Date: Mon, 8 Sep 2008 23:43:31 +0300 From: Kostik Belousov <kostikbel@gmail.com> To: "Sean C. Farley" <scf@freebsd.org> Cc: freebsd-emulation@freebsd.org, peter@freebsd.org Subject: Re: Linux applications core if running (k)qemu Message-ID: <20080908204331.GC39652@deviant.kiev.zoral.com.ua> In-Reply-To: <20080908203423.GA12147@saturn.kn-bremen.de> References: <alpine.BSF.1.10.0808291711580.5866@thor.farley.org> <20080830113448.GA2152@dchagin.dialup.corbina.ru> <alpine.BSF.2.00.0809021552040.7934@thor.farley.org> <20080906104659.GA2113@dchagin.dialup.corbina.ru> <200809062215.m86MF6NS040797@saturn.kn-bremen.de> <20080907215300.GH2038@deviant.kiev.zoral.com.ua> <20080908203423.GA12147@saturn.kn-bremen.de>
next in thread | previous in thread | raw e-mail | index | archive | help
--+xNpyl7Qekk2NvDX Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On Mon, Sep 08, 2008 at 10:34:23PM +0200, Juergen Lock wrote: > On Mon, Sep 08, 2008 at 12:53:00AM +0300, Kostik Belousov wrote: > > On Sun, Sep 07, 2008 at 12:15:06AM +0200, Juergen Lock wrote: > > > In article <20080906152929.GB2038@deviant.kiev.zoral.com.ua> you writ= e: > > > >-=3D-=3D-=3D-=3D-=3D- > > > > > > > >On Sat, Sep 06, 2008 at 02:46:59PM +0400, Chagin Dmitry wrote: > > > >> On Tue, Sep 02, 2008 at 03:56:33PM -0500, Sean C. Farley wrote: > > > >> > On Sat, 30 Aug 2008, Chagin Dmitry wrote: > > > >> >=20 > > > >> > >On Fri, Aug 29, 2008 at 05:29:09PM -0500, Sean C. Farley wrote: > > > >> > >>I am having trouble with kqemu.ko and linux.ko. If I run qemu= with > > > >> > >>the following command, Linux applications (chroot, acroread, l= s) will > > > >> > >>start core dumping: > > > >> > >> qemu-system-x86_64 -m 512 \ > > > >> > >> -drive file=3D/usr/QEMU/WinXP/c.img,if=3Dide,media=3Ddisk = -boot c \ > > > >> > >> -std-vga -parallel none -serial none -monitor stdio \ > > > >> > >> -net nic,model=3De1000 -net tap,ifname=3Dtap0,script=3Dno = -localtime > > > >> > >> > > > >> > >>Loading kqemu.ko does not cause the problem, but the cores sta= rt a > > > >> > >>little after WinXP starts running. Unloading kqemu.ko does no= t help; > > > >> > >>the cores still happen but more randomly. I even tried unload= ing all > > > >> > >>linux modules and reloading them without luck. It takes a reb= oot. > > > >> > >> > > > >> > >>Packages: > > > >> > >>qemu-devel-0.9.1s.20080620_1 > > > >> > >>kqemu-kmod-devel-1.4.0.p1 > > > >> > >>linux_base-f8-8_4 > > > >> > >> > > > >> > >>sysctl: > > > >> > >>compat.linux.osrelease: 2.6.16 > > > >> > >> > > > >> > >>dmesg: > > > >> > >>kqemu version 0x00010400 > > > >> > >>kqemu: KQEMU installed, max_locked_mem=3D1792492kB. > > > >> > >> > > > >> > >>System is 7-STABLE as of r181963 with or without the patch to = fix RT > > > >> > >>signals from Chagin. > > > >> > > > > > >> > >Interestingly... Sean, can you provide ktrace/kdump log of cori= ng > > > >> > >apps? thnx! > > > >> >=20 > > > >> > Here they are (good and bad): > > > >> > http://www.farley.org/freebsd/tmp/linuxulator_vs_kqemu/ > > > >> >=20 > > > >> > The good trace is after the bad trace. I just kept running ktra= ce > > > >> > /compat/linux/bin/date over and over until I got a good trace. = Before > > > >> > loading kqemu and running qemu, there were no core dumps. Also,= I > > > >> > compared two bad traces and they were basically the same except = for PID > > > >> > and a couple of addresses (still very close in value). > > > >> >=20 > > > >>=20 > > > >> Most likely it is a tls problem again, some days ago kib@ has made= MFC > > > >> r182684, probably it will help.. > > > > > > > >I doubt it. This seems to be an ingenious kqemu bug. As far as I rem= ember, > > > >it tries to use GDT/LDT. This probably has unwanted interaction with > > > >PCB_GS32BIT. > > >=20 > > > Wow. That corner of the code had escaped me so far, and yes this (in > > > amd64/linux32) looks like it won't like kqemu's seperating of the gdts > > > on SMP indeed. (it stores a pointer to &gdt[GUGS32_SEL] in pcb_gs32p= and > > > lets linux processes manipulate the segment pointed to by it, and when > > > kqemu is (or was) running this won't be used by all cpus, see older t= hreads > > > like > > > http://lists.freebsd.org/pipermail/freebsd-emulation/2008-May/004902= .html > > > for the reasons.) > > >=20 > > > What I wonder tho is, won't this also cause problems without kqemu w= hen > > > there are linux processes running on multiple cpus that manipulate th= is > > > segment because the gdt is then shared between the cpus? (like, linux > > > process on cpu 0 changes the segment, then linux process on cpu 1 com= es > > > along and changes it again and then the linux process on cpu 0 will p= ick > > > it up from cpu 1?) At least I must have somehow assumed the shared g= dt > > > wouldn't be changed later because of reasons like this... > >=20 > > Very nice catch! Me and Peter Wemm discussed the right approach, > > that consists of actually providing per-cpu GDT. Patch is at > > http://people.freebsd.org/~kib/misc/amd64_gdt.1.patch > >=20 > > Please, test and give a feedback. Even reports about thinks working > > the same as before the patch are important. >=20 > OK I just tested the patch on RELENG_7 (updated my amd64 SMP box from > RELENG_7_0) and found no problems. (I tested linux date(1), googleearth, > kqemu, and a few other non-linux things so far.) Thank you for the testing. Patch was committed to HEAD already (separated into four mostly self-contained commits). I expect the MFC in one week, your testing is important for MFC decision. --+xNpyl7Qekk2NvDX Content-Type: application/pgp-signature Content-Disposition: inline -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.9 (FreeBSD) iEYEARECAAYFAkjFjnIACgkQC3+MBN1Mb4jH6wCg70X/w6ynwSrLxDf6/XJ+uAY2 pYYAoJe0V3jDN/PmXny1BoV4gKZcGKOi =uwmL -----END PGP SIGNATURE----- --+xNpyl7Qekk2NvDX--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20080908204331.GC39652>