This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Problem in pthread_key_create

On Jan  9 18:31, Johnny Willemsen wrote:
> Hi,
> Posted this last Friday but didn't got anything back. I think this is a bug
> in the pthread library of Cygwin. Can someone have a look at this?
> Johnny 
> -----Original Message-----
> From: Johnny Willemsen [] 
> Sent: vrijdag 7 januari 2005 14:57
> To: ''
> Subject: Question about pthread_key_create
> Hi all,
> A question, I had a look at the implementation of pthread_key_create. When
> an invalid key is passed, a EBUSY is returned. This looks very strange to
> me, isn't it better to return EINVAL just as the pthread_key_delete does?

No, that's not a bug.  Please read the SUSv3 description for
pthread_key_create() here:

Please note especially the chapter

  RATIONALE/Non-Idempotent Data Key Creation.

It should sufficiently describe why returning EBUSY in this case isn't
such a bad idea.  Unfortunately, the SUSv3 definition left the actual
behaviour open to the implementation.  Another implementation could also
reset the data key to NULL in subsequent calls to pthread_key_create().
Either way, the bottom line is, don't call pthread_key_create() more
than once for the same data key.


Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Project Co-Leader
Red Hat, Inc.

Unsubscribe info:
Problem reports:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]