Re: [PATCH] configure: Check for GnuTLS block size function

[Alexandre Julliard]

Michael Cronenworth <mike(a)cchtml.com> writes:
> diff --git a/configure.ac b/configure.ac
> index 7f234b2..6ae8dde 100644
> --- a/configure.ac
> +++ b/configure.ac
> @@ -1268,6 +1268,9 @@ then
>                   WINE_CHECK_LIB_FUNCS(gnutls_hash,[$GNUTLS_LIBS],,
>                                        [WINE_NOTICE([libgnutls ${notice_platform}development files too old, no bcrypt hash support.])])])],
>              [GNUTLS_CFLAGS=""])])
> +    WINE_PACKAGE_FLAGS(GNUTLS,[gnutls],,,,
> +        [WINE_CHECK_LIB_FUNCS(gnutls_cipher_get_block_size,[$GNUTLS_LIBS],,
> +                              [WINE_NOTICE([libgnutls ${notice_platform}development files too old, no block size support.])])])

This won't have any effect, since you don't check the resulting define.

In any case I don't think this needs a configure check, since we already
have a fallback. You can simply declare the function pointer without
using the MAKE_FUNCPTR macro.

-- 
Alexandre Julliard
julliard(a)winehq.org

[PATCH] configure: Check for GnuTLS block size function

[Michael Cronenworth]

On 06/15/2016 10:39 PM, Alexandre Julliard wrote:
> This won't have any effect, since you don't check the resulting define.

Please excuse my ignorance of the WINE_* macros.
> In any case I don't think this needs a configure check, since we already
> have a fallback. You can simply declare the function pointer without
> using the MAKE_FUNCPTR macro.

I thought about adding a version check around the function, but I figured a 
configure check would be cleaner. Is removing the configure check and adding a 
version check (< GnuTLS 2.10) in secur32 acceptable?

[PATCH] configure: Check for GnuTLS block size function

[Alexandre Julliard]

Michael Cronenworth <mike(a)cchtml.com> writes:
> I thought about adding a version check around the function, but I
> figured a configure check would be cleaner. Is removing the configure
> check and adding a version check (< GnuTLS 2.10) in secur32
> acceptable?

No, you shouldn't need that. Something like this should be enough:

diff --git a/dlls/secur32/schannel_gnutls.c b/dlls/secur32/schannel_gnutls.c
index b10b62993326..bcadd47e7aaf 100644
--- a/dlls/secur32/schannel_gnutls.c
+++ b/dlls/secur32/schannel_gnutls.c
@@ -42,7 +42,7 @@ WINE_DEFAULT_DEBUG_CHANNEL(secur32);
 WINE_DECLARE_DEBUG_CHANNEL(winediag);
 
 /* Not present in gnutls version < 2.9.10. */
-extern int gnutls_cipher_get_block_size(gnutls_cipher_algorithm_t algorithm);
+static int (*pgnutls_cipher_get_block_size)(gnutls_cipher_algorithm_t algorithm);
 
 static void *libgnutls_handle;
 #define MAKE_FUNCPTR(f) static typeof(f) * p##f
@@ -52,7 +52,6 @@ MAKE_FUNCPTR(gnutls_certificate_allocate_credentials);
 MAKE_FUNCPTR(gnutls_certificate_free_credentials);
 MAKE_FUNCPTR(gnutls_certificate_get_peers);
 MAKE_FUNCPTR(gnutls_cipher_get);
-MAKE_FUNCPTR(gnutls_cipher_get_block_size);
 MAKE_FUNCPTR(gnutls_cipher_get_key_size);
 MAKE_FUNCPTR(gnutls_credentials_set);
 MAKE_FUNCPTR(gnutls_deinit);


-- 
Alexandre Julliard
julliard(a)winehq.org

[PATCH] configure: Check for GnuTLS block size function

[Michael Cronenworth]

On 06/16/2016 12:11 AM, Alexandre Julliard wrote:
> No, you shouldn't need that. Something like this should be enough:

OK. New patch submitted.