From: Jeff King <peff@peff.net>
To: "Michal Koutný" <mkoutny@suse.com>
Cc: git@vger.kernel.org, "Jean Delvare" <jdelvare@suse.de>,
"Elijah Newren" <newren@gmail.com>,
"Usman Akinyemi" <usmanakinyemi202@gmail.com>,
"Taylor Blau" <me@ttaylorr.com>,
"Junio C Hamano" <gitster@pobox.com>,
"René Scharfe" <l.s.r@web.de>
Subject: Re: [PATCH] merge-ll: Cleanup merge driver temporaries after interrupt
Date: Thu, 10 Sep 2026 12:22:42 -0400 [thread overview]
Message-ID: <20260910162242.GC251185@coredump.intra.peff.net> (raw)
In-Reply-To: <20260910150608.1867930-1-mkoutny@suse.com>
On Thu, Sep 10, 2026 at 05:06:07PM +0200, Michal Koutný wrote:
> When there's a long(er) running merge driver helper, the user may just
> decide to terminate it with Ctrl+C. That sends a signal to the driver
> prog and to the whole process group as well, including the git merge
> command proper. Hence the cleanup code would not run and .merge_file_*
> files are left behind.
>
> Transfer the idiom [1] from editor.c where the (process group) signal
> delivery is approximated from the return code of the child process and
> do the cleanup before going for good.
We have a temporary-file cleanup handler that we install already, which
handles signal propagation, atomicity, etc. It seems like it would be
simpler to just use that.
In the worst case we can just call register_tempfile() on each path, but
I think this code could be taught to use the actual creation. Something
like the patch below (only lightly tested).
---
diff --git a/merge-ll.c b/merge-ll.c
index ef5287dee8..d53f0fe4a6 100644
--- a/merge-ll.c
+++ b/merge-ll.c
@@ -17,6 +17,7 @@
#include "quote.h"
#include "strbuf.h"
#include "gettext.h"
+#include "tempfile.h"
struct ll_merge_driver;
@@ -174,15 +175,30 @@ static struct ll_merge_driver ll_merge_drv[] = {
{ "union", "built-in union merge", ll_union_merge },
};
-static void create_temp(mmfile_t *src, char *path, size_t len)
+static struct tempfile *create_temp(mmfile_t *src)
{
- int fd;
-
- xsnprintf(path, len, ".merge_file_XXXXXX");
- fd = xmkstemp(path);
- if (write_in_full(fd, src->ptr, src->size) < 0)
+ struct tempfile *t = xmks_tempfile(".merge_file_XXXXXX");
+ if (write_in_full(t->fd, src->ptr, src->size) < 0)
die_errno("unable to write temp-file");
- close(fd);
+ close(t->fd);
+ return t;
+}
+
+static const char *get_temp_path(struct tempfile *t)
+{
+ /*
+ * Tempfiles store the absolute path of the file, but
+ * we don't do any quoting against the shell, which
+ * can lead to problems if your path has spaces, etc, in it.
+ * Historically this was OK since we only provided relative
+ * paths which were fairly vanilla.
+ *
+ * We can work around it by going back to the relative path (since we
+ * know we created a tempfile in the cwd via create_temp() above).
+ * In the long run I think we ought to consider providing
+ * the absolute paths but correctly shell-quoting them.
+ */
+ return basename(get_tempfile_path(t));
}
/*
@@ -197,11 +213,11 @@ static enum ll_merge_result ll_ext_merge(const struct ll_merge_driver *fn,
const struct ll_merge_options *opts,
int marker_size)
{
- char temp[3][50];
+ struct tempfile *tmp_o, *tmp_a, *tmp_b;
struct strbuf cmd = STRBUF_INIT;
const char *format = fn->cmdline;
struct child_process child = CHILD_PROCESS_INIT;
- int status, fd, i;
+ int status, fd;
struct stat st;
enum ll_merge_result ret;
assert(opts);
@@ -211,19 +227,19 @@ static enum ll_merge_result ll_ext_merge(const struct ll_merge_driver *fn,
result->ptr = NULL;
result->size = 0;
- create_temp(orig, temp[0], sizeof(temp[0]));
- create_temp(src1, temp[1], sizeof(temp[1]));
- create_temp(src2, temp[2], sizeof(temp[2]));
+ tmp_o = create_temp(orig);
+ tmp_a = create_temp(src1);
+ tmp_b = create_temp(src2);
while (strbuf_expand_step(&cmd, &format)) {
if (skip_prefix(format, "%", &format))
strbuf_addch(&cmd, '%');
else if (skip_prefix(format, "O", &format))
- strbuf_addstr(&cmd, temp[0]);
+ strbuf_addstr(&cmd, get_temp_path(tmp_o));
else if (skip_prefix(format, "A", &format))
- strbuf_addstr(&cmd, temp[1]);
+ strbuf_addstr(&cmd, get_temp_path(tmp_a));
else if (skip_prefix(format, "B", &format))
- strbuf_addstr(&cmd, temp[2]);
+ strbuf_addstr(&cmd, get_temp_path(tmp_b));
else if (skip_prefix(format, "L", &format))
strbuf_addf(&cmd, "%d", marker_size);
else if (skip_prefix(format, "P", &format))
@@ -241,7 +257,8 @@ static enum ll_merge_result ll_ext_merge(const struct ll_merge_driver *fn,
child.use_shell = 1;
strvec_push(&child.args, cmd.buf);
status = run_command(&child);
- fd = open(temp[1], O_RDONLY);
+ /* really feels like we could just use strbuf_read_file() here? */
+ fd = open(get_tempfile_path(tmp_a), O_RDONLY);
if (fd < 0)
goto bad;
if (fstat(fd, &st))
@@ -255,8 +272,9 @@ static enum ll_merge_result ll_ext_merge(const struct ll_merge_driver *fn,
close_bad:
close(fd);
bad:
- for (i = 0; i < 3; i++)
- unlink_or_warn(temp[i]);
+ delete_tempfile(&tmp_o);
+ delete_tempfile(&tmp_a);
+ delete_tempfile(&tmp_b);
strbuf_release(&cmd);
if (!status)
ret = LL_MERGE_OK;
next prev parent reply other threads:[~2026-09-10 16:22 UTC|newest]
Thread overview: 11+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-10 15:06 [PATCH] merge-ll: Cleanup merge driver temporaries after interrupt Michal Koutný
2026-09-10 16:22 ` Jeff King [this message]
2026-09-11 14:43 ` Michal Koutný
2026-09-11 17:10 ` [PATCH v2 0/3] merge-ll: Cleanup merge driver temporaries after Jeff King
2026-09-11 17:11 ` [PATCH v2 1/3] merge-ll: use strbuf to read back external merge result Jeff King
2026-09-11 18:06 ` Elijah Newren
2026-09-11 18:32 ` Junio C Hamano
2026-09-11 17:11 ` [PATCH v2 2/3] merge-ll: catch close() errors when writing external tempfiles Jeff King
2026-09-11 18:06 ` Elijah Newren
2026-09-11 17:13 ` [PATCH v2 3/3] merge-ll: use tempfile API for external driver files Jeff King
2026-09-11 18:10 ` 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=20260910162242.GC251185@coredump.intra.peff.net \
--to=peff@peff.net \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=jdelvare@suse.de \
--cc=l.s.r@web.de \
--cc=me@ttaylorr.com \
--cc=mkoutny@suse.com \
--cc=newren@gmail.com \
--cc=usmanakinyemi202@gmail.com \
/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