public inbox for java-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* ARM EABI and eclipse-ecj
@ 2007-08-18 10:53 Steven Newbury
  2007-08-19 15:43 ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-18 10:53 UTC (permalink / raw)
  To: aph; +Cc: java-patches

Hi Andrew,
I've built gcj for ARM EABI using all relevant patches AFAICT, simple test java
programs are working fine both using gij and gcj.  I can't get eclipse-ecj to
work though, it is giving me this output with -verbose :

...
[Loaded (pre-compiled) gnu.java.security.hash.IMessageDigest from <no code
source>]
[Loaded (pre-compiled) gnu.gcj.runtime.PersistentByteMap from <no code source>]
[Loaded (pre-compiled) gnu.gcj.runtime.PersistentByteMap$AccessMode from <no
code source>]
[Loaded (pre-compiled) java.nio.channels.FileChannel$MapMode from <no code
source>]
[Loaded (pre-compiled) java.io.FileNotFoundException from <no code source>]
[Loaded (pre-compiled) java.util.Vector$1 from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.problem.ProblemSeverities
from (file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.util.SuffixConstants from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.io.DataOutputStream from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.io.FilenameFilter from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main$2 from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.env.AccessRule from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.core.compiler.InvalidInputException from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.lang.RuntimeException from <no code source>]
[Loaded (pre-compiled) java.io.ByteArrayInputStream from <no code source>]
[Loaded (pre-compiled) java.io.StringReader from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.ICompilerRequestor from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main$3 from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.env.IDependent from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.env.ICompilationUnit from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.CompilationUnit from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.IErrorHandlingPolicy from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main$4 from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.IProblemFactory from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode)
org.eclipse.jdt.internal.compiler.problem.DefaultProblemFactory from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.TypeConstants from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.TypeIds from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.ClassFile from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.ProblemReasons from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.LookupEnvironment
from (file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.env.INameEnvironment from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.FileSystem from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.Binding from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.TypeBinding from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.lookup.ReferenceBinding
from (file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.problem.AbortCompilation
from (file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.util.MissingResourceException from <no code
source>]
[Loaded (pre-compiled) java.util.ArrayList from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.FileSystem$Classpath
from (file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.io.PrintWriter from <no code source>]
[Loaded (pre-compiled) java.io.Writer from <no code source>]
[Loaded (pre-compiled) java.io.OutputStreamWriter from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main$Logger from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.lang.NoClassDefFoundError from <no code source>]
[Loaded (pre-compiled) java.lang.LinkageError from <no code source>]
[Loaded (pre-compiled) java.lang.IllegalAccessException from <no code source>]
[Loaded (pre-compiled) java.lang.IllegalArgumentException from <no code
source>]
[Loaded (pre-compiled) java.lang.SecurityException from <no code source>]
[Loaded (pre-compiled) java.io.StringWriter from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.batch.Main$1 from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.core.compiler.IProblem from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.core.compiler.CategorizedProblem from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.util.GenericXMLWriter from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
[Loaded (pre-compiled) java.lang.Float from <no code source>]
[Loaded (bytecode) org.eclipse.jdt.internal.compiler.util.HashtableOfInt from
(file:/usr/share/eclipse-ecj-3.3/lib/ecj.jar <no certificates>)]
/usr/lib/jvm/gcj-jdk-4.3/bin/javac: line 8: 16631 Aborted                
CLASSPATH="/usr/share/eclipse-ecj-3.3/lib/ecj.jar:${CLASSPATH}"
/usr/lib/gcj-4.3.0_alpha20070810/bin/gij
-Dgnu.gcj.precompiled.db.path=${JAVA_PKG_CLASSMAP} -verbose:class
org.eclipse.jdt.internal.compiler.batch.Main -nowarn "${@}"

Does it work for you?  Is there any way to to find out why it's aborting?  The
same snapshot (+patches) is working for me on x86_64 and i386.


Steve


      ___________________________________________________________ 
Want ideas for reducing your carbon footprint? Visit Yahoo! For Good  http://uk.promotions.yahoo.com/forgood/environment.html

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-18 10:53 ARM EABI and eclipse-ecj Steven Newbury
@ 2007-08-19 15:43 ` Andrew Haley
  2007-08-19 21:08   ` Steven Newbury
  0 siblings, 1 reply; 12+ messages in thread
From: Andrew Haley @ 2007-08-19 15:43 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > Hi Andrew,
 > I've built gcj for ARM EABI using all relevant patches AFAICT, simple test java
 > programs are working fine both using gij and gcj.  I can't get eclipse-ecj to
 > work though, it is giving me this output with -verbose :

It's still work in progress.  Once all of the tests pass I'll have
a look.

Did you try the branch at
svn+ssh://gcc.gnu.org/svn/gcc/branches/gcj/gcj-eabi-branch
?

Andrew.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-19 15:43 ` Andrew Haley
@ 2007-08-19 21:08   ` Steven Newbury
  2007-08-20  9:30     ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-19 21:08 UTC (permalink / raw)
  To: Andrew Haley; +Cc: aph, java-patches


--- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:

> Steven Newbury writes:
>  > Hi Andrew,
>  > I've built gcj for ARM EABI using all relevant patches AFAICT, simple test
> java
>  > programs are working fine both using gij and gcj.  I can't get eclipse-ecj
> to
>  > work though, it is giving me this output with -verbose :

I've found that if I use gcj as the jit:
        -Dgnu.gcj.jit.compiler=$(gcj-config --gcj) \ # <- Gentoo-ism
        -Dgnu.gcj.jit.cachedir=/var/tmp/jitcache/ \
        -Dgnu.gcj.jit.options="-Os"
Then it does work.  Something is wrong with the gij interpreter not gcj itself.
 This is the first time I've had a javac other than jikes running on EABI. 
Actually this is the most functional jvm I've got working, jamvm had problems
with doubles which I couldn't get to the bottom of.  Cacao and kaffe don't to
want to work either in various ways.

> 
> It's still work in progress.  Once all of the tests pass I'll have
> a look.
> 
> Did you try the branch at
> svn+ssh://gcc.gnu.org/svn/gcc/branches/gcj/gcj-eabi-branch
> ?
I will.  I wasn't aware of it.  At the moment I'm using all the patches you
have posted plus a patch based on the unwind-backtrace patch you sent to the
java list and the __clear_cache patch I posted.  It's also worth mentioning
that as I mentioned in the bugzilla entry, I'm using a pure iwmmxt non-multilib
toolchain with the associated patch I attached to that bug.


Steve


      ___________________________________________________________ 
Want ideas for reducing your carbon footprint? Visit Yahoo! For Good  http://uk.promotions.yahoo.com/forgood/environment.html

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-19 21:08   ` Steven Newbury
@ 2007-08-20  9:30     ` Andrew Haley
  2007-08-20 16:18       ` Steven Newbury
  0 siblings, 1 reply; 12+ messages in thread
From: Andrew Haley @ 2007-08-20  9:30 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > 
 > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > 
 > > Steven Newbury writes:
 > >  > Hi Andrew,
 > >  > I've built gcj for ARM EABI using all relevant patches AFAICT, simple test
 > > java
 > >  > programs are working fine both using gij and gcj.  I can't get eclipse-ecj
 > > to
 > >  > work though, it is giving me this output with -verbose :
 > 
 > I've found that if I use gcj as the jit:
 >         -Dgnu.gcj.jit.compiler=$(gcj-config --gcj) \ # <- Gentoo-ism
 >         -Dgnu.gcj.jit.cachedir=/var/tmp/jitcache/ \
 >         -Dgnu.gcj.jit.options="-Os"

 > Then it does work.  Something is wrong with the gij interpreter not
 > gcj itself.

OK, thanks.  That's really helpful.

 >  This is the first time I've had a javac other than jikes running on EABI. 
 > Actually this is the most functional jvm I've got working, jamvm had problems
 > with doubles which I couldn't get to the bottom of.  Cacao and kaffe don't to
 > want to work either in various ways.

I have found one or two more bugs in libffi and fixed them on the
branch.  It is possible that I forgot to post those patches.

 > > It's still work in progress.  Once all of the tests pass I'll have
 > > a look.
 > > 
 > > Did you try the branch at
 > > svn+ssh://gcc.gnu.org/svn/gcc/branches/gcj/gcj-eabi-branch
 > > ?

 > I will.  I wasn't aware of it.

Right, let's see how we get on with that.

Andrew.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-20  9:30     ` Andrew Haley
@ 2007-08-20 16:18       ` Steven Newbury
  2007-08-20 16:26         ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-20 16:18 UTC (permalink / raw)
  To: Andrew Haley; +Cc: aph, java-patches


--- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:

> Steven Newbury writes:
>  > 
>  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
>  > 
>  > > Steven Newbury writes:
>  > >  > Hi Andrew,
>  > >  > I've built gcj for ARM EABI using all relevant patches AFAICT, simple
> test
>  > > java
>  > >  > programs are working fine both using gij and gcj.  I can't get
> eclipse-ecj
>  > > to
>  > >  > work though, it is giving me this output with -verbose :
>  > 
>  > I've found that if I use gcj as the jit:
>  >         -Dgnu.gcj.jit.compiler=$(gcj-config --gcj) \ # <- Gentoo-ism
>  >         -Dgnu.gcj.jit.cachedir=/var/tmp/jitcache/ \
>  >         -Dgnu.gcj.jit.options="-Os"
> 
>  > Then it does work.  Something is wrong with the gij interpreter not
>  > gcj itself.
> 
> OK, thanks.  That's really helpful.
> 
>  >  This is the first time I've had a javac other than jikes running on EABI.
> 
>  > Actually this is the most functional jvm I've got working, jamvm had
> problems
>  > with doubles which I couldn't get to the bottom of.  Cacao and kaffe don't
> to
>  > want to work either in various ways.
> 
> I have found one or two more bugs in libffi and fixed them on the
> branch.  It is possible that I forgot to post those patches.
> 
OK, I'd put an || instead of an && for the while loop in the unwind loop, oops.
 As far as I can see I now have everything up to date with the gcj-eabi-branch.
 Now ecj1 doesn't abort, so it is probably fixed, except gij seems to be using
inordinate amounts of memory - is this expected?  I only have 64MB and it's
going deep into swap.  


Steve


      ___________________________________________________________
Yahoo! Answers - Got a question? Someone out there knows the answer. Try it
now.
http://uk.answers.yahoo.com/ 

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-20 16:18       ` Steven Newbury
@ 2007-08-20 16:26         ` Andrew Haley
  2007-08-20 16:29           ` Steven Newbury
  0 siblings, 1 reply; 12+ messages in thread
From: Andrew Haley @ 2007-08-20 16:26 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > 
 > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > 
 > > Steven Newbury writes:
 > >  > 
 > >  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > >  > 
 > >  > > Steven Newbury writes:
 > >  > >  > Hi Andrew,
 > >  > >  > I've built gcj for ARM EABI using all relevant patches AFAICT, simple
 > > test
 > >  > > java
 > >  > >  > programs are working fine both using gij and gcj.  I can't get
 > > eclipse-ecj
 > >  > > to
 > >  > >  > work though, it is giving me this output with -verbose :
 > >  > 
 > >  > I've found that if I use gcj as the jit:
 > >  >         -Dgnu.gcj.jit.compiler=$(gcj-config --gcj) \ # <- Gentoo-ism
 > >  >         -Dgnu.gcj.jit.cachedir=/var/tmp/jitcache/ \
 > >  >         -Dgnu.gcj.jit.options="-Os"
 > > 
 > >  > Then it does work.  Something is wrong with the gij interpreter not
 > >  > gcj itself.
 > > 
 > > OK, thanks.  That's really helpful.
 > > 
 > >  >  This is the first time I've had a javac other than jikes running on EABI.
 > > 
 > >  > Actually this is the most functional jvm I've got working, jamvm had
 > > problems
 > >  > with doubles which I couldn't get to the bottom of.  Cacao and kaffe don't
 > > to
 > >  > want to work either in various ways.
 > > 
 > > I have found one or two more bugs in libffi and fixed them on the
 > > branch.  It is possible that I forgot to post those patches.
 > > 
 > OK, I'd put an || instead of an && for the while loop in the unwind loop, oops.
 >  As far as I can see I now have everything up to date with the gcj-eabi-branch.
 >  Now ecj1 doesn't abort, so it is probably fixed, except gij seems to be using
 > inordinate amounts of memory - is this expected?  I only have 64MB and it's
 > going deep into swap.  

I found an unwinder bug that I'm working on.  It might cause the
behaviour you describe.

Andrew.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-20 16:26         ` Andrew Haley
@ 2007-08-20 16:29           ` Steven Newbury
  2007-08-24  9:47             ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-20 16:29 UTC (permalink / raw)
  To: Andrew Haley; +Cc: aph, java-patches


--- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:

>  > OK, I'd put an || instead of an && for the while loop in the unwind loop,
> oops.
>  >  As far as I can see I now have everything up to date with the
> gcj-eabi-branch.
>  >  Now ecj1 doesn't abort, so it is probably fixed, except gij seems to be
> using
>  > inordinate amounts of memory - is this expected?  I only have 64MB and
> it's
>  > going deep into swap.  
> 
> I found an unwinder bug that I'm working on.  It might cause the
> behaviour you describe.
Right, it's hitting the OOM killer before it completes so I don't know if it
would have worked.  I'll wait to see what you come up with.

Steve


      ___________________________________________________________
Yahoo! Answers - Got a question? Someone out there knows the answer. Try it
now.
http://uk.answers.yahoo.com/ 

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-20 16:29           ` Steven Newbury
@ 2007-08-24  9:47             ` Andrew Haley
  2007-08-24 10:43               ` Steven Newbury
  0 siblings, 1 reply; 12+ messages in thread
From: Andrew Haley @ 2007-08-24  9:47 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > 
 > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > 
 > >  > OK, I'd put an || instead of an && for the while loop in the unwind loop,
 > > oops.
 > >  >  As far as I can see I now have everything up to date with the
 > > gcj-eabi-branch.
 > >  >  Now ecj1 doesn't abort, so it is probably fixed, except gij seems to be
 > > using
 > >  > inordinate amounts of memory - is this expected?  I only have 64MB and
 > > it's
 > >  > going deep into swap.  
 > > 
 > > I found an unwinder bug that I'm working on.  It might cause the
 > > behaviour you describe.
 > Right, it's hitting the OOM killer before it completes so I don't know if it
 > would have worked.  I'll wait to see what you come up with.

The branch is looking better now.

Andrew.

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-24  9:47             ` Andrew Haley
@ 2007-08-24 10:43               ` Steven Newbury
  2007-08-28  8:33                 ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-24 10:43 UTC (permalink / raw)
  To: Andrew Haley; +Cc: aph, java-patches


--- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:

> Steven Newbury writes:
>  > 
>  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
>  > 
>  > >  > OK, I'd put an || instead of an && for the while loop in the unwind
> loop,
>  > > oops.
>  > >  >  As far as I can see I now have everything up to date with the
>  > > gcj-eabi-branch.
>  > >  >  Now ecj1 doesn't abort, so it is probably fixed, except gij seems to
> be
>  > > using
>  > >  > inordinate amounts of memory - is this expected?  I only have 64MB
> and
>  > > it's
>  > >  > going deep into swap.  
>  > > 
>  > > I found an unwinder bug that I'm working on.  It might cause the
>  > > behaviour you describe.
>  > Right, it's hitting the OOM killer before it completes so I don't know if
> it
>  > would have worked.  I'll wait to see what you come up with.
> 
> The branch is looking better now.

I'll give it a go and let you know how it goes.


Steve


      ___________________________________________________________ 
Want ideas for reducing your carbon footprint? Visit Yahoo! For Good  http://uk.promotions.yahoo.com/forgood/environment.html

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-24 10:43               ` Steven Newbury
@ 2007-08-28  8:33                 ` Andrew Haley
  2007-08-28 11:28                   ` Steven Newbury
  0 siblings, 1 reply; 12+ messages in thread
From: Andrew Haley @ 2007-08-28  8:33 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > 
 > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > 
 > > Steven Newbury writes:
 > >  > 
 > >  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 > >  > 
 > >  > >  > OK, I'd put an || instead of an && for the while loop in the unwind
 > > loop,
 > >  > > oops.
 > >  > >  >  As far as I can see I now have everything up to date with the
 > >  > > gcj-eabi-branch.
 > >  > >  >  Now ecj1 doesn't abort, so it is probably fixed, except gij seems to
 > > be
 > >  > > using
 > >  > >  > inordinate amounts of memory - is this expected?  I only have 64MB
 > > and
 > >  > > it's
 > >  > >  > going deep into swap.  
 > >  > > 
 > >  > > I found an unwinder bug that I'm working on.  It might cause the
 > >  > > behaviour you describe.
 > >  > Right, it's hitting the OOM killer before it completes so I don't know if
 > > it
 > >  > would have worked.  I'll wait to see what you come up with.
 > > 
 > > The branch is looking better now.
 > 
 > I'll give it a go and let you know how it goes.
 
I just got a nearly perfect set of test scores.  Throw_2 is expected
(indeed, designed) to fail, so no problem there.  TestClosureGC is a
weird one, but not worth worrying about ATM.  As far as I can see
we're good to go.

Let me know your results.

Thanks,
Andrew.


FAIL: /home/aph/gcc/trunk/libjava/testsuite/libjava.jar/TestClosureGC.jar execution - gij test
FAIL: Throw_2 execution - source compiled test
FAIL: Throw_2 -findirect-dispatch execution - source compiled test
FAIL: Throw_2 -O3 execution - source compiled test
FAIL: Throw_2 -O3 -findirect-dispatch execution - source compiled test

                === libjava Summary ===

# of expected passes            2528
# of unexpected failures        5
# of untested testcases         5

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-28  8:33                 ` Andrew Haley
@ 2007-08-28 11:28                   ` Steven Newbury
  2007-08-28 12:08                     ` Andrew Haley
  0 siblings, 1 reply; 12+ messages in thread
From: Steven Newbury @ 2007-08-28 11:28 UTC (permalink / raw)
  To: Andrew Haley; +Cc: aph, java-patches


--- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:

> Steven Newbury writes:
>  > 
>  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
>  > 
>  > > Steven Newbury writes:
>  > >  > 
>  > >  > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
>  > >  > 
>  > >  > >  > OK, I'd put an || instead of an && for the while loop in the
> unwind
>  > > loop,
>  > >  > > oops.
>  > >  > >  >  As far as I can see I now have everything up to date with the
>  > >  > > gcj-eabi-branch.
>  > >  > >  >  Now ecj1 doesn't abort, so it is probably fixed, except gij
> seems to
>  > > be
>  > >  > > using
>  > >  > >  > inordinate amounts of memory - is this expected?  I only have
> 64MB
>  > > and
>  > >  > > it's
>  > >  > >  > going deep into swap.  
>  > >  > > 
>  > >  > > I found an unwinder bug that I'm working on.  It might cause the
>  > >  > > behaviour you describe.
>  > >  > Right, it's hitting the OOM killer before it completes so I don't
> know if
>  > > it
>  > >  > would have worked.  I'll wait to see what you come up with.
>  > > 
>  > > The branch is looking better now.
>  > 
>  > I'll give it a go and let you know how it goes.
>  
> I just got a nearly perfect set of test scores.  Throw_2 is expected
> (indeed, designed) to fail, so no problem there.  TestClosureGC is a
> weird one, but not worth worrying about ATM.  As far as I can see
> we're good to go.
> 
> Let me know your results.

I won't be able to do much until later in the week.  I have noticed that gij is
quite a bit slower than jamvm (using libffi) on some benchmark tests, I'll see
if I can profile that when I get the chance.

> 
> Thanks,
> Andrew.
> 
> 
> FAIL: /home/aph/gcc/trunk/libjava/testsuite/libjava.jar/TestClosureGC.jar
> execution - gij test
> FAIL: Throw_2 execution - source compiled test
> FAIL: Throw_2 -findirect-dispatch execution - source compiled test
> FAIL: Throw_2 -O3 execution - source compiled test
> FAIL: Throw_2 -O3 -findirect-dispatch execution - source compiled test
> 
>                 === libjava Summary ===
> 
> # of expected passes            2528
> # of unexpected failures        5
> # of untested testcases         5
> 
That does look good! :)


Steve


      ___________________________________________________________
Yahoo! Answers - Got a question? Someone out there knows the answer. Try it
now.
http://uk.answers.yahoo.com/ 

^ permalink raw reply	[flat|nested] 12+ messages in thread

* Re: ARM EABI and eclipse-ecj
  2007-08-28 11:28                   ` Steven Newbury
@ 2007-08-28 12:08                     ` Andrew Haley
  0 siblings, 0 replies; 12+ messages in thread
From: Andrew Haley @ 2007-08-28 12:08 UTC (permalink / raw)
  To: Steven Newbury; +Cc: aph, java-patches

Steven Newbury writes:
 > 
 > --- Andrew Haley <aph-gcc@littlepinkcloud.COM> wrote:
 
 > > I just got a nearly perfect set of test scores.  Throw_2 is expected
 > > (indeed, designed) to fail, so no problem there.  TestClosureGC is a
 > > weird one, but not worth worrying about ATM.  As far as I can see
 > > we're good to go.
 > > 
 > > Let me know your results.
 > 
 > I won't be able to do much until later in the week.  I have noticed that gij is
 > quite a bit slower than jamvm (using libffi) on some benchmark tests,

That's true on all platforms: gij is a very simple bytecode
interpreter.  gcj wins with precompilation.

Andrew.

^ permalink raw reply	[flat|nested] 12+ messages in thread

end of thread, other threads:[~2007-08-28 12:08 UTC | newest]

Thread overview: 12+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2007-08-18 10:53 ARM EABI and eclipse-ecj Steven Newbury
2007-08-19 15:43 ` Andrew Haley
2007-08-19 21:08   ` Steven Newbury
2007-08-20  9:30     ` Andrew Haley
2007-08-20 16:18       ` Steven Newbury
2007-08-20 16:26         ` Andrew Haley
2007-08-20 16:29           ` Steven Newbury
2007-08-24  9:47             ` Andrew Haley
2007-08-24 10:43               ` Steven Newbury
2007-08-28  8:33                 ` Andrew Haley
2007-08-28 11:28                   ` Steven Newbury
2007-08-28 12:08                     ` Andrew Haley

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).