Skip to contentSkip to conversation

Improve ioprio tests

PATCH In review Damien Le Moal 4 participants 14 messages Jun 7, 2023 – Jun 20, 2023

lore
Share & tools
lore
Reviewed-by on v1 1/2
#2 lore raw

Re: [LTP] [PATCH 1/2] ioprio: use ioprio.h kernel header if it exists

In reply to Damien Le Moal
Quoted 8 lines · Damien Le Moal

For the ioprio system call test cases, avoid blindly defining the IOPRIO_XXX macro internally and instead use the kernel user API header file if it exists. Given that the definitions in this header file have changed over time, make sure to test for the existence of the macro IOPRIO_PRIO_LEVEL macro and define it if it does not exist. Similarly, use IOPRIO_NR_LEVELS to define IOPRIO_PRIO_NUM if that macro exists.

Yours,
Linus Walleij

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
Reviewed-by on v1 2/2
#4 lore raw

Re: [LTP] [PATCH 2/2] ioprio: Use IOPRIO_PRIO_NUM to check prio range

In reply to Damien Le Moal
Quoted 6 lines · Damien Le Moal

Use the macro IOPRIO_PRIO_NUM in prio_in_range() to check the upper bound of the valid range for priority levels. Similarly, in the test case ioprio_set03, use this macro to check for failures when the user attempts using a priority level out of range.

Yours,
Linus Walleij

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#5 lore raw

Re: [LTP] [PATCH 2/2] ioprio: Use IOPRIO_PRIO_NUM to check prio range

Quoted 10 lines · Linus Walleij
Quoted 6 lines · Damien Le Moal

Use the macro IOPRIO_PRIO_NUM in prio_in_range() to check the upper bound of the valid range for priority levels. Similarly, in the test case ioprio_set03, use this macro to check for failures when the user attempts using a priority level out of range.

Thanks Linus !

May I also ask for your review of the kernel ioprio API patch ? It is here:

https://lore.kernel.org/linux-block/[email protected]/

Quoted 3 lines

Yours,
Linus Walleij

Signature · 6 lines
Damien Le Moal
Western Digital Research


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#6 lore raw

Re: [LTP] [PATCH 2/2] ioprio: Use IOPRIO_PRIO_NUM to check prio range

Quoted 6 lines · Damien Le Moal

Thanks Linus !

May I also ask for your review of the kernel ioprio API patch ? It is here:

https://lore.kernel.org/linux-block/[email protected]/

Sadly I don't have this patch in my inbox, because it appears I fell off the block mailing list and wasn't on To or Cc. If you resend then include me on CC and I'll have a look!

Yours,
Linus Walleij

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw

PATCH 1/2 ioprio: use ioprio.h kernel header if it exists

In reply to Damien Le Moal

For the ioprio system call test cases, avoid blindly defining the IOPRIO_XXX macro internally and instead use the kernel user API header file if it exists. Given that the definitions in this header file have changed over time, make sure to test for the existence of the macro IOPRIO_PRIO_LEVEL macro and define it if it does not exist. Similarly, use IOPRIO_NR_LEVELS to define IOPRIO_PRIO_NUM if that macro exists.

---
 configure.ac                              |  1 +
 testcases/kernel/syscalls/ioprio/ioprio.h | 29 +++++++++++++++++------
 2 files changed, 23 insertions(+), 7 deletions(-)
configure.ac +1 -0
index 548288310..e4aa2cadf 100644
@@ -56,6 +56,7 @@ AC_CHECK_HEADERS_ONCE([ \
     linux/if_ether.h \
     linux/if_packet.h \
     linux/io_uring.h \
+    linux/ioprio.h \
     linux/keyctl.h \
     linux/mempolicy.h \
     linux/module.h \
testcases/kernel/syscalls/ioprio/ioprio.h +22 -7
index c74380475..6ca134a54 100644
@@ -6,6 +6,12 @@
 #ifndef LTP_IOPRIO_H
 #define LTP_IOPRIO_H
 
+#ifdef HAVE_LINUX_IOPRIO_H
+
+# include <linux/ioprio.h>
+
+#else
+
 enum {
 	IOPRIO_CLASS_NONE = 0,
 	IOPRIO_CLASS_RT,
@@ -19,15 +25,24 @@ enum {
 	IOPRIO_WHO_USER,
 };
 
-/* The I/O scheduler classes have 8 priorities 0..7 except for the IDLE class */
-#define IOPRIO_PRIO_NUM		8
+# define IOPRIO_CLASS_SHIFT	(13)
+# define IOPRIO_PRIO_MASK	((1UL << IOPRIO_CLASS_SHIFT) - 1)
+
+# define IOPRIO_PRIO_CLASS(data)	((data) >> IOPRIO_CLASS_SHIFT)
+# define IOPRIO_PRIO_VALUE(class, data)	(((class) << IOPRIO_CLASS_SHIFT) | data)
 
-#define IOPRIO_CLASS_SHIFT	(13)
-#define IOPRIO_PRIO_MASK	((1UL << IOPRIO_CLASS_SHIFT) - 1)
+#endif
+
+/* The RT and BE I/O priority classes have 8 priority levels 0..7 */
+#ifdef IOPRIO_NR_LEVELS
+# define IOPRIO_PRIO_NUM		IOPRIO_NR_LEVELS
+#else
+# define IOPRIO_PRIO_NUM		8
+#endif
 
-#define IOPRIO_PRIO_CLASS(data)	((data) >> IOPRIO_CLASS_SHIFT)
-#define IOPRIO_PRIO_LEVEL(data)	((data) & IOPRIO_PRIO_MASK)
-#define IOPRIO_PRIO_VALUE(class, data)	(((class) << IOPRIO_CLASS_SHIFT) | data)
+#ifndef IOPRIO_PRIO_LEVEL
+# define IOPRIO_PRIO_LEVEL(data)	((data) & IOPRIO_PRIO_MASK)
+#endif
 
 static const char * const to_class_str[] = {
 	[IOPRIO_CLASS_NONE] = "NONE",
Signature · 5 lines
2.40.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#0 lore raw

cover Improve ioprio tests

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

Damien Le Moal (2):
ioprio: use ioprio.h kernel header if it exists
ioprio: Use IOPRIO_PRIO_NUM to check prio range

 configure.ac                                  |  1 +
 testcases/kernel/syscalls/ioprio/ioprio.h     | 33 ++++++++++++++-----
 .../kernel/syscalls/ioprio/ioprio_set03.c     |  2 +-
 3 files changed, 26 insertions(+), 10 deletions(-)
Signature · 5 lines
2.40.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw

PATCH 2/2 ioprio: Use IOPRIO_PRIO_NUM to check prio range

Use the macro IOPRIO_PRIO_NUM in prio_in_range() to check the upper bound of the valid range for priority levels. Similarly, in the test case ioprio_set03, use this macro to check for failures when the user attempts using a priority level out of range.

---
 testcases/kernel/syscalls/ioprio/ioprio.h       | 4 ++--
 testcases/kernel/syscalls/ioprio/ioprio_set03.c | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)
testcases/kernel/syscalls/ioprio/ioprio.h +2 -2
index 6ca134a54..c2115bf20 100644
@@ -61,10 +61,10 @@ static inline int sys_ioprio_set(int which, int who, int ioprio)
 	return tst_syscall(__NR_ioprio_set, which, who, ioprio);
 }
 
-/* Priority range from 0 (highest) to 7 (lowest) */
+/* Priority range from 0 (highest) to IOPRIO_PRIO_NUM (lowest) */
 static inline int prio_in_range(int prio)
 {
-	if ((prio < 0) || (prio > 7))
+	if ((prio < 0) || (prio >= IOPRIO_PRIO_NUM))
 		return 0;
 	return 1;
 }
testcases/kernel/syscalls/ioprio/ioprio_set03.c +1 -1
index b2c962a6f..d6b44df85 100644
@@ -27,7 +27,7 @@ static void run(void)
 	sys_ioprio_set(IOPRIO_WHO_PROCESS, 0,
 		       IOPRIO_PRIO_VALUE(class, 4));
 	TEST(sys_ioprio_set(IOPRIO_WHO_PROCESS, 0,
-			    IOPRIO_PRIO_VALUE(class, 8)));
+			    IOPRIO_PRIO_VALUE(class, IOPRIO_PRIO_NUM)));
 	if (TST_RET == -1) {
 		ioprio_check_setting(class, 4, 1);
 		if (errno == EINVAL)
Signature · 5 lines
2.40.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#7 lore raw
In reply to Damien Le Moal

Hi Damien,

Quoted 10 lines

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

Quoted 4 lines

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

I haven't found this patchset on https://lore.kernel.org/linux-block/,
did you send it there?

Kind regards,
Petr

Quoted 8 lines

Damien Le Moal (2):
ioprio: use ioprio.h kernel header if it exists
ioprio: Use IOPRIO_PRIO_NUM to check prio range

 configure.ac                                  |  1 +
 testcases/kernel/syscalls/ioprio/ioprio.h     | 33 ++++++++++++++-----
 .../kernel/syscalls/ioprio/ioprio_set03.c     |  2 +-
 3 files changed, 26 insertions(+), 10 deletions(-)
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#8 lore raw
Quoted 12 lines · Petr Vorel

Hi Damien,

Quoted 9 lines

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

Quoted 7 lines
Quoted 4 lines

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

I haven't found this patchset on https://lore.kernel.org/linux-block/,
did you send it there?

Hello Petr,

The patch series for LTP can be found here:
https://lore.kernel.org/ltp/[email protected]/T/#t

I just checked LTP master, and it hasn't been applied by the LTP maintainers yet.

Kind regards,
Niklas

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#9 lore raw
Quoted 26 lines · Niklas Cassel
Quoted 12 lines · Petr Vorel

Hi Damien,

Quoted 9 lines

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

Quoted 7 lines
Quoted 4 lines

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

I haven't found this patchset on https://lore.kernel.org/linux-block/,
did you send it there?

Hello Petr,

The patch series for LTP can be found here:
https://lore.kernel.org/ltp/[email protected]/T/#t

Quoted 6 lines

I just checked LTP master, and it hasn't been applied by the LTP maintainers yet.

Kind regards,
Niklas

Oh, and in case you were referring to the prerequisite kernel patch, it is already in linux-next:

https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git/commit/?h=staging&id=01584c1e233740519d0e11aa20daa323d26bf598

So right now we are just waiting for the LTP patch series to get picked up.

Kind regards,
Niklas

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#13 lore raw
In reply to Niklas Cassel via ltp

Hi all,

Quoted 13 lines · Petr Vorel
Quoted 11 lines

Hi Damien,

Quoted 9 lines

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

I'm sorry, I overlooked Linus' review [2], because it's not visible in patchwork [3].

@Damien, @Linus, @Niklas FYI you need to be subscribed to post to LTP ML (unlike vger), otherwise the message is held (and we don't check for held messages very often). @Niklas I had to push your message due this.

Quoted 7 lines
Quoted 2 lines

I haven't found this patchset on https://lore.kernel.org/linux-block/,
did you send it there?

Hello Petr,

The patch series for LTP can be found here:
https://lore.kernel.org/ltp/[email protected]/T/#t

Yes, it's in our LTP mailing list, thus on LTP on lore. But I expected that linux-block ML would be in Cc (that's why Linus Walleij didn't get it in his mailbox [4] => @Damien, do not hesitate to Cc relevant kernel (or other mailing lists), if you expect kernel maintainers to do the review.

Quoted 2 lines

I just checked LTP master, and it hasn't been applied by the LTP maintainers yet.

Yes, but do not worry, I'll merge it soon (likely today).

Thank to all for your time.

Kind regards,
Petr

[2] https://lore.kernel.org/ltp/CACRpkdaBinsAofvQgLZ5u8ScR0+yWPnQCf6E7CPtn598PN0eoQ@mail.gmail.com/
[3] https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/
[4] https://lore.kernel.org/ltp/CACRpkdaD3xM=Z0xTzHa+yH-wZ+LyhrSb3ZzGzkGdnboCaHw_+A@mail.gmail.com/

Quoted 2 lines

Kind regards,
Niklas

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#10 lore raw
In reply to Niklas Cassel via ltp
Quoted 31 lines · Niklas Cassel
Quoted 28 lines · Petr Vorel
Quoted 12 lines

Hi Damien,

Quoted 9 lines

The ioprio syscall tests rely on ltp internal definitions of the IOPRIO_XXX() macro defining priority classes and levels. With changes to the ioprio API to support command duration limits, these internal definitions are incomplete, causing test case ioprio_set03 to fail.

Avoid this issue by having the iprio tests use the kernel header file definitions if that header exists. This enables additional checks in the header file [1] which restores the expected results with test ioprio_set03.

Quoted 7 lines
Quoted 4 lines

[1] https://lore.kernel.org/linux-block/[email protected]/

Note: a review of this patch on the kernel block mailing list would be very appreciated.

I haven't found this patchset on https://lore.kernel.org/linux-block/,
did you send it there?

Hello Petr,

The patch series for LTP can be found here:
https://lore.kernel.org/ltp/[email protected]/T/#t

I just checked LTP master, and it hasn't been applied by the LTP maintainers yet.

Quoted 7 lines
Quoted 2 lines

Kind regards,
Niklas

Oh, and in case you were referring to the prerequisite kernel patch, it is already in linux-next:

https://git.kernel.org/pub/scm/linux/kernel/git/mkp/scsi.git/commit/?h=staging&id=01584c1e233740519d0e11aa20daa323d26bf598

Good to know, thank you (referring to kernel commits is always considered useful).

No, I really searched if LTP patchset was also Cc'ed to linux-block. But as the maintainers were Cc'ed I now understand why linux-block was omitted. I'm sorry for the confusion.

Kind regards,
Petr

Quoted 5 lines

So right now we are just waiting for the LTP patch series to get picked up.

Kind regards,
Niklas

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#11 lore raw
In reply to Niklas Cassel via ltp

Hi all,

patchset merged with added missing #include "config.h".

I'll do the additional cleanup work and Cc you.

Kind regards,
Petr

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#12 lore raw
Quoted 5 lines · Petr Vorel

Hi all,

patchset merged with added missing #include "config.h".

I'll do the additional cleanup work and Cc you.

I was just about to send a v3 with #include "config.h", now I don't need to :)

Thank you for your help Petr!

Kind regards,
Niklas

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp