Linux Kernel Selftest development
 help / color / mirror / Atom feed
* [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED
@ 2026-05-13  9:56 Hongfu Li
  2026-05-13 10:05 ` David Hildenbrand (Arm)
  2026-05-13 10:57 ` Lorenzo Stoakes
  0 siblings, 2 replies; 5+ messages in thread
From: Hongfu Li @ 2026-05-13  9:56 UTC (permalink / raw)
  To: akpm, david, ljs, liam, vbabka, rppt, surenb, mhocko, shuah
  Cc: linux-mm, linux-kselftest, linux-kernel, Hongfu Li, Dev Jain

mmap() returns MAP_FAILED, which is defined as (void *)-1, on error,
not NULL.  Several selftests incorrectly check the return value of
mmap() using !ptr or ptr == NULL, which would erroneously treat
MAP_FAILED as a valid pointer since MAP_FAILED is non-zero and
non-NULL.  This can lead to segfaults when mmap() actually fails
under memory pressure.

Signed-off-by: Hongfu Li <lihongfu@kylinos.cn>
Reviewed-by: Dev Jain <dev.jain@arm.com>
---
v2:
- Add missing mmap() return value checks in pkey_sighandler_tests.c
  and protection_keys.c
---
 tools/testing/selftests/mm/ksm_tests.c             | 2 +-
 tools/testing/selftests/mm/madv_populate.c         | 2 +-
 tools/testing/selftests/mm/pkey_sighandler_tests.c | 2 ++
 tools/testing/selftests/mm/protection_keys.c       | 1 +
 tools/testing/selftests/mm/soft-dirty.c            | 4 ++--
 tools/testing/selftests/mm/vm_util.c               | 2 +-
 6 files changed, 8 insertions(+), 5 deletions(-)

diff --git a/tools/testing/selftests/mm/ksm_tests.c b/tools/testing/selftests/mm/ksm_tests.c
index a0b48b839d54..ed481f817282 100644
--- a/tools/testing/selftests/mm/ksm_tests.c
+++ b/tools/testing/selftests/mm/ksm_tests.c
@@ -174,7 +174,7 @@ static void  *allocate_memory(void *ptr, int prot, int mapping, char data, size_
 {
 	void *map_ptr = mmap(ptr, map_size, PROT_WRITE, mapping, -1, 0);
 
-	if (!map_ptr) {
+	if (map_ptr == MAP_FAILED) {
 		perror("mmap");
 		return NULL;
 	}
diff --git a/tools/testing/selftests/mm/madv_populate.c b/tools/testing/selftests/mm/madv_populate.c
index 88050e0f829a..7fce5d0b622b 100644
--- a/tools/testing/selftests/mm/madv_populate.c
+++ b/tools/testing/selftests/mm/madv_populate.c
@@ -34,7 +34,7 @@ static void sense_support(void)
 
 	addr = mmap(0, pagesize, PROT_READ | PROT_WRITE,
 		    MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
-	if (!addr)
+	if (addr == MAP_FAILED)
 		ksft_exit_fail_msg("mmap failed\n");
 
 	ret = madvise(addr, pagesize, MADV_POPULATE_READ);
diff --git a/tools/testing/selftests/mm/pkey_sighandler_tests.c b/tools/testing/selftests/mm/pkey_sighandler_tests.c
index 302fef54049c..4637809192f9 100644
--- a/tools/testing/selftests/mm/pkey_sighandler_tests.c
+++ b/tools/testing/selftests/mm/pkey_sighandler_tests.c
@@ -317,6 +317,7 @@ static void test_sigsegv_handler_with_different_pkey_for_stack(void)
 	/* Set up alternate signal stack that will use the default MPK */
 	sigstack.ss_sp = mmap(0, STACK_SIZE, PROT_READ | PROT_WRITE,
 			      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	assert(sigstack.ss_sp != MAP_FAILED);
 	sigstack.ss_flags = 0;
 	sigstack.ss_size = STACK_SIZE;
 
@@ -490,6 +491,7 @@ static void test_pkru_sigreturn(void)
 	/* Set up alternate signal stack that will use the default MPK */
 	sigstack.ss_sp = mmap(0, STACK_SIZE, PROT_READ | PROT_WRITE,
 			      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+	assert(sigstack.ss_sp != MAP_FAILED);
 	sigstack.ss_flags = 0;
 	sigstack.ss_size = STACK_SIZE;
 
diff --git a/tools/testing/selftests/mm/protection_keys.c b/tools/testing/selftests/mm/protection_keys.c
index 2085982dba69..580bf1668c71 100644
--- a/tools/testing/selftests/mm/protection_keys.c
+++ b/tools/testing/selftests/mm/protection_keys.c
@@ -1217,6 +1217,7 @@ static void arch_force_pkey_reg_init(void)
 	 * doing the XSAVE size enumeration dance.
 	 */
 	buf = mmap(NULL, 1*MB, PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
+	pkey_assert(buf != (void *)-1);
 
 	/* These __builtins require compiling with -mxsave */
 
diff --git a/tools/testing/selftests/mm/soft-dirty.c b/tools/testing/selftests/mm/soft-dirty.c
index bcfcac99b436..67c26c265880 100644
--- a/tools/testing/selftests/mm/soft-dirty.c
+++ b/tools/testing/selftests/mm/soft-dirty.c
@@ -143,7 +143,7 @@ static void test_mprotect(int pagemap_fd, int pagesize, bool anon)
 	if (anon) {
 		map = mmap(NULL, pagesize, PROT_READ|PROT_WRITE,
 			   MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
-		if (!map)
+		if (map == MAP_FAILED)
 			ksft_exit_fail_msg("anon mmap failed\n");
 	} else {
 		test_fd = open(fname, O_RDWR | O_CREAT, 0664);
@@ -155,7 +155,7 @@ static void test_mprotect(int pagemap_fd, int pagesize, bool anon)
 		ftruncate(test_fd, pagesize);
 		map = mmap(NULL, pagesize, PROT_READ|PROT_WRITE,
 			   MAP_SHARED, test_fd, 0);
-		if (!map)
+		if (map == MAP_FAILED)
 			ksft_exit_fail_msg("file mmap failed\n");
 	}
 
diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
index db94564f4431..63aaa2d9ec0b 100644
--- a/tools/testing/selftests/mm/vm_util.c
+++ b/tools/testing/selftests/mm/vm_util.c
@@ -463,7 +463,7 @@ bool softdirty_supported(void)
 	/* New mappings are expected to be marked with VM_SOFTDIRTY (sd). */
 	addr = mmap(0, pagesize, PROT_READ | PROT_WRITE,
 		    MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
-	if (!addr)
+	if (addr == MAP_FAILED)
 		ksft_exit_fail_msg("mmap failed\n");
 
 	supported = check_vmflag(addr, "sd");
-- 
2.25.1


^ permalink raw reply related	[flat|nested] 5+ messages in thread

* Re: [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED
  2026-05-13  9:56 [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED Hongfu Li
@ 2026-05-13 10:05 ` David Hildenbrand (Arm)
  2026-05-13 11:15   ` Hongfu Li
  2026-05-13 10:57 ` Lorenzo Stoakes
  1 sibling, 1 reply; 5+ messages in thread
From: David Hildenbrand (Arm) @ 2026-05-13 10:05 UTC (permalink / raw)
  To: Hongfu Li, akpm, ljs, liam, vbabka, rppt, surenb, mhocko, shuah
  Cc: linux-mm, linux-kselftest, linux-kernel, Dev Jain

On 5/13/26 11:56, Hongfu Li wrote:
> mmap() returns MAP_FAILED, which is defined as (void *)-1, on error,
> not NULL.  Several selftests incorrectly check the return value of
> mmap() using !ptr or ptr == NULL, which would erroneously treat
> MAP_FAILED as a valid pointer since MAP_FAILED is non-zero and
> non-NULL.  This can lead to segfaults when mmap() actually fails
> under memory pressure.

Well, your patch also adds more checks where we previously didn't have any checks?

> 
> Signed-off-by: Hongfu Li <lihongfu@kylinos.cn>
> Reviewed-by: Dev Jain <dev.jain@arm.com>
> ---
> v2:
> - Add missing mmap() return value checks in pkey_sighandler_tests.c
>   and protection_keys.c
> ---

[...]

> diff --git a/tools/testing/selftests/mm/protection_keys.c b/tools/testing/selftests/mm/protection_keys.c
> index 2085982dba69..580bf1668c71 100644
> --- a/tools/testing/selftests/mm/protection_keys.c
> +++ b/tools/testing/selftests/mm/protection_keys.c
> @@ -1217,6 +1217,7 @@ static void arch_force_pkey_reg_init(void)
>  	 * doing the XSAVE size enumeration dance.
>  	 */
>  	buf = mmap(NULL, 1*MB, PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
> +	pkey_assert(buf != (void *)-1);

What's the reason for not using MAP_FAILED?

-- 
Cheers,

David

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED
  2026-05-13  9:56 [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED Hongfu Li
  2026-05-13 10:05 ` David Hildenbrand (Arm)
@ 2026-05-13 10:57 ` Lorenzo Stoakes
  2026-05-13 11:20   ` Hongfu Li
  1 sibling, 1 reply; 5+ messages in thread
From: Lorenzo Stoakes @ 2026-05-13 10:57 UTC (permalink / raw)
  To: Hongfu Li
  Cc: akpm, david, liam, vbabka, rppt, surenb, mhocko, shuah, linux-mm,
	linux-kselftest, linux-kernel, Dev Jain

On Wed, May 13, 2026 at 05:56:09PM +0800, Hongfu Li wrote:
> mmap() returns MAP_FAILED, which is defined as (void *)-1, on error,
> not NULL.  Several selftests incorrectly check the return value of
> mmap() using !ptr or ptr == NULL, which would erroneously treat
> MAP_FAILED as a valid pointer since MAP_FAILED is non-zero and
> non-NULL.  This can lead to segfaults when mmap() actually fails
> under memory pressure.
>
> Signed-off-by: Hongfu Li <lihongfu@kylinos.cn>
> Reviewed-by: Dev Jain <dev.jain@arm.com>

You forgot my tag :)

Also, please do not send a respin right away if you can, multiple revisions in
one day can be tricky to deal with.

As per David's points - please update the commit message to reflect the fact
you've added a missing assert too and replace the (void *)-1 with MAP_FAILED.

Cheers, Lorenzo

> ---
> v2:
> - Add missing mmap() return value checks in pkey_sighandler_tests.c
>   and protection_keys.c
> ---
>  tools/testing/selftests/mm/ksm_tests.c             | 2 +-
>  tools/testing/selftests/mm/madv_populate.c         | 2 +-
>  tools/testing/selftests/mm/pkey_sighandler_tests.c | 2 ++
>  tools/testing/selftests/mm/protection_keys.c       | 1 +
>  tools/testing/selftests/mm/soft-dirty.c            | 4 ++--
>  tools/testing/selftests/mm/vm_util.c               | 2 +-
>  6 files changed, 8 insertions(+), 5 deletions(-)
>
> diff --git a/tools/testing/selftests/mm/ksm_tests.c b/tools/testing/selftests/mm/ksm_tests.c
> index a0b48b839d54..ed481f817282 100644
> --- a/tools/testing/selftests/mm/ksm_tests.c
> +++ b/tools/testing/selftests/mm/ksm_tests.c
> @@ -174,7 +174,7 @@ static void  *allocate_memory(void *ptr, int prot, int mapping, char data, size_
>  {
>  	void *map_ptr = mmap(ptr, map_size, PROT_WRITE, mapping, -1, 0);
>
> -	if (!map_ptr) {
> +	if (map_ptr == MAP_FAILED) {
>  		perror("mmap");
>  		return NULL;
>  	}
> diff --git a/tools/testing/selftests/mm/madv_populate.c b/tools/testing/selftests/mm/madv_populate.c
> index 88050e0f829a..7fce5d0b622b 100644
> --- a/tools/testing/selftests/mm/madv_populate.c
> +++ b/tools/testing/selftests/mm/madv_populate.c
> @@ -34,7 +34,7 @@ static void sense_support(void)
>
>  	addr = mmap(0, pagesize, PROT_READ | PROT_WRITE,
>  		    MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
> -	if (!addr)
> +	if (addr == MAP_FAILED)
>  		ksft_exit_fail_msg("mmap failed\n");
>
>  	ret = madvise(addr, pagesize, MADV_POPULATE_READ);
> diff --git a/tools/testing/selftests/mm/pkey_sighandler_tests.c b/tools/testing/selftests/mm/pkey_sighandler_tests.c
> index 302fef54049c..4637809192f9 100644
> --- a/tools/testing/selftests/mm/pkey_sighandler_tests.c
> +++ b/tools/testing/selftests/mm/pkey_sighandler_tests.c
> @@ -317,6 +317,7 @@ static void test_sigsegv_handler_with_different_pkey_for_stack(void)
>  	/* Set up alternate signal stack that will use the default MPK */
>  	sigstack.ss_sp = mmap(0, STACK_SIZE, PROT_READ | PROT_WRITE,
>  			      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
> +	assert(sigstack.ss_sp != MAP_FAILED);
>  	sigstack.ss_flags = 0;
>  	sigstack.ss_size = STACK_SIZE;
>
> @@ -490,6 +491,7 @@ static void test_pkru_sigreturn(void)
>  	/* Set up alternate signal stack that will use the default MPK */
>  	sigstack.ss_sp = mmap(0, STACK_SIZE, PROT_READ | PROT_WRITE,
>  			      MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
> +	assert(sigstack.ss_sp != MAP_FAILED);
>  	sigstack.ss_flags = 0;
>  	sigstack.ss_size = STACK_SIZE;
>
> diff --git a/tools/testing/selftests/mm/protection_keys.c b/tools/testing/selftests/mm/protection_keys.c
> index 2085982dba69..580bf1668c71 100644
> --- a/tools/testing/selftests/mm/protection_keys.c
> +++ b/tools/testing/selftests/mm/protection_keys.c
> @@ -1217,6 +1217,7 @@ static void arch_force_pkey_reg_init(void)
>  	 * doing the XSAVE size enumeration dance.
>  	 */
>  	buf = mmap(NULL, 1*MB, PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
> +	pkey_assert(buf != (void *)-1);

Should use MAP_FAILED here as David pointed out.

>
>  	/* These __builtins require compiling with -mxsave */
>
> diff --git a/tools/testing/selftests/mm/soft-dirty.c b/tools/testing/selftests/mm/soft-dirty.c
> index bcfcac99b436..67c26c265880 100644
> --- a/tools/testing/selftests/mm/soft-dirty.c
> +++ b/tools/testing/selftests/mm/soft-dirty.c
> @@ -143,7 +143,7 @@ static void test_mprotect(int pagemap_fd, int pagesize, bool anon)
>  	if (anon) {
>  		map = mmap(NULL, pagesize, PROT_READ|PROT_WRITE,
>  			   MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
> -		if (!map)
> +		if (map == MAP_FAILED)
>  			ksft_exit_fail_msg("anon mmap failed\n");
>  	} else {
>  		test_fd = open(fname, O_RDWR | O_CREAT, 0664);
> @@ -155,7 +155,7 @@ static void test_mprotect(int pagemap_fd, int pagesize, bool anon)
>  		ftruncate(test_fd, pagesize);
>  		map = mmap(NULL, pagesize, PROT_READ|PROT_WRITE,
>  			   MAP_SHARED, test_fd, 0);
> -		if (!map)
> +		if (map == MAP_FAILED)
>  			ksft_exit_fail_msg("file mmap failed\n");
>  	}
>
> diff --git a/tools/testing/selftests/mm/vm_util.c b/tools/testing/selftests/mm/vm_util.c
> index db94564f4431..63aaa2d9ec0b 100644
> --- a/tools/testing/selftests/mm/vm_util.c
> +++ b/tools/testing/selftests/mm/vm_util.c
> @@ -463,7 +463,7 @@ bool softdirty_supported(void)
>  	/* New mappings are expected to be marked with VM_SOFTDIRTY (sd). */
>  	addr = mmap(0, pagesize, PROT_READ | PROT_WRITE,
>  		    MAP_ANONYMOUS | MAP_PRIVATE, 0, 0);
> -	if (!addr)
> +	if (addr == MAP_FAILED)
>  		ksft_exit_fail_msg("mmap failed\n");
>
>  	supported = check_vmflag(addr, "sd");
> --
> 2.25.1
>

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED
  2026-05-13 10:05 ` David Hildenbrand (Arm)
@ 2026-05-13 11:15   ` Hongfu Li
  0 siblings, 0 replies; 5+ messages in thread
From: Hongfu Li @ 2026-05-13 11:15 UTC (permalink / raw)
  To: david
  Cc: akpm, dev.jain, liam, lihongfu, linux-kernel, linux-kselftest,
	linux-mm, ljs, mhocko, rppt, shuah, surenb, vbabka

Hi David,
Thanks for your review.

> > mmap() returns MAP_FAILED, which is defined as (void *)-1, on error,
> > not NULL.  Several selftests incorrectly check the return value of
> > mmap() using !ptr or ptr == NULL, which would erroneously treat
> > MAP_FAILED as a valid pointer since MAP_FAILED is non-zero and
> > non-NULL.  This can lead to segfaults when mmap() actually fails
> > under memory pressure.
> 
> Well, your patch also adds more checks where we previously didn't have any checks?

Yes, I added a few checks for the mmap return value, as they were not
handled previously. Perhaps I should also adjust the commit message to
match this change.

> > 
> > Signed-off-by: Hongfu Li <lihongfu@kylinos.cn>
> > Reviewed-by: Dev Jain <dev.jain@arm.com>
> > ---
> > v2:
> > - Add missing mmap() return value checks in pkey_sighandler_tests.c
> >   and protection_keys.c
> > ---
> 
> [...]
> 
> > diff --git a/tools/testing/selftests/mm/protection_keys.c b/tools/testing/selftests/mm/protection_keys.c
> > index 2085982dba69..580bf1668c71 100644
> > --- a/tools/testing/selftests/mm/protection_keys.c
> > +++ b/tools/testing/selftests/mm/protection_keys.c
> > @@ -1217,6 +1217,7 @@ static void arch_force_pkey_reg_init(void)
> >  	 * doing the XSAVE size enumeration dance.
> >  	 */
> >  	buf = mmap(NULL, 1*MB, PROT_READ|PROT_WRITE, MAP_ANONYMOUS|MAP_PRIVATE, -1, 0);
> > +	pkey_assert(buf != (void *)-1);
> 
> What's the reason for not using MAP_FAILED?

I tried to keep consistency with the existing mmap handling pattern in protection_keys.c.
That said, using MAP_FAILED would indeed be more appropriate.
I'll adjust this and send out a v3 shortly.

Best regards,
Hongfu

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED
  2026-05-13 10:57 ` Lorenzo Stoakes
@ 2026-05-13 11:20   ` Hongfu Li
  0 siblings, 0 replies; 5+ messages in thread
From: Hongfu Li @ 2026-05-13 11:20 UTC (permalink / raw)
  To: ljs
  Cc: akpm, david, dev.jain, liam, lihongfu, linux-kernel,
	linux-kselftest, linux-mm, mhocko, rppt, shuah, surenb, vbabka

> > mmap() returns MAP_FAILED, which is defined as (void *)-1, on error,
> > not NULL.  Several selftests incorrectly check the return value of
> > mmap() using !ptr or ptr == NULL, which would erroneously treat
> > MAP_FAILED as a valid pointer since MAP_FAILED is non-zero and
> > non-NULL.  This can lead to segfaults when mmap() actually fails
> > under memory pressure.
> >
> > Signed-off-by: Hongfu Li <lihongfu@kylinos.cn>
> > Reviewed-by: Dev Jain <dev.jain@arm.com>
> 
> You forgot my tag :)
> 
> Also, please do not send a respin right away if you can, multiple revisions in
> one day can be tricky to deal with.
> 
> As per David's points - please update the commit message to reflect the fact
> you've added a missing assert too and replace the (void *)-1 with MAP_FAILED.

Thanks for the reminder. I will avoid sending a quick respin within the
same day, and I'll update the commit message to mention the added assert
check, and replace (void *)-1 with MAP_FAILED in the next version.

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2026-05-13 11:19 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-05-13  9:56 [PATCH v2] selftests/mm: Fix incorrect mmap() error handling with NULL instead of MAP_FAILED Hongfu Li
2026-05-13 10:05 ` David Hildenbrand (Arm)
2026-05-13 11:15   ` Hongfu Li
2026-05-13 10:57 ` Lorenzo Stoakes
2026-05-13 11:20   ` Hongfu Li

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