- Devel - libvirt List Archives

[Libvir] Remote & the problem with qemu deadlock
by Richard W.M. Jones 19 Apr '07

19 Apr '07

As I wrote here: https://www.redhat.com/archives/libvir-list/2007-April/msg00114.html remote access to qemu:/// URLs currently deadlocks. A bit of explanation follows as to why this happens in the current architecture of the remote patch. If you apply the remote patch right now, you'll get a modified libvirt_qemud server which can handle both qemu and remote requests over the same socket. (Basically both qemu_internal.c and remote_internal.c connect to the same Unix domain socket, then depending on the "program number" encoded within the RPC messages, they get dispatched accordingly inside libvirt_qemud). The server is written to handle multiple connections at once using non-blocking poll[1], but once the server has assembled a whole incoming message, it then blocks while dealing with that message. The problem occurs in the qemu case when the remote server issues a call into qemu_internal (now linked into the server), and qemu_internal then tries to connect out to the qemu daemon. Unfortunately the qemu daemon /is/ the remote server, which is blocked handling the current call. Thus it is unable to accept the new connection (from itself) and completely deadlocks. One solution suggested was to have qemu_internal recognise when it is linked into the server and make local calls back to its counterpart in the server. However I think this is pretty ugly and unnecessarily complicates qemu_internal which shouldn't care how it is linked. My thoughts on the long term solution ------------------------------------- I'd really like to see the qemu case not require a daemon. At the moment (AFAICS) the qemu daemon serves two purposes: (a) it keeps track of the monitor file descriptors of the existing qemu processes, and (b) it handles the networking stuff, basically starting and stopping dnsmasq[2]. (a) is easily handled, I think, by having qemu processes put their monitor sockets into a well-known directory. This has the other advantage that qemu processes can survive and continue after qemu programs restart. For the (b) case, I don't think networking which is a general service should be mixed up with qemu code. I think (although probably I've missed some things) that this means that qemu_internal could be implemented entirely as a local service, requiring no daemon. (In the case where qemu_internal needs to control a root-owned system-wide set of qemu processes then we'd still have to have a daemon, but it can be the remote daemon, same way that Xen will work). My suggestion for a short term solution --------------------------------------- Since doing the above is way more than I'm going to do just to get remote working, my suggestion is that we have two daemons running, qemud and remote. Both would be built from the exact same codebase: the mode would be selected by having a 'libvirtd --remote' flag on the command line (or however, but the important thing is that they'd keep the same codebase). This immediately fixes the deadlock problem[3] because qemu_internal linked to the server is now talking to a different daemon. Rich. [1] As an aside, I don't particularly like this architecture. Linus has handed down to us an efficient operating system which is already capable of multiplexing different clients using a marvellous system called "multitasking". Hand-coding polling code inside a single process is really only necessary in very limited circumstances, such as highly tuned uniprocessor machines with enormous loads, and we are not likely to get into such a situation with libvirtd. If we ever did we'd need to benchmark a range of possible solutions. Processes on the other hand have real advantages, such as security through isolation and controlled sharing, and easy scalability onto SMP machines. [2] Although I don't really understand the networking stuff too well, so perhaps it does other things. [3] Except in the "remote-on-remote" case, but we should disallow this explicitly because it seems like it would be nothing but trouble. -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903 Directors: Michael Cunningham (USA), Charlie Peters (USA) and David Owens (Ireland)

2 4

[Libvir] Community IRC chat room for virt management tools/APIs/protocols
by Daniel P. Berrange 19 Apr '07

19 Apr '07

At the Xen summit a number of us thought it would be a good idea to have an IRC chat room for people interested in technical discussions relating to hypervisor agnostic management tools / APIs / protocols. This includes libvirt, virt-manager, VNC extensions for virtualization, and more. Thus we decided to colonise the channel #virt on irc.oftc.net - feel free to come and join us.... Regards, Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

1 0

[Libvir] ia64 support
by Aron Griffis 18 Apr '07

18 Apr '07

Hi Daniel, Yesterday in your libvirt presentation at Xen Summit, you listed sparc and ppc as architectures with explicit support in libvirt. I looked through the code to see how to add ia64 to that list, but I'm not seeing much mention of sparc or ppc except in the qemud backend. Is there somewhere else I should be looking? Thanks, Aron

2 3

[Libvir] Documentation for remote patch (for discussion only)
by Richard W.M. Jones 18 Apr '07

18 Apr '07

Attached, a patch which gives almost complete documentation for the remote patch. Highlights: * Remote URIs * Transports * Limitations * Implementation Rich. -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903 Directors: Michael Cunningham (USA), Charlie Peters (USA) and David Owens (Ireland)

2 4

[Libvir] [PATCH] Tidy up the error messages produced by test driver
by Richard W.M. Jones 18 Apr '07

18 Apr '07

This minor patch: * Introduces a VIR_FROM_TEST class of errors and makes errors generated by test come from that class. * Provides a reasonable error message if the user forgets to supply a path. Also stops the driver from trying to open ("/"). Example: $ src/virsh -c test:/// capabilities libvir: Test error : invalid argument in testOpen: supply a path or use test:///default lt-virsh: error: failed to connect to the hypervisor * Arranges that Xen-unified appears last in the list of drivers, because qemu and test have well-defined URI syntaxes so can be checked first. 'make check' passes all tests with this patch. Rich. -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903 Directors: Michael Cunningham (USA), Charlie Peters (USA) and David Owens (Ireland)

2 2

[Libvir] Remote patch 2007-04-17 (for discussion only)
by Richard W.M. Jones 17 Apr '07

17 Apr '07

Attached is the latest remote patch which should apply against either CVS or the just-released libvirt-0.2.2. It's only for discussion because it's not quite complete yet. This patch includes Dan's QEMUD/XDR patch. And it basically follows the plan outlined in [1]. Problems: * QEMUD deadlocks when you try to call qemu over remote (eg. try: virsh -c qemu+unix:///system?socket=/path/to/libvirt/qemud-sock capabilities). This is because qemud tries to connect back to itself, but it is in the middle of handling the request, so it doesn't respond. * Opening Xen over remote fails, because the Xen unified driver doesn't understand URLs like xen:///. I'm not sure if it ever did, or if this is a regresssion, but it certainly _ought_ to support this style of URLs. * The server will quite happily answer a mixture of remote and qemud requests even on the same connection (of course, libvirt would never generate such a mix). I'm not sure if this is a problem. To do: * Only supports a small subset of calls (not difficult to do but I'm trying to get the framework working first). * Needs GnuTLS support on the server side (copy & paste of the code from the old remote patch). * Needs to support TCP sockets on the server side. Rich. [1] https://www.redhat.com/archives/libvir-list/2007-March/msg00333.html -- Emerging Technologies, Red Hat http://et.redhat.com/~rjones/ 64 Baker Street, London, W1U 7DF Mobile: +44 7866 314 421 Registered Address: Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor, Berkshire, SL4 1TE, United Kingdom. Registered in England and Wales under Company Registration No. 3798903 Directors: Michael Cunningham (USA), Charlie Peters (USA) and David Owens (Ireland)

1 0

[Libvir] Release of libvirt-0.2.2
by Daniel Veillard 17 Apr '07

17 Apr '07

Time for the monthly release ! A lot of bug fixes have accumulated, and support for the upcoming Xen-3.0.5 should hopefully be ready: ftp://libvirt.org/libvirt/ Documentation: - fix errors due to Amaya (with Simon Hernandez) - virsh uses kB not bytes (Atsushi SAKAI) - add command line help to qemud (Richard Jones) - xenUnifiedRegister docs (Atsushi SAKAI) - strings typos (Nikolay Sivov) - ilocalization probalem raised by Thomas Canniot Bug fixes: - virsh memory values test (Masayuki Sunou) - operations without libvirt_qemud (Atsushi SAKAI) - fix spec file (Florian La Roche - Jeremy Katz - Michael Schwendt) - direct hypervisor call (Atsushi SAKAI) - buffer overflow on qemu networking command (Daniel Berrange) - buffer overflow in quemud (Daniel Berrange) - virsh vcpupin bug (Masayuki Sunou) - host PAE detections and strcuctures size (Richard Jones) - Xen PAE flag handling (Daniel Berrange) - bridged config configuration (Daniel Berrange) - erroneous XEN_V2_OP_SETMAXMEM value (Masayuki Sunou) - memory free error (Mark McLoughlin) - set VIR_CONNECT_RO on read-only connections (S.Sakamoto) - avoid memory explosion bug (Daniel Berrange) - integer overflow for qemu CPU time (Daniel Berrange) - QEMU binary path check (Daniel Berrange) Cleanups: - remove some global variables (Jim Meyering) - printf-style functions checks (Jim Meyering) - better virsh error messages - increase compiler checkings and security (Daniel Berrange) - virBufferGrow usage and docs - use calloc instead of malloc/memset - replace all sprintf by snprintf - avoid configure clobbering user's CTAGS (Jim Meyering) - signal handler error cleanup (Richard Jones) - iptables internal code claenup (Mark McLoughlin) - unified Xen driver (Richard Jones) - cleanup XPath libxml2 calls - IPTables rules tightening (Daniel Berrange) Improvements: - more regression tests on XML (Daniel Berrange) - Python bindings now generate exception in error cases (Richard Jones) - Python bindings for vir*GetAutoStart (Daniel Berrange) - handling of CD-Rom device without device name (Nobuhiro Itou) - fix hypervisor call to work with Xen 3.0.5 (Daniel Berrange) - DomainGetOSType for inactive domains (Daniel Berrange) - multiple boot devices for HVM (Daniel Berrange) thanks to everybody who helped, by raising problems or contributing ideas and patches ! Daniel -- Red Hat Virtualization group http://redhat.com/virtualization/ Daniel Veillard | virtualization library http://libvirt.org/ veillard(a)redhat.com | libxml GNOME XML XSLT toolkit http://xmlsoft.org/ http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/

1 0

[Libvir] [PATCH] About remote operation restrictions of a general user
by S.Sakamoto 17 Apr '07

17 Apr '07

Hi When I use virt-connect in a general user, and a general user can perform operation system such as "start". So, here's the patch adds to can't use operation system in "connect" in a general user. Signed-off-by: Shigeki Sakamoto <fj0588di(a)aa.jp.fujitsu.com> Thanks, Shigeki Sakamoto. Index: src/virsh.c =================================================================== RCS file: /data/cvs/libvirt/src/virsh.c,v retrieving revision 1.71 diff -u -p -r1.71 virsh.c --- src/virsh.c 2 Apr 2007 11:56:38 -0000 1.71 +++ src/virsh.c 4 Apr 2007 05:51:08 -0000 @@ -360,6 +360,8 @@ static int cmdConnect(vshControl * ctl, vshCmd * cmd) { int ro = vshCommandOptBool(cmd, "readonly"); + virDomainPtr dom; + char *ret; if (ctl->conn) { if (virConnectClose(ctl->conn) != 0) { @@ -374,12 +376,25 @@ cmdConnect(vshControl * ctl, vshCmd * cm free(ctl->name); ctl->name = vshStrdup(ctl, vshCommandOptString(cmd, "name", NULL)); + ctl->conn = virConnectOpenReadOnly(ctl->name); + ctl->readonly = 1; + if (!ro && ctl->uid != 0){ + dom = virDomainLookupByID(ctl->conn, 0); + if (dom==NULL) + return FALSE; + ret = virDomainGetXMLDesc(dom, 0); + if (NULL != strstr(ret, "domain type='xen'")) + ro = 1; + free(ret); + } if (!ro) { + if (virConnectClose(ctl->conn) != 0) { + vshError(ctl, FALSE, + _("Failed to disconnect from the hypervisor")); + return FALSE; + } ctl->conn = virConnectOpen(ctl->name); ctl->readonly = 0; - } else { - ctl->conn = virConnectOpenReadOnly(ctl->name); - ctl->readonly = 1; } if (!ctl->conn) @@ -3175,6 +3190,9 @@ _vshStrdup(vshControl * ctl, const char static int vshInit(vshControl * ctl) { + virDomainPtr dom; + char *ret; + if (ctl->conn) return FALSE; @@ -3188,8 +3206,26 @@ vshInit(vshControl * ctl) !strcasecmp(ctl->name, "xen")) && ctl->uid != 0) ctl->readonly = 1; - if (!ctl->readonly) + if (!ctl->readonly) { ctl->conn = virConnectOpen(ctl->name); + + if (ctl->uid != 0){ + dom = virDomainLookupByID(ctl->conn, 0); + if (dom==NULL) + return FALSE; + ret = virDomainGetXMLDesc(dom, 0); + if (NULL != strstr(ret, "domain type='xen'")) { + if (virConnectClose(ctl->conn) != 0) { + vshError(ctl, FALSE, + _("Failed to disconnect from the hypervisor")); + return FALSE; + } + ctl->conn = virConnectOpenReadOnly(ctl->name); + ctl->readonly = 1; + } + free(ret); + } + } else ctl->conn = virConnectOpenReadOnly(ctl->name);

4 9

[Libvir] [PATCH] virsh version
by Atsushi SAKAI 16 Apr '07

16 Apr '07

Hi, Currently virsh version is not work. Currently just show the libvirt version only. This is caused by mis-matching of xen(xen_unified.c) and Xen(libvirt.c) . I do not know the policy. So I attached 2 Pattern. Thanks Atsushi SAKAI =================Pattern1================ --- libvirt.orig/src/xen_unified.c 2007-04-13 09:43:57.000000000 +0900 +++ libvirt.Xen/src/xen_unified.c 2007-04-13 20:02:46.000000000 +0900 @@ -89,8 +89,8 @@ xenUnifiedOpen (virConnectPtr conn, cons /* If name == NULL, name == "", or begins with "xen", then it's for us. */ if (!name || name[0] == '\0') - name = "xen"; - if (strncasecmp (name, "xen", 3) != 0) + name = "Xen"; + if (strncasecmp (name, "Xen", 3) != 0) return VIR_DRV_OPEN_DECLINED; /* Allocate per-connection private data. */ @@ -719,7 +719,7 @@ xenUnifiedDomainSetAutostart (virDomainP /* The interface which we export upwards to libvirt.c. */ static virDriver xenUnifiedDriver = { .no = VIR_DRV_XEN_UNIFIED, - .name = "xen", + .name = "Xen", .ver = VERSION, .open = xenUnifiedOpen, .close = xenUnifiedClose, =================Pattern2================ --- libvirt.orig/src/libvirt.c 2007-04-12 23:18:39.000000000 +0900 +++ libvirt.Xen/src/libvirt.c 2007-04-13 20:09:02.000000000 +0900 @@ -246,7 +246,7 @@ virGetVersion(unsigned long *libVer, con if (typeVer != NULL) { if (type == NULL) - type = "Xen"; + type = "xen"; for (i = 0;i < virDriverTabCount;i++) { if ((virDriverTab[i] != NULL) && (!strcmp(virDriverTab[i]->name, type))) {

2 1

[Libvir] PATCH: Check for /usr/bin/qemu* binaries before starting guest
by Daniel P. Berrange 16 Apr '07

16 Apr '07

The attached patch fixes two issues - It explicitly checks to see if the requested /usr/bin/qemu* binary actually exists before fork()/exec()'ing it. While its technically possible to catch the execve() failure of ENOENT, its a real pain to feed this error back to the qemu daemon because we're in a sub-process at that point. The obvious & easy solution is to thus check to see if the binary exists before trying to fork. - It only passes the -no-kqemu flag if we're running matching arch to the host. Previously it would pass the -no-kqemu flag even if running an x86_64 guest on i686 platform - QEMU then exited with error because the -no-kqemu flag is not recognised on x86_64 when running on i686 host. Dan. -- |=- Red Hat, Engineering, Emerging Technologies, Boston. +1 978 392 2496 -=| |=- Perl modules: http://search.cpan.org/~danberr/ -=| |=- Projects: http://freshmeat.net/~danielpb/ -=| |=- GnuPG: 7D3B9505 F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 -=|

4 3