KVM (for Kernel-based Virtual Machine) is a full virtualization solution for Linux on x86 hardware. It consists of a loadable kernel module (kvm.ko) and a userspace component. Since the module will be included in the next 2.6.20 kernel, I think it will be nice to create and ebuild for the userspace part ;-) Leo
I see the following issues/TODO's so far. 1: It is a modified version of qemu and installs in the same place. 2: A user group kvm should be created for /dev/kvm 2.1: Or a more generic emul group should be created for all emulation /dev/ entries? 3: A udev rule should be created. If 2.1 is yes one set of rules can be used instead of every package using its own file in /etc/udev/rules.d I am not sure if the changed kvm made to qemu are being upstreamed but if they are it is probably better to wait for that.
Created attachment 108867 [details] first draft kvm ebuild
Created attachment 109180 [details] better version of kvm-12.ebuild
Hmm, the kvm qemu need gcc 3 according to the docs. Have you circumvented this somehow?
(In reply to comment #4) > Hmm, the kvm qemu need gcc 3 according to the docs. Have you circumvented this > somehow? > The ebuild fails with gcc 4.1. I don't think that it was circumvented at all. Ebuild works with gcc 3.3, however, it installs to /usr/local
kvm works here. :) thanks
Created attachment 109596 [details, diff] ebuild patch tp kvm-12-r1 There are several issues with the current ebuild. 1 --prefix is not passed to configure Fixed it now passed --prefix=/usr 2 It always assumes kvm is part of the kernel by paassing --with-patched-kernel Fixed by including ./configure inside the check for kvm NOT in the kernel and else it will pass --with-patched-kernel 3 It should not accept gcc-4 Fixed: Borrowed the check from the original qemu 4 The make install will always try to install the kernel modules failing with sanbox violations. Fixed: Very bad sed fix. Needs more love but my sed skills are almost zero 5 The udev rules where not correctly inserted because the path was wrong Fixed: Path is now set correctly 6 Usage of emake was last time I did an ebuild not permitted. fixed: substituted for make Tested on kernel 2.6.19 on ~amd64
Created attachment 109612 [details] ebuild kvm-12-r2.ebuild Make kvm qemu friendly
Created attachment 109613 [details] files/kvm_add-scripts-qemu-ifup.patch Make kvm qemu friendly
Created attachment 109614 [details] files/kvm_use_bios_files_in_usr_share_kvm.patch Make kvm qemu friendly
Created attachment 109615 [details] files/kvm_use_etc_kvm_kvm-ifup.patch Make kvm qemu friendly
(In reply to comment #8) > Created an attachment (id=109612) [edit] > New qemu friendly kvm > > Make kvm qemu friendly Thanks, but not yet amd64 friendly. The binary on amd64 is qemu-system-x86_64. replace the mv for the binary with the below. if use amd64; then mv ${D}/usr/bin/qemu-system-x86_64 ${D}/usr/bin/kvm-system-x86_64 else mv ${D}/usr/bin/qemu ${D}/usr/bin/kvm
Created attachment 109929 [details] Ebuild for kvm-13 I try to make an ebuild kvm-13. This version doesn't work with the modules in kernel 2.6.20
(In reply to comment #13) > I try to make an ebuild kvm-13. This version doesn't work with the modules in > kernel 2.6.20 KVM-12 works fine for me on AMD64. How does it fail for you?
(In reply to comment #14) > (In reply to comment #13) > > I try to make an ebuild kvm-13. This version doesn't work with the modules in > > kernel 2.6.20 > > KVM-12 works fine for me on AMD64. How does it fail for you? > I have right follow-up the recommendation of the developers see http://thread.gmane.org/gmane.comp.emulators.kvm.devel/1256
> > (In reply to comment #13158) > > > I try to make an ebuild kvm-13. This version doesn't work with the modules in > > > kernel 2.6.20 > > > > KVM-12 works fine for me on AMD64. How does it fail for you? > > > > I have right follow-up the recommendation of the developers see > http://thread.gmane.org/gmane.comp.emulators.kvm.devel/1256159 Oops, KVM-12 != KVM-13 so ignore my message ;-)
Created attachment 110542 [details, diff] fixes typo in kvm-12-r2.ebuild This patch fixes a typo in the kvm-12-r2.ebuild ("./confiure" -> "./configure"). I'm not sure if I'm following proper protocol (upping the revision number?), but I hope I'm not doing anything too egregious.
Created attachment 110544 [details, diff] modifies KVM_EXPECTED_API_VERSION for newer kernels, ebuild will need to be modified to handle this Also, the ebuild will need to be fixed to patch the user/kvmctl.c file for kernels that have (KVM_API_VERSION == 3). The source wouldn't compile on linux-2.6.20-mm1 until I applied the attached patch (from <a href=http://www.mail-archive.com/kvm-devel@lists.sourceforge.net/msg01046.html> http://www.mail-archive.com/kvm-devel@lists.sourceforge.net/msg01046.html</a>), which modifies KVM_EXPECTED_API_VERSION accordingly. I'm not sure how to do this, so I leave it for somebody more knowledgeable.
Why is qemu-img removed from the package? If it is a problem, is it still a problem with KVM-15?
(In reply to comment #19) > Why is qemu-img removed from the package? If it is a problem, is it still a > problem with KVM-15? Because it conflicts with the standard qemu. This way we can have both installed else they would block each other.
Created attachment 111591 [details] kvm-15.ebuild I'm not following why linux_chkconfig_present is used instead of linux_chkconfig_builtin because modules can be overwritten so I have changed that. kvm_add-scripts-qemu-ifup.patch is no longer necessary so that has been removed. I have updated the other two patches and I will post them as well.
Created attachment 111593 [details, diff] kvm_use_bios_files_in_usr_share_kvm-15.patch
Created attachment 111594 [details, diff] kvm_use_etc_kvm_kvm-ifup-15.patch
Created attachment 111721 [details] kvm-14.ebuild I'm obsoleting my kvm-15 ebuild since I couldn't get it to work. I have an even better ebuild using KVM-14 which compiles with GCC 4.1 and the driver modprobes (Yes it's possible) and everything works properly. :-D You should not compile the kernel virtualization modules with this ebuild. This ebuild also uses a lot of patches so make sure you download all the patches with a -14 suffix.
Created attachment 111722 [details, diff] kvm-configure-14.patch
Created attachment 111725 [details, diff] qemu-configure-14.patch
Created attachment 111726 [details, diff] qemu-helper.c-14.patch
Created attachment 111727 [details, diff] qemu-makefile.target-14.patch
Created attachment 111730 [details, diff] qemu-qemu-kvm.c-14.patch
Created attachment 111731 [details, diff] qemu-vl.c-14.patch
Created attachment 111732 [details, diff] scripts-qemu-ifup-14.patch
Also the executable binary is now called qemu-kvm.
kvm-14 doesn't compile here vl.o: In function `main': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:7063: undefined reference to `code_copy_enabled' libqemu.a(exec.o): In function `tb_invalidate_phys_page_range': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:681: undefined reference to `cpu_restore_state' /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:601: undefined reference to `cpu_x86_gen_code' libqemu.a(cpu-exec.o): In function `cpu_x86_exec': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:783: undefined reference to `cc_table' /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:140: undefined reference to `cpu_x86_gen_code' libqemu.a(helper.o): In function `tlb_fill': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:3866: undefined reference to `cpu_restore_state' libqemu.a(helper.o): In function `helper_verw': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2950: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_verr': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2916: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_lar': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2869: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_lsl': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2826: undefined reference to `cc_table' libqemu.a(helper.o): In function `compute_eflags': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: undefined reference to `cc_table' libqemu.a(helper.o):/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: more undefined references to `cc_table' follow libqemu.a(helper2.o): In function `cpu_x86_init': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper2.c:61: undefined reference to `optimize_flags_init' collect2: ld returned 1 exit status make[2]: *** [qemu-kvm] Error 1 make[2]: Leaving directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/i386-kvm' make[1]: *** [subdir-i386-kvm] Error 2 make[1]: Leaving directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu' make: *** [qemu] Error 2
(In reply to comment #33) > kvm-14 doesn't compile here > > vl.o: In function `main': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:7063: undefined > reference to `code_copy_enabled' > libqemu.a(exec.o): In function `tb_invalidate_phys_page_range': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:681: undefined > reference to `cpu_restore_state' > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:601: undefined > reference to `cpu_x86_gen_code' > libqemu.a(cpu-exec.o): In function `cpu_x86_exec': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:783: > undefined reference to `cc_table' > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:140: > undefined reference to `cpu_x86_gen_code' > libqemu.a(helper.o): In function `tlb_fill': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:3866: > undefined reference to `cpu_restore_state' > libqemu.a(helper.o): In function `helper_verw': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2950: > undefined reference to `cc_table' > libqemu.a(helper.o): In function `helper_verr': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2916: > undefined reference to `cc_table' > libqemu.a(helper.o): In function `helper_lar': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2869: > undefined reference to `cc_table' > libqemu.a(helper.o): In function `helper_lsl': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2826: > undefined reference to `cc_table' > libqemu.a(helper.o): In function `compute_eflags': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: > undefined reference to `cc_table' > libqemu.a(helper.o):/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: > more undefined references to `cc_table' follow > libqemu.a(helper2.o): In function `cpu_x86_init': > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper2.c:61: > undefined reference to `optimize_flags_init' > collect2: ld returned 1 exit status > make[2]: *** [qemu-kvm] Error 1 > make[2]: Leaving directory > `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/i386-kvm' > make[1]: *** [subdir-i386-kvm] Error 2 > make[1]: Leaving directory > `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu' > make: *** [qemu] Error 2 > emerge --info?
Same problem here: mythmaster src # emerge --info Portage 2.1.2-r9 (default-linux/amd64/2006.1, gcc-4.1.1, glibc-2.5-r0, 2.6.18.6 x86_64) ================================================================= System uname: 2.6.18.6 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ Gentoo Base System release 1.12.9 Timestamp of tree: Wed, 28 Feb 2007 00:00:10 +0000 distcc 2.18.3 x86_64-pc-linux-gnu (protocols 1 and 2) (default port 3632) [enabled] ccache version 2.4 [enabled] dev-lang/python: 2.4.3-r4 dev-python/pycrypto: 2.0.1-r5 dev-util/ccache: 2.4-r6 sys-apps/sandbox: 1.2.17 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.16.1-r3 sys-devel/gcc-config: 1.3.14 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.17-r1 ACCEPT_KEYWORDS="amd64" AUTOCLEAN="yes" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/gconf /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-O2 -pipe" DISTDIR="/misc/distfiles" FEATURES="autoconfig buildpkg ccache distcc distlocks metadata-transfer parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="http://open-systems.ufl.edu/mirrors/gentoo http://gentoo.oregonstate.edu http://www.ibiblio.org/pub/Linux/distributions/gentoo" LINGUAS="en en_US" MAKEOPTS="-j3" PKGDIR="/misc/packages/x86_64" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/overlays/ifolder-overlay /usr/local/overlays/other-overlay /usr/local/overlays/dhcp-overlay /usr/local/overlays/kvm-overlay" SYNC="rsync://hustle.ulmer.org/gentoo-portage" USE="X alsa amd64 berkdb bitmap-fonts cli cracklib crypt cups dri fortran gnome gpm gtk iconv ipv6 isdnlog libg++ midi ncurses nls nptl nptlonly pam pcre perl ppds pppd python readline reflection sasl session spl ssl tcpd truetype-fonts type1-fonts unicode xorg zlib" ALSA_CARDS="hda-intel usb-audio" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en en_US" USERLAND="GNU" VIDEO_CARDS="fbdev vesa nv nvidia" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, PORTAGE_RSYNC_EXTRA_OPTS mythmaster src #
Currently, I have been testing with a Core Duo 32-bit CPU on the x86 arch which does work. I'm going to have to test this out on a 64-bit system. (In reply to comment #35) > Same problem here: > > mythmaster src # emerge --info > Portage 2.1.2-r9 (default-linux/amd64/2006.1, gcc-4.1.1, glibc-2.5-r0, 2.6.18.6 > x86_64) > ================================================================= > System uname: 2.6.18.6 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ > Gentoo Base System release 1.12.9 > Timestamp of tree: Wed, 28 Feb 2007 00:00:10 +0000 > distcc 2.18.3 x86_64-pc-linux-gnu (protocols 1 and 2) (default port 3632) > [enabled] > ccache version 2.4 [enabled] > dev-lang/python: 2.4.3-r4 > dev-python/pycrypto: 2.0.1-r5 > dev-util/ccache: 2.4-r6 > sys-apps/sandbox: 1.2.17 > sys-devel/autoconf: 2.13, 2.61 > sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 > sys-devel/binutils: 2.16.1-r3 > sys-devel/gcc-config: 1.3.14 > sys-devel/libtool: 1.5.22 > virtual/os-headers: 2.6.17-r1 > ACCEPT_KEYWORDS="amd64" > AUTOCLEAN="yes" > CBUILD="x86_64-pc-linux-gnu" > CFLAGS="-O2 -pipe" > CHOST="x86_64-pc-linux-gnu" > CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config > /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" > CONFIG_PROTECT_MASK="/etc/env.d /etc/gconf /etc/revdep-rebuild /etc/terminfo" > CXXFLAGS="-O2 -pipe" > DISTDIR="/misc/distfiles" > FEATURES="autoconfig buildpkg ccache distcc distlocks metadata-transfer > parallel-fetch sandbox sfperms strict" > GENTOO_MIRRORS="http://open-systems.ufl.edu/mirrors/gentoo > http://gentoo.oregonstate.edu > http://www.ibiblio.org/pub/Linux/distributions/gentoo" > LINGUAS="en en_US" > MAKEOPTS="-j3" > PKGDIR="/misc/packages/x86_64" > PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress > --force --whole-file --delete --delete-after --stats --timeout=180 > --exclude=/distfiles --exclude=/local --exclude=/packages" > PORTAGE_TMPDIR="/var/tmp" > PORTDIR="/usr/portage" > PORTDIR_OVERLAY="/usr/local/overlays/ifolder-overlay > /usr/local/overlays/other-overlay /usr/local/overlays/dhcp-overlay > /usr/local/overlays/kvm-overlay" > SYNC="rsync://hustle.ulmer.org/gentoo-portage" > USE="X alsa amd64 berkdb bitmap-fonts cli cracklib crypt cups dri fortran gnome > gpm gtk iconv ipv6 isdnlog libg++ midi ncurses nls nptl nptlonly pam pcre perl > ppds pppd python readline reflection sasl session spl ssl tcpd truetype-fonts > type1-fonts unicode xorg zlib" ALSA_CARDS="hda-intel usb-audio" > ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file > hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route > share shm softvol" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" > LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses > text" LINGUAS="en en_US" USERLAND="GNU" VIDEO_CARDS="fbdev vesa nv nvidia" > Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, > PORTAGE_RSYNC_EXTRA_OPTS > > mythmaster src # >
I unbundled the source and applied the patches by hand and got: patching file qemu/target-i386/helper.c patch unexpectedly ends in middle of line and patching file qemu/Makefile.target patch unexpectedly ends in middle of line Hunk #2 succeeded at 235 with fuzz 1. I haven't examined it farther to see if it's actually related. Note that I applied them in what I guessed was the order of intrusiveness (may or may not explain the fuzz), and backed-out the ones with the above messages. I get a clean compile without the above patches -- you may want to check to see if those two were properly posted (and at least make ones that patch doesn't complain about :). Thanks for your efforts!
(In reply to comment #37) > I unbundled the source and applied the patches by hand and got: > > patching file qemu/target-i386/helper.c > patch unexpectedly ends in middle of line > > and > > patching file qemu/Makefile.target > patch unexpectedly ends in middle of line > Hunk #2 succeeded at 235 with fuzz 1. > > I haven't examined it farther to see if it's actually related. Note that I > applied them in what I guessed was the order of intrusiveness (may or may not > explain the fuzz), and backed-out the ones with the above messages. > > I get a clean compile without the above patches -- you may want to check to see > if those two were properly posted (and at least make ones that patch doesn't > complain about :). > > Thanks for your efforts! > Sounds good. Were you able to modprobe the kvm-amd driver after it compiled?
Yes, though I did no functional testing. Also, I got this when compiling a stage4 with catalyst (using the ebuild and all patches): * ERROR: --callback failed! * -- Grepping log... -- make[1]: Entering directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/k ernel' make -C /lib/modules/2.6.16.40-xen/build M=`pwd` "$@" make: Entering an unknown directory make: *** /lib/modules/2.6.16.40-xen/build: No such file or directory. Stop. make: Leaving an unknown directory make[1]: *** [all] Error 2 make[1]: Leaving directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/ke rnel' make: *** [kernel] Error 2 !!! ERROR: app-emulation/kvm-14 failed. -- note that 2.6.16.40 is the version of the host kernel, not the version that was compiled/installed inside the catalyst chroot. There is a (probably) a "uname" in there somewhere that should be worked-around...
(In reply to comment #34) > (In reply to comment #33) > > kvm-14 doesn't compile here > > > > vl.o: In function `main': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:7063: undefined > > reference to `code_copy_enabled' > > libqemu.a(exec.o): In function `tb_invalidate_phys_page_range': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:681: undefined > > reference to `cpu_restore_state' > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:601: undefined > > reference to `cpu_x86_gen_code' > > libqemu.a(cpu-exec.o): In function `cpu_x86_exec': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:783: > > undefined reference to `cc_table' > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:140: > > undefined reference to `cpu_x86_gen_code' > > libqemu.a(helper.o): In function `tlb_fill': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:3866: > > undefined reference to `cpu_restore_state' > > libqemu.a(helper.o): In function `helper_verw': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2950: > > undefined reference to `cc_table' > > libqemu.a(helper.o): In function `helper_verr': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2916: > > undefined reference to `cc_table' > > libqemu.a(helper.o): In function `helper_lar': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2869: > > undefined reference to `cc_table' > > libqemu.a(helper.o): In function `helper_lsl': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2826: > > undefined reference to `cc_table' > > libqemu.a(helper.o): In function `compute_eflags': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: > > undefined reference to `cc_table' > > libqemu.a(helper.o):/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: > > more undefined references to `cc_table' follow > > libqemu.a(helper2.o): In function `cpu_x86_init': > > /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper2.c:61: > > undefined reference to `optimize_flags_init' > > collect2: ld returned 1 exit status > > make[2]: *** [qemu-kvm] Error 1 > > make[2]: Leaving directory > > `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/i386-kvm' > > make[1]: *** [subdir-i386-kvm] Error 2 > > make[1]: Leaving directory > > `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu' > > make: *** [qemu] Error 2 > > > > emerge --info? > vrx@TiTaN ~ $ emerge --info Portage 2.1.2-r9 (default-linux/amd64/2006.1/desktop, gcc-4.1.1, glibc-2.5-r0, 2.6.20-gentoo x86_64) ================================================================= System uname: 2.6.20-gentoo x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 4200+ Gentoo Base System release 1.12.9 Timestamp of tree: Fri, 02 Mar 2007 19:01:01 +0000 ccache version 2.4 [enabled] dev-java/java-config: 1.3.7, 2.0.31 dev-lang/python: 2.4.3-r4 dev-python/pycrypto: 2.0.1-r5 dev-util/ccache: 2.4-r6 sys-apps/sandbox: 1.2.17 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.16.1-r3 sys-devel/gcc-config: 1.3.14 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.20-r1 ACCEPT_KEYWORDS="amd64" AUTOCLEAN="yes" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=athlon64 -Os -pipe -fomit-frame-pointer -msse3" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/revdep-rebuild /etc/splash /etc/terminfo" CXXFLAGS="-march=athlon64 -Os -pipe -fomit-frame-pointer -msse3" DISTDIR="/usr/portage/distfiles" FEATURES="autoconfig ccache distlocks metadata-transfer parallel-fetch sandbox sfperms strict userfetch" GENTOO_MIRRORS="http://ftp.belnet.be/mirror/rsync.gentoo.org/gentoo/ ftp://ftp.belnet.be/mirror/rsync.gentoo.org/gentoo/ " MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage /usr/portage/local/layman/kde /usr/portage/local/layman/sunrise" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X aac aalib alsa amd64 arts asf berkdb bitmap-fonts branding cairo cdr cli cracklib crypt cups dbus dri dvd dvdr dvdread eds emboss encode fam ffmpeg firefox flac fortran gdbm gif gpm gstreamer hal iconv ieee1394 ipod ipv6 isdnlog java jpeg kde kdehiddenvisibility ldap libg++ logitech-mouse mad midi mikmod mng mp3 mpeg mysql mysqli ncurses nls nptl nptlonly nvidia ogg opengl pam pcre pdf perl png ppds pppd python qt3 qt4 quicktime readline reflection sdl session sndfile spell spl ssl tcpd theora truetype truetype-fonts type1-fonts unicode vcd vorbis xml xorg xv xvid xvmc zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" USERLAND="GNU" VIDEO_CARDS="nvidia" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, LINGUAS, PORTAGE_RSYNC_EXTRA_OPTS
Fails here too: /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2552: warning: 'e1' may be used uninitialized in this function rm -f libqemu.a ar rcs libqemu.a exec.o kqemu.o qemu-kvm.o cpu-exec.o fpu/softfloat-native.o helper.o helper2.o translate-copy.o disas.o i386-dis.o gcc -L /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/../user -g -o qemu-kvm vl.o osdep.o readline.o monitor.o pci.o console.o loader.o isa_mmio.o cutils.o migration.o block.o block-raw.o block-cow.o block-qcow.o aes.o block-vmdk.o block-cloop.o block-dmg.o block-bochs.o block-vpc.o block-vvfat.o block-qcow2.o scsi-disk.o cdrom.o lsi53c895a.o usb.o usb-hub.o usb-linux.o usb-hid.o usb-ohci.o usb-msd.o ne2000.o rtl8139.o pcnet.o hypercall.o ide.o pckbd.o ps2.o vga.o sb16.o es1370.o dma.o audio.o noaudio.o wavaudio.o sdlaudio.o ossaudio.o alsaaudio.o wavcapture.o fdc.o mc146818rtc.o serial.o i8259.o i8254.o pcspk.o pc.o cirrus_vga.o mixeng.o apic.o parallel.o acpi.o piix_pci.o usb-uhci.o smbus_eeprom.o gdbstub.o sdl.o x_keymap.o vnc.o slirp/cksum.o slirp/if.o slirp/ip_icmp.o slirp/ip_input.o slirp/ip_output.o slirp/slirp.o slirp/mbuf.o slirp/misc.o slirp/sbuf.o slirp/socket.o slirp/tcp_input.o slirp/tcp_output.o slirp/tcp_subr.o slirp/tcp_timer.o slirp/udp.o slirp/bootp.o slirp/debug.o slirp/tftp.o libqemu.a /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/../user/libkvm.a -lm -lz -lasound -lkvm -lSDL -lpthread -lutil -lrt -luuid vl.o: In function `get_clock': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:652: undefined reference to `clock_gettime' vl.o: In function `init_get_clock': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:640: undefined reference to `clock_gettime' block-raw.o: In function `raw_aio_cancel': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:362: undefined reference to `aio_cancel64' /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:366: undefined reference to `aio_error64' /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:366: undefined reference to `aio_error64' block-raw.o: In function `raw_aio_write': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:349: undefined reference to `aio_write64' block-raw.o: In function `raw_aio_read': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:333: undefined reference to `aio_read64' block-raw.o: In function `qemu_aio_poll': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:225: undefined reference to `aio_error64' /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:233: undefined reference to `aio_return64' block-raw.o: In function `qemu_aio_init': /mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/block-raw.c:209: undefined reference to `aio_init' collect2: ld returned 1 exit status make[2]: *** [qemu-kvm] Error 1 make[2]: Leaving directory `/mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu/i386-kvm' make[1]: *** [subdir-i386-kvm] Error 2 make[1]: Leaving directory `/mnt/jaguar/portage/app-emulation/kvm-14/work/kvm-14/qemu' make: *** [qemu] Error 2 !!! ERROR: app-emulation/kvm-14 failed. Call stack: ebuild.sh, line 1614: Called dyn_compile ebuild.sh, line 971: Called qa_call 'src_compile' environment, line 4022: Called src_compile kvm-14.ebuild, line 57: Called die !!! make failed !!! If you need support, post the topmost build error, and the call stack if relevant. !!! A complete build log is located at '/mnt/jaguar/portage/app-emulation/kvm-14/temp/build.log'. !!! This ebuild is from an overlay: '/usr/local/portage' ---------- emerge --info Portage 2.1.2-r13 (default-linux/x86/2006.1, gcc-4.1.1, glibc-2.5-r0, 2.6.19-beyond1.2 i686) ================================================================= System uname: 2.6.19-beyond1.2 i686 Genuine Intel(R) CPU T2300 @ 1.66GHz Gentoo Base System release 1.12.9 Timestamp of tree: Fri, 02 Mar 2007 14:00:01 +0000 dev-java/java-config: 1.3.7, 2.0.31-r3 dev-lang/python: 2.4.4 dev-python/pycrypto: 2.0.1-r5 sys-apps/sandbox: 1.2.18.1 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.17 sys-devel/gcc-config: 1.3.14 sys-devel/libtool: 1.5.23b virtual/os-headers: 2.6.20-r1 ACCEPT_KEYWORDS="x86 ~x86" AUTOCLEAN="yes" CBUILD="i686-pc-linux-gnu" CFLAGS="-O3 -march=pentium4m -mfpmath=sse,387 -mmmx -msse -msse2 -pipe -fomit-frame-pointer -momit-leaf-frame-pointer" CHOST="i686-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-O3 -march=pentium4m -mfpmath=sse,387 -mmmx -msse -msse2 -pipe -fomit-frame-pointer -momit-leaf-frame-pointer" DISTDIR="/usr/portage/distfiles" FEATURES="autoconfig distlocks metadata-transfer parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="http://distfiles.gentoo.org http://distro.ibiblio.org/pub/linux/distributions/gentoo" MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*" PORTAGE_TMPDIR="/mnt/jaguar" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/portage/local/layman/iwlwifi /usr/local/portage /usr/local/portage-xgl" SYNC="rsync://rsync.gentoo.org/gentoo-portage" USE="X aac acpi alsa audacious avi bash-completion bitmap-fonts bzip2 cdparanoia cdr crypt cups curl divx4linux dri dvd dvdr dvdread encode exif firefox gcj gd gif gimpprint gphoto2 gpm gtk gtk2 hal iconv jabber jpeg kde mmx mp3 mpeg mplayer ncurses nptl nptlonly nsplugin offensive ogg oggvorbis openal opengl oss pam pcmcia pcre pdf perl png posix python qt3 qt4 quicktime real sse sse2 ssl tiff truetype truetype-fonts type1-fonts unicode usb v4l vorbis wifi win32codecs x86 xorg xvid xvmc zlib" ALSA_CARDS="hda-intel intel8x0m" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" USERLAND="GNU" VIDEO_CARDS="i810" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, LINGUAS, PORTAGE_RSYNC_EXTRA_OPTS
Shouldn't this be split up in 2 ebuilds? One for the kernel modules and one for the userspace tools?
kvm-14 fails for me with the same undefined references as already reported. However it seems to miss all DEPEND information. To get it to configure I had to merge: media-libs/libsdl media-libs/alsa-lib
(In reply to comment #33) Same here. gcc -L /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/../user -g -o qemu-kvm vl.o osdep.o readline.o monitor.o pci.o console.o loader.o isa_mmio.o cutils.o migration.o block.o block-raw.o block-cow.o block-qcow.o aes.o block-vmdk.o block-cloop.o block-dmg.o block-bochs.o block-vpc.o block-vvfat.o block-qcow2.o scsi-disk.o cdrom.o lsi53c895a.o usb.o usb-hub.o usb-linux.o usb-hid.o usb-ohci.o usb-msd.o ne2000.o rtl8139.o pcnet.o hypercall.o ide.o pckbd.o ps2.o vga.o sb16.o es1370.o dma.o audio.o noaudio.o wavaudio.o sdlaudio.o ossaudio.o alsaaudio.o wavcapture.o fdc.o mc146818rtc.o serial.o i8259.o i8254.o pcspk.o pc.o cirrus_vga.o mixeng.o apic.o parallel.o acpi.o piix_pci.o usb-uhci.o smbus_eeprom.o gdbstub.o sdl.o x_keymap.o vnc.o slirp/cksum.o slirp/if.o slirp/ip_icmp.o slirp/ip_input.o slirp/ip_output.o slirp/slirp.o slirp/mbuf.o slirp/misc.o slirp/sbuf.o slirp/socket.o slirp/tcp_input.o slirp/tcp_output.o slirp/tcp_subr.o slirp/tcp_timer.o slirp/udp.o slirp/bootp.o slirp/debug.o slirp/tftp.o libqemu.a -lm -lz -lasound -lSDL -lpthread -lutil -lrt -luuid vl.o: In function `main': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/vl.c:7063: undefined reference to `code_copy_enabled' libqemu.a(exec.o): In function `tb_invalidate_phys_page_range': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:681: undefined reference to `cpu_restore_state' /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/exec.c:601: undefined reference to `cpu_x86_gen_code' libqemu.a(cpu-exec.o): In function `cpu_x86_exec': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:783: undefined reference to `cc_table' /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/cpu-exec.c:140: undefined reference to `cpu_x86_gen_code' libqemu.a(helper.o): In function `tlb_fill': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:3866: undefined reference to `cpu_restore_state' libqemu.a(helper.o): In function `helper_verw': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2950: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_verr': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2916: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_lar': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2869: undefined reference to `cc_table' libqemu.a(helper.o): In function `helper_lsl': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper.c:2826: undefined reference to `cc_table' libqemu.a(helper.o): In function `compute_eflags': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: undefined reference to `cc_table' libqemu.a(helper.o):/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/exec.h:511: more undefined references to `cc_table' follow libqemu.a(helper2.o): In function `cpu_x86_init': /var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/target-i386/helper2.c:61: undefined reference to `optimize_flags_init' collect2: ld returned 1 exit status make[2]: *** [qemu-kvm] Error 1 make[2]: Leaving directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu/i386-kvm' make[1]: *** [subdir-i386-kvm] Error 2 make[1]: Leaving directory `/var/tmp/portage/app-emulation/kvm-14/work/kvm-14/qemu' make: *** [qemu] Error 2 !!! ERROR: app-emulation/kvm-14 failed. Call stack: ebuild.sh, line 1614: Called dyn_compile ebuild.sh, line 971: Called qa_call 'src_compile' environment, line 4121: Called src_compile kvm-14.ebuild, line 57: Called die !!! make failed !!! If you need support, post the topmost build error, and the call stack if relevant. !!! A complete build log is located at '/var/tmp/portage/app-emulation/kvm-14/temp/build.log'. !!! This ebuild is from an overlay: '/usr/portage/local/clement' bliss app-emulation/kvm # emerge --info Portage 2.1.2.2 (default-linux/amd64/2006.1/desktop, gcc-4.1.1, glibc-2.4-r3, 2.6.19-gentoo-r5 x86_64) ================================================================= System uname: 2.6.19-gentoo-r5 x86_64 Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz Gentoo Base System release 1.12.9 Timestamp of tree: Mon, 16 Apr 2007 01:46:01 +0000 dev-java/java-config: 1.3.7, 2.0.30 dev-lang/python: 2.4.3-r4 dev-python/pycrypto: 2.0.1-r5 sys-apps/sandbox: 1.2.17 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.16.1-r3 sys-devel/gcc-config: 1.3.13-r3 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.17-r2 ACCEPT_KEYWORDS="amd64" AUTOCLEAN="yes" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=nocona -O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/rc.d /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-march=nocona -O2 -pipe" DISTDIR="/usr/portage/distfiles" FEATURES="distlocks metadata-transfer multilib-pkg parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="http://ftp.uoi.gr/mirror/OS/gentoo/ ftp://distro.ibiblio.org/pub/linux/distributions/gentoo/ ftp://ftp.rhnet.is/pub/gentoo/ ftp://ftp.gentoo.mesh-solutions.com/gentoo/ http://distro.ibiblio.org/pub/linux/distributions/gentoo/" MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/portage/local/clement /usr/portage/local/layman/sunrise /usr/portage/local/layman/damn_bluetooth-alsa" SYNC="rsync://rsync.us.gentoo.org/gentoo-portage" USE="X a52 aac acpi aiglx alsa amd64 autoreplace berkdb bitmap-fonts bzip2 cairo cap caps cdr cjk cli cracklib crypt dbus dri dvb dvd dvdr encode fam ffmpeg firefox fortran gdbm gif glitz gpm gstreamer gtk gtk2 hal highlight history hou iconv immqt-bc ipv6 isdnlog jingle jpeg kde kdeenablefinal kdehiddenvisibility kdexdeltas lcms libg++ logrotate mad midi mikmod mmap mp3 mpeg ncurses nls nptl nptlonly nsplugin ogg opengl oss pam pcre perl png ppds pppd python qt3 qt4 quicktime readline reflection rtc sdl session skey sms sou spell spl ssl svg symlink tcpd theora threads timidity tk truetype truetype-fonts type1-fonts unicode vorbis wmp x264 xine xml xorg xv xvid xvmc zlib" ALSA_CARDS="hda-intel usb-audio" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" DVB_CARDS="nxt200x" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LIRC_DEVICES="kworld" USERLAND="GNU" VIDEO_CARDS="nvidia" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, LINGUAS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
The root cause of the undefined references appears to be that in qemu-qemu-kvm.c-14.patch, the new stuff are conditional on CONFIG_KVM. And due to reasons I have not figured out yet, CONFIG_KVM is not defined on 64-bit architectures.
I have redone the patches, there indeed was a mistake in them, which prevented building on amd64. As I'm lazy to upload this (with all the patches): http://g-overlays.googlecode.com/svn/trunk/mozilla/app-emulation/kvm/ I have made little corrections to the build, to install some python scripts, to install again qemu-img to not rename the binary and so on, as I didn't want to install qemu. I haven't look at it still but the udev rule seems not to trigger itself, so chown root:kvm /dev/kvm help in these cases should do it, without it - it will segfault (this is because the qemu without kvm is broken with these patches). Installed win2000, started some images and a gentoo cd - all seems to work with this release - kvm-20...
(In reply to comment #46) > I have redone the patches, there indeed was a mistake in them, which prevented > building on amd64. As I'm lazy to upload this (with all the patches): > > http://g-overlays.googlecode.com/svn/trunk/mozilla/app-emulation/kvm/ > > I have made little corrections to the build, to install some python scripts, to > install again qemu-img to not rename the binary and so on, as I didn't want to > install qemu. I haven't look at it still but the udev rule seems not to trigger > itself, so chown root:kvm /dev/kvm help in these cases should do it, without it > - it will segfault (this is because the qemu without kvm is broken with these > patches). > Installed win2000, started some images and a gentoo cd - all seems to work > with this release - kvm-20... > I have confirmed that these patches allows full compilation of kvm using gcc4 but when attempting to load the module with modprobe kvm-intel, it fails to load. What platform are you using? dmesg shows this: kvm: disabled by bios kvm_intel: Unknown symbol set_cr3 kvm_intel: Unknown symbol gfn_to_page kvm_intel: Unknown symbol segment_base kvm_intel: Unknown symbol kvm_get_msr_common kvm_intel: Unknown symbol kvm_mmu_free_some_pages kvm_intel: Unknown symbol load_msrs kvm_intel: Unknown symbol kvm_write_guest kvm_intel: Unknown symbol kvm_exit_arch kvm_intel: Unknown symbol emulate_instruction kvm_intel: Unknown symbol fx_init kvm_intel: Unknown symbol set_cr4 kvm_intel: Unknown symbol lmsw kvm_intel: Unknown symbol kvm_hypercall kvm_intel: Unknown symbol kvm_init_arch kvm_intel: Unknown symbol set_cr0 kvm_intel: Unknown symbol set_cr8 kvm_intel: Unknown symbol kvm_setup_pio kvm_intel: Unknown symbol kvm_read_guest kvm_intel: Unknown symbol kvm_emulate_cpuid kvm_intel: Unknown symbol kvm_set_msr_common kvm_intel: Unknown symbol save_msrs kvm_intel: Unknown symbol kvm_resched emerge --info: Portage 2.1.2.2 (default-linux/amd64/2006.1, gcc-4.1.1, glibc-2.5-r0, 2.6.20-gentoo-r6 x86_64) ================================================================= System uname: 2.6.20-gentoo-r6 x86_64 Intel(R) Core(TM)2 CPU T7200 @ 2.00GHz Gentoo Base System release 1.12.9 Timestamp of tree: Sun, 22 Apr 2007 19:20:01 +0000 dev-java/java-config: 1.3.7, 2.0.31-r5 dev-lang/python: 2.4.3-r4 dev-python/pycrypto: 2.0.1-r5 sys-apps/sandbox: 1.2.17 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.16.1-r3 sys-devel/gcc-config: 1.3.15-r1 sys-devel/libtool: 1.5.22 virtual/os-headers: 2.6.17-r2 ACCEPT_KEYWORDS="amd64" AUTOCLEAN="yes" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=nocona -O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/java-config/vms/ /etc/php/apache1-php5/ext-active/ /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-march=nocona -O2 -pipe" DISTDIR="/usr/portage/distfiles" FEATURES="distlocks metadata-transfer parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="http://gentoo.cites.uiuc.edu/pub/gentoo/ http://gentoo.mirrors.easynews.com/linux/gentoo/ http://mirror.phy.olemiss.edu/mirror/gentoo" LINGUAS="en ru" MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/local/portage" SYNC="rsync://rsync.namerica.gentoo.org/gentoo-portage" USE="X a52 aac acpi adns alsa amd64 apache2 audiofile avi bcmath bdf berkdb bitmap-fonts bzip2 calendar cdr cjk clamav cli cracklib crypt ctype cups curl curlwrappers dbus dio directfb dlloader dri dts dvb dvd dvdread emu-linux-x86 encode exif fbcon ffmpeg flac foomaticdb fortran ftp gd gdbm gif gmp gnutls gpm gsm gtk gtk2 hal hash iconv ieee1394 imagemagick imap imlib innodb ipv6 isdnlog ithreads java javascript jpeg jpeg2k json kde lcms ldap ldap-sasl libg++ logrotate lzo mad matroska mcal mhash midi mime mjpeg mng mp3 mp4 mpeg msession msn mysql mysqli ncurses network nls nptl nptlonly nsplugin nvidia ogg openal opengl osc oss pam pcntl pcre pdf pdo pear perl php png posix postgres ppds pppd python qt qt3support qt4 quicktime readline reflection samba sasl sdl seamonkey session simplexml snmp soap sockets spl sqlite ssl subversion suhosin svg tcpd theora threads tidy tiff tokenizer transcode truetype truetype-fonts type1-fonts unicode usb v4l v4l2 vcd vorbis wddx wifi wxwindows x264 xforms xine xml xorg xpm xv xvid zip zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="en ru" USERLAND="GNU" VIDEO_CARDS="nv nvidia vesa fbdev v4l" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
Have you loaded kvm before this - as I always loaded it before kvm-intel, I'm on Core 2 Duo with amd64 profile exactly as you.
(In reply to comment #47) > > kvm: disabled by bios I had this same message and was able to fix it by booting to my BIOS and enabling the CPU virtualization extensions and then POWERING DOWN. (I have an orig Core: Yonah though) I rebooted after changing the setting for the first time and was seeing the same exact symptoms. Both modules now load and I'm running QEMU/KVM and access to my wireless device works! (Something VMware couldn't do) - Hope this helps
(In reply to comment #49) > > kvm: disabled by bios > > I had this same message and was able to fix it by booting to my BIOS and > enabling the CPU virtualization extensions and then POWERING DOWN. You are lucky as there are also bioses that do not provide this option. So if someone can not find it, haunt the manufacturer for an update ;-)
(In reply to comment #50) > (In reply to comment #49) > > > kvm: disabled by bios > > > > I had this same message and was able to fix it by booting to my BIOS and > > enabling the CPU virtualization extensions and then POWERING DOWN. > > You are lucky as there are also bioses that do not provide this option. So if > someone can not find it, haunt the manufacturer for an update ;-) > You're right. Looks like I have the bios right before they updated it with the virtualization selection. Looks like I'm going to have to do a bios update. Thanks for pointing that out.
Ok, after a bios update, I got winxp sp2 installed but I haven't figured out how to get it to cooperate with a bridge using dhcp. Does anyone have a network setup that works?
(In reply to comment #52) > Ok, after a bios update, I got winxp sp2 installed but I haven't figured out > how to get it to cooperate with a bridge using dhcp. Does anyone have a > network setup that works? > NM, firewall was blocking it. I have it working.
http://g-overlays.googlecode.com/svn/trunk/mozilla/app-emulation/kvm/ updated to kvm-21, removed the kernel version checks, as there is no kernel with modules, which support kvm-21. linux-2.6.21 modules could be used with kvm-17 linux-2.6.20 modules could be used with kvm-13
i tried to use kvm-21 with kernel 2.6.21 on core2. emerging kvm was succeeded but /usr/bin/kvm failed like: insmod: can't read 'kernel/kvm.ko': No such file or directory insmod: can't read 'kernel/kvm-intel.ko': No such file or directory Traceback (most recent call last): File "/usr/bin/kvm", line 189, in ? raise Exception, 'Unable to determine eth0 mac address' Exception: Unable to determine eth0 mac address kvm.ko and kvm-intel.ko are placed in /lib/modules/2.6.21-gentoo/extra in my environment.
(In reply to comment #55) > i tried to use kvm-21 with kernel 2.6.21 on core2. > > emerging kvm was succeeded but /usr/bin/kvm failed like: > > insmod: can't read 'kernel/kvm.ko': No such file or directory > insmod: can't read 'kernel/kvm-intel.ko': No such file or directory > Traceback (most recent call last): > File "/usr/bin/kvm", line 189, in ? > raise Exception, 'Unable to determine eth0 mac address' > Exception: Unable to determine eth0 mac address > > kvm.ko and kvm-intel.ko are placed in /lib/modules/2.6.21-gentoo/extra in my > environment. > Yes I noticed that too. That file really isn't necessary anyway because it's just a bash script wrapper for qemu with trimmed down options no less. Just use /usr/bin/qemu-system-x86_64 or /usr/bin/qemu depending on what platform you're on instead.
Created attachment 117585 [details] kvm-21.ebuild New and improved kvm-21.ebuild. I have added a new patch to allow the kvm wrapper to work properly. I added usermode-utilities as a dependency since it is necessary for bringing up a tap interface in the qemu-ifup script. I improved the qemu-ifup script. I updated and checked all the patches to make sure they patch properly on this version. I removed all reference to the 2.6.20 kernel. I also added more configuration instructions after emerging the package.
Created attachment 117587 [details, diff] kvm-configure-21.patch
Created attachment 117589 [details, diff] kvm-kvm-21.patch
Created attachment 117590 [details, diff] qemu-configure-21.patch
Created attachment 117592 [details, diff] qemu-helper.c-21.patch
Created attachment 117594 [details, diff] qemu-makefile.target-21.patch
Created attachment 117595 [details, diff] qemu-qemu-kvm.c-21.patch
Created attachment 117596 [details, diff] qemu-vl.c-21.patch
Created attachment 117597 [details, diff] scripts-qemu-ifup-21.patch
It appears to be working fine but the wrapper still has a little problem. Here is a patch for the patched wrapper. --- kvm 2007-04-30 14:56:42.000000000 +0900 +++ /usr/bin/kvm 2007-04-30 15:36:51.000000000 +0900 @@ -137,7 +137,7 @@ if not options.notap: mac = options.mac - if not mac: + if mac: for line in commands.getoutput('/sbin/ip link show eth0').splitlines(): m = re.match(r'.*link/ether (..:..:..:..:..:..).*', line) if m: @@ -148,8 +148,8 @@ mac_components[0] = 'a0' mac = ':'.join(mac_components) - qemu_args += ('-net', 'nic,macaddr=%s,model=rtl8139' % (mac,), - '-net', 'tap,script=/etc/kvm/qemu-ifup',) + qemu_args += ('-net', 'nic,macaddr=%s,model=rtl8139' % (mac,)) + qemu_args += ('-net', 'tap,script=/etc/kvm/qemu-ifup') if options.vnc: qemu_args += ('-vnc', str(options.vnc)) In addtion, there is no consistency between man kvm and kvm --help. I think it's worth makeing a man for the wrapper if it is trivial.
sorry i have mistaken. the wrapper is correct
segment fault when I execute qemu (by the wrapper) with -no-kvm. it appears to be failed by sjlj in cpu_exec.
(In reply to comment #68) > segment fault when I execute qemu (by the wrapper) with -no-kvm. it appears to > be failed by sjlj in cpu_exec. > Yes, this will not work, as the patches remove some of the no-kvm functions to make qemu gcc-4 compatible. So no wonder that it segfaults.
kvm --install can not work because qemu/qemu-img is not found. /usr/bin/kvm:104 'qemu/qemu-img create -f qcow2 "%s" 10G' % disk)
Created attachment 119450 [details, diff] kvm-kvm-24.patch New patch for KVM-24 which will fix the errors with starting up. Removed all capabilities to disable kvm.
Created attachment 119452 [details, diff] qemu-qemu-kvm.c-24.patch This patch didn't patch properly for KVM-24. All other 21 patches that I submitted will work with KVM-24. You just need to change the file name suffix from 21 to 24.
Created attachment 119454 [details] kvm-24.ebuild
plase where i can found kvm-configure-24.patch ? kvm-24.ebuild, line 52: Called epatch '/usr/local/portage/app-emulation/kvm/files/kvm-configure-24.patch' Thank's
(In reply to comment #74) > plase where i can found kvm-configure-24.patch ? > > kvm-24.ebuild, line 52: Called epatch > '/usr/local/portage/app-emulation/kvm/files/kvm-configure-24.patch' > > Thank's > As I mentioned in my last comments, just download the 21 versions that aren't obsoleted since they didn't need to be changed and change the number to 24 in the filename. I saw no point in uploading new files just to change the name of the file since the patch hasn't changed.
(In reply to comment #75) > (In reply to comment #74) > > plase where i can found kvm-configure-24.patch ? > > > > kvm-24.ebuild, line 52: Called epatch > > '/usr/local/portage/app-emulation/kvm/files/kvm-configure-24.patch' > > > > Thank's > > > > As I mentioned in my last comments, just download the 21 versions that aren't > obsoleted since they didn't need to be changed and change the number to 24 in > the filename. I saw no point in uploading new files just to change the name of > the file since the patch hasn't changed. > Hi, I have downloaded the qemu-configure-21.patch and changed the name to 24 and it failed to apply the patch when I tried to: # ebuild kvm-24.ebuild unpack with the following message: * checking kvm-24.tar.gz ;-) ... [ ok ] >>> Checking kvm-24.tar.gz's mtime... >>> Not marked as unpacked; recreating WORKDIR... >>> Unpacking source... >>> Unpacking kvm-24.tar.gz to /var/tmp/portage/app-emulation/kvm-24/work * Applying kvm-configure-24.patch ... [ ok ] * Applying kvm-kvm-24.patch ... [ ok ] * Applying qemu-configure-24.patch ... * Failed Patch: qemu-configure-24.patch ! * ( /srv/overlay/app-emulation/kvm/files/qemu-configure-24.patch ) * * Include in your bugreport the contents of: * * /var/tmp/portage/app-emulation/kvm-24/temp/qemu-configure-24.patch-10018.out !!! ERROR: app-emulation/kvm-24 failed. Call stack: ebuild.sh, line 1615: Called dyn_unpack ebuild.sh, line 752: Called qa_call 'src_unpack' ebuild.sh, line 44: Called src_unpack kvm-24.ebuild, line 54: Called epatch '/srv/overlay/app-emulation/kvm/files/qemu-configure-24.patch' eutils.eclass, line 341: Called die !!! Failed Patch: qemu-configure-24.patch! The mentioned log file is in attachment. thank you very much! Guidolin !!! If you need support, post the topmost build error, and the call stack if relevant. !!! A complete build log is located at '/var/tmp/portage/app-emulation/kvm-24/temp/build.log'. !!! This ebuild is from an overlay: '/srv/overlay'
Created attachment 120155 [details] Error log mentioned on comment 76
Please try it again from <http://gentoo.panicode.com/overlay/app-emulation/kvm/> if you want. But there is NO WARRANTY.
(In reply to comment #77) > Created an attachment (id=120155) [edit] > Error log mentioned on comment 76 > I retested the patch and it patches fine on my end.
(In reply to comment #78) > Please try it again from > <http://gentoo.panicode.com/overlay/app-emulation/kvm/> if you want. But there > is NO WARRANTY. > Hi, I have downloaded all the files and emerged them. It worked for me! Thank you!
(In reply to comment #79) > (In reply to comment #77) > > Created an attachment (id=120155) [edit] > > Error log mentioned on comment 76 > > > > I retested the patch and it patches fine on my end. > Hi, I will try again and try to find out what was the problem on my machine! Thank you anyway!
Created attachment 120283 [details] kvm-25.ebuild I added nomirror to restrict since I got tired of it hanging on mirrors that don't have the file anyway. Once again I have tested all previous patches (they still work) and all they need is just a name change to the 25 version: 117587: kvm-configure-21.patch 117590: qemu-configure-21.patch 117592: qemu-helper.c-21.patch 117594: qemu-makefile.target-21.patch 117596: qemu-vl.c-21.patch 117597: scripts-qemu-ifup-21.patch 119450: kvm-kvm-24.patch 119452: qemu-qemu-kvm.c-24.patch
kvm25 faild to compile with gcc 4.1.2 on git-sources 2.6.22-rc3. please see a build.log.
Created attachment 120488 [details] error log with gcc 4.1.2 on git-sources 2.6.22-rc3
(In reply to comment #83) > kvm25 faild to compile with gcc 4.1.2 on git-sources 2.6.22-rc3. > please see a build.log. > Considering 2.6.22-rc3 is beta software, I wouldn't worry about compilation errors at this point since they go through multiple release candidates and that version is by no means stable.
I just tested out KVM-26. Nothing needs to be changed for it. Just change the kvm-25.ebuild to kvm-26.ebuild and all the latest patches to 26.
(In reply to comment #86) > I just tested out KVM-26. Nothing needs to be changed for it. Just change the > kvm-25.ebuild to kvm-26.ebuild and all the latest patches to 26. > I also tested KVM-26 in this manner on my Athlon 64 x2 running ~amd64. I can provide a emerge --info (mail me).
Hi, i tried kvm-26 as well and it seemed to work fine. However, there is one thing which confuses me: $kvm --image=/data/temp/xp.cow --install --cdrom=/dev/cdrom1 -m 384 --no-tap Warning: No DNS servers found Could not open '/dev/kqemu' - QEMU acceleration layer not activated Why does it complain about /dev/kqemu??? is that alright or does it currently not use acceleration. I thought kvm doesn't need/use kqemu, or does it just use the same /dev??? cheers, and well done with the ebuild, hope it'll be in portage soon florian
(In reply to comment #88) > Hi, > i tried kvm-26 as well and it seemed to work fine. > > However, there is one thing which confuses me: > > $kvm --image=/data/temp/xp.cow --install --cdrom=/dev/cdrom1 -m 384 --no-tap > Warning: No DNS servers found > Could not open '/dev/kqemu' - QEMU acceleration layer not activated > > Why does it complain about /dev/kqemu??? is that alright or does it currently > not use acceleration. I thought kvm doesn't need/use kqemu, or does it just use > the same /dev??? > > cheers, and well done with the ebuild, hope it'll be in portage soon > florian > I'm not sure how you're getting that error. I am not getting that error with those flags. KVM comes with a modified version of qemu which has code for kqemu in it but I'm not quite sure why it would be looking for kqemu. Do an lsmod and make sure kvm and kvm-intel or kvm-amd is loaded. I haven't tried this module with kqemu loaded as well and I'm not even sure if you can load those modules at the same time but just make sure you don't have any conflicts there.
Good news! I just tested the patches with KVM-27 and all the latest patches work without modifications. So just change the names of the patches and the ebuild to have a 27 version number instead of 26 to upgrade. changelog for KVM-27: - use slab cache for mmu shadow pages - make shadow pte updates atomic (for guest smp) - general mmu cleanups
(In reply to comment #90) > Good news! I just tested the patches with KVM-27 and all the latest patches > work without modifications. So just change the names of the patches and the > ebuild to have a 27 version number instead of 26 to upgrade. Easier is to change the naming in the ebuild. The patched don't change so hard code their names.
Created attachment 121167 [details] kvm-27.ebuild Removed patch versions from ebuild.
Created attachment 121170 [details, diff] kvm-configure.patch Removing version number.
Created attachment 121172 [details, diff] qemu-configure.patch Removing version number.
Created attachment 121174 [details, diff] qemu-helper.c.patch Removing version number.
Created attachment 121176 [details, diff] qemu-makefile.target.patch Removing version number.
Created attachment 121177 [details, diff] qemu-vl.c.patch Removing version number.
Created attachment 121179 [details, diff] scripts-qemu-ifup.patch Removing version number.
Created attachment 121181 [details, diff] kvm-kvm.patch Removing version number.
Created attachment 121183 [details, diff] qemu-qemu-kvm.c.patch Removing version number.
16:04 [0] hazynrg $ kvm -m 400 --no-tap --cdrom=/dev/hda --image=winlsd.raw exception 13 (0) rax 000000000000f001 rbx 000000000000d713 rcx 0000000000000001 rdx 0000000000000000 rsi 00000000ffff004c rdi 000000000008f7f4 rsp 000000000000ffb8 rbp 000000000000ffcc r8 0000000000000000 r9 0000000000000000 r10 0000000000000000 r11 0000000000000000 r12 0000000000000000 r13 0000000000000000 r14 0000000000000000 r15 0000000000000000 rip 0000000000000a45 rflags 00033002 cs f000 (000f0000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) ds 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) es 07c0 (00007c00/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) ss 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) fs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) gs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) tr 0000 (19850000/00002088 p 1 dpl 0 db 0 s 0 type b l 0 g 0 avl 0) ldt 0000 (00000000/0000ffff p 1 dpl 0 db 0 s 0 type 2 l 0 g 0 avl 0) gdt fa580/30 idt 0/3ff cr0 60000010 cr2 0 cr3 0 cr4 0 cr8 0 efer 0 code: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 Abandon 16:04 [134] hazynrg $ emerge --info Portage 2.1.2.7 (default-linux/amd64/2007.0/desktop, gcc-4.1.2, glibc-2.5-r2, 2.6.21-gentoo-r2 x86_64) ================================================================= System uname: 2.6.21-gentoo-r2 x86_64 Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz Gentoo Base System release 1.12.10 Timestamp of tree: Wed, 06 Jun 2007 17:20:01 +0000 dev-java/java-config: 1.3.7, 2.0.32 dev-lang/python: 2.4.4-r4, 2.5-r1 dev-python/pycrypto: 2.0.1-r5 sys-apps/sandbox: 1.2.18.1 sys-devel/autoconf: 2.13, 2.61 sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 sys-devel/binutils: 2.17.50.0.16 sys-devel/gcc-config: 1.3.16 sys-devel/libtool: 1.5.23b virtual/os-headers: 2.6.21 ACCEPT_KEYWORDS="amd64 ~amd64" AUTOCLEAN="yes" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=nocona -O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo" CXXFLAGS="-march=nocona -O2 -pipe" DISTDIR="/usr/portage/distfiles" FEATURES="distlocks metadata-transfer parallel-fetch sandbox sfperms strict" GENTOO_MIRRORS="ftp://ftp.free.fr/mirrors/ftp.gentoo.org/" LANG="fr_FR.UTF-8" LC_ALL="fr_FR.UTF-8" LDFLAGS="-Wl,--hash-style=both" LINGUAS="fr" MAKEOPTS="-j3" PKGDIR="/usr/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --delete-after --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages --filter=H_**/files/digest-*" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/usr/portage" PORTDIR_OVERLAY="/usr/portage/local/layman/enlightenment /usr/local/portage" SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage" USE="X acl acpi alsa amd64 bash-completion berkdb bitmap-fonts cairo cdr cli cracklib crypt cups dbus dri dvd dvdr dvdread eds emacs encode fam firefox fortran gdbm gif gpm gstreamer gtk hal hashstyle iconv ipv6 isdnlog jpeg kerberos ldap libg++ midi mmx mp3 mpeg mudflap ncurses nls nptl nptlonly ogg opengl openmp pam pcre pdf perl png pppd python qt3 qt3support qt4 readline reflection sdl session spell spl sse sse2 ssl svg tcpd tiff truetype truetype-fonts type1-fonts unicode usb vorbis xml xorg xv zlib" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" CAMERAS="kodak ptp2" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LINGUAS="fr" USERLAND="GNU" VIDEO_CARDS="nvidia" Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
(In reply to comment #101) > 16:04 [0] hazynrg $ kvm -m 400 --no-tap --cdrom=/dev/hda --image=winlsd.raw > exception 13 (0) > rax 000000000000f001 rbx 000000000000d713 rcx 0000000000000001 rdx > 0000000000000000 > rsi 00000000ffff004c rdi 000000000008f7f4 rsp 000000000000ffb8 rbp > 000000000000ffcc > r8 0000000000000000 r9 0000000000000000 r10 0000000000000000 r11 > 0000000000000000 > r12 0000000000000000 r13 0000000000000000 r14 0000000000000000 r15 > 0000000000000000 > rip 0000000000000a45 rflags 00033002 > cs f000 (000f0000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > ds 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > es 07c0 (00007c00/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > ss 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > fs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > gs 0000 (00000000/0000ffff p 1 dpl 3 db 0 s 1 type 3 l 0 g 0 avl 0) > tr 0000 (19850000/00002088 p 1 dpl 0 db 0 s 0 type b l 0 g 0 avl 0) > ldt 0000 (00000000/0000ffff p 1 dpl 0 db 0 s 0 type 2 l 0 g 0 avl 0) > gdt fa580/30 > idt 0/3ff > cr0 60000010 cr2 0 cr3 0 cr4 0 cr8 0 efer 0 > code: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 > 00 00 00 00 00 00 > Abandon > 16:04 [134] hazynrg $ emerge --info > Portage 2.1.2.7 (default-linux/amd64/2007.0/desktop, gcc-4.1.2, glibc-2.5-r2, > 2.6.21-gentoo-r2 x86_64) > ================================================================= > System uname: 2.6.21-gentoo-r2 x86_64 Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz > Gentoo Base System release 1.12.10 > Timestamp of tree: Wed, 06 Jun 2007 17:20:01 +0000 > dev-java/java-config: 1.3.7, 2.0.32 > dev-lang/python: 2.4.4-r4, 2.5-r1 > dev-python/pycrypto: 2.0.1-r5 > sys-apps/sandbox: 1.2.18.1 > sys-devel/autoconf: 2.13, 2.61 > sys-devel/automake: 1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10 > sys-devel/binutils: 2.17.50.0.16 > sys-devel/gcc-config: 1.3.16 > sys-devel/libtool: 1.5.23b > virtual/os-headers: 2.6.21 > ACCEPT_KEYWORDS="amd64 ~amd64" > AUTOCLEAN="yes" > CBUILD="x86_64-pc-linux-gnu" > CFLAGS="-march=nocona -O2 -pipe" > CHOST="x86_64-pc-linux-gnu" > CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config > /usr/kde/3.5/shutdown /usr/share/X11/xkb /usr/share/config" > CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/gconf > /etc/php/apache2-php5/ext-active/ /etc/php/cgi-php5/ext-active/ > /etc/php/cli-php5/ext-active/ /etc/revdep-rebuild /etc/terminfo" > CXXFLAGS="-march=nocona -O2 -pipe" > DISTDIR="/usr/portage/distfiles" > FEATURES="distlocks metadata-transfer parallel-fetch sandbox sfperms strict" > GENTOO_MIRRORS="ftp://ftp.free.fr/mirrors/ftp.gentoo.org/" > LANG="fr_FR.UTF-8" > LC_ALL="fr_FR.UTF-8" > LDFLAGS="-Wl,--hash-style=both" > LINGUAS="fr" > MAKEOPTS="-j3" > PKGDIR="/usr/portage/packages" > PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress > --force --whole-file --delete --delete-after --stats --timeout=180 > --exclude=/distfiles --exclude=/local --exclude=/packages > --filter=H_**/files/digest-*" > PORTAGE_TMPDIR="/var/tmp" > PORTDIR="/usr/portage" > PORTDIR_OVERLAY="/usr/portage/local/layman/enlightenment /usr/local/portage" > SYNC="rsync://rsync.europe.gentoo.org/gentoo-portage" > USE="X acl acpi alsa amd64 bash-completion berkdb bitmap-fonts cairo cdr cli > cracklib crypt cups dbus dri dvd dvdr dvdread eds emacs encode fam firefox > fortran gdbm gif gpm gstreamer gtk hal hashstyle iconv ipv6 isdnlog jpeg > kerberos ldap libg++ midi mmx mp3 mpeg mudflap ncurses nls nptl nptlonly ogg > opengl openmp pam pcre pdf perl png pppd python qt3 qt3support qt4 readline > reflection sdl session spell spl sse sse2 ssl svg tcpd tiff truetype > truetype-fonts type1-fonts unicode usb vorbis xml xorg xv zlib" > ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x > ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 > trident usb-audio via82xx via82xx-modem ymfpci" ALSA_PCM_PLUGINS="adpcm alaw > asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa > lfloat linear meter mulaw multi null plug rate route share shm softvol" > CAMERAS="kodak ptp2" ELIBC="glibc" INPUT_DEVICES="keyboard mouse" > KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 > mtxorb ncurses text" LINGUAS="fr" USERLAND="GNU" VIDEO_CARDS="nvidia" > Unset: CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, PORTAGE_COMPRESS, > PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS > It doesn't surprise me that that doesn't work. I have yet to get kvm to start up as anything but root and it looks like you're trying this as a regular user. Also it certainly can't help that you have ~amd64 in your keywords. You're just asking for problems when you always emerge testing packages if available.
compiles fine on ~amd64 with 2.6.21.1 kernel, can't test yet...
kvm-27.ebuild builds the kernel module against the currently running kernel, while it should be supposed to build it against whatever kernel is present in /usr/src/linux.
If anyone is interested, I modified the ebuild to be compatible with qemu (arm development, anyone?) It doesn't install qemu-img and the docs, and it make use of the existing files (/etc/qemu-ifup and /usr/share/qemu.) The ebuild: http://finiteless.net/~sio/misc/gentoo/kvm/kvm-28.ebuild It uses most of the patches used in kvm-27.ebuild, plus these three: http://finiteless.net/~sio/misc/gentoo/kvm/no-qemu-img.patch http://finiteless.net/~sio/misc/gentoo/kvm/bin-names.patch http://finiteless.net/~sio/misc/gentoo/kvm/kvm-bin-names.patch The binary name on amd64 is qemu-system-x86_64-kvm, it should be qemu-kvm on i386.
(In reply to comment #105) > If anyone is interested, I modified the ebuild to be compatible with qemu (arm > development, anyone?) It doesn't install qemu-img and the docs, and it make use > of the existing files (/etc/qemu-ifup and /usr/share/qemu.) > > The ebuild: > http://finiteless.net/~sio/misc/gentoo/kvm/kvm-28.ebuild This does not install bios.bin for some reason.
> This does not install bios.bin for some reason. If you don't have /usr/share/qemu/bios.bin at all you should try to reemerge app-emulation/qemu. If you mean the Vista patch... well, let's say I "forgot" about it. :P I'll look into this later today when I get back home. (or get something with ssh)
(In reply to comment #105) Your page does not work. Why not attach files to bugzilla?
Networking doesn't work. IP tunnelling is enabled as builtin in my kernel (2.6.21-gentoo-r3). I'm using kvm-27. (as root:) # kvm Bringing up interface tap0 TUNSETIFF: Device or resource busy /etc/kvm/qemu-ifup: line 19: /sbin/brctl: No such file or directory /etc/kvm/qemu-ifup: could not launch network script Could not initialize device 'tap' #ifconfig -a eth0 Link encap:Ethernet HWaddr 00:50:8D:95:85:FC inet addr:192.168.52.1 Bcast:192.168.52.255 Mask:255.255.255.0 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:223950 errors:0 dropped:0 overruns:0 frame:0 TX packets:214377 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:190609895 (181.7 Mb) TX bytes:72415029 (69.0 Mb) Interrupt:23 Base address:0xa000 eth1 Link encap:Ethernet HWaddr 00:50:8D:95:85:FD inet addr:10.0.0.1 Bcast:10.255.255.255 Mask:255.0.0.0 UP BROADCAST MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) Interrupt:22 Base address:0xc000 eth2 Link encap:UNSPEC HWaddr 00-50-8D-00-00-92-A5-3E-00-00-00-00-00-00-00-00 BROADCAST MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) lo Link encap:Local Loopback inet addr:127.0.0.1 Mask:255.0.0.0 UP LOOPBACK RUNNING MTU:16436 Metric:1 RX packets:749 errors:0 dropped:0 overruns:0 frame:0 TX packets:749 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:49943 (48.7 Kb) TX bytes:49943 (48.7 Kb) tunl0 Link encap:IPIP Tunnel HWaddr NOARP MTU:1480 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:0 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:0 (0.0 b) TX bytes:0 (0.0 b) (note that I have only 2 physical ethernet cards, I don't know what eth2 is supposed to be).
(note that I have only 2 physical ethernet cards, I don't know what eth2 is supposed to be). It could easily be a 1994 (firewire) network connection. Basically if you have a firewire port on your pc and you have the eth1394 module loaded you will get an entry for a firewire network.
(In reply to comment #110) > It could easily be a 1994 (firewire) network connection. Basically if you have > a firewire port on your pc and you have the eth1394 module loaded you will get > an entry for a firewire network. You're right, that must be it.
Created attachment 124934 [details] kvm-29.ebuild Version bump. A few patches had to be updated.
Created attachment 124936 [details, diff] kvm-kvm.patch
Created attachment 124937 [details, diff] qemu-configure.patch
Created attachment 124939 [details, diff] qemu-qemu-kvm.c.patch
Only 3 patches needed to be updated for KVM-29. All other previous patches will work with this version.
Hi, I installed the lastest linux-headers-2.6.22 and now I have a file collision with /usr/include/linux/kvm.h. Both linux-headers and kvm wants to install this file. Any idea?
Traceback (most recent call last): File "/usr/bin/kvm", line 20, in ? external_module = config.get('shell', 'want_module') File "/usr/lib/python2.4/ConfigParser.py", line 511, in get raise NoSectionError(section) ConfigParser.NoSectionError: No section: 'shell'
(In reply to comment #118) > Traceback (most recent call last): > File "/usr/bin/kvm", line 20, in ? > external_module = config.get('shell', 'want_module') > File "/usr/lib/python2.4/ConfigParser.py", line 511, in get > raise NoSectionError(section) > ConfigParser.NoSectionError: No section: 'shell' > I'm getting the same error here. I've fallen back to kvm-27, but now I'm getting this: Traceback (most recent call last): File "/usr/bin/kvm", line 138, in ? raise Exception, 'Unable to determine eth0 mac address' Exception: Unable to determine eth0 mac address Digging into the kvm script, it looks like it's trying to run /sbin/ip, whatever that is. If it wants to get eth0's MAC address, shouldn't it use some invocation of /sbin/ifconfig instead? As an alternative, what ebuild provides /sbin/ip?
(In reply to my previous comment) > Digging into the kvm script, it looks like it's trying to run /sbin/ip, > whatever that is. Turns out it's part of sys-apps/iproute2, which (along with net-misc/bridge-utils) should be marked in the ebuilds as dependencies.
They weren't marked as dependencies because you don't need network access to run kvm. It also requires extra configuration that the ebuild doesn't cover to set up a bridge interface.
Created attachment 125905 [details] KVM 32 ebuild works GCC 4.2 and module build Tested with GCC 4.1 and 4.2 on 2.6.22-gentoo-r1 kernel with -inkernel use flag. User space tools and kernel modules worked on this setup.
Created attachment 125989 [details] kvm-33.ebuild Old gcc patches not necessary anymore. Other patches have been redone where necessary.
Created attachment 125991 [details, diff] kvm-kvm.patch
Created attachment 125993 [details, diff] qemu-configure.patch
Created attachment 125994 [details, diff] qemu-vl.c.patch
Created attachment 125996 [details, diff] scripts-qemu-ifup.patch
In kvm-32.ebuild, I think you should enclose "linux-mod_src_install" with a "if ! use inkernel; then ... fi". Also, why the kvm-33 ebuild is so different from the kvm-32 one?
(In reply to comment #128) > In kvm-32.ebuild, I think you should enclose "linux-mod_src_install" with a "if > ! use inkernel; then ... fi". > > Also, why the kvm-33 ebuild is so different from the kvm-32 one? > The better question would be why 32 is different from 29. It was missing necessary patches and dependencies among other things. This ebuild can't be built with the module that comes with the kernel as no kernel uses this version of kvm and if you tried, kvm would compile but crash when you try to use it (probably due to the version mismatch).
kvm-32 compiles and works fine on ~amd64, 2.6.22-gentoo-r1 kernel with kvm-intel hardcoded. I used USE=inkernel. I'm going to test kvm-33 with the pkg_setup command removed, as it is weird imho.
Created attachment 126072 [details] kvm-33.ebuild ebuild wasn't dying when kvm was built-in or the kernel modules were compiled.
Hi, I use kvm-33.ebuild with a ~amd linux-2.6.22-gentoo-r2 kernel but I had to comment out the first part of the src_unpack to get it working. src_unpack() { # #Don't continue if the KVM modules that come with the kernel are compiled # if linux_chkconfig_present KVM ; then # eerror "kvm-${PV} doesn't work with the kernel modules" # die "kvm module is included in the kernel" # fi Are there some known issues with vde ? default uml network works fine but using tap0 via "vde qemu-system-x86_64 ..." produce a hang up of the guest in my env.
After one week of running some guests (mainly 32bit fedora rawhide) in my AMD64 qemu/kvm-33/linux-2.6.22-gentoo-r1/2 environment I can say it works but not as fine as I hoped. The first problem is the time in the guests. It seems that the guests clock speed is only 70% of the hosts clock speed. I played around a little bit with the hosts kernel options 300/1000HZ and the io-schedulers but no success. Perhaps qemu/kvm + guest have problems with the dynamic cpu speed. The second problem I have is still the network. I use now routed vde/tap configuration which works fine for transferring some hundreds of megabytes from guest to host/net vice versa. But when I try to transfer some gigabytes using scp the connection hangs (wireshark shows no longer packets for the interface) and I have to restart the guest's interface. Perhaps this is also a timer/scheduling problem...or a bug in vde ...
- the time problem is more or less a guest kernel problem. My fedora guest system has now the same time as the host after setting "acpi=force" for the guest kernel (vanilla 2.6.22.1). The the guest's clock source is no longer pit. It's now acpi_pm according to /sys/devices/system/clocksource/clocksource0/current_clocksource. here the concerning thread from the kvm list: http://www.mail-archive.com/kvm-devel@lists.sourceforge.net/msg05440.html
Just bumped kvm-33.ebuild to kvm-36.ebuild, works as expected. Matthew: Why die if KVM kernel module is built in? Where is the option to use just the userspace utils? Why not just notice that modules are not going to build because detected built-in modules and continue building? kvm-33 did'nt work for me (can't boot any OS) so I use kvm-33 userspace and (gentoo-sources 2.6.22-r{4,5}) built-in modules without any problem. http://www.flickr.com/photos/tacvbo/1266924511/
(In reply to comment #135) > Just bumped kvm-33.ebuild to kvm-36.ebuild, works as expected. > > Matthew: > > Why die if KVM kernel module is built in? > Where is the option to use just the userspace utils? > Why not just notice that modules are not going to build because detected > built-in modules and continue building? > > kvm-33 did'nt work for me (can't boot any OS) so I use kvm-33 userspace and > (gentoo-sources 2.6.22-r{4,5}) built-in modules without any problem. > > http://www.flickr.com/photos/tacvbo/1266924511/ > I suppose you've noticed that the ebuild is built with compiling only the userspace utilities in mind if you comment out the die. The reason why the die is still there is similar to comment #129 at this point. I have just tested KVM-36 with 2.6.22 (uses KVM-22) and it does not work. Right now it does not even compile where at least it compiled before but gave me a segfault when I attempted to run the kvm userspace utility which is really just a patched qemu. I have consistently been using the kvm module (not built into the kernel) since kvm-14 and it has not had any problems booting. My hardware consists of a 64-bit Intel Core 2 Duo using kvm-intel (I don't have an AMD-V processor to test with). I have also tested kvm-intel on a 32-bit Intel Core Processor and I can confirm that works as well. If you're having problems using the module, you're either misconfiguring something or there is a bug in kvm with whatever hardware/os you're using.
(In reply to comment #136) > (In reply to comment #135) > > Why die if KVM kernel module is built in? > The reason why the die > is still there is similar to comment #129 at this point. I have just tested > KVM-36 with 2.6.22 (uses KVM-22) and it does not work. Right now it does not > even compile where at least it compiled before but gave me a segfault when I > attempted to run the kvm userspace utility which is really just a patched qemu. the reason why kvm-36 wouldn't compile is that there is a slight change in the ABI for that version which is not reflected in kernel 2.6.22 or even 2.6.23 (currently in RC) as explained here : http://sourceforge.net/mailarchive/message.php?msg_id=20070825172144.GA13727%40dreamland.darkstar.lan the problem is self inflicted though, because the ebuild uses incorrectly --with-patched-kernel to override the compilation process to use the system headers instead of the ones that match the user space tools
(In reply to comment #137) Sorry, so how can we get kvm-36 compiling and working? I'm getting the same segfault and don't have KVM build in to my kernel (2.6.22).
(In reply to comment #137) > (In reply to comment #136) > > (In reply to comment #135) > > > Why die if KVM kernel module is built in? > > The reason why the die > > is still there is similar to comment #129 at this point. I have just tested > > KVM-36 with 2.6.22 (uses KVM-22) and it does not work. Right now it does not > > even compile where at least it compiled before but gave me a segfault when I > > attempted to run the kvm userspace utility which is really just a patched qemu. > > the reason why kvm-36 wouldn't compile is that there is a slight change in the > ABI for that version which is not reflected in kernel 2.6.22 or even 2.6.23 > (currently in RC) as explained here : > > http://sourceforge.net/mailarchive/message.php?msg_id=20070825172144.GA13727%40dreamland.darkstar.lan > > the problem is self inflicted though, because the ebuild uses incorrectly > --with-patched-kernel to override the compilation process to use the system > headers instead of the ones that match the user space tools > The short answer is KVM is changing so quickly that you shouldn't count on the userspace tools from one version working with a previous version of a kvm module. So the die will remain there for the foreseeable future and it will not make use of --with-patched-kernel as a result. So as far as I can tell, the only time you could comment out the die would be if the kvm userspace tools were the exact same version as the version in the kernel.
(In reply to comment #139) > (In reply to comment #137) > > (In reply to comment #136) > > > (In reply to comment #135) > > > > Why die if KVM kernel module is built in? > > > The reason why the die > > > is still there is similar to comment #129 at this point. I have just tested > > > KVM-36 with 2.6.22 (uses KVM-22) and it does not work. Right now it does not > > > even compile where at least it compiled before but gave me a segfault when I > > > attempted to run the kvm userspace utility which is really just a patched qemu. > > > > the reason why kvm-36 wouldn't compile is that there is a slight change in the > > ABI for that version which is not reflected in kernel 2.6.22 or even 2.6.23 > > (currently in RC) as explained here : > > > > http://sourceforge.net/mailarchive/message.php?msg_id=20070825172144.GA13727%40dreamland.darkstar.lan > > > > the problem is self inflicted though, because the ebuild uses incorrectly > > --with-patched-kernel to override the compilation process to use the system > > headers instead of the ones that match the user space tools > > > > The short answer is KVM is changing so quickly that you shouldn't count on the > userspace tools from one version working with a previous version of a kvm > module. agree, even though there is a binary compatibility version (KVM_API_VERSION) which should prevent incompatible changes to break in these cases as shown in kvm.h both: kvm-36 user space tools and gentoo-sources-2.6.22-r5 use version 12 and therefore are able to work together. sadly, even in that case you are probably better of by using the kvm module that cames with the kvm package as there is going to be missing functionality otherwise (using kvm-36 with 2.6.22-r5's module won't allow for SMP guests for example) > So the die will remain there for the foreseeable future and it will > not make use of --with-patched-kernel as a result. So as far as I can tell, > the only time you could comment out the die would be if the kvm userspace > tools were the exact same version as the version in the kernel. the use of --with-patched-kernel is only valid when the kernel used is the kvm.git repository itself, so it shouldn't be there. the die is also not valid, as it enforces a recommendation which otherwise would work fine.
Created attachment 130577 [details] an ebuild for kvm-37 and all supporting patches in files to use untar in your local overlay and then run : ebuild kvm-37.ebuild manifest emerge -Duv kvm add "app-emulation/kvm kvm" to /etc/portage/package.use if you want it also to build the kernel module instead of using the one in the currently installed kernel (recommended in most cases) this has been tested stable in amd64 using gcc-3.4.6-r2 for the qemu part (referenced by the ebuild but not as a dependency as per ebuild policies, so that needs to be installed independently for it to work)
some additional comments about the attached kvm-37 ebuild (which has a different objective than the one being used here, as it is complimentary to qemu instead of conflicting with it, and is not meant to be considered "official" but hope at least useful to some of you) 1) all the qemu system emulations generated are now named kvm-* (kvm-system-i386 and kvm-system-x86_64) instead of qemu-* to prevent conflicts and let you use qemu as well for all cases where it is more stable/faster. 2) there is no qemu-img so you would need qemu-img from qemu-softmmu to manage disk images 3) /usr/bin/kvm doesn't work so use kvm-system-* directly after loading the right kernel modules, I might fix it as it is a nice wrapper, but for now i'd rather keep the dependency list lean 4) it requires gcc-3.4 to compile (at least the qemu part of it) and will use it directly by name so no need to change your system CC, alsa is optional as per the corresponding use flag 5) the generated kernel modules will be in /lib/modules/`uname -r`/misc so insmod those if you want to use them instead of the ones from your kernel (if there is only 1 copy of them modprobe would be probably easier to use and this is recommended), remember to insmod kvm.ko fist before your CPU specific module (kvm-intel.ko or kvm-amd.ko) and some about kvm-37 itself : 1) since kvm-36 there is a buggy BIOS included which will break windows 2000 when using ACPI and windows vista at least (partially reverted in git already) and that has broken SMP support, this is not yet fixed upstream and could be worked around by using -no-acpi or using qemu's BIOS files in some cases. 2) qemu requires gcc 3.4 (at least to compile the dyngen code that it uses for emulation), this is still not fixed upstream even if some improvements had been done recently to improve the quality of gcc-4 generated code, but is still going to mean a performance issue at the least and a source for crashes in the most common cases. 3) kvm code is IMHO more unstable than qemu and because of the forked status might be missing important patches like qemu-0.9.0-atapi-hsm.patch which is required for guests using libata, so falling back to qemu using kqemu as a speedy replacement if needed is IMHO a must have for any kvm ebuild, and at least until kvm finishes merging their modified qemu upstream (it would seem that kvm is actually forking more parts of qemu instead at least for now, so this could take a while)
I'm also all in for a coexistence of both qemu and kvm. Please consider using this method in future ebuilds... Thanks...
(In reply to comment #142) > some additional comments about the attached kvm-37 ebuild (which has a > different objective than the one being used here, as it is complimentary to > qemu instead of conflicting with it, and is not meant to be considered > "official" but hope at least useful to some of you) > > 1) all the qemu system emulations generated are now named kvm-* > (kvm-system-i386 and kvm-system-x86_64) instead of qemu-* to prevent conflicts > and let you use qemu as well for all cases where it is more stable/faster. > 2) there is no qemu-img so you would need qemu-img from qemu-softmmu to manage > disk images > 3) /usr/bin/kvm doesn't work so use kvm-system-* directly after loading the > right kernel modules, I might fix it as it is a nice wrapper, but for now i'd > rather keep the dependency list lean > 4) it requires gcc-3.4 to compile (at least the qemu part of it) and will use > it directly by name so no need to change your system CC, alsa is optional as > per the corresponding use flag > 5) the generated kernel modules will be in /lib/modules/`uname -r`/misc so > insmod those if you want to use them instead of the ones from your kernel (if > there is only 1 copy of them modprobe would be probably easier to use and this > is recommended), remember to insmod kvm.ko fist before your CPU specific module > (kvm-intel.ko or kvm-amd.ko) > > and some about kvm-37 itself : > > 1) since kvm-36 there is a buggy BIOS included which will break windows 2000 > when using ACPI and windows vista at least (partially reverted in git already) > and that has broken SMP support, this is not yet fixed upstream and could be > worked around by using -no-acpi or using qemu's BIOS files in some cases. > 2) qemu requires gcc 3.4 (at least to compile the dyngen code that it uses for > emulation), this is still not fixed upstream even if some improvements had been > done recently to improve the quality of gcc-4 generated code, but is still > going to mean a performance issue at the least and a source for crashes in the > most common cases. > 3) kvm code is IMHO more unstable than qemu and because of the forked status > might be missing important patches like qemu-0.9.0-atapi-hsm.patch which is > required for guests using libata, so falling back to qemu using kqemu as a > speedy replacement if needed is IMHO a must have for any kvm ebuild, and at > least until kvm finishes merging their modified qemu upstream (it would seem > that kvm is actually forking more parts of qemu instead at least for now, so > this could take a while) > Apparently you haven't read all the comments to this bug. No one is interested in using gcc-3.4.6 to compile kvm. By removing qemu-img from the standard kvm build, you just created a dependency for gcc-3.4.6 because most people are going to need qemu-img. Due to the differences of the qemu bundled with kvm, I wouldn't count on any part of the upstream qemu working with kvm because you don't necessarily know what they changed and it's too much of a PITA to track every little change and update a bunch of patches as evidenced above. This is also the case with changing the name of the binary. So kvm should continue to block qemu until either kvm permanently forks from qemu (doubt it) or the kvm changes get merged into upstream and in this case, the ebuild will change from installing a module and userspace tools to just a module.
There is no technical reason why they need to block each other. Your other comments are fair enough but blocking each other is unnecessary if you just rename the binaries etc. Ubuntu takes this approach and it works fine.
Can anyone help with this error? CC [M] /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/lapic.o CC [M] /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.o /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c: In function 'ioapic_deliver': /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c:204: error: 'dest_LowestPrio' undeclared (first use in this function) /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c:204: error: (Each undeclared identifier is reported only once /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c:204: error: for each function it appears in.) /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c:215: error: 'dest_Fixed' undeclared (first use in this function) make[2]: *** [/var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.o] Error 1 make[1]: *** [_module_/var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel] Error 2 make[1]: Leaving directory `/usr/src/linux-2.6.21-gentoo-r4' make: *** [all] Error 2 !!! ERROR: app-emulation/kvm-37 failed. Call stack: ebuild.sh, line 1638: Called dyn_compile ebuild.sh, line 985: Called qa_call 'src_compile' ebuild.sh, line 44: Called src_compile kvm-37.ebuild, line 72: Called linux-mod_src_compile linux-mod.eclass, line 516: Called die !!! Unable to make all. !!! If you need support, post the topmost build error, and the call stack if relevant. !!! A complete build log is located at '/var/tmp/portage/app-emulation/kvm-37/temp/build.log'. --------------------------- ACCESS VIOLATION SUMMARY --------------------------- LOG FILE = "/var/log/sandbox/sandbox-app-emulation_-_kvm-37-27889.log" open_wr: /usr/src/linux-2.6.21-gentoo-r4/-.gcda -------------------------------------------------------------------------------- Thanks
(In reply to comment #146) > Can anyone help with this error? > > CC [M] /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/lapic.o > CC [M] /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.o > /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c: In function > 'ioapic_deliver': > /var/tmp/portage/app-emulation/kvm-37/work/kvm-37/kernel/ioapic.c:204: error: > 'dest_LowestPrio' undeclared (first use in this function) your kernel (if using x86 at least) should be built with CONFIG_X86_IO_APIC=yes and CONFIG_SMP=yes or apply the following patch (included in kvm-41) http://git.kernel.org/?p=linux/kernel/git/avi/kvm.git;a=commitdiff;h=1196dd4e2e6f09053335b0a91e3cc793808c00a7
Created attachment 131298 [details] an ebuild for kvm-41 (evolved from the unofficial kvm-37) to install and download all related files (including this ebuild) you can use layman : layman -f -o http://tapir.sajinet.com.pe/gentoo/layman.xml -a sajinet emerge -DNvu kvm if you don't have layman or would rather get the files the old fashioned way they are all available from : http://www.sajinet.com.pe/gentoo/portage/app-emulation/kvm/files/ you'll need to have gcc-3 installed to build the qemu part of it, but if you don't really want to emerge =sys-devel/gcc-3.4.6-r2 (there is no need to change the system gcc) there are also binary packages available from : http://www.sajinet.com.pe/gentoo/packages/
(In reply to comment #148) > Created an attachment (id=131298) [edit] and some comments on the ebuild : 1) the qemu system binary has been renamed as kvm-system-x86_64 so that it won't conflict with qemu's; qemu-img will be installed as part of this package or not depending on the USE='qemu' flag and will block or depend on qemu-softmmu as required. 2) the /usr/bin/kvm wrapper has been fixed so you can use it to load your modules and start your VM in some simple cases but it was never meant to be used out of a developer environment and therefore has very rough edges. use the kvm-system-x86_64 directly better and ensure your modules and other system emulation requirements (like bridges) are covered through gentoo native tools as suggested by upstream (who might even remove it in future versions) 3) the USE='kvm' flag lets you use the external module (recommended) or the one provided with your own kernel (it will only work for kernel 2.6.22 or higher) and some comments in the kvm-41 release : 1) most of the problems with the BIOS has been fixed, so SMP should work better now, but if you get problems use -no-acpi or -no-kvm-irqchip to disable ACPI or APIC kvm emulated implementations 2) problems with the APIC could result in lost interrupts or crashes, so consider -no-kvm-irqchip your friend until the APIC issues get sorted out; for linux guests the use of the "nolapic" boot parameter can help as well. 3) expect frequent releases and updates (which is why i recommended to use the overlay so you can be also in the bleeding edge), if all you want is stability kvm-36 is probably your best bet as well as keeping away from SMP guests.
paludis complains over the kvm-41 ebuild, that is has no whitespace after '(' and before ')' in the DEPEND String.
Created attachment 132381 [details, diff] kvm-45-qemu-configure.patch new configure patch
Created attachment 132382 [details, diff] kvm-45-qemu-no-img.patch updated qemu-no-img patch
I just posted 2 new patches for kvm-45 which synced with qemu cvs. The ide patch is not needed anymore as it was already fixed in the cvs snapshot.
(In reply to comment #153) > I just posted 2 new patches for kvm-45 which synced with qemu cvs. The ide > patch is not needed anymore as it was already fixed in the cvs snapshot. Which ebuild is this for? It isn't clear...
(In reply to comment #154) > Which ebuild is this for? > It isn't clear... No ebuild as this does not require major changing. The 2 patches replace the old ones. kvm-36-qemu-split.patch -> kvm-45-qemu-configure.patch kvm-36-qemu-no-img.patch -> kvm-45-qemu-no-img.patch
Sorry, I still don't understand. There are (at least) 2 sets of ebuilds in this bug, from Carlos and Matthew. I can't see kvm-36-* in this bug at all.
(In reply to comment #156) > Sorry, I still don't understand. There are (at least) 2 sets of ebuilds in > this bug, from Carlos and Matthew. I can't see kvm-36-* in this bug at all. Use either one I don't care and it also does not matter. If you had bothered to look at the actual patches you would know what they are for and which they replace. Anyway, I got the naming from the kvm-41 ebuild Carlos posted (Which you obviously have not looked at either). All the <whatever version>-no-img and <whatever version>-doc/-configure/-split patches dot not work on kvm-45 for the reason I already explained. so: <whatever version>-no-img.patch -> kvm-45-qemu-no-img.patch <whatever version>-doc/-configure/-split.patch -> kvm-45-qemu-configure.patch There is absolutely no sane reason for posting yet another complete new ebuild just because 2 patches needed updating and 1 had to be removed.
Created attachment 132422 [details] kvm-45.ebuild I'm taking an even more minimalist approach here by patching less and not renaming anything. This will block qemu but then again, if you're using kvm, why would you want to use qemu? The only patching that is involved now is to remove the ability to disable kvm (kvm-kvm.patch). I'm not sure why anyone would want to install kvm and not use it. It won't work when it's disabled anyway. The scripts-qemu-ifup.patch is something I wrote a while back to fix their broken interface script. You need to have a bridge interface set up for this. I'm providing and updated ebuild without the former patches for you lazy ones who don't want to edit files. ;-)
(In reply to comment #158) > I'm taking an even more minimalist approach here by patching less and not > renaming anything. This will block qemu but then again, if you're using kvm, > why would you want to use qemu? qemu has more options and features than kvm uses/enables. Plus we do not know if the kvm changes to qemu will break qemu in some weird and wonderfull way ;-) > I'm not sure why anyone would want to install kvm and not use it. It won't > work when it's disabled Because some will only want the userland tools as they use the in kernel version of kvm.
i'm trieing to get both of your approaches merged in one ebuild only, just adding some USE flags. It's rude for now, but it's usefull to know if this is an acceptable way to install kvm. Please let me know what you think about it :-)
Created attachment 132486 [details] kvm-45.ebuild with both approaches in
(In reply to comment #153) > I just posted 2 new patches for kvm-45 which synced with qemu cvs. The ide > patch is not needed anymore as it was already fixed in the cvs snapshot. > thanks Sanders, uploaded both to the overlay (had to change 1 line in kvm-45-qemu-no-img.patch though as there was a missing closing parenthesis that was making the tarbin target). the ide-cd patch was dropped in kvm-44 (an ebuild available in the overlay) when the patch was backported from qemu CVS and re-imported as shown by : http://git.kernel.org/?p=virt/kvm/kvm-userspace.git;a=commitdiff;h=667f2deff8cea6327f0c832afea2f4952b8f199e Carlo (without "S")
Created attachment 132866 [details] ebuild for kvm-45 with qemu compatibility tested in ~amd64 and ~x86 and adding the following changes : * the qemu provided binary is now named kvm (regardless of architecture) to prevent conflicts with qemu and to follow debian's naming and avoid unneeded confusion (plus help retiring the wrapper). * the kvm wrapper and kvm_stat script is now installed in /etc/kvm/utils (as done in debian) and eventhough kvm was fixed to work its use is discouraged (upstream never intended for it to be used in production) * avoid conflicts with kernel-headers (reported in comment #117) * fixes to /etc/kvm/qemu-ifup and /etc/kvm/utils/kvm to allow for setting up a bridge (reported in comment #118 to comment #120 and other) * add spaces to DEPEND to avoid paludis problem (reported in comment #150) * using rebased patches for kvm-45 from Sander (reported in comment #153)
(In reply to comment #160) > i'm trieing to get both of your approaches merged in one ebuild only, just > adding some USE flags. the compatible with qemu ebuild was based on the one from Matthew but just trying to wrap some of the alternatives with USE flags as you suggested, with the only one left to be support for gcc4 (which was just added to kvm-45-r1. > It's rude for now, but it's usefull to know if this is an acceptable way to > install kvm. from what I remember on gentoo ebuild policy, you can't depend on a gcc version (no idea why though since it definitely makes more sense that way) but the only options is "autodetect" (which is frowned upon as you can see in qemu's ebuild) or abort and force you to change your system CC (which is IMHO even worst and very difficult to do for kvm)
Created attachment 132899 [details] kvm-45-r1.ebuild the following changes : * added USE="gcc4" to use gcc4 when gcc3 can't be installed (use of gcc3 is recommended) * general cleanup as usual, the ebuild and all dependent patches available from the overlay
(In reply to comment #164) > > from what I remember on gentoo ebuild policy, you can't depend on a gcc version > (no idea why though since it definitely makes more sense that way) but the only > options is "autodetect" (which is frowned upon as you can see in qemu's ebuild) > or abort and force you to change your system CC (which is IMHO even worst and > very difficult to do for kvm) > Sorry, i didnt know about that policy restriction. Anyway, i think your last ebuild is the best approach for giving choices to final users about all of the aspects of the installation. I've just added your overlay to my favourite ones list ;-)
(In reply to comment #166) > Sorry, i didnt know about that policy restriction. That's not a policy restriction; that's a pure and simple fact that depending on some gcc version will NOT make sure that you are actually using it. The only 'policy' thing here is that you won't switch gcc version on unsuspecting users in ebuilds via gcc-config or whatnot...
Ok, i've got the point :-) But in this specific case we only need to be sure that gcc3 is emerged, we don't switch to it in a "subtle way" on unsuspecting users... Anyway, it may be even better to force this ebuild to compile with gcc4, while continuing to preserve the compatibility with the qemu ebuild... in this way we can have both installed at the same time (tbh, i don't understand why to use this ebuild without really using kvm... qemu is a better choice for that)
Created attachment 133157 [details, diff] updated kvm (python wrapper) patch for kvm 46 the default disk size is now defined to be 30GB instead of 10GB upstream so old patch will conflict. use of kvm (python wrapper) is discouraged though
Created attachment 133510 [details] ebuild for kvm 46 with the following changes : * added USE="gnutls" and corresponding dependency if qemu is to be compiled with support for TLS enabled VNC connections (as supported by the qemu version included since kvm-45) * removed dependency to >=sys-fs/e2fsprogs-1.39 which was obsoleted since >=app-emulation/kvm-29 * added build dependency for =sys-devel/gcc-3.4.6 which is stable in both amd64 and x86 if USE!="gcc4" and that is the recommended and supported compiler for qemu (suggested in comment #161 and confirmed valid in comment #167). ebuild with autodetect it and use it once it is available to build qemu.
Created attachment 133515 [details] ebuild for kvm 46 replaced the original ebuild uploaded which should had add a dependency to gcc-3.4.6-r2 instead of gcc-3.4.6 and which was used to test it for "amd64" and "x86". all supporting patches and this ebuild available as usual from the overlay
I don't understand how this ebuild works any more. Should we use the kernel drivers or the ebuild's one? Why it does need gcc-3 whereas a previous version compiles with gcc-4? Please help me ;(
imho, using the ebuild's kernel modules is a better choice because you'll be sure that you'll have no compatibility problems between kernel modules and userspace utilities. Compiling with gcc3 gives you the ability to run kvm without kvm kernel modules; gcc4 compiled kvm can only work with kernel modules (imho, this is the more logical choice). Anyway, just set use flags for your needs. Thanks again to Carlo Marcelo Arenas Belon for this ebuild, keep this good work up! :-)
Created attachment 133859 [details] ebuild for kvm 48 tested in ~amd64 and ~x86. important changes : * removed use alsa and gnutls patches (merged upstream) * add USE="sdl" to enable SDL support (used to be there by default) all patches and this ebuild (as well as one for kvm-47) available from the overlay. a description of all USE flags follows (recommended flags marked with +) : - - alsa : Adds support for media-libs/alsa-lib (Advanced Linux Sound Architecture) - - gcc4 : Add support for gcc 4 - - gnutls : Adds support for net-libs/gnutls + + kvm : Enables the kernel module + + qemu : Use QEMU's qemu-img + + sdl : Adds support for Simple Direct Layer (media library)
(In reply to comment #172) > I don't understand how this ebuild works any more. Hopefully the following documentation could help you in this case : http://gentoo-wiki.com/HOWTO_KVM > Should we use the kernel drivers or the ebuild's one? > Why it does need gcc-3 whereas a previous version compiles with gcc-4? It depends of what your environment is like, but if you are using gentoo stable (amd64 or x86) on your workstation with a gentoo-sources kernel then what I'd found that fits me better is: "kvm qemu sdl"
I'm building kvm-53 with a renamed kvm-48 ebuild but kvm-45-user-no-kernel.patch doesn't apply any more, so I get this collision: /usr/include/linux/kvm.h
(In reply to comment #176) > I'm building kvm-53 with a renamed kvm-48 ebuild there are actually tested and complete ebuilds for kvm up to 53 in the overlay, but I haven't been posting them here as there had been quite a number of changes in the codebase and the ebuilds themselves as patches had been adjusting and being upstreamed.
(In reply to comment #174) > Created an attachment (id=133859) [edit] > ebuild for kvm 48 > > tested in ~amd64 and ~x86. > > important changes : > > * removed use alsa and gnutls patches (merged upstream) > * add USE="sdl" to enable SDL support (used to be there by default) > > all patches and this ebuild (as well as one for kvm-47) available from the > overlay. > > a description of all USE flags follows (recommended flags marked with +) : > > - - alsa : Adds support for media-libs/alsa-lib (Advanced Linux Sound > Architecture) > - - gcc4 : Add support for gcc 4 > - - gnutls : Adds support for net-libs/gnutls > + + kvm : Enables the kernel module > + + qemu : Use QEMU's qemu-img > + + sdl : Adds support for Simple Direct Layer (media library) > With !gcc4 it depends on "=sys-devel/gcc-3.4.6-r2". Shouldn't it instead depend on "<sys-devel/gcc-4"?
(In reply to comment #178) > (In reply to comment #174) > > - - gcc4 : Add support for gcc 4 > > With !gcc4 it depends on "=sys-devel/gcc-3.4.6-r2". yes, because that compiler has been validated to work in both x86 and amd64 with the imported qemu. > Shouldn't it instead depend on "<sys-devel/gcc-4"? ideally, but qemu doesn't work with all compilers that match that criteria and since the ebuild isn't changing the default gcc you are using but using one directly by name, it needs to know what name to look for specifically, hence why it is just easier to depend on 1 specific version of gcc which is known to work in all cases and that we know the name for which can be hardcoded in the ebuild.
Created attachment 137185 [details] ebuild for kvm-54 as usual the ebuild and all supporting patches available from the overlay which also now includes a Changelog starting from kvm-50 and which hopefully would help tracking and documenting all required patches as they are being upstreamed
(In reply to comment #180) > as usual the ebuild and all supporting patches available from the overlay Could you please be so kind to point to it's URL / where to get it from ? Thank you!
(In reply to comment #181) > (In reply to comment #180) > > > as usual the ebuild and all supporting patches available from the overlay > > Could you please be so kind to point to it's URL / where to get it from ? sorry, I thought not to be repeating myself and get everyone bored everytime with the same explanation which was originally posted in comment 148. in any case, if you want some more documentation about the ebuild head to : http://gentoo-wiki.com/HOWTO_KVM if you want to get the ebuild and patches using http go to : http://www.sajinet.com.pe/gentoo/portage/app-emulation/kvm/ I would recommend though installing the overlay and syncing it periodically, as KVM releases are frequent (indeed there were 2 more just yesterdays with important fixes if running OpenSolaris or FreeBSD guests) Carlo
Thank you so much! Sorry; I had searched all these pages for "overlay" but not for "layman" too.
Created attachment 137723 [details] ebuild for kvm 55 ebuild for kvm-55 : * removes unused BLOBs and adds USE="bios" for rebuilding your BIOS (not recommended) as for kvm-55 : * lots of fixes to regressions in different guest OS and stabilization work from the new BIOS and in kernel APIC and IMHO one of the most stable releases since kvm-36
I've noticed that modules are being built for the currently running kernel, rather than the kernel pointed to by /usr/src/linux. I think that something like "--kerneldir=$KV_DIR" should be added to conf_opts. (Thanks for the overlay, BTW!)
(In reply to comment #185) > I've noticed that modules are being built for the currently running kernel, > rather than the kernel pointed to by /usr/src/linux. > > I think that something like "--kerneldir=$KV_DIR" should be added to conf_opts. yes, tested kvm-55-r2.ebuild with a fix and uploaded it to the overlay. > (Thanks for the overlay, BTW!) thanks for helping on making it better for everyone.
Created attachment 137946 [details] ebuild for kvm 55 updated ebuild for kvm-55 : * adds a configuration flag to report where is the kernel located as reported by gentoo's kernel eclass so it uses the eselected kernel instead of trying to guess for a valid one (reported in comment 185)
A few comments on your ebuild: - I see that there are dozens of patches. Have those been sent to the kvm developers? - you can check if you are using gcc4 with the following command (taken from qemu ebuild): if [ "$(gcc-major-version)" == "4" ]; then ... This makes the gcc4 useflag useless. - does econf not work? If not then document in the ebuild that it's not working. - take a look at use_enable. this avoids constructs like "if ! use gnutls; then conf_opts="$conf_opts --disable-vnc-tls"; fi" - *maybe* it's better to use 'module' instead of 'kvm' to build the kernel modules? It might be easier to understand for our users. - I think that ebuilds shouldn't set the compiler (i.e. do not set a vanilla compiler when the compiler is hardened). Just error out and let the user know why it's failing. - the install path for ${S}/kvm and ${S}/kvm_stat looks strange (/etc/kvm/utils/). Should these belong there? Thanks for the work you have put into this ebuild.
(In reply to comment #188) > - I see that there are dozens of patches. Have those been sent to the kvm > developers? yes > - you can check if you are using gcc4 with the following command (taken from > qemu ebuild): > > if [ "$(gcc-major-version)" == "4" ]; then ... > > This makes the gcc4 useflag useless. no, the gcc4 useflag is to tell the ebuild that you really want to use gcc4 to build kvm/qemu and so it does in a way that is consistent with what upstream will expect of qemu when compiled with gcc4 and gives you a more or less stable environment. I personally think that trying to force the use of gcc4 has no real value as gentoo gives you the flexibility to have gcc-3.x installed in a very simple way and like no other distribution (ask anyone using SuSE), but is there for anyone that would want that and understands the consequences. > - does econf not work? If not then document in the ebuild that it's not > working. kvm and qemu's configure are not generated from autotools, so econf is not the right way to manage them and adds no value. > - take a look at use_enable. this avoids constructs like "if ! use gnutls; then > conf_opts="$conf_opts --disable-vnc-tls"; fi" use_enable won't work because configure doesn't have both --disable and --enable flags for each option as it doesn't come from autoconf. qemu's ebuild will break in the next release because of that if that is what you are looking as a reference. > - *maybe* it's better to use 'module' instead of 'kvm' to build the kernel > modules? It might be easier to understand for our users. when trying to find a suitable USE flag, I looked at the full list from the ones available in the global use.desc and the ones from use.local.desc (including several overlays) and couldn't find one that was clearly defined to mean "kernel module" so I assumed that using "kvm" would be OK and descriptive enough as kqemu uses "kqemu". if you have any better option, please provide some references with your argument to why that would be better, but to help with the current use flags defined I provide a use.local.desc with the overlay which will be of use if it weren't for bug 138622 which has a patch for equery but no one looking at it from gentoo's development team. > - I think that ebuilds shouldn't set the compiler (i.e. do not set a vanilla > compiler when the compiler is hardened). Just error out and let the user know > why it's failing. the hardened correction is yet incomplete as it doesn't cover all angles and requires more testing but is useful as it is now (validated in amd64) and is technically the same logic as used by --filter-flags when some compiler options are not supported, but including the changes that are required in the compiler setup; with an ebuild version of the proposal in bug 166392 (for a more complete setup look at the qemu ebuilds provided in the overlay) it is important to note that since the compiler is being autodetected in most cases, it is not possible (or simple) for the user to change that setup without having to eselect a different system kernel to begin with, and therefore that is better avoided as it is fragile and prone to human error. > - the install path for ${S}/kvm and ${S}/kvm_stat looks strange > (/etc/kvm/utils/). Should these belong there? /etc/kvm/kvm is an script that is used upstream for testing and was never meant to be used in production or even get installed but sadly the first ebuild that was published here used it, so it was kept there for backward compatibility and following the filesystem layout that debian uses (they made the same mistake). kvm_stat is a helper script which should be instead on /usr/bin or be thrown away (which is basically what going temporarily to /etc/kvm/utils/ unless someone using it complained was meant to do). since this probably qualifies as a complain will move them around, but I really suspect that no one is really using them
Created attachment 139324 [details] ebuild for kvm 57 includes : * backward compatibility fixes for old kernels and non-smp configurations * some logic to try to prevent failed builds if using a hardened compiler (not supported upstream) * overall ebuild cleanup beware that kvm-57 includes a new version of qemu and is more likely to have problems than previous versions because of that, so if you don't want to risk your virtual machines to the bleeding edge use kvm-56-r1 instead. kvm-58 will most likely include initial support for ia64, so if you have some of that hardware and are willing to do some testing for the greater good this will be your once on a life opportunity to do so
Created attachment 139488 [details] ebuild for kvm 58 includes: * initial support for USE="test" * masked keyword for ia64 (not tested yet)
(In reply to comment #191) > Created an attachment (id=139488) [edit] > ebuild for kvm 58 > > includes: > > * initial support for USE="test" > * masked keyword for ia64 (not tested yet) > Could you please attach the patches this ebuild requires? Thanks :)
Created attachment 140913 [details] ebuild for kvm 59 Important changes : * Moved kvm_stat to /usr/bin, and discarded the python kvm wrapper that used to be in /etc/kvm/utils/kvm * A more complete workaround for hardened toolchain
(In reply to comment #192) > > Could you please attach the patches this ebuild requires? Thanks :) The last ebuild has 18 patches, and there had been some ebuilds with up to 23 of them. The patch list changes frequently and gets adapted as they are slowly upstreamed and new releases require new ones or rebased patches in some cases, so uploading them all in this ticket and keeping the ticket updated in each release will require too many ticket updates with will spam the long list that is CC here. KVM does releases frequently and if you look at the Changelog you are going to see that the ebuild and patches could change daily as well (I indeed don't upload all ebuilds but only the important ones, and sometimes forget to update the ChangeLog when doing updates that don't require a version bump), which is why I suggested using the overlay instead which will usually have the day of the upstream release a masked ebuild and will have its patch list adapted and stabilized from there as the new package is tested. If you want to get the files through http you can also use the http interface for the overlay with the files located in : http://tapir.sajinet.com.pe/gentoo/portage/app-emulation/kvm/files/ And if you feel so strongly about having them attached into this ticket then feel free to upload them as well
any news about the inclusion into the portage tree?
At my site it hangs the whole machine on booting :( (tested w/ win2k and plan9)
Well, this stuff seems to work very well for me; i've just installed app-emulation/kvm-68 USE="alsa gcc4 ncurses sdl -bios -gnutls -kvm -qemu -test" and was able to setup win-xp on top of linux with it. Thanks a lot Carlo! Portage 2.1.4.4 (default-linux/amd64/2007.0/desktop, gcc-4.1.2, glibc-2.6.1-r0, 2.6.24-gentoo-r7 x86_64) ================================================================= System uname: 2.6.24-gentoo-r7 x86_64 Intel(R) Core(TM)2 Duo CPU E6550 @ 2.33GHz Timestamp of tree: Tue, 13 May 2008 08:35:01 +0000 app-shells/bash: 3.2_p17-r1 dev-java/java-config: 1.3.7, 2.1.4 dev-lang/python: 2.4.4-r9 dev-python/pycrypto: 2.0.1-r6 sys-apps/baselayout: 1.12.11.1 sys-apps/sandbox: 1.2.18.1-r2 sys-devel/autoconf: 2.13, 2.61-r1 sys-devel/automake: 1.4_p6, 1.5, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2, 1.10.1 sys-devel/binutils: 2.18-r1 sys-devel/gcc-config: 1.4.0-r4 sys-devel/libtool: 1.5.26 virtual/os-headers: 2.6.23-r3 ACCEPT_KEYWORDS="amd64" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-march=nocona -O2 -pipe" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config /usr/kde/3.5/shutdown /usr/share/config /var/lib/hsqldb" CONFIG_PROTECT_MASK="/etc/env.d /etc/env.d/java/ /etc/fonts/fonts.conf /etc/gconf /etc/revdep-rebuild /etc/terminfo /etc/texmf/web2c /etc/udev/rules.d" CXXFLAGS="-march=nocona -O2 -pipe" DISTDIR="/var/portage/distfiles" FEATURES="distlocks metadata-transfer parallel-fetch sandbox sfperms strict unmerge-orphans userfetch" GENTOO_MIRRORS="ftp://gd.tuwien.ac.at/opsys/linux/gentoo/ http://mirror.uni-c.dk/pub/gentoo/ http://trumpetti.atm.tut.fi/gentoo/ " LANG="en_US.utf8" LC_ALL="en_US.utf8" LINGUAS="en de" MAKEOPTS="-j3" PKGDIR="/var/portage/packages" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress --force --whole-file --delete --stats --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/var/tmp" PORTDIR="/var/portage/repos/gentoo" PORTDIR_OVERLAY="/var/portage/repos/layman/sajinet /var/portage/repos/private" SYNC="rsync://192.168.0.1/gentoo-portage" USE="3dnow 3dnowext X a52 aac acpi alsa amd64 berkdb bzip2 cairo caps cddb cdparanoia cdr cli cracklib crypt cups dbus djvu dri dvd dvdr dvdread eds emboss encode evo exif fam ffmpeg firefox flac fortran gd gdbm gif gimp gnome gnome-keyring gphoto2 gpm gstreamer gtk hal hddtemp iconv icu ipod ipv6 isdnlog java jpeg jpeg2k lcms ldap libnotify lm_sensors mad matroska midi mikmod mmap mmx mmxext mono mp3 mpeg mudflap musicbrainz ncurses nls nptl nptlonly nvidia ogg opengl openmp pam pcre pdf perl plotutils png pppd pulseaudio python qt3support quicktime readline reflection ruby sdl session spell spl sse sse2 ssl ssse3 svg tcpd tetex theora threads tiff tracker truetype unicode usb vcd vim-syntax vorbis xattr xine xml xorg xv xvid zlib" ALSA_PCM_PLUGINS="adpcm alaw asym copy dmix dshare dsnoop empty extplug file hooks iec958 ioplug ladspa lfloat linear meter mulaw multi null plug rate route share shm softvol" APACHE2_MODULES="actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" CAMERAS="canon konica ptp2 kodak" ELIBC="glibc" INPUT_DEVICES="keyboard mouse evdev" KERNEL="linux" LINGUAS="en de" USERLAND="GNU" VIDEO_CARDS="nvidia nv" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LDFLAGS, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS
Hi, with the release of kvm-69 it's better to have CONFIG_PREEMPT_NOTIFIERS=y in kernel-config. This is only possible if you compile the kernel-internal-kvm as a module or buildin, but if you want to use the external modules the ebuild won't let you. dirty fix: --- kvm-69.ebuild 2008-05-24 13:19:16.000000000 +0200 +++ kvm-69.ebuild.orig 2008-05-23 04:27:05.000000000 +0200 @@ -74,7 +74,7 @@ if use kvm; then eerror "kvm module can't be used if your kernel is already" eerror "kvm enabled" - #die "incompatible module configuration, reconfig or USE='-kvm'" + die "incompatible module configuration, reconfig or USE='-kvm'" else ewarn "using the kernel provided module" ewarn "some features may only be available in newer releases"
(In reply to comment #198) > Hi, with the release of kvm-69 it's better to have CONFIG_PREEMPT_NOTIFIERS=y > in kernel-config. This is only possible if you compile the kernel-internal-kvm > as a module or buildin, that is a bug in the kernel because as you said prevent you to have PREEMPT_NOTIFIERS enabled if using the external module. the easiest workaround some distributions have adopted is to patch kvm into the kernel (instead of using the external module) but that is tricky to do with an unoficial package ;) > but if you want to use the external modules the ebuild > won't let you. it is a protection mechanism to avoid doing exactly that, which will otherwise result in conflicts resolving all needed symbols in the kernel for KVM. if you want to use the KVM that comes with your kernel then USE="-kvm" so you will only compile the userspace application. if you want to use the last module then you need to disable KVM in the kernel configuration and USE="kvm". > dirty fix: this is just overriding the protection, and if you compiled your kernel with KVM="y" will leave you with an extra module that can't be used, while if you use KVM="m" then will leave you with 2 sets of kvm modules (in different locations) and then force you to insmod the right one by hand (modprobe could get confused) as there are duplicated symbols that could lead into a broken kernel if the wrong mix is loaded.
In my opinion, i think you should fix the ebuild, because its not dangerous to have 2 sets of kvm, because you can't mix the modules (yes i am loading via insmod)
KVM is in the tree now.
(In reply to comment #201) > KVM is in the tree now. For everyone that was using the unofficial overlay packages and documentation. Stop. I had updated the wiki page in gentoo-wiki to reflect the official package instead, but I assume it won't be needed in the long run as well as the overlay since it might seem that now Gentoo got finally a "Virtualization Team" (IMHO: about time, since even Debian has one already that even got kvm backported to stable) Thanks for all great feedback and for making this such a fun ride and for helping stabilize kvm for the greater good. Be careful with KVM-70 as that is the first ever KVM release that has no more a dependency in the old gcc and is therefore rough in the edges (which is why I never felt like unleashing since its Jun 18 release and it will never be) but be also confident that whatever problems you might find will be promptly resolved. So long and thank for all the fish
Where I can find /etc/init.d/kvmd script ?
Nowhere, just add kvm-intel or kvm-amd to /etc/modules.autoload.d/kernel-2.6 if you build them as modules and run your virtual machines with /usr/bin/kvm.
