Re: [libvirt] [PATCH v2 00/14] Convert secret driver to use hashed object

v1: http://www.redhat.com/archives/libvir-list/2016-March/msg00099.html Differences high level: - Use virsecretobj.{c,h} instead of secret_conf.{c,h} - Work in v1 review comments Difference details for those that care to know: Note: Former patch2 has already been pushed. Patch 1: Is combination of v1's patch1 and patch3 w/ adjustments: - Create virsecretobj.c and virsecretobj.h and move sources there. - Add virsecretobj.{c,h} to src/Makefile.am - Use order as suggested by eblake review of patch3 for forward static decls, removed the != NULL from virSecretObjDispose, and add comment regarding why memset() is being done Patch 2: Former patch4 - Move code to virsecretobj.{c,h} instead of secret_conf.{c,h} - Change the virSecretObjFindByUUIDLocked to use the construct "return virObjectRef(virHashLookup(...));" as suggested by eblake Patch3: Former patch5 (split) - Split the format patch5 to have virSecretUsageIDForDef changes in one patch and the remainder in the followup patch. Patch4: Former patch5 (split) - The rest of former patch5 - Removed the extraneous "VIR_FREE(configFile);" as noted by eblake. - NOTE: eblake queried about the condition: if (secret->def->private && !def->private) in virSecretObjListAddLocked. This same check came from the former secretDefineXML in the else of the "secretFindByUUID(new_attrs->uuid)". IOW: Redefining a secret. The code didn't allow a "new" secret to be not private if the currently defined one was private. I left it as is. Patch5->Patch9: Former patch6->patch10: - All that's different is using virsecretobj instead of secret_conf Patch10: Former patch11 - Using virsecretobj instead of secret_conf Patch11: Former patch10 - Adjusted and moved the comment from virSecretObjDeleteConfig to virSecretObjDeleteData - Added a virReportSystemError for the DeleteConfig error Patch12: Former patch13: - All that's different is using virsecretobj instead of secret_conf Patch13: Former patch14 - Delete the extra space in the "return 0" as noted by Cole's review. Patch14: Former patch15: - All that's different is using virsecretobj instead of secret_conf After all is said done, I did a sdiff between the v1 secret_conf.h and v2 virsecretobj.h as well as the v1 secret_conf.c and v2 virsecretobj.c and found only the differences as noted above, plus removed a duplicated virSecretUsageIDForDef prototype I found in secret_conf.h. I also went through the painful process of make check syntax-check at each step of the way and built using Coverity. John Ferlan (14): secret: Create virsecretobj.c and virsecretconf.h secret: Introduce virSecretObjListFindBy{UUID|Usage} support secret: Introduce virSecretUsageIDForDef secret: Introduce virSecretObjListAdd* and virSecretObjListRemove secret: Introduce virSecretObjListNumOfSecrets secret: Introduce virSecretObjListExport secret: Introduce virSecretObjListGetUUIDs secret: Use the hashed virSecretObjList secret: Move and rename secretLoadAllConfigs secret: Introduce virSecretObjDelete{Config|Data} secret: Introduce virSecretObjSave{Config|Data} secret: Introduce virSecretObj{Get|Set}Def secret: Introduce virSecretObjGetValue and virSecretObjGetValueSize secret: Change virSecretDef variable names po/POTFILES.in | 1 + src/Makefile.am | 3 +- src/conf/secret_conf.c | 37 +- src/conf/secret_conf.h | 9 +- src/conf/virsecretobj.c | 1011 +++++++++++++++++++++++++++++++++++++++++ src/conf/virsecretobj.h | 110 +++++ src/libvirt_private.syms | 24 + src/secret/secret_driver.c | 823 ++++----------------------------- src/storage/storage_backend.c | 4 +- 9 files changed, 1279 insertions(+), 743 deletions(-) create mode 100644 src/conf/virsecretobj.c create mode 100644 src/conf/virsecretobj.h