Patchwork [RFC,2/2] Move gdb's xmalloc and friends to new file

login
register
mail settings
Submitter Tom Tromey
Date May 30, 2019, 9:30 p.m.
Message ID <20190530213046.20542-3-tom@tromey.com>
Download mbox | patch
Permalink /patch/32953/
State New
Headers show

Comments

Tom Tromey - May 30, 2019, 9:30 p.m.
When "common" becomes a library, linking will cause a symbol clash,
because "xmalloc" and some related symbols are defined in that
library, libiberty, and readline.

To work around this problem, this patch moves the clashing symbols to
a new file, which is then compiled separately for both gdb and
gdbserver.

gdb/ChangeLog
2019-05-30  Tom Tromey  <tom@tromey.com>

	* common/common-utils.c (xmalloc, xrealloc, xcalloc)
	(xmalloc_failed): Move to alloc.c.
	* alloc.c: New file.
	* Makefile.in (COMMON_SFILES): Add alloc.c.

gdb/gdbserver/ChangeLog
2019-05-30  Tom Tromey  <tom@tromey.com>

	* Makefile.in (SFILES): Add alloc.c.
	(OBS): Add alloc.o.
	(IPA_OBJS): Add alloc-ipa.o.
	(alloc-ipa.o): New target.
	(%.o: ../%.c): New pattern rule.
---
 gdb/ChangeLog             |  7 +++
 gdb/Makefile.in           |  1 +
 gdb/alloc.c               | 95 +++++++++++++++++++++++++++++++++++++++
 gdb/common/common-utils.c | 72 -----------------------------
 gdb/gdbserver/ChangeLog   |  8 ++++
 gdb/gdbserver/Makefile.in | 11 +++++
 6 files changed, 122 insertions(+), 72 deletions(-)
 create mode 100644 gdb/alloc.c
Simon Marchi - June 3, 2019, 3:03 p.m.
On 2019-05-30 5:30 p.m., Tom Tromey wrote:
> When "common" becomes a library, linking will cause a symbol clash,
> because "xmalloc" and some related symbols are defined in that
> library, libiberty, and readline.
> 
> To work around this problem, this patch moves the clashing symbols to
> a new file, which is then compiled separately for both gdb and
> gdbserver.

Hmm how does this work currently?  We have an xmalloc symbols both in
common/common-utils.o and ../libiberty/libiberty.a, why doesn't it clash?

Simon
Tom Tromey - June 3, 2019, 4:33 p.m.
>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:

Simon> On 2019-05-30 5:30 p.m., Tom Tromey wrote:
>> When "common" becomes a library, linking will cause a symbol clash,
>> because "xmalloc" and some related symbols are defined in that
>> library, libiberty, and readline.
>> 
>> To work around this problem, this patch moves the clashing symbols to
>> a new file, which is then compiled separately for both gdb and
>> gdbserver.

Simon> Hmm how does this work currently?  We have an xmalloc symbols both in
Simon> common/common-utils.o and ../libiberty/libiberty.a, why doesn't it clash?

If a symbol comes from a .o then it overrides the symbols coming from
libraries.  The error only occurs if the symbol is only provided by
multiple libraries.  The former is the case today, because we don't make
a library from common/.

Tom
Pedro Alves - June 5, 2019, 9:40 a.m.
On 5/30/19 10:30 PM, Tom Tromey wrote:
> When "common" becomes a library, linking will cause a symbol clash,
> because "xmalloc" and some related symbols are defined in that
> library, libiberty, and readline.
> 
> To work around this problem, this patch moves the clashing symbols to
> a new file, which is then compiled separately for both gdb and
> gdbserver.
> 
> gdb/ChangeLog
> 2019-05-30  Tom Tromey  <tom@tromey.com>
> 
> 	* common/common-utils.c (xmalloc, xrealloc, xcalloc)
> 	(xmalloc_failed): Move to alloc.c.
> 	* alloc.c: New file.
> 	* Makefile.in (COMMON_SFILES): Add alloc.c.
> 
> gdb/gdbserver/ChangeLog
> 2019-05-30  Tom Tromey  <tom@tromey.com>
> 
> 	* Makefile.in (SFILES): Add alloc.c.
> 	(OBS): Add alloc.o.
> 	(IPA_OBJS): Add alloc-ipa.o.
> 	(alloc-ipa.o): New target.
> 	(%.o: ../%.c): New pattern rule.

This will be the first case of gdbserver building a file
from gdb/ .  I suppose we could preserve the gdb/common/
directory for such files.  But I guess moving it out of the
way until gdb/common/ moves to top level helps.

> +++ b/gdb/alloc.c
> @@ -0,0 +1,95 @@
> +/* Shared allocation functions for GDB, the GNU debugger.
> +
> +   Copyright (C) 2019 Free Software Foundation, Inc.

The file is new, but the contents aren't, and it's the contents
that matter wrt to copyright years.

There should be some comment here about why this is in a separate file
instead of living in the common library, to help people that read
the code from the tree without having this commit in context.

> +
> +#include "common/common-defs.h"

There should also be a comment explaining why this includes common-defs.h
instead of defs.h.

> +#include "libiberty.h"
> +#include <cstdlib>

Including <cstdlib> looks strange, given common-defs.h includes stdlib.h.
Why did you need this?

> +#include "common/errors.h"
> +
> +/* The xmalloc() (libiberty.h) family of memory management routines.
> +

Thanks,
Pedro Alves
Tom Tromey - June 5, 2019, 10:33 p.m.
>>>>> "Pedro" == Pedro Alves <palves@redhat.com> writes:

Pedro> This will be the first case of gdbserver building a file
Pedro> from gdb/ .  I suppose we could preserve the gdb/common/
Pedro> directory for such files.  But I guess moving it out of the
Pedro> way until gdb/common/ moves to top level helps.

Yeah.  We can shuffle it around again later if we want.

>> +   Copyright (C) 2019 Free Software Foundation, Inc.

Pedro> The file is new, but the contents aren't, and it's the contents
Pedro> that matter wrt to copyright years.

Fixed.

Pedro> There should be some comment here about why this is in a separate file
Pedro> instead of living in the common library, to help people that read
Pedro> the code from the tree without having this commit in context.

Also fixed.

>> +#include "common/common-defs.h"

Pedro> There should also be a comment explaining why this includes common-defs.h
Pedro> instead of defs.h.

Fixed.

>> +#include "libiberty.h"
>> +#include <cstdlib>

Pedro> Including <cstdlib> looks strange, given common-defs.h includes stdlib.h.
Pedro> Why did you need this?

Removed.

I've appended the new file.

I'll probably push these patches sometime soon.

Tom

/* Shared allocation functions for GDB, the GNU debugger.

   Copyright (C) 1986-2019 Free Software Foundation, Inc.

   This file is part of GDB.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 3 of the License, or
   (at your option) any later version.

   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */

/* This file is unusual.

   Because both libiberty and readline define xmalloc and friends, the
   functions in this file can't appear in a library -- that will cause
   link errors.

   And, because we want to turn the common code into a library, this
   file can't live there.

   So, it lives in gdb and is built separately by gdb and gdbserver.
   Please be aware of this when modifying it.

   This also explains why this file includes common-defs.h and not
   defs.h or server.h -- we'd prefer to avoid depending on the
   GDBSERVER define when possible, and for this file it seemed
   simple to do so.  */

#include "common/common-defs.h"
#include "libiberty.h"
#include "common/errors.h"

/* The xmalloc() (libiberty.h) family of memory management routines.

   These are like the ISO-C malloc() family except that they implement
   consistent semantics and guard against typical memory management
   problems.  */

/* NOTE: These are declared using PTR to ensure consistency with
   "libiberty.h".  xfree() is GDB local.  */

PTR                            /* ARI: PTR */
xmalloc (size_t size)
{
  void *val;

  /* See libiberty/xmalloc.c.  This function need's to match that's
     semantics.  It never returns NULL.  */
  if (size == 0)
    size = 1;

  val = malloc (size);         /* ARI: malloc */
  if (val == NULL)
    malloc_failure (size);

  return val;
}

PTR                              /* ARI: PTR */
xrealloc (PTR ptr, size_t size)          /* ARI: PTR */
{
  void *val;

  /* See libiberty/xmalloc.c.  This function need's to match that's
     semantics.  It never returns NULL.  */
  if (size == 0)
    size = 1;

  if (ptr != NULL)
    val = realloc (ptr, size);	/* ARI: realloc */
  else
    val = malloc (size);	        /* ARI: malloc */
  if (val == NULL)
    malloc_failure (size);

  return val;
}

PTR                            /* ARI: PTR */
xcalloc (size_t number, size_t size)
{
  void *mem;

  /* See libiberty/xmalloc.c.  This function need's to match that's
     semantics.  It never returns NULL.  */
  if (number == 0 || size == 0)
    {
      number = 1;
      size = 1;
    }

  mem = calloc (number, size);      /* ARI: xcalloc */
  if (mem == NULL)
    malloc_failure (number * size);

  return mem;
}

void
xmalloc_failed (size_t size)
{
  malloc_failure (size);
}
Alan Hayward - June 17, 2019, 3:44 p.m.
Tom,

Looks like this breaks the building of alloc-ipa.o when using Make 3.81
I’ve tried this on a few different machines.

g++ -x c++  -g -O2    -I. -I../../../src/binutils-gdb/gdb/gdbserver -I../../../src/binutils-gdb/gdb/gdbserver/../regformats -I../../../src/binutils-gdb/gdb/gdbserver/.. -I../../../src/binutils-gdb/gdb/gdbserver/../../include -I../../../src/binutils-gdb/gdb/gdbserver/../../gnulib/import -Ibuild-gnulib-gdbserver/import  -Wall -Wpointer-arith -Wno-unused -Wunused-value -Wunused-variable -Wunused-function -Wno-switch -Wno-char-subscripts -Wempty-body -Wunused-but-set-parameter -Wunused-but-set-variable -Wno-sign-compare -Wno-error=maybe-uninitialized -Wsuggest-override -Wimplicit-fallthrough=3 -Wduplicated-cond -Wshadow=local -Wformat -Wformat-nonliteral -Werror -DGDBSERVER  -DCONFIG_UST_GDB_INTEGRATION -Drpl_strerror=strerror -fPIC -DIN_PROCESS_AGENT -fvisibility=hidden -c -o alloc-ipa.o -MT alloc-ipa.o -MMD -MP -MF ./.deps/alloc-ipa.Tpo `echo " -Wall -Wpointer-arith -Wno-unused -Wunused-value -Wunused-variable -Wunused-function -Wno-switch -Wno-char-subscripts -Wempty-body -Wunused-but-set-parameter -Wunused-but-set-variable -Wno-sign-compare -Wno-error=maybe-uninitialized -Wsuggest-override -Wimplicit-fallthrough=3 -Wduplicated-cond -Wshadow=local -Wformat -Wformat-nonliteral " | sed "s/ -Wformat-nonliteral / -Wno-format-nonliteral /g"`
g++: warning: ‘-x c++’ after last input file has no effect
g++: fatal error: no input files
compilation terminated.
make-3.81[4]: *** [alloc-ipa.o] Error 1
make-3.81[4]: *** Waiting for unfinished jobs....


Something has stripped the "../../../src/binutils-gdb/gdb/gdbserver/../alloc.c” from the build line.

Bizarrely, if you then run make a second time, everything works:

g++ -x c++  -g -O2    -I. -I../../../src/binutils-gdb/gdb/gdbserver -I../../../src/binutils-gdb/gdb/gdbserver/../regformats -I../../../src/binutils-gdb/gdb/gdbserver/.. -I../../../src/binutils-gdb/gdb/gdbserver/../../include -I../../../src/binutils-gdb/gdb/gdbserver/../../gnulib/import -Ibuild-gnulib-gdbserver/import  -Wall -Wpointer-arith -Wno-unused -Wunused-value -Wunused-variable -Wunused-function -Wno-switch -Wno-char-subscripts -Wempty-body -Wunused-but-set-parameter -Wunused-but-set-variable -Wno-sign-compare -Wno-error=maybe-uninitialized -Wsuggest-override -Wimplicit-fallthrough=3 -Wduplicated-cond -Wshadow=local -Wformat -Wformat-nonliteral -Werror -DGDBSERVER  -DCONFIG_UST_GDB_INTEGRATION -Drpl_strerror=strerror -fPIC -DIN_PROCESS_AGENT -fvisibility=hidden -c -o alloc-ipa.o -MT alloc-ipa.o -MMD -MP -MF ./.deps/alloc-ipa.Tpo `echo " -Wall -Wpointer-arith -Wno-unused -Wunused-value -Wunused-variable -Wunused-function -Wno-switch -Wno-char-subscripts -Wempty-body -Wunused-but-set-parameter -Wunused-but-set-variable -Wno-sign-compare -Wno-error=maybe-uninitialized -Wsuggest-override -Wimplicit-fallthrough=3 -Wduplicated-cond -Wshadow=local -Wformat -Wformat-nonliteral " | sed "s/ -Wformat-nonliteral / -Wno-format-nonliteral /g"` ../../../src/binutils-gdb/gdb/gdbserver/../alloc.c


Reverting your change fixes the issue.

I’ve no idea what’s causing it, and am assuming it’s a weird make bug.
I remember having a 3.81 issue around the same area:
a543c5ca7c1285548726e6d92ca6044dc1963340.


(Personally, I’ve no attachment to 3.81, and usually use 4.x.)



Alan.




> On 30 May 2019, at 22:30, Tom Tromey <tom@tromey.com> wrote:

> 

> When "common" becomes a library, linking will cause a symbol clash,

> because "xmalloc" and some related symbols are defined in that

> library, libiberty, and readline.

> 

> To work around this problem, this patch moves the clashing symbols to

> a new file, which is then compiled separately for both gdb and

> gdbserver.

> 

> gdb/ChangeLog

> 2019-05-30  Tom Tromey  <tom@tromey.com>

> 

> 	* common/common-utils.c (xmalloc, xrealloc, xcalloc)

> 	(xmalloc_failed): Move to alloc.c.

> 	* alloc.c: New file.

> 	* Makefile.in (COMMON_SFILES): Add alloc.c.

> 

> gdb/gdbserver/ChangeLog

> 2019-05-30  Tom Tromey  <tom@tromey.com>

> 

> 	* Makefile.in (SFILES): Add alloc.c.

> 	(OBS): Add alloc.o.

> 	(IPA_OBJS): Add alloc-ipa.o.

> 	(alloc-ipa.o): New target.

> 	(%.o: ../%.c): New pattern rule.

> ---

> gdb/ChangeLog             |  7 +++

> gdb/Makefile.in           |  1 +

> gdb/alloc.c               | 95 +++++++++++++++++++++++++++++++++++++++

> gdb/common/common-utils.c | 72 -----------------------------

> gdb/gdbserver/ChangeLog   |  8 ++++

> gdb/gdbserver/Makefile.in | 11 +++++

> 6 files changed, 122 insertions(+), 72 deletions(-)

> create mode 100644 gdb/alloc.c

> 

> diff --git a/gdb/Makefile.in b/gdb/Makefile.in

> index 0f495783600..15ec7a61b1c 100644

> --- a/gdb/Makefile.in

> +++ b/gdb/Makefile.in

> @@ -924,6 +924,7 @@ COMMON_SFILES = \

> 	ada-varobj.c \

> 	addrmap.c \

> 	agent.c \

> +	alloc.c \

> 	annotate.c \

> 	arch-utils.c \

> 	auto-load.c \

> diff --git a/gdb/alloc.c b/gdb/alloc.c

> new file mode 100644

> index 00000000000..e6c9c215435

> --- /dev/null

> +++ b/gdb/alloc.c

> @@ -0,0 +1,95 @@

> +/* Shared allocation functions for GDB, the GNU debugger.

> +

> +   Copyright (C) 2019 Free Software Foundation, Inc.

> +

> +   This file is part of GDB.

> +

> +   This program is free software; you can redistribute it and/or modify

> +   it under the terms of the GNU General Public License as published by

> +   the Free Software Foundation; either version 3 of the License, or

> +   (at your option) any later version.

> +

> +   This program is distributed in the hope that it will be useful,

> +   but WITHOUT ANY WARRANTY; without even the implied warranty of

> +   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

> +   GNU General Public License for more details.

> +

> +   You should have received a copy of the GNU General Public License

> +   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */

> +

> +#include "common/common-defs.h"

> +#include "libiberty.h"

> +#include <cstdlib>

> +#include "common/errors.h"

> +

> +/* The xmalloc() (libiberty.h) family of memory management routines.

> +

> +   These are like the ISO-C malloc() family except that they implement

> +   consistent semantics and guard against typical memory management

> +   problems.  */

> +

> +/* NOTE: These are declared using PTR to ensure consistency with

> +   "libiberty.h".  xfree() is GDB local.  */

> +

> +PTR                            /* ARI: PTR */

> +xmalloc (size_t size)

> +{

> +  void *val;

> +

> +  /* See libiberty/xmalloc.c.  This function need's to match that's

> +     semantics.  It never returns NULL.  */

> +  if (size == 0)

> +    size = 1;

> +

> +  val = malloc (size);         /* ARI: malloc */

> +  if (val == NULL)

> +    malloc_failure (size);

> +

> +  return val;

> +}

> +

> +PTR                              /* ARI: PTR */

> +xrealloc (PTR ptr, size_t size)          /* ARI: PTR */

> +{

> +  void *val;

> +

> +  /* See libiberty/xmalloc.c.  This function need's to match that's

> +     semantics.  It never returns NULL.  */

> +  if (size == 0)

> +    size = 1;

> +

> +  if (ptr != NULL)

> +    val = realloc (ptr, size);	/* ARI: realloc */

> +  else

> +    val = malloc (size);	        /* ARI: malloc */

> +  if (val == NULL)

> +    malloc_failure (size);

> +

> +  return val;

> +}

> +

> +PTR                            /* ARI: PTR */

> +xcalloc (size_t number, size_t size)

> +{

> +  void *mem;

> +

> +  /* See libiberty/xmalloc.c.  This function need's to match that's

> +     semantics.  It never returns NULL.  */

> +  if (number == 0 || size == 0)

> +    {

> +      number = 1;

> +      size = 1;

> +    }

> +

> +  mem = calloc (number, size);      /* ARI: xcalloc */

> +  if (mem == NULL)

> +    malloc_failure (number * size);

> +

> +  return mem;

> +}

> +

> +void

> +xmalloc_failed (size_t size)

> +{

> +  malloc_failure (size);

> +}

> diff --git a/gdb/common/common-utils.c b/gdb/common/common-utils.c

> index 74ca93810c7..dd839a0d4d1 100644

> --- a/gdb/common/common-utils.c

> +++ b/gdb/common/common-utils.c

> @@ -22,84 +22,12 @@

> #include "host-defs.h"

> #include <ctype.h>

> 

> -/* The xmalloc() (libiberty.h) family of memory management routines.

> -

> -   These are like the ISO-C malloc() family except that they implement

> -   consistent semantics and guard against typical memory management

> -   problems.  */

> -

> -/* NOTE: These are declared using PTR to ensure consistency with

> -   "libiberty.h".  xfree() is GDB local.  */

> -

> -PTR                            /* ARI: PTR */

> -xmalloc (size_t size)

> -{

> -  void *val;

> -

> -  /* See libiberty/xmalloc.c.  This function need's to match that's

> -     semantics.  It never returns NULL.  */

> -  if (size == 0)

> -    size = 1;

> -

> -  val = malloc (size);         /* ARI: malloc */

> -  if (val == NULL)

> -    malloc_failure (size);

> -

> -  return val;

> -}

> -

> -PTR                              /* ARI: PTR */

> -xrealloc (PTR ptr, size_t size)          /* ARI: PTR */

> -{

> -  void *val;

> -

> -  /* See libiberty/xmalloc.c.  This function need's to match that's

> -     semantics.  It never returns NULL.  */

> -  if (size == 0)

> -    size = 1;

> -

> -  if (ptr != NULL)

> -    val = realloc (ptr, size);	/* ARI: realloc */

> -  else

> -    val = malloc (size);	        /* ARI: malloc */

> -  if (val == NULL)

> -    malloc_failure (size);

> -

> -  return val;

> -}

> -

> -PTR                            /* ARI: PTR */           

> -xcalloc (size_t number, size_t size)

> -{

> -  void *mem;

> -

> -  /* See libiberty/xmalloc.c.  This function need's to match that's

> -     semantics.  It never returns NULL.  */

> -  if (number == 0 || size == 0)

> -    {

> -      number = 1;

> -      size = 1;

> -    }

> -

> -  mem = calloc (number, size);      /* ARI: xcalloc */

> -  if (mem == NULL)

> -    malloc_failure (number * size);

> -

> -  return mem;

> -}

> -

> void *

> xzalloc (size_t size)

> {

>   return xcalloc (1, size);

> }

> 

> -void

> -xmalloc_failed (size_t size)

> -{

> -  malloc_failure (size);

> -}

> -

> /* Like asprintf/vasprintf but get an internal_error if the call

>    fails. */

> 

> diff --git a/gdb/gdbserver/Makefile.in b/gdb/gdbserver/Makefile.in

> index f5fc55034ee..a0793e657c5 100644

> --- a/gdb/gdbserver/Makefile.in

> +++ b/gdb/gdbserver/Makefile.in

> @@ -197,6 +197,7 @@ SFILES = \

> 	$(srcdir)/arch/arm-get-next-pcs.c \

> 	$(srcdir)/arch/arm-linux.c \

> 	$(srcdir)/arch/ppc-linux-common.c \

> +	$(srcdir)/../alloc.c \

> 	$(srcdir)/common/btrace-common.c \

> 	$(srcdir)/common/buffer.c \

> 	$(srcdir)/common/cleanups.c \

> @@ -238,6 +239,7 @@ SOURCES = $(SFILES)

> TAGFILES = $(SOURCES) ${HFILES} ${ALLPARAM} ${POSSLIBS}

> 

> OBS = \

> +	alloc.o \

> 	ax.o \

> 	common/agent.o \

> 	common/btrace-common.o \

> @@ -413,6 +415,7 @@ gdbreplay$(EXEEXT): $(sort $(GDBREPLAY_OBS)) $(LIBGNU) $(LIBIBERTY)

> 		$(LIBIBERTY)

> 

> IPA_OBJS = \

> +	alloc-ipa.o \

> 	ax-ipa.o \

> 	common/common-utils-ipa.o \

> 	common/errors-ipa.o \

> @@ -568,6 +571,10 @@ ax.o: ax.c

> 	$(COMPILE) $(WARN_CFLAGS_NO_FORMAT) $<

> 	$(POSTCOMPILE)

> 

> +alloc-ipa.o: ../alloc.c

> +	$(IPAGENT_COMPILE) $(WARN_CFLAGS_NO_FORMAT) $<

> +	$(POSTCOMPILE)

> +

> # Rules for objects that go in the in-process agent.

> 

> arch/%-ipa.o: ../arch/%.c

> @@ -623,6 +630,10 @@ common/%.o: ../common/%.c

> 	$(COMPILE) $<

> 	$(POSTCOMPILE)

> 

> +%.o: ../%.c

> +	$(COMPILE) $<

> +	$(POSTCOMPILE)

> +

> # Rules for register format descriptions.  Suffix destination files with

> # -generated to identify and clean them easily.

> 

> -- 

> 2.17.2

>
Tom Tromey - June 17, 2019, 5:43 p.m.
>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:

Alan> Looks like this breaks the building of alloc-ipa.o when using Make 3.81
Alan> I’ve tried this on a few different machines.

Thanks.

My first thought is that maybe we should simply declare 3.81
unsupported.  It was apparently released in 2006:

    https://savannah.gnu.org/forum/forum.php?forum_id=4380

... so it is quite ancient.

Is there some compelling reason to keep supporting it?


If there is, I guess I can experiment to try to find a workaround.

thanks,
Tom
Pedro Alves - June 17, 2019, 6:37 p.m.
On 6/17/19 6:43 PM, Tom Tromey wrote:
>>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:
> 
> Alan> Looks like this breaks the building of alloc-ipa.o when using Make 3.81
> Alan> I’ve tried this on a few different machines.
> 
> Thanks.
> 
> My first thought is that maybe we should simply declare 3.81
> unsupported.  It was apparently released in 2006:
> 
>     https://savannah.gnu.org/forum/forum.php?forum_id=4380
> 
> ... so it is quite ancient.
> 
> Is there some compelling reason to keep supporting it?
> 

I think it depends more on what distributions ship than what
the release date was.  E.g., if you look around the last couple
stable releases of popular stable distros (e.g., ubuntu, debian, fedora),
which GNU Make version did they ship?  If the GNU Make version shipped
by default is not 4.x, is there an easy optional rpm/deb package
for GNU Make 4.x available?

This was the same kind of investigation that led to the GCC 4.8
minimum requirement.

Also, looking around the GCC compile farm machine (including the
/opt/ dirs) for what is available may be a good hint/proxy for
determining whether bumping the requirement could cause trouble
for people.

> If there is, I guess I can experiment to try to find a workaround.

Thanks,
Pedro Alves
Alan Hayward - June 18, 2019, 9:30 a.m.
> On 17 Jun 2019, at 19:37, Pedro Alves <palves@redhat.com> wrote:

> 

> On 6/17/19 6:43 PM, Tom Tromey wrote:

>>>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:

>> 

>> Alan> Looks like this breaks the building of alloc-ipa.o when using Make 3.81

>> Alan> I’ve tried this on a few different machines.

>> 

>> Thanks.

>> 

>> My first thought is that maybe we should simply declare 3.81

>> unsupported.  It was apparently released in 2006:

>> 

>>    https://savannah.gnu.org/forum/forum.php?forum_id=4380

>> 

>> ... so it is quite ancient.

>> 

>> Is there some compelling reason to keep supporting it?

>> 

> 

> I think it depends more on what distributions ship than what

> the release date was.  E.g., if you look around the last couple

> stable releases of popular stable distros (e.g., ubuntu, debian, fedora),

> which GNU Make version did they ship?  If the GNU Make version shipped

> by default is not 4.x, is there an easy optional rpm/deb package

> for GNU Make 4.x available?

> 

> This was the same kind of investigation that led to the GCC 4.8

> minimum requirement.

> 

> Also, looking around the GCC compile farm machine (including the

> /opt/ dirs) for what is available may be a good hint/proxy for

> determining whether bumping the requirement could cause trouble

> for people.



Glibc requires 4.0:
https://www.sourceware.org/ml/libc-alpha/2018-08/msg00003.html
	Changes to build and runtime requirements:
  	GNU make 4.0 or later is now required to build glibc.


GCC still allows 3.8:
https://gcc.gnu.org/install/prerequisites.html
	GNU make version 3.80 (or later)



Alan.
Alan Hayward - July 3, 2019, 4:18 p.m.
> On 18 Jun 2019, at 10:30, Alan Hayward <Alan.Hayward@arm.com> wrote:

> 

> 

> 

>> On 17 Jun 2019, at 19:37, Pedro Alves <palves@redhat.com> wrote:

>> 

>> On 6/17/19 6:43 PM, Tom Tromey wrote:

>>>>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:

>>> 

>>> Alan> Looks like this breaks the building of alloc-ipa.o when using Make 3.81

>>> Alan> I’ve tried this on a few different machines.

>>> 

>>> Thanks.

>>> 

>>> My first thought is that maybe we should simply declare 3.81

>>> unsupported.  It was apparently released in 2006:

>>> 

>>>   https://savannah.gnu.org/forum/forum.php?forum_id=4380

>>> 

>>> ... so it is quite ancient.

>>> 

>>> Is there some compelling reason to keep supporting it?

>>> 

>> 

>> I think it depends more on what distributions ship than what

>> the release date was.  E.g., if you look around the last couple

>> stable releases of popular stable distros (e.g., ubuntu, debian, fedora),

>> which GNU Make version did they ship?  If the GNU Make version shipped

>> by default is not 4.x, is there an easy optional rpm/deb package

>> for GNU Make 4.x available?

>> 

>> This was the same kind of investigation that led to the GCC 4.8

>> minimum requirement.

>> 

>> Also, looking around the GCC compile farm machine (including the

>> /opt/ dirs) for what is available may be a good hint/proxy for

>> determining whether bumping the requirement could cause trouble

>> for people.

> 

> 

> Glibc requires 4.0:

> https://www.sourceware.org/ml/libc-alpha/2018-08/msg00003.html

> 	Changes to build and runtime requirements:

>  	GNU make 4.0 or later is now required to build glibc.

> 

> 

> GCC still allows 3.8:

> https://gcc.gnu.org/install/prerequisites.html

> 	GNU make version 3.80 (or later)

> 


I’ve done a little more digging into as many machines as I could find:

Ubuntu 16.04, 18.04 - make 4.1
Ubuntu 14.04 - make 4.0
Fedora 27 - make 4.2.1
Redhat 7.5, Centos 7.5, Centos 6 - make 3.82
Redhat 5.8, 6.9 - make 3.81
OpenSuse 42.2 - 4.2.1

Also: https://software.opensuse.org/package/make confirms a mix of 4.21 and
3.82 as the versions in various latest distros.

I’ve tried building using make 3.82, and it works for me.
Make 3.82 was released in 2010.

Any objection upping the minimum make version to 3.82 ?

If so I can raise a patch to add a statement to the NEWS file (similar to
the one that already exists).


Alan.
Tom Tromey - July 13, 2019, 4:04 p.m.
>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:

Alan> Any objection upping the minimum make version to 3.82 ?

Alan> If so I can raise a patch to add a statement to the NEWS file (similar to
Alan> the one that already exists).

That would be fine by me.

Tom
Pedro Alves - July 16, 2019, 7:47 p.m.
On 7/3/19 5:18 PM, Alan Hayward wrote:
> 
> 
>> On 18 Jun 2019, at 10:30, Alan Hayward <Alan.Hayward@arm.com> wrote:
>>
>>
>>
>>> On 17 Jun 2019, at 19:37, Pedro Alves <palves@redhat.com> wrote:
>>>
>>> On 6/17/19 6:43 PM, Tom Tromey wrote:
>>>>>>>>> "Alan" == Alan Hayward <Alan.Hayward@arm.com> writes:
>>>>
>>>> Alan> Looks like this breaks the building of alloc-ipa.o when using Make 3.81
>>>> Alan> I’ve tried this on a few different machines.
>>>>
>>>> Thanks.
>>>>
>>>> My first thought is that maybe we should simply declare 3.81
>>>> unsupported.  It was apparently released in 2006:
>>>>
>>>>   https://savannah.gnu.org/forum/forum.php?forum_id=4380
>>>>
>>>> ... so it is quite ancient.
>>>>
>>>> Is there some compelling reason to keep supporting it?
>>>>
>>>
>>> I think it depends more on what distributions ship than what
>>> the release date was.  E.g., if you look around the last couple
>>> stable releases of popular stable distros (e.g., ubuntu, debian, fedora),
>>> which GNU Make version did they ship?  If the GNU Make version shipped
>>> by default is not 4.x, is there an easy optional rpm/deb package
>>> for GNU Make 4.x available?
>>>
>>> This was the same kind of investigation that led to the GCC 4.8
>>> minimum requirement.
>>>
>>> Also, looking around the GCC compile farm machine (including the
>>> /opt/ dirs) for what is available may be a good hint/proxy for
>>> determining whether bumping the requirement could cause trouble
>>> for people.
>>
>>
>> Glibc requires 4.0:
>> https://www.sourceware.org/ml/libc-alpha/2018-08/msg00003.html
>> 	Changes to build and runtime requirements:
>>  	GNU make 4.0 or later is now required to build glibc.
>>
>>
>> GCC still allows 3.8:
>> https://gcc.gnu.org/install/prerequisites.html
>> 	GNU make version 3.80 (or later)
>>
> 
> I’ve done a little more digging into as many machines as I could find:
> 
> Ubuntu 16.04, 18.04 - make 4.1
> Ubuntu 14.04 - make 4.0
> Fedora 27 - make 4.2.1
> Redhat 7.5, Centos 7.5, Centos 6 - make 3.82
> Redhat 5.8, 6.9 - make 3.81
> OpenSuse 42.2 - 4.2.1
> 
> Also: https://software.opensuse.org/package/make confirms a mix of 4.21 and
> 3.82 as the versions in various latest distros.
> 
> I’ve tried building using make 3.82, and it works for me.
> Make 3.82 was released in 2010.
> 
> Any objection upping the minimum make version to 3.82 ?
> 

Fine with me as well.  Thanks for digging into all those machines.

Thanks,
Pedro Alves

Patch

diff --git a/gdb/Makefile.in b/gdb/Makefile.in
index 0f495783600..15ec7a61b1c 100644
--- a/gdb/Makefile.in
+++ b/gdb/Makefile.in
@@ -924,6 +924,7 @@  COMMON_SFILES = \
 	ada-varobj.c \
 	addrmap.c \
 	agent.c \
+	alloc.c \
 	annotate.c \
 	arch-utils.c \
 	auto-load.c \
diff --git a/gdb/alloc.c b/gdb/alloc.c
new file mode 100644
index 00000000000..e6c9c215435
--- /dev/null
+++ b/gdb/alloc.c
@@ -0,0 +1,95 @@ 
+/* Shared allocation functions for GDB, the GNU debugger.
+
+   Copyright (C) 2019 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#include "common/common-defs.h"
+#include "libiberty.h"
+#include <cstdlib>
+#include "common/errors.h"
+
+/* The xmalloc() (libiberty.h) family of memory management routines.
+
+   These are like the ISO-C malloc() family except that they implement
+   consistent semantics and guard against typical memory management
+   problems.  */
+
+/* NOTE: These are declared using PTR to ensure consistency with
+   "libiberty.h".  xfree() is GDB local.  */
+
+PTR                            /* ARI: PTR */
+xmalloc (size_t size)
+{
+  void *val;
+
+  /* See libiberty/xmalloc.c.  This function need's to match that's
+     semantics.  It never returns NULL.  */
+  if (size == 0)
+    size = 1;
+
+  val = malloc (size);         /* ARI: malloc */
+  if (val == NULL)
+    malloc_failure (size);
+
+  return val;
+}
+
+PTR                              /* ARI: PTR */
+xrealloc (PTR ptr, size_t size)          /* ARI: PTR */
+{
+  void *val;
+
+  /* See libiberty/xmalloc.c.  This function need's to match that's
+     semantics.  It never returns NULL.  */
+  if (size == 0)
+    size = 1;
+
+  if (ptr != NULL)
+    val = realloc (ptr, size);	/* ARI: realloc */
+  else
+    val = malloc (size);	        /* ARI: malloc */
+  if (val == NULL)
+    malloc_failure (size);
+
+  return val;
+}
+
+PTR                            /* ARI: PTR */
+xcalloc (size_t number, size_t size)
+{
+  void *mem;
+
+  /* See libiberty/xmalloc.c.  This function need's to match that's
+     semantics.  It never returns NULL.  */
+  if (number == 0 || size == 0)
+    {
+      number = 1;
+      size = 1;
+    }
+
+  mem = calloc (number, size);      /* ARI: xcalloc */
+  if (mem == NULL)
+    malloc_failure (number * size);
+
+  return mem;
+}
+
+void
+xmalloc_failed (size_t size)
+{
+  malloc_failure (size);
+}
diff --git a/gdb/common/common-utils.c b/gdb/common/common-utils.c
index 74ca93810c7..dd839a0d4d1 100644
--- a/gdb/common/common-utils.c
+++ b/gdb/common/common-utils.c
@@ -22,84 +22,12 @@ 
 #include "host-defs.h"
 #include <ctype.h>
 
-/* The xmalloc() (libiberty.h) family of memory management routines.
-
-   These are like the ISO-C malloc() family except that they implement
-   consistent semantics and guard against typical memory management
-   problems.  */
-
-/* NOTE: These are declared using PTR to ensure consistency with
-   "libiberty.h".  xfree() is GDB local.  */
-
-PTR                            /* ARI: PTR */
-xmalloc (size_t size)
-{
-  void *val;
-
-  /* See libiberty/xmalloc.c.  This function need's to match that's
-     semantics.  It never returns NULL.  */
-  if (size == 0)
-    size = 1;
-
-  val = malloc (size);         /* ARI: malloc */
-  if (val == NULL)
-    malloc_failure (size);
-
-  return val;
-}
-
-PTR                              /* ARI: PTR */
-xrealloc (PTR ptr, size_t size)          /* ARI: PTR */
-{
-  void *val;
-
-  /* See libiberty/xmalloc.c.  This function need's to match that's
-     semantics.  It never returns NULL.  */
-  if (size == 0)
-    size = 1;
-
-  if (ptr != NULL)
-    val = realloc (ptr, size);	/* ARI: realloc */
-  else
-    val = malloc (size);	        /* ARI: malloc */
-  if (val == NULL)
-    malloc_failure (size);
-
-  return val;
-}
-
-PTR                            /* ARI: PTR */           
-xcalloc (size_t number, size_t size)
-{
-  void *mem;
-
-  /* See libiberty/xmalloc.c.  This function need's to match that's
-     semantics.  It never returns NULL.  */
-  if (number == 0 || size == 0)
-    {
-      number = 1;
-      size = 1;
-    }
-
-  mem = calloc (number, size);      /* ARI: xcalloc */
-  if (mem == NULL)
-    malloc_failure (number * size);
-
-  return mem;
-}
-
 void *
 xzalloc (size_t size)
 {
   return xcalloc (1, size);
 }
 
-void
-xmalloc_failed (size_t size)
-{
-  malloc_failure (size);
-}
-
 /* Like asprintf/vasprintf but get an internal_error if the call
    fails. */
 
diff --git a/gdb/gdbserver/Makefile.in b/gdb/gdbserver/Makefile.in
index f5fc55034ee..a0793e657c5 100644
--- a/gdb/gdbserver/Makefile.in
+++ b/gdb/gdbserver/Makefile.in
@@ -197,6 +197,7 @@  SFILES = \
 	$(srcdir)/arch/arm-get-next-pcs.c \
 	$(srcdir)/arch/arm-linux.c \
 	$(srcdir)/arch/ppc-linux-common.c \
+	$(srcdir)/../alloc.c \
 	$(srcdir)/common/btrace-common.c \
 	$(srcdir)/common/buffer.c \
 	$(srcdir)/common/cleanups.c \
@@ -238,6 +239,7 @@  SOURCES = $(SFILES)
 TAGFILES = $(SOURCES) ${HFILES} ${ALLPARAM} ${POSSLIBS}
 
 OBS = \
+	alloc.o \
 	ax.o \
 	common/agent.o \
 	common/btrace-common.o \
@@ -413,6 +415,7 @@  gdbreplay$(EXEEXT): $(sort $(GDBREPLAY_OBS)) $(LIBGNU) $(LIBIBERTY)
 		$(LIBIBERTY)
 
 IPA_OBJS = \
+	alloc-ipa.o \
 	ax-ipa.o \
 	common/common-utils-ipa.o \
 	common/errors-ipa.o \
@@ -568,6 +571,10 @@  ax.o: ax.c
 	$(COMPILE) $(WARN_CFLAGS_NO_FORMAT) $<
 	$(POSTCOMPILE)
 
+alloc-ipa.o: ../alloc.c
+	$(IPAGENT_COMPILE) $(WARN_CFLAGS_NO_FORMAT) $<
+	$(POSTCOMPILE)
+
 # Rules for objects that go in the in-process agent.
 
 arch/%-ipa.o: ../arch/%.c
@@ -623,6 +630,10 @@  common/%.o: ../common/%.c
 	$(COMPILE) $<
 	$(POSTCOMPILE)
 
+%.o: ../%.c
+	$(COMPILE) $<
+	$(POSTCOMPILE)
+
 # Rules for register format descriptions.  Suffix destination files with
 # -generated to identify and clean them easily.