Re: [Qemu-devel] [PATCH v5 6/6] block: Enable qemu_open/close to work with fd sets

[Kevin Wolf <kwolf@redhat.com>]

Am 26.07.2012 05:57, schrieb Corey Bryant:
> On 07/25/2012 03:43 PM, Eric Blake wrote:
>> On 07/23/2012 07:08 AM, Corey Bryant wrote:
>>> +int monitor_fdset_get_fd(Monitor *mon, int64_t fdset_id, int flags)
>>> +{
>>> +    mon_fdset_t *mon_fdset;
>>> +    mon_fdset_fd_t *mon_fdset_fd;
>>> +    int mon_fd_flags;
>>> +
>>> +    if (!mon) {
>>> +        errno = ENOENT;
>>> +        return -1;
>>> +    }
>>> +
>>> +    QLIST_FOREACH(mon_fdset, &mon->fdsets, next) {
>>> +        if (mon_fdset->id != fdset_id) {
>>> +            continue;
>>> +        }
>>> +        QLIST_FOREACH(mon_fdset_fd, &mon_fdset->fds, next) {
>>> +            if (mon_fdset_fd->removed) {
>>> +                continue;
>>> +            }
>>> +
>>> +            mon_fd_flags = fcntl(mon_fdset_fd->fd, F_GETFL);
>>> +            if (mon_fd_flags == -1) {
>>> +                return -1;
>>
>> This says we fail on the first fcntl() failure, instead of trying other
>> fds in the set.  Granted, an fcntl() failure is probably the sign of a
>> bigger bug (such as closing an fd at the wrong point in time), so I
>> guess trying to go on doesn't make much sense once we already know we
>> are hosed.
>>
> 
> I think I'll stick with it the way it is.  If fcntl() fails we might 
> have a tainted fd set so I think we should fail.

The alternative would be s/return 1/continue/, right? I think either way
is acceptable.

>>> +            }
>>> +
>>> +            switch (flags & O_ACCMODE) {
>>> +            case O_RDWR:
>>> +                if ((mon_fd_flags & O_ACCMODE) == O_RDWR) {
>>> +                    return mon_fdset_fd->fd;
>>> +                }
>>> +                break;
>>> +            case O_RDONLY:
>>> +                if ((mon_fd_flags & O_ACCMODE) == O_RDONLY) {
>>> +                    return mon_fdset_fd->fd;
>>> +                }
>>> +                break;
>>
>> Do we want to allow the case where the caller asked for O_RDONLY, but
>> the set only has O_RDWR?  After all, the caller is getting a compatible
>> subset of what the set offers.
> 
> I don't see a problem with it.

I would require exact matches like you implemented, in order to prevent
damage if we ever had a bug that writes to a read-only file. I believe
it also makes the semantics clearer and the code simpler, while it
shouldn't make much of a difference for clients.

Kevin