Re: [PATCH v6 3/3] io_uring: enable per-io hinting capability
Hannes Reinecke <[email protected]>
| Newsgroups | gmane.linux.kernel.aio.general,gmane.linux.file-systems,gmane.linux.kernel.io-uring,gmane.linux.block |
|---|---|
| Message-ID | <[email protected]> |
On 9/24/24 11:24, Kanchan Joshi wrote: > With F_SET_RW_HINT fcntl, user can set a hint on the file inode, and > all the subsequent writes on the file pass that hint value down. > This can be limiting for large files (and for block device) as all the > writes can be tagged with only one lifetime hint value. > Concurrent writes (with different hint values) are hard to manage. > Per-IO hinting solves that problem. > > Allow userspace to pass the write hint type and its value in the SQE. > Two new fields are carved in the leftover space of SQE: > __u8 hint_type; > __u64 hint_val; > > Adding the hint_type helps in keeping the interface extensible for future > use. > At this point only one type TYPE_WRITE_LIFETIME_HINT is supported. With > this type, user can pass the lifetime hint values that are currently > supported by F_SET_RW_HINT fcntl. > > The write handlers (io_prep_rw, io_write) process the hint type/value > and hint value is passed to lower-layer using kiocb. This is good for > supporting direct IO, but not when kiocb is not available (e.g., > buffered IO). > > In general, per-io hints take the precedence on per-inode hints. > Three cases to consider: > > Case 1: When hint_type is 0 (explicitly, or implicitly as SQE fields are > initialized to 0), this means user did not send any hint. The per-inode > hint values are set in the kiocb (as before). > > Case 2: When hint_type is TYPE_WRITE_LIFETIME_HINT, the hint_value is > set into the kiocb after sanity checking. > > Case 3: When hint_type is anything else, this is flagged as an error > and write is failed. > > Signed-off-by: Kanchan Joshi <[email protected]> > Signed-off-by: Nitesh Shetty <[email protected]> > --- > fs/fcntl.c | 22 ---------------------- > include/linux/rw_hint.h | 24 ++++++++++++++++++++++++ > include/uapi/linux/io_uring.h | 10 ++++++++++ > io_uring/rw.c | 21 ++++++++++++++++++++- > 4 files changed, 54 insertions(+), 23 deletions(-) > > diff --git a/fs/fcntl.c b/fs/fcntl.c > index 081e5e3d89ea..2eb78035a350 100644 > --- a/fs/fcntl.c > +++ b/fs/fcntl.c > @@ -334,28 +334,6 @@ static int f_getowner_uids(struct file *filp, unsigned long arg) > } > #endif > > -static bool rw_hint_valid(u64 hint) > -{ > - BUILD_BUG_ON(WRITE_LIFE_NOT_SET != RWH_WRITE_LIFE_NOT_SET); > - BUILD_BUG_ON(WRITE_LIFE_NONE != RWH_WRITE_LIFE_NONE); > - BUILD_BUG_ON(WRITE_LIFE_SHORT != RWH_WRITE_LIFE_SHORT); > - BUILD_BUG_ON(WRITE_LIFE_MEDIUM != RWH_WRITE_LIFE_MEDIUM); > - BUILD_BUG_ON(WRITE_LIFE_LONG != RWH_WRITE_LIFE_LONG); > - BUILD_BUG_ON(WRITE_LIFE_EXTREME != RWH_WRITE_LIFE_EXTREME); > - > - switch (hint) { > - case RWH_WRITE_LIFE_NOT_SET: > - case RWH_WRITE_LIFE_NONE: > - case RWH_WRITE_LIFE_SHORT: > - case RWH_WRITE_LIFE_MEDIUM: > - case RWH_WRITE_LIFE_LONG: > - case RWH_WRITE_LIFE_EXTREME: > - return true; > - default: > - return false; > - } > -} > - > static long fcntl_get_rw_hint(struct file *file, unsigned int cmd, > unsigned long arg) > { > diff --git a/include/linux/rw_hint.h b/include/linux/rw_hint.h > index 309ca72f2dfb..f4373a71ffed 100644 > --- a/include/linux/rw_hint.h > +++ b/include/linux/rw_hint.h > @@ -21,4 +21,28 @@ enum rw_hint { > static_assert(sizeof(enum rw_hint) == 1); > #endif > > +#define WRITE_LIFE_INVALID (RWH_WRITE_LIFE_EXTREME + 1) > + > +static inline bool rw_hint_valid(u64 hint) > +{ > + BUILD_BUG_ON(WRITE_LIFE_NOT_SET != RWH_WRITE_LIFE_NOT_SET); > + BUILD_BUG_ON(WRITE_LIFE_NONE != RWH_WRITE_LIFE_NONE); > + BUILD_BUG_ON(WRITE_LIFE_SHORT != RWH_WRITE_LIFE_SHORT); > + BUILD_BUG_ON(WRITE_LIFE_MEDIUM != RWH_WRITE_LIFE_MEDIUM); > + BUILD_BUG_ON(WRITE_LIFE_LONG != RWH_WRITE_LIFE_LONG); > + BUILD_BUG_ON(WRITE_LIFE_EXTREME != RWH_WRITE_LIFE_EXTREME); > + > + switch (hint) { > + case RWH_WRITE_LIFE_NOT_SET: > + case RWH_WRITE_LIFE_NONE: > + case RWH_WRITE_LIFE_SHORT: > + case RWH_WRITE_LIFE_MEDIUM: > + case RWH_WRITE_LIFE_LONG: > + case RWH_WRITE_LIFE_EXTREME: > + return true; > + default: > + return false; > + } > +} > + > #endif /* _LINUX_RW_HINT_H */ > diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h > index 1fe79e750470..e21a74dd0c49 100644 > --- a/include/uapi/linux/io_uring.h > +++ b/include/uapi/linux/io_uring.h > @@ -98,6 +98,11 @@ struct io_uring_sqe { > __u64 addr3; > __u64 __pad2[1]; > }; > + struct { > + /* To send per-io hint type/value with write command */ > + __u64 hint_val; > + __u8 hint_type; > + }; Why is 'hint_val' 64 bits? Everything else is 8 bytes, so wouldn't it be better to shorten that? As it stands the new struct will introduce a hole of 24 bytes after 'hint_type'. Cheers, Hannes -- Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich -- To unsubscribe, send a message with 'unsubscribe linux-aio' in the body to [email protected]. For more info on Linux AIO, see: http://www.kvack.org/aio/ Don't email: <a href=mailto:"[email protected]">[email protected]</a>