Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 29 Mar 2013 17:06:52 +0100
From:      =?ISO-8859-1?Q?Ren=E9_Ladan?= <rene@freebsd.org>
To:        Ian Lepore <ian@FreeBSD.org>
Cc:        arm@FreeBSD.org
Subject:   Re: signal 11 after kernel update r247742 -> r248706
Message-ID:  <5155BC1C.2000104@freebsd.org>
In-Reply-To: <515466B5.4020403@freebsd.org>
References:  <51545932.9050901@freebsd.org> <1364484652.36972.81.camel@revolution.hippie.lan> <515466B5.4020403@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 28-03-2013 16:50, René Ladan wrote:
> On 28-03-2013 16:30, Ian Lepore wrote:
>> On Thu, 2013-03-28 at 15:52 +0100, René Ladan wrote:
>>> Hi,
>>>
>>> after updating the kernel on my Raspberry PI (config RPI-B with ums
>>> added) from r247742 to r248706 a lot of programs (including the r247742
>>> init) started to crash with signal 11. Manually copying files from
>>> /usr/obj to the SD card (using another FreeBSD box) helped to get
>>> userland back up and running, but now a lot of ports have similar
>>> problems. For some binaries, copying the libraries was enough, other
>>> binaries had to be copied themselves as well.
>>>
>>> Some questions:
>>> - has anyone else experienced this (not necessarily on arm) ? I can try
>>> to repeat this on a i386 VirtualBox host.
>>> - the 20130304 entry in src/UPDATING about callout(9) looks a bit
>>> suspicious?
>>> - I have some kdump/core dump output available from python27 and pkgng
>>> for analysis, if anyone wants to take a look?
>>> - I could manually update pkgng and then portmaster -f all ports, but
>>> this would take some time on the Pi.
>>> - would a binary search through the revisions help here (I guess there
>>> should be one where the r248706 init will not work) ?
>>>
>> I'm at r248509 on my rpi and everything is good; maybe that'll help
>> narrow the binary search.
> Ok, I'll start looking from there.

The 'skeleton' unmapped IO commit was r248508, so that one is OK.
There seem to be some unmapped IO changes between r248509 and r248706,
(e.g. r248510-r248512, r248514-r248522, r248550, r248568-r248569,
r248596) but I think it is more related to something closer to userland?

But the other commits between r248508 and r248706 look OK to me
(including r248534 (jilles, SOCK_CLOEXEC, SOCK_NONLBOCK, MSG_CMSG_CLOEXEC))

René





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5155BC1C.2000104>