2020-08-31 20:40:59

by Jens Axboe

[permalink] [raw]
Subject: [PATCH] fs: align IOCB_* flags with RWF_* flags

We have a set of flags that are shared between the two and inherired
in kiocb_set_rw_flags(), but we check and set these individually.
Reorder the IOCB flags so that the bottom part of the space is synced
with the RWF flag space, and then we can do them all in one mask and
set operation.

The only exception is RWF_SYNC, which needs to mark IOCB_SYNC and
IOCB_DSYNC. Do that one separately.

This shaves 15 bytes of text from kiocb_set_rw_flags() for me.

Signed-off-by: Jens Axboe <[email protected]>

---

diff --git a/include/linux/fs.h b/include/linux/fs.h
index 7519ae003a08..c82360600ae4 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -310,17 +310,20 @@ enum rw_hint {
WRITE_LIFE_EXTREME = RWH_WRITE_LIFE_EXTREME,
};

-#define IOCB_EVENTFD (1 << 0)
-#define IOCB_APPEND (1 << 1)
-#define IOCB_DIRECT (1 << 2)
-#define IOCB_HIPRI (1 << 3)
-#define IOCB_DSYNC (1 << 4)
-#define IOCB_SYNC (1 << 5)
-#define IOCB_WRITE (1 << 6)
-#define IOCB_NOWAIT (1 << 7)
+/* Match RWF_* bits to IOCB bits */
+#define IOCB_HIPRI (__force int) RWF_HIPRI
+#define IOCB_DSYNC (__force int) RWF_DSYNC
+#define IOCB_SYNC (__force int) RWF_SYNC
+#define IOCB_NOWAIT (__force int) RWF_NOWAIT
+#define IOCB_APPEND (__force int) RWF_APPEND
+
+/* non-RWF related bits - start at 16 */
+#define IOCB_EVENTFD (1 << 16)
+#define IOCB_DIRECT (1 << 17)
+#define IOCB_WRITE (1 << 18)
/* iocb->ki_waitq is valid */
-#define IOCB_WAITQ (1 << 8)
-#define IOCB_NOIO (1 << 9)
+#define IOCB_WAITQ (1 << 19)
+#define IOCB_NOIO (1 << 20)

struct kiocb {
struct file *ki_filp;
@@ -3317,6 +3320,9 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
{
int kiocb_flags = 0;

+ /* make sure there's no overlap between RWF and private IOCB flags */
+ BUILD_BUG_ON((__force int) RWF_SUPPORTED & IOCB_EVENTFD);
+
if (!flags)
return 0;
if (unlikely(flags & ~RWF_SUPPORTED))
@@ -3325,16 +3331,11 @@ static inline int kiocb_set_rw_flags(struct kiocb *ki, rwf_t flags)
if (flags & RWF_NOWAIT) {
if (!(ki->ki_filp->f_mode & FMODE_NOWAIT))
return -EOPNOTSUPP;
- kiocb_flags |= IOCB_NOWAIT | IOCB_NOIO;
+ kiocb_flags |= IOCB_NOIO;
}
- if (flags & RWF_HIPRI)
- kiocb_flags |= IOCB_HIPRI;
- if (flags & RWF_DSYNC)
- kiocb_flags |= IOCB_DSYNC;
+ kiocb_flags |= (__force int) (flags & RWF_SUPPORTED);
if (flags & RWF_SYNC)
- kiocb_flags |= (IOCB_DSYNC | IOCB_SYNC);
- if (flags & RWF_APPEND)
- kiocb_flags |= IOCB_APPEND;
+ kiocb_flags |= IOCB_DSYNC;

ki->ki_flags |= kiocb_flags;
return 0;

--
Jens Axboe


2020-09-07 19:45:26

by Jens Axboe

[permalink] [raw]
Subject: Re: [PATCH] fs: align IOCB_* flags with RWF_* flags

On 8/31/20 12:08 PM, Jens Axboe wrote:
> We have a set of flags that are shared between the two and inherired
> in kiocb_set_rw_flags(), but we check and set these individually.
> Reorder the IOCB flags so that the bottom part of the space is synced
> with the RWF flag space, and then we can do them all in one mask and
> set operation.
>
> The only exception is RWF_SYNC, which needs to mark IOCB_SYNC and
> IOCB_DSYNC. Do that one separately.
>
> This shaves 15 bytes of text from kiocb_set_rw_flags() for me.

Al, are you OK with me queueing this one up? I'm fine either way,
obviously no dependencies for this one, just want to ensure it isn't
dropped.

--
Jens Axboe

2020-09-07 20:16:52

by Matthew Wilcox

[permalink] [raw]
Subject: Re: [PATCH] fs: align IOCB_* flags with RWF_* flags

On Mon, Aug 31, 2020 at 12:08:10PM -0600, Jens Axboe wrote:
> We have a set of flags that are shared between the two and inherired
> in kiocb_set_rw_flags(), but we check and set these individually.
> Reorder the IOCB flags so that the bottom part of the space is synced
> with the RWF flag space, and then we can do them all in one mask and
> set operation.
>
> The only exception is RWF_SYNC, which needs to mark IOCB_SYNC and
> IOCB_DSYNC. Do that one separately.
>
> This shaves 15 bytes of text from kiocb_set_rw_flags() for me.
>
> Signed-off-by: Jens Axboe <[email protected]>

Suggested-by: Matthew Wilcox (Oracle) <[email protected]>

2020-09-08 02:15:06

by Jens Axboe

[permalink] [raw]
Subject: Re: [PATCH] fs: align IOCB_* flags with RWF_* flags

On 9/7/20 2:15 PM, Matthew Wilcox wrote:
> On Mon, Aug 31, 2020 at 12:08:10PM -0600, Jens Axboe wrote:
>> We have a set of flags that are shared between the two and inherired
>> in kiocb_set_rw_flags(), but we check and set these individually.
>> Reorder the IOCB flags so that the bottom part of the space is synced
>> with the RWF flag space, and then we can do them all in one mask and
>> set operation.
>>
>> The only exception is RWF_SYNC, which needs to mark IOCB_SYNC and
>> IOCB_DSYNC. Do that one separately.
>>
>> This shaves 15 bytes of text from kiocb_set_rw_flags() for me.
>>
>> Signed-off-by: Jens Axboe <[email protected]>
>
> Suggested-by: Matthew Wilcox (Oracle) <[email protected]>

Added

--
Jens Axboe