- CIM - libvirt List Archives

[PATCH] ECTP: Adoption of changes to std_assoc struct in std_association
by Heidi Eckhart 29 Nov '07

29 Nov '07

# HG changeset patch # User Heidi Eckhart <heidieck(a)linux.vnet.ibm.com> # Date 1196287282 -3600 # Node ID 786976aa40c60088afe17571372b4a5d82a0d29e # Parent 535f219fe23fec8a5de725e0875e4fcdadbd584c ECTP: Adoption of changes to std_assoc struct in std_association This is the follow up patch to the API changes made in std_association. The struct std_assoc does now take lists for source, target and assoc classnames. It also adds the registration of ComputerSystem and HostSystem to the interop namespace. This is necessary to check the client's resultClass value. Signed-off-by: Heidi Eckhart <heidieck(a)linux.vnet.ibm.com> diff -r 535f219fe23f -r 786976aa40c6 Makefile.am --- a/Makefile.am Wed Nov 28 11:39:55 2007 +0100 +++ b/Makefile.am Wed Nov 28 23:01:22 2007 +0100 @@ -36,6 +36,8 @@ MOFS = \ schema/ElementSettingData.mof INTEROP_MOFS = \ + schema/ComputerSystem.mof \ + schema/HostSystem.mof \ schema/RegisteredProfile.mof \ schema/ElementConformsToProfile.mof diff -r 535f219fe23f -r 786976aa40c6 src/Virt_ElementConformsToProfile.c --- a/src/Virt_ElementConformsToProfile.c Wed Nov 28 11:39:55 2007 +0100 +++ b/src/Virt_ElementConformsToProfile.c Wed Nov 28 23:01:22 2007 +0100 @@ -224,27 +224,47 @@ static CMPIInstance *make_ref(const CMPI return assoc_inst; } +char* conformant_standard[] = { + "Xen_RegisteredProfile", + "KVM_RegisteredProfile", + NULL +}; + +char* managed_element[] = { + "Xen_HostSystem", + "Xen_ComputerSystem", + "KVM_HostSystem", + "KVM_ComputerSystem", + NULL +}; + +char* assoc_classname[] = { + "Xen_ElementConformsToProfile", + "KVM_ElementConformsToProfile", + NULL +}; + struct std_assoc forward = { - .source_class = "CIM_RegisteredProfile", + .source_class = (char**)&conformant_standard, .source_prop = "ConformantStandard", - .target_class = "CIM_ManagedElement", + .target_class = (char**)&managed_element, .target_prop = "ManagedElement", - .assoc_class = NULL, + .assoc_class = (char**)&assoc_classname, .handler = prof_to_elem, .make_ref = make_ref }; struct std_assoc backward = { - .source_class = "CIM_ManagedElement", + .source_class = (char**)&managed_element, .source_prop = "ManagedElement", - .target_class = "CIM_RegisteredProfile", + .target_class = (char**)&conformant_standard, .target_prop = "ConformantStandard", - .assoc_class = NULL, + .assoc_class = (char**)&assoc_classname, .handler = elem_to_prof, .make_ref = make_ref @@ -255,7 +275,6 @@ struct std_assoc *assoc_handlers[] = { &backward, NULL }; - STDA_AssocMIStub(, Virt_ElementConformsToProfileProvider, _BROKER, libvirt_cim_init(), assoc_handlers); /*

3 3

[PATCH 0 of 3] #2 - Enhance std association logic for reference lists and input parameter handling
by Heidi Eckhart 29 Nov '07

29 Nov '07

This patch set fixes: - enhance the std_assoc struct to support lists for source, target and association classnames - fixed a wrong error return for a NULL handler - reworked check for input parameter: assocClass, resultClass, role, resultRole

2 4

[PATCH 0 of 2] Fix VSMC enumeration failure.
by Kaitlin Rupert 29 Nov '07

29 Nov '07

wbemcli ein http://root:elm3b41@localhost/root/virt:Xen_VirtualSystemManagementCapabili… This currently returns: "Missing key: Name". This is because get_vsm_cap() is expecting a Name attribute in the ref, which doesn't exist on an enumeration query. This patch set fixes the enum issue in VSMC. Part of this fix is to add a new parameter to get_vsm_cap(). One of the handlers in ElementCapabilities calls get_vsm_cap(), so that function call has been updated as well.

2 3

[PATCH] Enhance handling of input parameter of std_association logic
by Heidi Eckhart 29 Nov '07

29 Nov '07

# HG changeset patch # User Heidi Eckhart <heidieck(a)linux.vnet.ibm.com> # Date 1196161060 -3600 # Node ID a9d4ef4c783c788db2b8ebd9da71255e492a0950 # Parent 16ac621eb3b34d11caca00100901d27bede7a7c7 Enhance handling of input parameter of std_association logic Signed-off-by: Heidi Eckhart <heidieck(a)linux.vnet.ibm.com> diff -r 16ac621eb3b3 -r a9d4ef4c783c std_association.c --- a/std_association.c Tue Nov 27 11:45:33 2007 +0100 +++ b/std_association.c Tue Nov 27 11:57:40 2007 +0100 @@ -67,14 +67,20 @@ static bool match_class(const CMPIBroker char *comp_class; int i; + if (test_class == NULL) + return true; + + if (comp_class_list == NULL) + return true; + rop = CMNewObjectPath(broker, ns, test_class, NULL); + if (CMIsNullObject(rop)) + return false; for (i = 0; comp_class_list[i]; i++) { comp_class = comp_class_list[i]; - if ((test_class == NULL) || - (comp_class == NULL) || - match_op(broker, rop, comp_class)) + if (!CMClassPathIsA(broker, rop, comp_class, NULL)) return true; } @@ -137,18 +143,82 @@ out: static struct std_assoc * std_assoc_get_handler(const struct std_assoc_ctx *ctx, + struct std_assoc_info *info, const CMPIObjectPath *ref) { struct std_assoc *ptr = NULL; int i; + bool rc; + + CU_DEBUG("Calling Provider: '%s'", + info->provider_name); for (i = 0; ctx->handlers[i]; i++) { ptr = ctx->handlers[i]; if (match_source_class(ctx->brkr, ref, ptr)) - return ptr; - } - + break; + } + + if (ptr) { + if (info->assoc_class) { + CU_DEBUG("Check client's assocClass: '%s'", + info->assoc_class); + + rc = match_class(ctx->brkr, + NAMESPACE(ref), + info->assoc_class, + ptr->assoc_class); + + if (!rc) { + CU_DEBUG("AssocClass not valid."); + goto out; + } + CU_DEBUG("AssocClass valid."); + } + + if (info->result_class) { + CU_DEBUG("Check client's resultClass: '%s'", + info->result_class); + + rc = match_class(ctx->brkr, + NAMESPACE(ref), + info->result_class, + ptr->target_class); + + if (!rc) { + CU_DEBUG("ResultClass not valid."); + goto out; + } + CU_DEBUG("ResultClass valid."); + } + + if (info->role) { + CU_DEBUG("Check client's role: '%s'", + info->role); + + if (!STREQC(info->role, ptr->source_prop)) { + CU_DEBUG("Role not valid."); + goto out; + } + CU_DEBUG("Role valid."); + } + + if (info->result_role) { + CU_DEBUG("Check client's resultRole: '%s'", + info->result_role); + + if (!STREQC(info->result_role, ptr->target_prop)) { + CU_DEBUG("ResultRole not valid."); + goto out; + } + CU_DEBUG("ResultRole valid."); + } + + return ptr; + } + + out: return NULL; } @@ -158,57 +228,21 @@ static CMPIStatus do_assoc(struct std_as const CMPIObjectPath *ref, bool names_only) { + CMPIStatus s = {CMPI_RC_OK, NULL}; struct inst_list list; - CMPIStatus s; struct std_assoc *handler; - bool rc; inst_list_init(&list); CU_DEBUG("Getting handler..."); - - handler = std_assoc_get_handler(ctx, ref); + handler = std_assoc_get_handler(ctx, info, ref); if (handler == NULL) { CU_DEBUG("No handler found."); goto out; } - - CU_DEBUG("OK.\n\tsource: '%s'\n\ttarget: '%s'\n\tassoc_class: '%s'", - handler->source_class, - handler->target_class, - handler->assoc_class); - CU_DEBUG("Calling match_class: \n\tinfo->result_class: '%s'", - info->result_class); - - rc = match_class(ctx->brkr, - NAMESPACE(ref), - info->result_class, - handler->target_class); - if (!rc) { - CU_DEBUG("Match_class failed."); - cu_statusf(ctx->brkr, &s, - CMPI_RC_ERR_FAILED, - "Result class is not valid for this association"); - goto out; - } - CU_DEBUG("Match_class succeeded."); - - CU_DEBUG("Calling match_class: \n\tinfo->assoc_class: '%s'", - info->assoc_class); - rc = match_class(ctx->brkr, - NAMESPACE(ref), - info->assoc_class, - handler->assoc_class); - if (!rc) { - CU_DEBUG("Match_class failed."); - cu_statusf(ctx->brkr, &s, - CMPI_RC_ERR_FAILED, - "Association class given is not valid for" - "this association"); - goto out; - } - CU_DEBUG("Match_class succeeded, calling handler->handler..."); - + CU_DEBUG("Getting handler succeeded."); + + CU_DEBUG("Calling handler->handler..."); s = handler->handler(ref, info, &list); if (s.rc != CMPI_RC_OK) { @@ -256,32 +290,22 @@ static CMPIStatus do_ref(struct std_asso const CMPIObjectPath *ref, bool names_only) { + CMPIStatus s = {CMPI_RC_OK, NULL}; struct inst_list list; - CMPIStatus s; - int i; struct std_assoc *handler; - bool rc; + int i; inst_list_init(&list); - handler = std_assoc_get_handler(ctx, ref); + CU_DEBUG("Getting handler..."); + handler = std_assoc_get_handler(ctx, info, ref); if (handler == NULL) { CU_DEBUG("No handler found."); goto out; } - - rc = match_class(ctx->brkr, - NAMESPACE(ref), - info->result_class, - handler->assoc_class); - if (!rc) { - cu_statusf(ctx->brkr, &s, - CMPI_RC_ERR_FAILED, - "Result class is not valid for this association"); - goto out; - } - - + CU_DEBUG("Getting handler succeeded."); + + CU_DEBUG("Calling handler->handler..."); s = handler->handler(ref, info, &list); if ((s.rc != CMPI_RC_OK) || (list.list == NULL)) goto out;

3 4

[PATCH] Fix AllocationCapabilities install issue
by Kaitlin Rupert 29 Nov '07

29 Nov '07

# HG changeset patch # User Kaitlin Rupert <karupert(a)us.ibm.com> # Date 1196276782 28800 # Node ID b6fdc32e25897a3cc0b38d31a738d4217c467328 # Parent f230e4d57a78a48efb5c32e70e048ce229c2cca4 Fix AllocationCapabilities install issue. The Virt_AllocationCapabilities provider install fails because it links with Virt_DevicePool, which is installed after Virt_AllocationCapabilities. When you run make install, you should see the following error: /usr/bin/ld: cannot find -lVirt_DevicePool collect2: ld returned 1 exit status libtool: install: error: relink `libVirt_AllocationCapabilities.la' with the above command before installing it Signed-off-by: Kaitlin Rupert <karupert(a)us.ibm.com> diff -r f230e4d57a78 -r b6fdc32e2589 src/Makefile.am --- a/src/Makefile.am Wed Nov 28 11:27:09 2007 +0100 +++ b/src/Makefile.am Wed Nov 28 11:06:22 2007 -0800 @@ -34,12 +34,12 @@ provider_LTLIBRARIES = libVirt_ComputerS libVirt_VirtualSystemManagementService.la \ libVirt_VirtualSystemManagementCapabilities.la \ libVirt_EnabledLogicalElementCapabilities.la \ + libVirt_DevicePool.la \ libVirt_AllocationCapabilities.la \ libVirt_VSSD.la \ libVirt_HostedDependency.la \ libVirt_RegisteredProfile.la \ libVirt_ElementConformsToProfile.la \ - libVirt_DevicePool.la \ libVirt_SettingsDefineCapabilities.la \ libVirt_HostedResourcePool.la \ libVirt_ElementCapabilities.la \

2 1

[PATCH 0 of 2] #2 - Check for matching hypervisor prefixes
by Heidi Eckhart 29 Nov '07

29 Nov '07

- renamed helper function to match_hypervisor_prefix to become clearer on the intent - updated ECTP to make use of match_hypervisor_prefix() function If this approach is ok, I can coop up a patch set for all association classes.

2 3

[PATCH 00 of 17] #2 Updates for libcmpiutil API change
by Dan Smith 29 Nov '07

29 Nov '07

Fixed ElementCapabilities after Kaitlin's recent fix patch. Fixed the STREQ check that was accidentally removed in the last version. This refactors all of the necessary functions in libvirt-cim for the new API. Things only got cleaner, I think. I also found a substantial number of memory leaks in the process, which are fundamentally fixed by the new API which I think is a validation of it as "the right thing to do". All required changes should have been easily spotted by the compiler, but some testing would be good.

2 19

[PATCH 0 of 4] #3 - Fixes to RegisteredProfile and ECTP depending on connect_by_classname approach
by Heidi Eckhart 29 Nov '07

29 Nov '07

-ixes duplicates of RegisteredProfile, adopt changes to RegisteredProfile to association ECTP - fix hypervisor prefix for ECTP; remove error label and let label out do all the free() stuff

2 5

Re: [Libvirt-cim] [PATCH 06 of 18] Fixes to VSMS for libcmpiutil API change
by Kaitlin Rupert 29 Nov '07

29 Nov '07

Kaitlin Rupert wrote: > Dan Smith wrote: >> No, that case is still handled. For full context: >> >> + if (cu_get_str_prop(inst, "NetworkType", &val) != >> CMPI_RC_OK) >> + val = "bridge"; >> + >> free(dev->dev.net.type); >> - cu_get_str_prop(inst, "NetworkType", >> &dev->dev.net.type); >> - if (dev->dev.net.type == NULL) >> - dev->dev.net.type = strdup("bridge"); >> + dev->dev.net.type = strdup(val); >> >> I moved the "not set" (NULL) case above, and set val to "bridge", so >> that below, I can assume it is a valid const string and strdup it >> directly. So val either gets the actual type, or "bridge" by the time >> I get to the strdup. >> >> I did the same reordering in disk with the Address field, making >> "/dev/null" the default if unspecified. Oops.. thanks for the clarification. I caught the Address part, but I missed it for the NetworkType part. Looks good then. =) -- Kaitlin Rupert IBM Linux Technology Center karupert(a)us.ibm.com

1 0

[PATCH 00 of 18] Updates for libcmpiutil API change
by Dan Smith 29 Nov '07

29 Nov '07

This refactors all of the necessary functions in libvirt-cim for the new API. Things only got cleaner, I think. I also found a substantial number of memory leaks in the process, which are fundamentally fixed by the new API which I think is a validation of it as "the right thing to do". All required changes should have been easily spotted by the compiler, but some testing would be good. Tip: To find the leaks, note every time we convert a variable from char * to const char * and don't remove a subsequent free() call :)

5 28