[PATCH v2 0/4] userfaultfd: selftest: Improve behavior with older kernels

[PATCH v2 0/4] userfaultfd: selftest: Improve behavior with older kernels

[Thiago Jung Bauermann]

Hello,

A tester ran the upstream selftest on a distro kernel and sounded the alarm when
it reported failures for features which aren't included in that kernel.

This patch set improves the test behavior in that scenario.

Changes since v1:

- Patch "userfaultfd: selftest: Skip test if userfaultfd() syscall
  not supported"
  - New patch, suggested by Mke Rapoport.

- Patch "userfaultfd: selftest: Skip test if a feature isn't supported"
  - Try running other tests even if one (or more) of them returns KSFT_SKIP.

- Patch "userfaultfd: selftest: Cope if shmem doesn't support zeropage"
  - Ignore lack of UFFDIO_ZEROPAGE in userfaultfd_events_test() and
    userfaultfd_stress().
  - Make userfaultfd_zeropage_test() return KSFT_SKIP if UFFDIO_ZEROPAGE
    isn't supported.

Thiago Jung Bauermann (4):
  userfaultfd: selftest: Fix checking of userfaultfd_open() result
  userfaultfd: selftest: Skip test if userfaultfd() syscall not
    supported
  userfaultfd: selftest: Skip test if a feature isn't supported
  userfaultfd: selftest: Cope if shmem doesn't support zeropage

 tools/testing/selftests/vm/userfaultfd.c | 86 ++++++++++++++++++++++++--------
 1 file changed, 66 insertions(+), 20 deletions(-)

[PATCH v2 1/4] userfaultfd: selftest: Fix checking of userfaultfd_open() result

[Thiago Jung Bauermann]

If the userfaultfd test is run on a kernel with CONFIG_USERFAULTFD=n, it
will report that the system call is not available yet go ahead and continue
anyway:

  # ./userfaultfd anon 30 1
  nr_pages: 480, nr_pages_per_cpu: 120
  userfaultfd syscall not available in this kernel
  bounces: 0, mode:, register failure

This is because userfaultfd_open() returns 0 on success and 1 on error but
all callers assume that it returns < 0 on error.

Since the convention of the test as a whole is the one used by
userfault_open(), fix its callers instead. Now the test behaves correctly:

  # ./userfaultfd anon 30 1
  nr_pages: 480, nr_pages_per_cpu: 120
  userfaultfd syscall not available in this kernel

Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
---
 tools/testing/selftests/vm/userfaultfd.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index 7b8171e3128a..e4099afe7557 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -859,7 +859,7 @@ static int userfaultfd_zeropage_test(void)
 	if (uffd_test_ops->release_pages(area_dst))
 		return 1;
 
-	if (userfaultfd_open(0) < 0)
+	if (userfaultfd_open(0))
 		return 1;
 	uffdio_register.range.start = (unsigned long) area_dst;
 	uffdio_register.range.len = nr_pages * page_size;
@@ -902,7 +902,7 @@ static int userfaultfd_events_test(void)
 
 	features = UFFD_FEATURE_EVENT_FORK | UFFD_FEATURE_EVENT_REMAP |
 		UFFD_FEATURE_EVENT_REMOVE;
-	if (userfaultfd_open(features) < 0)
+	if (userfaultfd_open(features))
 		return 1;
 	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
 
@@ -961,7 +961,7 @@ static int userfaultfd_sig_test(void)
 		return 1;
 
 	features = UFFD_FEATURE_EVENT_FORK|UFFD_FEATURE_SIGBUS;
-	if (userfaultfd_open(features) < 0)
+	if (userfaultfd_open(features))
 		return 1;
 	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
 
@@ -1027,7 +1027,7 @@ static int userfaultfd_stress(void)
 	if (!area_dst)
 		return 1;
 
-	if (userfaultfd_open(0) < 0)
+	if (userfaultfd_open(0))
 		return 1;
 
 	count_verify = malloc(nr_pages * sizeof(unsigned long long));

[PATCH v2 2/4] userfaultfd: selftest: Skip test if userfaultfd() syscall not supported

[Thiago Jung Bauermann]

Since there's no point in doing anything in this case, immediately exit the
process.

And take the opportunity to improve the error message.

Before:

  # ./userfaultfd shmem 10 10
  nr_pages: 160, nr_pages_per_cpu: 40
  userfaultfd syscall not available in this kernel
  # echo $?
  1

After:

  # ./userfaultfd shmem 10 10
  nr_pages: 160, nr_pages_per_cpu: 40
  userfaultfd syscall not available in this kernel: Function not implemented
  # echo $?
  4

Suggested-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
---
 tools/testing/selftests/vm/userfaultfd.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index e4099afe7557..d728bd1cb33c 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -636,9 +636,11 @@ static int userfaultfd_open(int features)
 
 	uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
 	if (uffd < 0) {
-		fprintf(stderr,
-			"userfaultfd syscall not available in this kernel\n");
-		return 1;
+		int errnum = errno;
+
+		perror("userfaultfd syscall not available in this kernel");
+
+		exit(errnum == ENOSYS ? KSFT_SKIP : 1);
 	}
 	uffd_flags = fcntl(uffd, F_GETFD, NULL);
 

[PATCH v2 3/4] userfaultfd: selftest: Skip test if a feature isn't supported

[Thiago Jung Bauermann]

If userfaultfd runs on a system that doesn't support some feature it is
trying to test, it currently ends with error code 1 which indicates
test failure:

  # ./userfaultfd anon 10 10
  nr_pages: 160, nr_pages_per_cpu: 80
  bounces: 9, mode: rnd poll, userfaults: 7 59
  bounces: 8, mode: poll, userfaults: 0 0
  bounces: 7, mode: rnd racing ver, userfaults: 45 2
  bounces: 6, mode: racing ver, userfaults: 3 1
  bounces: 5, mode: rnd ver, userfaults: 55 32
  bounces: 4, mode: ver, userfaults: 69 0
  bounces: 3, mode: rnd racing, userfaults: 1 1
  bounces: 2, mode: racing, userfaults: 65 0
  bounces: 1, mode: rnd, userfaults: 44 1
  bounces: 0, mode:, userfaults: 3 2
  testing UFFDIO_ZEROPAGE: done.
  testing signal delivery: UFFDIO_API
  # echo $?
  1

Make each test return KSFT_SKIP instead, which is more accurate since it is
not a real test failure and try to run the other tests which may still
work:

  # ./userfaultfd anon 10 10
  nr_pages: 160, nr_pages_per_cpu: 80
  bounces: 9, mode: rnd poll, userfaults: 65 27
  bounces: 8, mode: poll, userfaults: 0 0
  bounces: 7, mode: rnd racing ver, userfaults: 60 29
  bounces: 6, mode: racing ver, userfaults: 58 1
  bounces: 5, mode: rnd ver, userfaults: 69 3
  bounces: 4, mode: ver, userfaults: 35 27
  bounces: 3, mode: rnd racing, userfaults: 44 0
  bounces: 2, mode: racing, userfaults: 40 25
  bounces: 1, mode: rnd, userfaults: 2 1
  bounces: 0, mode:, userfaults: 0 1
  testing UFFDIO_ZEROPAGE: done.
  testing signal delivery: UFFDIO_API: Invalid argument
  testing events (fork, remap, remove): userfaults: 160
  # echo $?
  0

While at it, also improve the error message of the ioctl(UFFDIO_API) call.

Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
---
 tools/testing/selftests/vm/userfaultfd.c | 42 +++++++++++++++++++++++---------
 1 file changed, 30 insertions(+), 12 deletions(-)

diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index d728bd1cb33c..c84e44ddf314 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -647,8 +647,11 @@ static int userfaultfd_open(int features)
 	uffdio_api.api = UFFD_API;
 	uffdio_api.features = features;
 	if (ioctl(uffd, UFFDIO_API, &uffdio_api)) {
-		fprintf(stderr, "UFFDIO_API\n");
-		return 1;
+		int errnum = errno;
+
+		perror("UFFDIO_API");
+
+		return errnum == EINVAL ? KSFT_SKIP : 1;
 	}
 	if (uffdio_api.api != UFFD_API) {
 		fprintf(stderr, "UFFDIO_API error %Lu\n", uffdio_api.api);
@@ -854,6 +857,7 @@ static int userfaultfd_zeropage_test(void)
 {
 	struct uffdio_register uffdio_register;
 	unsigned long expected_ioctls;
+	int err;
 
 	printf("testing UFFDIO_ZEROPAGE: ");
 	fflush(stdout);
@@ -861,8 +865,10 @@ static int userfaultfd_zeropage_test(void)
 	if (uffd_test_ops->release_pages(area_dst))
 		return 1;
 
-	if (userfaultfd_open(0))
-		return 1;
+	err = userfaultfd_open(0);
+	if (err)
+		return err;
+
 	uffdio_register.range.start = (unsigned long) area_dst;
 	uffdio_register.range.len = nr_pages * page_size;
 	uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
@@ -904,8 +910,10 @@ static int userfaultfd_events_test(void)
 
 	features = UFFD_FEATURE_EVENT_FORK | UFFD_FEATURE_EVENT_REMAP |
 		UFFD_FEATURE_EVENT_REMOVE;
-	if (userfaultfd_open(features))
-		return 1;
+	err = userfaultfd_open(features);
+	if (err)
+		return err;
+
 	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
 
 	uffdio_register.range.start = (unsigned long) area_dst;
@@ -963,8 +971,9 @@ static int userfaultfd_sig_test(void)
 		return 1;
 
 	features = UFFD_FEATURE_EVENT_FORK|UFFD_FEATURE_SIGBUS;
-	if (userfaultfd_open(features))
-		return 1;
+	err = userfaultfd_open(features);
+	if (err)
+		return err;
 	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
 
 	uffdio_register.range.start = (unsigned long) area_dst;
@@ -1029,8 +1038,9 @@ static int userfaultfd_stress(void)
 	if (!area_dst)
 		return 1;
 
-	if (userfaultfd_open(0))
-		return 1;
+	err = userfaultfd_open(0);
+	if (err)
+		return err;
 
 	count_verify = malloc(nr_pages * sizeof(unsigned long long));
 	if (!count_verify) {
@@ -1201,8 +1211,16 @@ static int userfaultfd_stress(void)
 		return err;
 
 	close(uffd);
-	return userfaultfd_zeropage_test() || userfaultfd_sig_test()
-		|| userfaultfd_events_test();
+
+	err = userfaultfd_zeropage_test();
+	if (err && err != KSFT_SKIP)
+		return err;
+
+	err = userfaultfd_sig_test();
+	if (err && err != KSFT_SKIP)
+		return err;
+
+	return userfaultfd_events_test();
 }
 
 /*

[PATCH v2 4/4] userfaultfd: selftest: Cope if shmem doesn't support zeropage

[Thiago Jung Bauermann]

If userfaultfd runs on a system that doesn't support UFFDIO_ZEROPAGE for
shared memory, it currently ends with error code 1 which indicates test
failure:

  # ./userfaultfd shmem 10 10
  nr_pages: 160, nr_pages_per_cpu: 80
  bounces: 9, mode: rnd poll, unexpected missing ioctl for anon memory
  # echo $?
  1

Change userfaultfd_zeropage_test() to return KSFT_SKIP to indicate that
the test is being skipped.

Also change userfaultfd_events_test() and userfaultfd_stress() to ignore
the missing UFFDIO_ZEROPAGE bit from the list of supported ioctls since
these tests don't require that feature.

Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
---
 tools/testing/selftests/vm/userfaultfd.c | 36 +++++++++++++++++++++++++++-----
 1 file changed, 31 insertions(+), 5 deletions(-)

diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index c84e44ddf314..00f1ca663d67 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -852,6 +852,16 @@ static int uffdio_zeropage(int ufd, unsigned long offset)
 	return __uffdio_zeropage(ufd, offset, false);
 }
 
+/* Tells whether the kernel doesn't support ZEROPAGE on shared memory VMAs. */
+static bool shmem_without_zeropage_support(unsigned long expected_ioctls,
+					   unsigned long supported_ioctls)
+{
+	/* Turn off ZEROPAGE bit from expected_ioctls. */
+	expected_ioctls &= ~(1 << _UFFDIO_ZEROPAGE);
+
+	return test_type == TEST_SHMEM && supported_ioctls == expected_ioctls;
+}
+
 /* exercise UFFDIO_ZEROPAGE */
 static int userfaultfd_zeropage_test(void)
 {
@@ -877,10 +887,20 @@ static int userfaultfd_zeropage_test(void)
 
 	expected_ioctls = uffd_test_ops->expected_ioctls;
 	if ((uffdio_register.ioctls & expected_ioctls) !=
-	    expected_ioctls)
+	    expected_ioctls) {
+		close(uffd);
+
+		if (shmem_without_zeropage_support(expected_ioctls,
+						   uffdio_register.ioctls)) {
+			fprintf(stderr,
+				"UFFDIO_ZEROPAGE unsupported in shmem VMAs\n");
+			return KSFT_SKIP;
+		}
+
 		fprintf(stderr,
-			"unexpected missing ioctl for anon memory\n"),
-			exit(1);
+			"unexpected missing ioctl for anon memory\n");
+		return 1;
+	}
 
 	if (uffdio_zeropage(uffd, 0)) {
 		if (my_bcmp(area_dst, zeropage, page_size))
@@ -924,7 +944,10 @@ static int userfaultfd_events_test(void)
 
 	expected_ioctls = uffd_test_ops->expected_ioctls;
 	if ((uffdio_register.ioctls & expected_ioctls) !=
-	    expected_ioctls)
+	    expected_ioctls &&
+	    /* No need for zeropage support in this test, so ignore it. */
+	    !shmem_without_zeropage_support(expected_ioctls,
+					    uffdio_register.ioctls))
 		fprintf(stderr,
 			"unexpected missing ioctl for anon memory\n"),
 			exit(1);
@@ -1117,7 +1140,10 @@ static int userfaultfd_stress(void)
 		}
 		expected_ioctls = uffd_test_ops->expected_ioctls;
 		if ((uffdio_register.ioctls & expected_ioctls) !=
-		    expected_ioctls) {
+		    expected_ioctls &&
+		    /* No need for zeropage support in this test, so ignore it. */
+		    !shmem_without_zeropage_support(expected_ioctls,
+						    uffdio_register.ioctls)) {
 			fprintf(stderr,
 				"unexpected missing ioctl for anon memory\n");
 			return 1;

Re: [PATCH v2 1/4] userfaultfd: selftest: Fix checking of userfaultfd_open() result

[Mike Rapoport]

On Fri, Aug 03, 2018 at 07:00:43PM -0300, Thiago Jung Bauermann wrote:
> If the userfaultfd test is run on a kernel with CONFIG_USERFAULTFD=n, it
> will report that the system call is not available yet go ahead and continue
> anyway:
> 
>   # ./userfaultfd anon 30 1
>   nr_pages: 480, nr_pages_per_cpu: 120
>   userfaultfd syscall not available in this kernel
>   bounces: 0, mode:, register failure
> 
> This is because userfaultfd_open() returns 0 on success and 1 on error but
> all callers assume that it returns < 0 on error.
> 
> Since the convention of the test as a whole is the one used by
> userfault_open(), fix its callers instead. Now the test behaves correctly:
> 
>   # ./userfaultfd anon 30 1
>   nr_pages: 480, nr_pages_per_cpu: 120
>   userfaultfd syscall not available in this kernel
> 
> Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>

Reviewed-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

> ---
>  tools/testing/selftests/vm/userfaultfd.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
> index 7b8171e3128a..e4099afe7557 100644
> --- a/tools/testing/selftests/vm/userfaultfd.c
> +++ b/tools/testing/selftests/vm/userfaultfd.c
> @@ -859,7 +859,7 @@ static int userfaultfd_zeropage_test(void)
>  	if (uffd_test_ops->release_pages(area_dst))
>  		return 1;
> 
> -	if (userfaultfd_open(0) < 0)
> +	if (userfaultfd_open(0))
>  		return 1;
>  	uffdio_register.range.start = (unsigned long) area_dst;
>  	uffdio_register.range.len = nr_pages * page_size;
> @@ -902,7 +902,7 @@ static int userfaultfd_events_test(void)
> 
>  	features = UFFD_FEATURE_EVENT_FORK | UFFD_FEATURE_EVENT_REMAP |
>  		UFFD_FEATURE_EVENT_REMOVE;
> -	if (userfaultfd_open(features) < 0)
> +	if (userfaultfd_open(features))
>  		return 1;
>  	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
> 
> @@ -961,7 +961,7 @@ static int userfaultfd_sig_test(void)
>  		return 1;
> 
>  	features = UFFD_FEATURE_EVENT_FORK|UFFD_FEATURE_SIGBUS;
> -	if (userfaultfd_open(features) < 0)
> +	if (userfaultfd_open(features))
>  		return 1;
>  	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
> 
> @@ -1027,7 +1027,7 @@ static int userfaultfd_stress(void)
>  	if (!area_dst)
>  		return 1;
> 
> -	if (userfaultfd_open(0) < 0)
> +	if (userfaultfd_open(0))
>  		return 1;
> 
>  	count_verify = malloc(nr_pages * sizeof(unsigned long long));

-- 
Sincerely yours,
Mike.

Re: [PATCH v2 2/4] userfaultfd: selftest: Skip test if userfaultfd() syscall not supported

[Mike Rapoport]

On Fri, Aug 03, 2018 at 07:00:44PM -0300, Thiago Jung Bauermann wrote:
> Since there's no point in doing anything in this case, immediately exit the
> process.
> 
> And take the opportunity to improve the error message.
> 
> Before:
> 
>   # ./userfaultfd shmem 10 10
>   nr_pages: 160, nr_pages_per_cpu: 40
>   userfaultfd syscall not available in this kernel
>   # echo $?
>   1
> 
> After:
> 
>   # ./userfaultfd shmem 10 10
>   nr_pages: 160, nr_pages_per_cpu: 40
>   userfaultfd syscall not available in this kernel: Function not implemented
>   # echo $?
>   4
> 
> Suggested-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
> Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>

Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

> ---
>  tools/testing/selftests/vm/userfaultfd.c | 8 +++++---
>  1 file changed, 5 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
> index e4099afe7557..d728bd1cb33c 100644
> --- a/tools/testing/selftests/vm/userfaultfd.c
> +++ b/tools/testing/selftests/vm/userfaultfd.c
> @@ -636,9 +636,11 @@ static int userfaultfd_open(int features)
> 
>  	uffd = syscall(__NR_userfaultfd, O_CLOEXEC | O_NONBLOCK);
>  	if (uffd < 0) {
> -		fprintf(stderr,
> -			"userfaultfd syscall not available in this kernel\n");
> -		return 1;
> +		int errnum = errno;
> +
> +		perror("userfaultfd syscall not available in this kernel");
> +
> +		exit(errnum == ENOSYS ? KSFT_SKIP : 1);
>  	}
>  	uffd_flags = fcntl(uffd, F_GETFD, NULL);
> 

-- 
Sincerely yours,
Mike.

Re: [PATCH v2 3/4] userfaultfd: selftest: Skip test if a feature isn't supported

[Mike Rapoport]

On Fri, Aug 03, 2018 at 07:00:45PM -0300, Thiago Jung Bauermann wrote:
> If userfaultfd runs on a system that doesn't support some feature it is
> trying to test, it currently ends with error code 1 which indicates
> test failure:
> 
>   # ./userfaultfd anon 10 10
>   nr_pages: 160, nr_pages_per_cpu: 80
>   bounces: 9, mode: rnd poll, userfaults: 7 59
>   bounces: 8, mode: poll, userfaults: 0 0
>   bounces: 7, mode: rnd racing ver, userfaults: 45 2
>   bounces: 6, mode: racing ver, userfaults: 3 1
>   bounces: 5, mode: rnd ver, userfaults: 55 32
>   bounces: 4, mode: ver, userfaults: 69 0
>   bounces: 3, mode: rnd racing, userfaults: 1 1
>   bounces: 2, mode: racing, userfaults: 65 0
>   bounces: 1, mode: rnd, userfaults: 44 1
>   bounces: 0, mode:, userfaults: 3 2
>   testing UFFDIO_ZEROPAGE: done.
>   testing signal delivery: UFFDIO_API
>   # echo $?
>   1
> 
> Make each test return KSFT_SKIP instead, which is more accurate since it is
> not a real test failure and try to run the other tests which may still
> work:
> 
>   # ./userfaultfd anon 10 10
>   nr_pages: 160, nr_pages_per_cpu: 80
>   bounces: 9, mode: rnd poll, userfaults: 65 27
>   bounces: 8, mode: poll, userfaults: 0 0
>   bounces: 7, mode: rnd racing ver, userfaults: 60 29
>   bounces: 6, mode: racing ver, userfaults: 58 1
>   bounces: 5, mode: rnd ver, userfaults: 69 3
>   bounces: 4, mode: ver, userfaults: 35 27
>   bounces: 3, mode: rnd racing, userfaults: 44 0
>   bounces: 2, mode: racing, userfaults: 40 25
>   bounces: 1, mode: rnd, userfaults: 2 1
>   bounces: 0, mode:, userfaults: 0 1
>   testing UFFDIO_ZEROPAGE: done.
>   testing signal delivery: UFFDIO_API: Invalid argument
>   testing events (fork, remap, remove): userfaults: 160
>   # echo $?
>   0
> 
> While at it, also improve the error message of the ioctl(UFFDIO_API) call.
> 
> Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>

Acked-by: Mike Rapoport <rppt@linux.vnet.ibm.com>

> ---
>  tools/testing/selftests/vm/userfaultfd.c | 42 +++++++++++++++++++++++---------
>  1 file changed, 30 insertions(+), 12 deletions(-)
> 
> diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
> index d728bd1cb33c..c84e44ddf314 100644
> --- a/tools/testing/selftests/vm/userfaultfd.c
> +++ b/tools/testing/selftests/vm/userfaultfd.c
> @@ -647,8 +647,11 @@ static int userfaultfd_open(int features)
>  	uffdio_api.api = UFFD_API;
>  	uffdio_api.features = features;
>  	if (ioctl(uffd, UFFDIO_API, &uffdio_api)) {
> -		fprintf(stderr, "UFFDIO_API\n");
> -		return 1;
> +		int errnum = errno;
> +
> +		perror("UFFDIO_API");
> +
> +		return errnum == EINVAL ? KSFT_SKIP : 1;
>  	}
>  	if (uffdio_api.api != UFFD_API) {
>  		fprintf(stderr, "UFFDIO_API error %Lu\n", uffdio_api.api);
> @@ -854,6 +857,7 @@ static int userfaultfd_zeropage_test(void)
>  {
>  	struct uffdio_register uffdio_register;
>  	unsigned long expected_ioctls;
> +	int err;
> 
>  	printf("testing UFFDIO_ZEROPAGE: ");
>  	fflush(stdout);
> @@ -861,8 +865,10 @@ static int userfaultfd_zeropage_test(void)
>  	if (uffd_test_ops->release_pages(area_dst))
>  		return 1;
> 
> -	if (userfaultfd_open(0))
> -		return 1;
> +	err = userfaultfd_open(0);
> +	if (err)
> +		return err;
> +
>  	uffdio_register.range.start = (unsigned long) area_dst;
>  	uffdio_register.range.len = nr_pages * page_size;
>  	uffdio_register.mode = UFFDIO_REGISTER_MODE_MISSING;
> @@ -904,8 +910,10 @@ static int userfaultfd_events_test(void)
> 
>  	features = UFFD_FEATURE_EVENT_FORK | UFFD_FEATURE_EVENT_REMAP |
>  		UFFD_FEATURE_EVENT_REMOVE;
> -	if (userfaultfd_open(features))
> -		return 1;
> +	err = userfaultfd_open(features);
> +	if (err)
> +		return err;
> +
>  	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
> 
>  	uffdio_register.range.start = (unsigned long) area_dst;
> @@ -963,8 +971,9 @@ static int userfaultfd_sig_test(void)
>  		return 1;
> 
>  	features = UFFD_FEATURE_EVENT_FORK|UFFD_FEATURE_SIGBUS;
> -	if (userfaultfd_open(features))
> -		return 1;
> +	err = userfaultfd_open(features);
> +	if (err)
> +		return err;
>  	fcntl(uffd, F_SETFL, uffd_flags | O_NONBLOCK);
> 
>  	uffdio_register.range.start = (unsigned long) area_dst;
> @@ -1029,8 +1038,9 @@ static int userfaultfd_stress(void)
>  	if (!area_dst)
>  		return 1;
> 
> -	if (userfaultfd_open(0))
> -		return 1;
> +	err = userfaultfd_open(0);
> +	if (err)
> +		return err;
> 
>  	count_verify = malloc(nr_pages * sizeof(unsigned long long));
>  	if (!count_verify) {
> @@ -1201,8 +1211,16 @@ static int userfaultfd_stress(void)
>  		return err;
> 
>  	close(uffd);
> -	return userfaultfd_zeropage_test() || userfaultfd_sig_test()
> -		|| userfaultfd_events_test();
> +
> +	err = userfaultfd_zeropage_test();
> +	if (err && err != KSFT_SKIP)
> +		return err;
> +
> +	err = userfaultfd_sig_test();
> +	if (err && err != KSFT_SKIP)
> +		return err;
> +
> +	return userfaultfd_events_test();
>  }
> 
>  /*

-- 
Sincerely yours,
Mike.

Re: [PATCH v2 4/4] userfaultfd: selftest: Cope if shmem doesn't support zeropage

[Mike Rapoport]

Hi,

On Fri, Aug 03, 2018 at 07:00:46PM -0300, Thiago Jung Bauermann wrote:
> If userfaultfd runs on a system that doesn't support UFFDIO_ZEROPAGE for
> shared memory, it currently ends with error code 1 which indicates test
> failure:
> 
>   # ./userfaultfd shmem 10 10
>   nr_pages: 160, nr_pages_per_cpu: 80
>   bounces: 9, mode: rnd poll, unexpected missing ioctl for anon memory
>   # echo $?
>   1
> 
> Change userfaultfd_zeropage_test() to return KSFT_SKIP to indicate that
> the test is being skipped.

I took a deeper look at what userfaultfd_zeropage_test() does and,
apparently, I've mislead you. The test checks if the range has
UFFDIO_ZEROPAGE and verifies that it works if yes; otherwise the test
verifies that EINVAL is returned.

Can you please check if the patch below works in your environment?

From 7a34c84c0461b5073742275638c44b6535d19166 Mon Sep 17 00:00:00 2001
From: Mike Rapoport <rppt@linux.vnet.ibm.com>
Date: Tue, 7 Aug 2018 09:44:19 +0300
Subject: [PATCH] userfaultfd: selftest: make supported range ioctl
 verification more robust

When userfaultfd tests runs on older kernel that does not support
UFFDIO_ZEROPAGE for shared memory it fails at the ioctl verification.

Split out the verification that supported ioctls are superset of the
expected ioctls and relax the checks for UFFDIO_ZEROPAGE for shared memory
areas.

Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
---
 tools/testing/selftests/vm/userfaultfd.c | 63 +++++++++++++++++---------------
 1 file changed, 34 insertions(+), 29 deletions(-)

diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
index 7b8171e3128a..a64bc38bc0e1 100644
--- a/tools/testing/selftests/vm/userfaultfd.c
+++ b/tools/testing/selftests/vm/userfaultfd.c
@@ -271,6 +271,32 @@ static int my_bcmp(char *str1, char *str2, size_t n)
 	return 0;
 }
 
+static int verify_ioctls(unsigned long supported_ioctls)
+{
+	unsigned long expected_ioctls = uffd_test_ops->expected_ioctls;
+
+	if ((supported_ioctls & expected_ioctls) == expected_ioctls)
+		return 0;
+
+	/*
+	 * For older kernels shared memory may not have UFFDIO_ZEROPAGE.
+	 * In this case we just mask it out from the
+	 * expected_ioctls. The userfaultfd_zeropage_test will then
+	 * verify that an attempt to use UFFDIO_ZEROPAGE returns
+	 * EINVAL
+	 */
+	if (test_type == TEST_SHMEM) {
+		expected_ioctls &= ~(1 << _UFFDIO_ZEROPAGE);
+		if ((supported_ioctls & expected_ioctls) == expected_ioctls) {
+			uffd_test_ops->expected_ioctls = expected_ioctls;
+			return 0;
+		}
+	}
+
+	fprintf(stderr,	"unexpected missing ioctl\n");
+	return 1;
+}
+
 static void *locking_thread(void *arg)
 {
 	unsigned long cpu = (unsigned long) arg;
@@ -851,7 +877,6 @@ static int uffdio_zeropage(int ufd, unsigned long offset)
 static int userfaultfd_zeropage_test(void)
 {
 	struct uffdio_register uffdio_register;
-	unsigned long expected_ioctls;
 
 	printf("testing UFFDIO_ZEROPAGE: ");
 	fflush(stdout);
@@ -867,12 +892,8 @@ static int userfaultfd_zeropage_test(void)
 	if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register))
 		fprintf(stderr, "register failure\n"), exit(1);
 
-	expected_ioctls = uffd_test_ops->expected_ioctls;
-	if ((uffdio_register.ioctls & expected_ioctls) !=
-	    expected_ioctls)
-		fprintf(stderr,
-			"unexpected missing ioctl for anon memory\n"),
-			exit(1);
+	if (verify_ioctls(uffdio_register.ioctls))
+		return 1;
 
 	if (uffdio_zeropage(uffd, 0)) {
 		if (my_bcmp(area_dst, zeropage, page_size))
@@ -887,7 +908,6 @@ static int userfaultfd_zeropage_test(void)
 static int userfaultfd_events_test(void)
 {
 	struct uffdio_register uffdio_register;
-	unsigned long expected_ioctls;
 	unsigned long userfaults;
 	pthread_t uffd_mon;
 	int err, features;
@@ -912,12 +932,8 @@ static int userfaultfd_events_test(void)
 	if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register))
 		fprintf(stderr, "register failure\n"), exit(1);
 
-	expected_ioctls = uffd_test_ops->expected_ioctls;
-	if ((uffdio_register.ioctls & expected_ioctls) !=
-	    expected_ioctls)
-		fprintf(stderr,
-			"unexpected missing ioctl for anon memory\n"),
-			exit(1);
+	if (verify_ioctls(uffdio_register.ioctls))
+		return 1;
 
 	if (pthread_create(&uffd_mon, &attr, uffd_poll_thread, NULL))
 		perror("uffd_poll_thread create"), exit(1);
@@ -947,7 +963,6 @@ static int userfaultfd_events_test(void)
 static int userfaultfd_sig_test(void)
 {
 	struct uffdio_register uffdio_register;
-	unsigned long expected_ioctls;
 	unsigned long userfaults;
 	pthread_t uffd_mon;
 	int err, features;
@@ -971,12 +986,8 @@ static int userfaultfd_sig_test(void)
 	if (ioctl(uffd, UFFDIO_REGISTER, &uffdio_register))
 		fprintf(stderr, "register failure\n"), exit(1);
 
-	expected_ioctls = uffd_test_ops->expected_ioctls;
-	if ((uffdio_register.ioctls & expected_ioctls) !=
-	    expected_ioctls)
-		fprintf(stderr,
-			"unexpected missing ioctl for anon memory\n"),
-			exit(1);
+	if (verify_ioctls(uffdio_register.ioctls))
+		return 1;
 
 	if (faulting_process(1))
 		fprintf(stderr, "faulting process failed\n"), exit(1);
@@ -1076,8 +1087,6 @@ static int userfaultfd_stress(void)
 
 	err = 0;
 	while (bounces--) {
-		unsigned long expected_ioctls;
-
 		printf("bounces: %d, mode:", bounces);
 		if (bounces & BOUNCE_RANDOM)
 			printf(" rnd");
@@ -1103,13 +1112,9 @@ static int userfaultfd_stress(void)
 			fprintf(stderr, "register failure\n");
 			return 1;
 		}
-		expected_ioctls = uffd_test_ops->expected_ioctls;
-		if ((uffdio_register.ioctls & expected_ioctls) !=
-		    expected_ioctls) {
-			fprintf(stderr,
-				"unexpected missing ioctl for anon memory\n");
+
+		if (verify_ioctls(uffdio_register.ioctls))
 			return 1;
-		}
 
 		if (area_dst_alias) {
 			uffdio_register.range.start = (unsigned long)
-- 
2.7.4

 
> Also change userfaultfd_events_test() and userfaultfd_stress() to ignore
> the missing UFFDIO_ZEROPAGE bit from the list of supported ioctls since
> these tests don't require that feature.
> 
> Signed-off-by: Thiago Jung Bauermann <bauerman@linux.ibm.com>
> ---
>  tools/testing/selftests/vm/userfaultfd.c | 36 +++++++++++++++++++++++++++-----
>  1 file changed, 31 insertions(+), 5 deletions(-)
> 
> diff --git a/tools/testing/selftests/vm/userfaultfd.c b/tools/testing/selftests/vm/userfaultfd.c
> index c84e44ddf314..00f1ca663d67 100644
> --- a/tools/testing/selftests/vm/userfaultfd.c
> +++ b/tools/testing/selftests/vm/userfaultfd.c
> @@ -852,6 +852,16 @@ static int uffdio_zeropage(int ufd, unsigned long offset)
>  	return __uffdio_zeropage(ufd, offset, false);
>  }
> 
> +/* Tells whether the kernel doesn't support ZEROPAGE on shared memory VMAs. */
> +static bool shmem_without_zeropage_support(unsigned long expected_ioctls,
> +					   unsigned long supported_ioctls)
> +{
> +	/* Turn off ZEROPAGE bit from expected_ioctls. */
> +	expected_ioctls &= ~(1 << _UFFDIO_ZEROPAGE);
> +
> +	return test_type == TEST_SHMEM && supported_ioctls == expected_ioctls;
> +}
> +
>  /* exercise UFFDIO_ZEROPAGE */
>  static int userfaultfd_zeropage_test(void)
>  {
> @@ -877,10 +887,20 @@ static int userfaultfd_zeropage_test(void)
> 
>  	expected_ioctls = uffd_test_ops->expected_ioctls;
>  	if ((uffdio_register.ioctls & expected_ioctls) !=
> -	    expected_ioctls)
> +	    expected_ioctls) {
> +		close(uffd);
> +
> +		if (shmem_without_zeropage_support(expected_ioctls,
> +						   uffdio_register.ioctls)) {
> +			fprintf(stderr,
> +				"UFFDIO_ZEROPAGE unsupported in shmem VMAs\n");
> +			return KSFT_SKIP;
> +		}
> +
>  		fprintf(stderr,
> -			"unexpected missing ioctl for anon memory\n"),
> -			exit(1);
> +			"unexpected missing ioctl for anon memory\n");
> +		return 1;
> +	}
> 
>  	if (uffdio_zeropage(uffd, 0)) {
>  		if (my_bcmp(area_dst, zeropage, page_size))
> @@ -924,7 +944,10 @@ static int userfaultfd_events_test(void)
> 
>  	expected_ioctls = uffd_test_ops->expected_ioctls;
>  	if ((uffdio_register.ioctls & expected_ioctls) !=
> -	    expected_ioctls)
> +	    expected_ioctls &&
> +	    /* No need for zeropage support in this test, so ignore it. */
> +	    !shmem_without_zeropage_support(expected_ioctls,
> +					    uffdio_register.ioctls))
>  		fprintf(stderr,
>  			"unexpected missing ioctl for anon memory\n"),
>  			exit(1);
> @@ -1117,7 +1140,10 @@ static int userfaultfd_stress(void)
>  		}
>  		expected_ioctls = uffd_test_ops->expected_ioctls;
>  		if ((uffdio_register.ioctls & expected_ioctls) !=
> -		    expected_ioctls) {
> +		    expected_ioctls &&
> +		    /* No need for zeropage support in this test, so ignore it. */
> +		    !shmem_without_zeropage_support(expected_ioctls,
> +						    uffdio_register.ioctls)) {
>  			fprintf(stderr,
>  				"unexpected missing ioctl for anon memory\n");
>  			return 1;

-- 
Sincerely yours,
Mike.

Re: [PATCH v2 4/4] userfaultfd: selftest: Cope if shmem doesn't support zeropage

[Thiago Jung Bauermann]

Hello Mike,

Mike Rapoport <rppt@linux.vnet.ibm.com> writes:

> Hi,
>
> On Fri, Aug 03, 2018 at 07:00:46PM -0300, Thiago Jung Bauermann wrote:
>> If userfaultfd runs on a system that doesn't support UFFDIO_ZEROPAGE for
>> shared memory, it currently ends with error code 1 which indicates test
>> failure:
>>
>>   # ./userfaultfd shmem 10 10
>>   nr_pages: 160, nr_pages_per_cpu: 80
>>   bounces: 9, mode: rnd poll, unexpected missing ioctl for anon memory
>>   # echo $?
>>   1
>>
>> Change userfaultfd_zeropage_test() to return KSFT_SKIP to indicate that
>> the test is being skipped.
>
> I took a deeper look at what userfaultfd_zeropage_test() does and,
> apparently, I've mislead you. The test checks if the range has
> UFFDIO_ZEROPAGE and verifies that it works if yes; otherwise the test
> verifies that EINVAL is returned.
>
> Can you please check if the patch below works in your environment?
>
> From 7a34c84c0461b5073742275638c44b6535d19166 Mon Sep 17 00:00:00 2001
> From: Mike Rapoport <rppt@linux.vnet.ibm.com>
> Date: Tue, 7 Aug 2018 09:44:19 +0300
> Subject: [PATCH] userfaultfd: selftest: make supported range ioctl
>  verification more robust
>
> When userfaultfd tests runs on older kernel that does not support
> UFFDIO_ZEROPAGE for shared memory it fails at the ioctl verification.
>
> Split out the verification that supported ioctls are superset of the
> expected ioctls and relax the checks for UFFDIO_ZEROPAGE for shared memory
> areas.
>
> Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
> ---
>  tools/testing/selftests/vm/userfaultfd.c | 63 +++++++++++++++++---------------
>  1 file changed, 34 insertions(+), 29 deletions(-)

I'm sorry to take this long to respond, I was only able to get back to
this today.

Your patch does solve my problem. Thank you very much!

It has a trivial conflict in the second hunk with patch 3 in my series.
Should I repost the series with your patch in place of patch 4?

--
Thiago Jung Bauermann
IBM Linux Technology Center

Re: [PATCH v2 4/4] userfaultfd: selftest: Cope if shmem doesn't support zeropage

[Mike Rapoport]

On Mon, Aug 27, 2018 at 11:46:33PM -0300, Thiago Jung Bauermann wrote:
> 
> Hello Mike,
> 
> Mike Rapoport <rppt@linux.vnet.ibm.com> writes:
> 
> > Hi,
> >
> > On Fri, Aug 03, 2018 at 07:00:46PM -0300, Thiago Jung Bauermann wrote:
> >> If userfaultfd runs on a system that doesn't support UFFDIO_ZEROPAGE for
> >> shared memory, it currently ends with error code 1 which indicates test
> >> failure:
> >>
> >>   # ./userfaultfd shmem 10 10
> >>   nr_pages: 160, nr_pages_per_cpu: 80
> >>   bounces: 9, mode: rnd poll, unexpected missing ioctl for anon memory
> >>   # echo $?
> >>   1
> >>
> >> Change userfaultfd_zeropage_test() to return KSFT_SKIP to indicate that
> >> the test is being skipped.
> >
> > I took a deeper look at what userfaultfd_zeropage_test() does and,
> > apparently, I've mislead you. The test checks if the range has
> > UFFDIO_ZEROPAGE and verifies that it works if yes; otherwise the test
> > verifies that EINVAL is returned.
> >
> > Can you please check if the patch below works in your environment?
> >
> > From 7a34c84c0461b5073742275638c44b6535d19166 Mon Sep 17 00:00:00 2001
> > From: Mike Rapoport <rppt@linux.vnet.ibm.com>
> > Date: Tue, 7 Aug 2018 09:44:19 +0300
> > Subject: [PATCH] userfaultfd: selftest: make supported range ioctl
> >  verification more robust
> >
> > When userfaultfd tests runs on older kernel that does not support
> > UFFDIO_ZEROPAGE for shared memory it fails at the ioctl verification.
> >
> > Split out the verification that supported ioctls are superset of the
> > expected ioctls and relax the checks for UFFDIO_ZEROPAGE for shared memory
> > areas.
> >
> > Signed-off-by: Mike Rapoport <rppt@linux.vnet.ibm.com>
> > ---
> >  tools/testing/selftests/vm/userfaultfd.c | 63 +++++++++++++++++---------------
> >  1 file changed, 34 insertions(+), 29 deletions(-)
> 
> I'm sorry to take this long to respond, I was only able to get back to
> this today.

No problem :)

> Your patch does solve my problem. Thank you very much!
> 
> It has a trivial conflict in the second hunk with patch 3 in my series.
> Should I repost the series with your patch in place of patch 4?

Yep.
 
> --
> Thiago Jung Bauermann
> IBM Linux Technology Center

-- 
Sincerely yours,
Mike.