c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2026-05-22 10:37:35 +03:00
Code Issues Packages Projects Releases Wiki Activity

Wire O_DIRECT also to Uncached I/O (#17218)

Browse Source 
Before Direct I/O was implemented, I've implemented lighter version
I called Uncached I/O.  It uses normal DMU/ARC data path with some
optimizations, but evicts data from caches as soon as possible and
reasonable.  Originally I wired it only to a primarycache property,
but now completing the integration all the way up to the VFS.

While Direct I/O has the lowest possible memory bandwidth usage,
it also has a significant number of limitations.  It require I/Os
to be page aligned, does not allow speculative prefetch, etc.  The
Uncached I/O does not have those limitations, but instead require
additional memory copy, though still one less than regular cached
I/O.  As such it should fill the gap in between.  Considering this
I've disabled annoying EINVAL errors on misaligned requests, adding
a tunable for those who wants to test their applications.

To pass the information between the layers I had to change a number
of APIs.  But as side effect upper layers can now control not only
the caching, but also speculative prefetch.  I haven't wired it to
VFS yet, since it require looking on some OS specifics.  But while
there I've implemented speculative prefetch of indirect blocks for
Direct I/O, controllable via all the same mechanisms.

Signed-off-by:	Alexander Motin <mav@FreeBSD.org>
Sponsored by:	iXsystems, Inc.
Fixes #17027
Reviewed-by: Rob Norris <robn@despairlabs.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
This commit is contained in:
Alexander Motin
2025-05-13 17:26:55 -04:00
committed by GitHub
parent e2ba0f7643
commit 734eba251d
35 changed files with 397 additions and 294 deletions
Download Patch File Download Diff File Expand all files Collapse all files
module/os/freebsd/zfs/dmu_os.c
+11 -4
View File
@@ -41,7 +41,6 @@
#include <sys/dsl_pool.h>
#include <sys/dsl_synctask.h>
#include <sys/dsl_prop.h>
#include <sys/dmu_zfetch.h>
#include <sys/zfs_ioctl.h>
#include <sys/zap.h>
#include <sys/zio_checksum.h>
@@ -71,6 +70,7 @@ dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
struct sf_buf *sf;
int numbufs, i;
int err;
dmu_flags_t flags = 0;
if (size == 0)
return (0);
@@ -94,10 +94,17 @@ dmu_write_pages(objset_t *os, uint64_t object, uint64_t offset, uint64_t size,
ASSERT(i == 0 || i == numbufs-1 || tocpy == db->db_size);
if (tocpy == db->db_size)
if (tocpy == db->db_size) {
dmu_buf_will_fill(db, tx, B_FALSE);
else
dmu_buf_will_dirty(db, tx);
} else {
if (i == numbufs - 1 && bufoff + tocpy < db->db_size) {
if (bufoff == 0)
flags |= DMU_PARTIAL_FIRST;
else
flags |= DMU_PARTIAL_MORE;
}
dmu_buf_will_dirty_flags(db, tx, flags);
}
for (copied = 0; copied < tocpy; copied += PAGESIZE) {
ASSERT3U(ptoa((*ma)->pindex), ==,
module/os/freebsd/zfs/zfs_racct.c
+2 -2
View File
@@ -28,7 +28,7 @@
#include <sys/racct.h>
void
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
curthread->td_ru.ru_inblock += iops;
#ifdef RACCT
@@ -46,7 +46,7 @@ zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
}
void
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
curthread->td_ru.ru_oublock += iops;
#ifdef RACCT
module/os/freebsd/zfs/zfs_vnops_os.c
+1 -1
View File
@@ -530,7 +530,7 @@ mappedread(znode_t *zp, int nbytes, zfs_uio_t *uio)
page_unhold(pp);
} else {
error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
uio, bytes);
uio, bytes, DMU_READ_PREFETCH);
}
len -= bytes;
off = 0;
module/os/freebsd/zfs/zvol_os.c
+7 -4
View File
@@ -679,7 +679,7 @@ zvol_strategy_impl(zv_request_t *zvr)
while (resid != 0 && off < volsize) {
size_t size = MIN(resid, zvol_maxphys);
if (doread) {
error = dmu_read(os, ZVOL_OBJ, off, size, addr,
error = dmu_read_by_dnode(zv->zv_dn, off, size, addr,
DMU_READ_PREFETCH);
} else {
dmu_tx_t *tx = dmu_tx_create(os);
@@ -688,7 +688,8 @@ zvol_strategy_impl(zv_request_t *zvr)
if (error) {
dmu_tx_abort(tx);
} else {
dmu_write(os, ZVOL_OBJ, off, size, addr, tx);
dmu_write_by_dnode(zv->zv_dn, off, size, addr,
tx, DMU_READ_PREFETCH);
zvol_log_write(zv, tx, off, size, commit);
dmu_tx_commit(tx);
}
@@ -834,7 +835,8 @@ zvol_cdev_read(struct cdev *dev, struct uio *uio_s, int ioflag)
if (bytes > volsize - zfs_uio_offset(&uio))
bytes = volsize - zfs_uio_offset(&uio);
error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes);
error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes,
DMU_READ_PREFETCH);
if (error) {
/* Convert checksum errors into IO errors. */
if (error == ECKSUM)
@@ -893,7 +895,8 @@ zvol_cdev_write(struct cdev *dev, struct uio *uio_s, int ioflag)
dmu_tx_abort(tx);
break;
}
error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx);
error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx,
DMU_READ_PREFETCH);
if (error == 0)
zvol_log_write(zv, tx, off, bytes, commit);
dmu_tx_commit(tx);
module/os/linux/zfs/zfs_racct.c
+4 -4
View File
@@ -30,14 +30,14 @@
#include <linux/task_io_accounting_ops.h>
void
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
task_io_account_read(size);
spa_iostats_read_add(spa, size, iops, flags);
}
void
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
task_io_account_write(size);
spa_iostats_write_add(spa, size, iops, flags);
@@ -46,13 +46,13 @@ zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
#else
void
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_read(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
(void) spa, (void) size, (void) iops, (void) flags;
}
void
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, uint32_t flags)
zfs_racct_write(spa_t *spa, uint64_t size, uint64_t iops, dmu_flags_t flags)
{
(void) spa, (void) size, (void) iops, (void) flags;
}
module/os/linux/zfs/zfs_vnops_os.c
+1 -1
View File
@@ -329,7 +329,7 @@ mappedread(znode_t *zp, int nbytes, zfs_uio_t *uio)
put_page(pp);
} else {
error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
uio, bytes);
uio, bytes, DMU_READ_PREFETCH);
}
len -= bytes;
module/os/linux/zfs/zvol_os.c
+4 -2
View File
@@ -258,7 +258,8 @@ zvol_write(zv_request_t *zvr)
dmu_tx_abort(tx);
break;
}
error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx);
error = dmu_write_uio_dnode(zv->zv_dn, &uio, bytes, tx,
DMU_READ_PREFETCH);
if (error == 0) {
zvol_log_write(zv, tx, off, bytes, sync);
}
@@ -428,7 +429,8 @@ zvol_read(zv_request_t *zvr)
if (bytes > volsize - uio.uio_loffset)
bytes = volsize - uio.uio_loffset;
error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes);
error = dmu_read_uio_dnode(zv->zv_dn, &uio, bytes,
DMU_READ_PREFETCH);
if (error) {
/* convert checksum errors into IO errors */
if (error == ECKSUM)