Re: Porting sbcl to DragonFly

[vasily postnicov <shamaz.mazum@xxxxxxxxx>]

--047d7bdca0ba2082fa04d96e09b3
Content-Type: text/plain; charset=ISO-8859-1

> I am still concerned about sb-concurrency's symbol-value-in-thread.3 test
. It always causes sbcl to crash and always in different fashion (sometimes
it just prints "Illegal instruction" and crashes, sometimes sbcl debugger
is invoked with "The integrity of this image is possibly compromised"
message and sometimes there are "[diagnostic] cache_lock: blocked on
0xffffffe00653eac8 sbcl.core" kernel messages)

Cannot fix it. Still no luck. So I put sbcl binary (for x86-64), lisp file,
which can cause a crash, and 2 cores of crashed sbcl into archive and hope,
that someone can help me:

shamazmazum.users.sourceforge.net/sbcl-bin.tar.bz2 (23 M)

The bug has more probability to reproduce, when system is under load (I
used 'make buildworld'). On idle system this test often completes without
any error.

Backtrace usually looks like:
(gdb) bt
#0 0x000000080085f864 in _umtx_wakeup_err () from /usr/lib/libpthread.so.0
Cannot access memory at address 0x80ebcff08

And thread info usually looks like:
(gdb) info thread
   Id     Target     Id       Frame
   6       process   765    0x0000000800859dc8  in ?? () from
/usr/lib/libpthread.so.0
   5       process   711    0x000000080085f864   in  _umtx_wakeup_err
()from /usr/lib/libpthread.so.0
   4       process   705    0x0000000020100019  in  ?? ()
   3       process   1        0x0000001002c458fb   in  ?? ()
* 2       process   724    0x000000080085f864   in  _umtx_wakeup_err() from
/usr/lib/libpthread.so.0

Again, sources can be found here:
https://github.com/shamazmazum/sbcl-dragonfly

in 'dragonfly' branch.

Still no response from sbcl-devel, so you are my only hope ;)

--047d7bdca0ba2082fa04d96e09b3
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div><div><div><div><div>&gt; I am still=20
concerned about sb-concurrency&#39;s symbol-value-in-thread.3 test . It=20
always causes sbcl to crash and always in different fashion (sometimes=20
it just prints &quot;Illegal instruction&quot; and crashes, sometimes sbcl=
=20
debugger is invoked with &quot;The integrity of this image is possibly=20
compromised&quot; message and sometimes there are &quot;[diagnostic] cache_=
lock:=20
blocked on 0xffffffe00653eac8 sbcl.core&quot; kernel messages)<br><br></div=
>Cannot fix it. Still no luck. So I put sbcl binary (for x86-64), lisp file=
, which can cause a crash, and 2 cores of crashed sbcl into archive and hop=
e, that someone can help me:<br>
<br><a href=3D"http://shamazmazum.users.sourceforge.net/sbcl-bin.tar.bz2";>s=
hamazmazum.users.sourceforge.net/sbcl-bin.tar.bz2</a> (23 M)<br><br></div>T=
he bug has more probability to reproduce, when system is under load (I used=
 &#39;make buildworld&#39;). On idle system this test often completes witho=
ut any error.<br>
<br></div><div>Backtrace usually looks like:<br></div><div>(gdb) bt<br></di=
v><div>#0 0x000000080085f864 in _umtx_wakeup_err () from /usr/lib/libpthrea=
d.so.0<br></div><div>Cannot access memory at address 0x80ebcff08<br><br>
</div><div>And thread info usually looks like:<br>(gdb) info thread<br></di=
v><div>=A0=A0 Id=A0=A0=A0=A0 Target=A0=A0=A0=A0 Id=A0=A0=A0=A0=A0=A0 Frame<=
br></div><div>=A0=A0 6=A0=A0=A0=A0=A0=A0 process=A0=A0 765=A0=A0=A0 0x00000=
00800859dc8=A0 in ?? () from /usr/lib/libpthread.so.0<br>
</div><div>=A0=A0 5=A0=A0=A0=A0=A0=A0 process=A0=A0 711=A0=A0=A0 0x00000008=
0085f864=A0=A0 in=A0 _umtx_wakeup_err ()from /usr/lib/libpthread.so.0<br>=
=A0=A0 4=A0=A0=A0=A0=A0=A0 process=A0=A0 705 =A0=A0 0x0000000020100019=A0 i=
n=A0 ?? ()<br>=A0=A0 3=A0=A0=A0=A0=A0=A0 process=A0=A0 1 =A0=A0=A0=A0=A0=A0=
 0x0000001002c458fb=A0=A0 in=A0 ?? ()<br>
* 2=A0=A0=A0=A0=A0=A0 process=A0=A0 724 =A0=A0 0x000000080085f864=A0=A0 in=
=A0 _umtx_wakeup_err() from /usr/lib/libpthread.so.0<br><br>Again, sources =
can be found here:<br></div><a href=3D"https://github.com/shamazmazum/sbcl-=
dragonfly">https://github.com/shamazmazum/sbcl-dragonfly</a><br>
<br></div>in &#39;dragonfly&#39; branch.<br><br></div>Still no response fro=
m sbcl-devel, so you are my only hope ;)<br><br><br></div>

--047d7bdca0ba2082fa04d96e09b3--