--Signature_Thu__23_Jun_2005_07_00_15_-0400_J51ZHBCBi5mKbpFp
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Hello,

> I've spent the better part of this week trying to test some "useful"
> if not "heavy" X apps:
>=20
> gedit
> galeon
> firefox
> firefox-gtk1
> mozilla-gtk2
>=20
> I've managed to compile all of the above from pkgsrc on an
> ultra5/400mhz:
>=20
> NetBSD sparc 3.99.6 NetBSD 3.99.6 (GENERIC) #0: Sun Jun 19 16:08:44
> UTC 2005=20
jmc@faith.netbsd.org:/home/builds/ab/HEAD/sparc64/200506180000Z-obj/hom
e/builds/ab/HEAD/src/sys/arch/sparc64/compile/GENERIC
> sparc64
>=20
> Both firefox's load initially, but then core dump (memory fault or bus
> error) fairly soon. There's nothing particular that forces them to do
> this, as sometimes I can access pages that would core dump previously.

There's a bug that leads to register content during syscalls, Martin is
working on it. This would affect all gecko-based applications and almost
certainly others too, but gecko seems to trigger it the most. The
symptom would be a very short or completely bogus stacktrace after such
a crash and the crashes themselves would be pretty random. Besides that
the gecko code contains various 64bit-related bugs, most should be fixed
in mozilla but I'm not sure all patches made it into the firefox
package.

> However mozilla, gedit and galeon never manage to load before core
dumping.


> All of them have:
>=20
> thread_resume_suspend_cb: td_thr_suspend(0x459840): generic error.
> [Switching to LWP 1]
> Stopped due to shared library event
> (gdb) cont
> Continuing.
> thread_resume_suspend_cb: td_thr_suspend(0x459840): generic error.
> 0x000000004030530c in _rtld_debug_state () from /usr/libexec/ld.elf_so

that's not what makes the apps crash. Start them and attach gdb later to
get past this and see the real crash ( at least that's the case here -
I get this error immediately after starting ff or mozilla in gdb ).
Anything that uses gecko should start slow enough to give you plenty of
time to do that.

> Does anyone else have these apps running OK?

I doubt it. Mozilla starts and even works for a while here but sooner or
later it will run into the syscall bug and crash.

have fun
Michael

--Signature_Thu__23_Jun_2005_07_00_15_-0400_J51ZHBCBi5mKbpFp
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (NetBSD)

iQEVAwUBQrqWP8pnzkX8Yg2nAQI4dAf/UXGyk6tAiUb6O7RUBadzoxYWOMw0pbCb
j/MhucfUaKY9WEmO1sPuZn/mb1UmLnhumd1HpZggznd9t7kHFNPuF72G/KEtQLAM
X1Je712QyX5MVLPOCqWzZhEBiO8IZrecMm1koNDtabOCEaVxwSo3x149QPJ8WWpO
FYtqGT4GkrN8F/0afevnSP5ybJydgfpfL8xJSwwE6y8vC+5U7V20lhOPxEZCbsjm
6vK5BTkTv3iVuvqBuVtXUG6ILKu1y4+btZCndE1V40P+XhoKTxLSTaGZV6Js2z6o
2XRMhKtc1dnohVO/ARf4QiRpE2w1x/b/nXXm2PHmiXctAnz6JzYX/w==
=Xmqy
-----END PGP SIGNATURE-----

--Signature_Thu__23_Jun_2005_07_00_15_-0400_J51ZHBCBi5mKbpFp--