post 2.6.7 BK change breaks Java?

post 2.6.7 BK change breaks Java?

[Matthias Andree]

Hi,

I've pulled from the linux-2.6 BK tree some post-2.6.7 version, compiled
and installed it, and it breaks Java, standalone or plugged into
firefox, the symptom is that the application catches SIGKILL. This
didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.

Is there any particular change I should try backing out?

TIA,

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95 (PGP/MIME preferred)

Re: post 2.6.7 BK change breaks Java?

[John Richard Moser]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

The only thing I've seen kill java like that would be NX things, such as
the NX patch mentioned in an earlier thread; execshield; or PaX.  I saw
some talk about possibly enabling NX by default; but I don't see this in
the -mm6 list, and I have no idea where the bk patch list is.  I
wouldn't expect either Linus or Andrew to have decided to merge an NX
patch in at this stage; but it's a possibility.

Andrew?  Has anything like that been added in the bk tree?

Matthias Andree wrote:
| Hi,
|
| I've pulled from the linux-2.6 BK tree some post-2.6.7 version, compiled
| and installed it, and it breaks Java, standalone or plugged into
| firefox, the symptom is that the application catches SIGKILL. This
| didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.
|
| Is there any particular change I should try backing out?
|
| TIA,
|
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFA6stihDd4aOud5P8RAnSWAJsGGWL61RC+GIiKk083w6tN5minSQCfSzb4
tstDqu+7FnIyeCSrfSBPrS8=
=pOq9
-----END PGP SIGNATURE-----

Re: post 2.6.7 BK change breaks Java?

[Matthias Andree]

On Tue, 06 Jul 2004, John Richard Moser wrote:

> The only thing I've seen kill java like that would be NX things, such as
> the NX patch mentioned in an earlier thread; execshield; or PaX.  I saw
> some talk about possibly enabling NX by default; but I don't see this in
> the -mm6 list, and I have no idea where the bk patch list is.  I
> wouldn't expect either Linus or Andrew to have decided to merge an NX
> patch in at this stage; but it's a possibility.

I've been pointed to the NX feature off-list and investigated, my CPU
(AMD Athlon XP 2500+ Model 10 "Barton") doesn't support the noexec flag,
and dmesg does not contain any output that MX was enabled, and the Java
"Killed" problem persists when the kernel is booted with noexec=off.

It must have entered the tree between v2.6.7 and revision 1.1757 in
Linus' tree.

BTW, how do I tell BitKeeper "pull up to revision..."?  bk pull and bk
undo -aREV is a way, but it's wasteful.

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95 (PGP/MIME preferred)

Re: post 2.6.7 BK change breaks Java?

[John Richard Moser]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1



Matthias Andree wrote:
| On Tue, 06 Jul 2004, John Richard Moser wrote:
|
| I've been pointed to the NX feature off-list and investigated, my CPU
| (AMD Athlon XP 2500+ Model 10 "Barton") doesn't support the noexec flag,

Your CPU doesn't need to support an NX flag.  See
http://en.wikipedia.net/wiki/PaX for two ways to mimic the same
functionality.  I'm not sure if any noexec patches other than PaX can do
this though; the thing IS about 3 years old.

Anyway, it was just a longshot.  I've got nothing else to offer on the
subject of java dying because of in-kernel issues.

| and dmesg does not contain any output that MX was enabled, and the Java
| "Killed" problem persists when the kernel is booted with noexec=off.
|
| It must have entered the tree between v2.6.7 and revision 1.1757 in
| Linus' tree.
|
| BTW, how do I tell BitKeeper "pull up to revision..."?  bk pull and bk
| undo -aREV is a way, but it's wasteful.
|
Dunno.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFA6uYrhDd4aOud5P8RAsDwAKCANubhFZRy8cMywXF+/VFzX+sNNgCfY5l9
0u+mKXtBIimd4xQSEVIplJE=
=RXue
-----END PGP SIGNATURE-----

Re: post 2.6.7 BK change breaks Java?

[Ray Lee]

> BTW, how do I tell BitKeeper "pull up to revision..."? bk pull and bk
> undo -aREV is a way, but it's wasteful.

Don't pull to a version. Pull the latest, keep it, and get or export
different versions instead:

	bk -r get -q -rREV

...Or Incantations To That Effect.

Ray

Re: post 2.6.7 BK change breaks Java?

[Andy Isaacson]

On Tue, Jul 06, 2004 at 06:14:51PM +0200, Matthias Andree wrote:
> I've been pointed to the NX feature off-list and investigated, my CPU
> (AMD Athlon XP 2500+ Model 10 "Barton") doesn't support the noexec flag,
> and dmesg does not contain any output that MX was enabled, and the Java
> "Killed" problem persists when the kernel is booted with noexec=off.
> 
> It must have entered the tree between v2.6.7 and revision 1.1757 in
> Linus' tree.

BK revision numbers aren't stable, so "1.1757" doesn't say much.
Instead, quote keys, either :KEY: or :MD5KEY:, like so:
% bk prs -hnd:KEY: -r1.1657 ChangeSet
akpm@osdl.org[torvalds]|ChangeSet|20040323152307|55600
% bk prs -hnd:MD5KEY: -r1.1657 ChangeSet
4060565bRhJji9RfHpiUg8dYxnHR1A

Those identifiers are eternal and unchanging, and can be used almost
anywhere a revision number can be used.  (Note that I used a different
rev, as I don't have 1.1757 in my tree at the moment.)

> BTW, how do I tell BitKeeper "pull up to revision..."?  bk pull and bk
> undo -aREV is a way, but it's wasteful.

bk clone has a -r option, but it just does an undo internally.  You
should definitely have a local mirror of the kernel source and make
temporary clones to work in, then the only things that you're wasting
are compute cycles and disk IO (rather than network bandwidth).

% (cd mirror/linux-2.5; bk pull)
% bk clone -ql -r4060565bRhJji9RfHpiUg8dYxnHR1A mirror/linux-2.5 tmptree

This works better if you have enough RAM to cache the entire BK tree
comfortably (at least 768MB, preferably a gig).

-andy (not speaking for or associated with BitMover)

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Tue, 6 Jul 2004, Matthias Andree wrote:

> Hi,
>
> I've pulled from the linux-2.6 BK tree some post-2.6.7 version, compiled
> and installed it, and it breaks Java, standalone or plugged into
> firefox, the symptom is that the application catches SIGKILL. This
> didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.
>
I'm seeing the same thing. I'm using Eclipse a lot which is Java based,
and I noticed that wen I went from plain 2.6.7 to 2.6.7-mm3 Eclipse
started dying shortly after launch (it only manages to get the splash
screen up) with a message about the JVM dying. Since I had also upgraded
my Sun Java at the same time I initially suspected that and back down to
my old version, but the problem persisted. Then I tried the latest Java
release from Sun, with same result. Then I started suspecting the kernel
and tried 2.6.7-mm6, 2.6.7-bk20 and 2.6.7-mm7 - all with the same result
that Java breaks. Finally I went back to a plain 2.6.7 and the problem
went away - so it certainly looks kernel related.
I was using the same .config with all kernels (copied from my plain 2.6.7
kernel to the others and then running 'make oldconfig'), so I'm also
pretty sure it's not due to some new kernel option I've enabled that I
don't usually use.

My hardware is AMD Athlon (t-bird) 1.4GHz CPU in a ASUS A7M266 mobo with
512MB of DDR266 RAM.


> Is there any particular change I should try backing out?
>
I'm looking for the same thing, haven't found it yet unfortunately.

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Con Kolivas]

[-- Attachment #1: Type: text/plain, Size: 2404 bytes --]

Jesper Juhl wrote:
> On Tue, 6 Jul 2004, Matthias Andree wrote:
> 
> 
>>Hi,
>>
>>I've pulled from the linux-2.6 BK tree some post-2.6.7 version, compiled
>>and installed it, and it breaks Java, standalone or plugged into
>>firefox, the symptom is that the application catches SIGKILL. This
>>didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.
>>
> 
> I'm seeing the same thing. I'm using Eclipse a lot which is Java based,
> and I noticed that wen I went from plain 2.6.7 to 2.6.7-mm3 Eclipse
> started dying shortly after launch (it only manages to get the splash
> screen up) with a message about the JVM dying. Since I had also upgraded
> my Sun Java at the same time I initially suspected that and back down to
> my old version, but the problem persisted. Then I tried the latest Java
> release from Sun, with same result. Then I started suspecting the kernel
> and tried 2.6.7-mm6, 2.6.7-bk20 and 2.6.7-mm7 - all with the same result
> that Java breaks. Finally I went back to a plain 2.6.7 and the problem
> went away - so it certainly looks kernel related.
> I was using the same .config with all kernels (copied from my plain 2.6.7
> kernel to the others and then running 'make oldconfig'), so I'm also
> pretty sure it's not due to some new kernel option I've enabled that I
> don't usually use.
> 
> My hardware is AMD Athlon (t-bird) 1.4GHz CPU in a ASUS A7M266 mobo with
> 512MB of DDR266 RAM.
> 
> 
> 
>>Is there any particular change I should try backing out?
>>
> 
> I'm looking for the same thing, haven't found it yet unfortunately.

Hello

I've just started having a java application bomb out not long into 
running as well where previously it would run for hours without 
problems. However, unlike  yourselves I'm running -ck and the only 
change between the last working -ck and this kernel are the 3 security 
patches. I haven't investigated because I cant take the machine offline, 
but I suspect it's one of those possibly interfering. Looking at the 
patches in question I have no idea how they could do it. I guess if you 
can try backing them out it would be helpful. Here are links to the 
patches in question.

http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tables.patch
http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-2004-0497.patch
http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.patch

Con

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: post 2.6.7 BK change breaks Java?

[José de Paula]

I found that compiling 2.6.7-bk18 and 2.6.7-bk20 with  GCC 2.95 does
away with these random Java segmentation faults. Kernels compiled with
GCC 3.3, 3.2 and 3.0 on the same system, however, have this Java
problem. I hope this helps.

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Fri, 9 Jul 2004, Con Kolivas wrote:

> Jesper Juhl wrote:
> > On Tue, 6 Jul 2004, Matthias Andree wrote:
> >
> >
> >>Hi,
> >>
> >>I've pulled from the linux-2.6 BK tree some post-2.6.7 version,
> compiled
> >>and installed it, and it breaks Java, standalone or plugged into
> >>firefox, the symptom is that the application catches SIGKILL. This
> >>didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.
> >>
> >
> > I'm seeing the same thing. I'm using Eclipse a lot which is Java
> based,
> > and I noticed that wen I went from plain 2.6.7 to 2.6.7-mm3 Eclipse
> > started dying shortly after launch (it only manages to get the splash
> > screen up) with a message about the JVM dying. Since I had also
> upgraded
> > my Sun Java at the same time I initially suspected that and back down
> to
> > my old version, but the problem persisted. Then I tried the latest
> Java
> > release from Sun, with same result. Then I started suspecting the
> kernel
> > and tried 2.6.7-mm6, 2.6.7-bk20 and 2.6.7-mm7 - all with the same
> result
> > that Java breaks. Finally I went back to a plain 2.6.7 and the problem
> > went away - so it certainly looks kernel related.
> > I was using the same .config with all kernels (copied from my plain
> 2.6.7
> > kernel to the others and then running 'make oldconfig'), so I'm also
> > pretty sure it's not due to some new kernel option I've enabled that I
> > don't usually use.
> >
> > My hardware is AMD Athlon (t-bird) 1.4GHz CPU in a ASUS A7M266 mobo
> with
> > 512MB of DDR266 RAM.
> >
> >
> >
> >>Is there any particular change I should try backing out?
> >>
> >
> > I'm looking for the same thing, haven't found it yet unfortunately.
> Hello
> I've just started having a java application bomb out not long into
> running as well where previously it would run for hours without
> problems. However, unlike  yourselves I'm running -ck and the only
> change between the last working -ck and this kernel are the 3 security
> patches. I haven't investigated because I cant take the machine offline,
>
> but I suspect it's one of those possibly interfering. Looking at the
> patches in question I have no idea how they could do it. I guess if you
> can try backing them out it would be helpful. Here are links to the
> patches in question.
> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> es.patch
> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> 4-0497.patch
> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> tch

Thanks Con, I'll try playing with those tomorrow (got no time tonight),
and report back.

Ohh, and in case it's significant, I'm using gcc 3.4.0


--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Go Taniguchi]

Hi list.

Jesper Juhl wrote:
> On Fri, 9 Jul 2004, Con Kolivas wrote:
> 
> 
>>>On Tue, 6 Jul 2004, Matthias Andree wrote:
>>>>Hi,
>>>>
>>>>I've pulled from the linux-2.6 BK tree some post-2.6.7 version,
>>compiled
>>>>and installed it, and it breaks Java, standalone or plugged into
>>>>firefox, the symptom is that the application catches SIGKILL. This
>>>>didn't happen with stock 2.6.7 and doesn't happen with 2.6.6 either.

If I removed this changeset, java worked.
http://linux.bkbits.net:8080/linux-2.6/cset@1.1743

Is this "rt_sigsuspend() and sigaltstack() prototype change" right?
If so, how do we resolve user side?

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Fri, 9 Jul 2004, Jesper Juhl wrote:

> On Fri, 9 Jul 2004, Con Kolivas wrote:
>
> >
> > but I suspect it's one of those possibly interfering. Looking at the
> > patches in question I have no idea how they could do it. I guess if you
> > can try backing them out it would be helpful. Here are links to the
> > patches in question.
> > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> > es.patch
> > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> > 4-0497.patch
> > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> > tch
>
> Thanks Con, I'll try playing with those tomorrow (got no time tonight),
> and report back.
>
Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
thing. The JVM still dies when I try to run eclipse.

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Con Kolivas]

[-- Attachment #1: Type: text/plain, Size: 1053 bytes --]

Jesper Juhl wrote:
> On Fri, 9 Jul 2004, Jesper Juhl wrote:
> 
> 
>>On Fri, 9 Jul 2004, Con Kolivas wrote:
>>
>>
>>>but I suspect it's one of those possibly interfering. Looking at the
>>>patches in question I have no idea how they could do it. I guess if you
>>>can try backing them out it would be helpful. Here are links to the
>>>patches in question.
>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
>>>es.patch
>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
>>>4-0497.patch
>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
>>>tch
>>
>>Thanks Con, I'll try playing with those tomorrow (got no time tonight),
>>and report back.
>>
> 
> Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> thing. The JVM still dies when I try to run eclipse.

Sorry someone else reported success with this:

quote:
------
If I removed this changeset, java worked.
http://linux.bkbits.net:8080/linux-2.6/cset@1.1743
------

Sorry I was hoping others saw this.

Con

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

Re: post 2.6.7 BK change breaks Java?

[Grzegorz Kulewski]

On Sun, 11 Jul 2004, Jesper Juhl wrote:

> On Fri, 9 Jul 2004, Jesper Juhl wrote:
> 
> > On Fri, 9 Jul 2004, Con Kolivas wrote:
> >
> > >
> > > but I suspect it's one of those possibly interfering. Looking at the
> > > patches in question I have no idea how they could do it. I guess if you
> > > can try backing them out it would be helpful. Here are links to the
> > > patches in question.
> > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> > > es.patch
> > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> > > 4-0497.patch
> > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> > > tch
> >
> > Thanks Con, I'll try playing with those tomorrow (got no time tonight),
> > and report back.
> >
> Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> thing. The JVM still dies when I try to run eclipse.

I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other not 
related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try 
reverting some patches. Maybe there is some other problem in -mm that 
gives similar results?

PS. I am using Gentoo and it compiles eclipse from sources. But all others 
Java apps (including Azureus - built very similar to eclipse and installed 
from binaries) work for me too.


Grzegorz Kulewski

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Sun, 11 Jul 2004, Con Kolivas wrote:
> Jesper Juhl wrote:
> > Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> > thing. The JVM still dies when I try to run eclipse.
> Sorry someone else reported success with this:
> quote:
> ------
> If I removed this changeset, java worked.
> http://linux.bkbits.net:8080/linux-2.6/cset@1.1743
> ------
> Sorry I was hoping others saw this.

Yeah, now that you mention it I see the mail from Go Taniguchi, missed it
before.

I can confirm his findings. I just now grabbed that cset as a diff and
backed it out of 2.6.7-mm7 and that fixes the problem.

Now all we need it someone knowledgable about the changes made by that
cset to review it and find out exactely what causes the breakage.

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:

> On Sun, 11 Jul 2004, Jesper Juhl wrote:
>
> > On Fri, 9 Jul 2004, Jesper Juhl wrote:
> >
> > > On Fri, 9 Jul 2004, Con Kolivas wrote:
> > >
> > > >
> > > > but I suspect it's one of those possibly interfering. Looking at the
> > > > patches in question I have no idea how they could do it. I guess if you
> > > > can try backing them out it would be helpful. Here are links to the
> > > > patches in question.
> > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> > > > es.patch
> > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> > > > 4-0497.patch
> > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> > > > tch
> > >
> > > Thanks Con, I'll try playing with those tomorrow (got no time tonight),
> > > and report back.
> > >
> > Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> > thing. The JVM still dies when I try to run eclipse.
>
> I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other not
> related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
> reverting some patches. Maybe there is some other problem in -mm that
> gives similar results?
>

with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
back on the results of that in a few minutes.

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Sun, 11 Jul 2004, Jesper Juhl wrote:

> On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
>
> > On Sun, 11 Jul 2004, Jesper Juhl wrote:
> >
> > > On Fri, 9 Jul 2004, Jesper Juhl wrote:
> > >
> > > > On Fri, 9 Jul 2004, Con Kolivas wrote:
> > > >
> > > > >
> > > > > but I suspect it's one of those possibly interfering. Looking at the
> > > > > patches in question I have no idea how they could do it. I guess if you
> > > > > can try backing them out it would be helpful. Here are links to the
> > > > > patches in question.
> > > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> > > > > es.patch
> > > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> > > > > 4-0497.patch
> > > > > http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> > > > > tch
> > > >
> > > > Thanks Con, I'll try playing with those tomorrow (got no time tonight),
> > > > and report back.
> > > >
> > > Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> > > thing. The JVM still dies when I try to run eclipse.
> >
> > I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other not
> > related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
> > reverting some patches. Maybe there is some other problem in -mm that
> > gives similar results?
> >
>
> with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
> http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
> issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
> back on the results of that in a few minutes.
>
2.6.7-bk20 minus the cset works.

Testing with 2.6.8-rc1 and backing out one or both of the changes in the
cset I get these results :
2.6.8-rc1       - vanilla                                       - breaks Java
2.6.8-rc1-jju1  - both changes backed out                       - works
2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  - breaks Java

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Brian Gerst]

Jesper Juhl wrote:
> On Sun, 11 Jul 2004, Jesper Juhl wrote:
> 
> 
>>On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
>>
>>
>>>On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>
>>>
>>>>On Fri, 9 Jul 2004, Jesper Juhl wrote:
>>>>
>>>>
>>>>>On Fri, 9 Jul 2004, Con Kolivas wrote:
>>>>>
>>>>>
>>>>>>but I suspect it's one of those possibly interfering. Looking at the
>>>>>>patches in question I have no idea how they could do it. I guess if you
>>>>>>can try backing them out it would be helpful. Here are links to the
>>>>>>patches in question.
>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
>>>>>>es.patch
>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
>>>>>>4-0497.patch
>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
>>>>>>tch
>>>>>
>>>>>Thanks Con, I'll try playing with those tomorrow (got no time tonight),
>>>>>and report back.
>>>>>
>>>>
>>>>Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
>>>>thing. The JVM still dies when I try to run eclipse.
>>>
>>>I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other not
>>>related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
>>>reverting some patches. Maybe there is some other problem in -mm that
>>>gives similar results?
>>>
>>
>>with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
>>http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
>>issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
>>back on the results of that in a few minutes.
>>
> 
> 2.6.7-bk20 minus the cset works.
> 
> Testing with 2.6.8-rc1 and backing out one or both of the changes in the
> cset I get these results :
> 2.6.8-rc1       - vanilla                                       - breaks Java
> 2.6.8-rc1-jju1  - both changes backed out                       - works
> 2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
> 2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  - breaks Java
> 
> --
> Jesper Juhl <juhl-lkml@dif.dk>

Looks like a GCC (gcc version 3.4.1 20040702 (Red Hat Linux 3.4.1-2)) 
screwup:

sys_sigaltstack:
         movl    4(%esp), %eax
         movl    8(%esp), %edx
         movl    56(%esp), %ecx
         jmp     do_sigaltstack

The offsets should be 4 more, to account for the return address on the 
stack.

--
				Brian Gerst

Re: post 2.6.7 BK change breaks Java?

[Brian Gerst]

Brian Gerst wrote:
> Jesper Juhl wrote:
> 
>> On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>
>>
>>> On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
>>>
>>>
>>>> On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>>
>>>>
>>>>> On Fri, 9 Jul 2004, Jesper Juhl wrote:
>>>>>
>>>>>
>>>>>> On Fri, 9 Jul 2004, Con Kolivas wrote:
>>>>>>
>>>>>>
>>>>>>> but I suspect it's one of those possibly interfering. Looking at the
>>>>>>> patches in question I have no idea how they could do it. I guess 
>>>>>>> if you
>>>>>>> can try backing them out it would be helpful. Here are links to the
>>>>>>> patches in question.
>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl 
>>>>>>>
>>>>>>> es.patch
>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200 
>>>>>>>
>>>>>>> 4-0497.patch
>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa 
>>>>>>>
>>>>>>> tch
>>>>>>
>>>>>>
>>>>>> Thanks Con, I'll try playing with those tomorrow (got no time 
>>>>>> tonight),
>>>>>> and report back.
>>>>>>
>>>>>
>>>>> Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
>>>>> thing. The JVM still dies when I try to run eclipse.
>>>>
>>>>
>>>> I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other 
>>>> not
>>>> related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
>>>> reverting some patches. Maybe there is some other problem in -mm that
>>>> gives similar results?
>>>>
>>>
>>> with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
>>> http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
>>> issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
>>> back on the results of that in a few minutes.
>>>
>>
>> 2.6.7-bk20 minus the cset works.
>>
>> Testing with 2.6.8-rc1 and backing out one or both of the changes in the
>> cset I get these results :
>> 2.6.8-rc1       - vanilla                                       - 
>> breaks Java
>> 2.6.8-rc1-jju1  - both changes backed out                       - works
>> 2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
>> 2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  - 
>> breaks Java
>>
>> -- 
>> Jesper Juhl <juhl-lkml@dif.dk>
> 
> 
> Looks like a GCC (gcc version 3.4.1 20040702 (Red Hat Linux 3.4.1-2)) 
> screwup:
> 
> sys_sigaltstack:
>         movl    4(%esp), %eax
>         movl    8(%esp), %edx
>         movl    56(%esp), %ecx
>         jmp     do_sigaltstack
> 
> The offsets should be 4 more, to account for the return address on the 
> stack.

Nevermind, I should have looked more carefully.  The offsets are fine in 
my example.  What version of GCC are you using?

-- 
				Brian Gerst

Re: post 2.6.7 BK change breaks Java?

[Jesper Juhl]

On Sun, 11 Jul 2004, Brian Gerst wrote:

> Brian Gerst wrote:
> > Jesper Juhl wrote:
> >
> >> On Sun, 11 Jul 2004, Jesper Juhl wrote:
> >>
> >>
> >>> On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
> >>>
> >>>
> >>>> On Sun, 11 Jul 2004, Jesper Juhl wrote:
> >>>>
> >>>>
> >>>>> On Fri, 9 Jul 2004, Jesper Juhl wrote:
> >>>>>
> >>>>>
> >>>>>> On Fri, 9 Jul 2004, Con Kolivas wrote:
> >>>>>>
> >>>>>>
> >>>>>>> but I suspect it's one of those possibly interfering. Looking at the
> >>>>>>> patches in question I have no idea how they could do it. I guess
> >>>>>>> if you
> >>>>>>> can try backing them out it would be helpful. Here are links to the
> >>>>>>> patches in question.
> >>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
> >>>>>>>
> >>>>>>> es.patch
> >>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
> >>>>>>>
> >>>>>>> 4-0497.patch
> >>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
> >>>>>>>
> >>>>>>> tch
> >>>>>>
> >>>>>>
> >>>>>> Thanks Con, I'll try playing with those tomorrow (got no time
> >>>>>> tonight),
> >>>>>> and report back.
> >>>>>>
> >>>>>
> >>>>> Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
> >>>>> thing. The JVM still dies when I try to run eclipse.
> >>>>
> >>>>
> >>>> I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other
> >>>> not
> >>>> related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
> >>>> reverting some patches. Maybe there is some other problem in -mm that
> >>>> gives similar results?
> >>>>
> >>>
> >>> with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
> >>> http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
> >>> issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
> >>> back on the results of that in a few minutes.
> >>>
> >>
> >> 2.6.7-bk20 minus the cset works.
> >>
> >> Testing with 2.6.8-rc1 and backing out one or both of the changes in the
> >> cset I get these results :
> >> 2.6.8-rc1       - vanilla                                       -
> >> breaks Java
> >> 2.6.8-rc1-jju1  - both changes backed out                       - works
> >> 2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
> >> 2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  -
> >> breaks Java
> >>
> >> --
> >> Jesper Juhl <juhl-lkml@dif.dk>
> >
> >
> > Looks like a GCC (gcc version 3.4.1 20040702 (Red Hat Linux 3.4.1-2))
> > screwup:
> >
> > sys_sigaltstack:
> >         movl    4(%esp), %eax
> >         movl    8(%esp), %edx
> >         movl    56(%esp), %ecx
> >         jmp     do_sigaltstack
> >
> > The offsets should be 4 more, to account for the return address on the
> > stack.
>
> Nevermind, I should have looked more carefully.  The offsets are fine in
> my example.  What version of GCC are you using?
>
gcc 3.4.0

I got a patch from Linus yesterday that seems to fix it nicely on top of
2.6.8-rc1. I guess he has his reasons for not CC'ing it to the list, but
I've given him feedback on my testing of it, so I hope it'll surface as
soon as he's happy with it.

--
Jesper Juhl <juhl-lkml@dif.dk>

Re: post 2.6.7 BK change breaks Java?

[Brian Gerst]

Jesper Juhl wrote:
> On Sun, 11 Jul 2004, Brian Gerst wrote:
> 
> 
>>Brian Gerst wrote:
>>
>>>Jesper Juhl wrote:
>>>
>>>
>>>>On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>>
>>>>
>>>>
>>>>>On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
>>>>>
>>>>>
>>>>>
>>>>>>On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>>>On Fri, 9 Jul 2004, Jesper Juhl wrote:
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>>On Fri, 9 Jul 2004, Con Kolivas wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>>>but I suspect it's one of those possibly interfering. Looking at the
>>>>>>>>>patches in question I have no idea how they could do it. I guess
>>>>>>>>>if you
>>>>>>>>>can try backing them out it would be helpful. Here are links to the
>>>>>>>>>patches in question.
>>>>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl
>>>>>>>>>
>>>>>>>>>es.patch
>>>>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200
>>>>>>>>>
>>>>>>>>>4-0497.patch
>>>>>>>>>http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa
>>>>>>>>>
>>>>>>>>>tch
>>>>>>>>
>>>>>>>>
>>>>>>>>Thanks Con, I'll try playing with those tomorrow (got no time
>>>>>>>>tonight),
>>>>>>>>and report back.
>>>>>>>>
>>>>>>>
>>>>>>>Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't change a
>>>>>>>thing. The JVM still dies when I try to run eclipse.
>>>>>>
>>>>>>
>>>>>>I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few other
>>>>>>not
>>>>>>related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
>>>>>>reverting some patches. Maybe there is some other problem in -mm that
>>>>>>gives similar results?
>>>>>>
>>>>>
>>>>>with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
>>>>>http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
>>>>>issue. I'm currently building 2.6.7-bk20 minus that cset and I'll report
>>>>>back on the results of that in a few minutes.
>>>>>
>>>>
>>>>2.6.7-bk20 minus the cset works.
>>>>
>>>>Testing with 2.6.8-rc1 and backing out one or both of the changes in the
>>>>cset I get these results :
>>>>2.6.8-rc1       - vanilla                                       -
>>>>breaks Java
>>>>2.6.8-rc1-jju1  - both changes backed out                       - works
>>>>2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
>>>>2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  -
>>>>breaks Java
>>>>
>>>>--
>>>>Jesper Juhl <juhl-lkml@dif.dk>
>>>
>>>
>>>Looks like a GCC (gcc version 3.4.1 20040702 (Red Hat Linux 3.4.1-2))
>>>screwup:
>>>
>>>sys_sigaltstack:
>>>        movl    4(%esp), %eax
>>>        movl    8(%esp), %edx
>>>        movl    56(%esp), %ecx
>>>        jmp     do_sigaltstack
>>>
>>>The offsets should be 4 more, to account for the return address on the
>>>stack.
>>
>>Nevermind, I should have looked more carefully.  The offsets are fine in
>>my example.  What version of GCC are you using?
>>
> 
> gcc 3.4.0
> 
> I got a patch from Linus yesterday that seems to fix it nicely on top of
> 2.6.8-rc1. I guess he has his reasons for not CC'ing it to the list, but
> I've given him feedback on my testing of it, so I hope it'll surface as
> soon as he's happy with it.

I see Linus commited a changeset that avoids a tailcall from this 
function, which messes up the stack if CONFIG_REGPARM=n.  Specifically, 
it clobbers %edx in the pt_regs image:

sys_sigaltstack:
         movl    56(%esp), %eax
         movl    %eax, 12(%esp)
         jmp     do_sigaltstack

--
				Brian Gerst

Re: post 2.6.7 BK change breaks Java?

[Matthias Andree]

On Mon, 12 Jul 2004, Brian Gerst wrote:

[a hell of a quote that got snipped]

> >gcc 3.4.0
> >
> >I got a patch from Linus yesterday that seems to fix it nicely on top of
> >2.6.8-rc1. I guess he has his reasons for not CC'ing it to the list, but
> >I've given him feedback on my testing of it, so I hope it'll surface as
> >soon as he's happy with it.
> 
> I see Linus commited a changeset that avoids a tailcall from this 
> function, which messes up the stack if CONFIG_REGPARM=n.  Specifically, 
> it clobbers %edx in the pt_regs image:
> 
> sys_sigaltstack:
>         movl    56(%esp), %eax
>         movl    %eax, 12(%esp)
>         jmp     do_sigaltstack

After a BK pull and a rebuild, Java is well again.

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95 (PGP/MIME preferred)

Re: post 2.6.7 BK change breaks Java?

[Linus Torvalds]

On Mon, 12 Jul 2004, Brian Gerst wrote:
> 
> I see Linus commited a changeset that avoids a tailcall from this 
> function, which messes up the stack if CONFIG_REGPARM=n.  Specifically, 
> it clobbers %edx in the pt_regs image

Yes. 

There's more information about why this is needed in the commit message,
but it basically boils down to an optimization in the system call path
where the save-register-area is used _directly_ as the system call
argument frame on the stack. It's a very worthwhile optimization, since it 
works very well in 99% of all cases and it makes a very hot path faster, 
but gcc thinking that the callee owns the arguments has bitten us before, 
and there is no way to tell gcc that it is wrong.

Personally, I think that if you wanted the callee to own the arguments, 
you should use the "stdcall" attribute, which _literally_ is all about the 
callee owning the stack (and can often generate better code for _other_ 
reasons than tail-calls), but lacking that, I'd have preferred some other 
explicit way to tell gcc about the special "asmlinkage" rules.

No such way exists, so I just fooled gcc by hand. Ugly.

		Linus

Re: post 2.6.7 BK change breaks Java?

[Bill Davidsen]

Brian Gerst wrote:
> Brian Gerst wrote:
> 
>> Jesper Juhl wrote:
>>
>>> On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>
>>>
>>>> On Sun, 11 Jul 2004, Grzegorz Kulewski wrote:
>>>>
>>>>
>>>>> On Sun, 11 Jul 2004, Jesper Juhl wrote:
>>>>>
>>>>>
>>>>>> On Fri, 9 Jul 2004, Jesper Juhl wrote:
>>>>>>
>>>>>>
>>>>>>> On Fri, 9 Jul 2004, Con Kolivas wrote:
>>>>>>>
>>>>>>>
>>>>>>>> but I suspect it's one of those possibly interfering. Looking at 
>>>>>>>> the
>>>>>>>> patches in question I have no idea how they could do it. I guess 
>>>>>>>> if you
>>>>>>>> can try backing them out it would be helpful. Here are links to the
>>>>>>>> patches in question.
>>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1100_ip_tabl 
>>>>>>>>
>>>>>>>> es.patch
>>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1105_CAN-200 
>>>>>>>>
>>>>>>>> 4-0497.patch
>>>>>>>> http://ck.kolivas.org/patches/2.6/2.6.7/2.6.7-ck5/split-out/1110_proc.pa 
>>>>>>>>
>>>>>>>> tch
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Thanks Con, I'll try playing with those tomorrow (got no time 
>>>>>>> tonight),
>>>>>>> and report back.
>>>>>>>
>>>>>>
>>>>>> Ok, got them all 3 backed out of 2.6.7-mm7 , but that doesn't 
>>>>>> change a
>>>>>> thing. The JVM still dies when I try to run eclipse.
>>>>>
>>>>>
>>>>>
>>>>> I can run Eclipse without any problems on 2.6.7-bk20-ck5 + few 
>>>>> other not
>>>>> related patches. Maybe try using non -mm? Try 2.6.7-bk20 and then try
>>>>> reverting some patches. Maybe there is some other problem in -mm that
>>>>> gives similar results?
>>>>>
>>>>
>>>> with plain 2.6.7-bk20 I see the issue, same with 2.6.7-mm7. Reverting
>>>> http://linux.bkbits.net:8080/linux-2.6/cset@1.1743 from -mm7 fixes the
>>>> issue. I'm currently building 2.6.7-bk20 minus that cset and I'll 
>>>> report
>>>> back on the results of that in a few minutes.
>>>>
>>>
>>> 2.6.7-bk20 minus the cset works.
>>>
>>> Testing with 2.6.8-rc1 and backing out one or both of the changes in the
>>> cset I get these results :
>>> 2.6.8-rc1       - vanilla                                       - 
>>> breaks Java
>>> 2.6.8-rc1-jju1  - both changes backed out                       - works
>>> 2.6.8-rc1-jju2  - only first change (sys_rt_sigsuspend) applied - works
>>> 2.6.8-rc1-jju3  - only second change (sys_sigaltstack) applied  - 
>>> breaks Java
>>>
>>> -- 
>>> Jesper Juhl <juhl-lkml@dif.dk>
>>
>>
>>
>> Looks like a GCC (gcc version 3.4.1 20040702 (Red Hat Linux 3.4.1-2)) 
>> screwup:
>>
>> sys_sigaltstack:
>>         movl    4(%esp), %eax
>>         movl    8(%esp), %edx
>>         movl    56(%esp), %ecx
>>         jmp     do_sigaltstack
>>
>> The offsets should be 4 more, to account for the return address on the 
>> stack.
> 
> 
> Nevermind, I should have looked more carefully.  The offsets are fine in 
> my example.  What version of GCC are you using?
> 
Not related to passing args in registers or some such, is it? I did a 
brief flirtation with 7-bk20 and didn't see a problem. I have NOT tried 
that option, perhaps some folks did?

-- 
    -bill davidsen (davidsen@tmr.com)
"The secret to procrastination is to put things off until the
  last possible moment - but no longer"  -me