Date: Thu, 15 Oct 2015 14:38:22 +0200 From: Mattia Rossi <mattia.rossi.mailinglists@gmail.com> To: Konstantin Belousov <kostikbel@gmail.com>, Mattia Rossi <mattia.rossi.mate@gmail.com> Cc: Jason Harmening <jason.harmening@gmail.com>, ARM <freebsd-arm@freebsd.org> Subject: Re: testing of busdma updates for armv5 Message-ID: <561F9E3E.3040903@gmail.com> In-Reply-To: <20151015120015.GA2257@kib.kiev.ua> References: <561AA64C.5010001@gmail.com> <561B679D.8020501@gmail.com> <CAM=8qa=fHXjX7qGeCSc=M2d7jPLb7D5hfR01mV7X5rgOSbOh2g@mail.gmail.com> <561CEBE4.7060304@gmail.com> <561CEE85.1000301@gmail.com> <561E1312.1030908@gmail.com> <CAM=8qaki5FVKTE=oq7AEBNczTF0Wp8yx73BKVGaymTP8j=ORQw@mail.gmail.com> <561F3B00.80404@gmail.com> <20151015120015.GA2257@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
Am 15.10.2015 um 14:00 schrieb Konstantin Belousov: > On Thu, Oct 15, 2015 at 07:34:56AM +0200, Mattia Rossi wrote: >> Hi Jason, >> >> sorry, I don't think your patch has anything to do with it. Since I >> updated the source tree from scratch after the previous patch mess, I >> must have deleted some patch that I had applied and was using for hte >> last year or so, because currently I simply can't get the system to boot >> anymore. Something to do with mounting my ufs root fs... >> >> Fatal kernel mode data abort: 'Alignment Fault 1' >> trapframe: 0xd733a720 >> FSR=00000001, FAR=0000012b, spsr=60000013 > Faulted address is reported as 0x12b, which sounds as if some sort of > NULL or NULL-like pointer dereference was performed. > > The trace below is suspicious, most of the frames are consistent with an > attempt to load a module. spa_* symbols should not have anything common > with UFS, and runtime module load could only happen after the root is > mounted. > > Do you use zfs, or have zfs module loaded, or zfs compiled into the kernel ? > Can you boot into single-user mode ? > > Please look up the source line for the vaccess+0x58 in your kernel. Konstantin, thanks for your Feedback. I realised in the meantime that it was ZFS breaking things here. I'm currently trying to fix it, but I need to perform the following steps, as I can't get into single user mode (loading kernel.bin from uboot) 1) rebuild world and kernel for my external usb stick 2) install world on the usb stick and boot from it 3) reinstall world and kernel (including new zfs modules) on the Dreamplug 4) reboot the Dreamplug and cross my fingers The issue is quite obvious: the kernel modules on the Dreamplug are older than the current kernel (given I updated my svn tree) but given that they're there, the kernel will load them, as there are some ZFS infos on my rootfs which tell that there is/was a ZFS device attached. This obviously results in something bad. After I have performed all the described steps (which between compiling and installing world on flash devices and all the rest, takes a day or two) I'll let you all know whether things work correctly again. Cheers, Mat
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?561F9E3E.3040903>