syscalls/execve06: Add test for argv[0] = NULL
PATCH 1/1 syscalls/execve06: Add test for argv[0] = NULL
Adds a test that kernel sets argv[0] to a dummy empty string if NULL was passed to the execve() syscall. This was introduced in commit:
commit dcd46d897adb70d63e025f175a00a89797d31a43
Author: Kees Cook <[email protected]> Date: Mon Jan 31 16:09:47 2022 -0800
exec: Force single empty string when argv is empty
in order to fix all potential CVEs where userspace programs attempt to blindly process the argv[] list starting at argv[1]. There was at least one example of this caught in the wild CVE-2021-4034 in polkit but there are likely more.
--- testcases/kernel/syscalls/execve/.gitignore | 2 + testcases/kernel/syscalls/execve/execve06.c | 49 +++++++++++++++++++ .../kernel/syscalls/execve/execve06_child.c | 27 ++++++++++ 3 files changed, 78 insertions(+) create mode 100644 testcases/kernel/syscalls/execve/execve06.c create mode 100644 testcases/kernel/syscalls/execve/execve06_child.c
testcases/kernel/syscalls/execve/.gitignore +2 -0
@@ -4,4 +4,6 @@ /execve03 /execve04 /execve05 +/execve06 +/execve06_child /execve_child
testcases/kernel/syscalls/execve/execve06.c new file +49 -0
49 lines not on this page · raw message
testcases/kernel/syscalls/execve/execve06_child.c new file +27 -0
@@ -0,0 +1,27 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* +* Copyright (C) 2022 Cyril Hrubis [email protected] +*/ + +#define TST_NO_DEFAULT_MAIN +#include <stdlib.h> +#include "tst_test.h" + +int main (int argc,char *argv[]) +{ +tst_reinit (); + +if (argc !=1 ) { +tst_res (TFAIL ,"argc is %d, expected 1" , argc); +return 0 ; + } + +if (!argv[0 ]) { +tst_res (TFAIL ,"argv[0] == NULL" ); +return 0 ; + } + +tst_res (TPASS ,"argv[0] was filled in by kernel" ); + +return 0 ; +}
Signature · 5 lines
2.35.1 -- Mailing list info: https://lists.linux.it/listinfo/ltp
HI Cyril,
Quoted 7 lines
Adds a test that kernel sets argv[0] to a dummy empty string if NULL was passed to the
execve()syscall. This was introduced in commit:commit
dcd46d897adb70d63e025f175a00a89797d31a43Author: Kees Cook <[email protected]> Date: Mon Jan 31 16:09:47 2022 -0800
Quoted 6 lines
exec: Force single empty string when argv is empty
in order to fix all potential CVEs where userspace programs attempt to blindly process the argv[] list starting at argv[1]. There was at least one example of this caught in the wild CVE-2021-4034 in polkit but there are likely more.
Great, thanks for addressing this.
Quoted 8 lines
Fixes #911testcases/kernel/syscalls/execve/.gitignore | 2 + testcases/kernel/syscalls/execve/execve06.c | 49 +++++++++++++++++++ .../kernel/syscalls/execve/execve06_child.c | 27 ++++++++++ 3 files changed, 78 insertions(+) create mode 100644 testcases/kernel/syscalls/execve/execve06.c create mode 100644 testcases/kernel/syscalls/execve/execve06_child.c
This should go to runtest/syscalls and runtest/cve, right? (can be fixed before merge).
Kind regards,
Petr
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
Hi!
Quoted 16 lines
Quoted 4 lines
in order to fix all potential CVEs where userspace programs attempt to blindly process the argv[] list starting at argv[1]. There was at least one example of this caught in the wild CVE-2021-4034 in polkit but there are likely more.
Great, thanks for addressing this.
Quoted 8 lines
Fixes #911testcases/kernel/syscalls/execve/.gitignore | 2 + testcases/kernel/syscalls/execve/execve06.c | 49 +++++++++++++++++++ .../kernel/syscalls/execve/execve06_child.c | 27 ++++++++++ 3 files changed, 78 insertions(+) create mode 100644 testcases/kernel/syscalls/execve/execve06.c create mode 100644 testcases/kernel/syscalls/execve/execve06_child.c
Quoted 2 lines
This should go to runtest/syscalls and runtest/cve, right? (can be fixed before merge).
Sigh, sorry, I did forget about these.
Signature · 5 lines
Cyril Hrubis [email protected] -- Mailing list info: https://lists.linux.it/listinfo/ltp
Hi Cyril,
...
Quoted 4 lines
in order to fix all potential CVEs where userspace programs attempt to blindly process the argv[] list starting at argv[1]. There was at least one example of this caught in the wild CVE-2021-4034 in polkit but there are likely more.
Thanks for implementing this!
Quoted 1 line
Fixes #911
...
Quoted 8 lines
testcases/kernel/syscalls/execve/execve06.c new file +2
-0@@ -0,0 +1,49 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/*
testcases/kernel/syscalls/execve/execve06.c line 12
Quoted 10 lines
testcases/kernel/syscalls/execve/execve06.c new file +10
-0+* Copyright (c) 2022 Cyril Hrubis <[email protected]> +*/ + +/*\ +* [Description] +* +* Test that kernel adds dummy argv[0] if empty argument list was passed to +* execve(). This fixes at least one CVE where userspace programs start to +* process argument list blindly from argv[1] such as polkit pkexec +* CVE-2021-4034.
We might also add link to LWM article :)
https://lwn.net/Articles/883547/
Quoted 9 lines
testcases/kernel/syscalls/execve/execve06.c new file +9
-0+*/ + +#include <stdlib.h> +#include <stdio.h> +#include "tst_test.h" + +static void verify_execve (void ) +{ +pid_t pid;
testcases/kernel/syscalls/execve/execve06.c line 31
Quoted 10 lines
testcases/kernel/syscalls/execve/execve06.c new file +10
-0+char path[512 ]; +char ipc_env_var[1024 ]; + +sprintf (ipc_env_var,IPC_ENV_VAR "=%s" ,getenv (IPC_ENV_VAR )); + +char *const envp[] = {ipc_env_var,NULL }; +char *const argv[] = {NULL }; + +if (tst_get_path ("execve06_child" , path,sizeof (path))) +tst_brk (TCONF ,"Couldn't find execve06_child in $PATH" );
I wonder why this does not work:
.needs_cmds = (const char *[]) {
"execve06_child",
NULL
},tst_test.c:1526: TINFO: Timeout per run is 0h 00m 30s execve06.c:33: TFAIL: Failed to execute execl01_child: ENOENT (2) tst_test.c:395: TBROK: Invalid child (23451) exit value 1
Quoted 24 lines
testcases/kernel/syscalls/execve/execve06.c new file +18
-0+ + pid =SAFE_FORK (); +if (pid ==0 ) { +execve (path, argv, envp); +tst_brk (TFAIL |TERRNO ,"Failed to execute execl01_child" ); + } +} + +static struct tst_test test = { + .forks_child =1 , + .child_needs_reinit =1 , + .test_all = verify_execve, + .tags = (const struct tst_tag []) { + {"linux-git" ,"dcd46d897adb" }, + {"CVE" ,"2021-4034" }, + {} + } +};testcases/kernel/syscalls/execve/execve06_child.c new file +0
-0
testcases/kernel/syscalls/execve/execve06_child.c line 3
Quoted 3 lines
testcases/kernel/syscalls/execve/execve06_child.c new file +3
-0@@ -0,0 +1,27 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* +* Copyright (C) 2022 Cyril Hrubis [email protected]
very nit: <[email protected]> (missing < >)
Quoted 11 lines
testcases/kernel/syscalls/execve/execve06_child.c new file +11
-0+*/ + +#define TST_NO_DEFAULT_MAIN +#include <stdlib.h> +#include "tst_test.h" + +int main (int argc,char *argv[]) +{ +tst_reinit (); + +if (argc !=1 ) {
testcases/kernel/syscalls/execve/execve06_child.c line 24
Quoted 10 lines
testcases/kernel/syscalls/execve/execve06_child.c new file +10
-0+tst_res (TFAIL ,"argc is %d, expected 1" , argc); +return 0 ; + } + +if (!argv[0 ]) { +tst_res (TFAIL ,"argv[0] == NULL" ); +return 0 ; + } + +tst_res (TPASS ,"argv[0] was filled in by kernel" );
Testing matches the description from kernel commit.
Maybe also test for argv[0] being empty string (to make sure behavior does not change, although unlikely it'd change)?
I tested it on various kernels, works as expected.
Kind regards,
Petr
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
Hi!
Quoted 9 lines
I wonder why this does not work:
.needs_cmds = (const char *[]) { "execve06_child", NULL },tst_test.c:1526: TINFO: Timeout per run is 0h 00m 30s execve06.c:33: TFAIL: Failed to execute
execl01_child: ENOENT (2) tst_test.c:395: TBROK: Invalid child (23451) exit value 1
The difference is because the test library adds current directory and test start directory into $PATH before a test is executed, see add_paths() in the tst_test.c.
The .needs_cmds is for binaries installed on the system in /usr/bin/ etc. so it does not make much sense to call the add_paths() before we process the array.
Quoted 20 lines
Quoted 20 lines
testcases/kernel/syscalls/execve/execve06.c new file +18
-0+ + pid =SAFE_FORK (); +if (pid ==0 ) { +execve (path, argv, envp); +tst_brk (TFAIL |TERRNO ,"Failed to execute execl01_child" ); + } +} + +static struct tst_test test = { + .forks_child =1 , + .child_needs_reinit =1 , + .test_all = verify_execve, + .tags = (const struct tst_tag []) { + {"linux-git" ,"dcd46d897adb" }, + {"CVE" ,"2021-4034" }, + {} + } +};testcases/kernel/syscalls/execve/execve06_child.c new file +0
-0
Quoted 8 lines
testcases/kernel/syscalls/execve/execve06_child.c line 3
Quoted 7 lines
index 000000000..17280d58a
testcases/kernel/syscalls/execve/execve06_child.c new file +3
-0@@ -0,0 +1,27 @@ +// SPDX-License-Identifier: GPL-2.0-or-later +/* +* Copyright (C) 2022 Cyril Hrubis [email protected] very nit: <[email protected]> (missing < >)
Will fix.
Quoted 17 lines
Quoted 17 lines
testcases/kernel/syscalls/execve/execve06_child.c new file +17
-0+*/ + +#define TST_NO_DEFAULT_MAIN +#include <stdlib.h> +#include "tst_test.h" + +int main (int argc,char *argv[]) +{ +tst_reinit (); + +if (argc !=1 ) { +tst_res (TFAIL ,"argc is %d, expected 1" , argc); +return 0 ; + } + +if (!argv[0 ]) { +tst_res (TFAIL ,"argv[0] == NULL" );
Quoted 8 lines
testcases/kernel/syscalls/execve/execve06_child.c line 24
Quoted 4 lines
testcases/kernel/syscalls/execve/execve06_child.c new file +4
-0+return 0 ; + } + +tst_res (TPASS ,"argv[0] was filled in by kernel" );Testing matches the description from kernel commit.
Maybe also test for argv[0] being empty string (to make sure behavior does not change, although unlikely it'd change)?
It should be safe either way but I decided to be less specific and only check that something have been added there.
Quoted 4 lines
I tested it on various kernels, works as expected.
Reviewed-by Petr Vorel <[email protected]>Tested-by Petr Vorel <[email protected]>
I will push it with the added runtest entries and fixed copyright then.
Signature · 5 lines
Cyril Hrubis [email protected] -- Mailing list info: https://lists.linux.it/listinfo/ltp
Hi Cyril,
Quoted 10 lines
Quoted 9 lines
I wonder why this does not work:
.needs_cmds = (const char *[]) { "execve06_child", NULL },tst_test.c:1526: TINFO: Timeout per run is 0h 00m 30s execve06.c:33: TFAIL: Failed to execute
execl01_child: ENOENT (2) tst_test.c:395: TBROK: Invalid child (23451) exit value 1
Quoted 7 lines
The difference is because the test library adds current directory and test start directory into $PATH before a test is executed, see
add_paths()in thetst_test.c.The .needs_cmds is for binaries installed on the system in
/usr/bin/etc. so it does not make much sense to call theadd_paths()before we process the array.
Thx!
...
Quoted 6 lines
Quoted 3 lines
Testing matches the description from kernel commit.
Maybe also test for argv[0] being empty string (to make sure behavior does not change, although unlikely it'd change)?It should be safe either way but I decided to be less specific and only check that something have been added there.
Sure.
Quoted 6 lines
Quoted 4 lines
I tested it on various kernels, works as expected.
Reviewed-by Petr Vorel <[email protected]>Tested-by Petr Vorel <[email protected]>I will push it with the added runtest entries and fixed copyright then.
Thx!
Kind regards,
Petr
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp