Static variables and "gp"

Static variables and "gp"

[Gilad Benjamini]

Hi,
How can I force a specific static variable to be used by the "gp" register
rather than by a two-step load ?
Compiling a mips-linux using gcc.

I was poking around for information and stumbled across the "-G" flag,
the "__attribute__ ((section (".sdata")))" (or sbss), the "-msdata" but to no avail.
I only got more confused.

Any help will be appreciated

Re: Static variables and "gp"

[Ralf Baechle]

On Sun, Mar 02, 2003 at 09:07:28AM +0000, Gilad Benjamini wrote:

> How can I force a specific static variable to be used by the "gp" register
> rather than by a two-step load ?
> Compiling a mips-linux using gcc.

That feature doesn't work in gcc for userspace code ...

  Ralf

Re: Static variables and "gp"

[Johannes Stezenbach]

On Sun, Mar 02, 2003 at 12:18:20PM +0100, Ralf Baechle wrote:
> On Sun, Mar 02, 2003 at 09:07:28AM +0000, Gilad Benjamini wrote:
> 
> > How can I force a specific static variable to be used by the "gp" register
> > rather than by a two-step load ?
> > Compiling a mips-linux using gcc.
> 
> That feature doesn't work in gcc for userspace code ...

It works with diet libc (http://dietlibc.org/).
You need a static, non-PIC libgcc.a which supports it, though,
which means a bit of gcc config hackery...

Regards,
Johannes

JVM under Linux on MIPS

[Rajesh Palani]

[-- Attachment #1: Type: text/plain, Size: 279 bytes --]


Hi,
   Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under Linux/MIPS.
   Thanks and regards,
   Rajesh



---------------------------------
Do you Yahoo!?
Yahoo! Tax Center - forms, calculators, tips, and more

[-- Attachment #2: Type: text/html, Size: 392 bytes --]

Re: JVM under Linux on MIPS

[Kevin D. Kissell]

[-- Attachment #1: Type: text/plain, Size: 1481 bytes --]

I developed a patch to post-Transvirtual Kaffe 1.0.7 that allows it to pass 
all its internal regression tests and run the Spec JVM98 benchmarks
on MIPS/Linux on a MIPS Malta/5Kc system (little endian)
using H.K.'s "Red Hat" distribution (with an update to the X11
stuff).  The necessary code has been integrated into the Kaffe
CVS repository if you check out the current sources, or you
can apply the patch at
ftp://ftp.paralogos.com/pub/kaffe/mips/kaffe-1.0.7.patch_kevink_021001 
to the 1.0.7 source release on the kaffe.org site.

Note that I did *not* get the JIT functioning fully.  It will run some
programs (including Caffeinemark) but contains the same bugs in 
argument passing that I fixed for the interpreter linkage to native 
methods, but the same fix doesn't quite work there, and I ran out of time 
to look at it.  You therefore need to run configure with --with-engine=intrp
and to not expect miraculous performance.

  ----- Original Message ----- 
  From: Rajesh Palani 
  To: linux-mips@linux-mips.org 
  Sent: Tuesday, March 04, 2003 2:14 AM
  Subject: JVM under Linux on MIPS


  Hi, 

     Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under Linux/MIPS. 

     Thanks and regards, 

     Rajesh





------------------------------------------------------------------------------
  Do you Yahoo!?
  Yahoo! Tax Center - forms, calculators, tips, and more

[-- Attachment #2: Type: text/html, Size: 3409 bytes --]

Re: JVM under Linux on MIPS

[Kevin D. Kissell]

[-- Attachment #1: Type: text/plain, Size: 1481 bytes --]

I developed a patch to post-Transvirtual Kaffe 1.0.7 that allows it to pass 
all its internal regression tests and run the Spec JVM98 benchmarks
on MIPS/Linux on a MIPS Malta/5Kc system (little endian)
using H.K.'s "Red Hat" distribution (with an update to the X11
stuff).  The necessary code has been integrated into the Kaffe
CVS repository if you check out the current sources, or you
can apply the patch at
ftp://ftp.paralogos.com/pub/kaffe/mips/kaffe-1.0.7.patch_kevink_021001 
to the 1.0.7 source release on the kaffe.org site.

Note that I did *not* get the JIT functioning fully.  It will run some
programs (including Caffeinemark) but contains the same bugs in 
argument passing that I fixed for the interpreter linkage to native 
methods, but the same fix doesn't quite work there, and I ran out of time 
to look at it.  You therefore need to run configure with --with-engine=intrp
and to not expect miraculous performance.

  ----- Original Message ----- 
  From: Rajesh Palani 
  To: linux-mips@linux-mips.org 
  Sent: Tuesday, March 04, 2003 2:14 AM
  Subject: JVM under Linux on MIPS


  Hi, 

     Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under Linux/MIPS. 

     Thanks and regards, 

     Rajesh





------------------------------------------------------------------------------
  Do you Yahoo!?
  Yahoo! Tax Center - forms, calculators, tips, and more

[-- Attachment #2: Type: text/html, Size: 3409 bytes --]

Re: JVM under Linux on MIPS

[TAKANO Ryousei]

Hi,

On Mon, 3 Mar 2003 17:14:59 -0800 (PST)
Rajesh Palani <rpalani2@yahoo.com> wrote:
>    Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under Linux/MIPS.
>

I have succeeded in running Kaffe 1.0.7 with --with-engine=jit3
on a TANBAC TB0193 board (VR4181/66MHz).
However, it was very poor performance. A hello world execution,
for example, takes about 30 seconds.
I think that a JIT initialization has taken a lot of time.

Thanks,
TAKANO Ryousei

Re: JVM under Linux on MIPS

[Kevin D. Kissell]

> On Mon, 3 Mar 2003 17:14:59 -0800 (PST)
> Rajesh Palani <rpalani2@yahoo.com> wrote:
> >    Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under
Linux/MIPS.
> >
>
> I have succeeded in running Kaffe 1.0.7 with --with-engine=jit3
> on a TANBAC TB0193 board (VR4181/66MHz).
> However, it was very poor performance. A hello world execution,
> for example, takes about 30 seconds.
> I think that a JIT initialization has taken a lot of time.

I'm very pleased to hear that you got it running on a Vr41xx,
but I'm curious about the JIT behavior you saw.  I can believe
that it could run "hello world", but does it really pass all the
internal regression tests ("make check")?  Are you running
a "normal" MIPS/Linux distribution which assumes a
hardware FPU and does kernel emulation where necessary,
or are you using a purely soft-float environment?  I ask
this because most of the problems I have with the JIT are
in areas where mixed integer/floating arguments are being
passed, and those might not be an issue with soft-float.

As for the performance you observed, how much memory
did you have on the board, and what kind of secondary storage
(disk?) hardware was used?  66MHz isn't fast, but the combined
compile-and-run time for Caffeinemark for the patched
kaffe 1.0.7 on a MIPS 5Kc core at 160MHz was in fact
pretty good, better than 3 Embedded Caffeienmarks
per megahertz, which isn't as fast as commercial dynmic
compilers, but which is still several times faster than most
commercial interpretive JVMs.  Running fully interpretive,
kaffe's performance is mediocre but reasonable, I certainly
wasn't seeing delays of 10 seconds to run "hello world",
which is roughly what one would expect scaling your reported
run time by the frequency.  I really think that you are far more
likely to have been I/O bound, either from paging or from file I/O.

            Kevin K.

Re: JVM under Linux on MIPS

[Kevin D. Kissell]

> On Mon, 3 Mar 2003 17:14:59 -0800 (PST)
> Rajesh Palani <rpalani2@yahoo.com> wrote:
> >    Has anyone had any success running any open source JVMs (other than Cobalt machines running Transvirtual's Kaffe) under
Linux/MIPS.
> >
>
> I have succeeded in running Kaffe 1.0.7 with --with-engine=jit3
> on a TANBAC TB0193 board (VR4181/66MHz).
> However, it was very poor performance. A hello world execution,
> for example, takes about 30 seconds.
> I think that a JIT initialization has taken a lot of time.

I'm very pleased to hear that you got it running on a Vr41xx,
but I'm curious about the JIT behavior you saw.  I can believe
that it could run "hello world", but does it really pass all the
internal regression tests ("make check")?  Are you running
a "normal" MIPS/Linux distribution which assumes a
hardware FPU and does kernel emulation where necessary,
or are you using a purely soft-float environment?  I ask
this because most of the problems I have with the JIT are
in areas where mixed integer/floating arguments are being
passed, and those might not be an issue with soft-float.

As for the performance you observed, how much memory
did you have on the board, and what kind of secondary storage
(disk?) hardware was used?  66MHz isn't fast, but the combined
compile-and-run time for Caffeinemark for the patched
kaffe 1.0.7 on a MIPS 5Kc core at 160MHz was in fact
pretty good, better than 3 Embedded Caffeienmarks
per megahertz, which isn't as fast as commercial dynmic
compilers, but which is still several times faster than most
commercial interpretive JVMs.  Running fully interpretive,
kaffe's performance is mediocre but reasonable, I certainly
wasn't seeing delays of 10 seconds to run "hello world",
which is roughly what one would expect scaling your reported
run time by the frequency.  I really think that you are far more
likely to have been I/O bound, either from paging or from file I/O.

            Kevin K.

Re: JVM under Linux on MIPS

[TAKANO Ryousei]

Hi Kevin,

> I'm very pleased to hear that you got it running on a Vr41xx,
> but I'm curious about the JIT behavior you saw.  I can believe
> that it could run "hello world", but does it really pass all the
> internal regression tests ("make check")?  Are you running
> a "normal" MIPS/Linux distribution which assumes a
> hardware FPU and does kernel emulation where necessary,
> or are you using a purely soft-float environment?  I ask
> this because most of the problems I have with the JIT are
> in areas where mixed integer/floating arguments are being
> passed, and those might not be an issue with soft-float.
> 
I have cross-compiled Kaffe, so it did not pass "make check".
I tried it under a Linux-VR kernel(kernel-2.4.0-test9) which is
enabled with a kernel FPU emulation.
I have not tried under a Linux/MIPS kernel.

> As for the performance you observed, how much memory
> did you have on the board, and what kind of secondary storage
> (disk?) hardware was used?  66MHz isn't fast, but the combined
> compile-and-run time for Caffeinemark for the patched
> kaffe 1.0.7 on a MIPS 5Kc core at 160MHz was in fact
> pretty good, better than 3 Embedded Caffeienmarks
> per megahertz, which isn't as fast as commercial dynmic
> compilers, but which is still several times faster than most
> commercial interpretive JVMs.  Running fully interpretive,
> kaffe's performance is mediocre but reasonable, I certainly
> wasn't seeing delays of 10 seconds to run "hello world",
> which is roughly what one would expect scaling your reported
> run time by the frequency.  I really think that you are far more
> likely to have been I/O bound, either from paging or from file I/O.
> 
TANBAC TB0193 has 16MB SDRAM, and it is using Compact Flash
as a secondary storage.

I try to make jar files compact (strips unused packages) 
for a faster initialization.

Thanks,
TAKANO Ryousei

Re: JVM under Linux on MIPS

[Jun Sun]

On Tue, Mar 04, 2003 at 07:09:02PM +0900, TAKANO Ryousei wrote:
> Hi Kevin,
> 
> > I'm very pleased to hear that you got it running on a Vr41xx,
> > but I'm curious about the JIT behavior you saw.  I can believe
> > that it could run "hello world", but does it really pass all the
> > internal regression tests ("make check")?  Are you running
> > a "normal" MIPS/Linux distribution which assumes a
> > hardware FPU and does kernel emulation where necessary,
> > or are you using a purely soft-float environment?  I ask
> > this because most of the problems I have with the JIT are
> > in areas where mixed integer/floating arguments are being
> > passed, and those might not be an issue with soft-float.
> > 
> I have cross-compiled Kaffe, so it did not pass "make check".
> I tried it under a Linux-VR kernel(kernel-2.4.0-test9) which is
> enabled with a kernel FPU emulation.
> I have not tried under a Linux/MIPS kernel.
>

Java VM typically uses massive threading.  Emulating ll/sc instructions
for vr4181a might one of the biggest causes of delay.

Jun