| Message ID | 20231024095820.1068949-1-thaller@redhat.com |
|---|---|
| Headers | show |
| Series | remove xfree() and add free_const()+nft_gmp_free() | expand |
On Tue, Oct 24, 2023 at 11:57:06AM +0200, Thomas Haller wrote: > RESENT of v1. > > Also rebased on top of current `master`, which required minor > adjustments. > > Also minor adjustments to the commit messages. I will put this in the tree this evening, after with the recent fixes I have posted. > Thomas Haller (4): > datatype: don't return a const string from cgroupv2_get_path() > gmputil: add nft_gmp_free() to free strings from mpz_get_str() > all: add free_const() and use it instead of xfree() > all: remove xfree() and use plain free() > > include/gmputil.h | 2 + > include/nft.h | 6 ++ > include/utils.h | 1 - > src/cache.c | 6 +- > src/ct.c | 2 +- > src/datatype.c | 18 ++--- > src/erec.c | 6 +- > src/evaluate.c | 18 ++--- > src/expression.c | 6 +- > src/gmputil.c | 21 +++++- > src/json.c | 2 +- > src/libnftables.c | 24 +++--- > src/meta.c | 4 +- > src/misspell.c | 2 +- > src/mnl.c | 16 ++-- > src/netlink_linearize.c | 4 +- > src/optimize.c | 12 +-- > src/parser_bison.y | 158 ++++++++++++++++++++-------------------- > src/rule.c | 68 ++++++++--------- > src/scanner.l | 6 +- > src/segtree.c | 4 +- > src/statement.c | 4 +- > src/utils.c | 5 -- > src/xt.c | 10 +-- > 24 files changed, 213 insertions(+), 192 deletions(-) > > -- > 2.41.0 >
On Tue, Oct 24, 2023 at 11:57:06AM +0200, Thomas Haller wrote: > RESENT of v1. > > Also rebased on top of current `master`, which required minor > adjustments. > > Also minor adjustments to the commit messages. Applied. To be honest, reading the longish commit descriptions several times, I doubt there is any benefit in this, I might end up myself using free_const() everywhere not to figure out if it is const or not, because I don't really care. But now this series are in master.
On Thu, 2023-11-09 at 17:05 +0100, Pablo Neira Ayuso wrote: > On Tue, Oct 24, 2023 at 11:57:06AM +0200, Thomas Haller wrote: > > RESENT of v1. > > > > Also rebased on top of current `master`, which required minor > > adjustments. > > > > Also minor adjustments to the commit messages. > > Applied. Thanks. > > To be honest, reading the longish commit descriptions several times, > I > doubt there is any benefit in this, I claim, there was no benefit in xfree(). that's why it was dropped in favor of standard free(). The special function xfree() needs a justification, not the other way around. > I might end up myself using > free_const() everywhere not to figure out if it is const or not, > because I don't really care. That seems not a good practice. Const-correctness may help you to catch bugs via unwanted modifications. If constness is unnecessarily cast away, it's looses such hints from the compiler. Thomas
On Thu, Nov 09, 2023 at 06:14:56PM +0100, Thomas Haller wrote: > On Thu, 2023-11-09 at 17:05 +0100, Pablo Neira Ayuso wrote: [...] > > I might end up myself using > > free_const() everywhere not to figure out if it is const or not, > > because I don't really care. > > That seems not a good practice. Const-correctness may help you to catch > bugs via unwanted modifications. If constness is unnecessarily cast > away, it's looses such hints from the compiler. Why should I care if the pointer is const or not if what I need to free it?
On Thu, 2023-11-09 at 20:15 +0100, Pablo Neira Ayuso wrote: > On Thu, Nov 09, 2023 at 06:14:56PM +0100, Thomas Haller wrote: > > On Thu, 2023-11-09 at 17:05 +0100, Pablo Neira Ayuso wrote: > [...] > > > I might end up myself using > > > free_const() everywhere not to figure out if it is const or not, > > > because I don't really care. > > > > That seems not a good practice. Const-correctness may help you to > > catch > > bugs via unwanted modifications. If constness is unnecessarily cast > > away, it's looses such hints from the compiler. > > Why should I care if the pointer is const or not if what I need to > free it? There might be a mistake/bug, and the thing should not actually be freed. We need every help from the compiler we can get (compiler warnings, strong typing/const-correctness, static assertions). Normally, the compiler would help and complain against free() of a const pointer. With xfree()/free_const() used eagerly, it does not help. Thomas