Re: generic crypto - remarks

2005-10-21 Thread Paul Eggert
Jim Meyering <[EMAIL PROTECTED]> writes: > verify (offsetof (struct S, member_m) % 4 == 0); > > Hmm... that assumes 8-bit bytes. And it also assumes no holes in integer representations. This is more portable: verify (offsetof (struct S, member_m) % alignof (uint32_t) == 0); where alignof is

Re: generic crypto - remarks

2005-10-21 Thread Ben Pfaff
Jim Meyering <[EMAIL PROTECTED]> writes: > Putting a compile-time assertion like the following, > right after the struct definition, adds some clout :-) > > verify (offsetof (struct S, member_m) % 4 == 0); > > Hmm... that assumes 8-bit bytes. You could multiply by CHAR_BIT and check for a multi

Re: generic crypto - remarks

2005-10-21 Thread Jim Meyering
Bruno Haible <[EMAIL PROTECTED]> wrote: > Simon Josefsson wrote: >> > you don't need it, >> > because the rules for struct layout in C guarantee that a structure field >> > is aligned to a multiple of the alignment of the previous field. >> >> Are you saying that even if we don't change the type o

Re: generic crypto - remarks

2005-10-21 Thread Bruno Haible
Simon Josefsson wrote: > > you don't need it, > > because the rules for struct layout in C guarantee that a structure field > > is aligned to a multiple of the alignment of the previous field. > > Are you saying that even if we don't change the type of buffer to > char, we don't need that alignment

Re: arctwo, arctwo-tests, gc-arctwo, gc-arctwo-tests

2005-10-21 Thread Simon Josefsson
Ralf Wildenhues <[EMAIL PROTECTED]> writes: > Hi Simon, >> +#define arctwo_setkey(context,keylen,key) \ >> + arctwo_setkey_ekb (context, keylen, key, 8 * keylen) > > Ouch. `8 * (keylen)' would be safer. Thanks, Ralf. I have installed this. ___ bug-

Re: arctwo, arctwo-tests, gc-arctwo, gc-arctwo-tests

2005-10-21 Thread Ralf Wildenhues
Hi Simon, * Simon Josefsson wrote on Fri, Oct 21, 2005 at 02:02:58PM CEST: > I have installed this. A quick glance showed up the nit below. Cheers, Ralf *snip* > Index: lib/arctwo.h > === > RCS file: lib/arctwo.h > diff -N lib/arct

fix bug in rijndael

2005-10-21 Thread Simon Josefsson
I have installed the patch below. It seems the example Rijndael implementation didn't implement CBC mode correctly... --- rijndael-api-fst.c 14 Oct 2005 15:33:22 +0200 1.1 +++ rijndael-api-fst.c 19 Oct 2005 16:08:10 +0200 @@ -244,7 +244,7 @@ ((uint32_t *) block)[3] = ((uint

Re: generic crypto - remarks

2005-10-21 Thread Simon Josefsson
Bruno Haible <[EMAIL PROTECTED]> writes: > Hi Simon, > > Here are a few random remarks regarding the new code. It's a bit terse, but > I hope you can decipher the meaning. Hi! Many thanks. I have been pushing gc into gnulib rather rapidly because we want to release GnuTLS 1.2.9 with that stuff

des, des-tests, gc-des, gc-des-tests (was: Re: des)

2005-10-21 Thread Simon Josefsson
I have installed this. Index: ChangeLog === RCS file: /cvsroot/gnulib/gnulib/ChangeLog,v retrieving revision 1.433 diff -u -p -r1.433 ChangeLog --- ChangeLog 21 Oct 2005 12:03:17 - 1.433 +++ ChangeLog 21 Oct 2005 12:27:10

generic crypto - remarks

2005-10-21 Thread Bruno Haible
Hi Simon, Here are a few random remarks regarding the new code. It's a bit terse, but I hope you can decipher the meaning. 1) crc.h #if HAVE_INTTYPES_H # include #endif #if HAVE_STDINT_H # include #endif Correct this to #include because - On old platforms with neither nor , these includ

arctwo, arctwo-tests, gc-arctwo, gc-arctwo-tests (was: Re: arctwo)

2005-10-21 Thread Simon Josefsson
I have installed this. Index: ChangeLog === RCS file: /cvsroot/gnulib/gnulib/ChangeLog,v retrieving revision 1.432 diff -u -p -r1.432 ChangeLog --- ChangeLog 19 Oct 2005 15:40:26 - 1.432 +++ ChangeLog 21 Oct 2005 12:02:14

gc-libgcrypt.c fix

2005-10-21 Thread Simon Josefsson
I have installed this; ECB modes didn't work for libgcrypt through gc. 2005-10-21 Simon Josefsson <[EMAIL PROTECTED]> * gc-libgcrypt.c (gc_cipher_open): Handle ECB. Index: gc-libgcrypt.c === RCS file: /cvsroot/gnulib/gnuli