Date: Wed, 22 Jul 2015 12:08:37 +0200 From: Mattia Rossi <mattia.rossi.mailinglists@gmail.com> To: Warner Losh <imp@bsdimp.com> Cc: Ronald Klop <ronald-lists@klop.ws>, "freebsd-arm@freebsd.org" <freebsd-arm@freebsd.org> Subject: Re: executable gone after fsck - Fwd: Cron <root@sheeva> newsyslog Message-ID: <55AF6BA5.4070400@gmail.com> In-Reply-To: <4CF1F0A4-9486-46DC-AF73-3E98DAB284DD@bsdimp.com> References: <20150709200019.E7710D975C@sheeva.klop.ws> <op.x1js2ee0kndu52@ronaldradial.radialsg.local> <55A652D6.7010401@gmail.com> <4CF1F0A4-9486-46DC-AF73-3E98DAB284DD@bsdimp.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Warner, thanks for fixing this. I can confirm that fsck works on the Dreamplug. Cheers, Mat Am 15.07.2015 um 18:16 schrieb Warner Losh: >> On Jul 15, 2015, at 6:32 AM, Mattia Rossi <mattia.rossi.mailinglists@gmail.com> wrote: >> >> Hi Ronald, >> >> I might have missed some information on the list about the topic, and unless the unmapped i/o was the cause, fsck is pretty much broken at least on the Dreamplug.. and I would suspect also on the Sheevaplug. Wih the first few fsck you do you will lose random files, later on you won't be able to boot anymore, as fsck won't be able to fix fs errors anymore… > It was broken on atmel until I turned of unmapped I/O. then it was rock solid on atmel, at least as rock solid as it is on other platforms. Ian told me his dream plug was also broken in a similar way, but I can’t recall if he tested it for me there. I crashed my Atmel AT91SAM9G20 system about 100 times chasing unrelated bugs after this change. Before I turned off unmapped I/O it was always a disaster (never once did fsck do something useful on a dirty fs). Afterwards every time i was paying attention, it just worked (and when I wasn’t paying attention, it didn’t have the mandatory trip into single user mode). > > fsck definitely does unaligned direct I/O, which is what I think is broken with unmapped I/O. I’m confident that I’ve fixed this. > > The kernel Ron is using should be plenty new enough since I fixed this back in march, iirc. > > Warner > >> I'm currently using the fix suggested by Ian at the time: set the dump and pass values of your root fs in fstab to 0 0 and live with a dirty fs after a crash. Unless the system is crashing often, you'll be quite fine ;-) >> >> Cheers, >> >> Mat >> >> Am 10.07.2015 um 10:12 schrieb Ronald Klop: >>> Hi, >>> >>> I used fsck after a panic on my sheevaplug. >>> FreeBSD sheeva.klop.ws 11.0-CURRENT FreeBSD 11.0-CURRENT #8 r284499M: Wed Jun 17 12:54:02 CEST 2015 root@sjakie.klop.ws:/usr/obj-arm/arm.arm/usr/src-arm/sys/SHEEVAPLUG arm >>> >>> If I did not made a mistake somewhere I'm running with the unmapped i/o turned off. >>> http://www.secnetix.de/olli/FreeBSD/svnews/index.py?r=283931 >>> >>> NB: In the case below bzip2 does not exist anymore. Which is not something _just_ written to disk. >>> >>> Unfortunately I don't have more information about the crash. >>> >>> Ronald. >>> >>> >>> ------- Forwarded message ------- >>> From: "Cron Daemon" <root@sheeva.klop.ws> >>> To: root@klop.ws >>> Cc: >>> Subject: Cron <root@sheeva> newsyslog >>> Date: Thu, 09 Jul 2015 22:00:10 +0200 >>> >>> newsyslog: execl(`/usr/bin/bzip2 -f /var/log/console.log.0'): No such file >>> or directory >>> newsyslog: `bzip2 -f /var/log/console.log.0' terminated with a non-zero >>> status (1) >>> _______________________________________________ >>> freebsd-arm@freebsd.org mailing list >>> http://lists.freebsd.org/mailman/listinfo/freebsd-arm >>> To unsubscribe, send any mail to "freebsd-arm-unsubscribe@freebsd.org" >> _______________________________________________ >> freebsd-arm@freebsd.org mailing list >> http://lists.freebsd.org/mailman/listinfo/freebsd-arm >> To unsubscribe, send any mail to "freebsd-arm-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?55AF6BA5.4070400>