Ticket #308 (closed defect: wontfix)

Opened 4 years ago

Last modified 4 years ago

possible race condition on static buffer in HMAC-SHA1

Reported by: neil Owned by: jaymccarthy
Priority: minor Milestone:
Component: jaymccarthy/hmac-sha1.plt Keywords:
Cc: Version: (1 1)
Racket Version: 4.2.5

Description

Looks like the way that the "HMAC" C function is used in this code -- using a static buffer on the C side for the result -- presents a race condition, in that some code in another Scheme thread could conceivably stomp on the buffer in between the C call being made through FFI and the result being copied out of the static buffer on the Scheme side.

I thought of a mutex lock in the Scheme side, but two potential problems occurred to me: (1) there still might be other concurrent calls into the shared library from Scheme code that don't use this particular FFI binding; (2) C code in native threads might also inadvisably use the static buffer.

Perhaps passing in a fresh buffer each time is the best solution?

I am actually using this library now, so please advise whether you expect to get a chance to look at this in the next few days or I should clone the code and develop my own fix. Thank you.

 http://www.openssl.org/docs/crypto/hmac.html

Change History

Changed 4 years ago by neil

  • priority changed from critical to minor
  • status changed from new to closed
  • resolution set to wontfix

Jay informs me that a newer version of this code is in core, as part of collects/web-server/stuffers/hmac-sha1, and that the PLaneT package is considered deprecated.

Note: See TracTickets for help on using tickets.