linux/fs/nfs/read.c
Linus Torvalds 4f40c636b2 NFS Client Updates for Linux 6.11
New Features:
   * Add support for large folios
   * Implement rpcrdma generic device removal notification
   * Add client support for attribute delegations
   * Use a LAYOUTRETURN during reboot recovery to report layoutstats and errors
   * Improve throughput for random buffered writes
   * Add NVMe support to pnfs/blocklayout
 
 Bugfixes:
   * Fix rpcrdma_reqs_reset()
   * Avoid soft lockups when using UDP
   * Fix an nfs/blocklayout premature PR key unregestration
   * Another fix for EXCHGID4_FLAG_USE_PNFS_DS for DS server
   * Do not extend writes to the entire folio
   * Pass explicit offset and count values to tracepoints
   * Fix a race to wake up sleeping SUNRPC sync tasks
   * Fix gss_status tracepoint output
 
 Cleanups:
   * Add missing MODULE_DESCRIPTION() macros
   * Add blocklayout / SCSI layout tracepoints
   * Remove asm-generic headers from xprtrdma verbs.c
   * Remove unused 'struct mnt_fhstatus'
   * Other delegation related cleanups
   * Other folio related cleanups
   * Other pNFS related cleanups
   * Other xprtrdma cleanups
 -----BEGIN PGP SIGNATURE-----
 
 iQIzBAABCAAdFiEEnZ5MQTpR7cLU7KEp18tUv7ClQOsFAmaZgr0ACgkQ18tUv7Cl
 QOv8FxAAnUyYG7Kdbv+5Ko/SFv0imxCb5DQh2XC/hSHNrlKBlDnqe2PANXR9XocL
 mS0Wry5tZf/T+o+QoKv0HQUdWFlnqKzwclggrekf/lkioU1feWsLe2RzDl1iUh0V
 6fwcCyWXW1mYX2CtCaDe+/ZFcoZOMD+bItNHt/RdDScSnS9Jd8GSyocsVKsqaBx6
 3wub0FJ4UBgYNoX2T3YyK2JwvO9GLaKIQRJV74rjgPJKjcjhptbcb5MKBmOZrF95
 UCcpl4CwvD9RTsSEp0B98UbAFFpk8Nw1tmHF3GmyG/nsrJomDuLKFvbsiq23eHUf
 XeULZIbjMEzU56vjoTglZA4s7JYx17D0vzdPGUqU4mLN3LPm5LtGLBg2uQoPw/xW
 50euLU+ol36mfnQlBsuM/tAXgtoAcT63aNeNRNp8aOL47xA+PC6kWTBK9OaR5+x6
 w+d22Dpy+riMk1TRaAVt0ANcENKELsWRFvxkuWCpQhVoQ1h8LigQJzeggEEK7Sa6
 5u9H6wCTee2wz746uwA43koj1utuyrLq/5S+qEtCY1pbP3U0A+Gh0Xh00OXiYuzL
 TgRdksmiAL8cA51WjSrq6HhGLOUJAYLfbdKaVhW+fULxUVwzWhFFaFbbdiq/e4OR
 0pfqls8UZWICE51GeTfalEidpKZgV/LxU3QOuVoalWBULyj/TeI=
 =avTW
 -----END PGP SIGNATURE-----

Merge tag 'nfs-for-6.11-1' of git://git.linux-nfs.org/projects/anna/linux-nfs

Pull NFS client updates from Anna Schumaker:
 "New Features:
   - Add support for large folios
   - Implement rpcrdma generic device removal notification
   - Add client support for attribute delegations
   - Use a LAYOUTRETURN during reboot recovery to report layoutstats
     and errors
   - Improve throughput for random buffered writes
   - Add NVMe support to pnfs/blocklayout

  Bugfixes:
   - Fix rpcrdma_reqs_reset()
   - Avoid soft lockups when using UDP
   - Fix an nfs/blocklayout premature PR key unregestration
   - Another fix for EXCHGID4_FLAG_USE_PNFS_DS for DS server
   - Do not extend writes to the entire folio
   - Pass explicit offset and count values to tracepoints
   - Fix a race to wake up sleeping SUNRPC sync tasks
   - Fix gss_status tracepoint output

  Cleanups:
   - Add missing MODULE_DESCRIPTION() macros
   - Add blocklayout / SCSI layout tracepoints
   - Remove asm-generic headers from xprtrdma verbs.c
   - Remove unused 'struct mnt_fhstatus'
   - Other delegation related cleanups
   - Other folio related cleanups
   - Other pNFS related cleanups
   - Other xprtrdma cleanups"

* tag 'nfs-for-6.11-1' of git://git.linux-nfs.org/projects/anna/linux-nfs: (63 commits)
  SUNRPC: Fixup gss_status tracepoint error output
  SUNRPC: Fix a race to wake a sync task
  nfs: split nfs_read_folio
  nfs: pass explicit offset/count to trace events
  nfs: do not extend writes to the entire folio
  nfs/blocklayout: add support for NVMe
  nfs: remove nfs_page_length
  nfs: remove the unused max_deviceinfo_size field from struct pnfs_layoutdriver_type
  nfs: don't reuse partially completed requests in nfs_lock_and_join_requests
  nfs: move nfs_wait_on_request to write.c
  nfs: fold nfs_page_group_lock_subrequests into nfs_lock_and_join_requests
  nfs: fold nfs_folio_find_and_lock_request into nfs_lock_and_join_requests
  nfs: simplify nfs_folio_find_and_lock_request
  nfs: remove nfs_folio_private_request
  nfs: remove dead code for the old swap over NFS implementation
  NFSv4.1 another fix for EXCHGID4_FLAG_USE_PNFS_DS for DS server
  nfs: Block on write congestion
  nfs: Properly initialize server->writeback
  nfs: Drop pointless check from nfs_commit_release_pages()
  nfs/blocklayout: SCSI layout trace points for reservation key reg/unreg
  ...
2024-07-18 17:17:30 -07:00

477 lines
12 KiB
C

// SPDX-License-Identifier: GPL-2.0-only
/*
* linux/fs/nfs/read.c
*
* Block I/O for NFS
*
* Partial copy of Linus' read cache modifications to fs/nfs/file.c
* modified for async RPC by okir@monad.swb.de
*/
#include <linux/time.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/fcntl.h>
#include <linux/stat.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/task_io_accounting_ops.h>
#include <linux/pagemap.h>
#include <linux/sunrpc/clnt.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
#include <linux/module.h>
#include "nfs4_fs.h"
#include "internal.h"
#include "iostat.h"
#include "fscache.h"
#include "pnfs.h"
#include "nfstrace.h"
#include "delegation.h"
#define NFSDBG_FACILITY NFSDBG_PAGECACHE
const struct nfs_pgio_completion_ops nfs_async_read_completion_ops;
static const struct nfs_rw_ops nfs_rw_read_ops;
static struct kmem_cache *nfs_rdata_cachep;
static struct nfs_pgio_header *nfs_readhdr_alloc(void)
{
struct nfs_pgio_header *p = kmem_cache_zalloc(nfs_rdata_cachep, GFP_KERNEL);
if (p)
p->rw_mode = FMODE_READ;
return p;
}
static void nfs_readhdr_free(struct nfs_pgio_header *rhdr)
{
if (rhdr->res.scratch != NULL)
kfree(rhdr->res.scratch);
kmem_cache_free(nfs_rdata_cachep, rhdr);
}
static int nfs_return_empty_folio(struct folio *folio)
{
folio_zero_segment(folio, 0, folio_size(folio));
folio_mark_uptodate(folio);
folio_unlock(folio);
return 0;
}
void nfs_pageio_init_read(struct nfs_pageio_descriptor *pgio,
struct inode *inode, bool force_mds,
const struct nfs_pgio_completion_ops *compl_ops)
{
struct nfs_server *server = NFS_SERVER(inode);
const struct nfs_pageio_ops *pg_ops = &nfs_pgio_rw_ops;
#ifdef CONFIG_NFS_V4_1
if (server->pnfs_curr_ld && !force_mds)
pg_ops = server->pnfs_curr_ld->pg_read_ops;
#endif
nfs_pageio_init(pgio, inode, pg_ops, compl_ops, &nfs_rw_read_ops,
server->rsize, 0);
}
EXPORT_SYMBOL_GPL(nfs_pageio_init_read);
void nfs_pageio_complete_read(struct nfs_pageio_descriptor *pgio)
{
struct nfs_pgio_mirror *pgm;
unsigned long npages;
nfs_pageio_complete(pgio);
/* It doesn't make sense to do mirrored reads! */
WARN_ON_ONCE(pgio->pg_mirror_count != 1);
pgm = &pgio->pg_mirrors[0];
NFS_I(pgio->pg_inode)->read_io += pgm->pg_bytes_written;
npages = (pgm->pg_bytes_written + PAGE_SIZE - 1) >> PAGE_SHIFT;
nfs_add_stats(pgio->pg_inode, NFSIOS_READPAGES, npages);
}
void nfs_pageio_reset_read_mds(struct nfs_pageio_descriptor *pgio)
{
struct nfs_pgio_mirror *mirror;
if (pgio->pg_ops && pgio->pg_ops->pg_cleanup)
pgio->pg_ops->pg_cleanup(pgio);
pgio->pg_ops = &nfs_pgio_rw_ops;
/* read path should never have more than one mirror */
WARN_ON_ONCE(pgio->pg_mirror_count != 1);
mirror = &pgio->pg_mirrors[0];
mirror->pg_bsize = NFS_SERVER(pgio->pg_inode)->rsize;
}
EXPORT_SYMBOL_GPL(nfs_pageio_reset_read_mds);
bool nfs_read_alloc_scratch(struct nfs_pgio_header *hdr, size_t size)
{
WARN_ON(hdr->res.scratch != NULL);
hdr->res.scratch = kmalloc(size, GFP_KERNEL);
return hdr->res.scratch != NULL;
}
EXPORT_SYMBOL_GPL(nfs_read_alloc_scratch);
static void nfs_readpage_release(struct nfs_page *req, int error)
{
struct folio *folio = nfs_page_to_folio(req);
if (nfs_page_group_sync_on_bit(req, PG_UNLOCKPAGE))
if (nfs_netfs_folio_unlock(folio))
folio_unlock(folio);
nfs_release_request(req);
}
static void nfs_page_group_set_uptodate(struct nfs_page *req)
{
if (nfs_page_group_sync_on_bit(req, PG_UPTODATE))
folio_mark_uptodate(nfs_page_to_folio(req));
}
static void nfs_read_completion(struct nfs_pgio_header *hdr)
{
unsigned long bytes = 0;
int error;
if (test_bit(NFS_IOHDR_REDO, &hdr->flags))
goto out;
while (!list_empty(&hdr->pages)) {
struct nfs_page *req = nfs_list_entry(hdr->pages.next);
struct folio *folio = nfs_page_to_folio(req);
unsigned long start = req->wb_pgbase;
unsigned long end = req->wb_pgbase + req->wb_bytes;
if (test_bit(NFS_IOHDR_EOF, &hdr->flags)) {
/* note: regions of the page not covered by a
* request are zeroed in nfs_read_add_folio
*/
if (bytes > hdr->good_bytes) {
/* nothing in this request was good, so zero
* the full extent of the request */
folio_zero_segment(folio, start, end);
} else if (hdr->good_bytes - bytes < req->wb_bytes) {
/* part of this request has good bytes, but
* not all. zero the bad bytes */
start += hdr->good_bytes - bytes;
WARN_ON(start < req->wb_pgbase);
folio_zero_segment(folio, start, end);
}
}
error = 0;
bytes += req->wb_bytes;
if (test_bit(NFS_IOHDR_ERROR, &hdr->flags)) {
if (bytes <= hdr->good_bytes)
nfs_page_group_set_uptodate(req);
else {
error = hdr->error;
xchg(&nfs_req_openctx(req)->error, error);
}
} else
nfs_page_group_set_uptodate(req);
nfs_list_remove_request(req);
nfs_readpage_release(req, error);
}
nfs_netfs_read_completion(hdr);
out:
hdr->release(hdr);
}
static void nfs_initiate_read(struct nfs_pgio_header *hdr,
struct rpc_message *msg,
const struct nfs_rpc_ops *rpc_ops,
struct rpc_task_setup *task_setup_data, int how)
{
rpc_ops->read_setup(hdr, msg);
nfs_netfs_initiate_read(hdr);
trace_nfs_initiate_read(hdr);
}
static void
nfs_async_read_error(struct list_head *head, int error)
{
struct nfs_page *req;
while (!list_empty(head)) {
req = nfs_list_entry(head->next);
nfs_list_remove_request(req);
nfs_readpage_release(req, error);
}
}
const struct nfs_pgio_completion_ops nfs_async_read_completion_ops = {
.error_cleanup = nfs_async_read_error,
.completion = nfs_read_completion,
};
/*
* This is the callback from RPC telling us whether a reply was
* received or some error occurred (timeout or socket shutdown).
*/
static int nfs_readpage_done(struct rpc_task *task,
struct nfs_pgio_header *hdr,
struct inode *inode)
{
int status = NFS_PROTO(inode)->read_done(task, hdr);
if (status != 0)
return status;
nfs_add_stats(inode, NFSIOS_SERVERREADBYTES, hdr->res.count);
trace_nfs_readpage_done(task, hdr);
if (task->tk_status == -ESTALE) {
nfs_set_inode_stale(inode);
nfs_mark_for_revalidate(inode);
}
return 0;
}
static void nfs_readpage_retry(struct rpc_task *task,
struct nfs_pgio_header *hdr)
{
struct nfs_pgio_args *argp = &hdr->args;
struct nfs_pgio_res *resp = &hdr->res;
/* This is a short read! */
nfs_inc_stats(hdr->inode, NFSIOS_SHORTREAD);
trace_nfs_readpage_short(task, hdr);
/* Has the server at least made some progress? */
if (resp->count == 0) {
nfs_set_pgio_error(hdr, -EIO, argp->offset);
return;
}
/* For non rpc-based layout drivers, retry-through-MDS */
if (!task->tk_ops) {
hdr->pnfs_error = -EAGAIN;
return;
}
/* Yes, so retry the read at the end of the hdr */
hdr->mds_offset += resp->count;
argp->offset += resp->count;
argp->pgbase += resp->count;
argp->count -= resp->count;
resp->count = 0;
resp->eof = 0;
rpc_restart_call_prepare(task);
}
static void nfs_readpage_result(struct rpc_task *task,
struct nfs_pgio_header *hdr)
{
if (hdr->res.eof) {
loff_t pos = hdr->args.offset + hdr->res.count;
unsigned int new = pos - hdr->io_start;
if (hdr->good_bytes > new) {
hdr->good_bytes = new;
set_bit(NFS_IOHDR_EOF, &hdr->flags);
clear_bit(NFS_IOHDR_ERROR, &hdr->flags);
}
} else if (hdr->res.count < hdr->args.count)
nfs_readpage_retry(task, hdr);
}
int nfs_read_add_folio(struct nfs_pageio_descriptor *pgio,
struct nfs_open_context *ctx,
struct folio *folio)
{
struct inode *inode = folio->mapping->host;
struct nfs_server *server = NFS_SERVER(inode);
size_t fsize = folio_size(folio);
unsigned int rsize = server->rsize;
struct nfs_page *new;
unsigned int len, aligned_len;
int error;
len = nfs_folio_length(folio);
if (len == 0)
return nfs_return_empty_folio(folio);
aligned_len = min_t(unsigned int, ALIGN(len, rsize), fsize);
new = nfs_page_create_from_folio(ctx, folio, 0, aligned_len);
if (IS_ERR(new)) {
error = PTR_ERR(new);
if (nfs_netfs_folio_unlock(folio))
folio_unlock(folio);
goto out;
}
if (len < fsize)
folio_zero_segment(folio, len, fsize);
if (!nfs_pageio_add_request(pgio, new)) {
nfs_list_remove_request(new);
error = pgio->pg_error;
nfs_readpage_release(new, error);
goto out;
}
return 0;
out:
return error;
}
/*
* Actually read a folio over the wire.
*/
static int nfs_do_read_folio(struct file *file, struct folio *folio)
{
struct inode *inode = file_inode(file);
struct nfs_pageio_descriptor pgio;
struct nfs_open_context *ctx;
int ret;
ctx = get_nfs_open_context(nfs_file_open_context(file));
xchg(&ctx->error, 0);
nfs_pageio_init_read(&pgio, inode, false,
&nfs_async_read_completion_ops);
ret = nfs_read_add_folio(&pgio, ctx, folio);
if (ret)
goto out_put;
nfs_pageio_complete_read(&pgio);
nfs_update_delegated_atime(inode);
if (pgio.pg_error < 0) {
ret = pgio.pg_error;
goto out_put;
}
ret = folio_wait_locked_killable(folio);
if (!folio_test_uptodate(folio) && !ret)
ret = xchg(&ctx->error, 0);
out_put:
put_nfs_open_context(ctx);
return ret;
}
/*
* Synchronously read a folio.
*
* This is not heavily used as most users to try an asynchronous
* large read through ->readahead first.
*/
int nfs_read_folio(struct file *file, struct folio *folio)
{
struct inode *inode = file_inode(file);
loff_t pos = folio_pos(folio);
size_t len = folio_size(folio);
int ret;
trace_nfs_aop_readpage(inode, pos, len);
nfs_inc_stats(inode, NFSIOS_VFSREADPAGE);
task_io_account_read(len);
/*
* Try to flush any pending writes to the file..
*
* NOTE! Because we own the folio lock, there cannot
* be any new pending writes generated at this point
* for this folio (other folios can be written to).
*/
ret = nfs_wb_folio(inode, folio);
if (ret)
goto out_unlock;
if (folio_test_uptodate(folio))
goto out_unlock;
ret = -ESTALE;
if (NFS_STALE(inode))
goto out_unlock;
ret = nfs_netfs_read_folio(file, folio);
if (ret)
ret = nfs_do_read_folio(file, folio);
out:
trace_nfs_aop_readpage_done(inode, pos, len, ret);
return ret;
out_unlock:
folio_unlock(folio);
goto out;
}
void nfs_readahead(struct readahead_control *ractl)
{
struct nfs_pageio_descriptor pgio;
struct nfs_open_context *ctx;
unsigned int nr_pages = readahead_count(ractl);
struct file *file = ractl->file;
struct inode *inode = ractl->mapping->host;
struct folio *folio;
int ret;
trace_nfs_aop_readahead(inode, readahead_pos(ractl), nr_pages);
nfs_inc_stats(inode, NFSIOS_VFSREADPAGES);
task_io_account_read(readahead_length(ractl));
ret = -ESTALE;
if (NFS_STALE(inode))
goto out;
ret = nfs_netfs_readahead(ractl);
if (!ret)
goto out;
if (file == NULL) {
ret = -EBADF;
ctx = nfs_find_open_context(inode, NULL, FMODE_READ);
if (ctx == NULL)
goto out;
} else
ctx = get_nfs_open_context(nfs_file_open_context(file));
nfs_pageio_init_read(&pgio, inode, false,
&nfs_async_read_completion_ops);
while ((folio = readahead_folio(ractl)) != NULL) {
ret = nfs_read_add_folio(&pgio, ctx, folio);
if (ret)
break;
}
nfs_pageio_complete_read(&pgio);
nfs_update_delegated_atime(inode);
put_nfs_open_context(ctx);
out:
trace_nfs_aop_readahead_done(inode, nr_pages, ret);
}
int __init nfs_init_readpagecache(void)
{
nfs_rdata_cachep = kmem_cache_create("nfs_read_data",
sizeof(struct nfs_pgio_header),
0, SLAB_HWCACHE_ALIGN,
NULL);
if (nfs_rdata_cachep == NULL)
return -ENOMEM;
return 0;
}
void nfs_destroy_readpagecache(void)
{
kmem_cache_destroy(nfs_rdata_cachep);
}
static const struct nfs_rw_ops nfs_rw_read_ops = {
.rw_alloc_header = nfs_readhdr_alloc,
.rw_free_header = nfs_readhdr_free,
.rw_done = nfs_readpage_done,
.rw_result = nfs_readpage_result,
.rw_initiate = nfs_initiate_read,
};