Pure space overflows

Pure space overflows

[Po Lu]

I get pure space overflows when building with X and XInput 2 on the
following systems:

  i586-pc-linux-gnu
  i386-pc-solaris2.11

Would anyone mind me changing BASE_PURESIZE to a value which results in
no overflow on both those systems?  Thanks.

Re: Pure space overflows

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Date: Sat, 22 Jan 2022 15:34:38 +0800
> 
> I get pure space overflows when building with X and XInput 2 on the
> following systems:
> 
>   i586-pc-linux-gnu
>   i386-pc-solaris2.11
> 
> Would anyone mind me changing BASE_PURESIZE to a value which results in
> no overflow on both those systems?  Thanks.

How much would you need to enlarge that?

And why BASE_PURESIZE and not SYSTEM_PURESIZE_EXTRA for those
configurations?

Re: Pure space overflows

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

> How much would you need to enlarge that?

100000 bytes seems to be enough to make the build work on both these
systems.

> And why BASE_PURESIZE and not SYSTEM_PURESIZE_EXTRA for those
> configurations?

Isn't that a relic from when we had separate headers for each machine
and configuration?  I couldn't find anywhere they are set today.

Thanks.

Re: Pure space overflows

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: emacs-devel@gnu.org
> Date: Sat, 22 Jan 2022 17:38:25 +0800
> 
> > And why BASE_PURESIZE and not SYSTEM_PURESIZE_EXTRA for those
> > configurations?
> 
> Isn't that a relic from when we had separate headers for each machine
> and configuration?

No, not yet a relic.

> I couldn't find anywhere they are set today.

See src/conf_post.h.

Re: Pure space overflows

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Po Lu <luangruo@yahoo.com>
>> Cc: emacs-devel@gnu.org
>> Date: Sat, 22 Jan 2022 17:38:25 +0800
>> 
>> > And why BASE_PURESIZE and not SYSTEM_PURESIZE_EXTRA for those
>> > configurations?
>> 
>> Isn't that a relic from when we had separate headers for each machine
>> and configuration?
>
> No, not yet a relic.
>
>> I couldn't find anywhere they are set today.
>
> See src/conf_post.h.

I wonder if the following code being under an MSDOS conditional is an
oversight, since it mentions GNU/Linux:

  #ifdef HAVE_X_WINDOWS
  /* We need a little extra space, see ../../lisp/loadup.el and the
     commentary below, in the non-X branch.  The 140KB number was
     measured on GNU/Linux and on MS-Windows.  */
  #define SYSTEM_PURESIZE_EXTRA (-170000+140000)

Thanks.

Re: Pure space overflows

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: emacs-devel@gnu.org
> Date: Sat, 22 Jan 2022 19:02:49 +0800
> 
> > See src/conf_post.h.
> 
> I wonder if the following code being under an MSDOS conditional is an
> oversight, since it mentions GNU/Linux:
> 
>   #ifdef HAVE_X_WINDOWS
>   /* We need a little extra space, see ../../lisp/loadup.el and the
>      commentary below, in the non-X branch.  The 140KB number was
>      measured on GNU/Linux and on MS-Windows.  */
>   #define SYSTEM_PURESIZE_EXTRA (-170000+140000)

No, not an oversight.  This is for a MS-DOS build with X support.
(Yes, we actually had that at some point, and we still have *.inp
files in the msdos/ directory to support it, although I'd be very
surprised if such a build could be made today.)

Re: Pure space overflows

[Eli Zaretskii]

> Date: Sat, 22 Jan 2022 13:13:14 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-devel@gnu.org
> 
> >   #ifdef HAVE_X_WINDOWS
> >   /* We need a little extra space, see ../../lisp/loadup.el and the
> >      commentary below, in the non-X branch.  The 140KB number was
> >      measured on GNU/Linux and on MS-Windows.  */
> >   #define SYSTEM_PURESIZE_EXTRA (-170000+140000)
> 
> No, not an oversight.  This is for a MS-DOS build with X support.
> (Yes, we actually had that at some point, and we still have *.inp
> files in the msdos/ directory to support it, although I'd be very
> surprised if such a build could be made today.)

See the --with-x option to config.bat.

Re: Pure space overflows

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

> No, not an oversight.

Thanks.

Re: Pure space overflows

[Po Lu]

Po Lu <luangruo@yahoo.com> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> No, not an oversight.
>
> Thanks.

powerpc-unknown-linux-gnu needs a SYSTEM_PURESIZE_EXTRA of 250000 to
build without pure space overflows, which seems a little excessive to
me.

Any ideas here?
Thanks.

Re: Pure space overflows

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: emacs-devel@gnu.org
> Date: Sun, 23 Jan 2022 16:10:48 +0800
> 
> powerpc-unknown-linux-gnu needs a SYSTEM_PURESIZE_EXTRA of 250000 to
> build without pure space overflows, which seems a little excessive to
> me.

With what configure options and which optional features?

Re: Pure space overflows

[Po Lu]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Po Lu <luangruo@yahoo.com>
>> Cc: emacs-devel@gnu.org
>> Date: Sun, 23 Jan 2022 16:10:48 +0800
>> 
>> powerpc-unknown-linux-gnu needs a SYSTEM_PURESIZE_EXTRA of 250000 to
>> build without pure space overflows, which seems a little excessive to
>> me.
>
> With what configure options and which optional features?

Sorry, I should've mentioned that it's the same as the other machines I
tested: XInput 2 and the GTK 3 toolkit, and xpm, jpeg, tiff, gif, png,
rsvg, and webp support.

Though on that machine, turning XInput 2 off makes no difference.

Thanks.

Re: Pure space overflows

[Eli Zaretskii]

> From: Po Lu <luangruo@yahoo.com>
> Cc: emacs-devel@gnu.org
> Date: Sun, 23 Jan 2022 17:28:24 +0800
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> powerpc-unknown-linux-gnu needs a SYSTEM_PURESIZE_EXTRA of 250000 to
> >> build without pure space overflows, which seems a little excessive to
> >> me.
> >
> > With what configure options and which optional features?
> 
> Sorry, I should've mentioned that it's the same as the other machines I
> tested: XInput 2 and the GTK 3 toolkit, and xpm, jpeg, tiff, gif, png,
> rsvg, and webp support.
> 
> Though on that machine, turning XInput 2 off makes no difference.

It is strange that PPC needs significantly more pure space.  Does it
have some unique alignment requirements, perhaps?

Andreas, any ideas why additional pure space would be needed for PPC?

Re: Pure space overflows

[Andreas Schwab]

On Jan 23 2022, Eli Zaretskii wrote:

> Andreas, any ideas why additional pure space would be needed for PPC?

Is it?

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: Pure space overflows

[Po Lu]

Andreas Schwab <schwab@linux-m68k.org> writes:

> On Jan 23 2022, Eli Zaretskii wrote:
>
>> Andreas, any ideas why additional pure space would be needed for PPC?
>
> Is it?

I needed to set SYSTEM_PURESIZE_EXTRA to 250000 to produce a build with
X that didn't cause an overflow.

Thanks.

Re: Pure space overflows

[Andreas Schwab]

Worksforme.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: Pure space overflows

[Po Lu]

Andreas Schwab <schwab@linux-m68k.org> writes:

> Worksforme.

What system did you use to test?

Thanks.

Re: Pure space overflows

[Andreas Schwab]

https://build.opensuse.org/project/show/home:AndreasSchwab:emacs:master

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510  2552 DF73 E780 A9DA AEC1
"And now for something completely different."

Re: Pure space overflows

[Po Lu]

Andreas Schwab <schwab@linux-m68k.org> writes:

> https://build.opensuse.org/project/show/home:AndreasSchwab:emacs:master

I tried on an old system running Fedora 14, FWIW.