Fixing ant-bootstrap / jamvm segfault (i686)

Fixing ant-bootstrap / jamvm segfault (i686)

[Ricardo Wurmus]

Hi Guix,

here’s a probably needlessly complicated patch to fix the build of
ant-bootstrap.  It currently fails to build because jamvm segfaults.

--8<---------------cut here---------------start------------->8---
diff --git a/gnu/packages/java.scm b/gnu/packages/java.scm
index 403c446a82..6bb4f8167f 100644
--- a/gnu/packages/java.scm
+++ b/gnu/packages/java.scm
@@ -186,10 +186,15 @@ language.")
     (arguments
      `(#:configure-flags
        (list (string-append "--with-classpath-install-dir="
-                            (assoc-ref %build-inputs "classpath")))))
+                            (assoc-ref %build-inputs "classpath"))
+             "--disable-int-caching"
+             "--enable-runtime-reloc-checks"
+             "--disable-int-inlining"
+             "--enable-ffi")))
     (inputs
      `(("classpath" ,classpath-bootstrap)
        ("jikes" ,jikes)
+       ("libffi" ,libffi)
        ("zlib" ,zlib)))
     (home-page "http://jamvm.sourceforge.net/")
     (synopsis "Small Java Virtual Machine")
@@ -258,7 +263,7 @@ JNI.")
                 ,@(if (string-prefix? "armhf" (or (%current-system)
                                                   (%current-target-system)))
                       `((string-append m "-Xnocompact "))
-                      `((string-append m "-Xnocompact -Xnoinlining ")))))
+                      `((string-append m "-Xnocompact ")))))

              ;; Disable tests because we are bootstrapping and thus don't have
              ;; any of the dependencies required to build and run the tests.
--8<---------------cut here---------------end--------------->8---

I’m almost certain that not all of these configure flags are needed.
What do you think?  Would someone like to play a little with this patch
and see if we can figure out why exactly jamvm segfaults?

I used this command to build ant-bootstrap:

    ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu packages java) ant-bootstrap)'

--
Ricardo

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Gábor Boskovits]

[-- Attachment #1: Type: text/plain, Size: 2239 bytes --]

Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 16.,
Hét 15:20):

> Hi Guix,
>
> here’s a probably needlessly complicated patch to fix the build of
> ant-bootstrap.  It currently fails to build because jamvm segfaults.
>
> --8<---------------cut here---------------start------------->8---
> diff --git a/gnu/packages/java.scm b/gnu/packages/java.scm
> index 403c446a82..6bb4f8167f 100644
> --- a/gnu/packages/java.scm
> +++ b/gnu/packages/java.scm
> @@ -186,10 +186,15 @@ language.")
>      (arguments
>       `(#:configure-flags
>         (list (string-append "--with-classpath-install-dir="
> -                            (assoc-ref %build-inputs "classpath")))))
> +                            (assoc-ref %build-inputs "classpath"))
> +             "--disable-int-caching"
> +             "--enable-runtime-reloc-checks"
> +             "--disable-int-inlining"
> +             "--enable-ffi")))
>      (inputs
>       `(("classpath" ,classpath-bootstrap)
>         ("jikes" ,jikes)
> +       ("libffi" ,libffi)
>         ("zlib" ,zlib)))
>      (home-page "http://jamvm.sourceforge.net/")
>      (synopsis "Small Java Virtual Machine")
> @@ -258,7 +263,7 @@ JNI.")
>                  ,@(if (string-prefix? "armhf" (or (%current-system)
>
>  (%current-target-system)))
>                        `((string-append m "-Xnocompact "))
> -                      `((string-append m "-Xnocompact -Xnoinlining ")))))
> +                      `((string-append m "-Xnocompact ")))))
>
>               ;; Disable tests because we are bootstrapping and thus don't
> have
>               ;; any of the dependencies required to build and run the
> tests.
> --8<---------------cut here---------------end--------------->8---
>
> I’m almost certain that not all of these configure flags are needed.
> What do you think?  Would someone like to play a little with this patch
> and see if we can figure out why exactly jamvm segfaults?
>
I will try to have a look at this again in the coming days.

>
> I used this command to build ant-bootstrap:
>
>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu packages
> java) ant-bootstrap)'
>
> --
> Ricardo
>
>

[-- Attachment #2: Type: text/html, Size: 3306 bytes --]

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Ricardo Wurmus]

Hi Gábor,

> Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 16.,
> Hét 15:20):
>
>> Hi Guix,
>>
>> here’s a probably needlessly complicated patch to fix the build of
>> ant-bootstrap.  It currently fails to build because jamvm
>> segfaults.

[…]

>> I’m almost certain that not all of these configure flags are needed.
>> What do you think?  Would someone like to play a little with this patch
>> and see if we can figure out why exactly jamvm segfaults?
>>
> I will try to have a look at this again in the coming days.
>
>>
>> I used this command to build ant-bootstrap:
>>
>>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu packages
>> java) ant-bootstrap)'

Have you been able to take a look at this?  I’d like to push this or a variant
of it to the core-updates branch soon.

-- 
Ricardo

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Gábor Boskovits]

[-- Attachment #1: Type: text/plain, Size: 1118 bytes --]

Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 27.,
Pén 9:14):

>
> Hi Gábor,
>
> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 16.,
> > Hét 15:20):
> >
> >> Hi Guix,
> >>
> >> here’s a probably needlessly complicated patch to fix the build of
> >> ant-bootstrap.  It currently fails to build because jamvm
> >> segfaults.
>
> […]
>
> >> I’m almost certain that not all of these configure flags are needed.
> >> What do you think?  Would someone like to play a little with this patch
> >> and see if we can figure out why exactly jamvm segfaults?
> >>
> > I will try to have a look at this again in the coming days.
> >
> >>
> >> I used this command to build ant-bootstrap:
> >>
> >>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu
> packages
> >> java) ant-bootstrap)'
>
> Have you been able to take a look at this?  I’d like to push this or a
> variant
> of it to the core-updates branch soon.
>
Not yet. I will have a look tomorrow and report back if that is not too
late.

>
> --
> Ricardo
>
>

[-- Attachment #2: Type: text/html, Size: 1829 bytes --]

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Ricardo Wurmus]

Gábor Boskovits <boskovits@gmail.com> writes:

> Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 27.,
> Pén 9:14):
>
>>
>> Hi Gábor,
>>
>> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 16.,
>> > Hét 15:20):
>> >
>> >> Hi Guix,
>> >>
>> >> here’s a probably needlessly complicated patch to fix the build of
>> >> ant-bootstrap.  It currently fails to build because jamvm
>> >> segfaults.
>>
>> […]
>>
>> >> I’m almost certain that not all of these configure flags are needed.
>> >> What do you think?  Would someone like to play a little with this patch
>> >> and see if we can figure out why exactly jamvm segfaults?
>> >>
>> > I will try to have a look at this again in the coming days.
>> >
>> >>
>> >> I used this command to build ant-bootstrap:
>> >>
>> >>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu
>> packages
>> >> java) ant-bootstrap)'
>>
>> Have you been able to take a look at this?  I’d like to push this or a
>> variant
>> of it to the core-updates branch soon.
>>
> Not yet. I will have a look tomorrow and report back if that is not too
> late.

Tomorrow would be just fine.  There’s no real rush, but I’d like to
avoid forgetting about this when we already have the means to fix lots
of broken packages on core-updates.

Thanks for taking a look at this!

--
Ricardo

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Gábor Boskovits]

[-- Attachment #1: Type: text/plain, Size: 1807 bytes --]

Hello,

Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 27., P,
9:52):

>
> Gábor Boskovits <boskovits@gmail.com> writes:
>
> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 27.,
> > Pén 9:14):
> >
> >>
> >> Hi Gábor,
> >>
> >> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept.
> 16.,
> >> > Hét 15:20):
> >> >
> >> >> Hi Guix,
> >> >>
> >> >> here’s a probably needlessly complicated patch to fix the build of
> >> >> ant-bootstrap.  It currently fails to build because jamvm
> >> >> segfaults.
> >>
> >> […]
> >>
> >> >> I’m almost certain that not all of these configure flags are needed.
> >> >> What do you think?  Would someone like to play a little with this
> patch
> >> >> and see if we can figure out why exactly jamvm segfaults?
> >> >>
> >> > I will try to have a look at this again in the coming days.
> >> >
> >> >>
> >> >> I used this command to build ant-bootstrap:
> >> >>
> >> >>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu
> >> packages
> >> >> java) ant-bootstrap)'
> >>
> >> Have you been able to take a look at this?  I’d like to push this or a
> >> variant
> >> of it to the core-updates branch soon.
> >>
> > Not yet. I will have a look tomorrow and report back if that is not too
> > late.
>
> Tomorrow would be just fine.  There’s no real rush, but I’d like to
> avoid forgetting about this when we already have the means to fix lots
> of broken packages on core-updates.
>
> Thanks for taking a look at this!
>
> --
> Ricardo
>
>
So far I can confirm that the patch as is works fine for me. I will now
have a look at reducing this.

-- 
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21

[-- Attachment #2: Type: text/html, Size: 2801 bytes --]

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Gábor Boskovits]

[-- Attachment #1.1: Type: text/plain, Size: 2133 bytes --]

Hello,

Gábor Boskovits <boskovits@gmail.com> ezt írta (időpont: 2019. szept. 28.,
Szo, 23:14):

> Hello,
>
> Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept. 27.,
> P, 9:52):
>
>>
>> Gábor Boskovits <boskovits@gmail.com> writes:
>>
>> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept.
>> 27.,
>> > Pén 9:14):
>> >
>> >>
>> >> Hi Gábor,
>> >>
>> >> > Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. szept.
>> 16.,
>> >> > Hét 15:20):
>> >> >
>> >> >> Hi Guix,
>> >> >>
>> >> >> here’s a probably needlessly complicated patch to fix the build of
>> >> >> ant-bootstrap.  It currently fails to build because jamvm
>> >> >> segfaults.
>> >>
>> >> […]
>> >>
>> >> >> I’m almost certain that not all of these configure flags are needed.
>> >> >> What do you think?  Would someone like to play a little with this
>> patch
>> >> >> and see if we can figure out why exactly jamvm segfaults?
>> >> >>
>> >> > I will try to have a look at this again in the coming days.
>> >> >
>> >> >>
>> >> >> I used this command to build ant-bootstrap:
>> >> >>
>> >> >>     ./pre-inst-env guix build -K --system=i686-linux -e '(@@ (gnu
>> >> packages
>> >> >> java) ant-bootstrap)'
>> >>
>> >> Have you been able to take a look at this?  I’d like to push this or a
>> >> variant
>> >> of it to the core-updates branch soon.
>> >>
>> > Not yet. I will have a look tomorrow and report back if that is not too
>> > late.
>>
>> Tomorrow would be just fine.  There’s no real rush, but I’d like to
>> avoid forgetting about this when we already have the means to fix lots
>> of broken packages on core-updates.
>>
>> Thanks for taking a look at this!
>>
>> --
>> Ricardo
>>
>>
> So far I can confirm that the patch as is works fine for me. I will now
> have a look at reducing this.
>
> --
> OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21
>

The attached patch also works for me. Can you check?

g_bor
-- 
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21

[-- Attachment #1.2: Type: text/html, Size: 3474 bytes --]

[-- Attachment #2: 0001-gnu-Fix-jamvm-segfault-on-i686.patch --]
[-- Type: text/x-patch, Size: 1173 bytes --]

From 286b3934f71c0abeb66266c4a8fdbb119dea714e Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?G=C3=A1bor=20Boskovits?= <boskovits@gmail.com>
Date: Sat, 28 Sep 2019 23:35:02 +0200
Subject: [PATCH] gnu: Fix jamvm segfault on i686.

* gnu/packages/java.scm (jamvm-1-bootstrap): Fix segfault.
[arguments]: Add --enable-ffi to #:configure-flags.
[inputs]: Add libffi.
---
 gnu/packages/java.scm | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/gnu/packages/java.scm b/gnu/packages/java.scm
index def2c8d92d..55cbb33ac9 100644
--- a/gnu/packages/java.scm
+++ b/gnu/packages/java.scm
@@ -197,10 +197,12 @@ language.")
     (arguments
      `(#:configure-flags
        (list (string-append "--with-classpath-install-dir="
-                            (assoc-ref %build-inputs "classpath")))))
+                            (assoc-ref %build-inputs "classpath"))
+             "--enable-ffi")))
     (inputs
      `(("classpath" ,classpath-bootstrap)
        ("jikes" ,jikes)
+       ("libffi" ,libffi)
        ("zlib" ,zlib)))
     ;; When built with a recent GCC and glibc the configure step of icedtea-6
     ;; fails with an invalid instruction error.
-- 
2.23.0

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Ricardo Wurmus]

Hi Gábor,

> The attached patch also works for me. Can you check?

I just noticed that this is not enough for me to build
classpath-0.99-1.e7c13ee0c.  ant-bootstrap builds fine but Classpath
fails.  (This is with --system=i686-linux.)

I haven’t tried to build all the way to the final JDK with my previous
patch yet.

--
Ricardo

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Ricardo Wurmus]

Ricardo Wurmus <rekado@elephly.net> writes:

> Hi Gábor,
>
>> The attached patch also works for me. Can you check?
>
> I just noticed that this is not enough for me to build
> classpath-0.99-1.e7c13ee0c.  ant-bootstrap builds fine but Classpath
> fails.  (This is with --system=i686-linux.)
>
> I haven’t tried to build all the way to the final JDK with my previous
> patch yet.

I can build up to the first icedtea package with this patch, whereas the
reduced patch leads to a failure to build classpath.

--8<---------------cut here---------------start------------->8---
diff --git a/gnu/packages/java.scm b/gnu/packages/java.scm
index 403c446a82..597bd8733b 100644
--- a/gnu/packages/java.scm
+++ b/gnu/packages/java.scm
@@ -186,10 +186,15 @@ language.")
     (arguments
      `(#:configure-flags
        (list (string-append "--with-classpath-install-dir="
-                            (assoc-ref %build-inputs "classpath")))))
+                            (assoc-ref %build-inputs "classpath"))
+             "--disable-int-caching"
+             "--enable-runtime-reloc-checks"
+             "--disable-int-inlining"
+             "--enable-ffi")))
     (inputs
      `(("classpath" ,classpath-bootstrap)
        ("jikes" ,jikes)
+       ("libffi" ,libffi)
        ("zlib" ,zlib)))
     (home-page "http://jamvm.sourceforge.net/")
     (synopsis "Small Java Virtual Machine")
@@ -258,7 +263,7 @@ JNI.")
                 ,@(if (string-prefix? "armhf" (or (%current-system)
                                                   (%current-target-system)))
                       `((string-append m "-Xnocompact "))
-                      `((string-append m "-Xnocompact -Xnoinlining ")))))
+                      `((string-append m "-Xnocompact ")))))

              ;; Disable tests because we are bootstrapping and thus don't have
              ;; any of the dependencies required to build and run the tests.
@@ -537,7 +542,7 @@ gnu.classpath.tools.~a.~a $@"
                                    (if (string=? "native2ascii" tool)
                                        "Native2ASCII" "Main")))
                                  `((format #t "#!~a/bin/sh
-~a/bin/jamvm -Xnocompact -Xnoinlining -classpath ~a/share/classpath/tools.zip \
+~a/bin/jamvm -Xnocompact -classpath ~a/share/classpath/tools.zip \
 gnu.classpath.tools.~a.~a $@"
                                    bash jamvm classpath tool
                                    (if (string=? "native2ascii" tool)
--8<---------------cut here---------------end--------------->8---

I would like to push this.  Not sure if just to core-updates or to
master as well, because Java on i686-linux is currently broken on
master.

--
Ricardo

Re: Fixing ant-bootstrap / jamvm segfault (i686)

[Gábor Boskovits]

[-- Attachment #1: Type: text/plain, Size: 3094 bytes --]

Ricardo Wurmus <rekado@elephly.net> ezt írta (időpont: 2019. okt. 1., Ke
3:02):

>
> Ricardo Wurmus <rekado@elephly.net> writes:
>
> > Hi Gábor,
> >
> >> The attached patch also works for me. Can you check?
> >
> > I just noticed that this is not enough for me to build
> > classpath-0.99-1.e7c13ee0c.  ant-bootstrap builds fine but Classpath
> > fails.  (This is with --system=i686-linux.)
> >
> > I haven’t tried to build all the way to the final JDK with my previous
> > patch yet.
>
> I can build up to the first icedtea package with this patch, whereas the
> reduced patch leads to a failure to build classpath.
>
> --8<---------------cut here---------------start------------->8---
> diff --git a/gnu/packages/java.scm b/gnu/packages/java.scm
> index 403c446a82..597bd8733b 100644
> --- a/gnu/packages/java.scm
> +++ b/gnu/packages/java.scm
> @@ -186,10 +186,15 @@ language.")
>      (arguments
>       `(#:configure-flags
>         (list (string-append "--with-classpath-install-dir="
> -                            (assoc-ref %build-inputs "classpath")))))
> +                            (assoc-ref %build-inputs "classpath"))
> +             "--disable-int-caching"
> +             "--enable-runtime-reloc-checks"
> +             "--disable-int-inlining"
> +             "--enable-ffi")))
>      (inputs
>       `(("classpath" ,classpath-bootstrap)
>         ("jikes" ,jikes)
> +       ("libffi" ,libffi)
>         ("zlib" ,zlib)))
>      (home-page "http://jamvm.sourceforge.net/")
>      (synopsis "Small Java Virtual Machine")
> @@ -258,7 +263,7 @@ JNI.")
>                  ,@(if (string-prefix? "armhf" (or (%current-system)
>
>  (%current-target-system)))
>                        `((string-append m "-Xnocompact "))
> -                      `((string-append m "-Xnocompact -Xnoinlining ")))))
> +                      `((string-append m "-Xnocompact ")))))
>
>               ;; Disable tests because we are bootstrapping and thus don't
> have
>               ;; any of the dependencies required to build and run the
> tests.
> @@ -537,7 +542,7 @@ gnu.classpath.tools.~a.~a $@"
>                                     (if (string=? "native2ascii" tool)
>                                         "Native2ASCII" "Main")))
>                                   `((format #t "#!~a/bin/sh
> -~a/bin/jamvm -Xnocompact -Xnoinlining -classpath
> ~a/share/classpath/tools.zip \
> +~a/bin/jamvm -Xnocompact -classpath ~a/share/classpath/tools.zip \
>  gnu.classpath.tools.~a.~a $@"
>                                     bash jamvm classpath tool
>                                     (if (string=? "native2ascii" tool)
> --8<---------------cut here---------------end--------------->8---
>
> I would like to push this.  Not sure if just to core-updates or to
> master as well, because Java on i686-linux is currently broken on
> master.
>
I did not check this yet, but it looks good to me. I will have a look
later, but you can push this. If I find anything we can reduce this in
staging.

>
> --
> Ricardo
>
>

[-- Attachment #2: Type: text/html, Size: 4443 bytes --]