From: Phillip Wood <phillip.wood123@gmail.com>
To: Jeff King <peff@peff.net>
Cc: Git Mailing List <git@vger.kernel.org>,
Phillip Wood <phillip.wood@dunelm.org.uk>,
Junio C Hamano <gitster@pobox.com>
Subject: Re: [PATCH 7/6] xdiff: drop unused mmfile parameters from xdl_do_patience_diff()
Date: Fri, 26 Aug 2022 14:15:20 +0100 [thread overview]
Message-ID: <CAPoeCOZFua+VvwBPix2M-8sRhfhymFCpejOPbCERtCp7fZ9FWQ@mail.gmail.com> (raw)
In-Reply-To: <YwCO+WE2Z+sJofEb@coredump.intra.peff.net>
Hi Peff
On Sat, 20 Aug 2022 at 08:36, Jeff King <peff@peff.net> wrote:
>
> On Sat, Aug 20, 2022 at 02:58:44AM -0400, Jeff King wrote:
>
> > > Looking at the xpatience.c I think we can remove the mmfile_t
> > > parameters there as well, they are only end up being used because
> > > patience_diff() gets called recursively. I'm about to go off list for
> > > a week, but I can look at putting a patch together for that when I get
> > > back unless you want to.
> >
> > I wondered that, too, but they also get passed in to fill_hashmap(),
> > which records the pointers in its "struct hashmap". It looks like those
> > fields are never accessed, though, and could even be removed from the
> > struct entirely!
> >
> > So I think there is some room for cleanup.
>
> And here's a patch, so we don't forget about it. It could be applied
> independently of this series, but there's a small textual dependency
> since we're touching a line close to the related histogram cleanup.
>
> Thanks for mentioning this. I had given only a cursory glance before,
> and just assumed it would be hard to trace. It turned out to be fun. :)
Yes, it took me a couple of minutes to realize that it wasn't doing
anything useful with the pointers in "struct hashmap" but it was quite
satisfying when I did. The patch looks good, thanks for putting it
together.
Best Wishes
Phillip
> Junio: this could go on top of the jk/unused-fixes topic.
>
> -- >8 --
> Subject: xdiff: drop unused mmfile parameters from xdl_do_patience_diff()
>
> The entry point to the patience-diff algorithm takes two mmfile_t
> structs with the original file contents, but it doesn't actually do
> anything useful with them. This is similar to the case recently cleaned
> up in the histogram code via f1d019071e (xdiff: drop unused mmfile
> parameters from xdl_do_histogram_diff(), 2022-08-19), but there's a bit
> more subtlety going on.
>
> We pass them into the recursive patience_diff(), which in turn passes
> them into fill_hashmap(), which stuffs the pointers into a struct. But
> the only thing which reads the struct fields is our recursion into
> patience_diff()!
>
> So it's unlikely that something like -Wunused-parameter could find this
> case: it would have to detect the circular dependency caused by the
> recursion (not to mention tracing across struct field assignments).
>
> But once found, it's easy to have the compiler confirm what's going on:
>
> 1. Drop the "file1" and "file2" fields from the hashmap struct
> definition. Remove the assignments in fill_hashmap(), and
> temporarily substitute NULL in the recursive call to
> patience_diff(). Compiling shows that no other code touched those
> fields.
>
> 2. Now fill_hashmap() will trigger -Wunused-parameter. Drop "file1"
> and "file2" from its definition and callsite.
>
> 3. Now patience_diff() will trigger -Wunused-parameter. Drop them
> there, too. One of the callsites is the recursion with our
> NULL values, so those temporary values go away.
>
> 4. Now xdl_do_patience_diff() will trigger -Wunused-parameter. Drop
> them there. And we're done.
>
> Suggested-by: Phillip Wood <phillip.wood@dunelm.org.uk>
> Signed-off-by: Jeff King <peff@peff.net>
> ---
> Explaining the 4 steps is probably overkill, but I found it a little
> interesting how one can bend the compiler to their will here. At least I
> didn't make it a series of 4 patches. ;)
>
> xdiff/xdiffi.c | 2 +-
> xdiff/xdiffi.h | 3 +--
> xdiff/xpatience.c | 23 +++++++----------------
> 3 files changed, 9 insertions(+), 19 deletions(-)
>
> diff --git a/xdiff/xdiffi.c b/xdiff/xdiffi.c
> index 8c64519eac..32652ded2d 100644
> --- a/xdiff/xdiffi.c
> +++ b/xdiff/xdiffi.c
> @@ -321,7 +321,7 @@ int xdl_do_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
> return -1;
>
> if (XDF_DIFF_ALG(xpp->flags) == XDF_PATIENCE_DIFF) {
> - res = xdl_do_patience_diff(mf1, mf2, xpp, xe);
> + res = xdl_do_patience_diff(xpp, xe);
> goto out;
> }
>
> diff --git a/xdiff/xdiffi.h b/xdiff/xdiffi.h
> index 9d988e0263..126c9d8ff4 100644
> --- a/xdiff/xdiffi.h
> +++ b/xdiff/xdiffi.h
> @@ -56,8 +56,7 @@ int xdl_build_script(xdfenv_t *xe, xdchange_t **xscr);
> void xdl_free_script(xdchange_t *xscr);
> int xdl_emit_diff(xdfenv_t *xe, xdchange_t *xscr, xdemitcb_t *ecb,
> xdemitconf_t const *xecfg);
> -int xdl_do_patience_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp,
> - xdfenv_t *env);
> +int xdl_do_patience_diff(xpparam_t const *xpp, xdfenv_t *env);
> int xdl_do_histogram_diff(xpparam_t const *xpp, xdfenv_t *env);
>
> #endif /* #if !defined(XDIFFI_H) */
> diff --git a/xdiff/xpatience.c b/xdiff/xpatience.c
> index fe39c2978c..a2d8955537 100644
> --- a/xdiff/xpatience.c
> +++ b/xdiff/xpatience.c
> @@ -69,7 +69,6 @@ struct hashmap {
> } *entries, *first, *last;
> /* were common records found? */
> unsigned long has_matches;
> - mmfile_t *file1, *file2;
> xdfenv_t *env;
> xpparam_t const *xpp;
> };
> @@ -139,13 +138,10 @@ static void insert_record(xpparam_t const *xpp, int line, struct hashmap *map,
> *
> * It is assumed that env has been prepared using xdl_prepare().
> */
> -static int fill_hashmap(mmfile_t *file1, mmfile_t *file2,
> - xpparam_t const *xpp, xdfenv_t *env,
> +static int fill_hashmap(xpparam_t const *xpp, xdfenv_t *env,
> struct hashmap *result,
> int line1, int count1, int line2, int count2)
> {
> - result->file1 = file1;
> - result->file2 = file2;
> result->xpp = xpp;
> result->env = env;
>
> @@ -254,8 +250,7 @@ static int match(struct hashmap *map, int line1, int line2)
> return record1->ha == record2->ha;
> }
>
> -static int patience_diff(mmfile_t *file1, mmfile_t *file2,
> - xpparam_t const *xpp, xdfenv_t *env,
> +static int patience_diff(xpparam_t const *xpp, xdfenv_t *env,
> int line1, int count1, int line2, int count2);
>
> static int walk_common_sequence(struct hashmap *map, struct entry *first,
> @@ -286,8 +281,7 @@ static int walk_common_sequence(struct hashmap *map, struct entry *first,
>
> /* Recurse */
> if (next1 > line1 || next2 > line2) {
> - if (patience_diff(map->file1, map->file2,
> - map->xpp, map->env,
> + if (patience_diff(map->xpp, map->env,
> line1, next1 - line1,
> line2, next2 - line2))
> return -1;
> @@ -326,8 +320,7 @@ static int fall_back_to_classic_diff(struct hashmap *map,
> *
> * This function assumes that env was prepared with xdl_prepare_env().
> */
> -static int patience_diff(mmfile_t *file1, mmfile_t *file2,
> - xpparam_t const *xpp, xdfenv_t *env,
> +static int patience_diff(xpparam_t const *xpp, xdfenv_t *env,
> int line1, int count1, int line2, int count2)
> {
> struct hashmap map;
> @@ -346,7 +339,7 @@ static int patience_diff(mmfile_t *file1, mmfile_t *file2,
> }
>
> memset(&map, 0, sizeof(map));
> - if (fill_hashmap(file1, file2, xpp, env, &map,
> + if (fill_hashmap(xpp, env, &map,
> line1, count1, line2, count2))
> return -1;
>
> @@ -374,9 +367,7 @@ static int patience_diff(mmfile_t *file1, mmfile_t *file2,
> return result;
> }
>
> -int xdl_do_patience_diff(mmfile_t *file1, mmfile_t *file2,
> - xpparam_t const *xpp, xdfenv_t *env)
> +int xdl_do_patience_diff(xpparam_t const *xpp, xdfenv_t *env)
> {
> - return patience_diff(file1, file2, xpp, env,
> - 1, env->xdf1.nrec, 1, env->xdf2.nrec);
> + return patience_diff(xpp, env, 1, env->xdf1.nrec, 1, env->xdf2.nrec);
> }
> --
> 2.37.2.931.g60e3983abc
>
next prev parent reply other threads:[~2022-08-26 13:15 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-19 8:48 [PATCH 0/6] unused function parameter potpourri Jeff King
2022-08-19 8:49 ` [PATCH 1/6] xdiff: drop unused mmfile parameters from xdl_do_histogram_diff() Jeff King
2022-08-19 13:32 ` Phillip Wood
2022-08-20 6:58 ` Jeff King
2022-08-20 7:36 ` [PATCH 7/6] xdiff: drop unused mmfile parameters from xdl_do_patience_diff() Jeff King
2022-08-26 13:15 ` Phillip Wood [this message]
2022-08-19 8:50 ` [PATCH 2/6] log-tree: drop unused commit param in remerge_diff() Jeff King
2022-08-19 23:05 ` Elijah Newren
2022-08-20 7:04 ` Jeff King
2022-08-19 8:50 ` [PATCH 3/6] match_pathname(): drop unused "flags" parameter Jeff King
2022-08-19 8:51 ` [PATCH 4/6] verify_one_sparse(): drop unused repository parameter Jeff King
2022-08-19 19:04 ` Derrick Stolee
2022-08-20 7:01 ` Jeff King
2022-08-20 8:48 ` René Scharfe
2022-08-20 9:02 ` [PATCH v2 " Jeff King
2022-08-19 8:54 ` [PATCH 5/6] reftable: drop unused parameter from reader_seek_linear() Jeff King
2022-08-19 19:06 ` Derrick Stolee
2022-08-22 7:46 ` Han-Wen Nienhuys
2022-08-19 8:55 ` [PATCH 6/6] reflog: assert PARSE_OPT_NONEG in parse-options callbacks Jeff King
2022-08-19 19:08 ` [PATCH 0/6] unused function parameter potpourri Derrick Stolee
2022-08-19 23:07 ` Elijah Newren
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAPoeCOZFua+VvwBPix2M-8sRhfhymFCpejOPbCERtCp7fZ9FWQ@mail.gmail.com \
--to=phillip.wood123@gmail.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=peff@peff.net \
--cc=phillip.wood@dunelm.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).