Skip to contentSkip to conversation

syscalls/pidfd_open01.c: Add check for close-on-exec flag

PATCH Superseded Xiao Yang 4 participants 17 messages May 13, 2020 – May 14, 2020

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Xiao Yang

author
#0 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

pidfd_open(2) will set close-on-exec flag on the file descriptor as it manpage states, so check close-on-exec flag by fcntl(2).

BTW:
I tried to pass (long) TST_RET to fcntl() but triggered the following
compiler warning, so pass (int) pidfd instead.
------------------------------------------------------
In file included from pidfd_open01.c:9:
pidfd_open01.c: In function ?run?:
../../../../include/tst_test.h:76:41: warning: format ?%i? expects argument of type ?int?, but argument 5 has type ?long int? [-Wformat=]
76 | tst_brk_(__FILE__, __LINE__, (ttype), (arg_fmt), ##__VA_ARGS__);\
| ^~~~~~~~~
../../../../include/tst_safe_macros.h:224:5: note: in expansion of macro ?tst_brk?

  224 |     tst_brk(TBROK | TERRNO,                          \
      |     ^~~~~~~

pidfd_open01.c:20:9: note: in expansion of macro ?SAFE_FCNTL?
20 | flag = SAFE_FCNTL(TST_RET, F_GETFD);
------------------------------------------------------

---
 testcases/kernel/syscalls/pidfd_open/pidfd_open01.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
testcases/kernel/syscalls/pidfd_open/pidfd_open01.c +8 -1
index 93bb86687..ba1580bc7 100644
@@ -11,13 +11,20 @@
 
 static void run(void)
 {
-	TEST(pidfd_open(getpid(), 0));
+	int pidfd, flag;
+
+	TEST(pidfd = pidfd_open(getpid(), 0));
 
 	if (TST_RET == -1)
 		tst_brk(TFAIL | TTERRNO, "pidfd_open(getpid(), 0) failed");
 
+	flag = SAFE_FCNTL(pidfd, F_GETFD);
+
 	SAFE_CLOSE(TST_RET);
 
+	if (!(flag & FD_CLOEXEC))
+		tst_brk(TFAIL, "pidfd_open(getpid(), 0) didn't set close-on-exec flag");
+
 	tst_res(TPASS, "pidfd_open(getpid(), 0) passed");
 }
 
Signature · 1 line
2.21.0
author
#1 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Hi Cyril, Petr

For the patch set, I and Viresh have the following doubts so do you have any suggestion about them?
1) I keep TEST() in pidfd_open01/pidfd_open03 for now but I think it is surplus because pidfd/fd and TERRNO are enough to check return value and errno.
I wonder if it is necessary to keep TEST()?
2) tst_syscall() is enough to check the support of pidfd_open() and I don't want to define check function as fsopen_supported_by_kernel() does.
Do you think so?

BTW:
I don't like the implementation of fsopen_supported_by_kernel():
a) syscall()/tst_syscall() is enough to check the support of pidfd_open(2) and 'tst_kvercmp(5, 2, 0)) < 0' will skip the check if a kernel on distribution is newer than v5.2 but drop the support of pidfd_open(2) on purpose.
b) tst_syscall() has checked ENOSYS error so we can simple fsopen_supported_by_kernel() by replacing syscall() with tst_syscalls().

Best Regards,
Xiao Yang

Quoted 51 lines · Xiao Yang

pidfd_open(2) will set close-on-exec flag on the file descriptor as it manpage states, so check close-on-exec flag by fcntl(2).

BTW:
I tried to pass (long) TST_RET to fcntl() but triggered the following
compiler warning, so pass (int) pidfd instead.
------------------------------------------------------
In file included from pidfd_open01.c:9:
pidfd_open01.c: In function ?run?:
../../../../include/tst_test.h:76:41: warning: format ?%i? expects argument of type ?int?, but argument 5 has type ?long int? [-Wformat=]

    76 |   tst_brk_(__FILE__, __LINE__, (ttype), (arg_fmt), ##__VA_ARGS__);\
       |                                         ^~~~~~~~~
../../../../include/tst_safe_macros.h:224:5: note: in expansion of macro ?tst_brk?
   224 |     tst_brk(TBROK | TERRNO,                          \
       |     ^~~~~~~
pidfd_open01.c:20:9: note: in expansion of macro ?SAFE_FCNTL?
    20 |  flag = SAFE_FCNTL(TST_RET, F_GETFD);

------------------------------------------------------

---
  testcases/kernel/syscalls/pidfd_open/pidfd_open01.c | 9 ++++++++-
  1 file changed, 8 insertions(+), 1 deletion(-)
testcases/kernel/syscalls/pidfd_open/pidfd_open01.c +8 -1
index 93bb86687..ba1580bc7 100644
@@ -11,13 +11,20 @@
 
  static void run(void)
  {
-	TEST(pidfd_open(getpid(), 0));
+	int pidfd, flag;
+
+	TEST(pidfd = pidfd_open(getpid(), 0));
 
  	if (TST_RET == -1)
  		tst_brk(TFAIL | TTERRNO, "pidfd_open(getpid(), 0) failed");
 
+	flag = SAFE_FCNTL(pidfd, F_GETFD);
+
  	SAFE_CLOSE(TST_RET);
 
+	if (!(flag&  FD_CLOEXEC))
+		tst_brk(TFAIL, "pidfd_open(getpid(), 0) didn't set close-on-exec flag");
+
  	tst_res(TPASS, "pidfd_open(getpid(), 0) passed");
  }
 
#2 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Hi Yang,

Quoted 6 lines

For the patch set, I and Viresh have the following doubts so do you have any suggestion about them?
1) I keep TEST() in pidfd_open01/pidfd_open03 for now but I think it is surplus because pidfd/fd and TERRNO are enough to check return value and errno.
I wonder if it is necessary to keep TEST()?

yes, I've noticed your discussion at v1, sorry I wasn't able to follow.
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/
Just to get context, We're talking about part of the changes between v1 and v2:

+++ b/testcases/kernel/syscalls/pidfd_open/pidfd_open03.c @@ -27,9 +27,11 @@ static void run(void)

                exit(EXIT_SUCCESS);
        }

- fd = pidfd_open(pid, 0);
+ TEST(pidfd_open(pid, 0));
+

+       fd = TST_RET;
        if (fd == -1)
-               tst_brk(TFAIL | TERRNO, "pidfd_open() failed");
+               tst_brk(TFAIL | TTERRNO, "pidfd_open() failed");
        TST_CHECKPOINT_WAKE(0);

I haven't found Cyril's request to use TEST(). To be honest, not sure if it was meant to make sure that errno needs to be reset before (which TEST()) does. If not, using pidfd_open() directly would be ok for me.

Quoted 9 lines

2) tst_syscall() is enough to check the support of pidfd_open() and I don't want to define check function as fsopen_supported_by_kernel() does.
Do you think so?

BTW:
I don't like the implementation of fsopen_supported_by_kernel():
a) syscall()/tst_syscall() is enough to check the support of pidfd_open(2) and 'tst_kvercmp(5, 2, 0)) < 0' will skip the check if

+1 for tst_syscall()

Quoted 2 lines

a kernel on distribution is newer than v5.2 but drop the support of pidfd_open(2) on purpose.

"drop support of pidfd_open(2) on purpose": would anybody has a reason to do that?

Quoted 2 lines

b) tst_syscall() has checked ENOSYS error so we can simple fsopen_supported_by_kernel() by replacing syscall() with tst_syscalls().

Well, one of the benefits of fsopen_supported_by_kernel() was to reduce a bit of duplicity. Even if the implementation is like TEST and SAFE_CLOSE(), it's a fewer lines (+ function name as a self docs).

void fsopen_supported_by_kernel(void)

{
	TEST(tst_syscall(__NR_fsopen, NULL, 0));
	if (TST_RET != -1)
		SAFE_CLOSE(TST_RET);
}

For your change for pidfd_open03.c:

static void setup(void)

{
	int pidfd = -1;

	// Check if pidfd_open(2) is not supported
	pidfd = tst_syscall(__NR_pidfd_open, getpid(), 0);
	if (pidfd != -1)
		SAFE_CLOSE(pidfd);
}

static struct tst_test test = {
- .min_kver = "5.3",
+ .setup = setup,

How about to call the function pidfd_open_supported_by_kernel()? Than you can remove the comment (which BTW should use C style /* */). And IMHO you don't have to assign pidfd to -1.

Kind regards,
Petr

author
#3 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Quoted 28 lines · Petr Vorel

Hi Yang,

Quoted 6 lines

For the patch set, I and Viresh have the following doubts so do you have any suggestion about them?
1) I keep TEST() in pidfd_open01/pidfd_open03 for now but I think it is surplus because pidfd/fd and TERRNO are enough to check return value and errno.
I wonder if it is necessary to keep TEST()?

yes, I've noticed your discussion at v1, sorry I wasn't able to follow.
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/
Just to get context, We're talking about part of the changes between v1 and v2:

+++ b/testcases/kernel/syscalls/pidfd_open/pidfd_open03.c

(quoted diff) +4 -2
@@ -27,9 +27,11 @@ static void run(void)
                 exit(EXIT_SUCCESS);
         }
 
-       fd = pidfd_open(pid, 0);
+       TEST(pidfd_open(pid, 0));
+
+       fd = TST_RET;
         if (fd == -1)
-               tst_brk(TFAIL | TERRNO, "pidfd_open() failed");
+               tst_brk(TFAIL | TTERRNO, "pidfd_open() failed");
 
         TST_CHECKPOINT_WAKE(0);
 
Quoted 3 lines

I haven't found Cyril's request to use TEST(). To be honest, not sure if it was meant to make sure that errno needs to be reset before (which TEST()) does. If not, using pidfd_open() directly would be ok for me.

Hi Petr,

Thanks a lot for your quick reply.

Resetting errno may not necessary because errno will be set again when fd == -1.

Quoted 11 lines
Quoted 11 lines

2) tst_syscall() is enough to check the support of pidfd_open() and I don't want to define check function as fsopen_supported_by_kernel() does.
Do you think so?

BTW:
I don't like the implementation of fsopen_supported_by_kernel():
a) syscall()/tst_syscall() is enough to check the support of pidfd_open(2) and 'tst_kvercmp(5, 2, 0))< 0' will skip the check if

Quoted 6 lines

+1 for tst_syscall()

Quoted 2 lines

a kernel on distribution is newer than v5.2 but drop the support of pidfd_open(2) on purpose.

"drop support of pidfd_open(2) on purpose": would anybody has a reason to do that?

As my pervious mail said, It is just a possible situation? for example:
Upstream kernel introduces btrfs filesystem long long ago but the kernel of RHEL8 drops btrfs filesystem because of some reasons.

It is just a reason used to explain why I want to drop the kernel version check.

Quoted 31 lines
Quoted 3 lines

b) tst_syscall() has checked ENOSYS error so we can simple fsopen_supported_by_kernel() by replacing syscall() with tst_syscalls().

Well, one of the benefits of fsopen_supported_by_kernel() was to reduce a bit of duplicity. Even if the implementation is like TEST and SAFE_CLOSE(), it's a fewer lines (+ function name as a self docs).

void fsopen_supported_by_kernel(void)

{
	TEST(tst_syscall(__NR_fsopen, NULL, 0));
	if (TST_RET != -1)
		SAFE_CLOSE(TST_RET);
}

For your change for pidfd_open03.c:

static void setup(void)

{
	int pidfd = -1;

	// Check if pidfd_open(2) is not supported
	pidfd = tst_syscall(__NR_pidfd_open, getpid(), 0);
	if (pidfd != -1)
		SAFE_CLOSE(pidfd);
}
  static struct tst_test test = {
(quoted diff) +1 -1
-       .min_kver = "5.3",
+       .setup = setup,
Quoted 1 line

How about to call the function pidfd_open_supported_by_kernel()?

OK

Quoted 1 line

Than you can remove the comment (which BTW should use C style /* */).

OK

Quoted 1 line

And IMHO you don't have to assign pidfd to -1.

In pidfd_open_supported_by_kernel(), do you want to drop 'pidfd = -1' directly or drop 'pidfd = -1' by using TEST()?

Best Regards,
Xiao Yang

Quoted 7 lines

Kind regards,
Petr

.

#4 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Hi Yang,

Quoted 1 line

Thanks a lot for your quick reply.

Thanks for a patience (we're not working just on LTP unfortunately).

Quoted 2 lines

Resetting errno may not necessary because errno will be set again when fd == -1.

Agree, I'm just careful, thus asking :).

Quoted 11 lines
Quoted 10 lines
Quoted 9 lines

2) tst_syscall() is enough to check the support of pidfd_open() and I don't want to define check function as fsopen_supported_by_kernel() does.
Do you think so?

BTW:
I don't like the implementation of fsopen_supported_by_kernel():
a) syscall()/tst_syscall() is enough to check the support of pidfd_open(2) and 'tst_kvercmp(5, 2, 0))< 0' will skip the check if

+1 for tst_syscall()

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

a kernel on distribution is newer than v5.2 but drop the support of pidfd_open(2) on purpose.

"drop support of pidfd_open(2) on purpose": would anybody has a reason to do that?

As my pervious mail said, It is just a possible situation? for example:
Upstream kernel introduces btrfs filesystem long long ago but the kernel of RHEL8 drops btrfs filesystem because of some reasons.

I guess filesystem changes are the most frequent. But as I said, I wouldn't mind this implementation:
void fsopen_supported_by_kernel(void)

{
	TEST(tst_syscall(__NR_fsopen, NULL, 0));
	if (TST_RET != -1)
		SAFE_CLOSE(TST_RET);
}
Quoted 2 lines

It is just a reason used to explain why I want to drop the kernel version check.

...

Quoted 4 lines
Quoted 1 line

How about to call the function pidfd_open_supported_by_kernel()?

OK

Quoted 8 lines
Quoted 1 line

Than you can remove the comment (which BTW should use C style /* */).

OK

Quoted 1 line

And IMHO you don't have to assign pidfd to -1.

In pidfd_open_supported_by_kernel(), do you want to drop 'pidfd = -1' directly or drop 'pidfd = -1' by using TEST()?

I meant (as it's always assigned by the call, it's just a nit.):
-int pidfd = -1;
+int pidfd;

Kind regards,
Petr

#5 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Hi Yang,

one more note:

Quoted 11 lines
Quoted 3 lines

As my pervious mail said, It is just a possible situation? for example:
Upstream kernel introduces btrfs filesystem long long ago but the kernel of RHEL8 drops btrfs filesystem because of some reasons.

I guess filesystem changes are the most frequent. But as I said, I wouldn't mind this implementation:
void fsopen_supported_by_kernel(void)

{
	TEST(tst_syscall(__NR_fsopen, NULL, 0));
	if (TST_RET != -1)
		SAFE_CLOSE(TST_RET);
}

BTW the same approach is used in include/lapi/openat2.h

void openat2_supported_by_kernel(void)

{
	if ((tst_kvercmp(5, 6, 0)) < 0) {
		/* Check if the syscall is backported on an older kernel */
		TEST(syscall(__NR_openat2, -1, NULL, NULL, 0));
		if (TST_RET == -1 && TST_ERR == ENOSYS)
			tst_brk(TCONF, "Test not supported on kernel version < v5.6");
	}
}

and clone3_supported_by_kernel(). Both merged by Cyril.

To be honest I like this approach, because 1) it defines when new syscall was backported 2) if there is really problem that some functionality was removed, we can always handle it. But IMHO that's going to be rare (btrfs removed in RHEL 8 is IMHO because RHEL does not want to support it, but that would not happen for syscalls).

I'd also like to be consistent how we handle these new syscalls.

Kind regards,
Petr

author
#6 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Quoted 27 lines · Petr Vorel

Hi Yang,

one more note:

Quoted 11 lines
Quoted 3 lines

As my pervious mail said, It is just a possible situation? for example:
Upstream kernel introduces btrfs filesystem long long ago but the kernel of RHEL8 drops btrfs filesystem because of some reasons.

I guess filesystem changes are the most frequent. But as I said, I wouldn't mind this implementation:
void fsopen_supported_by_kernel(void)

{
	TEST(tst_syscall(__NR_fsopen, NULL, 0));
	if (TST_RET != -1)
		SAFE_CLOSE(TST_RET);
}

BTW the same approach is used in include/lapi/openat2.h

void openat2_supported_by_kernel(void)

{
	if ((tst_kvercmp(5, 6, 0))<  0) {
		/* Check if the syscall is backported on an older kernel */
		TEST(syscall(__NR_openat2, -1, NULL, NULL, 0));
		if (TST_RET == -1&&  TST_ERR == ENOSYS)
			tst_brk(TCONF, "Test not supported on kernel version<  v5.6");
	}
}
Quoted 4 lines

and clone3_supported_by_kernel(). Both merged by Cyril.

To be honest I like this approach, because 1) it defines when new syscall was backported

Hi Petr?

Hmm, the reason seems a little weak, it can be done by adding a comment(e.g. "the syscall is introduced since v5.6.0").

2) if there is really problem that some functionality was removed, we

Quoted 3 lines

can always handle it. But IMHO that's going to be rare (btrfs removed in RHEL 8 is IMHO because RHEL does not want to support it, but that would not happen for syscalls).

Without the rare situation, I also think tst_syscall() is enough to check the support of syscall.

Quoted 2 lines

I'd also like to be consistent how we handle these new syscalls.

Agreed.

I also think if we can implement common func(e.g.
syscall_supported_by_kernel()).

Best Regards,
Xiao Yang

Quoted 7 lines

Kind regards,
Petr

.

#7 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Hi Yang,

Quoted 5 lines
Quoted 2 lines

To be honest I like this approach, because 1) it defines when new syscall was backported

Hmm, the reason seems a little weak, it can be done by adding a comment(e.g. "the syscall is introduced since v5.6.0").

Sure, that would work as well.

Quoted 7 lines

2) if there is really problem that some functionality was removed, we

Quoted 3 lines

can always handle it. But IMHO that's going to be rare (btrfs removed in RHEL 8 is IMHO because RHEL does not want to support it, but that would not happen for syscalls).

Without the rare situation, I also think tst_syscall() is enough to check the support of syscall.

Well, nothing that much important, but I'd like to hear the opinion of other maintainers. BTW We now concentrate on pre-release fixes.

Quoted 5 lines
Quoted 1 line

I'd also like to be consistent how we handle these new syscalls.

Agreed.

I also think if we can implement common func(e.g.
syscall_supported_by_kernel()).

Sure, feel free to send a patch (could be a macro).

Kind regards,
Petr

author
#8 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Quoted 10 lines · Petr Vorel

Hi Yang,

Quoted 5 lines
Quoted 2 lines

To be honest I like this approach, because 1) it defines when new syscall was backported

Hmm, the reason seems a little weak, it can be done by adding a comment(e.g. "the syscall is introduced since v5.6.0").

Sure, that would work as well.

Quoted 1 line

2) if there is really problem that some functionality was removed, we

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

can always handle it. But IMHO that's going to be rare (btrfs removed in RHEL 8 is IMHO because RHEL does not want to support it, but that would not happen for syscalls).

Without the rare situation, I also think tst_syscall() is enough to check the support of syscall.

Well, nothing that much important, but I'd like to hear the opinion of other maintainers. BTW We now concentrate on pre-release fixes.

Hi Petr,

Sure.

Quoted 7 lines
Quoted 6 lines
Quoted 2 lines

I'd also like to be consistent how we handle these new syscalls.

Agreed.

I also think if we can implement common func(e.g.
syscall_supported_by_kernel()).

Sure, feel free to send a patch (could be a macro).

OK, I will send a patch after the release.

Best Regards,
Xiao Yang

Quoted 8 lines

Kind regards,
Petr

.

author
#9 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

In reply to Xiao Yang

1) Drop .min_kver flag directly because of two following reasons:
a) pidfd_open(2) may be backported to old kernel which is less than v5.3 so kernel version check is meaningless.
b) tst_syscall() can report TCONF if pidfd_open(2) is not supported.
2) For pidfd_open03.c, check if pidfd_open(2) is not supported before calling fork().

---
 testcases/kernel/syscalls/pidfd_open/pidfd_open01.c |  1 -
 testcases/kernel/syscalls/pidfd_open/pidfd_open02.c |  1 -
 testcases/kernel/syscalls/pidfd_open/pidfd_open03.c | 12 +++++++++++-
 3 files changed, 11 insertions(+), 3 deletions(-)
testcases/kernel/syscalls/pidfd_open/pidfd_open01.c +0 -1
index ba1580bc7..6f5114f68 100644
@@ -29,6 +29,5 @@ static void run(void)
 }
 
 static struct tst_test test = {
-	.min_kver = "5.3",
 	.test_all = run,
 };
testcases/kernel/syscalls/pidfd_open/pidfd_open02.c +0 -1
index dc86cae7a..a7328ddfe 100644
@@ -51,7 +51,6 @@ static void run(unsigned int n)
 }
 
 static struct tst_test test = {
-	.min_kver = "5.3",
 	.tcnt = ARRAY_SIZE(tcases),
 	.test = run,
 	.setup = setup,
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +11 -1
index 48470e5e1..9e27ee75e 100644
@@ -49,8 +49,18 @@ static void run(void)
 		tst_res(TPASS, "pidfd_open() passed");
 }
 
+static void setup(void)
+{
+	int pidfd = -1;
+
+	// Check if pidfd_open(2) is not supported
+	pidfd = tst_syscall(__NR_pidfd_open, getpid(), 0);
+	if (pidfd != -1)
+		SAFE_CLOSE(pidfd);
+}
+
 static struct tst_test test = {
-	.min_kver = "5.3",
+	.setup = setup,
 	.test_all = run,
 	.forks_child = 1,
 	.needs_checkpoints = 1,
Signature · 1 line
2.21.0
#10 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

Quoted 9 lines · Xiao Yang

1) Drop .min_kver flag directly because of two following reasons:
a) pidfd_open(2) may be backported to old kernel which is less than v5.3 so kernel version check is meaningless.
b) tst_syscall() can report TCONF if pidfd_open(2) is not supported.
2) For pidfd_open03.c, check if pidfd_open(2) is not supported before calling fork().

You can not apply someone's tag without them explicitly asking you to.

Quoted 37 lines
---
 testcases/kernel/syscalls/pidfd_open/pidfd_open01.c |  1 -
 testcases/kernel/syscalls/pidfd_open/pidfd_open02.c |  1 -
 testcases/kernel/syscalls/pidfd_open/pidfd_open03.c | 12 +++++++++++-
 3 files changed, 11 insertions(+), 3 deletions(-)
testcases/kernel/syscalls/pidfd_open/pidfd_open01.c +0 -1
index ba1580bc7..6f5114f68 100644
@@ -29,6 +29,5 @@ static void run(void)
 }
 
 static struct tst_test test = {
-	.min_kver = "5.3",
 	.test_all = run,
 };
testcases/kernel/syscalls/pidfd_open/pidfd_open02.c +0 -1
index dc86cae7a..a7328ddfe 100644
@@ -51,7 +51,6 @@ static void run(unsigned int n)
 }
 
 static struct tst_test test = {
-	.min_kver = "5.3",
 	.tcnt = ARRAY_SIZE(tcases),
 	.test = run,
 	.setup = setup,
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +0 -0
index 48470e5e1..9e27ee75e 100644
@@ -49,8 +49,18 @@ static void run(void)
 		tst_res(TPASS, "pidfd_open() passed");
 }
 

testcases/kernel/syscalls/pidfd_open/pidfd_open03.c static void run(void) · line 61

Quoted 10 lines
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +10 -0
+static void setup(void)
+{
+	int pidfd = -1;
+
+	// Check if pidfd_open(2) is not supported
+	pidfd = tst_syscall(__NR_pidfd_open, getpid(), 0);
+	if (pidfd != -1)
+		SAFE_CLOSE(pidfd);
+}
+

This only fixes the 3rd test, the other two will have an issue now.

Quoted 10 lines
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +1 -1
 static struct tst_test test = {
-	.min_kver = "5.3",
+	.setup = setup,
 	.test_all = run,
 	.forks_child = 1,
 	.needs_checkpoints = 1,
Signature · 1 line
2.21.0
Signature · 1 line
viresh
author
#11 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

Quoted 8 lines · Viresh Kumar
Quoted 7 lines · Xiao Yang

1) Drop .min_kver flag directly because of two following reasons:
a) pidfd_open(2) may be backported to old kernel which is less than v5.3 so kernel version check is meaningless.
b) tst_syscall() can report TCONF if pidfd_open(2) is not supported.
2) For pidfd_open03.c, check if pidfd_open(2) is not supported before calling fork().

Quoted 4 lines
Quoted 2 lines

You can not apply someone's tag without them explicitly asking you to.

Hi Viresh,

OK, I can remove it.

Quoted 38 lines
Quoted 38 lines
---
  testcases/kernel/syscalls/pidfd_open/pidfd_open01.c |  1 -
  testcases/kernel/syscalls/pidfd_open/pidfd_open02.c |  1 -
  testcases/kernel/syscalls/pidfd_open/pidfd_open03.c | 12 +++++++++++-
  3 files changed, 11 insertions(+), 3 deletions(-)
testcases/kernel/syscalls/pidfd_open/pidfd_open01.c +0 -1
index ba1580bc7..6f5114f68 100644
@@ -29,6 +29,5 @@ static void run(void)
  }
 
  static struct tst_test test = {
-	.min_kver = "5.3",
  	.test_all = run,
  };
testcases/kernel/syscalls/pidfd_open/pidfd_open02.c +0 -1
index dc86cae7a..a7328ddfe 100644
@@ -51,7 +51,6 @@ static void run(unsigned int n)
  }
 
  static struct tst_test test = {
-	.min_kver = "5.3",
  	.tcnt = ARRAY_SIZE(tcases),
  	.test = run,
  	.setup = setup,
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +0 -0
index 48470e5e1..9e27ee75e 100644
@@ -49,8 +49,18 @@ static void run(void)
  		tst_res(TPASS, "pidfd_open() passed");
  }
 
Quoted 12 lines

testcases/kernel/syscalls/pidfd_open/pidfd_open03.c static void run(void) · line 61

Quoted 10 lines
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +10 -0
+static void setup(void)
+{
+	int pidfd = -1;
+
+	// Check if pidfd_open(2) is not supported
+	pidfd = tst_syscall(__NR_pidfd_open, getpid(), 0);
+	if (pidfd != -1)
+		SAFE_CLOSE(pidfd);
+}
+

This only fixes the 3rd test, the other two will have an issue now.

Could you tell which issue happen? Thanks a lot.
The other two don't need the extra check because the implementation of pidfd_open() can do it well. For 3rd test, I want to check the support of pidfs_open() before doing fork().

Thanks,
Xiao Yang

Quoted 12 lines
Quoted 11 lines
testcases/kernel/syscalls/pidfd_open/pidfd_open03.c +1 -2
  static struct tst_test test = {
-	.min_kver = "5.3",
+	.setup = setup,
  	.test_all = run,
  	.forks_child = 1,
  	.needs_checkpoints = 1,
--

2.21.0

#12 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

Quoted 4 lines · Xiao Yang

Could you tell which issue happen? Thanks a lot.
The other two don't need the extra check because the implementation of pidfd_open() can do it well. For 3rd test, I want to check the support of pidfs_open() before doing fork().

What I meant was that the solution needs to be consistent across tests. For example, with the current change the run() function will run for all tests in pidfd_open02.c and print the message that syscall isn't supported, while it would be better to run it only once in setup and get done with it. i.e. 1 message instead of 3 similar ones.

Signature · 1 line
viresh
author
#13 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

? 2020/5/13 14:13, Viresh Kumar ??:

Quoted 7 lines · Xiao Yang
Quoted 4 lines

Could you tell which issue happen? Thanks a lot.
The other two don't need the extra check because the implementation of pidfd_open() can do it well. For 3rd test, I want to check the support of pidfs_open() before doing fork().

What I meant was that the solution needs to be consistent across

Hi Viresh,

Current change can do correct check for pidfd_open[1-3] so don't need to add redundant check.

Quoted 5 lines

tests. For example, with the current change the run() function will run for all tests in pidfd_open02.c and print the message that syscall isn't supported, while it would be better to run it only once in setup and get done with it. i.e. 1 message instead of 3 similar ones.

Are you sure?

Triggering first tst_brk(TCONF, ...) will break the whole test instead of a subtest.

Thanks,
Xiao Yang

#14 lore raw

PATCH v2 2/2 syscalls/pidfd_open*.c: Drop .min_kver flag

Quoted 7 lines · Xiao Yang

? 2020/5/13 14:13, Viresh Kumar ??:

Quoted 6 lines · Xiao Yang
Quoted 4 lines

Could you tell which issue happen? Thanks a lot.
The other two don't need the extra check because the implementation of pidfd_open() can do it well. For 3rd test, I want to check the support of pidfs_open() before doing fork().

Quoted 5 lines
Quoted 1 line

What I meant was that the solution needs to be consistent across

Hi Viresh,

Current change can do correct check for pidfd_open[1-3] so don't need to add redundant check.

How will a check from file 03.c check it for 01.c or 02.c ?

Quoted 10 lines
Quoted 5 lines

tests. For example, with the current change the run() function will run for all tests in pidfd_open02.c and print the message that syscall isn't supported, while it would be better to run it only once in setup and get done with it. i.e. 1 message instead of 3 similar ones.

Are you sure?

Triggering first tst_brk(TCONF, ...) will break the whole test instead of a subtest.

Ahh, yes. I missed the tst_brk part :(

But I will still like it to be consistent across files. But anyway, whatever you and Cyril decide is fine :)

Thanks.

Signature · 1 line
viresh
#15 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

In reply to Xiao Yang

Hi!

Quoted 13 lines

pidfd_open(2) will set close-on-exec flag on the file descriptor as it manpage states, so check close-on-exec flag by fcntl(2).

BTW:
I tried to pass (long) TST_RET to fcntl() but triggered the following
compiler warning, so pass (int) pidfd instead.
------------------------------------------------------
In file included from pidfd_open01.c:9:
pidfd_open01.c: In function ???run???:
../../../../include/tst_test.h:76:41: warning: format ???%i??? expects argument of type ???int???, but argument 5 has type ???long int??? [-Wformat=]
76 | tst_brk_(__FILE__, __LINE__, (ttype), (arg_fmt), ##__VA_ARGS__);\
| ^~~~~~~~~
../../../../include/tst_safe_macros.h:224:5: note: in expansion of macro ???tst_brk???

Quoted 4 lines
  224 |     tst_brk(TBROK | TERRNO,                          \
      |     ^~~~~~~

pidfd_open01.c:20:9: note: in expansion of macro ???SAFE_FCNTL???
20 | flag = SAFE_FCNTL(TST_RET, F_GETFD);

You can either cast the TST_RET to int as SAFE_FCNTL((int)TST_RET, ...) or if you decide to store the pidfd into an int variable there is no added value in using the TEST() macro here so the code should just do pidfd = pidfd_open(...) and use the pidfd instead of TST_RET.

Signature · 2 lines
Cyril Hrubis
[email protected]
author
#16 lore raw

PATCH v2 1/2 syscalls/pidfd_open01.c: Add check for close-on-exec flag

Quoted 19 lines · Cyril Hrubis

Hi!

Quoted 10 lines

pidfd_open(2) will set close-on-exec flag on the file descriptor as it manpage states, so check close-on-exec flag by fcntl(2).

BTW:
I tried to pass (long) TST_RET to fcntl() but triggered the following
compiler warning, so pass (int) pidfd instead.
------------------------------------------------------
In file included from pidfd_open01.c:9:
pidfd_open01.c: In function ???run???:
../../../../include/tst_test.h:76:41: warning: format ???%i??? expects argument of type ???int???, but argument 5 has type ???long int??? [-Wformat=]

Quoted 7 lines
    76 |   tst_brk_(__FILE__, __LINE__, (ttype), (arg_fmt), ##__VA_ARGS__);\
       |                                         ^~~~~~~~~
../../../../include/tst_safe_macros.h:224:5: note: in expansion of macro ???tst_brk???
   224 |     tst_brk(TBROK | TERRNO,                          \
       |     ^~~~~~~
pidfd_open01.c:20:9: note: in expansion of macro ???SAFE_FCNTL???
    20 |  flag = SAFE_FCNTL(TST_RET, F_GETFD);
Quoted 4 lines

You can either cast the TST_RET to int as SAFE_FCNTL((int)TST_RET, ...) or if you decide to store the pidfd into an int variable there is no added value in using the TEST() macro here so the code should just do pidfd = pidfd_open(...) and use the pidfd instead of TST_RET.

Hi Cyril,

Thanks for your reply.
I also think it is OK to replace TEST() and TST_RET directly.

Thanks,
Xiao Yang

Quoted 1 line