[racket] can custodian-shutdown-all interrupt an #:atomic callback?

It looks to me as though when a custodian-shutdown-all occurs, the custodian actions associated using scheme_add_managed() can be triggered while another thread is in the middle of a callback to an FFI-generated callback that includes the #:atomic declaration.  Is this true? If true, is it only true for some certain set of (avoidable) circumstances?

Apologies in advance for not constructing the experiments necessary to prove by example that it can occur; I believe I've observed this, and I will certainly try to construct a small example if the answer to this question isn't on the tip of anyone's tongue.

John

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4624 bytes
Desc: not available
URL: <http://lists.racket-lang.org/users/archive/attachments/20110909/9fa3ed9c/attachment.p7s>