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

Re: [libvirt] [PATCH 08/10] Introduce new APIs for spawning processes



On 11/18/2010 02:55 AM, Daniel P. Berrange wrote:
> On Wed, Nov 17, 2010 at 09:29:00PM -0700, Eric Blake wrote:
>> From: Daniel P. Berrange <berrange redhat com>
>>
>> This introduces a new set of APIs in src/util/command.h
>> to use for invoking commands. This is intended to replace
>> all current usage of virRun and virExec variants, with a
>> more flexible and less error prone API.
>>
> 
> 
> My code forgot to ever close() the fds in cmd->preserve. We definitely
> need todo it in virCommandFree(), but there's a small argument to say
> we should also do it in virCommandRun/virCommandRunAsync so that if
> the caller keeps the virCommandPtr alive for a long time, we don't
> have the open FDs.

I'll look into this more.

> 
> It would also be useful to have a generic API for logging info about
> the command to an FD (to let us remove that logging code from UML
> and QEMU & LXC drivers).
> 
> eg
> 
> +void virCommandWriteArgLog(virCommandPtr cmd, int logfd)

Would this be called before virCommandRun and do the logging immediately
at that point (or, if something has gone wrong such as allocation
failure, then nothing is logged, and the user will eventually learn that
nothing was logged when virCommandRun fails), or is it something where
you call it once to register the logfd, then calling virCommandRun is
what actually writes to the registered fd at the time the child is
actually spawned?

-- 
Eric Blake   eblake redhat com    +1-801-349-2682
Libvirt virtualization library http://libvirt.org

Attachment: signature.asc
Description: OpenPGP digital signature


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