iov_iter: Improve page extraction (pin or just list)
Subsystems (7)
v12 cover iov_iter: Improve page extraction (pin or just list)
Hi Jens, Al, Christoph,
Here are patches to provide support for extracting pages from an iov_iter and to use this in the extraction functions in the block layer bio code.
The patches make the following changes:
(1) Change generic_file_splice_read() to load up an ITER_BVEC iterator with sufficient pages and use that rather than using an ITER_PIPE. This avoids a problem[2] when __iomap_dio_rw() calls iov_iter_revert() to shorten an iterator when it races with truncation. The reversion causes the pipe iterator to prematurely release the pages it was retaining - despite the read still being in progress. This caused memory corruption.
(2) Remove ITER_PIPE and its paraphernalia as generic_file_splice_read() was the only user.
(3) Add a function, iov_iter_extract_pages() to replace iov_iter_get_pages*() that gets refs, pins or just lists the pages as appropriate to the iterator type.
Add a function, iov_iter_extract_will_pin() that will indicate from the iterator type how the cleanup is to be performed, returning true if the pages will need unpinning, false otherwise.
(4) Make the bio struct carry a pair of flags to indicate the cleanup mode. BIO_NO_PAGE_REF is replaced with BIO_PAGE_REFFED (indicating FOLL_GET was used) and BIO_PAGE_PINNED (indicating FOLL_PIN was used) is added.
BIO_PAGE_REFFED will go away, but at the moment fs/direct-io.c sets it and this series does not fully address that file.
(5) Add a function, bio_release_page(), to release a page appropriately to the cleanup mode indicated by the BIO_PAGE_* flags.
(6) Make the iter-to-bio code use iov_iter_extract_pages() to retain the pages appropriately and clean them up later.
(7) Fix bio_flagged() so that it doesn't prevent a gcc optimisation.
I've pushed the patches here also:
https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/log/?h=iov-extract
and tagged it with
iov-extract-20230207
David
Changes:
========
ver #12)
- Added the missing __bitwise on the iov_iter_extraction_t typedef.
- Rebased on -rc7.
- Don't specify FOLL_PIN to pin_user_pages_fast().
- Inserted patch at front to fix race between DIO read and truncation that caused memory corruption when iov_iter_revert() got called on an ITER_PIPE iterator[2].
- Inserted a patch after that to remove the now-unused ITER_PIPE and its helper functions.
- Removed the ITER_PIPE bits from iov_iter_extract_pages().
ver #11)
- Fix iov_iter_extract_kvec_pages() to include the offset into the page in the returned starting offset.
- Use __bitwise for the extraction flags
ver #10)
- Fix use of i->kvec in iov_iter_extract_bvec_pages() to be i->bvec.
- Drop bio_set_cleanup_mode(), open coding it instead.
ver #9)
- It's now not permitted to use FOLL_PIN outside of mm/, so:
- Change iov_iter_extract_mode() into iov_iter_extract_will_pin() and return true/false instead of FOLL_PIN/0.
- Drop of folio_put_unpin() and page_put_unpin() and instead call unpin_user_page() (and put_page()) directly as necessary.
- Make __bio_release_pages() call bio_release_page() instead of unpin_user_page() as there's no BIO_* -> FOLL_* translation to do.
- Drop the FOLL_* renumbering patch.
- Change extract_flags to extraction_flags.
ver #8)
- Import Christoph Hellwig's changes.
- Split the conversion-to-extraction patch.
- Drop the extract_flags arg from iov_iter_extract_mode().
- Don't default bios to BIO_PAGE_REFFED, but set explicitly.
- Switch FOLL_PIN and FOLL_GET when renumbering so PIN is at bit 0.
- Switch BIO_PAGE_PINNED and BIO_PAGE_REFFED so PINNED is at bit 0.
- We should always be using FOLL_PIN (not FOLL_GET) for DIO, so adjust the patches for that.
ver #7)
- For now, drop the parts to pass the I/O direction to iov_iter_*pages*() as it turned out to be a lot more complicated, with places not setting IOCB_WRITE when they should, for example.
- Drop all the patches that changed things other then the block layer's bio handling. The netfslib and cifs changes can go into a separate patchset.
- Add support for extracting pages from KVEC-type iterators.
- When extracting from BVEC/KVEC, skip over empty vecs at the front.
ver #6)
- Fix write() syscall and co. not setting IOCB_WRITE.
- Added iocb_is_read() and iocb_is_write() to check IOCB_WRITE.
- Use op_is_write() in bio_copy_user_iov().
- Drop the iterator direction checks from smbd_recv().
- Define FOLL_SOURCE_BUF and FOLL_DEST_BUF and pass them in as part of gup_flags to iov_iter_get/extract_pages*().
- Replace iov_iter_get_pages*2() with iov_iter_get_pages*() and remove.
- Add back the function to indicate the cleanup mode.
- Drop the cleanup_mode return arg to iov_iter_extract_pages().
- Provide a helper to clean up a page.
- Renumbered FOLL_GET and FOLL_PIN and made BIO_PAGE_REFFED/PINNED have the same numerical values, enforced with an assertion.
- Converted AF_ALG, SCSI vhost, generic DIO, FUSE, splice to pipe, 9P and NFS.
- Added in the patches to make CIFS do top-to-bottom iterators and use various of the added extraction functions.
- Added a pair of work-in-progess patches to make sk_buff fragments store FOLL_GET and FOLL_PIN.
ver #5)
- Replace BIO_NO_PAGE_REF with BIO_PAGE_REFFED and split into own patch.
- Transcribe FOLL_GET/PIN into BIO_PAGE_REFFED/PINNED flags.
- Add patch to allow bio_flagged() to be combined by gcc.
ver #4)
- Drop the patch to move the FOLL_* flags to linux/mm_types.h as they're no longer referenced by linux/uio.h.
- Add ITER_SOURCE/DEST cleanup patches.
- Make iov_iter/netfslib iter extraction patches use ITER_SOURCE/DEST.
- Allow additional gup_flags to be passed into iov_iter_extract_pages().
- Add struct bio patch.
ver #3)
- Switch to using EXPORT_SYMBOL_GPL to prevent indirect 3rd-party access to get/pin_user_pages_fast()[1].
ver #2)
- Rolled the extraction cleanup mode query function into the extraction function, returning the indication through the argument list.
- Fixed patch 4 (extract to scatterlist) to actually use the new extraction API.
Christoph Hellwig (1):
block: Replace BIO_NO_PAGE_REF with BIO_PAGE_REFFED with inverted
logic
David Howells (9):
vfs, iomap: Fix generic_file_splice_read() to avoid reversion of
ITER_PIPE
iov_iter: Kill ITER_PIPE
iov_iter: Define flags to qualify page extraction.
iov_iter: Add a function to extract a page list from an iterator
iomap: Don't get an reference on ZERO_PAGE for direct I/O block
zeroing
block: Fix bio_flagged() so that gcc can better optimise it
block: Add BIO_PAGE_PINNED and associated infrastructure
block: Convert bio_iov_iter_get_pages to use iov_iter_extract_pages
block: convert bio_map_user_iov to use iov_iter_extract_pages
block/bio.c | 33 +- block/blk-map.c | 26 +- block/blk.h | 12 + fs/cifs/file.c | 8 +- fs/direct-io.c | 2 + fs/iomap/direct-io.c | 1 - fs/splice.c | 76 +++- include/linux/bio.h | 5 +- include/linux/blk_types.h | 3 +- include/linux/uio.h | 49 ++- lib/iov_iter.c | 713 +++++++++++++++----------------------- mm/filemap.c | 3 - 12 files changed, 426 insertions(+), 505 deletions(-)
PATCH v12 1/10 vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
With the new iov_iter_extract_pages() function (added in a later patch), pages extracted from a non-user-backed iterator, such as ITER_PIPE, aren't pinned. __iomap_dio_rw(), however, calls iov_iter_revert() to shorten the iterator to just the data it is going to use - which causes the pipe buffers to be freed, even though they're attached to a bio and may get written to by DMA (thanks to Hillf Danton for spotting this[1]).
This then causes massive memory corruption that is particularly noticable when the syzbot test[2] is run. The test boils down to:
out = creat(argv[1], 0666); ftruncate(out, 0x800); lseek(out, 0x200, SEEK_SET); in = open(argv[1], O_RDONLY | O_DIRECT | O_NOFOLLOW); sendfile(out, in, NULL, 0x1dd00);
run repeatedly in parallel. What I think is happening is that ftruncate() occasionally shortens the DIO read that's about to be made by sendfile's splice core by reducing i_size.
Fix this by replacing the use of an ITER_PIPE iterator with an ITER_BVEC iterator for which reversion won't free the buffers. Bulk allocate all the buffers we think we're going to use in advance, do the read synchronously and only then trim the buffer down. The pages we did use get pushed into the pipe.
This is more efficient by virtue of doing a bulk page allocation, but slightly less efficient by ignoring any partial page in the pipe.
Note that this removes the only user of ITER_PIPE.
--- fs/splice.c | 76 +++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 68 insertions(+), 8 deletions(-)
fs/splice.c +68 -8
@@ -295,24 +295,62 @@ void splice_shrink_spd(struct splice_pipe_desc *spd)* used as long as it has more or less sane ->read_iter(). * */ -ssize_t generic_file_splice_read (struct file *in,loff_t *ppos, +ssize_t generic_file_splice_read (struct file *file,loff_t *ppos,struct pipe_inode_info *pipe,size_t len,unsigned int flags) { +LIST_HEAD (pages);struct iov_iter to; +struct bio_vec *bv;struct kiocb kiocb; -int ret; +struct page *page; +unsigned int head; +ssize_t ret; +size_t used, npages, chunk, remain, reclaim; +int i; + +/* Work out how much data we can actually add into the pipe */ + used =pipe_occupancy (pipe->head, pipe->tail); + npages =max_t (ssize_t, pipe->max_usage - used,0 ); + len =min_t (size_t, len, npages *PAGE_SIZE ); + npages =DIV_ROUND_UP (len,PAGE_SIZE ); + + bv =kmalloc (array_size (npages,sizeof (bv[0 ])),GFP_KERNEL ); +if (!bv) +return -ENOMEM ; + + npages =alloc_pages_bulk_list (GFP_USER , npages, &pages); +if (!npages) { +kfree (bv); +return -ENOMEM ; + } -iov_iter_pipe (&to,ITER_DEST , pipe, len); -init_sync_kiocb (&kiocb, in); + remain = len =min_t (size_t, len, npages *PAGE_SIZE ); + +for (i =0 ; i < npages; i++) { + chunk =min_t (size_t,PAGE_SIZE , remain); + page = list_first_entry(&pages, struct page, lru); +list_del_init (&page->lru); + bv[i].bv_page = page; + bv[i].bv_offset =0 ; + bv[i].bv_len = chunk; + remain -= chunk; + } + +/* Do the I/O */ +iov_iter_bvec (&to,ITER_DEST , bv, npages, len); +init_sync_kiocb (&kiocb, file); kiocb.ki_pos = *ppos; - ret =call_read_iter (in, &kiocb, &to); + ret =call_read_iter (file, &kiocb, &to); + + reclaim = npages *PAGE_SIZE ; + remain =0 ;if (ret >0 ) { + reclaim -= ret; + remain = ret; *ppos = kiocb.ki_pos; -file_accessed (in); +file_accessed (file); }else if (ret <0 ) { -/* free what was emitted */ -pipe_discard_from (pipe, to.start_head);/* * callers of ->splice_read() expect -EAGAIN on * "can't put anything in there", rather than -EFAULT.
@@ -321,6 +359,28 @@ ssize_t generic_file_splice_read(struct file *in, loff_t *ppos, ret = -EAGAIN ; } +/* Free any pages that didn't get touched at all. */ +for (; reclaim >=PAGE_SIZE ; reclaim -=PAGE_SIZE ) +__free_page (bv[--npages].bv_page); + +/* Push the remaining pages into the pipe. */ + head = pipe->head; +for (i =0 ; i < npages; i++) { +struct pipe_buffer *buf = &pipe->bufs[head & (pipe->ring_size -1 )]; + + chunk =min_t (size_t, remain,PAGE_SIZE ); + *buf = (struct pipe_buffer ) { + .ops = &default_pipe_buf_ops, + .page = bv[i].bv_page, + .offset =0 , + .len = chunk, + }; + head++; + remain -= chunk; + } + pipe->head = head; + +kfree (bv);return ret; }EXPORT_SYMBOL (generic_file_splice_read);
Re: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
Subject nitpick: this does not ouch iomap at all.
Quoted 8 lines
Fix this by replacing the use of an
ITER_PIPEiterator with anITER_BVECiterator for which reversion won't free the buffers. Bulk allocate all the buffers we think we're going to use in advance, do the read synchronously and only then trim the buffer down. The pages we did use get pushed into the pipe.This is more efficient by virtue of doing a bulk page allocation, but slightly less efficient by ignoring any partial page in the pipe.
For the usual case of a buffered read into the iter, this completely changes the semantics:
- before the pagecache pages just grew a new reference and were added to the pipe buffer, and I/O was done without an extra copy
- with this patch you always allocate an extra set of pages for the pipe and copy to it
So I very much doubt this makes anything more efficient, and I don't think we can just do it.
We'll have to fix reverting of pipe buffers, just as I already pointed out in your cifs series that tries to play the same game.
Re: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
Quoted 2 lines · Christoph Hellwig
We'll have to fix reverting of pipe buffers, just as I already pointed out in your cifs series that tries to play the same game.
Fixing ITER_PIPE reversion isn't very straightforward, unfortunately. It's possible for a partial direct I/O read to use reversion to trim the iterator (thereby discarding anon pages from a pipe) and then fall back to reading the rest by buffered I/O. I suppose I could set a flag on the pipe_buffers indicating whether iov_iter_revert() is allowed to free them (if they were spliced in from the pagecache, then they must be freed by reverting them).
How about one of two different solutions?
(1) Repurpose the function I proposed for generic_file_splice_read() but only for splicing from O_DIRECT files; reading from non-O_DIRECT files would use an ITER_PIPE as upstream.
(2) Repurpose the function I proposed for generic_file_splice_read() but only for splicing from O_DIRECT files, as (1), but also replace the splice from a buffered file with something like the patch below. This uses filemap_get_pages() to do the reading and to get a bunch of folios from the pagecache that we can then splice into the pipe directly.
David
---
splice: Do splice read from a buffered file without using ITER_PIPE
Provide a function to do splice read from a buffered file, pulling the folios out of the pagecache directly by calling filemap_get_pages() to do any required reading and then pasting the returned folios into the pipe.
A helper function is provided to do the actual folio pasting and will handle multipage folios by splicing as many of the relevant subpages as will fit into the pipe.
The ITER_BVEC-based splicing previously added is then only used for splicing from O_DIRECT files.
The code is loosely based on filemap_read() and might belong in mm/filemap.c with that as it needs to use filemap_get_pages().
filemap_get_pages() and some of the functions it uses are changed to take the required byte count rather than an iterator (which was only being used for iter->count). This should be split into a separate file.
--- fs/splice.c | 177 +++++++++++++++++++++++++++++++++++++++++++----- include/linux/pagemap.h | 2 mm/filemap.c | 23 +++--- 3 files changed, 176 insertions(+), 26 deletions(-)
fs/splice.c +162 -15
@@ -22,6 +22,7 @@#include <linux/fs.h> #include <linux/file.h> #include <linux/pagemap.h> +#include <linux/pagevec.h> #include <linux/splice.h> #include <linux/memcontrol.h> #include <linux/mm_inline.h>
@@ -282,22 +283,13 @@ void splice_shrink_spd(struct splice_pipe_desc *spd)kfree (spd->partial); } -/** -* generic_file_splice_read - splice data from file to a pipe -* @in: file to splice from -* @ppos: position in @in -* @pipe: pipe to splice to -* @len: number of bytes to splice -* @flags: splice modifier flags -* -* Description: -* Will read pages from given file and fill them into a pipe. Can be -* used as long as it has more or less sane ->read_iter(). -* +/* +* Splice data from an O_DIRECT file into pages and then add them to the output +* pipe. */ -ssize_t generic_file_splice_read (struct file *in,loff_t *ppos, -struct pipe_inode_info *pipe,size_t len, -unsigned int flags) +static ssize_t generic_file_direct_splice_read (struct file *in,loff_t *ppos, +struct pipe_inode_info *pipe, +size_t len,unsigned int flags) {LIST_HEAD (pages);struct iov_iter to;
@@ -383,6 +375,161 @@ ssize_t generic_file_splice_read(struct file *in, loff_t *ppos,kfree (bv);return ret; } + +/* +* Splice subpages from a folio into a pipe. +*/ +static size_t splice_folio_into_pipe (struct pipe_inode_info *pipe, +struct folio *folio, +loff_t fpos,size_t size) +{ +struct page *page; +size_t spliced =0 , offset =offset_in_folio (folio, fpos); + + page =folio_page (folio, offset /PAGE_SIZE ); + size =min (size,folio_size (folio) - offset); + offset %=PAGE_SIZE ; + +while (spliced < size && + !pipe_full (pipe->head, pipe->tail, pipe->max_usage)) { +struct pipe_buffer *buf = &pipe->bufs[pipe->head & (pipe->ring_size -1 )]; +size_t part =min_t (size_t,PAGE_SIZE - offset, size - spliced); + + *buf = (struct pipe_buffer ) { + .ops = &page_cache_pipe_buf_ops, + .page = page, + .offset = offset, + .len = part, + }; +folio_get (folio); + pipe->head++; + page++; + spliced += part; + offset =0 ; + } + +return spliced; +} + +/* +* Splice folios from the pagecache of a buffered (ie. non-O_DIRECT) file into +* a pipe. +*/ +static ssize_t generic_file_buffered_splice_read (struct file *in,loff_t *ppos, +struct pipe_inode_info *pipe, +size_t len, +unsigned int flags) +{ +struct folio_batch fbatch; +size_t total_spliced =0 , used, npages; +loff_t isize, end_offset; +bool writably_mapped; +int i, error =0 ; + +struct kiocb iocb = { + .ki_filp = in, + .ki_pos = *ppos, + }; + +/* Work out how much data we can actually add into the pipe */ + used =pipe_occupancy (pipe->head, pipe->tail); + npages =max_t (ssize_t, pipe->max_usage - used,0 ); + len =min_t (size_t, len, npages *PAGE_SIZE ); + +folio_batch_init (&fbatch); + +do { +cond_resched (); + +if (*ppos >=i_size_read (file_inode (in))) +break ; + + iocb.ki_pos = *ppos; + error =filemap_get_pages (&iocb, len, &fbatch); +if (error <0 ) +break ; + +/* +* i_size must be checked after we know the pages are Uptodate. +* +* Checking i_size after the check allows us to calculate +* the correct value for "nr", which means the zero-filled +* part of the page is not copied back to userspace (unless +* another truncate extends the file - this is desired though). +*/ + isize =i_size_read (file_inode (in)); +if (unlikely (*ppos >= isize)) +break ; + end_offset =min_t (loff_t, isize, *ppos + len); + +/* +* Once we start copying data, we don't want to be touching any +* cachelines that might be contended: +*/ + writably_mapped =mapping_writably_mapped (in->f_mapping); + +for (i =0 ; i <folio_batch_count (&fbatch); i++) { +struct folio *folio = fbatch.folios[i]; +size_t n; + +if (folio_pos (folio) >= end_offset) + goto out; +folio_mark_accessed (folio); + +/* +* If users can be writing to this folio using arbitrary +* virtual addresses, take care of potential aliasing +* before reading the folio on the kernel side. +*/ +if (writably_mapped) +flush_dcache_folio (folio); + + n =splice_folio_into_pipe (pipe, folio, *ppos, len); +if (!n) + goto out; + len -= n; + total_spliced += n; + *ppos += n; + in->f_ra.prev_pos = *ppos; +if (pipe_full (pipe->head, pipe->tail, pipe->max_usage)) + goto out; + } + +folio_batch_release (&fbatch); + }while (len); + +out: +folio_batch_release (&fbatch); +file_accessed (in); + +return total_spliced ? total_spliced : error; +} + +/** +* generic_file_splice_read - splice data from file to a pipe +* @in: file to splice from +* @ppos: position in @in +* @pipe: pipe to splice to +* @len: number of bytes to splice +* @flags: splice modifier flags +* +* Description: +* Will read pages from given file and fill them into a pipe. Can be +* used as long as it has more or less sane ->read_iter(). +* +*/ +ssize_t generic_file_splice_read (struct file *in,loff_t *ppos, +struct pipe_inode_info *pipe,size_t len, +unsigned int flags) +{ +if (unlikely (*ppos >=file_inode (in)->i_sb->s_maxbytes)) +return 0 ; +if (unlikely (!len)) +return 0 ; +if (in->f_flags &O_DIRECT ) +return generic_file_direct_splice_read (in, ppos, pipe, len, flags); +return generic_file_buffered_splice_read (in, ppos, pipe, len, flags); +}EXPORT_SYMBOL (generic_file_splice_read);const struct pipe_buf_operations default_pipe_buf_ops = {
include/linux/pagemap.h +2 -0
@@ -748,6 +748,8 @@ struct page *read_cache_page(struct address_space *, pgoff_t index, filler_t *filler, struct file *file);extern struct page *read_cache_page_gfp (struct address_space *mapping,pgoff_t index,gfp_t gfp_mask); +int filemap_get_pages (struct kiocb *iocb,size_t count, +struct folio_batch *fbatch);static inline struct page *read_mapping_page (struct address_space *mapping,pgoff_t index,struct file *file)
mm/filemap.c +12 -11
@@ -2440,10 +2440,8 @@ static int filemap_read_folio(struct file *file, filler_t filler, }static bool filemap_range_uptodate (struct address_space *mapping, -loff_t pos,struct iov_iter *iter,struct folio *folio) +loff_t pos,size_t count,struct folio *folio) { -int count; -if (folio_test_uptodate (folio))return true;if (!mapping->a_ops->is_partially_uptodate)
@@ -2451,7 +2449,6 @@ static bool filemap_range_uptodate(struct address_space *mapping,if (mapping->host->i_blkbits >=folio_shift (folio))return false; - count = iter->count;if (folio_pos (folio) > pos) { count -=folio_pos (folio) - pos; pos =0 ;
@@ -2463,7 +2460,7 @@ static bool filemap_range_uptodate(struct address_space *mapping, }static int filemap_update_page (struct kiocb *iocb, -struct address_space *mapping,struct iov_iter *iter, +struct address_space *mapping,size_t count,struct folio *folio) {int error;
@@ -2498,7 +2495,7 @@ static int filemap_update_page(struct kiocb *iocb, goto unlock; error =0 ; -if (filemap_range_uptodate (mapping, iocb->ki_pos, iter, folio)) +if (filemap_range_uptodate (mapping, iocb->ki_pos, count, folio)) goto unlock; error = -EAGAIN ;
@@ -2574,8 +2571,12 @@ static int filemap_readahead(struct kiocb *iocb, struct file *file,return 0 ; } -static int filemap_get_pages (struct kiocb *iocb,struct iov_iter *iter, -struct folio_batch *fbatch) +/* +* Extract some folios from the pagecache of a file, reading those pages from +* the backing store if necessary and waiting for them. +*/ +int filemap_get_pages (struct kiocb *iocb,size_t count, +struct folio_batch *fbatch) {struct file *filp = iocb->ki_filp;struct address_space *mapping = filp->f_mapping;
@@ -2585,7 +2586,7 @@ static int filemap_get_pages(struct kiocb *iocb, struct iov_iter *iter,struct folio *folio;int err =0 ; - last_index =DIV_ROUND_UP (iocb->ki_pos + iter->count,PAGE_SIZE ); + last_index =DIV_ROUND_UP (iocb->ki_pos + count,PAGE_SIZE ); retry:if (fatal_signal_pending (current))return -EINTR ;
@@ -2618,7 +2619,7 @@ static int filemap_get_pages(struct kiocb *iocb, struct iov_iter *iter,if ((iocb->ki_flags &IOCB_WAITQ ) &&folio_batch_count (fbatch) >1 ) iocb->ki_flags |=IOCB_NOWAIT ; - err =filemap_update_page (iocb, mapping, iter, folio); + err =filemap_update_page (iocb, mapping, count, folio);if (err) goto err; }
@@ -2688,7 +2689,7 @@ ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *iter,if (unlikely (iocb->ki_pos >=i_size_read (inode)))break ; - error =filemap_get_pages (iocb, iter, &fbatch); + error =filemap_get_pages (iocb,iov_iter_count (iter), &fbatch);if (error <0 )break ;
Re: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
Quoted 5 lines · David Howells
How about one of two different solutions?
(1) Repurpose the function I proposed for
generic_file_splice_read()but only for splicing fromO_DIRECTfiles; reading from non-O_DIRECT files would use anITER_PIPEas upstream.
Given the amounts of problems we had with O_DIRECT vs splice, and the fact that even doing this is a bit pointless that seems sensible to me.
Quoted 4 lines
for splicing from
O_DIRECTfiles, as (1), but also replace the splice from a buffered file with something like the patch below. This usesfilemap_get_pages()to do the reading and to get a bunch of folios from the pagecache that we can then splice into the pipe directly.
I defintively like the idea of killing ITER_PIPE. Isn't the 16 folios in a folio tree often much less than what we could fit into a single pipe buf? Unless you have a file system that can use huge folios for buffered I/O and actually does this might significantly limit performance.
With that in mind I'll try to find some time to review your actual patch, but I'm a little busy at the moment.
Re: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
Quoted 5 lines · Christoph Hellwig
I defintively like the idea of killing
ITER_PIPE. Isn't the 16 folios in a folio tree often much less than what we could fit into a single pipe buf? Unless you have a file system that can use huge folios for buffered I/O and actually does this might significantly limit performance.
There's a loop there that repeats the filemap_get_pages() until either the pipe is full or we hit EOF, the same as in filemap_read() (upon which this is based).
I want to use filemap_get_pages() if I can as that does all the readahead stuff. What might be nice, though, is if I could tell it to return rather than waiting for a folio to come uptodate if it has already returned a folio so that I can push the other side of the splice along whilst the read is in progress.
David
Matthew Wilcox
maintainerRe: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
In reply to David Howellsmm/filemap.c ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *iter, · line 2692
Quoted 6 lines · David Howells
mm/filemap.c +1
-1@@ -2688,7 +2689,7 @@ ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *iter,if (unlikely (iocb->ki_pos >=i_size_read (inode)))break ; - error =filemap_get_pages (iocb, iter, &fbatch); + error =filemap_get_pages (iocb,iov_iter_count (iter), &fbatch);
What's the point in iov_iter_count() anyway? It's more characters than iter->count, so why not use it directly?
Re: [PATCH v12 01/10] vfs, iomap: Fix generic_file_splice_read() to avoid reversion of ITER_PIPE
Quoted 2 lines · Matthew Wilcox
What's the point in
iov_iter_count()anyway? It's more characters thaniter->count, so why not use it directly?
I presume it got added for a reason and should be used in preference to direct access.
David
PATCH v12 2/10 iov_iter: Kill ITER_PIPE
In reply to David HowellsThe ITER_PIPE-type iterator was only used for generic_file_splice_read(), but that has now been switched to using ITER_BVEC instead, leaving ITER_PIPE unused - so get rid of it.
--- fs/cifs/file.c | 8 +- include/linux/uio.h | 14 -- lib/iov_iter.c | 435 +------------------------------------------- mm/filemap.c | 3 - 4 files changed, 4 insertions(+), 456 deletions(-)
fs/cifs/file.c +1 -7
@@ -3806,13 +3806,7 @@ cifs_readdata_to_iov(struct cifs_readdata *rdata, struct iov_iter *iter)size_t copy =min_t (size_t, remaining,PAGE_SIZE );size_t written; -if (unlikely (iov_iter_is_pipe (iter))) { -void *addr =kmap_atomic (page); - - written =copy_to_iter (addr, copy, iter); -kunmap_atomic (addr); - }else - written =copy_page_to_iter (page,0 , copy, iter); + written =copy_page_to_iter (page,0 , copy, iter); remaining -= written;if (written < copy &&iov_iter_count (iter) >0 )break ;
include/linux/uio.h +0 -14
@@ -11,7 +11,6 @@#include <uapi/linux/uio.h> struct page ; -struct pipe_inode_info ;struct kvec {void *iov_base;/* and that should *never* hold a userland pointer */
@@ -23,7 +22,6 @@ enum iter_type {ITER_IOVEC ,ITER_KVEC ,ITER_BVEC , -ITER_PIPE ,ITER_XARRAY ,ITER_DISCARD , ITER_UBUF,
@@ -53,15 +51,10 @@ struct iov_iter {const struct kvec *kvec;const struct bio_vec *bvec;struct xarray *xarray; -struct pipe_inode_info *pipe;void __user *ubuf; };union {unsigned long nr_segs; -struct { -unsigned int head; -unsigned int start_head; - };loff_t xarray_start; }; };
@@ -99,11 +92,6 @@ static inline bool iov_iter_is_bvec(const struct iov_iter *i)return iov_iter_type (i) ==ITER_BVEC ; } -static inline bool iov_iter_is_pipe (const struct iov_iter *i) -{ -return iov_iter_type (i) ==ITER_PIPE ; -} -static inline bool iov_iter_is_discard (const struct iov_iter *i) {return iov_iter_type (i) ==ITER_DISCARD ;
@@ -245,8 +233,6 @@ void iov_iter_kvec(struct iov_iter *i, unsigned int direction, const struct kvec unsigned long nr_segs, size_t count);void iov_iter_bvec (struct iov_iter *i,unsigned int direction,const struct bio_vec *bvec,unsigned long nr_segs,size_t count); -void iov_iter_pipe (struct iov_iter *i,unsigned int direction,struct pipe_inode_info *pipe, -size_t count);void iov_iter_discard (struct iov_iter *i,unsigned int direction,size_t count);void iov_iter_xarray (struct iov_iter *i,unsigned int direction,struct xarray *xarray,loff_t start,size_t count);
lib/iov_iter.c +3 -432
561 lines not on this page · raw message
mm/filemap.c +0 -3
@@ -2446,9 +2446,6 @@ static bool filemap_range_uptodate(struct address_space *mapping,if (folio_test_uptodate (folio))return true; -/* pipes can't handle partially uptodate pages */ -if (iov_iter_is_pipe (iter)) -return false;if (!mapping->a_ops->is_partially_uptodate)return false;if (mapping->host->i_blkbits >=folio_shift (folio))
PATCH v12 3/10 iov_iter: Define flags to qualify page extraction.
In reply to David HowellsDefine flags to qualify page extraction to pass into iov_iter_*_pages*() rather than passing in FOLL_* flags.
For now only a flag to allow peer-to-peer DMA is supported.
---
Notes:
ver #12)
- Use __bitwise for the extraction flags typedef.
ver #11)
- Use __bitwise for the extraction flags.
ver #9)
- Change extract_flags to extraction_flags.
ver #7)
- Don't use FOLL_* as a parameter, but rather define constants specifically to use with iov_iter_*_pages*().
- Drop the I/O direction constants for now.
block/bio.c | 6 +++--- block/blk-map.c | 8 ++++---- include/linux/uio.h | 10 ++++++++-- lib/iov_iter.c | 14 ++++++++------ 4 files changed, 23 insertions(+), 15 deletions(-)
block/bio.c +3 -3
@@ -1245,11 +1245,11 @@ static int bio_iov_add_zone_append_page(struct bio *bio, struct page *page,*/ static int __bio_iov_iter_get_pages (struct bio *bio,struct iov_iter *iter) { +iov_iter_extraction_t extraction_flags =0 ;unsigned short nr_pages = bio->bi_max_vecs - bio->bi_vcnt;unsigned short entries_left = bio->bi_max_vecs - bio->bi_vcnt;struct bio_vec *bv = bio->bi_io_vec + bio->bi_vcnt;struct page **pages = (struct page **)bv; -unsigned int gup_flags =0 ;ssize_t size, left;unsigned len, i =0 ;size_t offset, trim;
@@ -1264,7 +1264,7 @@ static int __bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter) pages += entries_left * (PAGE_PTRS_PER_BVEC -1 );if (bio->bi_bdev &&blk_queue_pci_p2pdma (bio->bi_bdev->bd_disk->queue)) - gup_flags |=FOLL_PCI_P2PDMA ; + extraction_flags |=ITER_ALLOW_P2PDMA ;/* * Each segment in the iov is required to be a block size multiple.
@@ -1275,7 +1275,7 @@ static int __bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)*/ size =iov_iter_get_pages (iter, pages,UINT_MAX - bio->bi_iter.bi_size, - nr_pages, &offset, gup_flags); + nr_pages, &offset, extraction_flags);if (unlikely (size <=0 ))return size ? size : -EFAULT ;
block/blk-map.c +4 -4
@@ -265,9 +265,9 @@ static struct bio *blk_rq_map_bio_alloc(struct request *rq,static int bio_map_user_iov (struct request *rq,struct iov_iter *iter,gfp_t gfp_mask) { +iov_iter_extraction_t extraction_flags =0 ;unsigned int max_sectors =queue_max_hw_sectors (rq->q);unsigned int nr_vecs =iov_iter_npages (iter,BIO_MAX_VECS ); -unsigned int gup_flags =0 ;struct bio *bio;int ret;int j;
@@ -280,7 +280,7 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,return -ENOMEM ;if (blk_queue_pci_p2pdma (rq->q)) - gup_flags |=FOLL_PCI_P2PDMA ; + extraction_flags |=ITER_ALLOW_P2PDMA ;while (iov_iter_count (iter)) {struct page **pages, *stack_pages[UIO_FASTIOV ];
@@ -291,10 +291,10 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,if (nr_vecs <=ARRAY_SIZE (stack_pages)) { pages = stack_pages; bytes =iov_iter_get_pages (iter, pages,LONG_MAX , - nr_vecs, &offs, gup_flags); + nr_vecs, &offs, extraction_flags); }else { bytes =iov_iter_get_pages_alloc (iter, &pages, -LONG_MAX , &offs, gup_flags); +LONG_MAX , &offs, extraction_flags); }if (unlikely (bytes <=0 )) { ret = bytes ? bytes : -EFAULT ;
include/linux/uio.h +8 -2
@@ -12,6 +12,8 @@struct page ; +typedef unsigned int __bitwise iov_iter_extraction_t ; +struct kvec {void *iov_base;/* and that should *never* hold a userland pointer */ size_t iov_len;
@@ -238,12 +240,12 @@ void iov_iter_xarray(struct iov_iter *i, unsigned int direction, struct xarray * loff_t start, size_t count);ssize_t iov_iter_get_pages (struct iov_iter *i,struct page **pages,size_t maxsize,unsigned maxpages,size_t *start, -unsigned gup_flags); +iov_iter_extraction_t extraction_flags);ssize_t iov_iter_get_pages2 (struct iov_iter *i,struct page **pages,size_t maxsize,unsigned maxpages,size_t *start);ssize_t iov_iter_get_pages_alloc (struct iov_iter *i,struct page ***pages,size_t maxsize,size_t *start, -unsigned gup_flags); +iov_iter_extraction_t extraction_flags);ssize_t iov_iter_get_pages_alloc2 (struct iov_iter *i,struct page ***pages,size_t maxsize,size_t *start);int iov_iter_npages (const struct iov_iter *i,int maxpages);
@@ -346,4 +348,8 @@ static inline void iov_iter_ubuf(struct iov_iter *i, unsigned int direction, }; } +/* Flags for iov_iter_get/extract_pages*() */ +/* Allow P2PDMA on the extracted pages */ +#define ITER_ALLOW_P2PDMA ((__force iov_iter_extraction_t)0x01) +#endif
lib/iov_iter.c +8 -6
@@ -1020,9 +1020,9 @@ static struct page *first_bvec_segment(const struct iov_iter *i,static ssize_t __iov_iter_get_pages_alloc (struct iov_iter *i,struct page ***pages,size_t maxsize,unsigned int maxpages,size_t *start, -unsigned int gup_flags) +iov_iter_extraction_t extraction_flags) { -unsigned int n; +unsigned int n, gup_flags =0 ;if (maxsize > i->count) maxsize = i->count;
@@ -1030,6 +1030,8 @@ static ssize_t __iov_iter_get_pages_alloc(struct iov_iter *i,return 0 ;if (maxsize >MAX_RW_COUNT ) maxsize =MAX_RW_COUNT ; +if (extraction_flags &ITER_ALLOW_P2PDMA ) + gup_flags |=FOLL_PCI_P2PDMA ;if (likely (user_backed_iter (i))) {unsigned long addr;
@@ -1081,14 +1083,14 @@ static ssize_t __iov_iter_get_pages_alloc(struct iov_iter *i,ssize_t iov_iter_get_pages (struct iov_iter *i,struct page **pages,size_t maxsize,unsigned maxpages, -size_t *start,unsigned gup_flags) +size_t *start,iov_iter_extraction_t extraction_flags) {if (!maxpages)return 0 ;BUG_ON (!pages);return __iov_iter_get_pages_alloc (i, &pages, maxsize, maxpages, - start, gup_flags); + start, extraction_flags); }EXPORT_SYMBOL_GPL (iov_iter_get_pages);
@@ -1101,14 +1103,14 @@ EXPORT_SYMBOL(iov_iter_get_pages2);ssize_t iov_iter_get_pages_alloc (struct iov_iter *i,struct page ***pages,size_t maxsize, -size_t *start,unsigned gup_flags) +size_t *start,iov_iter_extraction_t extraction_flags) {ssize_t len; *pages =NULL ; len =__iov_iter_get_pages_alloc (i, pages, maxsize, ~0U , start, - gup_flags); + extraction_flags);if (len <=0 ) {kvfree (*pages); *pages =NULL ;
PATCH v12 4/10 iov_iter: Add a function to extract a page list from an iterator
In reply to David HowellsAdd a function, iov_iter_extract_pages(), to extract a list of pages from an iterator. The pages may be returned with a pin added or nothing, depending on the type of iterator.
Add a second function, iov_iter_extract_will_pin(), to determine how the cleanup should be done.
There are two cases:
(1) ITER_IOVEC or ITER_UBUF iterator.
Extracted pages will have pins (FOLL_PIN) obtained on them so that a concurrent fork() will forcibly copy the page so that DMA is done to/from the parent's buffer and is unavailable to/unaffected by the child process.
iov_iter_extract_will_pin() will return true for this case. The caller should use something like unpin_user_page() to dispose of the page.
(2) Any other sort of iterator.
No refs or pins are obtained on the page, the assumption is made that the caller will manage page retention.
iov_iter_extract_will_pin() will return false. The pages don't need additional disposal.
---
Notes:
ver #12)
- ITER_PIPE is gone, so drop related bits.
- Don't specify FOLL_PIN as that's implied by pin_user_pages_fast().
ver #11)
- Fix iov_iter_extract_kvec_pages() to include the offset into the page in the returned starting offset.
- Use __bitwise for the extraction flags
ver #10)
- Fix use of i->kvec in iov_iter_extract_bvec_pages() to be i->bvec.
ver #9)
- Rename iov_iter_extract_mode() to iov_iter_extract_will_pin() and make it return true/false not FOLL_PIN/0 as FOLL_PIN is going to be made private to mm/.
- Change extract_flags to extraction_flags.
ver #8)
- It seems that all DIO is supposed to be done under FOLL_PIN now, and not FOLL_GET, so switch to only using pin_user_pages() for user-backed iters.
- Wrap an argument in brackets in the iov_iter_extract_mode() macro.
- Drop the extract_flags argument to iov_iter_extract_mode() for now [hch].
ver #7)
- Switch to passing in iter-specific flags rather than FOLL_* flags.
- Drop the direction flags for now.
- Use ITER_ALLOW_P2PDMA to request FOLL_PCI_P2PDMA.
- Disallow use of ITER_ALLOW_P2PDMA with non-user-backed iter.
- Add support for extraction from KVEC-type iters.
- Use iov_iter_advance() rather than open-coding it.
- Make BVEC- and KVEC-type skip over initial empty vectors.
ver #6)
- Add back the function to indicate the cleanup mode.
- Drop the cleanup_mode return arg to iov_iter_extract_pages().
- Pass FOLL_SOURCE/DEST_BUF in gup_flags. Check this against the iter data_source.
ver #4)
- Use ITER_SOURCE/DEST instead of WRITE/READ.
- Allow additional FOLL_* flags, such as FOLL_PCI_P2PDMA to be passed in.
ver #3)
- Switch to using EXPORT_SYMBOL_GPL to prevent indirect 3rd-party access to get/pin_user_pages_fast()[1].
include/linux/uio.h | 27 ++++- lib/iov_iter.c | 264 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 290 insertions(+), 1 deletion(-)
include/linux/uio.h +26 -1
@@ -347,9 +347,34 @@ static inline void iov_iter_ubuf(struct iov_iter *i, unsigned int direction, .count = count }; } -/* Flags for iov_iter_get/extract_pages*() */ /* Allow P2PDMA on the extracted pages */ #define ITER_ALLOW_P2PDMA ((__force iov_iter_extraction_t)0x01) +ssize_t iov_iter_extract_pages (struct iov_iter *i,struct page ***pages, +size_t maxsize,unsigned int maxpages, +iov_iter_extraction_t extraction_flags, +size_t *offset0); + +/** +* iov_iter_extract_will_pin - Indicate how pages from the iterator will be retained +* @iter: The iterator +* +* Examine the iterator and indicate by returning true or false as to how, if +* at all, pages extracted from the iterator will be retained by the extraction +* function. +* +* %true indicates that the pages will have a pin placed in them that the +* caller must unpin. This is must be done for DMA/async DIO to force fork() +* to forcibly copy a page for the child (the parent must retain the original +* page). +* +* %false indicates that no measures are taken and that it's up to the caller +* to retain the pages. +*/ +static inline bool iov_iter_extract_will_pin (const struct iov_iter *iter) +{ +return user_backed_iter (iter); +} +#endif
lib/iov_iter.c +264 -0
267 lines not on this page · raw message
PATCH v12 5/10 iomap: Don't get an reference on ZERO_PAGE for direct I/O block zeroing
In reply to David HowellsZERO_PAGE can't go away, no need to hold an extra reference.
--- fs/iomap/direct-io.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
fs/iomap/direct-io.c +1 -1
@@ -202,7 +202,7 @@ static void iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio, bio->bi_private = dio; bio->bi_end_io = iomap_dio_bio_end_io; -get_page (page); +bio_set_flag (bio,BIO_NO_PAGE_REF );__bio_add_page (bio, page, len,0 );iomap_dio_submit_bio (iter, dio, bio, pos); }
PATCH v12 6/10 block: Fix bio_flagged() so that gcc can better optimise it
In reply to David HowellsFix bio_flagged() so that multiple instances of it, such as:
if (bio_flagged(bio, BIO_PAGE_REFFED) || bio_flagged(bio, BIO_PAGE_PINNED))
can be combined by the gcc optimiser into a single test in assembly (arguably, this is a compiler optimisation issue[1]).
The missed optimisation stems from bio_flagged() comparing the result of the bitwise-AND to zero. This results in an out-of-line bio_release_page() being compiled to something like:
<+0>: mov 0x14(%rdi),%eax <+3>: test $0x1,%al <+5>: jne 0xffffffff816dac53 <bio_release_pages+11> <+7>: test $0x2,%al <+9>: je 0xffffffff816dac5c <bio_release_pages+20> <+11>: movzbl %sil,%esi <+15>: jmp 0xffffffff816daba1 <__bio_release_pages> <+20>: jmp 0xffffffff81d0b800 <__x86_return_thunk>
However, the test is superfluous as the return type is bool. Removing it results in:
<+0>: testb $0x3,0x14(%rdi) <+4>: je 0xffffffff816e4af4 <bio_release_pages+15> <+6>: movzbl %sil,%esi <+10>: jmp 0xffffffff816dab7c <__bio_release_pages> <+15>: jmp 0xffffffff81d0b7c0 <__x86_return_thunk>
instead.
Also, the MOVZBL instruction looks unnecessary[2] - I think it's just 're-booling' the mark_dirty parameter.
--- include/linux/bio.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/bio.h +1 -1
@@ -227,7 +227,7 @@ static inline void bio_cnt_set(struct bio *bio, unsigned int count)static inline bool bio_flagged (struct bio *bio,unsigned int bit) { -return (bio->bi_flags & (1U << bit)) !=0 ; +return bio->bi_flags & (1U << bit); }static inline void bio_set_flag (struct bio *bio,unsigned int bit)
PATCH v12 7/10 block: Replace BIO_NO_PAGE_REF with BIO_PAGE_REFFED with inverted logic
In reply to David HowellsFrom: Christoph Hellwig <[email protected]>
Replace BIO_NO_PAGE_REF with a BIO_PAGE_REFFED flag that has the inverted meaning is only set when a page reference has been acquired that needs to be released by bio_release_pages().
---
Notes:
ver #8)
- Split out from another patch [hch].
- Don't default to BIO_PAGE_REFFED [hch].
ver #5)
- Split from patch that uses iov_iter_extract_pages().
block/bio.c | 2 +- block/blk-map.c | 1 + fs/direct-io.c | 2 ++ fs/iomap/direct-io.c | 1 - include/linux/bio.h | 2 +- include/linux/blk_types.h | 2 +- 6 files changed, 6 insertions(+), 4 deletions(-)
block/bio.c +1 -1
@@ -1198,7 +1198,6 @@ void bio_iov_bvec_set(struct bio *bio, struct iov_iter *iter) bio->bi_io_vec = (struct bio_vec *)iter->bvec; bio->bi_iter.bi_bvec_done = iter->iov_offset; bio->bi_iter.bi_size = size; -bio_set_flag (bio,BIO_NO_PAGE_REF );bio_set_flag (bio,BIO_CLONED ); }
@@ -1343,6 +1342,7 @@ int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)return 0 ; } +bio_set_flag (bio,BIO_PAGE_REFFED );do { ret =__bio_iov_iter_get_pages (bio, iter); }while (!ret &&iov_iter_count (iter) && !bio_full (bio,0 ));
block/blk-map.c +1 -0
@@ -282,6 +282,7 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,if (blk_queue_pci_p2pdma (rq->q)) extraction_flags |=ITER_ALLOW_P2PDMA ; +bio_set_flag (bio,BIO_PAGE_REFFED );while (iov_iter_count (iter)) {struct page **pages, *stack_pages[UIO_FASTIOV ];ssize_t bytes;
fs/direct-io.c +2 -0
@@ -403,6 +403,8 @@ dio_bio_alloc(struct dio *dio, struct dio_submit *sdio, bio->bi_end_io = dio_bio_end_aio; else bio->bi_end_io = dio_bio_end_io; +/* for now require references for all pages */ +bio_set_flag (bio,BIO_PAGE_REFFED ); sdio->bio = bio; sdio->logical_offset_in_bio = sdio->cur_page_fs_offset; }
fs/iomap/direct-io.c +0 -1
@@ -202,7 +202,6 @@ static void iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio, bio->bi_private = dio; bio->bi_end_io = iomap_dio_bio_end_io; -bio_set_flag (bio,BIO_NO_PAGE_REF );__bio_add_page (bio, page, len,0 );iomap_dio_submit_bio (iter, dio, bio, pos); }
include/linux/bio.h +1 -1
@@ -484,7 +484,7 @@ void zero_fill_bio(struct bio *bio);static inline void bio_release_pages (struct bio *bio,bool mark_dirty) { -if (!bio_flagged (bio,BIO_NO_PAGE_REF )) +if (bio_flagged (bio,BIO_PAGE_REFFED ))__bio_release_pages (bio, mark_dirty); }
include/linux/blk_types.h +1 -1
@@ -318,7 +318,7 @@ struct bio {* bio flags */ enum { -BIO_NO_PAGE_REF ,/* don't put release vec pages */ +BIO_PAGE_REFFED ,/* put pages in bio_release_pages() */ BIO_CLONED ,/* doesn't own data */ BIO_BOUNCED ,/* bio is a bounce bio */ BIO_QUIET ,/* Make BIO Quiet */
PATCH v12 8/10 block: Add BIO_PAGE_PINNED and associated infrastructure
In reply to David HowellsAdd BIO_PAGE_PINNED to indicate that the pages in a bio are pinned (FOLL_PIN) and that the pin will need removing.
---
Notes:
ver #10)
- Drop bio_set_cleanup_mode(), open coding it instead.
ver #9)
- Only consider pinning in bio_set_cleanup_mode(). Ref'ing pages in struct bio is going away.
- page_put_unpin() is removed; call unpin_user_page() and put_page() directly.
- Use bio_release_page() in __bio_release_pages().
- BIO_PAGE_PINNED and BIO_PAGE_REFFED can't both be set, so use if-else when testing both of them.
ver #8)
- Move the infrastructure to clean up pinned pages to this patch [hch].
- Put BIO_PAGE_PINNED before BIO_PAGE_REFFED as the latter should probably be removed at some point. FOLL_PIN can then be renumbered first.
block/bio.c | 6 +++--- block/blk.h | 12 ++++++++++++ include/linux/bio.h | 3 ++- include/linux/blk_types.h | 1 + 4 files changed, 18 insertions(+), 4 deletions(-)
block/bio.c +3 -3
@@ -1176,7 +1176,7 @@ void __bio_release_pages(struct bio *bio, bool mark_dirty)bio_for_each_segment_all (bvec, bio, iter_all) {if (mark_dirty && !PageCompound (bvec->bv_page))set_page_dirty_lock (bvec->bv_page); -put_page (bvec->bv_page); +bio_release_page (bio, bvec->bv_page); } }EXPORT_SYMBOL_GPL (__bio_release_pages);
@@ -1496,8 +1496,8 @@ void bio_set_pages_dirty(struct bio *bio)* the BIO and re-dirty the pages in process context. * * It is expected that bio_check_pages_dirty() will wholly own the BIO from -* here on. It will run one put_page() against each page and will run one -* bio_put() against the BIO. +* here on. It will unpin each page and will run one bio_put() against the +* BIO. */ static void bio_dirty_fn (struct work_struct *work);
block/blk.h +12 -0
@@ -425,6 +425,18 @@ int bio_add_hw_page(struct request_queue *q, struct bio *bio, struct page *page, unsigned int len, unsigned int offset, unsigned int max_sectors, bool *same_page); +/* +* Clean up a page appropriately, where the page may be pinned, may have a +* ref taken on it or neither. +*/ +static inline void bio_release_page (struct bio *bio,struct page *page) +{ +if (bio_flagged (bio,BIO_PAGE_PINNED )) +unpin_user_page (page); +else if (bio_flagged (bio,BIO_PAGE_REFFED )) +put_page (page); +} +struct request_queue *blk_alloc_queue (int node_id);int disk_scan_partitions (struct gendisk *disk,fmode_t mode,void *owner);
include/linux/bio.h +2 -1
@@ -484,7 +484,8 @@ void zero_fill_bio(struct bio *bio);static inline void bio_release_pages (struct bio *bio,bool mark_dirty) { -if (bio_flagged (bio,BIO_PAGE_REFFED )) +if (bio_flagged (bio,BIO_PAGE_REFFED ) || +bio_flagged (bio,BIO_PAGE_PINNED ))__bio_release_pages (bio, mark_dirty); }
include/linux/blk_types.h +1 -0
@@ -318,6 +318,7 @@ struct bio {* bio flags */ enum { +BIO_PAGE_PINNED ,/* Unpin pages in bio_release_pages() */ BIO_PAGE_REFFED ,/* put pages in bio_release_pages() */ BIO_CLONED ,/* doesn't own data */ BIO_BOUNCED ,/* bio is a bounce bio */
PATCH v12 9/10 block: Convert bio_iov_iter_get_pages to use iov_iter_extract_pages
In reply to David HowellsThis will pin pages or leave them unaltered rather than getting a ref on them as appropriate to the iterator.
The pages need to be pinned for DIO rather than having refs taken on them to prevent VM copy-on-write from malfunctioning during a concurrent fork() (the result of the I/O could otherwise end up being affected by/visible to the child process).
---
Notes:
ver #10)
- Drop bio_set_cleanup_mode(), open coding it instead.
ver #8)
- Split the patch up a bit [hch].
- We should only be using pinned/non-pinned pages and not ref'd pages, so adjust the comments appropriately.
ver #7)
- Don't treat BIO_PAGE_REFFED/PINNED as being the same as FOLL_GET/PIN.
ver #5)
- Transcribe the FOLL_* flags returned by iov_iter_extract_pages() to BIO_* flags and got rid of bi_cleanup_mode.
- Replaced BIO_NO_PAGE_REF to BIO_PAGE_REFFED in the preceding patch.
block/bio.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-)
block/bio.c +12 -11
@@ -1212,7 +1212,7 @@ static int bio_iov_add_page(struct bio *bio, struct page *page, }if (same_page) -put_page (page); +bio_release_page (bio, page);return 0 ; }
@@ -1226,7 +1226,7 @@ static int bio_iov_add_zone_append_page(struct bio *bio, struct page *page, queue_max_zone_append_sectors(q), &same_page) != len)return -EINVAL ;if (same_page) -put_page (page); +bio_release_page (bio, page);return 0 ; }
@@ -1237,10 +1237,10 @@ static int bio_iov_add_zone_append_page(struct bio *bio, struct page *page,* @bio: bio to add pages to * @iter: iov iterator describing the region to be mapped * -* Pins pages from *iter and appends them to @bio's bvec array. The -* pages will have to be released using put_page() when done. -* For multi-segment *iter, this function only adds pages from the -* next non-empty segment of the iov iterator. +* Extracts pages from *iter and appends them to @bio's bvec array. The pages +* will have to be cleaned up in the way indicated by the BIO_PAGE_PINNED flag. +* For a multi-segment *iter, this function only adds pages from the next +* non-empty segment of the iov iterator. */ static int __bio_iov_iter_get_pages (struct bio *bio,struct iov_iter *iter) {
@@ -1272,9 +1272,9 @@ static int __bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)* result to ensure the bio's total size is correct. The remainder of * the iov data will be picked up in the next bio iteration. */ - size =iov_iter_get_pages (iter, pages, -UINT_MAX - bio->bi_iter.bi_size, - nr_pages, &offset, extraction_flags); + size =iov_iter_extract_pages (iter, &pages, +UINT_MAX - bio->bi_iter.bi_size, + nr_pages, extraction_flags, &offset);if (unlikely (size <=0 ))return size ? size : -EFAULT ;
@@ -1307,7 +1307,7 @@ static int __bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)iov_iter_revert (iter, left); out:while (i < nr_pages) -put_page (pages[i++]); +bio_release_page (bio, pages[i++]);return ret; }
@@ -1342,7 +1342,8 @@ int bio_iov_iter_get_pages(struct bio *bio, struct iov_iter *iter)return 0 ; } -bio_set_flag (bio,BIO_PAGE_REFFED ); +if (iov_iter_extract_will_pin (iter)) +bio_set_flag (bio,BIO_PAGE_PINNED );do { ret =__bio_iov_iter_get_pages (bio, iter); }while (!ret &&iov_iter_count (iter) && !bio_full (bio,0 ));
PATCH v12 10/10 block: convert bio_map_user_iov to use iov_iter_extract_pages
In reply to David HowellsThis will pin pages or leave them unaltered rather than getting a ref on them as appropriate to the iterator.
The pages need to be pinned for DIO rather than having refs taken on them to prevent VM copy-on-write from malfunctioning during a concurrent fork() (the result of the I/O could otherwise end up being visible to/affected by the child process).
---
Notes:
ver #10)
- Drop bio_set_cleanup_mode(), open coding it instead.
ver #8)
- Split the patch up a bit [hch].
- We should only be using pinned/non-pinned pages and not ref'd pages, so adjust the comments appropriately.
ver #7)
- Don't treat BIO_PAGE_REFFED/PINNED as being the same as FOLL_GET/PIN.
ver #5)
- Transcribe the FOLL_* flags returned by iov_iter_extract_pages() to BIO_* flags and got rid of bi_cleanup_mode.
- Replaced BIO_NO_PAGE_REF to BIO_PAGE_REFFED in the preceding patch.
block/blk-map.c | 23 +++++++++++------------ 1 file changed, 11 insertions(+), 12 deletions(-)
block/blk-map.c +11 -12
@@ -281,22 +281,21 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,if (blk_queue_pci_p2pdma (rq->q)) extraction_flags |=ITER_ALLOW_P2PDMA ; +if (iov_iter_extract_will_pin (iter)) +bio_set_flag (bio,BIO_PAGE_PINNED ); -bio_set_flag (bio,BIO_PAGE_REFFED );while (iov_iter_count (iter)) { -struct page **pages, *stack_pages[UIO_FASTIOV ]; +struct page *stack_pages[UIO_FASTIOV ]; +struct page **pages = stack_pages;ssize_t bytes;size_t offs;int npages; -if (nr_vecs <=ARRAY_SIZE (stack_pages)) { - pages = stack_pages; - bytes =iov_iter_get_pages (iter, pages,LONG_MAX , - nr_vecs, &offs, extraction_flags); - }else { - bytes =iov_iter_get_pages_alloc (iter, &pages, -LONG_MAX , &offs, extraction_flags); - } +if (nr_vecs >ARRAY_SIZE (stack_pages)) + pages =NULL ; + + bytes =iov_iter_extract_pages (iter, &pages,LONG_MAX , + nr_vecs, extraction_flags, &offs);if (unlikely (bytes <=0 )) { ret = bytes ? bytes : -EFAULT ; goto out_unmap;
@@ -318,7 +317,7 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,if (!bio_add_hw_page (rq->q, bio, page, n, offs, max_sectors, &same_page)) {if (same_page) -put_page (page); +bio_release_page (bio, page);break ; }
@@ -330,7 +329,7 @@ static int bio_map_user_iov(struct request *rq, struct iov_iter *iter,* release the pages we didn't map into the bio, if any */ while (j < npages) -put_page (pages[j++]); +bio_release_page (bio, pages[j++]);if (pages != stack_pages)kvfree (pages);/* couldn't stuff something into bio? */
Jens Axboe
maintainerQuoted 34 lines · David Howells
Hi Jens, Al, Christoph,
Here are patches to provide support for extracting pages from an
iov_iterand to use this in the extraction functions in the block layer bio code.The patches make the following changes:
(1) Change
generic_file_splice_read()to load up anITER_BVECiterator with sufficient pages and use that rather than using anITER_PIPE. This avoids a problem[2] when__iomap_dio_rw()callsiov_iter_revert()to shorten an iterator when it races with truncation. The reversion causes the pipe iterator to prematurely release the pages it was retaining - despite the read still being in progress. This caused memory corruption.(2) Remove
ITER_PIPEand its paraphernalia asgeneric_file_splice_read()was the only user.(3) Add a function,
iov_iter_extract_pages()to replace iov_iter_get_pages*() that gets refs, pins or just lists the pages as appropriate to the iterator type.Add a function,
iov_iter_extract_will_pin()that will indicate from the iterator type how the cleanup is to be performed, returning true if the pages will need unpinning, false otherwise.(4) Make the bio struct carry a pair of flags to indicate the cleanup mode.
BIO_NO_PAGE_REFis replaced withBIO_PAGE_REFFED(indicatingFOLL_GETwas used) andBIO_PAGE_PINNED(indicatingFOLL_PINwas used) is added.
BIO_PAGE_REFFEDwill go away, but at the momentfs/direct-io.csets it and this series does not fully address that file.
Quoted 7 lines
(5) Add a function,
bio_release_page(), to release a page appropriately to the cleanup mode indicated by the BIO_PAGE_* flags.(6) Make the iter-to-bio code use
iov_iter_extract_pages()to retain the pages appropriately and clean them up later.(7) Fix
bio_flagged()so that it doesn't prevent a gcc optimisation.
I've updated the for-6.3/iov-extract branch and the for-next branch. This isn't done to bypass any review, just so we can get some more testing on this (and because the old one is known broken).
Signature · 1 line
Jens Axboe
Quoted 3 lines · Jens Axboe
I've updated the for-6.3/iov-extract branch and the for-next branch. This isn't done to bypass any review, just so we can get some more testing on this (and because the old one is known broken).
Thanks!
David