Git development
 help / color / mirror / Atom feed
* Git crashed on `git apply --check`
@ 2026-10-17 15:17 Gingrspacecadet
  2026-10-24  6:10 ` Tanga Joris
  2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
  0 siblings, 2 replies; 23+ messages in thread
From: Gingrspacecadet @ 2026-10-17 15:17 UTC (permalink / raw)
  To: git

Thank you for filling out a Git bug report!
Please answer the following questions to help us understand your issue.

What did you do before the bug happened? (Steps to reproduce your issue)
Checking the following patch:
diff --git a/tests/constraint_type_signature_lifting.test
b/tests/constraint_type_signature_lifting.test

new file mode 100644
index 0000000..0000000
--- /dev/null
+++ b/tests/constraint_type_signature_lifting.test
@@ -0,0 +1,40 @@
+name: constraint types in signatures are lifted to generic parameters
+expect: pass
+stop: run
+errors: 0
+
+[source]
+constraint Readable = {
+    fn uint64 Readable.read(Readable *self);
+}
+
+type File = struct {
+    uint64 size;
+};
+
+fn uint64 File.read(File *self) {
+    return self.size;
+}
+
+fn Readable identity(Readable value) {
+    return value;
+}
+
+fn uint64 read_size(Readable file) {
+    return file.read();
+}
+
+fn int64 main() {
+    File file = {42};
+    File copy = identity(file);
+    if (copy.size != 42) return 1;
+    if (read_size(file) != 42) return 2;
+    return 0;
+}
+
+[stdout]
+
+[stderr]
+
+[exit]
+0

What did you expect to happen?
The patch check to pass

What happened instead?
"git: apply.c:3856: check_preimage: Assertion `patch->is_new <= 0' failed.i"

[System Info]
git version:
git version 2.54.0
cpu: x86_64
built from commit: 94f057755b7941b321fd11fec1b2e3ca5313a4e0
sizeof-long: 8
sizeof-size_t: 8
shell-path: /bin/sh
rust: enabled
gettext: enabled
libcurl: 8.19.0
OpenSSL: OpenSSL 3.6.2 7 Apr 2026
zlib-ng: 2.3.3
SHA-1: SHA1_DC
SHA-256: SHA256_BLK
default-ref-format: files
default-hash: sha1
uname: Linux 7.0.10-arch1-1 #1 SMP PREEMPT_DYNAMIC Sat, 23 May 2026
14:21:20 +0000 x86_64
compiler info: gnuc: 15.2
libc info: glibc: 2.43
$SHELL (typically, interactive shell): /usr/bin/bash


[Enabled Hooks]

^ permalink raw reply related	[flat|nested] 23+ messages in thread

* Re: Git crashed on `git apply --check`
  2026-10-17 15:17 Git crashed on `git apply --check` Gingrspacecadet
@ 2026-10-24  6:10 ` Tanga Joris
       [not found]   ` <CALdwsC11JHUBBTt3e_RFixLYSF0VU1C+mG7bQsO0vV=a68XHNw@mail.gmail.com>
  2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
  1 sibling, 1 reply; 23+ messages in thread
From: Tanga Joris @ 2026-10-24  6:10 UTC (permalink / raw)
  To: jackrawlings2011; +Cc: git

Hi,

Thanks for the report. I was unable to reproduce it with the patch as
quoted (the "diff --git" line is wrapped in your mail), but I can
trigger the same assertion on current master, and on 2.43, when the
"--- /dev/null" line comes before "new file mode" in the header:

    diff --git a/n.txt b/n.txt
    --- /dev/null
    +++ b/n.txt
    new file mode 100644
    @@ -0,0 +1 @@
    +hello

Was your patch handwritten or generated by a tool, and could you
please check whether its header lines are in that order? If you can,
please attach the exact file so that we can confirm it is the same
problem.

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: Git crashed on `git apply --check`
       [not found]   ` <CALdwsC11JHUBBTt3e_RFixLYSF0VU1C+mG7bQsO0vV=a68XHNw@mail.gmail.com>
@ 2026-10-25 22:56     ` Tanga Joris
  0 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-10-25 22:56 UTC (permalink / raw)
  To: Gingrspacecadet, git

Adding the list back

On Sat, 24 Oct 2026 at 20:08, Gingrspacecadet
<jackrawlings2011@gmail.com> wrote:
>
> hi,
>
> I've attached the patch that faulted.
>
> It was generated by an LLM.

Hello,

Thank you for sending the original patch file. The crash was caused
by a blank line after the diff --git line in the new-file section,
and it is reproducible on master. I am currently working on a fix.

Best,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-10-17 15:17 Git crashed on `git apply --check` Gingrspacecadet
  2026-10-24  6:10 ` Tanga Joris
@ 2026-11-07 19:54 ` Tanga Joris
  2026-11-07 19:54   ` [Outreachy PATCH 2/2] apply: reject new and deleted files that have both names Tanga Joris
                     ` (2 more replies)
  1 sibling, 3 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-07 19:54 UTC (permalink / raw)
  To: git
  Cc: gitster, jackrawlings2011, christian.couder, siddharthasthana31,
	kaartic.sivaraam, usmanakinyemi202, Tanga Joris

When a "diff --git" line is followed by a line that is not a git
header line, such as a blank line or an indented "new file mode"
line, parse_git_diff_header() stops right after the "diff --git"
line. find_header() then treats it as an empty git header and keeps
looking, as it has done since 206de27efa (git-apply: ignore empty git
headers, 2005-06-12). It finds the "--- /dev/null" and "+++ b/<path>"
lines and reads the rest as a traditional patch.

But parse_git_diff_header() has already set old_name and new_name
from the "diff --git" line. These names stay in the patch.
parse_traditional_patch() sets is_new for "--- /dev/null", but it
does not clear old_name. So "git apply --check" dies on the assertion
that 5c47f4c6e7 (builtin-apply: accept patch to an empty file,
2008-05-17) added to check_preimage():

  check_preimage: Assertion `patch->is_new <= 0' failed.

The stale new_name causes the same crash when a deletion with such a
header is applied with "git apply -R". The stale names also break an
empty "diff --git" section that is followed by another section. They
do not match the names in the next section, so "git apply" fails with
a "bad git-diff" error.

Free the names when the "diff --git" line is not used. A new file is
then created, and a deletion can be reversed. An empty section is
ignored, as 206de27efa intended. As with any traditional patch, the
"new file mode" line is not used, so a new file gets the default mode.

parse_git_diff_header() has also set is_toplevel_relative. Keep it,
so that the paths stay relative to the top of the repository, as
before.

Reported-by: Gingrspacecadet <jackrawlings2011@gmail.com>
Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
This fixes the crash reported in [1]. While looking into it, I found
two more ways to trigger the same assertion: an indented "new file
mode" line, which this patch also fixes, and a "---" line before
"new file mode", which 2/2 handles. Git 2.43 crashes on all three.

I based this on maint (v2.56.0), since the crash is in released
versions. It merges cleanly with master, next and seen.

[1] https://lore.kernel.org/git/CALdwsC1cMKZgRPZA+pBaJe4PPs9q9W0K9CQLmM2bgUCyysTB-w@mail.gmail.com/

 apply.c                    | 10 ++++-
 t/t4133-apply-filenames.sh | 90 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 99 insertions(+), 1 deletion(-)

diff --git a/apply.c b/apply.c
index f00b7ba4d3..b56b363963 100644
--- a/apply.c
+++ b/apply.c
@@ -1651,8 +1651,16 @@ static int find_header(struct apply_state *state,
 								size, patch);
 			if (git_hdr_len < 0)
 				return -128;
-			if (git_hdr_len <= len)
+			if (git_hdr_len <= len) {
+				/*
+				 * Not a git header after all. Do not keep the
+				 * names it took from the "diff --git" line.
+				 */
+				FREE_AND_NULL(patch->def_name);
+				FREE_AND_NULL(patch->old_name);
+				FREE_AND_NULL(patch->new_name);
 				continue;
+			}
 			*hdrsize = git_hdr_len;
 			return offset;
 		}
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index 3cab1038cf..727c36e7ea 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -60,4 +60,94 @@ test_expect_success 'apply diff with old filename missing from headers' '
 	test_grep "lacks filename information" err
 '
 
+test_expect_success 'apply new file with a blank line after "diff --git"' '
+	test_when_finished "rm -f f" &&
+	cat >blank_line.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	git apply --check blank_line.diff &&
+	git apply blank_line.diff &&
+	echo 1 >expect &&
+	test_cmp expect f
+'
+
+test_expect_success 'apply new file with an indented "new file mode" line' '
+	test_when_finished "rm -f f" &&
+	cat >indented_mode.diff <<-\EOF &&
+	diff --git a/f b/f
+	 new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	git apply --check indented_mode.diff &&
+	git apply indented_mode.diff &&
+	echo 1 >expect &&
+	test_cmp expect f
+'
+
+test_expect_success 'reverse a deletion with a blank line after "diff --git"' '
+	test_when_finished "rm -f f" &&
+	cat >blank_line_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	deleted file mode 100644
+	index d00491f..0000000
+	--- a/f
+	+++ /dev/null
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	git apply -R --check blank_line_delete.diff &&
+	git apply -R blank_line_delete.diff &&
+	echo 1 >expect &&
+	test_cmp expect f
+'
+
+test_expect_success 'ignore an empty "diff --git" section' '
+	test_when_finished "rm -f g" &&
+	echo 1 >g &&
+	cat >empty_section.diff <<-\EOF &&
+	diff --git a/f b/f
+	diff --git a/g b/g
+	index d00491f..0cfbf08 100644
+	--- a/g
+	+++ b/g
+	@@ -1 +1 @@
+	-1
+	+2
+	EOF
+	git apply --check empty_section.diff &&
+	git apply empty_section.diff &&
+	echo 2 >expect &&
+	test_cmp expect g
+'
+
+test_expect_success 'blank line after "diff --git" keeps paths relative to the top' '
+	test_when_finished "rm -rf sub" &&
+	mkdir sub &&
+	cat >sub/blank_line.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	git -C sub apply blank_line.diff &&
+	test_path_is_missing sub/f &&
+	test_path_is_missing f
+'
+
 test_done

base-commit: a018953688f1b10bddf91bff8747068f5f4746a4
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* [Outreachy PATCH 2/2] apply: reject new and deleted files that have both names
  2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
@ 2026-11-07 19:54   ` Tanga Joris
  2026-11-09 10:52   ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Phillip Wood
  2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
  2 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-07 19:54 UTC (permalink / raw)
  To: git
  Cc: gitster, jackrawlings2011, christian.couder, siddharthasthana31,
	kaartic.sivaraam, usmanakinyemi202, Tanga Joris

Git writes "new file mode" or "deleted file mode" before the "---"
and "+++" lines. The "new file mode" line sets is_new. Then
gitdiff_oldname() leaves old_name empty for "--- /dev/null".

When the "---" line comes first, is_new is still 0. So
gitdiff_verify_name() takes the name on that line as old_name. For
"--- /dev/null" this gives "dev/null", because one leading directory
is stripped. The "new file mode" line then sets is_new. The result is
a new file that also has an old name. "git apply --check" dies on it,
on the same assertion in check_preimage(). In the same way,
"+++ /dev/null" before "deleted file mode" gives a deleted file with
a new name. "git apply -R" dies on that one.

d70e9c5c8c (apply: check git diffs for mutually exclusive header
lines, 2017-06-27) made apply report other conflicting header lines,
instead of leaving them to an assertion. Do the same here. After the
header is read, reject a new file that has an old name. Reject a
deleted file that has a new name too.

Patches made by Git are therefore not affected. A patch with
"+++ /dev/null" before "deleted file mode" used to apply without -R.
It is now rejected too.

Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
I chose to reject these headers, like d70e9c5c8c does for other
conflicting lines. Another option would be to accept "--- /dev/null"
before "new file mode" as a new file, since "/dev/null" means no file
there. I'm happy to change it if that is preferred.

 apply.c                    | 20 ++++++++++++++++++++
 t/t4133-apply-filenames.sh | 31 +++++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+)

diff --git a/apply.c b/apply.c
index b56b363963..76cdacff33 100644
--- a/apply.c
+++ b/apply.c
@@ -1466,6 +1466,26 @@ int parse_git_diff_header(struct strbuf *root,
 			      *linenr);
 		return -128;
 	}
+	if (patch->is_new > 0 && patch->old_name) {
+		if (patch_input_file)
+			error(_("git diff header for new file %s names an old file %s at %s:%d"),
+			      patch->new_name, patch->old_name,
+			      patch_input_file, *linenr);
+		else
+			error(_("git diff header for new file %s names an old file %s (line %d)"),
+			      patch->new_name, patch->old_name, *linenr);
+		return -128;
+	}
+	if (patch->is_delete > 0 && patch->new_name) {
+		if (patch_input_file)
+			error(_("git diff header for deleted file %s names a new file %s at %s:%d"),
+			      patch->old_name, patch->new_name,
+			      patch_input_file, *linenr);
+		else
+			error(_("git diff header for deleted file %s names a new file %s (line %d)"),
+			      patch->old_name, patch->new_name, *linenr);
+		return -128;
+	}
 	patch->is_toplevel_relative = 1;
 	return offset;
 }
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index 727c36e7ea..ac9090d903 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -150,4 +150,35 @@ test_expect_success 'blank line after "diff --git" keeps paths relative to the t
 	test_path_is_missing f
 '
 
+for old in /dev/null a/f
+do
+	test_expect_success "reject \"--- $old\" before \"new file mode\"" '
+		cat >old_before_new.diff <<-EOF &&
+		diff --git a/f b/f
+		--- $old
+		+++ b/f
+		new file mode 100644
+		@@ -0,0 +1 @@
+		+1
+		EOF
+		test_must_fail git apply --check old_before_new.diff 2>err &&
+		test_grep "header for new file f names an old file" err
+	'
+done
+
+test_expect_success 'reject "+++ /dev/null" before "deleted file mode"' '
+	cat >new_before_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+	--- a/f
+	+++ /dev/null
+	deleted file mode 100644
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	test_must_fail git apply --check new_before_delete.diff 2>err &&
+	test_grep "header for deleted file f names a new file" err &&
+	test_must_fail git apply --check -R new_before_delete.diff 2>err &&
+	test_grep "header for deleted file f names a new file" err
+'
+
 test_done
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
  2026-11-07 19:54   ` [Outreachy PATCH 2/2] apply: reject new and deleted files that have both names Tanga Joris
@ 2026-11-09 10:52   ` Phillip Wood
  2026-11-09 13:13     ` Tanga Joris
  2026-11-09 19:56     ` Junio C Hamano
  2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
  2 siblings, 2 replies; 23+ messages in thread
From: Phillip Wood @ 2026-11-09 10:52 UTC (permalink / raw)
  To: Tanga Joris, git
  Cc: gitster, jackrawlings2011, christian.couder, siddharthasthana31,
	kaartic.sivaraam, usmanakinyemi202

Hi Tanga

On 07/11/2026 19:54, Tanga Joris wrote:
> When a "diff --git" line is followed by a line that is not a git
> header line, such as a blank line or an indented "new file mode"
> line,

These sound like corrupt patches which we should be rejecting with a 
better error message, rather than applying.

> +test_expect_success 'apply new file with a blank line after "diff --git"' '
> +	test_when_finished "rm -f f" &&
> +	cat >blank_line.diff <<-\EOF &&
> +	diff --git a/f b/f
> +
> +	new file mode 100644
> +	index 0000000..d00491f
> +	--- /dev/null
> +	+++ b/f
> +	@@ -0,0 +1 @@
> +	+1
> +	EOF

To me this looks like a corrupt patch because we have a blank line after 
the "diff --git" header followed by valid extended diff headers. Now 
that we accept this do we respect the file mode?

> +test_expect_success 'apply new file with an indented "new file mode" line' '
> +	test_when_finished "rm -f f" &&
> +	cat >indented_mode.diff <<-\EOF &&
> +	diff --git a/f b/f
> +	 new file mode 100644

This looks like a corrupt patch as well where a space has been 
accidentally introduced at the start of the line.

> +test_expect_success 'ignore an empty "diff --git" section' '
> +	test_when_finished "rm -f g" &&
> +	echo 1 >g &&
> +	cat >empty_section.diff <<-\EOF &&
> +	diff --git a/f b/f
> +	diff --git a/g b/g
> +	index d00491f..0cfbf08 100644
> +	--- a/g
> +	+++ b/g
> +	@@ -1 +1 @@
> +	-1
> +	+2
> +	EOF

This is the only one that I think it is reasonable to accept - we have 
an empty patch for "f" followed by a valid patch for "g"

Thanks

Phillip


^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 10:52   ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Phillip Wood
@ 2026-11-09 13:13     ` Tanga Joris
  2026-11-09 14:31       ` Phillip Wood
  2026-11-09 19:56     ` Junio C Hamano
  1 sibling, 1 reply; 23+ messages in thread
From: Tanga Joris @ 2026-11-09 13:13 UTC (permalink / raw)
  To: phillip.wood
  Cc: git, gitster, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Hey Phillip,

Thank you for reviewing my patch.

On Mon, 9 Nov 2026 at 13:52, Phillip Wood <phillip.wood123@gmail.com> wrote:
> These sound like corrupt patches which we should be rejecting with a
> better error message, rather than applying.

I now see and agree that accepting invalid patches was the wrong
call. Giving a clearer error seems to be a better direction to take.

> To me this looks like a corrupt patch because we have a blank line after
> the "diff --git" header followed by valid extended diff headers. Now
> that we accept this do we respect the file mode?

No, it does not respect the file mode. "new file mode 100755" is
ignored and the file gets the default mode.

> This looks like a corrupt patch as well where a space has been
> accidentally introduced at the start of the line.

Yes, I will treat this one the same way as the blank line.

> This is the only one that I think it is reasonable to accept - we have
> an empty patch for "f" followed by a valid patch for "g"

Alright, I will keep accepting this one.

For the next version, I plan to die with an error when a "diff --git"
line has no header lines after it, and a "---" and "+++" patch follows
before the next "diff --git" line. The error would name the line of
the "diff --git". An empty section followed by another "diff --git"
line would still be ignored.

One thing I would like to check with you. On master, a blank line
after "diff --git" is accepted in a patch that modifies a file. If
that patch has "old mode" and "new mode" lines, the mode change is
lost. The rule above would reject these patches too. What is your view
on this?

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 13:13     ` Tanga Joris
@ 2026-11-09 14:31       ` Phillip Wood
  2026-11-09 20:54         ` Tanga Joris
  0 siblings, 1 reply; 23+ messages in thread
From: Phillip Wood @ 2026-11-09 14:31 UTC (permalink / raw)
  To: Tanga Joris, phillip.wood
  Cc: git, gitster, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Hi Tanga

On 09/11/2026 13:13, Tanga Joris wrote:
> On Mon, 9 Nov 2026 at 13:52, Phillip Wood <phillip.wood123@gmail.com> wrote:

It would have been helpful to retain a bit more context here, I've 
restored it so anyone reading this could see the patch that I was 
referring to.

>>> diff --git a/f b/f
>>> 
>>> new file mode 100644
>>> index 0000000..d00491f
>>> --- /dev/null
>>> +++ b/f
>>> @@ -0,0 +1 @@
>>> +1 

>> To me this looks like a corrupt patch because we have a blank line after
>> the "diff --git" header followed by valid extended diff headers. Now
>> that we accept this do we respect the file mode?
> 
> No, it does not respect the file mode. "new file mode 100755" is
> ignored and the file gets the default mode.

So that is a good reason to reject this, rather than silently applying a 
corrupt patch.

 >>> diff --git a/f b/f
 >>> diff --git a/g b/g

>> This is the only one that I think it is reasonable to accept - we have
>> an empty patch for "f" followed by a valid patch for "g"
> 
> Alright, I will keep accepting this one.

Do we accept that now, or only with your patch?

> For the next version, I plan to die with an error when a "diff --git"
> line has no header lines after it, and a "---" and "+++" patch follows
> before the next "diff --git" line. The error would name the line of
> the "diff --git". An empty section followed by another "diff --git"
> line would still be ignored.
> 
> One thing I would like to check with you. On master, a blank line
> after "diff --git" is accepted in a patch that modifies a file. If
> that patch has "old mode" and "new mode" lines, the mode change is
> lost. The rule above would reject these patches too. What is your view
> on this?
Oh, so

diff --git a/f b/f

index 573541a..d00491f 100644
--- a/f
+++ b/f
@@ -1 +1 @@
-0
+1

applies. That looks like a corrupt patch as well to me and if we're 
ignoring the mode headers I'd say it is definitely a bug and rejecting 
it is the right thing to do.

Thanks

Phillip

^ permalink raw reply related	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 10:52   ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Phillip Wood
  2026-11-09 13:13     ` Tanga Joris
@ 2026-11-09 19:56     ` Junio C Hamano
  1 sibling, 0 replies; 23+ messages in thread
From: Junio C Hamano @ 2026-11-09 19:56 UTC (permalink / raw)
  To: Phillip Wood
  Cc: Tanga Joris, git, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Phillip Wood <phillip.wood123@gmail.com> writes:

>> +	cat >empty_section.diff <<-\EOF &&
>> +	diff --git a/f b/f
>> +	diff --git a/g b/g
>> +	index d00491f..0cfbf08 100644
>> +	--- a/g
>> +	+++ b/g
>> +	@@ -1 +1 @@
>> +	-1
>> +	+2
>> +	EOF
>
> This is the only one that I think it is reasonable to accept - we have 
> an empty patch for "f" followed by a valid patch for "g"

We shouldn't accept anything that "git format-patch" (or "git diff")
would not produce, I would have to say.

Thanks.

^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 14:31       ` Phillip Wood
@ 2026-11-09 20:54         ` Tanga Joris
  2026-11-09 21:06           ` Junio C Hamano
  0 siblings, 1 reply; 23+ messages in thread
From: Tanga Joris @ 2026-11-09 20:54 UTC (permalink / raw)
  To: phillip.wood
  Cc: git, gitster, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Hey Phillip,

Thanks, and sorry for trimming too much context last time.

On Mon, 9 Nov 2026 at 17:31, Phillip Wood <phillip.wood123@gmail.com> wrote:
>  >>> diff --git a/f b/f
>  >>> diff --git a/g b/g
>
> >> This is the only one that I think it is reasonable to accept - we have
> >> an empty patch for "f" followed by a valid patch for "g"
> >
> > Alright, I will keep accepting this one.
>
> Do we accept that now, or only with your patch?

Only with my patch. On master your example fails with "error: git
apply: bad git-diff - inconsistent old filename", because the names
from the empty "diff --git" line are kept. A new or deleted file
after it fails in a similar way. It does work on master when the next
section names the same file, or is a rename.

Junio wrote in reply to your first mail:

> We shouldn't accept anything that "git format-patch" (or "git diff")
> would not produce, I would have to say.

So I plan to drop this part and leave this case as it is on master.
The next version will only turn the corrupt patches into an error.

One thing I noticed while checking this. "git format-patch" can
produce a "diff --git" line with nothing after it, when a line in the
body of the commit message looks like "diff --git a/x b/x". On
master, "git apply" fails on that patch with the same "inconsistent
old filename" error, unless the line names the same file as the real
patch. This is not new, so I would leave it out of this series,
unless you or Junio think otherwise.

> Oh, so
>
> diff --git a/f b/f
>
> index 573541a..d00491f 100644
> --- a/f
> +++ b/f
> @@ -1 +1 @@
> -0
> +1
>
> applies. That looks like a corrupt patch as well to me and if we're
> ignoring the mode headers I'd say it is definitely a bug and rejecting
> it is the right thing to do.

Yes, it applies on master. Alright, I will reject this one too in the
next version, along with the blank line and the indented "new file
mode" cases.

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 20:54         ` Tanga Joris
@ 2026-11-09 21:06           ` Junio C Hamano
  2026-11-09 23:03             ` Tanga Joris
  0 siblings, 1 reply; 23+ messages in thread
From: Junio C Hamano @ 2026-11-09 21:06 UTC (permalink / raw)
  To: Tanga Joris
  Cc: phillip.wood, git, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Tanga Joris <tangajoris@gmail.com> writes:

> One thing I noticed while checking this. "git format-patch" can
> produce a "diff --git" line with nothing after it, when a line in the
> body of the commit message looks like "diff --git a/x b/x".

A garbage in the commit log message does not count as coming from
"git diff".  It may take some effort to teach "git am" to avoid
feeding such a false "diff --git" line down to "git apply", but an
important thing to keep in mind is that it is NOT "git apply"'s job
to worry about such a garbage input.


^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line
  2026-11-09 21:06           ` Junio C Hamano
@ 2026-11-09 23:03             ` Tanga Joris
  0 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-09 23:03 UTC (permalink / raw)
  To: Junio C Hamano
  Cc: phillip.wood, git, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202

Hello Junio,

On Tue, 10 Nov 2026 at 00:06, Junio C Hamano <gitster@pobox.com> wrote:
> A garbage in the commit log message does not count as coming from
> "git diff".  It may take some effort to teach "git am" to avoid
> feeding such a false "diff --git" line down to "git apply", but an
> important thing to keep in mind is that it is NOT "git apply"'s job
> to worry about such a garbage input.

That makes sense. I had thought of it as something "git format-patch"
produces, but it is only the log message passing through. I will
leave it out, and the next version will only reject the corrupt
headers.

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line
  2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
  2026-11-07 19:54   ` [Outreachy PATCH 2/2] apply: reject new and deleted files that have both names Tanga Joris
  2026-11-09 10:52   ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Phillip Wood
@ 2026-11-14  1:20   ` Tanga Joris
  2026-11-14  1:20     ` [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names Tanga Joris
                       ` (2 more replies)
  2 siblings, 3 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-14  1:20 UTC (permalink / raw)
  To: git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202,
	Tanga Joris

When a "diff --git" line is followed by a line that is not a git
header line, such as a blank line or an indented "new file mode"
line, parse_git_diff_header() stops right after the "diff --git"
line. find_header() then treats it as an empty git header and keeps
looking, as it has done since 206de27efa (git-apply: ignore empty git
headers, 2005-06-12). It finds the "--- /dev/null" and "+++ b/<path>"
lines and reads the rest as a traditional patch.

But parse_git_diff_header() has already set old_name and new_name
from the "diff --git" line. These names stay in the patch.
parse_traditional_patch() sets is_new for "--- /dev/null", but it
does not clear old_name. So "git apply --check" dies on the assertion
that 5c47f4c6e7 (builtin-apply: accept patch to an empty file,
2008-05-17) added to check_preimage():

  check_preimage: Assertion `patch->is_new <= 0' failed.

The stale new_name causes the same crash when a deletion with such a
header is applied with "git apply -R". When such a patch modifies a
file, it applies. But its "old mode" and "new mode" lines are
ignored.

These are corrupt patches. "git diff" does not produce them. So
report the empty header as an error, instead of reading the patch
that follows it. This also rejects a "diff --git" line that is
followed by other text and then a non-git patch, which used to apply.
A "diff --git" line followed by another "diff --git" line is handled
as before, because 206de27efa ignores empty headers on purpose. Only
a patch that is read under an empty header is a problem.

Reported-by: Gingrspacecadet <jackrawlings2011@gmail.com>
Helped-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
Changes since v1:
 - 1/2 no longer reads the patch that follows an empty "diff --git"
   line. It reports the empty header as an error instead, as Phillip
   and Junio suggested:

     error: corrupt patch: empty "diff --git" header at <file>:<line>

 - A blank line after "diff --git" in a patch that modifies a file is
   now rejected too, as Phillip suggested. v1 and master apply it
   without its "old mode" and "new mode" lines.
 - One case that works on master is now rejected: a "diff --git" line
   followed by other text and then a non-git patch.
 - Dropped the test for an empty "diff --git" section followed by
   another section. That case behaves as on master, but the test fails
   under SANITIZE=leak, because of a leak in git_header_name() that is
   already on master. Also dropped the subdirectory test, which only
   made sense while the patch was applied.
 - 2/2: the code is unchanged. Its commit message now says why patches
   made by Git are not affected.

 apply.c                    | 13 +++++-
 t/t4133-apply-filenames.sh | 86 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 98 insertions(+), 1 deletion(-)

diff --git a/apply.c b/apply.c
index f00b7ba4d3..c5d4fe7887 100644
--- a/apply.c
+++ b/apply.c
@@ -1604,6 +1604,7 @@ static int find_header(struct apply_state *state,
 		       struct patch *patch)
 {
 	unsigned long offset, len;
+	int empty_git_hdr_linenr = 0;
 
 	patch->is_toplevel_relative = 0;
 	patch->is_rename = patch->is_copy = 0;
@@ -1644,6 +1645,7 @@ static int find_header(struct apply_state *state,
 		 * or mode change, so we handle that specially
 		 */
 		if (!memcmp("diff --git ", line, 11)) {
+			int linenr = state->linenr;
 			int git_hdr_len = parse_git_diff_header(&state->root,
 								state->patch_input_file,
 								&state->linenr,
@@ -1651,8 +1653,10 @@ static int find_header(struct apply_state *state,
 								size, patch);
 			if (git_hdr_len < 0)
 				return -128;
-			if (git_hdr_len <= len)
+			if (git_hdr_len <= len) {
+				empty_git_hdr_linenr = linenr;
 				continue;
+			}
 			*hdrsize = git_hdr_len;
 			return offset;
 		}
@@ -1670,6 +1674,13 @@ static int find_header(struct apply_state *state,
 		if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
 			continue;
 
+		/* Git does not write a patch under an empty git header */
+		if (empty_git_hdr_linenr) {
+			error(_("corrupt patch: empty \"diff --git\" header at %s:%d"),
+			      state->patch_input_file, empty_git_hdr_linenr);
+			return -128;
+		}
+
 		/* Ok, we'll consider it a patch */
 		if (parse_traditional_patch(state, line, line+len, patch))
 			return -128;
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index 3cab1038cf..751a01f92f 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -60,4 +60,90 @@ test_expect_success 'apply diff with old filename missing from headers' '
 	test_grep "lacks filename information" err
 '
 
+reject_empty_git_header () {
+	file=$1 &&
+	line=$2 &&
+	shift 2 &&
+	test_must_fail git apply --check "$@" "$file" 2>err &&
+	test_grep "corrupt patch: empty \"diff --git\" header at $file:$line" err
+}
+
+test_expect_success 'reject a new file with a blank line after "diff --git"' '
+	cat >blank_line.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header blank_line.diff 1 &&
+	test_must_fail git apply blank_line.diff &&
+	test_path_is_missing f
+'
+
+test_expect_success 'reject an indented "new file mode" line' '
+	cat >indented_mode.diff <<-\EOF &&
+	diff --git a/f b/f
+	 new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header indented_mode.diff 1
+'
+
+test_expect_success 'reject a deletion with a blank line after "diff --git"' '
+	cat >blank_line_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	deleted file mode 100644
+	index d00491f..0000000
+	--- a/f
+	+++ /dev/null
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	reject_empty_git_header blank_line_delete.diff 1 &&
+	reject_empty_git_header blank_line_delete.diff 1 -R
+'
+
+test_expect_success 'reject a modification with a blank line after "diff --git"' '
+	test_when_finished "rm -f g" &&
+	echo 1 >g &&
+	cat >blank_line_modify.diff <<-\EOF &&
+	diff --git a/g b/g
+
+	old mode 100644
+	new mode 100755
+	index d00491f..0cfbf08
+	--- a/g
+	+++ b/g
+	@@ -1 +1 @@
+	-1
+	+2
+	EOF
+	reject_empty_git_header blank_line_modify.diff 1 &&
+	echo 1 >expect &&
+	test_cmp expect g
+'
+
+test_expect_success 'name the empty "diff --git" line in the error' '
+	cat >later_line.diff <<-\EOF &&
+	some text before the patch
+
+	diff --git a/f b/f
+	some text after it
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header later_line.diff 3
+'
+
 test_done

base-commit: a018953688f1b10bddf91bff8747068f5f4746a4
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names
  2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
@ 2026-11-14  1:20     ` Tanga Joris
  2026-11-15 12:47       ` Siddharth Asthana
  2026-11-15 12:39     ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line Siddharth Asthana
  2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
  2 siblings, 1 reply; 23+ messages in thread
From: Tanga Joris @ 2026-11-14  1:20 UTC (permalink / raw)
  To: git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	siddharthasthana31, kaartic.sivaraam, usmanakinyemi202,
	Tanga Joris

Git writes "new file mode" or "deleted file mode" before the "---"
and "+++" lines. The "new file mode" line sets is_new. Then
gitdiff_oldname() leaves old_name empty for "--- /dev/null".

When the "---" line comes first, is_new is still 0. So
gitdiff_verify_name() takes the name on that line as old_name. For
"--- /dev/null" this gives "dev/null", because one leading directory
is stripped. The "new file mode" line then sets is_new. The result is
a new file that also has an old name. "git apply --check" dies on it,
on the same assertion in check_preimage(). In the same way,
"+++ /dev/null" before "deleted file mode" gives a deleted file with
a new name. "git apply -R" dies on that one.

d70e9c5c8c (apply: check git diffs for mutually exclusive header
lines, 2017-06-27) made apply report other conflicting header lines,
instead of leaving them to an assertion. Do the same here. After the
header is read, reject a new file that has an old name. Reject a
deleted file that has a new name too.

Git always writes the mode line first, so patches made by Git are not
affected. A patch with
"+++ /dev/null" before "deleted file mode" used to apply without -R.
It is now rejected too.

Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
 apply.c                    | 20 ++++++++++++++++++++
 t/t4133-apply-filenames.sh | 31 +++++++++++++++++++++++++++++++
 2 files changed, 51 insertions(+)

diff --git a/apply.c b/apply.c
index c5d4fe7887..0fa6a15a65 100644
--- a/apply.c
+++ b/apply.c
@@ -1466,6 +1466,26 @@ int parse_git_diff_header(struct strbuf *root,
 			      *linenr);
 		return -128;
 	}
+	if (patch->is_new > 0 && patch->old_name) {
+		if (patch_input_file)
+			error(_("git diff header for new file %s names an old file %s at %s:%d"),
+			      patch->new_name, patch->old_name,
+			      patch_input_file, *linenr);
+		else
+			error(_("git diff header for new file %s names an old file %s (line %d)"),
+			      patch->new_name, patch->old_name, *linenr);
+		return -128;
+	}
+	if (patch->is_delete > 0 && patch->new_name) {
+		if (patch_input_file)
+			error(_("git diff header for deleted file %s names a new file %s at %s:%d"),
+			      patch->old_name, patch->new_name,
+			      patch_input_file, *linenr);
+		else
+			error(_("git diff header for deleted file %s names a new file %s (line %d)"),
+			      patch->old_name, patch->new_name, *linenr);
+		return -128;
+	}
 	patch->is_toplevel_relative = 1;
 	return offset;
 }
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index 751a01f92f..ff48e68da0 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -146,4 +146,35 @@ test_expect_success 'name the empty "diff --git" line in the error' '
 	reject_empty_git_header later_line.diff 3
 '
 
+for old in /dev/null a/f
+do
+	test_expect_success "reject \"--- $old\" before \"new file mode\"" '
+		cat >old_before_new.diff <<-EOF &&
+		diff --git a/f b/f
+		--- $old
+		+++ b/f
+		new file mode 100644
+		@@ -0,0 +1 @@
+		+1
+		EOF
+		test_must_fail git apply --check old_before_new.diff 2>err &&
+		test_grep "header for new file f names an old file" err
+	'
+done
+
+test_expect_success 'reject "+++ /dev/null" before "deleted file mode"' '
+	cat >new_before_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+	--- a/f
+	+++ /dev/null
+	deleted file mode 100644
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	test_must_fail git apply --check new_before_delete.diff 2>err &&
+	test_grep "header for deleted file f names a new file" err &&
+	test_must_fail git apply --check -R new_before_delete.diff 2>err &&
+	test_grep "header for deleted file f names a new file" err
+'
+
 test_done
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line
  2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
  2026-11-14  1:20     ` [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names Tanga Joris
@ 2026-11-15 12:39     ` Siddharth Asthana
  2026-11-16 17:54       ` Tanga Joris
  2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
  2 siblings, 1 reply; 23+ messages in thread
From: Siddharth Asthana @ 2026-11-15 12:39 UTC (permalink / raw)
  To: Tanga Joris, git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	kaartic.sivaraam, usmanakinyemi202



On 14/11/26 06:50, Tanga Joris wrote:
> When a "diff --git" line is followed by a line that is not a git
> header line, such as a blank line or an indented "new file mode"
> line, parse_git_diff_header() stops right after the "diff --git"
> line. find_header() then treats it as an empty git header and keeps
> looking, as it has done since 206de27efa (git-apply: ignore empty git
> headers, 2005-06-12). It finds the "--- /dev/null" and "+++ b/<path>"
> lines and reads the rest as a traditional patch.
> 
> But parse_git_diff_header() has already set old_name and new_name
> from the "diff --git" line. These names stay in the patch.
> parse_traditional_patch() sets is_new for "--- /dev/null", but it
> does not clear old_name. So "git apply --check" dies on the assertion
> that 5c47f4c6e7 (builtin-apply: accept patch to an empty file,
> 2008-05-17) added to check_preimage():
> 
>    check_preimage: Assertion `patch->is_new <= 0' failed.
> 
> The stale new_name causes the same crash when a deletion with such a
> header is applied with "git apply -R". When such a patch modifies a
> file, it applies. But its "old mode" and "new mode" lines are
> ignored.
> 
> These are corrupt patches. "git diff" does not produce them. So
> report the empty header as an error, instead of reading the patch
> that follows it. This also rejects a "diff --git" line that is
> followed by other text and then a non-git patch, which used to apply.
> A "diff --git" line followed by another "diff --git" line is handled
> as before, because 206de27efa ignores empty headers on purpose. Only
> a patch that is read under an empty header is a problem.
> 
> Reported-by: Gingrspacecadet <jackrawlings2011@gmail.com>
> Helped-by: Phillip Wood <phillip.wood@dunelm.org.uk>
> Signed-off-by: Tanga Joris <tangajoris@gmail.com>
> ---
> Changes since v1:
>   - 1/2 no longer reads the patch that follows an empty "diff --git"
>     line. It reports the empty header as an error instead, as Phillip
>     and Junio suggested:
> 
>       error: corrupt patch: empty "diff --git" header at <file>:<line>
> 
>   - A blank line after "diff --git" in a patch that modifies a file is
>     now rejected too, as Phillip suggested. v1 and master apply it
>     without its "old mode" and "new mode" lines.
>   - One case that works on master is now rejected: a "diff --git" line
>     followed by other text and then a non-git patch.
>   - Dropped the test for an empty "diff --git" section followed by
>     another section. That case behaves as on master, but the test fails
>     under SANITIZE=leak, because of a leak in git_header_name() that is
>     already on master. Also dropped the subdirectory test, which only
>     made sense while the patch was applied.


The leak is actually in parse_git_diff_header(), we overwrite 
patch->def_name without freeing the old value, so it leaks when the 
function runs again for the next "diff --git". A free() before the 
assignment fixes it, and both callers zero the patch first so it's safe. 
Can't we do that in a prep patch and keep the test?

Nit: the notes under "---" describe both patches, so a cover letter with 
"--cover-letter --range-diff=<v2>" would be nicer for v3.

Thanks,
Siddharth


>   - 2/2: the code is unchanged. Its commit message now says why patches
>     made by Git are not affected.
> 
>   apply.c                    | 13 +++++-
>   t/t4133-apply-filenames.sh | 86 ++++++++++++++++++++++++++++++++++++++
>   2 files changed, 98 insertions(+), 1 deletion(-)
> 
> diff --git a/apply.c b/apply.c
> index f00b7ba4d3..c5d4fe7887 100644
> --- a/apply.c
> +++ b/apply.c
> @@ -1604,6 +1604,7 @@ static int find_header(struct apply_state *state,
>   		       struct patch *patch)
>   {
>   	unsigned long offset, len;
> +	int empty_git_hdr_linenr = 0;
>   
>   	patch->is_toplevel_relative = 0;
>   	patch->is_rename = patch->is_copy = 0;
> @@ -1644,6 +1645,7 @@ static int find_header(struct apply_state *state,
>   		 * or mode change, so we handle that specially
>   		 */
>   		if (!memcmp("diff --git ", line, 11)) {
> +			int linenr = state->linenr;
>   			int git_hdr_len = parse_git_diff_header(&state->root,
>   								state->patch_input_file,
>   								&state->linenr,
> @@ -1651,8 +1653,10 @@ static int find_header(struct apply_state *state,
>   								size, patch);
>   			if (git_hdr_len < 0)
>   				return -128;
> -			if (git_hdr_len <= len)
> +			if (git_hdr_len <= len) {
> +				empty_git_hdr_linenr = linenr;
>   				continue;
> +			}
>   			*hdrsize = git_hdr_len;
>   			return offset;
>   		}
> @@ -1670,6 +1674,13 @@ static int find_header(struct apply_state *state,
>   		if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
>   			continue;
>   
> +		/* Git does not write a patch under an empty git header */
> +		if (empty_git_hdr_linenr) {
> +			error(_("corrupt patch: empty \"diff --git\" header at %s:%d"),
> +			      state->patch_input_file, empty_git_hdr_linenr);
> +			return -128;
> +		}
> +
>   		/* Ok, we'll consider it a patch */
>   		if (parse_traditional_patch(state, line, line+len, patch))
>   			return -128;
> diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
> index 3cab1038cf..751a01f92f 100755
> --- a/t/t4133-apply-filenames.sh
> +++ b/t/t4133-apply-filenames.sh
> @@ -60,4 +60,90 @@ test_expect_success 'apply diff with old filename missing from headers' '
>   	test_grep "lacks filename information" err
>   '
>   
> +reject_empty_git_header () {
> +	file=$1 &&
> +	line=$2 &&
> +	shift 2 &&
> +	test_must_fail git apply --check "$@" "$file" 2>err &&
> +	test_grep "corrupt patch: empty \"diff --git\" header at $file:$line" err
> +}
> +
> +test_expect_success 'reject a new file with a blank line after "diff --git"' '
> +	cat >blank_line.diff <<-\EOF &&
> +	diff --git a/f b/f
> +
> +	new file mode 100644
> +	index 0000000..d00491f
> +	--- /dev/null
> +	+++ b/f
> +	@@ -0,0 +1 @@
> +	+1
> +	EOF
> +	reject_empty_git_header blank_line.diff 1 &&
> +	test_must_fail git apply blank_line.diff &&
> +	test_path_is_missing f
> +'
> +
> +test_expect_success 'reject an indented "new file mode" line' '
> +	cat >indented_mode.diff <<-\EOF &&
> +	diff --git a/f b/f
> +	 new file mode 100644
> +	index 0000000..d00491f
> +	--- /dev/null
> +	+++ b/f
> +	@@ -0,0 +1 @@
> +	+1
> +	EOF
> +	reject_empty_git_header indented_mode.diff 1
> +'
> +
> +test_expect_success 'reject a deletion with a blank line after "diff --git"' '
> +	cat >blank_line_delete.diff <<-\EOF &&
> +	diff --git a/f b/f
> +
> +	deleted file mode 100644
> +	index d00491f..0000000
> +	--- a/f
> +	+++ /dev/null
> +	@@ -1 +0,0 @@
> +	-1
> +	EOF
> +	reject_empty_git_header blank_line_delete.diff 1 &&
> +	reject_empty_git_header blank_line_delete.diff 1 -R
> +'
> +
> +test_expect_success 'reject a modification with a blank line after "diff --git"' '
> +	test_when_finished "rm -f g" &&
> +	echo 1 >g &&
> +	cat >blank_line_modify.diff <<-\EOF &&
> +	diff --git a/g b/g
> +
> +	old mode 100644
> +	new mode 100755
> +	index d00491f..0cfbf08
> +	--- a/g
> +	+++ b/g
> +	@@ -1 +1 @@
> +	-1
> +	+2
> +	EOF
> +	reject_empty_git_header blank_line_modify.diff 1 &&
> +	echo 1 >expect &&
> +	test_cmp expect g
> +'
> +
> +test_expect_success 'name the empty "diff --git" line in the error' '
> +	cat >later_line.diff <<-\EOF &&
> +	some text before the patch
> +
> +	diff --git a/f b/f
> +	some text after it
> +	--- /dev/null
> +	+++ b/f
> +	@@ -0,0 +1 @@
> +	+1
> +	EOF
> +	reject_empty_git_header later_line.diff 3
> +'
> +
>   test_done
> 
> base-commit: a018953688f1b10bddf91bff8747068f5f4746a4


^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names
  2026-11-14  1:20     ` [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names Tanga Joris
@ 2026-11-15 12:47       ` Siddharth Asthana
  2026-11-16 18:06         ` Tanga Joris
  0 siblings, 1 reply; 23+ messages in thread
From: Siddharth Asthana @ 2026-11-15 12:47 UTC (permalink / raw)
  To: Tanga Joris, git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	kaartic.sivaraam, usmanakinyemi202



On 14/11/26 06:50, Tanga Joris wrote:
> Git writes "new file mode" or "deleted file mode" before the "---"
> and "+++" lines. The "new file mode" line sets is_new. Then
> gitdiff_oldname() leaves old_name empty for "--- /dev/null".
> 
> When the "---" line comes first, is_new is still 0. So
> gitdiff_verify_name() takes the name on that line as old_name. For
> "--- /dev/null" this gives "dev/null", because one leading directory
> is stripped. The "new file mode" line then sets is_new. The result is
> a new file that also has an old name. "git apply --check" dies on it,
> on the same assertion in check_preimage(). In the same way,
> "+++ /dev/null" before "deleted file mode" gives a deleted file with
> a new name. "git apply -R" dies on that one.
> 
> d70e9c5c8c (apply: check git diffs for mutually exclusive header
> lines, 2017-06-27) made apply report other conflicting header lines,
> instead of leaving them to an assertion. Do the same here. After the
> header is read, reject a new file that has an old name. Reject a
> deleted file that has a new name too.
> 
> Git always writes the mode line first, so patches made by Git are not
> affected. A patch with
> "+++ /dev/null" before "deleted file mode" used to apply without -R.
> It is now rejected too.
> 
> Signed-off-by: Tanga Joris <tangajoris@gmail.com>
> ---
>   apply.c                    | 20 ++++++++++++++++++++
>   t/t4133-apply-filenames.sh | 31 +++++++++++++++++++++++++++++++
>   2 files changed, 51 insertions(+)
> 
> diff --git a/apply.c b/apply.c
> index c5d4fe7887..0fa6a15a65 100644
> --- a/apply.c
> +++ b/apply.c
> @@ -1466,6 +1466,26 @@ int parse_git_diff_header(struct strbuf *root,
>   			      *linenr);
>   		return -128;
>   	}
> +	if (patch->is_new > 0 && patch->old_name) {
> +		if (patch_input_file)
> +			error(_("git diff header for new file %s names an old file %s at %s:%d"),
> +			      patch->new_name, patch->old_name,


Nit: for "--- /dev/null" this prints "names an old file dev/null", but 
the users never write "dev/null", that's what's left after one leading 
component is stripped. And "--- a/f" gives "names an old file f". The 
only way to reach here is a "---" before "new file mode", so can't we 
name that instead of printing the old name? Same for the deleted case 
with "+++" before "deleted file mode".

Thanks,
Siddharth


> +			      patch_input_file, *linenr);
> +		else
> +			error(_("git diff header for new file %s names an old file %s (line %d)"),
> +			      patch->new_name, patch->old_name, *linenr);
> +		return -128;
> +	}
> +	if (patch->is_delete > 0 && patch->new_name) {
> +		if (patch_input_file)
> +			error(_("git diff header for deleted file %s names a new file %s at %s:%d"),
> +			      patch->old_name, patch->new_name,
> +			      patch_input_file, *linenr);
> +		else
> +			error(_("git diff header for deleted file %s names a new file %s (line %d)"),
> +			      patch->old_name, patch->new_name, *linenr);
> +		return -128;
> +	}
>   	patch->is_toplevel_relative = 1;
>   	return offset;
>   }
> diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
> index 751a01f92f..ff48e68da0 100755
> --- a/t/t4133-apply-filenames.sh
> +++ b/t/t4133-apply-filenames.sh
> @@ -146,4 +146,35 @@ test_expect_success 'name the empty "diff --git" line in the error' '
>   	reject_empty_git_header later_line.diff 3
>   '
>   
> +for old in /dev/null a/f
> +do
> +	test_expect_success "reject \"--- $old\" before \"new file mode\"" '
> +		cat >old_before_new.diff <<-EOF &&
> +		diff --git a/f b/f
> +		--- $old
> +		+++ b/f
> +		new file mode 100644
> +		@@ -0,0 +1 @@
> +		+1
> +		EOF
> +		test_must_fail git apply --check old_before_new.diff 2>err &&
> +		test_grep "header for new file f names an old file" err
> +	'
> +done
> +
> +test_expect_success 'reject "+++ /dev/null" before "deleted file mode"' '
> +	cat >new_before_delete.diff <<-\EOF &&
> +	diff --git a/f b/f
> +	--- a/f
> +	+++ /dev/null
> +	deleted file mode 100644
> +	@@ -1 +0,0 @@
> +	-1
> +	EOF
> +	test_must_fail git apply --check new_before_delete.diff 2>err &&
> +	test_grep "header for deleted file f names a new file" err &&
> +	test_must_fail git apply --check -R new_before_delete.diff 2>err &&
> +	test_grep "header for deleted file f names a new file" err
> +'
> +
>   test_done


^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line
  2026-11-15 12:39     ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line Siddharth Asthana
@ 2026-11-16 17:54       ` Tanga Joris
  0 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-16 17:54 UTC (permalink / raw)
  To: Siddharth Asthana, git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	kaartic.sivaraam, usmanakinyemi202

Hi Siddharth,

Thank you for reviewing this.

On Sun, 15 Nov 2026 at 15:39, Siddharth Asthana
<siddharthasthana31@gmail.com> wrote:
> The leak is actually in parse_git_diff_header(), we overwrite
> patch->def_name without freeing the old value, so it leaks when the
> function runs again for the next "diff --git". A free() before the
> assignment fixes it, and both callers zero the patch first so it's safe.
> Can't we do that in a prep patch and keep the test?

Yes, I tried it. With the free(), the test passes under SANITIZE=leak,
so I'll add it as a preparatory patch in v3 and keep the test.

> Nit: the notes under "---" describe both patches, so a cover letter with
> "--cover-letter --range-diff=<v2>" would be nicer for v3.

Alright, I'll send v3 with a cover letter and a range-diff.

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names
  2026-11-15 12:47       ` Siddharth Asthana
@ 2026-11-16 18:06         ` Tanga Joris
  0 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-16 18:06 UTC (permalink / raw)
  To: Siddharth Asthana, git
  Cc: phillip.wood, gitster, jackrawlings2011, christian.couder,
	kaartic.sivaraam, usmanakinyemi202

Hi Siddharth,

On Sun, 15 Nov 2026 at 15:47, Siddharth Asthana
<siddharthasthana31@gmail.com> wrote:
> Nit: for "--- /dev/null" this prints "names an old file dev/null", but
> the users never write "dev/null", that's what's left after one leading
> component is stripped. And "--- a/f" gives "names an old file f". The
> only way to reach here is a "---" before "new file mode", so can't we
> name that instead of printing the old name? Same for the deleted case
> with "+++" before "deleted file mode".

Alright, in v3 the error will name the "---" line before "new file
mode", and the "+++" line before "deleted file mode".

Thanks,
Tanga

^ permalink raw reply	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers
  2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
  2026-11-14  1:20     ` [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names Tanga Joris
  2026-11-15 12:39     ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line Siddharth Asthana
@ 2026-11-17  0:43     ` Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header() Tanga Joris
                         ` (2 more replies)
  2 siblings, 3 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-17  0:43 UTC (permalink / raw)
  To: git
  Cc: siddharthasthana31, phillip.wood, gitster, jackrawlings2011,
	christian.couder, kaartic.sivaraam, usmanakinyemi202, Tanga Joris

"git apply --check" dies on an assertion in check_preimage() for some
patches with a broken "diff --git" header. This series reports these
patches as errors instead.

1/3 fixes a leak in parse_git_diff_header().
2/3 rejects a patch that is read under an empty "diff --git" line,
like a "diff --git" line with a blank line after it.
3/3 rejects a new file with a "---" line before "new file mode", and
a deleted file with a "+++" line before "deleted file mode".

Changes since v2:
 - Added 1/3, which frees patch->def_name before setting it, as
   Siddharth suggested. It brings back the test for an empty
   "diff --git" line followed by another one, which v2 dropped.
 - 2/3 is unchanged, apart from where its tests sit in the file.
 - 3/3 names the "---" or "+++" line in the error, instead of the
   file name, which was "dev/null" for "/dev/null", as Siddharth
   suggested.
 - Fixed a broken line in the 3/3 commit message.

Tanga Joris (3):
  apply: free the old def_name in parse_git_diff_header()
  apply: reject a patch after an empty "diff --git" line
  apply: reject new and deleted files that have both names

 apply.c                    |  32 ++++++++-
 t/t4133-apply-filenames.sh | 135 +++++++++++++++++++++++++++++++++++++
 2 files changed, 166 insertions(+), 1 deletion(-)

Range-diff against v2:
-:  ---------- > 1:  8136f837a9 apply: free the old def_name in parse_git_diff_header()
1:  3b802e3053 ! 2:  048574a48f apply: reject a patch after an empty "diff --git" line
    @@ t/t4133-apply-filenames.sh: test_expect_success 'apply diff with old filename mi
     +	reject_empty_git_header later_line.diff 3
     +'
     +
    - test_done
    + test_expect_success 'ignore an empty "diff --git" line before a git patch' '
    + 	test_when_finished "rm -f g" &&
    + 	echo 1 >g &&
2:  f2d40ed931 ! 3:  b3da07db98 apply: reject new and deleted files that have both names
    @@ Commit message
         lines, 2017-06-27) made apply report other conflicting header lines,
         instead of leaving them to an assertion. Do the same here. After the
         header is read, reject a new file that has an old name. Reject a
    -    deleted file that has a new name too.
    +    deleted file that has a new name too. The error names the "---" or
    +    "+++" line that came before the mode line. It does not print the
    +    name, because for "/dev/null" that would be "dev/null".
     
         Git always writes the mode line first, so patches made by Git are not
    -    affected. A patch with
    -    "+++ /dev/null" before "deleted file mode" used to apply without -R.
    -    It is now rejected too.
    +    affected. A patch with "+++ /dev/null" before "deleted file mode" used
    +    to apply without -R. It is now rejected too.
     
    +    Helped-by: Siddharth Asthana <siddharthasthana31@gmail.com>
         Signed-off-by: Tanga Joris <tangajoris@gmail.com>
     
      ## apply.c ##
    @@ apply.c: int parse_git_diff_header(struct strbuf *root,
      	}
     +	if (patch->is_new > 0 && patch->old_name) {
     +		if (patch_input_file)
    -+			error(_("git diff header for new file %s names an old file %s at %s:%d"),
    -+			      patch->new_name, patch->old_name,
    -+			      patch_input_file, *linenr);
    ++			error(_("git diff header has \"---\" before \"new file mode\" at %s:%d"),
    ++			      patch_input_file, patch->extension_linenr);
     +		else
    -+			error(_("git diff header for new file %s names an old file %s (line %d)"),
    -+			      patch->new_name, patch->old_name, *linenr);
    ++			error(_("git diff header has \"---\" before \"new file mode\" (line %d)"),
    ++			      patch->extension_linenr);
     +		return -128;
     +	}
     +	if (patch->is_delete > 0 && patch->new_name) {
     +		if (patch_input_file)
    -+			error(_("git diff header for deleted file %s names a new file %s at %s:%d"),
    -+			      patch->old_name, patch->new_name,
    -+			      patch_input_file, *linenr);
    ++			error(_("git diff header has \"+++\" before \"deleted file mode\" at %s:%d"),
    ++			      patch_input_file, patch->extension_linenr);
     +		else
    -+			error(_("git diff header for deleted file %s names a new file %s (line %d)"),
    -+			      patch->old_name, patch->new_name, *linenr);
    ++			error(_("git diff header has \"+++\" before \"deleted file mode\" (line %d)"),
    ++			      patch->extension_linenr);
     +		return -128;
     +	}
      	patch->is_toplevel_relative = 1;
    @@ apply.c: int parse_git_diff_header(struct strbuf *root,
      }
     
      ## t/t4133-apply-filenames.sh ##
    -@@ t/t4133-apply-filenames.sh: test_expect_success 'name the empty "diff --git" line in the error' '
    - 	reject_empty_git_header later_line.diff 3
    +@@ t/t4133-apply-filenames.sh: test_expect_success 'ignore an empty "diff --git" line before a git patch' '
    + 	test_cmp expect g
      '
      
     +for old in /dev/null a/f
    @@ t/t4133-apply-filenames.sh: test_expect_success 'name the empty "diff --git" lin
     +		+1
     +		EOF
     +		test_must_fail git apply --check old_before_new.diff 2>err &&
    -+		test_grep "header for new file f names an old file" err
    ++		test_grep "has \"---\" before \"new file mode\" at old_before_new.diff:4" err
     +	'
     +done
     +
    @@ t/t4133-apply-filenames.sh: test_expect_success 'name the empty "diff --git" lin
     +	-1
     +	EOF
     +	test_must_fail git apply --check new_before_delete.diff 2>err &&
    -+	test_grep "header for deleted file f names a new file" err &&
    ++	test_grep "has \"+++\" before \"deleted file mode\" at new_before_delete.diff:4" err &&
     +	test_must_fail git apply --check -R new_before_delete.diff 2>err &&
    -+	test_grep "header for deleted file f names a new file" err
    ++	test_grep "has \"+++\" before \"deleted file mode\" at new_before_delete.diff:4" err
     +'
     +
      test_done

base-commit: a018953688f1b10bddf91bff8747068f5f4746a4
-- 
2.43.0


^ permalink raw reply	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header()
  2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
@ 2026-11-17  0:43       ` Tanga Joris
  2026-11-18 20:51         ` Junio C Hamano
  2026-11-17  0:43       ` [Outreachy PATCH v3 2/3] apply: reject a patch after an empty "diff --git" line Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 3/3] apply: reject new and deleted files that have both names Tanga Joris
  2 siblings, 1 reply; 23+ messages in thread
From: Tanga Joris @ 2026-11-17  0:43 UTC (permalink / raw)
  To: git
  Cc: siddharthasthana31, phillip.wood, gitster, jackrawlings2011,
	christian.couder, kaartic.sivaraam, usmanakinyemi202, Tanga Joris

parse_git_diff_header() sets patch->def_name from the "diff --git"
line, without freeing the name it already holds. find_header() calls
it again for each "diff --git" line it finds, with the same patch. So
when an empty "diff --git" line is followed by another "diff --git"
line, the first name leaks.

Free the old name before setting the new one. Both callers start with
a zeroed patch, so def_name is NULL the first time.

Add a test that such a patch still applies. Without this change, the
test fails under SANITIZE=leak.

Suggested-by: Siddharth Asthana <siddharthasthana31@gmail.com>
Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
 apply.c                    |  1 +
 t/t4133-apply-filenames.sh | 18 ++++++++++++++++++
 2 files changed, 19 insertions(+)

diff --git a/apply.c b/apply.c
index f00b7ba4d3..0847835357 100644
--- a/apply.c
+++ b/apply.c
@@ -1373,6 +1373,7 @@ int parse_git_diff_header(struct strbuf *root,
 	 * or removing or adding empty files), so we get
 	 * the default name from the header.
 	 */
+	free(patch->def_name);
 	patch->def_name = git_header_name(p_value, line, len);
 	if (patch->def_name && root->len) {
 		char *s = xstrfmt("%s%s", root->buf, patch->def_name);
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index 3cab1038cf..f8158faec0 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -60,4 +60,22 @@ test_expect_success 'apply diff with old filename missing from headers' '
 	test_grep "lacks filename information" err
 '
 
+test_expect_success 'ignore an empty "diff --git" line before a git patch' '
+	test_when_finished "rm -f g" &&
+	echo 1 >g &&
+	cat >empty_section.diff <<-\EOF &&
+	diff --git a/g b/g
+	diff --git a/g b/g
+	index d00491f..0cfbf08 100644
+	--- a/g
+	+++ b/g
+	@@ -1 +1 @@
+	-1
+	+2
+	EOF
+	git apply empty_section.diff &&
+	echo 2 >expect &&
+	test_cmp expect g
+'
+
 test_done
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v3 2/3] apply: reject a patch after an empty "diff --git" line
  2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header() Tanga Joris
@ 2026-11-17  0:43       ` Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 3/3] apply: reject new and deleted files that have both names Tanga Joris
  2 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-17  0:43 UTC (permalink / raw)
  To: git
  Cc: siddharthasthana31, phillip.wood, gitster, jackrawlings2011,
	christian.couder, kaartic.sivaraam, usmanakinyemi202, Tanga Joris

When a "diff --git" line is followed by a line that is not a git
header line, such as a blank line or an indented "new file mode"
line, parse_git_diff_header() stops right after the "diff --git"
line. find_header() then treats it as an empty git header and keeps
looking, as it has done since 206de27efa (git-apply: ignore empty git
headers, 2005-06-12). It finds the "--- /dev/null" and "+++ b/<path>"
lines and reads the rest as a traditional patch.

But parse_git_diff_header() has already set old_name and new_name
from the "diff --git" line. These names stay in the patch.
parse_traditional_patch() sets is_new for "--- /dev/null", but it
does not clear old_name. So "git apply --check" dies on the assertion
that 5c47f4c6e7 (builtin-apply: accept patch to an empty file,
2008-05-17) added to check_preimage():

  check_preimage: Assertion `patch->is_new <= 0' failed.

The stale new_name causes the same crash when a deletion with such a
header is applied with "git apply -R". When such a patch modifies a
file, it applies. But its "old mode" and "new mode" lines are
ignored.

These are corrupt patches. "git diff" does not produce them. So
report the empty header as an error, instead of reading the patch
that follows it. This also rejects a "diff --git" line that is
followed by other text and then a non-git patch, which used to apply.
A "diff --git" line followed by another "diff --git" line is handled
as before, because 206de27efa ignores empty headers on purpose. Only
a patch that is read under an empty header is a problem.

Reported-by: Gingrspacecadet <jackrawlings2011@gmail.com>
Helped-by: Phillip Wood <phillip.wood@dunelm.org.uk>
Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
 apply.c                    | 13 +++++-
 t/t4133-apply-filenames.sh | 86 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 98 insertions(+), 1 deletion(-)

diff --git a/apply.c b/apply.c
index 0847835357..77a665a71a 100644
--- a/apply.c
+++ b/apply.c
@@ -1605,6 +1605,7 @@ static int find_header(struct apply_state *state,
 		       struct patch *patch)
 {
 	unsigned long offset, len;
+	int empty_git_hdr_linenr = 0;
 
 	patch->is_toplevel_relative = 0;
 	patch->is_rename = patch->is_copy = 0;
@@ -1645,6 +1646,7 @@ static int find_header(struct apply_state *state,
 		 * or mode change, so we handle that specially
 		 */
 		if (!memcmp("diff --git ", line, 11)) {
+			int linenr = state->linenr;
 			int git_hdr_len = parse_git_diff_header(&state->root,
 								state->patch_input_file,
 								&state->linenr,
@@ -1652,8 +1654,10 @@ static int find_header(struct apply_state *state,
 								size, patch);
 			if (git_hdr_len < 0)
 				return -128;
-			if (git_hdr_len <= len)
+			if (git_hdr_len <= len) {
+				empty_git_hdr_linenr = linenr;
 				continue;
+			}
 			*hdrsize = git_hdr_len;
 			return offset;
 		}
@@ -1671,6 +1675,13 @@ static int find_header(struct apply_state *state,
 		if (size < nextlen + 14 || memcmp("@@ -", line + len + nextlen, 4))
 			continue;
 
+		/* Git does not write a patch under an empty git header */
+		if (empty_git_hdr_linenr) {
+			error(_("corrupt patch: empty \"diff --git\" header at %s:%d"),
+			      state->patch_input_file, empty_git_hdr_linenr);
+			return -128;
+		}
+
 		/* Ok, we'll consider it a patch */
 		if (parse_traditional_patch(state, line, line+len, patch))
 			return -128;
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index f8158faec0..b6b9195924 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -60,6 +60,92 @@ test_expect_success 'apply diff with old filename missing from headers' '
 	test_grep "lacks filename information" err
 '
 
+reject_empty_git_header () {
+	file=$1 &&
+	line=$2 &&
+	shift 2 &&
+	test_must_fail git apply --check "$@" "$file" 2>err &&
+	test_grep "corrupt patch: empty \"diff --git\" header at $file:$line" err
+}
+
+test_expect_success 'reject a new file with a blank line after "diff --git"' '
+	cat >blank_line.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header blank_line.diff 1 &&
+	test_must_fail git apply blank_line.diff &&
+	test_path_is_missing f
+'
+
+test_expect_success 'reject an indented "new file mode" line' '
+	cat >indented_mode.diff <<-\EOF &&
+	diff --git a/f b/f
+	 new file mode 100644
+	index 0000000..d00491f
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header indented_mode.diff 1
+'
+
+test_expect_success 'reject a deletion with a blank line after "diff --git"' '
+	cat >blank_line_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+
+	deleted file mode 100644
+	index d00491f..0000000
+	--- a/f
+	+++ /dev/null
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	reject_empty_git_header blank_line_delete.diff 1 &&
+	reject_empty_git_header blank_line_delete.diff 1 -R
+'
+
+test_expect_success 'reject a modification with a blank line after "diff --git"' '
+	test_when_finished "rm -f g" &&
+	echo 1 >g &&
+	cat >blank_line_modify.diff <<-\EOF &&
+	diff --git a/g b/g
+
+	old mode 100644
+	new mode 100755
+	index d00491f..0cfbf08
+	--- a/g
+	+++ b/g
+	@@ -1 +1 @@
+	-1
+	+2
+	EOF
+	reject_empty_git_header blank_line_modify.diff 1 &&
+	echo 1 >expect &&
+	test_cmp expect g
+'
+
+test_expect_success 'name the empty "diff --git" line in the error' '
+	cat >later_line.diff <<-\EOF &&
+	some text before the patch
+
+	diff --git a/f b/f
+	some text after it
+	--- /dev/null
+	+++ b/f
+	@@ -0,0 +1 @@
+	+1
+	EOF
+	reject_empty_git_header later_line.diff 3
+'
+
 test_expect_success 'ignore an empty "diff --git" line before a git patch' '
 	test_when_finished "rm -f g" &&
 	echo 1 >g &&
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* [Outreachy PATCH v3 3/3] apply: reject new and deleted files that have both names
  2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header() Tanga Joris
  2026-11-17  0:43       ` [Outreachy PATCH v3 2/3] apply: reject a patch after an empty "diff --git" line Tanga Joris
@ 2026-11-17  0:43       ` Tanga Joris
  2 siblings, 0 replies; 23+ messages in thread
From: Tanga Joris @ 2026-11-17  0:43 UTC (permalink / raw)
  To: git
  Cc: siddharthasthana31, phillip.wood, gitster, jackrawlings2011,
	christian.couder, kaartic.sivaraam, usmanakinyemi202, Tanga Joris

Git writes "new file mode" or "deleted file mode" before the "---"
and "+++" lines. The "new file mode" line sets is_new. Then
gitdiff_oldname() leaves old_name empty for "--- /dev/null".

When the "---" line comes first, is_new is still 0. So
gitdiff_verify_name() takes the name on that line as old_name. For
"--- /dev/null" this gives "dev/null", because one leading directory
is stripped. The "new file mode" line then sets is_new. The result is
a new file that also has an old name. "git apply --check" dies on it,
on the same assertion in check_preimage(). In the same way,
"+++ /dev/null" before "deleted file mode" gives a deleted file with
a new name. "git apply -R" dies on that one.

d70e9c5c8c (apply: check git diffs for mutually exclusive header
lines, 2017-06-27) made apply report other conflicting header lines,
instead of leaving them to an assertion. Do the same here. After the
header is read, reject a new file that has an old name. Reject a
deleted file that has a new name too. The error names the "---" or
"+++" line that came before the mode line. It does not print the
name, because for "/dev/null" that would be "dev/null".

Git always writes the mode line first, so patches made by Git are not
affected. A patch with "+++ /dev/null" before "deleted file mode" used
to apply without -R. It is now rejected too.

Helped-by: Siddharth Asthana <siddharthasthana31@gmail.com>
Signed-off-by: Tanga Joris <tangajoris@gmail.com>
---
 apply.c                    | 18 ++++++++++++++++++
 t/t4133-apply-filenames.sh | 31 +++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+)

diff --git a/apply.c b/apply.c
index 77a665a71a..b920dbbbc7 100644
--- a/apply.c
+++ b/apply.c
@@ -1467,6 +1467,24 @@ int parse_git_diff_header(struct strbuf *root,
 			      *linenr);
 		return -128;
 	}
+	if (patch->is_new > 0 && patch->old_name) {
+		if (patch_input_file)
+			error(_("git diff header has \"---\" before \"new file mode\" at %s:%d"),
+			      patch_input_file, patch->extension_linenr);
+		else
+			error(_("git diff header has \"---\" before \"new file mode\" (line %d)"),
+			      patch->extension_linenr);
+		return -128;
+	}
+	if (patch->is_delete > 0 && patch->new_name) {
+		if (patch_input_file)
+			error(_("git diff header has \"+++\" before \"deleted file mode\" at %s:%d"),
+			      patch_input_file, patch->extension_linenr);
+		else
+			error(_("git diff header has \"+++\" before \"deleted file mode\" (line %d)"),
+			      patch->extension_linenr);
+		return -128;
+	}
 	patch->is_toplevel_relative = 1;
 	return offset;
 }
diff --git a/t/t4133-apply-filenames.sh b/t/t4133-apply-filenames.sh
index b6b9195924..b85dcb57bd 100755
--- a/t/t4133-apply-filenames.sh
+++ b/t/t4133-apply-filenames.sh
@@ -164,4 +164,35 @@ test_expect_success 'ignore an empty "diff --git" line before a git patch' '
 	test_cmp expect g
 '
 
+for old in /dev/null a/f
+do
+	test_expect_success "reject \"--- $old\" before \"new file mode\"" '
+		cat >old_before_new.diff <<-EOF &&
+		diff --git a/f b/f
+		--- $old
+		+++ b/f
+		new file mode 100644
+		@@ -0,0 +1 @@
+		+1
+		EOF
+		test_must_fail git apply --check old_before_new.diff 2>err &&
+		test_grep "has \"---\" before \"new file mode\" at old_before_new.diff:4" err
+	'
+done
+
+test_expect_success 'reject "+++ /dev/null" before "deleted file mode"' '
+	cat >new_before_delete.diff <<-\EOF &&
+	diff --git a/f b/f
+	--- a/f
+	+++ /dev/null
+	deleted file mode 100644
+	@@ -1 +0,0 @@
+	-1
+	EOF
+	test_must_fail git apply --check new_before_delete.diff 2>err &&
+	test_grep "has \"+++\" before \"deleted file mode\" at new_before_delete.diff:4" err &&
+	test_must_fail git apply --check -R new_before_delete.diff 2>err &&
+	test_grep "has \"+++\" before \"deleted file mode\" at new_before_delete.diff:4" err
+'
+
 test_done
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 23+ messages in thread

* Re: [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header()
  2026-11-17  0:43       ` [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header() Tanga Joris
@ 2026-11-18 20:51         ` Junio C Hamano
  0 siblings, 0 replies; 23+ messages in thread
From: Junio C Hamano @ 2026-11-18 20:51 UTC (permalink / raw)
  To: Tanga Joris
  Cc: git, siddharthasthana31, phillip.wood, jackrawlings2011,
	christian.couder, kaartic.sivaraam, usmanakinyemi202

Tanga Joris <tangajoris@gmail.com> writes:

> diff --git a/apply.c b/apply.c
> index f00b7ba4d3..0847835357 100644
> --- a/apply.c
> +++ b/apply.c
> @@ -1373,6 +1373,7 @@ int parse_git_diff_header(struct strbuf *root,
>  	 * or removing or adding empty files), so we get
>  	 * the default name from the header.
>  	 */
> +	free(patch->def_name);
>  	patch->def_name = git_header_name(p_value, line, len);
>  	if (patch->def_name && root->len) {
>  		char *s = xstrfmt("%s%s", root->buf, patch->def_name);

This looks seriously inadequate, though.

After seeing "diff --git" line and extracting patch->def_name by
calling git_header_name(), suppose there is an unrecognised line
(e.g., "diff --git a/f b/f" followed by "diff --git a/g b/g").  We
set patch->old_name and patch->new_name both to copy of
patch->def_name (e.g., "f" taken from "diff --git a/f b/f") and
patch->is_toplevel_relative set to 1 before we return from
parse_git_diff_header() to find_header().  Which causes
find_header() to "continue" when git_hdr_len <= len, without
clearing patch->old_name and patch->new_name and others.

The test below only "passes" because the bogus "diff --git a/g b/g"
line names the same file "g" used for the next one, I suspect.  If
you replace 'g' with 'f' on the bogus "diff --git" line, you will
see an "inconsistent old filename" error, so this change is not
making us ignore the bogus "diff --git" line, either.  Not that I
think it is a good idea to ignore such a corrupt patch input.

> +test_expect_success 'ignore an empty "diff --git" line before a git patch' '
> +	test_when_finished "rm -f g" &&
> +	echo 1 >g &&
> +	cat >empty_section.diff <<-\EOF &&
> +	diff --git a/g b/g
> +	diff --git a/g b/g
> +	index d00491f..0cfbf08 100644
> +	--- a/g
> +	+++ b/g
> +	@@ -1 +1 @@
> +	-1
> +	+2
> +	EOF
> +	git apply empty_section.diff &&
> +	echo 2 >expect &&
> +	test_cmp expect g
> +'

I do not think we should ignore such a bogus "diff --git" line;
shouldn't we diagnose it as a corrupt patch input?  I do not think
we are there yet without making find_header() do more work when
punting, though.

Thanks.

^ permalink raw reply	[flat|nested] 23+ messages in thread

end of thread, other threads:[~2026-11-18 20:51 UTC | newest]

Thread overview: 23+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-10-17 15:17 Git crashed on `git apply --check` Gingrspacecadet
2026-10-24  6:10 ` Tanga Joris
     [not found]   ` <CALdwsC11JHUBBTt3e_RFixLYSF0VU1C+mG7bQsO0vV=a68XHNw@mail.gmail.com>
2026-10-25 22:56     ` Tanga Joris
2026-11-07 19:54 ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Tanga Joris
2026-11-07 19:54   ` [Outreachy PATCH 2/2] apply: reject new and deleted files that have both names Tanga Joris
2026-11-09 10:52   ` [Outreachy PATCH 1/2] apply: do not keep names from an unused "diff --git" line Phillip Wood
2026-11-09 13:13     ` Tanga Joris
2026-11-09 14:31       ` Phillip Wood
2026-11-09 20:54         ` Tanga Joris
2026-11-09 21:06           ` Junio C Hamano
2026-11-09 23:03             ` Tanga Joris
2026-11-09 19:56     ` Junio C Hamano
2026-11-14  1:20   ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty " Tanga Joris
2026-11-14  1:20     ` [Outreachy PATCH v2 2/2] apply: reject new and deleted files that have both names Tanga Joris
2026-11-15 12:47       ` Siddharth Asthana
2026-11-16 18:06         ` Tanga Joris
2026-11-15 12:39     ` [Outreachy PATCH v2 1/2] apply: reject a patch after an empty "diff --git" line Siddharth Asthana
2026-11-16 17:54       ` Tanga Joris
2026-11-17  0:43     ` [Outreachy PATCH v3 0/3] apply: reject corrupt "diff --git" headers Tanga Joris
2026-11-17  0:43       ` [Outreachy PATCH v3 1/3] apply: free the old def_name in parse_git_diff_header() Tanga Joris
2026-11-18 20:51         ` Junio C Hamano
2026-11-17  0:43       ` [Outreachy PATCH v3 2/3] apply: reject a patch after an empty "diff --git" line Tanga Joris
2026-11-17  0:43       ` [Outreachy PATCH v3 3/3] apply: reject new and deleted files that have both names Tanga Joris

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox