Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

toolchain/46385: gcc internal error compiling boost-libs on sparc64

2 views
Skip to first unread message

mar...@netbsd.org

unread,
Apr 27, 2012, 6:25:32 AM4/27/12
to toolchai...@netbsd.org, gnats...@netbsd.org, netbs...@netbsd.org
>Number: 46385
>Category: toolchain
>Synopsis: gcc internal error compiling boost-libs on sparc64
>Confidential: no
>Severity: serious
>Priority: medium
>Responsible: toolchain-manager
>State: open
>Class: sw-bug
>Submitter-Id: net
>Arrival-Date: Fri Apr 27 10:25:00 +0000 2012
>Originator: Martin Husemann
>Release: NetBSD 6.99.5
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD whoever-brings-the-night.aprisoft.de 6.99.5 NetBSD 6.99.5 (MODULAR) #166: Wed Apr 25 10:56:51 CEST 2012 mar...@emmas.aprisoft.de:/nelly/usr/src/sys/arch/sparc64/compile/MODULAR sparc64
Architecture: sparc64
Machine: sparc64
>Description:

Trying to build devel/boost-libs on sparc64 -current breaks with an internal
gcc compiler error:

ore was generated by `cc1plus'.
Program terminated with signal 4, Illegal instruction.
#0 0x00000000001ba7a0 in gt_ggc_mx_lang_tree_node ()

Note that the compile uses g++ -O3.

However, it is not clear to me why it gets a signal:

(gdb) x/16i 0x00000000001ba780
0x1ba780 <gt_ggc_mx_lang_tree_node+4480>: nop
0x1ba784 <gt_ggc_mx_lang_tree_node+4484>: ldx [ %l0 + 0x50 ], %o0
0x1ba788 <gt_ggc_mx_lang_tree_node+4488>:
brz,a,pn %o0, 0x1ba79c <gt_ggc_mx_lang_tree_node+4508>
0x1ba78c <gt_ggc_mx_lang_tree_node+4492>: ld [ %l0 + 0x58 ], %g1
0x1ba790 <gt_ggc_mx_lang_tree_node+4496>:
call 0x1b9600 <gt_ggc_mx_lang_tree_node>
0x1ba794 <gt_ggc_mx_lang_tree_node+4500>: nop
0x1ba798 <gt_ggc_mx_lang_tree_node+4504>: ld [ %l0 + 0x58 ], %g1
0x1ba79c <gt_ggc_mx_lang_tree_node+4508>:
brz,pn %g1, 0x1b96fc <gt_ggc_mx_lang_tree_node+252>
=> 0x1ba7a0 <gt_ggc_mx_lang_tree_node+4512>: stx %g1, [ %fp + 0x7f7 ]
0x1ba7a4 <gt_ggc_mx_lang_tree_node+4516>: mov %l0, %g1
0x1ba7a8 <gt_ggc_mx_lang_tree_node+4520>: clr %i1
0x1ba7ac <gt_ggc_mx_lang_tree_node+4524>: ldx [ %g1 + 0x60 ], %o0
0x1ba7b0 <gt_ggc_mx_lang_tree_node+4528>:
brz,pn %o0, 0x1ba7c8 <gt_ggc_mx_lang_tree_node+4552>
0x1ba7b4 <gt_ggc_mx_lang_tree_node+4532>: ldx [ %fp + 0x7f7 ], %g2
0x1ba7b8 <gt_ggc_mx_lang_tree_node+4536>:
call 0x1b9600 <gt_ggc_mx_lang_tree_node>
0x1ba7bc <gt_ggc_mx_lang_tree_node+4540>: stx %g1, [ %fp + 0x7ef ]

Nothing illegal in this instructions, AFAICT, but %fp looks bogus:

g0 *value not available*
g1 0x0 0
g2 0x3bfffffffc000000 4323455642208567296
g3 0x30 48
g4 0x3bfffffffc000000 4323455642208567296
g5 0x1 1
g6 0x0 0
g7 0x40a18040 1084325952
o0 0x0 0
o1 0x1 1
o2 0x7b5000 8081408
o3 0x7f0cc8 8326344
o4 0x7b5c00 8084480
o5 0x200 512
sp 0xffffffffffdfd621 0xffffffffffdfd621
o7 0x1ba704 1812228
l0 0x1 1
l1 0x1 1
l2 0x1 1
l3 0x1 1
l4 0x1 1
l5 0x1 1
l6 0x1 1
l7 0x1 1
i0 0x1 1
i1 0x1 1
i2 0x1 1
i3 0x1 1
i4 0x1 1
i5 0x1 1
fp 0x1 0x1
i7 0x1 1
pc 0x1ba7a0 0x1ba7a0 <gt_ggc_mx_lang_tree_node+4512>
npc 0x1b96fc 0x1b96fc <gt_ggc_mx_lang_tree_node+252>
state 0x4482008204 294238847492
fsr 0x820 [ NXC #11 ]
fprs *value not available*
y 0x0 0
cwp 0x4 4
pstate 0x82 [ IE #7 ]
asi 0x82 130
ccr 0x44 68


>How-To-Repeat:

Seems to be repeatable for me, I'll try on another machine to make sure.

>Fix:
n/a

Martin Husemann

unread,
Apr 27, 2012, 10:05:33 AM4/27/12
to toolchai...@netbsd.org, gnats...@netbsd.org, netbs...@netbsd.org, mar...@netbsd.org
The following reply was made to PR toolchain/46385; it has been noted by GNATS.

From: Martin Husemann <mar...@duskware.de>
To: gnats...@NetBSD.org
Cc:
Subject: Re: toolchain/46385: gcc internal error compiling boost-libs on sparc64
Date: Fri, 27 Apr 2012 16:04:55 +0200

I patched the boost-libs build and see different (but alike) failures
with -O2.

It works with -O1.

Martin

matthew green

unread,
Oct 1, 2016, 4:50:08 PM10/1/16
to toolchai...@netbsd.org, gnats...@netbsd.org, netbs...@netbsd.org, mar...@netbsd.org
The following reply was made to PR toolchain/46385; it has been noted by GNATS.

From: matthew green <m...@eterna.com.au>
To: gnats...@NetBSD.org
Cc:
Subject: Re: toolchain/46385: gcc internal error compiling boost-libs on sparc64
Date: Sun, 02 Oct 2016 07:46:32 +1100

it seems the compiler error is gone on modern boost-libs but now
there is missing "context" (ucontext(2)?) support for sparc*.
it ends up chosing an "unsupported.cpp" from
libs/context/build/Jamfile.v2.


.mrg.

Martin Husemann

unread,
Oct 11, 2016, 11:45:09 AM10/11/16
to toolchai...@netbsd.org, gnats...@netbsd.org, netbs...@netbsd.org, mar...@netbsd.org
The following reply was made to PR toolchain/46385; it has been noted by GNATS.

From: Martin Husemann <mar...@duskware.de>
To: gnats...@NetBSD.org
Cc:
Subject: Re: toolchain/46385: gcc internal error compiling boost-libs on sparc64
Date: Tue, 11 Oct 2016 17:43:39 +0200

On Sat, Oct 01, 2016 at 08:50:01PM +0000, matthew green wrote:
> it seems the compiler error is gone on modern boost-libs but now
> there is missing "context" (ucontext(2)?) support for sparc*.
> it ends up chosing an "unsupported.cpp" from
> libs/context/build/Jamfile.v2.

Oliver asked me to take a look at it again after optimzations, but at that
time pkgsrc jam was too old - I'll check again now.

Guess this PR should be closed.

Martin

0 new messages