October 2009 - Devel - libvirt List Archives

[libvirt] [PATCH] create() and destroy() support for Power Hypervisor
by Eduardo Otubo 05 Nov '09

05 Nov '09

Hello all, This patch includes a lot of changes: * I changed all references of uuid_db to uuid_table. Bigger name, but this semantic has a better understanding. * Now we have a little control of UUID generated for each partition. It's based on a table that matches UUID with LPAR's IDs, I keep it in a file that is stored in the managed system (HMC or IVM). Even having isolated functions to manipulate this file I still need to implement a way to lock it and make the operation atomic to avoid corruptions. * The XML file used in the create() function still must be improved. I used the <disk> tag to make a work around to handle storage pools. Now I ask for some help, how do I use the <pool> tag if there is no reference to it at the virDomainDef structure? I've been told that libvirt possibly would make a mini-release this week to push some major fixes on Fedora 12. Is there a little possibility to push these new phyp features too? This would be very important for the phyp project, since this functions are not just to manage partitions but manipulate them. Any comments are always welcome. []'s -- Eduardo Otubo Software Engineer Linux Technology Center IBM Systems & Technology Group Mobile: +55 19 8135 0885 eotubo(a)linux.vnet.ibm.com

5 13

[libvirt] [PATCH 0/3] IPv6 / multi address / report live config in virInterface
by laine＠laine.org 03 Nov '09

03 Nov '09

This is the 2nd iteration of this patchset incorporating danpb's suggestions. Along with a couple other minor things, the type attribute is again a required attribute, and netcf support will be omitted in the build unless you have netcf-devel-0.1.3 or greater. Since there is no netcf-devel-0.1.3 yet, I tested this by changing the appropriate strings to 0.1.2. 0.1.3 will be released "soon" (I want to try and get IPv6 address reporting in before lutter cuts another release). I just wanted to get the patches to the list early to get some visual review in case I screwed up anything new ;-) Note that (unless you're building/installing netcf from the head of git) if you try to fake out the version of netcf yourself you'll get runtime errors from virInterfaceGetXMLDesc (iface-dumpxml) due to there being no "type" attribute in the interface xml.

4 21

[libvirt] [PATCH] Allow NULL mac address in virGetInterface.
by Laine Stump 03 Nov '09

03 Nov '09

There are places where an interface will not have a mac address, and netcf returns this as a NULL pointer rather than a pointer to an empty string. Rather than checking for this all over the place in libvirt, just save it in the virInterface object as an empty string. --- src/datatypes.c | 7 ++++++- 1 files changed, 6 insertions(+), 1 deletions(-) diff --git a/src/datatypes.c b/src/datatypes.c index 89ad309..ecefc59 100644 --- a/src/datatypes.c +++ b/src/datatypes.c @@ -588,10 +588,15 @@ virInterfacePtr virGetInterface(virConnectPtr conn, const char *name, const char *mac) { virInterfacePtr ret = NULL; - if ((!VIR_IS_CONNECT(conn)) || (name == NULL) || (mac == NULL)) { + if ((!VIR_IS_CONNECT(conn)) || (name == NULL)) { virLibConnError(NULL, VIR_ERR_INVALID_ARG, __FUNCTION__); return(NULL); } + + /* a NULL mac from caller is okay. Treat it as blank */ + if (mac == NULL) + mac = ""; + virMutexLock(&conn->lock); ret = (virInterfacePtr) virHashLookup(conn->interfaces, name); -- 1.6.5.15.gc274d

2 3

[libvirt] [PATCH 00/21] Rewrite the QEMU monitor handling
by Daniel P. Berrange 02 Nov '09

02 Nov '09

This patch series rewrites the QEMU monitor handling almost completely. The key theme here is to move from a totally synchronous way of interacting with the monitor, to a totally asynchronous way. This allows us to handle receipt & dispatch of asychronous events from QEMU. For example a notification of a disk-full error, or VM state change. In the process of doing this re-factoring I have also dropped in basic support/infrastructure for the JSON based monitor. The basic JSON parsing/formatting is working, but this needs to wait until QEMU community finalize their monitor syntax for the new commands & what args they accept. Daniel

5 45

[libvirt] [PATCH] ESX: Don't automatically follow redirects.
by Matthias Bolte 02 Nov '09

02 Nov '09

The default transport for the VI API is HTTPS. If the server redirects from HTTPS to HTTP the driver would silently follow that redirection. The user assumes to communicate with the server over a secure transport but isn't. This patch disables automatical redirection following. The driver reports an error if the server tries to redirect. * src/esx/esx_vi.c: refactor the call to curl_easy_perform() into a function and do error handling there, disable automatical redirection following for curl * src/esx/esx_vi.h: change the type of responseCode to int --- src/esx/esx_vi.c | 166 +++++++++++++++++++++++++++--------------------------- src/esx/esx_vi.h | 2 +- 2 files changed, 84 insertions(+), 84 deletions(-) diff --git a/src/esx/esx_vi.c b/src/esx/esx_vi.c index dc51e20..bcf110f 100644 --- a/src/esx/esx_vi.c +++ b/src/esx/esx_vi.c @@ -216,6 +216,68 @@ esxVI_CURL_Debug(CURL *curl ATTRIBUTE_UNUSED, curl_infotype type, } #endif +static int +esxVI_CURL_Perform(virConnectPtr conn, esxVI_Context *ctx, const char *url) +{ + CURLcode errorCode; + long responseCode = 0; +#if LIBCURL_VERSION_NUM >= 0x071202 /* 7.18.2 */ + const char *redirectUrl = NULL; +#endif + + errorCode = curl_easy_perform(ctx->curl_handle); + + if (errorCode != CURLE_OK) { + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "curl_easy_perform() returned an error: %s (%d)", + curl_easy_strerror(errorCode), errorCode); + return -1; + } + + errorCode = curl_easy_getinfo(ctx->curl_handle, CURLINFO_RESPONSE_CODE, + &responseCode); + + if (errorCode != CURLE_OK) { + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "curl_easy_getinfo(CURLINFO_RESPONSE_CODE) returned an " + "error: %s (%d)", curl_easy_strerror(errorCode), + errorCode); + return -1; + } + + if (responseCode < 0) { + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "curl_easy_getinfo(CURLINFO_RESPONSE_CODE) returned a " + "negative response code"); + return -1; + } + + if (responseCode == 301) { +#if LIBCURL_VERSION_NUM >= 0x071202 /* 7.18.2 */ + errorCode = curl_easy_getinfo(ctx->curl_handle, CURLINFO_REDIRECT_URL, + &redirectUrl); + + if (errorCode != CURLE_OK) { + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "curl_easy_getinfo(CURLINFO_REDIRECT_URL) returned " + "an error: %s (%d)", curl_easy_strerror(errorCode), + errorCode); + } else { + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "The server redirects from '%s' to '%s'", url, + redirectUrl); + } +#else + ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, + "The server redirects from '%s'", url); +#endif + + return -1; + } + + return responseCode; +} + int esxVI_Context_Connect(virConnectPtr conn, esxVI_Context *ctx, const char *url, const char *ipAddress, const char *username, @@ -269,7 +331,7 @@ esxVI_Context_Connect(virConnectPtr conn, esxVI_Context *ctx, const char *url, curl_easy_setopt(ctx->curl_handle, CURLOPT_URL, ctx->url); curl_easy_setopt(ctx->curl_handle, CURLOPT_USERAGENT, "libvirt-esx"); curl_easy_setopt(ctx->curl_handle, CURLOPT_HEADER, 0); - curl_easy_setopt(ctx->curl_handle, CURLOPT_FOLLOWLOCATION, 1); + curl_easy_setopt(ctx->curl_handle, CURLOPT_FOLLOWLOCATION, 0); curl_easy_setopt(ctx->curl_handle, CURLOPT_SSL_VERIFYPEER, noVerify ? 0 : 1); curl_easy_setopt(ctx->curl_handle, CURLOPT_SSL_VERIFYHOST, noVerify ? 0 : 2); curl_easy_setopt(ctx->curl_handle, CURLOPT_COOKIEFILE, ""); @@ -433,8 +495,7 @@ esxVI_Context_DownloadFile(virConnectPtr conn, esxVI_Context *ctx, const char *url, char **content) { virBuffer buffer = VIR_BUFFER_INITIALIZER; - CURLcode errorCode; - long responseCode; + int responseCode = 0; if (content == NULL || *content != NULL) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "Invalid argument"); @@ -448,27 +509,19 @@ esxVI_Context_DownloadFile(virConnectPtr conn, esxVI_Context *ctx, curl_easy_setopt(ctx->curl_handle, CURLOPT_UPLOAD, 0); curl_easy_setopt(ctx->curl_handle, CURLOPT_HTTPGET, 1); - errorCode = curl_easy_perform(ctx->curl_handle); - - if (errorCode != CURLE_OK) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_perform() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; - } + responseCode = esxVI_CURL_Perform(conn, ctx, url); - errorCode = curl_easy_getinfo(ctx->curl_handle, CURLINFO_RESPONSE_CODE, - &responseCode); + virMutexUnlock(&ctx->curl_lock); - if (errorCode != CURLE_OK) { + if (responseCode < 0) { + goto failure; + } else if (responseCode != 200) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_getinfo() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; + "HTTP response code %d while trying to download '%s'", + responseCode, url); + goto failure; } - virMutexUnlock(&ctx->curl_lock); - if (virBufferError(&buffer)) { virReportOOMError(conn); goto failure; @@ -476,32 +529,19 @@ esxVI_Context_DownloadFile(virConnectPtr conn, esxVI_Context *ctx, *content = virBufferContentAndReset(&buffer); - if (responseCode != 200) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "HTTP response code %d while trying to download '%s'", - (int)responseCode, url); - goto failure; - } - return 0; failure: free(virBufferContentAndReset(&buffer)); return -1; - - unlock: - virMutexUnlock(&ctx->curl_lock); - - goto failure; } int esxVI_Context_UploadFile(virConnectPtr conn, esxVI_Context *ctx, const char *url, const char *content) { - CURLcode errorCode; - long responseCode; + int responseCode = 0; if (content == NULL) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "Invalid argument"); @@ -515,40 +555,20 @@ esxVI_Context_UploadFile(virConnectPtr conn, esxVI_Context *ctx, curl_easy_setopt(ctx->curl_handle, CURLOPT_UPLOAD, 1); curl_easy_setopt(ctx->curl_handle, CURLOPT_INFILESIZE, strlen(content)); - errorCode = curl_easy_perform(ctx->curl_handle); - - if (errorCode != CURLE_OK) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_perform() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; - } - - errorCode = curl_easy_getinfo(ctx->curl_handle, CURLINFO_RESPONSE_CODE, - &responseCode); - - if (errorCode != CURLE_OK) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_getinfo() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; - } + responseCode = esxVI_CURL_Perform(conn, ctx, url); virMutexUnlock(&ctx->curl_lock); - if (responseCode != 200 && responseCode != 201) { + if (responseCode < 0) { + return -1; + } else if (responseCode != 200 && responseCode != 201) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "HTTP response code %d while trying to upload to '%s'", - (int)responseCode, url); + responseCode, url); return -1; } return 0; - - unlock: - virMutexUnlock(&ctx->curl_lock); - - return -1; } int @@ -558,7 +578,6 @@ esxVI_Context_Execute(virConnectPtr conn, esxVI_Context *ctx, { virBuffer buffer = VIR_BUFFER_INITIALIZER; esxVI_Fault *fault = NULL; - CURLcode errorCode; if (request == NULL || response == NULL || *response != NULL) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, "Invalid argument"); @@ -577,27 +596,14 @@ esxVI_Context_Execute(virConnectPtr conn, esxVI_Context *ctx, curl_easy_setopt(ctx->curl_handle, CURLOPT_POSTFIELDS, request); curl_easy_setopt(ctx->curl_handle, CURLOPT_POSTFIELDSIZE, strlen(request)); - errorCode = curl_easy_perform(ctx->curl_handle); - - if (errorCode != CURLE_OK) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_perform() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; - } + (*response)->responseCode = esxVI_CURL_Perform(conn, ctx, ctx->url); - errorCode = curl_easy_getinfo(ctx->curl_handle, CURLINFO_RESPONSE_CODE, - &(*response)->responseCode); + virMutexUnlock(&ctx->curl_lock); - if (errorCode != CURLE_OK) { - ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "curl_easy_getinfo() returned an error: %s (%d)", - curl_easy_strerror(errorCode), errorCode); - goto unlock; + if ((*response)->responseCode < 0) { + goto failure; } - virMutexUnlock(&ctx->curl_lock); - if (virBufferError(&buffer)) { virReportOOMError(conn); goto failure; @@ -695,8 +701,7 @@ esxVI_Context_Execute(virConnectPtr conn, esxVI_Context *ctx, } } else if ((*response)->responseCode != 200) { ESX_VI_ERROR(conn, VIR_ERR_INTERNAL_ERROR, - "HTTP response code %d", (int)(*response)->responseCode); - + "HTTP response code %d", (*response)->responseCode); goto failure; } @@ -708,11 +713,6 @@ esxVI_Context_Execute(virConnectPtr conn, esxVI_Context *ctx, esxVI_Fault_Free(&fault); return -1; - - unlock: - virMutexUnlock(&ctx->curl_lock); - - goto failure; } diff --git a/src/esx/esx_vi.h b/src/esx/esx_vi.h index 4892fde..4b3005e 100644 --- a/src/esx/esx_vi.h +++ b/src/esx/esx_vi.h @@ -110,7 +110,7 @@ int esxVI_Context_Execute(virConnectPtr conn, esxVI_Context *ctx, */ struct _esxVI_Response { - long responseCode; /* required */ + int responseCode; /* required */ char *content; /* required */ xmlDocPtr document; /* optional */ xmlXPathContextPtr xpathContext; /* optional */ -- 1.6.0.4

3 5

[libvirt] [patch] Add openvzDomainSetMemoryInternal
by Yuji NISHIDA 02 Nov '09

02 Nov '09

Hi all, This patch is to set KMEMSIZE for OpenVZ. This function is used for initializing the parameter of KMEMSIZE to start container. openvzDomainSetMemory should be left for another purpose so I added new one. I think users should specify memory as kbyte ( or bigger ). --- src/openvz/openvz_driver.c | 32 ++++++++++++++++++++++++++++++++ 1 files changed, 32 insertions(+), 0 deletions(-) diff --git a/src/openvz/openvz_driver.c b/src/openvz/openvz_driver.c index f64ad1e..5c1fefa 100644 --- a/src/openvz/openvz_driver.c +++ b/src/openvz/openvz_driver.c @@ -69,6 +69,7 @@ static int openvzGetMaxVCPUs(virConnectPtr conn, const char *type); static int openvzDomainGetMaxVcpus(virDomainPtr dom); static int openvzDomainSetVcpus(virDomainPtr dom, unsigned int nvcpus); static int openvzDomainSetVcpusInternal(virConnectPtr conn, virDomainObjPtr vm, unsigned int nvcpus); +static int openvzDomainSetMemoryInternal(virConnectPtr conn, virDomainObjPtr vm, unsigned long memory); static void openvzDriverLock(struct openvz_driver *driver) { @@ -803,6 +804,14 @@ openvzDomainDefineXML(virConnectPtr conn, const char *xml) } } + if (vm->def->memory > 0) { + if (openvzDomainSetMemoryInternal(conn, vm, vm->def->memory) < 0) { + openvzError(conn, VIR_ERR_INTERNAL_ERROR, + "%s", _("Could not set memory size")); + goto cleanup; + } + } + dom = virGetDomain(conn, vm->def->name, vm->def->uuid); if (dom) dom->id = -1; @@ -1364,6 +1373,29 @@ static int openvzNumDefinedDomains (virConnectPtr conn) { return ninactive; } +static int openvzDomainSetMemoryInternal(virConnectPtr conn, virDomainObjPtr vm, + unsigned long mem) { + struct openvz_driver *driver = conn->privateData; + char str_mem[16]; + const char *prog[] = { VZCTL, "--quiet", "set", PROGRAM_SENTINAL, + "--kmemsize", str_mem, "--save", NULL }; + + /* memory has to be changed its format from kbyte to byte */ + snprintf( str_mem, sizeof(str_mem), "%lu", mem * 1024 ); + + openvzSetProgramSentinal(prog, vm->def->name); + if (virRun(conn, prog, NULL) < 0) { + openvzError(conn, VIR_ERR_INTERNAL_ERROR, + _("Could not exec %s"), VZCTL); + goto cleanup; + } + + return 0; + +cleanup: + return -1; +} + static virDriver openvzDriver = { VIR_DRV_OPENVZ, "OPENVZ", -- 1.5.3.4 ----- Yuji Nishida nishidy(a)nict.go.jp

2 1

[libvirt] Don't use private struct member names of in6_addr
by Matthias Bolte 02 Nov '09

02 Nov '09

__in6_u.__u6_addr16 is the private name for this struct member, s6_addr16 is the public one. Our buildserver revealed this problem, because for some reason the internal names are missing the __ prefix there. Matthias

2 2

Re: [libvirt] [PATCH 10/21] Don't let parent of daemon exit until basic initialization is done
by Laurent Léonard 31 Oct '09

31 Oct '09

Hi, I'm not a C expert, but I see there is a use of an unitialized variable: default: { int got, exitstatus = 0; int ret; char status; close(statuspipe[1]); /* We wait to make sure the first child forked successfully */ if ((got = waitpid(pid, &exitstatus, 0)) < 0 || got != pid || status != 0) { <<< here return -1; } That explains the issue described by Guido, because he uses an i386 system. It was impossible for me to reproduce the issue because I use an amd64 system. char test; if (test == 0) puts("ok"); else puts("ko"); produces: - "ok" on amd64 system - "ko" on i386 system If I use the exitstatus variable in the test it seems to work great on both architectures. -- Laurent Léonard

2 1

[libvirt] [PATCH] network utilities
by Daniel Veillard 30 Oct '09

30 Oct '09

Revamping my previous patches to parse IPv4/6 numeric addresses, adding range computation and netmask checking entries too. As suggested I used an union for the socket address type: typedef union { struct sockaddr_storage stor; struct sockaddr_in inet4; struct sockaddr_in6 inet6; } virSocketAddr; typedef virSocketAddr *virSocketAddrPtr; But I wonder if it's not better to make it a struct and add the lenght of the address as this is usually needed when passing one of the struct sockaddr_* and that lenght is returned as part of parsing, keeping both together is probably more useful. A macro to get the type AF_INET/AF_INET6 out of a virSocketAddr might be useful instead of having the users manually guess that ->stor.ss_family is teh filed to look at. Also I wonder how useful the Ipv6 code for netmask and ranges really is, it seems that IPv6 mostly go with prefix and never use netmasks, maybe that's just dead code, maybe that's just wrong, anyway here it is. I named the header and file network.[ch] as I think other network related utilities like validation or generation of MAC addresses could well be moved in that file too. Daniel -- Daniel Veillard | libxml Gnome XML XSLT toolkit http://xmlsoft.org/ daniel(a)veillard.com | Rpmfind RPM search engine http://rpmfind.net/ http://veillard.com/ | virtualization library http://libvirt.org/

3 8

[libvirt] [PATCH] Fix up a bogus TLS message.
by Chris Lalancette 30 Oct '09

30 Oct '09

When working with the TLS transport, I noticed that every single time a remote stream was closed, I would get a message like: 09:09:40.793: error : remoteIOReadBuffer:7328 : failed to read from TLS socket A TLS packet with unexpected length was received. libvir: QEMU error : failed to read from TLS socket A TLS packet with unexpected length was received. in the logs. This happens because of a race in libvirtd; one thread handles the doRemoteClose(), which calls gnutls_bye() followed by close() while another thread is poll()'ing on the same fd. Once the close() happens, the poll returns with revents set to POLLIN, and we would poll one more time for data from the now-closed fd. Fix this situation by setting poll->session to NULL when we clean up, and check for that in remoteIOHandleInput(). Signed-off-by: Chris Lalancette <clalance(a)redhat.com> --- src/remote/remote_driver.c | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-) diff --git a/src/remote/remote_driver.c b/src/remote/remote_driver.c index bf001eb..335f44b 100644 --- a/src/remote/remote_driver.c +++ b/src/remote/remote_driver.c @@ -1389,6 +1389,7 @@ doRemoteClose (virConnectPtr conn, struct private_data *priv) if (priv->uses_tls && priv->session) { gnutls_bye (priv->session, GNUTLS_SHUT_RDWR); gnutls_deinit (priv->session); + priv->session = NULL; } #if HAVE_SASL if (priv->saslconn) @@ -7223,6 +7224,12 @@ remoteIOReadBuffer(virConnectPtr conn, int ret; if (priv->uses_tls) { + if (!priv->session) { + /* we may have reached here one more time after gnutls_bye() + * was called, so just return here + */ + return 0; + } tls_resend: ret = gnutls_record_recv (priv->session, bytes, len); if (ret == GNUTLS_E_INTERRUPTED) -- 1.6.0.6

3 3