From mboxrd@z Thu Jan 1 00:00:00 1970 Message-ID: <4BA74820.2090401@domain.hid> Date: Mon, 22 Mar 2010 11:36:16 +0100 From: Stefan Kisdaroczi MIME-Version: 1.0 References: <4BA39C3D.1040802@domain.hid> <4BA4EF75.7010202@domain.hid> In-Reply-To: <4BA4EF75.7010202@domain.hid> Content-Type: multipart/signed; micalg=pgp-sha1; protocol="application/pgp-signature"; boundary="------------enigCA6217EB41BB095D5680F42C" Subject: Re: [Xenomai-help] crash after termination List-Id: Help regarding installation and common use of Xenomai List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Jan Kiszka Cc: xenomai-help This is an OpenPGP/MIME signed message (RFC 2440 and 3156) --------------enigCA6217EB41BB095D5680F42C Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: quoted-printable Am 20.03.2010 16:53, schrieb Jan Kiszka: > Stefan Kisdaroczi wrote: >> hi all, >> >> after termination of my xenomai applications, my box crashed often (3 = of 4 times) with xenomai 2.5.1. >> Today I updated to current git-2.5, rebuilt xenomai and the kernel. St= ill crashes. >> I have captured two traces over a serial connection. I have attached t= he kernel config, the two >> traces and the cpuinfo (UP x86 Pentium III). Help is welcome. Thank yo= u. >=20 > Both crashes seem to be caused by corrupted skbs. There are (at least) > two uncommon network component in your kernel: ARCNET and ISDN. Do you > make use of them or are they just passively loaded? If the former: Can Just passively loaded. > you retry without them? >(The Kernel's ISDN stuff is of, well, mixed > quality, and ARCNET is so old and uncommon that the code /may/ have > bit-rotted - I disable them. > just an attempt to detract the attention from Xenomai. :) ) I think the problem is in my apps, but I had some hope that some changes for 2.5.2 (about thread cleanup,fork) would help. > What other network components are actively involved in your scenario? Only TCP/IP (non-realtime). Currently a dozen xenomai native tasks are doing standard socket calls. If everything works, there will be about 60 tasks using socket calls, but its still a long way. I focus now first on another problem I have in my apps with a shared heap= , so one question about that: There was no H_NONCACHED flag at the time I did write some library code. I use a rt_heap as shared memory between 8 processes on x86. According to the documentation H_NONCACHED is only relevant for ARM, is that true? Thanks for the help Jan. Stefan --------------enigCA6217EB41BB095D5680F42C Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iD8DBQFLp0glIPTw9rIdn6oRApCeAJ9szaQsxvTnUT0tVDKd8m3sm2lEcQCfbwAe wnr0dc+wXVPB/XfTOjmfNKQ= =QL1+ -----END PGP SIGNATURE----- --------------enigCA6217EB41BB095D5680F42C--