 
>  https://sourceware.org/pipermail/gdb-patches/2008-June/058209.html
> 
> Since HAVE_PERSONALITY is only checked in Linux code, and
> ADDR_NO_RANDOMIZE exists for over 15 years, I propose just completely
> removing the configure checks.
> 
> If for some odd reason, some remotely modern system still needs a
> configure check, then we can revert this commit but drop the
> AC_RUN_IFELSE in favor of always doing the AC_LINK_IFELSE
> cross-compile fallback.
> 
> gdb/ChangeLog:
> 
> 	* linux-nat.c (linux_nat_target::supports_disable_randomization):
> 	Remove references to HAVE_PERSONALITY.
> 	* nat/linux-personality.c: Remove references to HAVE_PERSONALITY.
> 	(maybe_disable_address_space_randomization)
> 	(~maybe_disable_address_space_randomizatio): Remove references to
> 	HAVE_PERSONALITY.
> 	* config.in, configure: Regenerate.
> 
> gdbserver/ChangeLog:
> 
> 	* linux-low.cc: Remove references to HAVE_PERSONALITY.
> 	(linux_process_target::supports_disable_randomization): Remove
> 	reference to HAVE_PERSONALITY.
> 	* config.in, configure: Regenerate.
> 
> gdbsupport/ChangeLog:
> 
> 	* common.m4 (personality test): Remove.

Hi Pedro,

The patch LGTM.

However, I have some trouble re-generating gdb/configure, so I can't
verify that I get the same result as you, I'd like if we could quickly
fix that first.  With current master, I get:

    gdb/ $ autoreconf -vf
    autoreconf: Entering directory `.'
    autoreconf: configure.ac: not using Gettext
    autoreconf: running: aclocal --force 
    autoreconf: configure.ac: tracing
    autoreconf: configure.ac: adding subdirectory testsuite to autoreconf
    autoreconf: Entering directory `testsuite'
    autoreconf: configure.ac: not using aclocal
    autoreconf: configure.ac: subdirectory gdb.gdbtk not present
    autoreconf: configure.ac: not using Libtool
    autoreconf: running: /opt/autostuff/2.69/bin/autoconf --force
    autoreconf: configure.ac: not using Autoheader
    autoreconf: configure.ac: not using Automake
    autoreconf: Leaving directory `testsuite'
    autoreconf: configure.ac: subdirectory gdbtk not present
    configure.ac:161: error: possibly undefined macro: AC_MSG_ERROR
          If this token and others are legitimate, please use m4_pattern_allow.
          See the Autoconf documentation.
    configure.ac:329: error: possibly undefined macro: AC_MSG_WARN
    configure.ac:409: error: possibly undefined macro: AC_DEFINE
    autoreconf: /opt/autostuff/2.69/bin/autoconf failed with exit status: 1

This happens since commit adeab0c5b33f ("config/debuginfod: do not
include pkg.m4 directly").  The commit message hints that we should run
autoconf with -I ../config, and indeed it works:

    gdb/ $ autoreconf -vf -I ../config
    autoreconf: Entering directory `.'
    autoreconf: configure.ac: not using Gettext
    autoreconf: running: aclocal -I ../config --force 
    autoreconf: configure.ac: tracing
    autoreconf: configure.ac: adding subdirectory testsuite to autoreconf
    autoreconf: Entering directory `testsuite'
    autoreconf: configure.ac: not using aclocal
    autoreconf: configure.ac: subdirectory gdb.gdbtk not present
    autoreconf: configure.ac: not using Libtool
    autoreconf: running: /opt/autostuff/2.69/bin/autoconf --include=../config --force
    autoreconf: configure.ac: not using Autoheader
    autoreconf: configure.ac: not using Automake
    autoreconf: Leaving directory `testsuite'
    autoreconf: configure.ac: subdirectory gdbtk not present
    autoreconf: running: /opt/autostuff/2.69/bin/autoheader --include=../config --force
    autoreconf: Leaving directory `.'

However, I never needed to use -I ../config before, so I don't know why
it would be needed now (and we use plenty of other macros from config/).
Do we need to list ../config/pkg.m4 in our acinclude.m4?  Or, isn't
there a way in our configure.ac to say that we always want to include
../config?

Simon

Re: [PATCH v2] Don't run personality syscall at configure time; don't check it at all

[Tom Tromey]

Simon> This happens since commit adeab0c5b33f ("config/debuginfod: do not
Simon> include pkg.m4 directly").  The commit message hints that we should run
Simon> autoconf with -I ../config, and indeed it works:

Yeah, aclocal needs to pick up the -I options.  If we used automake, it
would pick them out of Makefile.am, but since we don't I guess it
doesn't know how.

Simon> However, I never needed to use -I ../config before, so I don't know why
Simon> it would be needed now (and we use plenty of other macros from config/).
Simon> Do we need to list ../config/pkg.m4 in our acinclude.m4?

I think we could do that.

Simon> Or, isn't
Simon> there a way in our configure.ac to say that we always want to include
Simon> ../config?

This area remains under-developed AFAIK.

Tom

Re: [PATCH v2] Don't run personality syscall at configure time; don't check it at all

[Pedro Alves]

On 2021-05-07 3:32 p.m., Simon Marchi wrote:
> 
> The patch LGTM.
> 
> However, I have some trouble re-generating gdb/configure, so I can't
> verify that I get the same result as you, I'd like if we could quickly
> fix that first.  With current master, I get:

I've never gotten into the habit of using autoreconf.  I had just
run autoconf and autoheader directly.  I tried it again and it still
works fine here.

From the discussion, it doesn't look like this patch makes anything
better or worse, so I went ahead and merged it.

Pedro Alves

Re: [PATCH v2] Don't run personality syscall at configure time; don't check it at all

[Simon Marchi]

On 2021-05-07 4:37 p.m., Tom Tromey wrote:
> Simon> However, I never needed to use -I ../config before, so I don't know why
> Simon> it would be needed now (and we use plenty of other macros from config/).
> Simon> Do we need to list ../config/pkg.m4 in our acinclude.m4?
> 
> I think we could do that.

Ok.  My understanding is that this is our standard way of doing things,
otherwise we wouldn't have "../config/enable.m4" and others in
acinclude.m4.

I'll submit a patch for that.

Simon