LibTomCrypt bug
From: Tom St Denis (tomstdenis@yahoo.com)
Date: 03/29/03
- Next message: rjh: "Re: MD5 spec's C-code is wrong"
- Previous message: Andrew Swallow: "Re: MD5 spec's C-code is wrong"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
From: "Tom St Denis" <tomstdenis@yahoo.com> Date: Sat, 29 Mar 2003 16:15:04 GMT
There is a minor bug in LibTomCrypt v0.82 and previous with regards to the
ECC multiplier. In the routine I make a copy of the input point so that the
destination of the multiplier can be the input point [e.g. X = kX].
When it allocs a point to store the copy if its NULL [e.g. no mem] my
routine will jump to the end and free up all the memory used so far. It
will then try to free a NULL pointer [and dereference it too].
I've fixed this bug and v0.83 will sport the fix. I don't know of a way to
maliciously cause this bug and as far as I know it would cause a GPF on most
systems if it occured.
On a positive note I've optimized the mulmod function
Tom
- Next message: rjh: "Re: MD5 spec's C-code is wrong"
- Previous message: Andrew Swallow: "Re: MD5 spec's C-code is wrong"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|
|