[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH 0/5] Make virDomainObjListFindByName run in O(1)



On 2015/4/24 1:14, Michal Privoznik wrote:

> As discussed here:
> 
> https://www.redhat.com/archives/libvir-list/2015-April/msg01135.html
> 
> Michal Privoznik (5):
>   virDomainObjListAddLocked: s/false/NULL/ for @oldDef
>   virDomainObjListNew: Use virObjectFreeHashData
>   Introduce virDomainObjEndAPI
>   virDomainObjListFindByName: Return referenced object
>   virDomainObjList: Introduce yet another hash table
> 
>  src/bhyve/bhyve_driver.c         |  3 +-
>  src/conf/domain_conf.c           | 83 +++++++++++++++++++++++------------
>  src/conf/domain_conf.h           |  2 +
>  src/libvirt_private.syms         |  1 +
>  src/libxl/libxl_driver.c         | 10 ++---
>  src/lxc/lxc_driver.c             |  3 +-
>  src/openvz/openvz_driver.c       | 11 +++--
>  src/parallels/parallels_driver.c |  3 +-
>  src/qemu/qemu_domain.c           |  7 +--
>  src/qemu/qemu_driver.c           | 14 ++----
>  src/test/test_driver.c           | 93 +++++++++++++---------------------------
>  src/uml/uml_driver.c             | 15 +++----
>  12 files changed, 110 insertions(+), 135 deletions(-)
> 


BTW, we just dealt with virDomainObjListFindByName() here, without caring about virDomainObjListFindByID().
virDomainObjListFindByID() is now called by umlDomainDestroyFlags() and umlDomainShutdownFlags(),
if we simultaneously shutdown multiple vms on hypervisor UML, then similar problem comes out: 
some guest maybe unable to shutdown immediately, until other guests get shutoff, and even 'virsh list' blocks.

so, shall we:
1 don't take such problem as a problem
2 use virDomainObjListFindByName() instead of virDomainObjListFindByID() there.
3 add a third hash table for domid

If we use virDomainObjListFindByName() there, and forbids developers from using **ByID(), it seems unacceptable for
any developers.
If we add a third hash table, it may become not easy to maintain the high-complexity codes. Is there a better solution other
than adding more hash tables?


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]