nisse at lysator.liu.se
Sun Jan 1 18:30:55 UTC 2017
"Marco Bodrato" <bodrato at mail.dm.unipi.it> writes:
> While we are looking at gcdext, I propose a small change to save an
> allocation when t must be computed. Can you review it, to check if the
> size assumptions are safe?
Sorry, I haven't looked too closely at this code...
I guess it still allocates an extra limb for some arguments, which no
longer is required by mpn_gcdext.
> PTR (>mp) = tmp_gp;
> SIZ (>mp) = gsize;
> PTR (&stmp) = tmp_sp;
> SIZ (&stmp) = tmp_ssize;
This style makes my uneasy... I'd prefer to either use real
reallocatable mpz_t for destination operands, or use mpn functions
directly... But that's unrelated to this change.
You did one more thing in the committed change 84ce52f0d360:
--- a/mpz/gcdext.c Tue Dec 13 19:20:31 2016 +0100
+++ b/mpz/gcdext.c Sat Dec 17 21:26:50 2016 +0100
@@ -64,7 +64,7 @@ mpz_gcdext (mpz_ptr g, mpz_ptr s, mpz_pt
if (g != NULL)
- gp = MPZ_REALLOC (g, asize);
+ gp = MPZ_NEWALLOC (g, asize);
MPN_COPY (gp, PTR (a), asize);
SIZ (g) = asize;
That will break in case g == a, right? So we'd need to either keep
MPZ_REALLOC, or check for that special case. So I'm afraid the reuse
test doesn't hit this corner...
Regards, and Happy New Year,
Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.
More information about the gmp-devel