random: Admit that we will never add error checking everywhere

[secnet.git] / random.c

diff --git a/random.c b/random.c
index cdb0dfd9ce751cf1d5a7b7c6cb158e679ae672d3..323fffd692e4677732edef90fe5d5a4106261872 100644 (file)
--- a/random.c
+++ b/random.c
@@ -33,7 +33,7 @@ struct rgen_data {
 };
 
 static random_fn random_generate;
-static bool_t random_generate(void *data, int32_t bytes, uint8_t *buff)
+static void random_generate(void *data, int32_t bytes, uint8_t *buff)
 {
     struct rgen_data *st=data;
     int r;
@@ -41,26 +41,10 @@ static bool_t random_generate(void *data, int32_t bytes, uint8_t *buff)
     r= read(st->fd,buff,bytes);
 
     assert(r == bytes);
-    /* This is totally crap error checking, but AFAICT many callers of
-     * this function do not check the return value.  This is a minimal
-     * change to make the code not fail silently-but-insecurely.
-     *
-     * A proper fix requires either:
-     *  - Declare all random number generation failures as fatal
-     *    errors, and make this return void, and fix all callers,
-     *    and make this call some appropriate function if it fails.
-     *  - Make this have proper error checking (and reporting!)
-     *    and make all callers check the error (and report!);
-     *    this will be tricky, I think, because you have to report
-     *    the errno somewhere.
-     *
-     * There's also the issue that this is only one possible
-     * implementation of a random number source; others may not rely
-     * on reading from a file descriptor, and may not produce
-     * appropriate settings of errno.
+    /* This is totally crap error checking, but callers of
+     * this function do not check the return value and dealing
+     * with failure of this everywhere would be very inconvenient.
      */
-
-    return True;
 }
 
 static list_t *random_apply(closure_t *self, struct cloc loc,