Re: [LTP] coredump01: New core_pattern specifiers test

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Aug 20, 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

> 	ssize_t rval, i;
> 	int fd, elf;
>
> 	if (bytes >= 4 && !memcmp(magic, "\177ELF", 4))
> 		elf = 1;
>
> 	dprintf(fd, "exe=%s pid=%s sig=%s bytes=%lld elf=%d\n",
> 		argv[1], argv[2], argv[3], bytes, elf);

Could `elf` be initialized to zero? For short or non-ELF input the
condition does not assign it, so `dprintf()` reads an indeterminate value.
A nonzero value can make the test accept a malformed core stream as ELF.

> 	SAFE_PRCTL(PR_GET_DUMPABLE, 1, 0, 0, 0);

Should this use `PR_SET_DUMPABLE`? `PR_GET_DUMPABLE` only returns the
current state and ignores arg2, so this call does not ensure that
`abort()` can produce a core dump.

> 	/* the kernel spawns the helper asynchronously */
> 	if (TST_RETRY_FN_EXP_BACKOFF(access(res, F_OK), TST_RETVAL_EQ0, HELPER_TIMEOUT)) {
> 		tst_res(TFAIL, "%s did not report any core dump", HELPER);

Could kernels with `CONFIG_STATIC_USERMODEHELPER` be rejected with TCONF
before this check? In particular, an empty
`CONFIG_STATIC_USERMODEHELPER_PATH` intentionally disables the helper, so
this timeout reports TFAIL without testing specifier expansion.

> 	set_pattern("|%s/%s %%e %%p %%s %s/res.%%p", helper_dir, HELPER, cwd);
>
> 	len = SAFE_READLINK("/proc/self/exe", helper_dir, sizeof(helper_dir) - 1);

Could the pipe case require the initial mount namespace, or otherwise use
paths visible there? The kernel resolves and runs a `core_pattern` pipe
handler in the initial mount namespace. If LTP runs in another mount
namespace, these helper and result paths may not exist there, producing a
spurious TFAIL.

> /*
>  * Core dump collector for the piped core_pattern tested by coredump01.
>  *
>  * Avoiding the LTP API here is correct, since the kernel spawns the helper
>  * through ``call_usermodehelper()`` without the LTP IPC environment.
>  */
>
> int main(int argc, char *argv[])

Could this helper follow the LTP helper-binary convention by defining
`TST_NO_DEFAULT_MAIN` and including `tst_test.h`? The missing IPC
environment means it should avoid IPC-backed result calls, but does not
require avoiding the helper API form itself.

Verdict - Needs revision

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[Andrea Cervesato via ltp]

> > 	ssize_t rval, i;
> > 	int fd, elf;
> >
> > 	if (bytes >= 4 && !memcmp(magic, "\177ELF", 4))
> > 		elf = 1;
> >
> > 	dprintf(fd, "exe=%s pid=%s sig=%s bytes=%lld elf=%d\n",
> > 		argv[1], argv[2], argv[3], bytes, elf);
> 
> Could `elf` be initialized to zero? For short or non-ELF input the
> condition does not assign it, so `dprintf()` reads an indeterminate value.
> A nonzero value can make the test accept a malformed core stream as ELF.

This can be fixed.

> 
> > 	SAFE_PRCTL(PR_GET_DUMPABLE, 1, 0, 0, 0);
> 
> Should this use `PR_SET_DUMPABLE`? `PR_GET_DUMPABLE` only returns the
> current state and ignores arg2, so this call does not ensure that
> `abort()` can produce a core dump.
> 
> > 	/* the kernel spawns the helper asynchronously */
> > 	if (TST_RETRY_FN_EXP_BACKOFF(access(res, F_OK), TST_RETVAL_EQ0, HELPER_TIMEOUT)) {
> > 		tst_res(TFAIL, "%s did not report any core dump", HELPER);
> 
> Could kernels with `CONFIG_STATIC_USERMODEHELPER` be rejected with TCONF
> before this check? In particular, an empty
> `CONFIG_STATIC_USERMODEHELPER_PATH` intentionally disables the helper, so
> this timeout reports TFAIL without testing specifier expansion.

This can be added in needs_kconfig.

> 
> > 	set_pattern("|%s/%s %%e %%p %%s %s/res.%%p", helper_dir, HELPER, cwd);
> >
> > 	len = SAFE_READLINK("/proc/self/exe", helper_dir, sizeof(helper_dir) - 1);
> 
> Could the pipe case require the initial mount namespace, or otherwise use
> paths visible there? The kernel resolves and runs a `core_pattern` pipe
> handler in the initial mount namespace. If LTP runs in another mount
> namespace, these helper and result paths may not exist there, producing a
> spurious TFAIL.
> 
> > /*
> >  * Core dump collector for the piped core_pattern tested by coredump01.
> >  *
> >  * Avoiding the LTP API here is correct, since the kernel spawns the helper
> >  * through ``call_usermodehelper()`` without the LTP IPC environment.
> >  */
> >
> > int main(int argc, char *argv[])
> 
> Could this helper follow the LTP helper-binary convention by defining
> `TST_NO_DEFAULT_MAIN` and including `tst_test.h`? The missing IPC
> environment means it should avoid IPC-backed result calls, but does not
> require avoiding the helper API form itself.

It doesn't make any sense. We can skip this one.

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On August 24, 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

> static struct tst_test test = {
> 	.test = run,
> 	.tcnt = 2,
> 	.setup = setup,
> 	.needs_root = 1,
> 	.needs_tmpdir = 1,
> 	.forks_child = 1,
> 	.save_restore = (const struct tst_path_val[]) {

Could this add a CONFIG_COREDUMP=y entry to .needs_kconfigs?
Without it, the ELF binary format has no core-dump callback, so
vfs_coredump() returns without setting the WCOREDUMP status bit.
crash_child() then reports TFAIL rather than TCONF on a valid kernel
configuration.

Verdict - Needs revision

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Mon, 24 Aug 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

Verdict - Reviewed

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On September 2, 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

> 	if (tst_get_path(HELPER, helper_path, sizeof(helper_path)))
> 		tst_brk(TCONF, "'%s' not found in $PATH", HELPER);

`tst_get_path()` can return a relative path such as
`./coredump01_helper`, but a piped core_pattern helper is resolved from
the root directory. Canonicalize `helper_path` to an absolute path before
writing it to core_pattern, otherwise the helper may not start and the
test reports a false failure.

Verdict - Needs revision

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[Andrea Cervesato via ltp]

> Hi Andrea,
> 
> On September 2, 2026, Andrea Cervesato wrote:
> > coredump01: New core_pattern specifiers test
> 
> > 	if (tst_get_path(HELPER, helper_path, sizeof(helper_path)))
> > 		tst_brk(TCONF, "'%s' not found in $PATH", HELPER);
> 
> `tst_get_path()` can return a relative path such as
> `./coredump01_helper`, but a piped core_pattern helper is resolved from
> the root directory. Canonicalize `helper_path` to an absolute path before
> writing it to core_pattern, otherwise the helper may not start and the
> test reports a false failure.
> 
> Verdict - Needs revision
> 
> ---
> Note:
> 
> The agent can sometimes produce false positives although often its
> findings are genuine. If you find issues with the review, please
> comment this email or ignore the suggestions.
> 
> Regards,
> LTP AI Reviewer
> 
> -- 
> Mailing list info: https://lists.linux.it/listinfo/ltp

This is valid. I sent a new version

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Thu, 3 Sep 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

Verdict - Reviewed

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Fri, Sep 4, 2026 at 09:15, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

--- [PATCH 2/2] ---

> Verify that a kernel-generated core dump contains a valid ELF header,
> expected program headers (PT_NOTE and PT_LOAD), and valid process
> metadata notes (NT_PRPSINFO and NT_PRSTATUS).

The commit message only restates what the test verifies. Please explain why
this coverage is needed or what problem it addresses.

Verdict - Needs revision

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v7 0/2] coredump testing suite

[Andrea Cervesato]

Add a simple coredump01 test that simulate a coredump and verify that it
actually works. This is a small coverage that is missing inside LTP
testing suite.

Add a coredump02 test that verifies ELF data after a coredump.

Signed-off-by: Andrea Cervesato <andrea.cervesato@suse.com>
---
Changes in v7:
- define a common header for the testing suite
- update commit message for coredump02
- Link to v6: https://lore.kernel.org/20260904-coredump-v6-0-b8e3947cc646@suse.com

Changes in v6:
- coredump01 check Elf
- add coredump02 test for Elf validation
- Link to v5: https://lore.kernel.org/20260903-coredump-v5-1-2ea86fa29107@suse.com

Changes in v5:
- resolve path from root for helper
- Link to v4: https://lore.kernel.org/20260902-coredump-v4-1-3f511ea1c96a@suse.com

Changes in v4:
- read back core_pattern after writing and abort with TBROK on mismatch
- read CONFIG_STATIC_USERMODEHELPER once in setup() and cache the result
- remove namespace visibility scaffolding and skip logic from verify_pipe_pattern()
- resolve helper path with tst_get_path() instead of /proc/self/exe readlink
- Link to v3: https://lore.kernel.org/20260824-coredump-v3-1-6429da674718@suse.com

Changes in v3:
- check for CONFIG_COREDUMP=y
- Link to v2: https://lore.kernel.org/20260824-coredump-v2-1-42428acf5216@suse.com

Changes in v2:
- fix CI agent spotted errors
- Link to v1: https://lore.kernel.org/20260820-coredump-v1-1-7ce51603cefb@suse.com

To: Linux Test Project <ltp@lists.linux.it>

---
Andrea Cervesato (2):
      coredump01: New core_pattern specifiers test
      coredump02: Verify ELF structure and notes

 runtest/kernel_misc                           |   2 +
 testcases/kernel/Makefile                     |   1 +
 testcases/kernel/coredump/.gitignore          |   3 +
 testcases/kernel/coredump/Makefile            |   7 +
 testcases/kernel/coredump/coredump01.c        | 158 ++++++++++++++++++++
 testcases/kernel/coredump/coredump01_helper.c |  73 ++++++++++
 testcases/kernel/coredump/coredump02.c        | 198 ++++++++++++++++++++++++++
 testcases/kernel/coredump/coredump_common.h   |  69 +++++++++
 8 files changed, 511 insertions(+)
---
base-commit: 043bdc49af410f55ece6f61abcdaba7d9580f86a
change-id: 20260820-coredump-22fc731bde08

Best regards,
--  
Andrea Cervesato <andrea.cervesato@suse.com>


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v7 1/2] coredump01: New core_pattern specifiers test

[Andrea Cervesato]

From: Andrea Cervesato <andrea.cervesato@suse.com>

LTP has no coverage for the core_pattern specifiers expansion, neither
for the pipe flavor of core_pattern, where the kernel spawns a user
space helper and writes the core dump into its standard input.

The new test verifies that %e, %p and %s are expanded both in the core
file name and in the arguments of a piped helper, checking also that
the helper really receives an ELF core dump.

Root is required to rewrite the system wide core_pattern, which is
restored by the test library on all exit paths.

Signed-off-by: Andrea Cervesato <andrea.cervesato@suse.com>
---
 runtest/kernel_misc                           |   1 +
 testcases/kernel/Makefile                     |   1 +
 testcases/kernel/coredump/.gitignore          |   2 +
 testcases/kernel/coredump/Makefile            |   7 ++
 testcases/kernel/coredump/coredump01.c        | 158 ++++++++++++++++++++++++++
 testcases/kernel/coredump/coredump01_helper.c |  73 ++++++++++++
 testcases/kernel/coredump/coredump_common.h   |  69 +++++++++++
 7 files changed, 311 insertions(+)

diff --git a/runtest/kernel_misc b/runtest/kernel_misc
index cc3562cb7..ecf9ee2a2 100644
--- a/runtest/kernel_misc
+++ b/runtest/kernel_misc
@@ -17,3 +17,4 @@ zram02 zram02.sh
 zram03 zram03
 umip_basic_test umip_basic_test
 aslr01 aslr01
+coredump01 coredump01
diff --git a/testcases/kernel/Makefile b/testcases/kernel/Makefile
index ac816e4e8..6b303c14b 100644
--- a/testcases/kernel/Makefile
+++ b/testcases/kernel/Makefile
@@ -19,6 +19,7 @@ SUBDIRS			+= lib
 SUBDIRS			+= connectors \
 			   containers \
 			   controllers \
+			   coredump \
 			   crypto \
 			   device-drivers \
 			   firmware \
diff --git a/testcases/kernel/coredump/.gitignore b/testcases/kernel/coredump/.gitignore
new file mode 100644
index 000000000..cd0b51700
--- /dev/null
+++ b/testcases/kernel/coredump/.gitignore
@@ -0,0 +1,2 @@
+/coredump01
+/coredump01_helper
diff --git a/testcases/kernel/coredump/Makefile b/testcases/kernel/coredump/Makefile
new file mode 100644
index 000000000..951a3e976
--- /dev/null
+++ b/testcases/kernel/coredump/Makefile
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+# Copyright (c) 2026 Linux Test Project
+
+top_srcdir		?= ../../..
+
+include $(top_srcdir)/include/mk/testcases.mk
+include $(top_srcdir)/include/mk/generic_leaf_target.mk
diff --git a/testcases/kernel/coredump/coredump01.c b/testcases/kernel/coredump/coredump01.c
new file mode 100644
index 000000000..ad690e95c
--- /dev/null
+++ b/testcases/kernel/coredump/coredump01.c
@@ -0,0 +1,158 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Linux Test Project
+ */
+
+/*\
+ * Verify that the kernel expands the :manpage:`core(5)` specifiers %e
+ * (executable name), %p (PID) and %s (signal number) when it creates a
+ * core dump.
+ *
+ * Both flavors of ``/proc/sys/kernel/core_pattern`` are tested:
+ *
+ * - a plain file name, where the specifiers become part of the core file
+ *   name
+ * - a pipe to a user space helper, where the specifiers are expanded in
+ *   the helper arguments and the core dump is written to the helper
+ *   standard input
+ *
+ * The test needs root because it rewrites the system wide core_pattern.
+ * The original value is saved and restored by the test library on all
+ * exit paths.
+ *
+ * [Algorithm]
+ *
+ * - Point core_pattern into the test temporary directory
+ * - Fork a child which aborts itself to produce a core dump
+ * - For the file pattern check that ``core.<exe>.<pid>.<signal>`` was created
+ * - For the pipe pattern check the arguments the helper was called with
+ *   and that it received an ELF core dump on its standard input
+ */
+
+#include "coredump_common.h"
+#include "tst_kconfig.h"
+
+#define HELPER "coredump01_helper"
+#define HELPER_TIMEOUT 10
+
+static char helper_path[PATH_MAX];
+static int static_usermodehelper;
+
+/*
+ * e_ident[] and e_type live at the same file offset in ELF32 and ELF64,
+ * so we can validate both without picking a class.
+ */
+static void verify_elf_core(const char *path)
+{
+	unsigned char hdr[EI_NIDENT + sizeof(Elf32_Half)];
+	Elf32_Half e_type;
+	int fd;
+
+	fd = SAFE_OPEN(path, O_RDONLY);
+	SAFE_READ(1, fd, hdr, sizeof(hdr));
+	SAFE_CLOSE(fd);
+
+	memcpy(&e_type, hdr + EI_NIDENT, sizeof(e_type));
+
+	TST_EXP_EXPR(!memcmp(hdr, ELFMAG, SELFMAG), "%s starts with ELF magic", path);
+	TST_EXP_EQ_LI(e_type, ET_CORE);
+}
+
+static void verify_file_pattern(void)
+{
+	char dump[PATH_MAX + 32];
+	pid_t pid;
+
+	tst_res(TINFO, "Testing file core_pattern");
+
+	set_pattern("%s/core.%%e.%%p.%%s", cwd);
+
+	pid = crash_child();
+
+	snprintf(dump, sizeof(dump), "%s/core.coredump01.%d.%d", cwd, pid, SIGABRT);
+
+	TST_EXP_PASS(access(dump, F_OK), "core.%%e.%%p.%%s expanded to core.coredump01.%d.%d",
+		     pid, SIGABRT);
+	if (!TST_PASS)
+		return;
+
+	verify_elf_core(dump);
+}
+
+static void verify_pipe_pattern(void)
+{
+	char res[PATH_MAX + 32], exe[PATH_MAX];
+	int pid_seen, sig_seen, elf, et_core;
+	long long bytes;
+	pid_t pid;
+
+	if (static_usermodehelper) {
+		tst_res(TCONF, "CONFIG_STATIC_USERMODEHELPER is enabled, skipping pipe core_pattern");
+		return;
+	}
+
+	tst_res(TINFO, "Testing pipe core_pattern");
+
+	set_pattern("|%s %%e %%p %%s %s/res.%%p", helper_path, cwd);
+
+	pid = crash_child();
+
+	snprintf(res, sizeof(res), "%s/res.%d", cwd, pid);
+
+	/* the kernel spawns the helper asynchronously */
+	if (TST_RETRY_FN_EXP_BACKOFF(access(res, F_OK), TST_RETVAL_EQ0, HELPER_TIMEOUT)) {
+		tst_res(TFAIL, "%s did not report any core dump", HELPER);
+		return;
+	}
+
+	SAFE_FILE_SCANF(res, "exe=%15s pid=%d sig=%d bytes=%lld elf=%d et_core=%d",
+			exe, &pid_seen, &sig_seen, &bytes, &elf, &et_core);
+
+	TST_EXP_EQ_STR(exe, "coredump01");
+	TST_EXP_EQ_LI(pid_seen, pid);
+	TST_EXP_EQ_LI(sig_seen, SIGABRT);
+	TST_EXP_EXPR(elf && bytes > 0, "%s read %lli bytes of ELF core dump", HELPER, bytes);
+	TST_EXP_EQ_LI(et_core, 1);
+}
+
+static void run(unsigned int n)
+{
+	if (n)
+		verify_pipe_pattern();
+	else
+		verify_file_pattern();
+}
+
+static void setup(void)
+{
+	char path[PATH_MAX];
+	struct tst_kconfig_var kconfig = TST_KCONFIG_INIT("CONFIG_STATIC_USERMODEHELPER");
+
+	coredump_setup();
+
+	tst_kconfig_read(&kconfig, 1);
+	static_usermodehelper = (kconfig.choice == 'y');
+
+	if (tst_get_path(HELPER, path, sizeof(path)))
+		tst_brk(TCONF, "'%s' not found in $PATH", HELPER);
+
+	if (!realpath(path, helper_path))
+		tst_brk(TBROK | TERRNO, "realpath(%s) failed", path);
+}
+
+static struct tst_test test = {
+	.test = run,
+	.tcnt = 2,
+	.setup = setup,
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.forks_child = 1,
+	.needs_kconfigs = (const char* []) {
+		"CONFIG_COREDUMP=y",
+		NULL,
+	},
+	.save_restore = (const struct tst_path_val[]) {
+		{PATH_KERN_CORE_PATTERN, NULL, TST_SR_TCONF},
+		{}
+	},
+};
diff --git a/testcases/kernel/coredump/coredump01_helper.c b/testcases/kernel/coredump/coredump01_helper.c
new file mode 100644
index 000000000..b454de476
--- /dev/null
+++ b/testcases/kernel/coredump/coredump01_helper.c
@@ -0,0 +1,73 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Linux Test Project
+ */
+
+/*
+ * Core dump collector for the piped core_pattern tested by coredump01.
+ *
+ * Avoiding the LTP API here is correct, since the kernel spawns the helper
+ * through ``call_usermodehelper()`` without the LTP IPC environment.
+ *
+ * The helper is called as:
+ *
+ *	coredump01_helper <exe> <pid> <signal> <result file>
+ *
+ * and stores what it received in the result file, which is published with
+ * :manpage:`rename()` so that the polling test never reads a partial line.
+ */
+
+#include <elf.h>
+#include <fcntl.h>
+#include <limits.h>
+#include <stdio.h>
+#include <string.h>
+#include <unistd.h>
+
+int main(int argc, char *argv[])
+{
+	unsigned char buf[4096], hdr[EI_NIDENT + sizeof(Elf32_Half)];
+	char tmp[PATH_MAX];
+	unsigned int hdr_len = 0;
+	long long bytes = 0;
+	ssize_t rval, i;
+	int fd, elf = 0, et_core = 0;
+	Elf32_Half e_type;
+
+	if (argc < 5)
+		return 1;
+
+	/*
+	 * The ELF header is only meaningful at the very beginning of the
+	 * stream. Collect enough bytes to read e_ident and e_type.
+	 */
+	while ((rval = read(STDIN_FILENO, buf, sizeof(buf))) > 0) {
+		for (i = 0; i < rval && hdr_len < sizeof(hdr); i++)
+			hdr[hdr_len++] = buf[i];
+
+		bytes += rval;
+	}
+
+	if (rval < 0)
+		return 1;
+
+	if (hdr_len >= sizeof(hdr)) {
+		if (!memcmp(hdr, ELFMAG, SELFMAG))
+			elf = 1;
+		memcpy(&e_type, hdr + EI_NIDENT, sizeof(e_type));
+		et_core = (e_type == ET_CORE);
+	}
+
+	snprintf(tmp, sizeof(tmp), "%s.tmp", argv[4]);
+
+	fd = open(tmp, O_WRONLY | O_CREAT | O_TRUNC, 0644);
+	if (fd < 0)
+		return 1;
+
+	dprintf(fd, "exe=%s pid=%s sig=%s bytes=%lld elf=%d et_core=%d\n",
+		argv[1], argv[2], argv[3], bytes, elf, et_core);
+
+	close(fd);
+
+	return rename(tmp, argv[4]) ? 1 : 0;
+}
diff --git a/testcases/kernel/coredump/coredump_common.h b/testcases/kernel/coredump/coredump_common.h
new file mode 100644
index 000000000..7c04a659c
--- /dev/null
+++ b/testcases/kernel/coredump/coredump_common.h
@@ -0,0 +1,69 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Linux Test Project
+ */
+
+#ifndef COREDUMP_COMMON_H
+#define COREDUMP_COMMON_H
+
+#include <elf.h>
+#include <sys/wait.h>
+
+#include "tst_test.h"
+#include "lapi/prctl.h"
+
+#define PATTERN_MAX 128
+
+static char cwd[PATH_MAX];
+
+static void set_pattern(const char *fmt, ...)
+{
+	char pattern[PATTERN_MAX];
+	char readback[PATTERN_MAX];
+	va_list va;
+	int len;
+
+	va_start(va, fmt);
+	len = vsnprintf(pattern, sizeof(pattern), fmt, va);
+	va_end(va);
+
+	if (len >= PATTERN_MAX)
+		tst_brk(TCONF, "core_pattern does not fit into %i bytes", PATTERN_MAX - 1);
+
+	SAFE_FILE_PRINTF(PATH_KERN_CORE_PATTERN, "%s", pattern);
+	SAFE_FILE_LINES_SCANF(PATH_KERN_CORE_PATTERN, "%127[^\n]", readback);
+
+	if (strcmp(pattern, readback))
+		tst_brk(TBROK, "core_pattern readback mismatch: wrote '%s', read '%s'",
+			pattern, readback);
+
+	tst_res(TINFO, "core_pattern is '%s'", pattern);
+}
+
+static pid_t crash_child(void)
+{
+	int status;
+	pid_t pid;
+
+	pid = SAFE_FORK();
+	if (!pid)
+		abort();
+
+	SAFE_WAITPID(pid, &status, 0);
+
+	if (!WIFSIGNALED(status) || !WCOREDUMP(status))
+		tst_brk(TFAIL, "Child did not dump core");
+
+	return pid;
+}
+
+static void coredump_setup(void)
+{
+	struct rlimit rl = {RLIM_INFINITY, RLIM_INFINITY};
+
+	SAFE_SETRLIMIT(RLIMIT_CORE, &rl);
+	SAFE_PRCTL(PR_SET_DUMPABLE, 1, 0, 0, 0);
+	SAFE_GETCWD(cwd, sizeof(cwd));
+}
+
+#endif /* COREDUMP_COMMON_H */

-- 
2.51.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v7 2/2] coredump02: Verify ELF structure and notes

[Andrea Cervesato]

From: Andrea Cervesato <andrea.cervesato@suse.com>

LTP currently only tests core_pattern specifier expansion and basic
ELF magic in coredump01, leaving the internal ELF structure and notes
generated by the kernel unverified.

Add a test to verify that the kernel produces a valid ET_CORE ELF file
with the expected PT_NOTE and PT_LOAD segments, and that the
NT_PRPSINFO and NT_PRSTATUS notes contain the expected process name,
PID, and terminating signal.

Root is required to set the system-wide core_pattern into the test's
temporary directory.

Signed-off-by: Andrea Cervesato <andrea.cervesato@suse.com>
---
 runtest/kernel_misc                    |   1 +
 testcases/kernel/coredump/.gitignore   |   1 +
 testcases/kernel/coredump/coredump02.c | 198 +++++++++++++++++++++++++++++++++
 3 files changed, 200 insertions(+)

diff --git a/runtest/kernel_misc b/runtest/kernel_misc
index ecf9ee2a2..3311e1929 100644
--- a/runtest/kernel_misc
+++ b/runtest/kernel_misc
@@ -18,3 +18,4 @@ zram03 zram03
 umip_basic_test umip_basic_test
 aslr01 aslr01
 coredump01 coredump01
+coredump02 coredump02
diff --git a/testcases/kernel/coredump/.gitignore b/testcases/kernel/coredump/.gitignore
index cd0b51700..e241a112e 100644
--- a/testcases/kernel/coredump/.gitignore
+++ b/testcases/kernel/coredump/.gitignore
@@ -1,2 +1,3 @@
 /coredump01
 /coredump01_helper
+/coredump02
diff --git a/testcases/kernel/coredump/coredump02.c b/testcases/kernel/coredump/coredump02.c
new file mode 100644
index 000000000..2980bc8ca
--- /dev/null
+++ b/testcases/kernel/coredump/coredump02.c
@@ -0,0 +1,198 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2026 Linux Test Project
+ */
+
+/*\
+ * Verify the ELF structure and note content of a kernel-generated core
+ * dump.
+ *
+ * A core dump written by the kernel is an ELF file of type ET_CORE. It
+ * contains:
+ *
+ * - one PT_NOTE program header holding process metadata
+ * - one or more PT_LOAD program headers for the mapped memory segments
+ *
+ * The PT_NOTE segment carries a stream of notes. Two of them describe
+ * the crashed process:
+ *
+ * - NT_PRPSINFO, whose descriptor is a struct elf_prpsinfo. The pr_fname
+ *   field holds the executable name.
+ * - NT_PRSTATUS, whose descriptor is a struct elf_prstatus. The pr_pid
+ *   field holds the PID and pr_cursig holds the terminating signal.
+ *
+ * The crashed child is a fork of the test binary, so the core dump has
+ * the same ELF class as the test. ElfW() is therefore safe here.
+ *
+ * The test needs root because it rewrites the system-wide core_pattern.
+ * The original value is saved and restored by the test library on all
+ * exit paths.
+ *
+ * [Algorithm]
+ *
+ * - Point core_pattern into the test temporary directory
+ * - Fork a child which aborts itself to produce a core dump
+ * - Read the core file into memory and parse the ELF header
+ * - Walk the program headers and verify PT_NOTE and PT_LOAD are present
+ * - Walk the notes in every PT_NOTE segment
+ * - Check that NT_PRPSINFO carries the expected executable name
+ * - Check that NT_PRSTATUS carries the expected PID and signal
+ */
+
+#include <link.h>
+#include <sys/procfs.h>
+
+#include "coredump_common.h"
+
+#define NOTE_ALIGN(x) (((x) + 3) & ~3U)
+
+static char *core_buf;
+static size_t core_len;
+static int prpsinfo_seen, prstatus_seen;
+
+static void load_core(const char *path)
+{
+	struct stat st;
+	int fd;
+
+	SAFE_STAT(path, &st);
+	if (st.st_size <= 0)
+		tst_brk(TFAIL, "core file %s is empty", path);
+
+	core_len = st.st_size;
+	core_buf = SAFE_MALLOC(core_len);
+
+	fd = SAFE_OPEN(path, O_RDONLY);
+	SAFE_READ(1, fd, core_buf, core_len);
+	SAFE_CLOSE(fd);
+}
+
+static void unload_core(void)
+{
+	free(core_buf);
+	core_buf = NULL;
+	core_len = 0;
+}
+
+static const void *core_at(size_t off, size_t need)
+{
+	if (off > core_len || need > core_len - off)
+		tst_brk(TFAIL, "core file truncated at %zu (need %zu, have %zu)",
+			off, need, core_len);
+
+	return core_buf + off;
+}
+
+static void handle_note(uint32_t type, const void *desc, size_t descsz, pid_t pid)
+{
+	if (type == NT_PRPSINFO && descsz >= sizeof(struct elf_prpsinfo)) {
+		struct elf_prpsinfo info;
+		char name[sizeof(info.pr_fname) + 1] = {0};
+
+		memcpy(&info, desc, sizeof(info));
+		memcpy(name, info.pr_fname, sizeof(info.pr_fname));
+
+		TST_EXP_EQ_STR(name, "coredump02");
+
+		prpsinfo_seen = 1;
+	} else if (type == NT_PRSTATUS && descsz >= sizeof(struct elf_prstatus)) {
+		struct elf_prstatus st;
+
+		memcpy(&st, desc, sizeof(st));
+
+		TST_EXP_EQ_LI(st.pr_pid, pid);
+		TST_EXP_EQ_LI(st.pr_cursig, SIGABRT);
+
+		prstatus_seen = 1;
+	}
+}
+
+static void walk_notes(size_t off, size_t size, pid_t pid)
+{
+	size_t pos = 0;
+
+	while (pos + sizeof(ElfW(Nhdr)) <= size) {
+		const ElfW(Nhdr) *nh = core_at(off + pos, sizeof(*nh));
+		size_t np = NOTE_ALIGN(nh->n_namesz);
+		size_t dp = NOTE_ALIGN(nh->n_descsz);
+		size_t total = sizeof(*nh) + np + dp;
+
+		if (pos + total > size)
+			tst_brk(TFAIL, "note extends past PT_NOTE (pos=%zu total=%zu size=%zu)",
+				pos, total, size);
+
+		handle_note(nh->n_type,
+			    core_at(off + pos + sizeof(*nh) + np, nh->n_descsz),
+			    nh->n_descsz, pid);
+
+		pos += total;
+	}
+}
+
+static void run(void)
+{
+	char dump[PATH_MAX + 32];
+	int pt_note_cnt = 0, have_load = 0;
+	const ElfW(Ehdr) *eh;
+	const ElfW(Phdr) *ph;
+	pid_t pid;
+	size_t i;
+
+	prpsinfo_seen = prstatus_seen = 0;
+
+	set_pattern("%s/core.%%p", cwd);
+
+	pid = crash_child();
+
+	snprintf(dump, sizeof(dump), "%s/core.%d", cwd, pid);
+	load_core(dump);
+
+	eh = core_at(0, sizeof(*eh));
+
+	TST_EXP_EXPR(!memcmp(eh->e_ident, ELFMAG, SELFMAG), "core has ELF magic");
+	TST_EXP_EQ_LI(eh->e_type, ET_CORE);
+
+	if (!eh->e_phnum)
+		tst_brk(TFAIL, "core has no program headers");
+
+	ph = core_at(eh->e_phoff, (size_t)eh->e_phnum * sizeof(*ph));
+
+	for (i = 0; i < eh->e_phnum; i++) {
+		if (ph[i].p_type == PT_NOTE) {
+			pt_note_cnt++;
+			walk_notes(ph[i].p_offset, ph[i].p_filesz, pid);
+		} else if (ph[i].p_type == PT_LOAD) {
+			have_load = 1;
+		}
+	}
+
+	TST_EXP_EQ_LI(pt_note_cnt, 1);
+	TST_EXP_EQ_LI(have_load, 1);
+	TST_EXP_EQ_LI(prpsinfo_seen, 1);
+	TST_EXP_EQ_LI(prstatus_seen, 1);
+
+	SAFE_UNLINK(dump);
+	unload_core();
+}
+
+static void cleanup(void)
+{
+	unload_core();
+}
+
+static struct tst_test test = {
+	.test_all = run,
+	.setup = coredump_setup,
+	.cleanup = cleanup,
+	.needs_root = 1,
+	.needs_tmpdir = 1,
+	.forks_child = 1,
+	.needs_kconfigs = (const char *[]) {
+		"CONFIG_COREDUMP=y",
+		NULL,
+	},
+	.save_restore = (const struct tst_path_val[]) {
+		{PATH_KERN_CORE_PATTERN, NULL, TST_SR_TCONF},
+		{}
+	},
+};

-- 
2.51.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Fri, 4 Sep 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

--- [PATCH 1/2] ---

> static void run(unsigned int n)
> {
> 	if (n)
> 		verify_pipe_pattern();
> 	else
> 		verify_file_pattern();
> }
>
> static struct tst_test test = {
> 	.test = run,
> 	.tcnt = 2,

This manually dispatches separate functions for the two test cases. Use a
single struct tcase array and a generic .test callback, with
.tcnt = ARRAY_SIZE(tcases), as required for multi-case tests.

--- [PATCH 2/2] ---

> 		} else if (ph[i].p_type == PT_LOAD) {
> 			have_load = 1;
> 		}

This only proves that a PT_LOAD header exists, not that its data is present.
A short dump_emit() after the headers and notes leaves WCOREDUMP set, so a
truncated core can pass every check here. Validate p_offset + p_filesz against
core_len with overflow-safe arithmetic for every file-backed segment.

Verdict - Needs revision

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[Andrea Cervesato via ltp]

> 
> This manually dispatches separate functions for the two test cases. Use a
> single struct tcase array and a generic .test callback, with
> .tcnt = ARRAY_SIZE(tcases), as required for multi-case tests.

no need for this.

> 
> --- [PATCH 2/2] ---
> 
> > 		} else if (ph[i].p_type == PT_LOAD) {
> > 			have_load = 1;
> > 		}
> 
> This only proves that a PT_LOAD header exists, not that its data is present.
> A short dump_emit() after the headers and notes leaves WCOREDUMP set, so a
> truncated core can pass every check here. Validate p_offset + p_filesz against
> core_len with overflow-safe arithmetic for every file-backed segment.
> 

this is valid

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v7 1/2] coredump01: New core_pattern specifiers test

[Li Wang]

Reviewed-by: Li Wang <li.wang@linux.dev>

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v7 2/2] coredump02: Verify ELF structure and notes

[Li Wang]

Reviewed-by: Li Wang <li.wang@linux.dev>

-- 
Regards,
Li Wang

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[Andrea Cervesato via ltp]

> --- [PATCH 2/2] ---
> 
> > 		} else if (ph[i].p_type == PT_LOAD) {
> > 			have_load = 1;
> > 		}
> 
> This only proves that a PT_LOAD header exists, not that its data is present.
> A short dump_emit() after the headers and notes leaves WCOREDUMP set, so a
> truncated core can pass every check here. Validate p_offset + p_filesz against
> core_len with overflow-safe arithmetic for every file-backed segment.

Sent a v8 fixing this

--
Andrea Cervesato
SUSE QE Automation Engineer Linux
andrea.cervesato@suse.com

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] coredump01: New core_pattern specifiers test

[linuxtestproject.agent]

Hi Andrea,

On Sep 8, 2026, Andrea Cervesato wrote:
> coredump01: New core_pattern specifiers test

Verdict - Reviewed

---
Note:

The agent can sometimes produce false positives although often its
findings are genuine. If you find issues with the review, please
comment this email or ignore the suggestions.

Regards,
LTP AI Reviewer

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp