Varnish LINGER crash on Solaris (Ticket #649)

[Poul-Henning Kamp]

Can I get you to open a ticket with this text in it, so we have
a record of why we are doing this ?

Poul-Henning

In message <AANLkTik-i3Kh45hQTP4zVYzw50Kz2zGvT8xWdpYQS--q at mail.gmail.com>, =?IS
O-8859-1?Q?Jorge_D=EDaz?= writes:
>--===============8061307876579297173==
>Content-Type: multipart/alternative; boundary=0016e659f6d29162560487e1a818
>
>--0016e659f6d29162560487e1a818
>Content-Type: text/plain; charset=ISO-8859-1
>
>Hello,
>
>I am testing Varnish (r4576 <http://varnish-cache.org/changeset/4576>) in
>Solaris 10 5.10 Generic_120011-14 sun4v sparc SUNW,Sun-Fire-T2000.
>We are planning to use a cache like Varnish or Squid and I have followed the
>instructions in
>http://letsgetdugg.com/2009/12/04/varnish-on-solaris/<http://letsgetdugg.com/2009/12/04/varnish-on-solaris/>
>I have the same LINGER crash like in
>#660<http://varnish-cache.org/ticket/660>that has the same root cause
>in
>#649 <http://varnish-cache.org/ticket/649>
>
>I have trying to fix the bug and I have found *the problem is that solaris
>setsockopt returns sometimes EINVAL* when it is no invalid parameters,
>problem found in Java JVM in Solaris:
>*  http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6378870<http://bugs.sun.com/bugdatabase/view_bug.do?bug_id=6378870>
>*
>http://bugs.opensolaris.org/bugdatabase/view_bug.do;jsessionid=7141b1811572e415779f4a711a96?bug_id=6850464<http://bugs.opensolaris.org/bugdatabase/view_bug.do;jsessionid=7141b1811572e415779f4a711a96?bug_id=6850464>
>
>I think the solution is to changed the definition of "TCP_Check" in *
>libvarnish.h* only for Solaris*
>
>/* In Solaris OS, errno == EINVAL is OK because setsockopt(3SOCKET) call
>returns EINVAL when the connection is reset. */
>#if defined (__SVR4) && defined (__sun)
>#define TCP_Check(a) ((a) == 0 || errno == ECONNRESET || errno == ENOTCONN
>|| errno == EINVAL)
>#else
>#define TCP_Check(a) ((a) == 0 || errno == ECONNRESET || errno == ENOTCONN)
>#endif
>
>Do you think it is ok to commit it to trunk ?
>
>Thank you
>*
>
>--0016e659f6d29162560487e1a818
>Content-Type: text/html; charset=ISO-8859-1
>Content-Transfer-Encoding: quoted-printable
>
>Hello,<br><br><p>
>I am testing Varnish (<a class=3D"changeset" href=3D"http://varnish-cache.o=
>rg/changeset/4576" title=3D"Tell FlexeLint=20
>that we ignore returnvalues on purpose.">r4576</a>) in Solaris 10 5.10=20
>Generic_120011-14 sun4v sparc SUNW,Sun-Fire-T2000. <br>
>We are planning to use a cache like Varnish or Squid and I have followed
> the instructions in <a class=3D"ext-link" href=3D"http://letsgetdugg.com/2=
>009/12/04/varnish-on-solaris/"><span class=3D"icon">=A0</span>http://letsge=
>tdugg.com/2009/12/04/varnish-on-solaris/</a>
></p>
>
>I have the same LINGER crash like in <a class=3D"closed ticket" href=3D"htt=
>p://varnish-cache.org/ticket/660" title=3D"defect: Varnish LINGER
> crash in tcp.c (closed: duplicate)">#660</a> that has the same root=20
>cause in <a class=3D"new ticket" href=3D"http://varnish-cache.org/ticket/64=
>9" title=3D"defect: Varnish LINGER
> crash on Solaris (new)">#649</a> <br><br>I have trying to fix the bug and =
>I have found <strong>the problem is=20
>that solaris setsockopt returns sometimes EINVAL</strong> when it is no=20
>invalid parameters, problem found in Java JVM in Solaris:<br>
>* <a class=3D"ext-link" href=3D"http://bugs.sun.com/bugdatabase/view_bug.do=
>?bug_id=3D6378870"><span class=3D"icon">=A0</span>http://bugs.sun.com/bugda=
>tabase/view_bug.do?bug_id=3D6378870</a>
>  <br>
>* <a class=3D"ext-link" href=3D"http://bugs.opensolaris.org/bugdatabase/vie=
>w_bug.do;jsessionid=3D7141b1811572e415779f4a711a96?bug_id=3D6850464"><span =
>class=3D"icon">=A0</span>http://bugs.opensolaris.org/bugdatabase/view_bug.d=
>o;jsessionid=3D7141b1811572e415779f4a711a96?bug_id=3D6850464</a>
><br><br>I think the solution is to changed the definition of "TCP_Chec=
>k" in <b>libvarnish.h</b> only for Solaris<strong><br><br><span style=
>=3D"font-family: courier new,monospace;">/* In Solaris OS, errno =3D=3D EIN=
>VAL is OK because setsockopt(3SOCKET) call returns EINVAL when the connecti=
>on is reset. */</span><br style=3D"font-family: courier new,monospace;">
><span style=3D"font-family: courier new,monospace;">#if defined (__SVR4) &a=
>mp;& defined (__sun) </span><br style=3D"font-family: courier new,monos=
>pace;"><span style=3D"font-family: courier new,monospace;">#define TCP_Chec=
>k(a) ((a) =3D=3D 0 || errno =3D=3D ECONNRESET || errno =3D=3D ENOTCONN || e=
>rrno =3D=3D EINVAL)</span><br style=3D"font-family: courier new,monospace;"=
>>
><span style=3D"font-family: courier new,monospace;">#else</span><br style=
>=3D"font-family: courier new,monospace;"><span style=3D"font-family: courie=
>r new,monospace;">#define TCP_Check(a) ((a) =3D=3D 0 || errno =3D=3D ECONNR=
>ESET || errno =3D=3D ENOTCONN)</span><br style=3D"font-family: courier new,=
>monospace;">
><span style=3D"font-family: courier new,monospace;">#endif</span><br style=
>=3D"font-family: courier new,monospace;"><br><span style=3D"font-weight: no=
>rmal;">Do you think it is ok to commit it to trunk ?<br><br>Thank you<br></=
>span></strong>
>
>--0016e659f6d29162560487e1a818--
>
>
>--===============8061307876579297173==
>Content-Type: text/plain; charset="us-ascii"
>MIME-Version: 1.0
>Content-Transfer-Encoding: 7bit
>Content-Disposition: inline
>
>_______________________________________________
>varnish-dev mailing list
>varnish-dev at varnish-cache.org
>http://lists.varnish-cache.org/mailman/listinfo/varnish-dev
>--===============8061307876579297173==--
>

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.