/**
* Prepare partially written-to page for a write.
*/
static int ll_prepare_partial_page(const struct lu_env *env, struct cl_io *io,
struct cl_page *pg)
{
struct cl_attr *attr = vvp_env_thread_attr(env);
struct cl_object *obj = io->ci_obj;
struct vvp_page *vpg = cl_object_page_slice(obj, pg);
loff_t offset = cl_offset(obj, vvp_index(vpg));
int result;
cl_object_attr_lock(obj);
result = cl_object_attr_get(env, obj, attr);
cl_object_attr_unlock(obj);
if (result == 0) {
/*
* If are writing to a new page, no need to read old data.
* The extent locking will have updated the KMS, and for our
* purposes here we can treat it like i_size.
*/
if (attr->cat_kms <= offset) {
char *kaddr = ll_kmap_atomic(vpg->vpg_page, KM_USER0);
memset(kaddr, 0, cl_page_size(obj));
ll_kunmap_atomic(kaddr, KM_USER0);
} else if (vpg->vpg_defer_uptodate)
vpg->vpg_ra_used = 1;
else
result = ll_page_sync_io(env, io, pg, CRT_READ);
}
return result;
}
/**
* Helper function that if necessary adjusts file size (inode->i_size), when
* position at the offset \a pos is accessed. File size can be arbitrary stale
* on a Lustre client, but client at least knows KMS. If accessed area is
* inside [0, KMS], set file size to KMS, otherwise glimpse file size.
*
* Locking: cl_isize_lock is used to serialize changes to inode size and to
* protect consistency between inode size and cl_object
* attributes. cl_object_size_lock() protects consistency between cl_attr's of
* top-object and sub-objects.
*/
static int vvp_prep_size(const struct lu_env *env, struct cl_object *obj,
struct cl_io *io, loff_t start, size_t count,
int *exceed)
{
struct cl_attr *attr = vvp_env_thread_attr(env);
struct inode *inode = vvp_object_inode(obj);
loff_t pos = start + count - 1;
loff_t kms;
int result;
/*
* Consistency guarantees: following possibilities exist for the
* relation between region being accessed and real file size at this
* moment:
*
* (A): the region is completely inside of the file;
*
* (B-x): x bytes of region are inside of the file, the rest is
* outside;
*
* (C): the region is completely outside of the file.
*
* This classification is stable under DLM lock already acquired by
* the caller, because to change the class, other client has to take
* DLM lock conflicting with our lock. Also, any updates to ->i_size
* by other threads on this client are serialized by
* ll_inode_size_lock(). This guarantees that short reads are handled
* correctly in the face of concurrent writes and truncates.
*/
vvp_object_size_lock(obj);
result = cl_object_attr_get(env, obj, attr);
if (result == 0) {
kms = attr->cat_kms;
if (pos > kms) {
/*
* A glimpse is necessary to determine whether we
* return a short read (B) or some zeroes at the end
* of the buffer (C)
*/
vvp_object_size_unlock(obj);
result = cl_glimpse_lock(env, io, inode, obj, 0);
if (result == 0 && exceed) {
/* If objective page index exceed end-of-file
* page index, return directly. Do not expect
* kernel will check such case correctly.
* linux-2.6.18-128.1.1 miss to do that.
* --bug 17336
*/
loff_t size = i_size_read(inode);
loff_t cur_index = start >> PAGE_SHIFT;
loff_t size_index = (size - 1) >> PAGE_SHIFT;
if ((size == 0 && cur_index != 0) ||
size_index < cur_index)
*exceed = 1;
}
/**
* Prepare partially written-to page for a write.
*/
static int vvp_io_prepare_partial(const struct lu_env *env, struct cl_io *io,
struct cl_object *obj, struct cl_page *pg,
struct ccc_page *cp,
unsigned from, unsigned to)
{
struct cl_attr *attr = ccc_env_thread_attr(env);
loff_t offset = cl_offset(obj, pg->cp_index);
int result;
cl_object_attr_lock(obj);
result = cl_object_attr_get(env, obj, attr);
cl_object_attr_unlock(obj);
if (result == 0) {
/*
* If are writing to a new page, no need to read old data.
* The extent locking will have updated the KMS, and for our
* purposes here we can treat it like i_size.
*/
if (attr->cat_kms <= offset) {
char *kaddr = ll_kmap_atomic(cp->cpg_page, KM_USER0);
memset(kaddr, 0, cl_page_size(obj));
ll_kunmap_atomic(kaddr, KM_USER0);
} else if (cp->cpg_defer_uptodate)
cp->cpg_ra_used = 1;
else
result = vvp_page_sync_io(env, io, pg, cp, CRT_READ);
/*
* In older implementations, obdo_refresh_inode is called here
* to update the inode because the write might modify the
* object info at OST. However, this has been proven useless,
* since LVB functions will be called when user space program
* tries to retrieve inode attribute. Also, see bug 15909 for
* details. -jay
*/
if (result == 0)
cl_page_export(env, pg, 1);
}
return result;
}
static int osc_io_write_start(const struct lu_env *env,
const struct cl_io_slice *slice)
{
struct osc_io *oio = cl2osc_io(env, slice);
struct cl_object *obj = slice->cis_obj;
struct cl_attr *attr = &osc_env_info(env)->oti_attr;
int result = 0;
ENTRY;
if (oio->oi_lockless == 0) {
OBD_FAIL_TIMEOUT(OBD_FAIL_OSC_DELAY_SETTIME, 1);
cl_object_attr_lock(obj);
result = cl_object_attr_get(env, obj, attr);
if (result == 0) {
attr->cat_mtime = attr->cat_ctime =
LTIME_S(CFS_CURRENT_TIME);
result = cl_object_attr_set(env, obj, attr,
CAT_MTIME | CAT_CTIME);
}
cl_object_attr_unlock(obj);
}
RETURN(result);
}
/* Retrieve object striping information.
*
* @lump is a pointer to an in-core struct with lmm_ost_count indicating
* the maximum number of OST indices which will fit in the user buffer.
* lmm_magic must be LOV_USER_MAGIC.
*
* If @size > 0, User specified limited buffer size, usually the buffer is from
* ll_lov_setstripe(), and the buffer can only hold basic layout template info.
*/
int lov_getstripe(const struct lu_env *env, struct lov_object *obj,
struct lov_stripe_md *lsm, struct lov_user_md __user *lump,
size_t size)
{
/* we use lov_user_md_v3 because it is larger than lov_user_md_v1 */
struct lov_mds_md *lmmk, *lmm;
struct lov_user_md_v1 lum;
size_t lmmk_size;
ssize_t lmm_size, lum_size = 0;
static bool printed;
int rc = 0;
ENTRY;
if (lsm->lsm_magic != LOV_MAGIC_V1 && lsm->lsm_magic != LOV_MAGIC_V3 &&
lsm->lsm_magic != LOV_MAGIC_COMP_V1) {
CERROR("bad LSM MAGIC: 0x%08X != 0x%08X nor 0x%08X\n",
lsm->lsm_magic, LOV_MAGIC_V1, LOV_MAGIC_V3);
GOTO(out, rc = -EIO);
}
if (!printed) {
LCONSOLE_WARN("%s: using old ioctl(LL_IOC_LOV_GETSTRIPE) on "
DFID", use llapi_layout_get_by_path()\n",
current->comm,
PFID(&obj->lo_cl.co_lu.lo_header->loh_fid));
printed = true;
}
lmmk_size = lov_comp_md_size(lsm);
OBD_ALLOC_LARGE(lmmk, lmmk_size);
if (lmmk == NULL)
GOTO(out, rc = -ENOMEM);
lmm_size = lov_lsm_pack(lsm, lmmk, lmmk_size);
if (lmm_size < 0)
GOTO(out_free, rc = lmm_size);
if (cpu_to_le32(LOV_MAGIC) != LOV_MAGIC) {
if (lmmk->lmm_magic == cpu_to_le32(LOV_MAGIC_V1) ||
lmmk->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) {
lustre_swab_lov_mds_md(lmmk);
lustre_swab_lov_user_md_objects(
(struct lov_user_ost_data *)lmmk->lmm_objects,
lmmk->lmm_stripe_count);
} else if (lmmk->lmm_magic == cpu_to_le32(LOV_MAGIC_COMP_V1)) {
lustre_swab_lov_comp_md_v1(
(struct lov_comp_md_v1 *)lmmk);
}
}
/* Legacy appication passes limited buffer, we need to figure out
* the user buffer size by the passed in lmm_stripe_count. */
if (copy_from_user(&lum, lump, sizeof(struct lov_user_md_v1)))
GOTO(out_free, rc = -EFAULT);
if (lum.lmm_magic == LOV_USER_MAGIC_V1 ||
lum.lmm_magic == LOV_USER_MAGIC_V3)
lum_size = lov_user_md_size(lum.lmm_stripe_count,
lum.lmm_magic);
if (lum_size != 0) {
struct lov_mds_md *comp_md = lmmk;
/* Legacy app (ADIO for instance) treats the layout as V1/V3
* blindly, we'd return a reasonable V1/V3 for them. */
if (lmmk->lmm_magic == LOV_MAGIC_COMP_V1) {
struct lov_comp_md_v1 *comp_v1;
struct cl_object *cl_obj;
struct cl_attr attr;
int i;
attr.cat_size = 0;
cl_obj = cl_object_top(&obj->lo_cl);
cl_object_attr_get(env, cl_obj, &attr);
/* return the last instantiated component if file size
* is non-zero, otherwise, return the last component.*/
comp_v1 = (struct lov_comp_md_v1 *)lmmk;
i = attr.cat_size == 0 ? comp_v1->lcm_entry_count : 0;
for (; i < comp_v1->lcm_entry_count; i++) {
if (!(comp_v1->lcm_entries[i].lcme_flags &
LCME_FL_INIT))
break;
}
if (i > 0)
i--;
comp_md = (struct lov_mds_md *)((char *)comp_v1 +
comp_v1->lcm_entries[i].lcme_offset);
}
lmm = comp_md;
lmm_size = lum_size;
} else {
lmm = lmmk;
lmm_size = lmmk_size;
}
/**
* User specified limited buffer size, usually the buffer is
* from ll_lov_setstripe(), and the buffer can only hold basic
* layout template info.
*/
if (size == 0 || size > lmm_size)
size = lmm_size;
if (copy_to_user(lump, lmm, size))
GOTO(out_free, rc = -EFAULT);
out_free:
OBD_FREE_LARGE(lmmk, lmmk_size);
out:
RETURN(rc);
}
static int osc_io_setattr_start(const struct lu_env *env,
const struct cl_io_slice *slice)
{
struct cl_io *io = slice->cis_io;
struct osc_io *oio = cl2osc_io(env, slice);
struct cl_object *obj = slice->cis_obj;
struct lov_oinfo *loi = cl2osc(obj)->oo_oinfo;
struct cl_attr *attr = &osc_env_info(env)->oti_attr;
struct obdo *oa = &oio->oi_oa;
struct osc_async_cbargs *cbargs = &oio->oi_cbarg;
__u64 size = io->u.ci_setattr.sa_attr.lvb_size;
unsigned int ia_valid = io->u.ci_setattr.sa_valid;
int result = 0;
struct obd_info oinfo = { { { 0 } } };
/* truncate cache dirty pages first */
if (cl_io_is_trunc(io))
result = osc_cache_truncate_start(env, oio, cl2osc(obj), size);
if (result == 0 && oio->oi_lockless == 0) {
cl_object_attr_lock(obj);
result = cl_object_attr_get(env, obj, attr);
if (result == 0) {
struct ost_lvb *lvb = &io->u.ci_setattr.sa_attr;
unsigned int cl_valid = 0;
if (ia_valid & ATTR_SIZE) {
attr->cat_size = attr->cat_kms = size;
cl_valid = (CAT_SIZE | CAT_KMS);
}
if (ia_valid & ATTR_MTIME_SET) {
attr->cat_mtime = lvb->lvb_mtime;
cl_valid |= CAT_MTIME;
}
if (ia_valid & ATTR_ATIME_SET) {
attr->cat_atime = lvb->lvb_atime;
cl_valid |= CAT_ATIME;
}
if (ia_valid & ATTR_CTIME_SET) {
attr->cat_ctime = lvb->lvb_ctime;
cl_valid |= CAT_CTIME;
}
result = cl_object_attr_set(env, obj, attr, cl_valid);
}
cl_object_attr_unlock(obj);
}
memset(oa, 0, sizeof(*oa));
if (result == 0) {
oa->o_oi = loi->loi_oi;
oa->o_mtime = attr->cat_mtime;
oa->o_atime = attr->cat_atime;
oa->o_ctime = attr->cat_ctime;
oa->o_valid = OBD_MD_FLID | OBD_MD_FLGROUP | OBD_MD_FLATIME |
OBD_MD_FLCTIME | OBD_MD_FLMTIME;
if (ia_valid & ATTR_SIZE) {
oa->o_size = size;
oa->o_blocks = OBD_OBJECT_EOF;
oa->o_valid |= OBD_MD_FLSIZE | OBD_MD_FLBLOCKS;
if (oio->oi_lockless) {
oa->o_flags = OBD_FL_SRVLOCK;
oa->o_valid |= OBD_MD_FLFLAGS;
}
} else {
LASSERT(oio->oi_lockless == 0);
}
oinfo.oi_oa = oa;
oinfo.oi_capa = io->u.ci_setattr.sa_capa;
init_completion(&cbargs->opc_sync);
if (ia_valid & ATTR_SIZE)
result = osc_punch_base(osc_export(cl2osc(obj)),
&oinfo, osc_async_upcall,
cbargs, PTLRPCD_SET);
else
result = osc_setattr_async_base(osc_export(cl2osc(obj)),
&oinfo, NULL,
osc_async_upcall,
cbargs, PTLRPCD_SET);
cbargs->opc_rpc_sent = result == 0;
}
return result;
}