Date: Fri, 11 Nov 2016 18:20:59 +0300 From: Michael Zhilin <mizhka@gmail.com> To: Adrian Chadd <adrian.chadd@gmail.com> Cc: "freebsd-mips@freebsd.org" <freebsd-mips@freebsd.org>, Brooks Davis <brooks@freebsd.org>, Ruslan Bukin <br@freebsd.org> Subject: Re: mips24k (and others?) no longer works Message-ID: <CAF19XBLWZDyifwmQZjbDu111u795KXttM4BW-oi0rJT3C%2BEpRw@mail.gmail.com> In-Reply-To: <CAF19XBLdQ28qbSg=D2EgdUqH1KuYm=UODkuB1U8=ioEXAr8tVA@mail.gmail.com> References: <CAJ-VmomDvwQgW4Bs52NtKqv25Lzb%2BejpXApGNcP0hA3s4te4qg@mail.gmail.com> <CAJ-Vmom5ifdn92GiMDwNzkeuRgM=1Rx3_QTe8P3f_%2BedCZkNaQ@mail.gmail.com> <CAF19XBLdQ28qbSg=D2EgdUqH1KuYm=UODkuB1U8=ioEXAr8tVA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi, geom is fine. badaddr = 0x533f44, pc = 0x4001e4, ra = 0, sp = 0x7ffeee90, sr = b413, pid = 1, ASID = 3 bio_request(0x809bd240) from 0x80956b00(ffs.ufs/FBSD) to 0x80894900(ufs/FBSD) cmd 1 bio_request(0x809bd120) from 0x80956dc0(flash/spi0s.rootfs.uzip) to 0x80894d80(flash/spi0s.rootfs.uzip) cmd 1 g_io_deliver(0x809bd120) from 0x80956dc0(flash/spi0s.rootfs.uzip) to 0x80894d80(flash/spi0s.rootfs.uzip) cmd 1 error 0 off 19902464 len 4096 g_io_deliver(0x809bd240) from 0x80956b00(ffs.ufs/FBSD) to 0x80894900(ufs/FBSD) cmd 1 error 0 off 19902464 len 4096 vm_fault(0x80852a50 (pmap 0x80852b0c), 0x533000 (0x533f44), 1, 0) -> 0 at pc 0x4001e4 0x533f44: 0 <----- to be 0x00557ee0 trap type 22 (TLB miss (load or instr. fetch) - user mode) badaddr = 0, pc = 0x4001e8, ra = 0, sp = 0x7ffeee90, sr = b413, pid = 1, ASID = 3 ... db> x/wx 0x533f44 0x533f44: 557ee0 My next check will be CPU cache (KSEG0 vs KSEG1). I'll check it tomorrow ;) Thanks! On Fri, Nov 11, 2016 at 5:49 PM, Michael Zhilin <mizhka@gmail.com> wrote: > Hi Adrian, > > I see bad fault a bit earlier: just after dozens of cycles of "init": at > pc = 0x4001e4, but it cou > > What found by me: after page fault on file objects the page content is > filled with small delay, i.e. it results in race. If you add small piece of > work (black-hole-cpu) after fault (or context switch?), then the page is > filled with data (magic). > > I want to find some time this weekend (tomorrow) to trace geom layer > (ktrace?). > > Thanks! > > On Fri, Nov 11, 2016 at 11:07 AM, Adrian Chadd <adrian.chadd@gmail.com> > wrote: > >> Hiya, >> >> So mips on the atheros 24kc and 74kc parts is just completely borked. >> It varies between "this doesn't even boot" to "all I get are spams of >> failure exceptions", and if i back out a variety of hilarity, i end up >> with a userland that just plainly hangs after userland runs for a >> little while. No CTRL-T, nothing obvious in ddb. But hey, 'cont' in >> ddb now works. >> >> The working release for me is 304552, which is when I finished >> updating the atheros config files to use std.XXXX instead of >> ARXXXX_BASE. >> >> It's going to take some time to find the first point where things >> broke. It's likely going to be a combination of everyone's recent mips >> platform changes (most of the exceptions, the not-booting-at-all) and >> other unrelated kernel changes to other areas (likely the hang.) >> >> I ask that people hold off on the MIPS side of work until I figure out >> what's going on here. >> >> Also - I really do suggest you invest $40 in a carambola 2 so you have >> the AR9331 as a reference platform to do MIPS testing on. These >> mips24kc, FPU-less cut down SoCs are in a lot of things and having now >> heavily broken support for them is kinda sad. :( It's also putting a >> cramp on my ability to finish my net80211 and 11ac changes as a whole >> lot of it depends upon working AP hardware to test against. >> >> Thanks, >> >> >> -adrian >> _______________________________________________ >> freebsd-mips@freebsd.org mailing list >> https://lists.freebsd.org/mailman/listinfo/freebsd-mips >> To unsubscribe, send any mail to "freebsd-mips-unsubscribe@freebsd.org" >> > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAF19XBLWZDyifwmQZjbDu111u795KXttM4BW-oi0rJT3C%2BEpRw>