From: Pankaj Raghav <[email protected]>
This is the sixth version of the series that enables block size > page size
(Large Block Size) in XFS targetted for inclusion in 6.11.
The context and motivation can be seen in cover letter of the RFC v1 [0].
We also recorded a talk about this effort at LPC [1], if someone would
like more context on this effort.
The major change on this v6 is max order is respected by the page cache
and iomap direct IO zeroing will be using 64k buffer instead of looping
through ZERO_PAGE.
A lot of emphasis has been put on testing using kdevops, starting with an XFS
baseline [3]. The testing has been split into regression and progression.
Regression testing:
In regression testing, we ran the whole test suite to check for regressions on
existing profiles due to the page cache changes.
No regressions were found with these patches added on top.
Progression testing:
For progression testing, we tested for 8k, 16k, 32k and 64k block sizes. To
compare it with existing support, an ARM VM with 64k base page system (without
our patches) was used as a reference to check for actual failures due to LBS
support in a 4k base page size system.
There are some tests that assumes block size < page size that needs to be fixed.
We have a tree with fixes for xfstests [4], most of the changes have been posted
already, and only a few minor changes need to be posted. Already part of these
changes has been upstreamed to fstests, and new tests have also been written and
are out for review, namely for mmap zeroing-around corner cases, compaction
and fsstress races on mm, and stress testing folio truncation on file mapped
folios.
No new failures were found with the LBS support.
We've done some preliminary performance tests with fio on XFS on 4k block size
against pmem and NVMe with buffered IO and Direct IO on vanilla Vs + these
patches applied, and detected no regressions.
We also wrote an eBPF tool called blkalgn [5] to see if IO sent to the device
is aligned and at least filesystem block size in length.
For those who want this in a git tree we have this up on a kdevops
20240503-large-block-minorder branch [6].
[0] https://lore.kernel.org/lkml/[email protected]/
[1] https://www.youtube.com/watch?v=ar72r5Xf7x4
[2] https://lkml.kernel.org/r/[email protected]
[3] https://github.com/linux-kdevops/kdevops/blob/master/docs/xfs-bugs.md
489 non-critical issues and 55 critical issues. We've determined and reported
that the 55 critical issues have all fall into 5 common XFS asserts or hung
tasks and 2 memory management asserts.
[4] https://github.com/linux-kdevops/fstests/tree/lbs-fixes
[5] https://github.com/iovisor/bcc/pull/4813
[6] https://github.com/linux-kdevops/linux/tree/large-block-minorder-next-20240528
Changes since v6:
- Max order is resptected by the page cache
- No LBS support for V4 format in XFS
- Use a 64k zeroed buffer for iomap direct io zeroing
Dave Chinner (1):
xfs: use kvmalloc for xattr buffers
Hannes Reinecke (1):
readahead: rework loop in page_cache_ra_unbounded()
Luis Chamberlain (1):
mm: split a folio in minimum folio order chunks
Matthew Wilcox (Oracle) (1):
fs: Allow fine-grained control of folio sizes
Pankaj Raghav (7):
filemap: allocate mapping_min_order folios in the page cache
readahead: allocate folios with mapping_min_order in readahead
filemap: cap PTE range to be created to allowed zero fill in
folio_map_range()
iomap: fix iomap_dio_zero() for fs bs > system page size
xfs: expose block size in stat
xfs: make the calculation generic in xfs_sb_validate_fsb_count()
xfs: enable block size larger than page size support
fs/internal.h | 8 +++
fs/iomap/buffered-io.c | 5 ++
fs/iomap/direct-io.c | 9 ++-
fs/xfs/libxfs/xfs_attr_leaf.c | 15 ++---
fs/xfs/libxfs/xfs_ialloc.c | 5 ++
fs/xfs/libxfs/xfs_shared.h | 3 +
fs/xfs/xfs_icache.c | 6 +-
fs/xfs/xfs_iops.c | 2 +-
fs/xfs/xfs_mount.c | 11 +++-
fs/xfs/xfs_super.c | 18 +++---
include/linux/huge_mm.h | 14 +++--
include/linux/pagemap.h | 106 +++++++++++++++++++++++++++++-----
mm/filemap.c | 36 ++++++++----
mm/huge_memory.c | 50 +++++++++++++++-
mm/readahead.c | 98 ++++++++++++++++++++++++-------
15 files changed, 310 insertions(+), 76 deletions(-)
base-commit: 6dc544b66971c7f9909ff038b62149105272d26a
--
2.34.1
From: Pankaj Raghav <[email protected]>
Usually the page cache does not extend beyond the size of the inode,
therefore, no PTEs are created for folios that extend beyond the size.
But with LBS support, we might extend page cache beyond the size of the
inode as we need to guarantee folios of minimum order. Cap the PTE range
to be created for the page cache up to the max allowed zero-fill file
end, which is aligned to the PAGE_SIZE.
An fstests test has been created to trigger this edge case [0].
[0] https://lore.kernel.org/fstests/[email protected]/
Signed-off-by: Luis Chamberlain <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
mm/filemap.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/mm/filemap.c b/mm/filemap.c
index 0914ef2e8256..e398fa7b2ef6 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -3610,7 +3610,7 @@ vm_fault_t filemap_map_pages(struct vm_fault *vmf,
struct vm_area_struct *vma = vmf->vma;
struct file *file = vma->vm_file;
struct address_space *mapping = file->f_mapping;
- pgoff_t last_pgoff = start_pgoff;
+ pgoff_t file_end, last_pgoff = start_pgoff;
unsigned long addr;
XA_STATE(xas, &mapping->i_pages, start_pgoff);
struct folio *folio;
@@ -3636,6 +3636,10 @@ vm_fault_t filemap_map_pages(struct vm_fault *vmf,
goto out;
}
+ file_end = DIV_ROUND_UP(i_size_read(mapping->host), PAGE_SIZE) - 1;
+ if (end_pgoff > file_end)
+ end_pgoff = file_end;
+
folio_type = mm_counter_file(folio);
do {
unsigned long end;
--
2.34.1
From: Pankaj Raghav <[email protected]>
filemap_create_folio() and do_read_cache_folio() were always allocating
folio of order 0. __filemap_get_folio was trying to allocate higher
order folios when fgp_flags had higher order hint set but it will default
to order 0 folio if higher order memory allocation fails.
Supporting mapping_min_order implies that we guarantee each folio in the
page cache has at least an order of mapping_min_order. When adding new
folios to the page cache we must also ensure the index used is aligned to
the mapping_min_order as the page cache requires the index to be aligned
to the order of the folio.
Signed-off-by: Pankaj Raghav <[email protected]>
Co-developed-by: Luis Chamberlain <[email protected]>
Signed-off-by: Luis Chamberlain <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
---
include/linux/pagemap.h | 20 ++++++++++++++++++++
mm/filemap.c | 24 +++++++++++++++++-------
2 files changed, 37 insertions(+), 7 deletions(-)
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index 228275e7049f..899b8d751768 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -439,6 +439,26 @@ unsigned int mapping_min_folio_order(const struct address_space *mapping)
return (mapping->flags & AS_FOLIO_ORDER_MIN_MASK) >> AS_FOLIO_ORDER_MIN;
}
+static inline unsigned long mapping_min_folio_nrpages(struct address_space *mapping)
+{
+ return 1UL << mapping_min_folio_order(mapping);
+}
+
+/**
+ * mapping_align_start_index() - Align starting index based on the min
+ * folio order of the page cache.
+ * @mapping: The address_space.
+ *
+ * Ensure the index used is aligned to the minimum folio order when adding
+ * new folios to the page cache by rounding down to the nearest minimum
+ * folio number of pages.
+ */
+static inline pgoff_t mapping_align_start_index(struct address_space *mapping,
+ pgoff_t index)
+{
+ return round_down(index, mapping_min_folio_nrpages(mapping));
+}
+
/*
* Large folio support currently depends on THP. These dependencies are
* being worked on but are not yet fixed.
diff --git a/mm/filemap.c b/mm/filemap.c
index 308714a44a0f..0914ef2e8256 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -859,6 +859,8 @@ noinline int __filemap_add_folio(struct address_space *mapping,
VM_BUG_ON_FOLIO(!folio_test_locked(folio), folio);
VM_BUG_ON_FOLIO(folio_test_swapbacked(folio), folio);
+ VM_BUG_ON_FOLIO(folio_order(folio) < mapping_min_folio_order(mapping),
+ folio);
mapping_set_update(&xas, mapping);
VM_BUG_ON_FOLIO(index & (folio_nr_pages(folio) - 1), folio);
@@ -1919,8 +1921,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
folio_wait_stable(folio);
no_page:
if (!folio && (fgp_flags & FGP_CREAT)) {
- unsigned order = FGF_GET_ORDER(fgp_flags);
+ unsigned int min_order = mapping_min_folio_order(mapping);
+ unsigned int order = max(min_order, FGF_GET_ORDER(fgp_flags));
int err;
+ index = mapping_align_start_index(mapping, index);
if ((fgp_flags & FGP_WRITE) && mapping_can_writeback(mapping))
gfp |= __GFP_WRITE;
@@ -1958,7 +1962,7 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
break;
folio_put(folio);
folio = NULL;
- } while (order-- > 0);
+ } while (order-- > min_order);
if (err == -EEXIST)
goto repeat;
@@ -2447,13 +2451,16 @@ static int filemap_update_page(struct kiocb *iocb,
}
static int filemap_create_folio(struct file *file,
- struct address_space *mapping, pgoff_t index,
+ struct address_space *mapping, loff_t pos,
struct folio_batch *fbatch)
{
struct folio *folio;
int error;
+ unsigned int min_order = mapping_min_folio_order(mapping);
+ pgoff_t index;
- folio = filemap_alloc_folio(mapping_gfp_mask(mapping), 0);
+ folio = filemap_alloc_folio(mapping_gfp_mask(mapping),
+ min_order);
if (!folio)
return -ENOMEM;
@@ -2471,6 +2478,8 @@ static int filemap_create_folio(struct file *file,
* well to keep locking rules simple.
*/
filemap_invalidate_lock_shared(mapping);
+ /* index in PAGE units but aligned to min_order number of pages. */
+ index = (pos >> (PAGE_SHIFT + min_order)) << min_order;
error = filemap_add_folio(mapping, folio, index,
mapping_gfp_constraint(mapping, GFP_KERNEL));
if (error == -EEXIST)
@@ -2531,8 +2540,7 @@ static int filemap_get_pages(struct kiocb *iocb, size_t count,
if (!folio_batch_count(fbatch)) {
if (iocb->ki_flags & (IOCB_NOWAIT | IOCB_WAITQ))
return -EAGAIN;
- err = filemap_create_folio(filp, mapping,
- iocb->ki_pos >> PAGE_SHIFT, fbatch);
+ err = filemap_create_folio(filp, mapping, iocb->ki_pos, fbatch);
if (err == AOP_TRUNCATED_PAGE)
goto retry;
return err;
@@ -3748,9 +3756,11 @@ static struct folio *do_read_cache_folio(struct address_space *mapping,
repeat:
folio = filemap_get_folio(mapping, index);
if (IS_ERR(folio)) {
- folio = filemap_alloc_folio(gfp, 0);
+ folio = filemap_alloc_folio(gfp,
+ mapping_min_folio_order(mapping));
if (!folio)
return ERR_PTR(-ENOMEM);
+ index = mapping_align_start_index(mapping, index);
err = filemap_add_folio(mapping, folio, index, gfp);
if (unlikely(err)) {
folio_put(folio);
--
2.34.1
From: Pankaj Raghav <[email protected]>
page_cache_ra_unbounded() was allocating single pages (0 order folios)
if there was no folio found in an index. Allocate mapping_min_order folios
as we need to guarantee the minimum order if it is set.
When read_pages() is triggered and if a page is already present, check
for truncation and move the ractl->_index by mapping_min_nrpages if that
folio was truncated. This is done to ensure we keep the alignment
requirement while adding a folio to the page cache.
page_cache_ra_order() tries to allocate folio to the page cache with a
higher order if the index aligns with that order. Modify it so that the
order does not go below the mapping_min_order requirement of the page
cache. This function will do the right thing even if the new_order passed
is less than the mapping_min_order.
When adding new folios to the page cache we must also ensure the index
used is aligned to the mapping_min_order as the page cache requires the
index to be aligned to the order of the folio.
readahead_expand() is called from readahead aops to extend the range of
the readahead so this function can assume ractl->_index to be aligned with
min_order.
Signed-off-by: Pankaj Raghav <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
---
mm/readahead.c | 85 +++++++++++++++++++++++++++++++++++++++++---------
1 file changed, 71 insertions(+), 14 deletions(-)
diff --git a/mm/readahead.c b/mm/readahead.c
index da34b28da02c..389cd802da63 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -206,9 +206,10 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
unsigned long nr_to_read, unsigned long lookahead_size)
{
struct address_space *mapping = ractl->mapping;
- unsigned long index = readahead_index(ractl);
+ unsigned long ra_folio_index, index = readahead_index(ractl);
gfp_t gfp_mask = readahead_gfp_mask(mapping);
- unsigned long i = 0;
+ unsigned long mark, i = 0;
+ unsigned int min_nrpages = mapping_min_folio_nrpages(mapping);
/*
* Partway through the readahead operation, we will have added
@@ -223,6 +224,22 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
unsigned int nofs = memalloc_nofs_save();
filemap_invalidate_lock_shared(mapping);
+ index = mapping_align_start_index(mapping, index);
+
+ /*
+ * As iterator `i` is aligned to min_nrpages, round_up the
+ * difference between nr_to_read and lookahead_size to mark the
+ * index that only has lookahead or "async_region" to set the
+ * readahead flag.
+ */
+ ra_folio_index = round_up(readahead_index(ractl) + nr_to_read - lookahead_size,
+ min_nrpages);
+ mark = ra_folio_index - index;
+ if (index != readahead_index(ractl)) {
+ nr_to_read += readahead_index(ractl) - index;
+ ractl->_index = index;
+ }
+
/*
* Preallocate as many pages as we will need.
*/
@@ -230,7 +247,9 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
struct folio *folio = xa_load(&mapping->i_pages, index + i);
int ret;
+
if (folio && !xa_is_value(folio)) {
+ long nr_pages = folio_nr_pages(folio);
/*
* Page already present? Kick off the current batch
* of contiguous pages before continuing with the
@@ -240,12 +259,24 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
* not worth getting one just for that.
*/
read_pages(ractl);
- ractl->_index += folio_nr_pages(folio);
+
+ /*
+ * Move the ractl->_index by at least min_pages
+ * if the folio got truncated to respect the
+ * alignment constraint in the page cache.
+ *
+ */
+ if (mapping != folio->mapping)
+ nr_pages = min_nrpages;
+
+ VM_BUG_ON_FOLIO(nr_pages < min_nrpages, folio);
+ ractl->_index += nr_pages;
i = ractl->_index + ractl->_nr_pages - index;
continue;
}
- folio = filemap_alloc_folio(gfp_mask, 0);
+ folio = filemap_alloc_folio(gfp_mask,
+ mapping_min_folio_order(mapping));
if (!folio)
break;
@@ -255,11 +286,11 @@ void page_cache_ra_unbounded(struct readahead_control *ractl,
if (ret == -ENOMEM)
break;
read_pages(ractl);
- ractl->_index++;
+ ractl->_index += min_nrpages;
i = ractl->_index + ractl->_nr_pages - index;
continue;
}
- if (i == nr_to_read - lookahead_size)
+ if (i == mark)
folio_set_readahead(folio);
ractl->_workingset |= folio_test_workingset(folio);
ractl->_nr_pages += folio_nr_pages(folio);
@@ -493,13 +524,19 @@ void page_cache_ra_order(struct readahead_control *ractl,
{
struct address_space *mapping = ractl->mapping;
pgoff_t index = readahead_index(ractl);
+ unsigned int min_order = mapping_min_folio_order(mapping);
pgoff_t limit = (i_size_read(mapping->host) - 1) >> PAGE_SHIFT;
pgoff_t mark = index + ra->size - ra->async_size;
unsigned int nofs;
int err = 0;
gfp_t gfp = readahead_gfp_mask(mapping);
+ unsigned int min_ra_size = max(4, mapping_min_folio_nrpages(mapping));
- if (!mapping_large_folio_support(mapping) || ra->size < 4)
+ /*
+ * Fallback when size < min_nrpages as each folio should be
+ * at least min_nrpages anyway.
+ */
+ if (!mapping_large_folio_support(mapping) || ra->size < min_ra_size)
goto fallback;
limit = min(limit, index + ra->size - 1);
@@ -508,11 +545,20 @@ void page_cache_ra_order(struct readahead_control *ractl,
new_order += 2;
new_order = min(mapping_max_folio_order(mapping), new_order);
new_order = min_t(unsigned int, new_order, ilog2(ra->size));
+ new_order = max(new_order, min_order);
}
/* See comment in page_cache_ra_unbounded() */
nofs = memalloc_nofs_save();
filemap_invalidate_lock_shared(mapping);
+ /*
+ * If the new_order is greater than min_order and index is
+ * already aligned to new_order, then this will be noop as index
+ * aligned to new_order should also be aligned to min_order.
+ */
+ ractl->_index = mapping_align_start_index(mapping, index);
+ index = readahead_index(ractl);
+
while (index <= limit) {
unsigned int order = new_order;
@@ -520,7 +566,7 @@ void page_cache_ra_order(struct readahead_control *ractl,
if (index & ((1UL << order) - 1))
order = __ffs(index);
/* Don't allocate pages past EOF */
- while (index + (1UL << order) - 1 > limit)
+ while (order > min_order && index + (1UL << order) - 1 > limit)
order--;
err = ra_alloc_folio(ractl, index, mark, order, gfp);
if (err)
@@ -784,8 +830,15 @@ void readahead_expand(struct readahead_control *ractl,
struct file_ra_state *ra = ractl->ra;
pgoff_t new_index, new_nr_pages;
gfp_t gfp_mask = readahead_gfp_mask(mapping);
+ unsigned long min_nrpages = mapping_min_folio_nrpages(mapping);
+ unsigned int min_order = mapping_min_folio_order(mapping);
new_index = new_start / PAGE_SIZE;
+ /*
+ * Readahead code should have aligned the ractl->_index to
+ * min_nrpages before calling readahead aops.
+ */
+ VM_BUG_ON(!IS_ALIGNED(ractl->_index, min_nrpages));
/* Expand the leading edge downwards */
while (ractl->_index > new_index) {
@@ -795,9 +848,11 @@ void readahead_expand(struct readahead_control *ractl,
if (folio && !xa_is_value(folio))
return; /* Folio apparently present */
- folio = filemap_alloc_folio(gfp_mask, 0);
+ folio = filemap_alloc_folio(gfp_mask, min_order);
if (!folio)
return;
+
+ index = mapping_align_start_index(mapping, index);
if (filemap_add_folio(mapping, folio, index, gfp_mask) < 0) {
folio_put(folio);
return;
@@ -807,7 +862,7 @@ void readahead_expand(struct readahead_control *ractl,
ractl->_workingset = true;
psi_memstall_enter(&ractl->_pflags);
}
- ractl->_nr_pages++;
+ ractl->_nr_pages += min_nrpages;
ractl->_index = folio->index;
}
@@ -822,9 +877,11 @@ void readahead_expand(struct readahead_control *ractl,
if (folio && !xa_is_value(folio))
return; /* Folio apparently present */
- folio = filemap_alloc_folio(gfp_mask, 0);
+ folio = filemap_alloc_folio(gfp_mask, min_order);
if (!folio)
return;
+
+ index = mapping_align_start_index(mapping, index);
if (filemap_add_folio(mapping, folio, index, gfp_mask) < 0) {
folio_put(folio);
return;
@@ -834,10 +891,10 @@ void readahead_expand(struct readahead_control *ractl,
ractl->_workingset = true;
psi_memstall_enter(&ractl->_pflags);
}
- ractl->_nr_pages++;
+ ractl->_nr_pages += min_nrpages;
if (ra) {
- ra->size++;
- ra->async_size++;
+ ra->size += min_nrpages;
+ ra->async_size += min_nrpages;
}
}
}
--
2.34.1
From: Luis Chamberlain <[email protected]>
split_folio() and split_folio_to_list() assume order 0, to support
minorder we must expand these to check the folio mapping order and use
that.
Set new_order to be at least minimum folio order if it is set in
split_huge_page_to_list() so that we can maintain minimum folio order
requirement in the page cache.
Update the debugfs write files used for testing to ensure the order
is respected as well. We simply enforce the min order when a file
mapping is used.
Signed-off-by: Luis Chamberlain <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
include/linux/huge_mm.h | 14 ++++++++----
mm/huge_memory.c | 50 ++++++++++++++++++++++++++++++++++++++---
2 files changed, 57 insertions(+), 7 deletions(-)
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
index 87682498a5af..6a8e527b78a2 100644
--- a/include/linux/huge_mm.h
+++ b/include/linux/huge_mm.h
@@ -88,6 +88,8 @@ extern struct kobj_attribute shmem_enabled_attr;
#define thp_vma_allowable_order(vma, vm_flags, tva_flags, order) \
(!!thp_vma_allowable_orders(vma, vm_flags, tva_flags, BIT(order)))
+#define split_folio(f) split_folio_to_list(f, NULL)
+
#ifdef CONFIG_PGTABLE_HAS_HUGE_LEAVES
#define HPAGE_PMD_SHIFT PMD_SHIFT
#define HPAGE_PUD_SHIFT PUD_SHIFT
@@ -307,9 +309,10 @@ unsigned long thp_get_unmapped_area_vmflags(struct file *filp, unsigned long add
bool can_split_folio(struct folio *folio, int *pextra_pins);
int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
unsigned int new_order);
+int split_folio_to_list(struct folio *folio, struct list_head *list);
static inline int split_huge_page(struct page *page)
{
- return split_huge_page_to_list_to_order(page, NULL, 0);
+ return split_folio(page_folio(page));
}
void deferred_split_folio(struct folio *folio);
@@ -474,6 +477,12 @@ static inline int split_huge_page(struct page *page)
{
return 0;
}
+
+static inline int split_folio_to_list(struct folio *folio, struct list_head *list)
+{
+ return 0;
+}
+
static inline void deferred_split_folio(struct folio *folio) {}
#define split_huge_pmd(__vma, __pmd, __address) \
do { } while (0)
@@ -578,7 +587,4 @@ static inline int split_folio_to_order(struct folio *folio, int new_order)
return split_folio_to_list_to_order(folio, NULL, new_order);
}
-#define split_folio_to_list(f, l) split_folio_to_list_to_order(f, l, 0)
-#define split_folio(f) split_folio_to_order(f, 0)
-
#endif /* _LINUX_HUGE_MM_H */
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index cf9ead052d2a..e4e0b3431dc6 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -3068,6 +3068,9 @@ bool can_split_folio(struct folio *folio, int *pextra_pins)
* released, or if some unexpected race happened (e.g., anon VMA disappeared,
* truncation).
*
+ * Callers should ensure that the order respects the address space mapping
+ * min-order if one is set.
+ *
* Returns -EINVAL when trying to split to an order that is incompatible
* with the folio. Splitting to order 0 is compatible with all folios.
*/
@@ -3143,6 +3146,7 @@ int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
mapping = NULL;
anon_vma_lock_write(anon_vma);
} else {
+ unsigned int min_order;
gfp_t gfp;
mapping = folio->mapping;
@@ -3153,6 +3157,14 @@ int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
goto out;
}
+ min_order = mapping_min_folio_order(folio->mapping);
+ if (new_order < min_order) {
+ VM_WARN_ONCE(1, "Cannot split mapped folio below min-order: %u",
+ min_order);
+ ret = -EINVAL;
+ goto out;
+ }
+
gfp = current_gfp_context(mapping_gfp_mask(mapping) &
GFP_RECLAIM_MASK);
@@ -3264,6 +3276,21 @@ int split_huge_page_to_list_to_order(struct page *page, struct list_head *list,
return ret;
}
+int split_folio_to_list(struct folio *folio, struct list_head *list)
+{
+ unsigned int min_order = 0;
+
+ if (!folio_test_anon(folio)) {
+ if (!folio->mapping) {
+ count_vm_event(THP_SPLIT_PAGE_FAILED);
+ return -EBUSY;
+ }
+ min_order = mapping_min_folio_order(folio->mapping);
+ }
+
+ return split_huge_page_to_list_to_order(&folio->page, list, min_order);
+}
+
void __folio_undo_large_rmappable(struct folio *folio)
{
struct deferred_split *ds_queue;
@@ -3493,6 +3520,7 @@ static int split_huge_pages_pid(int pid, unsigned long vaddr_start,
struct vm_area_struct *vma = vma_lookup(mm, addr);
struct page *page;
struct folio *folio;
+ unsigned int target_order = new_order;
if (!vma)
break;
@@ -3529,7 +3557,7 @@ static int split_huge_pages_pid(int pid, unsigned long vaddr_start,
if (!folio_trylock(folio))
goto next;
- if (!split_folio_to_order(folio, new_order))
+ if (!split_folio_to_order(folio, target_order))
split++;
folio_unlock(folio);
@@ -3572,14 +3600,19 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
for (index = off_start; index < off_end; index += nr_pages) {
struct folio *folio = filemap_get_folio(mapping, index);
+ unsigned int min_order, target_order = new_order;
nr_pages = 1;
if (IS_ERR(folio))
continue;
- if (!folio_test_large(folio))
+ if (!folio->mapping || !folio_test_large(folio))
goto next;
+ min_order = mapping_min_folio_order(mapping);
+ if (new_order < min_order)
+ target_order = min_order;
+
total++;
nr_pages = folio_nr_pages(folio);
@@ -3589,7 +3622,18 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
if (!folio_trylock(folio))
goto next;
- if (!split_folio_to_order(folio, new_order))
+ if (!folio_test_anon(folio)) {
+ unsigned int min_order;
+
+ if (!folio->mapping)
+ goto next;
+
+ min_order = mapping_min_folio_order(folio->mapping);
+ if (new_order < target_order)
+ target_order = min_order;
+ }
+
+ if (!split_folio_to_order(folio, target_order))
split++;
folio_unlock(folio);
--
2.34.1
From: "Matthew Wilcox (Oracle)" <[email protected]>
We need filesystems to be able to communicate acceptable folio sizes
to the pagecache for a variety of uses (e.g. large block sizes).
Support a range of folio sizes between order-0 and order-31.
Signed-off-by: Matthew Wilcox (Oracle) <[email protected]>
Co-developed-by: Pankaj Raghav <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
include/linux/pagemap.h | 86 ++++++++++++++++++++++++++++++++++-------
mm/filemap.c | 6 +--
mm/readahead.c | 4 +-
3 files changed, 77 insertions(+), 19 deletions(-)
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index 8f09ed4a4451..228275e7049f 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -204,14 +204,21 @@ enum mapping_flags {
AS_EXITING = 4, /* final truncate in progress */
/* writeback related tags are not used */
AS_NO_WRITEBACK_TAGS = 5,
- AS_LARGE_FOLIO_SUPPORT = 6,
- AS_RELEASE_ALWAYS, /* Call ->release_folio(), even if no private data */
- AS_STABLE_WRITES, /* must wait for writeback before modifying
+ AS_RELEASE_ALWAYS = 6, /* Call ->release_folio(), even if no private data */
+ AS_STABLE_WRITES = 7, /* must wait for writeback before modifying
folio contents */
- AS_UNMOVABLE, /* The mapping cannot be moved, ever */
- AS_INACCESSIBLE, /* Do not attempt direct R/W access to the mapping */
+ AS_UNMOVABLE = 8, /* The mapping cannot be moved, ever */
+ AS_INACCESSIBLE = 9, /* Do not attempt direct R/W access to the mapping */
+ /* Bits 16-25 are used for FOLIO_ORDER */
+ AS_FOLIO_ORDER_BITS = 5,
+ AS_FOLIO_ORDER_MIN = 16,
+ AS_FOLIO_ORDER_MAX = AS_FOLIO_ORDER_MIN + AS_FOLIO_ORDER_BITS,
};
+#define AS_FOLIO_ORDER_MASK ((1u << AS_FOLIO_ORDER_BITS) - 1)
+#define AS_FOLIO_ORDER_MIN_MASK (AS_FOLIO_ORDER_MASK << AS_FOLIO_ORDER_MIN)
+#define AS_FOLIO_ORDER_MAX_MASK (AS_FOLIO_ORDER_MASK << AS_FOLIO_ORDER_MAX)
+
/**
* mapping_set_error - record a writeback error in the address_space
* @mapping: the mapping in which an error should be set
@@ -360,9 +367,49 @@ static inline void mapping_set_gfp_mask(struct address_space *m, gfp_t mask)
#define MAX_PAGECACHE_ORDER 8
#endif
+/*
+ * mapping_set_folio_order_range() - Set the orders supported by a file.
+ * @mapping: The address space of the file.
+ * @min: Minimum folio order (between 0-MAX_PAGECACHE_ORDER inclusive).
+ * @max: Maximum folio order (between @min-MAX_PAGECACHE_ORDER inclusive).
+ *
+ * The filesystem should call this function in its inode constructor to
+ * indicate which base size (min) and maximum size (max) of folio the VFS
+ * can use to cache the contents of the file. This should only be used
+ * if the filesystem needs special handling of folio sizes (ie there is
+ * something the core cannot know).
+ * Do not tune it based on, eg, i_size.
+ *
+ * Context: This should not be called while the inode is active as it
+ * is non-atomic.
+ */
+static inline void mapping_set_folio_order_range(struct address_space *mapping,
+ unsigned int min,
+ unsigned int max)
+{
+ if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
+ return;
+
+ if (min > MAX_PAGECACHE_ORDER)
+ min = MAX_PAGECACHE_ORDER;
+ if (max > MAX_PAGECACHE_ORDER)
+ max = MAX_PAGECACHE_ORDER;
+ if (max < min)
+ max = min;
+
+ mapping->flags = (mapping->flags & ~AS_FOLIO_ORDER_MASK) |
+ (min << AS_FOLIO_ORDER_MIN) | (max << AS_FOLIO_ORDER_MAX);
+}
+
+static inline void mapping_set_folio_min_order(struct address_space *mapping,
+ unsigned int min)
+{
+ mapping_set_folio_order_range(mapping, min, MAX_PAGECACHE_ORDER);
+}
+
/**
* mapping_set_large_folios() - Indicate the file supports large folios.
- * @mapping: The file.
+ * @mapping: The address space of the file.
*
* The filesystem should call this function in its inode constructor to
* indicate that the VFS can use large folios to cache the contents of
@@ -373,7 +420,23 @@ static inline void mapping_set_gfp_mask(struct address_space *m, gfp_t mask)
*/
static inline void mapping_set_large_folios(struct address_space *mapping)
{
- __set_bit(AS_LARGE_FOLIO_SUPPORT, &mapping->flags);
+ mapping_set_folio_order_range(mapping, 0, MAX_PAGECACHE_ORDER);
+}
+
+static inline
+unsigned int mapping_max_folio_order(const struct address_space *mapping)
+{
+ if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
+ return 0;
+ return (mapping->flags & AS_FOLIO_ORDER_MAX_MASK) >> AS_FOLIO_ORDER_MAX;
+}
+
+static inline
+unsigned int mapping_min_folio_order(const struct address_space *mapping)
+{
+ if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
+ return 0;
+ return (mapping->flags & AS_FOLIO_ORDER_MIN_MASK) >> AS_FOLIO_ORDER_MIN;
}
/*
@@ -382,16 +445,13 @@ static inline void mapping_set_large_folios(struct address_space *mapping)
*/
static inline bool mapping_large_folio_support(struct address_space *mapping)
{
- return IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) &&
- test_bit(AS_LARGE_FOLIO_SUPPORT, &mapping->flags);
+ return mapping_max_folio_order(mapping) > 0;
}
/* Return the maximum folio size for this pagecache mapping, in bytes. */
-static inline size_t mapping_max_folio_size(struct address_space *mapping)
+static inline size_t mapping_max_folio_size(const struct address_space *mapping)
{
- if (mapping_large_folio_support(mapping))
- return PAGE_SIZE << MAX_PAGECACHE_ORDER;
- return PAGE_SIZE;
+ return PAGE_SIZE << mapping_max_folio_order(mapping);
}
static inline int filemap_nr_thps(struct address_space *mapping)
diff --git a/mm/filemap.c b/mm/filemap.c
index ba06237b942d..308714a44a0f 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1933,10 +1933,8 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
if (WARN_ON_ONCE(!(fgp_flags & (FGP_LOCK | FGP_FOR_MMAP))))
fgp_flags |= FGP_LOCK;
- if (!mapping_large_folio_support(mapping))
- order = 0;
- if (order > MAX_PAGECACHE_ORDER)
- order = MAX_PAGECACHE_ORDER;
+ if (order > mapping_max_folio_order(mapping))
+ order = mapping_max_folio_order(mapping);
/* If we're not aligned, allocate a smaller folio */
if (index & ((1UL << order) - 1))
order = __ffs(index);
diff --git a/mm/readahead.c b/mm/readahead.c
index 75e934a1fd78..da34b28da02c 100644
--- a/mm/readahead.c
+++ b/mm/readahead.c
@@ -504,9 +504,9 @@ void page_cache_ra_order(struct readahead_control *ractl,
limit = min(limit, index + ra->size - 1);
- if (new_order < MAX_PAGECACHE_ORDER) {
+ if (new_order < mapping_max_folio_order(mapping)) {
new_order += 2;
- new_order = min_t(unsigned int, MAX_PAGECACHE_ORDER, new_order);
+ new_order = min(mapping_max_folio_order(mapping), new_order);
new_order = min_t(unsigned int, new_order, ilog2(ra->size));
}
--
2.34.1
From: Pankaj Raghav <[email protected]>
For block size larger than page size, the unit of efficient IO is
the block size, not the page size. Leaving stat() to report
PAGE_SIZE as the block size causes test programs like fsx to issue
illegal ranges for operations that require block size alignment
(e.g. fallocate() insert range). Hence update the preferred IO size
to reflect the block size in this case.
This change is based on a patch originally from Dave Chinner.[1]
[1] https://lwn.net/ml/linux-fsdevel/[email protected]/
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Luis Chamberlain <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
fs/xfs/xfs_iops.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/xfs/xfs_iops.c b/fs/xfs/xfs_iops.c
index ff222827e550..a7883303dee8 100644
--- a/fs/xfs/xfs_iops.c
+++ b/fs/xfs/xfs_iops.c
@@ -560,7 +560,7 @@ xfs_stat_blksize(
return 1U << mp->m_allocsize_log;
}
- return PAGE_SIZE;
+ return max_t(uint32_t, PAGE_SIZE, mp->m_sb.sb_blocksize);
}
STATIC int
--
2.34.1
From: Pankaj Raghav <[email protected]>
Page cache now has the ability to have a minimum order when allocating
a folio which is a prerequisite to add support for block size > page
size.
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Luis Chamberlain <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
fs/xfs/libxfs/xfs_ialloc.c | 5 +++++
fs/xfs/libxfs/xfs_shared.h | 3 +++
fs/xfs/xfs_icache.c | 6 ++++--
fs/xfs/xfs_mount.c | 1 -
fs/xfs/xfs_super.c | 18 ++++++++++--------
5 files changed, 22 insertions(+), 11 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index 14c81f227c5b..1e76431d75a4 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -3019,6 +3019,11 @@ xfs_ialloc_setup_geometry(
igeo->ialloc_align = mp->m_dalign;
else
igeo->ialloc_align = 0;
+
+ if (mp->m_sb.sb_blocksize > PAGE_SIZE)
+ igeo->min_folio_order = mp->m_sb.sb_blocklog - PAGE_SHIFT;
+ else
+ igeo->min_folio_order = 0;
}
/* Compute the location of the root directory inode that is laid out by mkfs. */
diff --git a/fs/xfs/libxfs/xfs_shared.h b/fs/xfs/libxfs/xfs_shared.h
index 34f104ed372c..e67a1c7cc0b0 100644
--- a/fs/xfs/libxfs/xfs_shared.h
+++ b/fs/xfs/libxfs/xfs_shared.h
@@ -231,6 +231,9 @@ struct xfs_ino_geometry {
/* precomputed value for di_flags2 */
uint64_t new_diflags2;
+ /* minimum folio order of a page cache allocation */
+ unsigned int min_folio_order;
+
};
#endif /* __XFS_SHARED_H__ */
diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index 0953163a2d84..5ed3dc9e7d90 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -89,7 +89,8 @@ xfs_inode_alloc(
/* VFS doesn't initialise i_mode or i_state! */
VFS_I(ip)->i_mode = 0;
VFS_I(ip)->i_state = 0;
- mapping_set_large_folios(VFS_I(ip)->i_mapping);
+ mapping_set_folio_min_order(VFS_I(ip)->i_mapping,
+ M_IGEO(mp)->min_folio_order);
XFS_STATS_INC(mp, vn_active);
ASSERT(atomic_read(&ip->i_pincount) == 0);
@@ -324,7 +325,8 @@ xfs_reinit_inode(
inode->i_rdev = dev;
inode->i_uid = uid;
inode->i_gid = gid;
- mapping_set_large_folios(inode->i_mapping);
+ mapping_set_folio_min_order(inode->i_mapping,
+ M_IGEO(mp)->min_folio_order);
return error;
}
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 46cb0384143b..a99454208807 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -135,7 +135,6 @@ xfs_sb_validate_fsb_count(
uint64_t max_index;
uint64_t max_bytes;
- ASSERT(PAGE_SHIFT >= sbp->sb_blocklog);
ASSERT(sbp->sb_blocklog >= BBSHIFT);
if (check_shl_overflow(nblocks, sbp->sb_blocklog, &max_bytes))
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 27e9f749c4c7..b8a93a8f35ca 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1638,16 +1638,18 @@ xfs_fs_fill_super(
goto out_free_sb;
}
- /*
- * Until this is fixed only page-sized or smaller data blocks work.
- */
if (mp->m_sb.sb_blocksize > PAGE_SIZE) {
- xfs_warn(mp,
- "File system with blocksize %d bytes. "
- "Only pagesize (%ld) or less will currently work.",
+ if (!xfs_has_crc(mp)) {
+ xfs_warn(mp,
+"V4 Filesystem with blocksize %d bytes. Only pagesize (%ld) or less is supported.",
mp->m_sb.sb_blocksize, PAGE_SIZE);
- error = -ENOSYS;
- goto out_free_sb;
+ error = -ENOSYS;
+ goto out_free_sb;
+ }
+
+ xfs_warn(mp,
+"EXPERIMENTAL: V5 Filesystem with Large Block Size (%d bytes) enabled.",
+ mp->m_sb.sb_blocksize);
}
/* Ensure this filesystem fits in the page cache limits */
--
2.34.1
From: Pankaj Raghav <[email protected]>
iomap_dio_zero() will pad a fs block with zeroes if the direct IO size
< fs block size. iomap_dio_zero() has an implicit assumption that fs block
size < page_size. This is true for most filesystems at the moment.
If the block size > page size, this will send the contents of the page
next to zero page(as len > PAGE_SIZE) to the underlying block device,
causing FS corruption.
iomap is a generic infrastructure and it should not make any assumptions
about the fs block size and the page size of the system.
Signed-off-by: Pankaj Raghav <[email protected]>
---
After disucssing a bit in LSFMM about this, it was clear that using a
PMD sized zero folio might not be a good idea[0], especially in platforms
with 64k base page size, the huge zero folio can be as high as
512M just for zeroing small block sizes in the direct IO path.
The idea to use iomap_init to allocate 64k zero buffer was suggested by
Dave Chinner as it gives decent tradeoff between memory usage and efficiency.
This is a good enough solution for now as moving beyond 64k block size
in XFS might take a while. We can work on a more generic solution in the
future to offer different sized zero folio that can go beyond 64k.
[0] https://lore.kernel.org/linux-fsdevel/[email protected]/
fs/internal.h | 8 ++++++++
fs/iomap/buffered-io.c | 5 +++++
fs/iomap/direct-io.c | 9 +++++++--
3 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/fs/internal.h b/fs/internal.h
index 84f371193f74..18eedbb82c50 100644
--- a/fs/internal.h
+++ b/fs/internal.h
@@ -35,6 +35,14 @@ static inline void bdev_cache_init(void)
int __block_write_begin_int(struct folio *folio, loff_t pos, unsigned len,
get_block_t *get_block, const struct iomap *iomap);
+/*
+ * iomap/buffered-io.c
+ */
+
+#define ZERO_FSB_SIZE (65536)
+#define ZERO_FSB_ORDER (get_order(ZERO_FSB_SIZE))
+extern struct page *zero_fs_block;
+
/*
* char_dev.c
*/
diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
index c5802a459334..2c0149c827cd 100644
--- a/fs/iomap/buffered-io.c
+++ b/fs/iomap/buffered-io.c
@@ -42,6 +42,7 @@ struct iomap_folio_state {
};
static struct bio_set iomap_ioend_bioset;
+struct page *zero_fs_block;
static inline bool ifs_is_fully_uptodate(struct folio *folio,
struct iomap_folio_state *ifs)
@@ -1998,6 +1999,10 @@ EXPORT_SYMBOL_GPL(iomap_writepages);
static int __init iomap_init(void)
{
+ zero_fs_block = alloc_pages(GFP_KERNEL | __GFP_ZERO, ZERO_FSB_ORDER);
+ if (!zero_fs_block)
+ return -ENOMEM;
+
return bioset_init(&iomap_ioend_bioset, 4 * (PAGE_SIZE / SECTOR_SIZE),
offsetof(struct iomap_ioend, io_bio),
BIOSET_NEED_BVECS);
diff --git a/fs/iomap/direct-io.c b/fs/iomap/direct-io.c
index f3b43d223a46..50c2bca8a347 100644
--- a/fs/iomap/direct-io.c
+++ b/fs/iomap/direct-io.c
@@ -236,17 +236,22 @@ static void iomap_dio_zero(const struct iomap_iter *iter, struct iomap_dio *dio,
loff_t pos, unsigned len)
{
struct inode *inode = file_inode(dio->iocb->ki_filp);
- struct page *page = ZERO_PAGE(0);
struct bio *bio;
+ /*
+ * Max block size supported is 64k
+ */
+ WARN_ON_ONCE(len > ZERO_FSB_SIZE);
+
bio = iomap_dio_alloc_bio(iter, dio, 1, REQ_OP_WRITE | REQ_SYNC | REQ_IDLE);
fscrypt_set_bio_crypt_ctx(bio, inode, pos >> inode->i_blkbits,
GFP_KERNEL);
+
bio->bi_iter.bi_sector = iomap_sector(&iter->iomap, pos);
bio->bi_private = dio;
bio->bi_end_io = iomap_dio_bio_end_io;
- __bio_add_page(bio, page, len, 0);
+ __bio_add_page(bio, zero_fs_block, len, 0);
iomap_dio_submit_bio(iter, dio, bio, pos);
}
--
2.34.1
From: Dave Chinner <[email protected]>
Pankaj Raghav reported that when filesystem block size is larger
than page size, the xattr code can use kmalloc() for high order
allocations. This triggers a useless warning in the allocator as it
is a __GFP_NOFAIL allocation here:
static inline
struct page *rmqueue(struct zone *preferred_zone,
struct zone *zone, unsigned int order,
gfp_t gfp_flags, unsigned int alloc_flags,
int migratetype)
{
struct page *page;
/*
* We most definitely don't want callers attempting to
* allocate greater than order-1 page units with __GFP_NOFAIL.
*/
>>>> WARN_ON_ONCE((gfp_flags & __GFP_NOFAIL) && (order > 1));
..
Fix this by changing all these call sites to use kvmalloc(), which
will strip the NOFAIL from the kmalloc attempt and if that fails
will do a __GFP_NOFAIL vmalloc().
This is not an issue that productions systems will see as
filesystems with block size > page size cannot be mounted by the
kernel; Pankaj is developing this functionality right now.
Reported-by: Pankaj Raghav <[email protected]>
Fixes: f078d4ea8276 ("xfs: convert kmem_alloc() to kmalloc()")
Signed-off-by: Dave Chinner <[email protected]>
Reviewed-by: Darrick J. Wong <[email protected]>
Reviewed-by: Pankaj Raghav <[email protected]>
---
fs/xfs/libxfs/xfs_attr_leaf.c | 15 ++++++---------
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/fs/xfs/libxfs/xfs_attr_leaf.c b/fs/xfs/libxfs/xfs_attr_leaf.c
index b9e98950eb3d..09f4cb061a6e 100644
--- a/fs/xfs/libxfs/xfs_attr_leaf.c
+++ b/fs/xfs/libxfs/xfs_attr_leaf.c
@@ -1138,10 +1138,7 @@ xfs_attr3_leaf_to_shortform(
trace_xfs_attr_leaf_to_sf(args);
- tmpbuffer = kmalloc(args->geo->blksize, GFP_KERNEL | __GFP_NOFAIL);
- if (!tmpbuffer)
- return -ENOMEM;
-
+ tmpbuffer = kvmalloc(args->geo->blksize, GFP_KERNEL | __GFP_NOFAIL);
memcpy(tmpbuffer, bp->b_addr, args->geo->blksize);
leaf = (xfs_attr_leafblock_t *)tmpbuffer;
@@ -1205,7 +1202,7 @@ xfs_attr3_leaf_to_shortform(
error = 0;
out:
- kfree(tmpbuffer);
+ kvfree(tmpbuffer);
return error;
}
@@ -1613,7 +1610,7 @@ xfs_attr3_leaf_compact(
trace_xfs_attr_leaf_compact(args);
- tmpbuffer = kmalloc(args->geo->blksize, GFP_KERNEL | __GFP_NOFAIL);
+ tmpbuffer = kvmalloc(args->geo->blksize, GFP_KERNEL | __GFP_NOFAIL);
memcpy(tmpbuffer, bp->b_addr, args->geo->blksize);
memset(bp->b_addr, 0, args->geo->blksize);
leaf_src = (xfs_attr_leafblock_t *)tmpbuffer;
@@ -1651,7 +1648,7 @@ xfs_attr3_leaf_compact(
*/
xfs_trans_log_buf(trans, bp, 0, args->geo->blksize - 1);
- kfree(tmpbuffer);
+ kvfree(tmpbuffer);
}
/*
@@ -2330,7 +2327,7 @@ xfs_attr3_leaf_unbalance(
struct xfs_attr_leafblock *tmp_leaf;
struct xfs_attr3_icleaf_hdr tmphdr;
- tmp_leaf = kzalloc(state->args->geo->blksize,
+ tmp_leaf = kvzalloc(state->args->geo->blksize,
GFP_KERNEL | __GFP_NOFAIL);
/*
@@ -2371,7 +2368,7 @@ xfs_attr3_leaf_unbalance(
}
memcpy(save_leaf, tmp_leaf, state->args->geo->blksize);
savehdr = tmphdr; /* struct copy */
- kfree(tmp_leaf);
+ kvfree(tmp_leaf);
}
xfs_attr3_leaf_hdr_to_disk(state->args->geo, save_leaf, &savehdr);
--
2.34.1
From: Pankaj Raghav <[email protected]>
Instead of assuming that PAGE_SHIFT is always higher than the blocklog,
make the calculation generic so that page cache count can be calculated
correctly for LBS.
Reviewed-by: Darrick J. Wong <[email protected]>
Signed-off-by: Pankaj Raghav <[email protected]>
---
fs/xfs/xfs_mount.c | 10 +++++++++-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 09eef1721ef4..46cb0384143b 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -132,11 +132,19 @@ xfs_sb_validate_fsb_count(
xfs_sb_t *sbp,
uint64_t nblocks)
{
+ uint64_t max_index;
+ uint64_t max_bytes;
+
ASSERT(PAGE_SHIFT >= sbp->sb_blocklog);
ASSERT(sbp->sb_blocklog >= BBSHIFT);
+ if (check_shl_overflow(nblocks, sbp->sb_blocklog, &max_bytes))
+ return -EFBIG;
+
/* Limited by ULONG_MAX of page cache index */
- if (nblocks >> (PAGE_SHIFT - sbp->sb_blocklog) > ULONG_MAX)
+ max_index = max_bytes >> PAGE_SHIFT;
+
+ if (max_index > ULONG_MAX)
return -EFBIG;
return 0;
}
--
2.34.1
On Wed, May 29, 2024 at 03:45:05PM +0200, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <[email protected]>
>
> iomap_dio_zero() will pad a fs block with zeroes if the direct IO size
> < fs block size. iomap_dio_zero() has an implicit assumption that fs block
> size < page_size. This is true for most filesystems at the moment.
>
> If the block size > page size, this will send the contents of the page
> next to zero page(as len > PAGE_SIZE) to the underlying block device,
> causing FS corruption.
>
> iomap is a generic infrastructure and it should not make any assumptions
> about the fs block size and the page size of the system.
>
> Signed-off-by: Pankaj Raghav <[email protected]>
> ---
>
> After disucssing a bit in LSFMM about this, it was clear that using a
> PMD sized zero folio might not be a good idea[0], especially in platforms
> with 64k base page size, the huge zero folio can be as high as
> 512M just for zeroing small block sizes in the direct IO path.
>
> The idea to use iomap_init to allocate 64k zero buffer was suggested by
> Dave Chinner as it gives decent tradeoff between memory usage and efficiency.
>
> This is a good enough solution for now as moving beyond 64k block size
> in XFS might take a while. We can work on a more generic solution in the
> future to offer different sized zero folio that can go beyond 64k.
>
> [0] https://lore.kernel.org/linux-fsdevel/[email protected]/
>
> fs/internal.h | 8 ++++++++
> fs/iomap/buffered-io.c | 5 +++++
> fs/iomap/direct-io.c | 9 +++++++--
> 3 files changed, 20 insertions(+), 2 deletions(-)
>
> diff --git a/fs/internal.h b/fs/internal.h
> index 84f371193f74..18eedbb82c50 100644
> --- a/fs/internal.h
> +++ b/fs/internal.h
> @@ -35,6 +35,14 @@ static inline void bdev_cache_init(void)
> int __block_write_begin_int(struct folio *folio, loff_t pos, unsigned len,
> get_block_t *get_block, const struct iomap *iomap);
>
> +/*
> + * iomap/buffered-io.c
> + */
> +
> +#define ZERO_FSB_SIZE (65536)
> +#define ZERO_FSB_ORDER (get_order(ZERO_FSB_SIZE))
> +extern struct page *zero_fs_block;
This is really iomap direct IO private stuff. It should be visible
anywhere else...
> +
> /*
> * char_dev.c
> */
> diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
> index c5802a459334..2c0149c827cd 100644
> --- a/fs/iomap/buffered-io.c
> +++ b/fs/iomap/buffered-io.c
> @@ -42,6 +42,7 @@ struct iomap_folio_state {
> };
>
> static struct bio_set iomap_ioend_bioset;
> +struct page *zero_fs_block;
>
> static inline bool ifs_is_fully_uptodate(struct folio *folio,
> struct iomap_folio_state *ifs)
> @@ -1998,6 +1999,10 @@ EXPORT_SYMBOL_GPL(iomap_writepages);
>
> static int __init iomap_init(void)
> {
> + zero_fs_block = alloc_pages(GFP_KERNEL | __GFP_ZERO, ZERO_FSB_ORDER);
> + if (!zero_fs_block)
> + return -ENOMEM;
> +
> return bioset_init(&iomap_ioend_bioset, 4 * (PAGE_SIZE / SECTOR_SIZE),
> offsetof(struct iomap_ioend, io_bio),
> BIOSET_NEED_BVECS);
just create an iomap_dio_init() function in iomap/direct-io.c
and call that from here. Then everything can be private to
iomap/direct-io.c...
-Dave.
--
Dave Chinner
[email protected]
On 5/29/24 15:45, Pankaj Raghav (Samsung) wrote:
> From: "Matthew Wilcox (Oracle)" <[email protected]>
>
> We need filesystems to be able to communicate acceptable folio sizes
> to the pagecache for a variety of uses (e.g. large block sizes).
> Support a range of folio sizes between order-0 and order-31.
>
> Signed-off-by: Matthew Wilcox (Oracle) <[email protected]>
> Co-developed-by: Pankaj Raghav <[email protected]>
> Signed-off-by: Pankaj Raghav <[email protected]>
> ---
> include/linux/pagemap.h | 86 ++++++++++++++++++++++++++++++++++-------
> mm/filemap.c | 6 +--
> mm/readahead.c | 4 +-
> 3 files changed, 77 insertions(+), 19 deletions(-)
>
Reviewed-by: Hannes Reinecke <[email protected]>
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
On 5/29/24 15:45, Pankaj Raghav (Samsung) wrote:
> From: Luis Chamberlain <[email protected]>
>
> split_folio() and split_folio_to_list() assume order 0, to support
> minorder we must expand these to check the folio mapping order and use
> that.
>
> Set new_order to be at least minimum folio order if it is set in
> split_huge_page_to_list() so that we can maintain minimum folio order
> requirement in the page cache.
>
> Update the debugfs write files used for testing to ensure the order
> is respected as well. We simply enforce the min order when a file
> mapping is used.
>
> Signed-off-by: Luis Chamberlain <[email protected]>
> Signed-off-by: Pankaj Raghav <[email protected]>
> ---
> include/linux/huge_mm.h | 14 ++++++++----
> mm/huge_memory.c | 50 ++++++++++++++++++++++++++++++++++++++---
> 2 files changed, 57 insertions(+), 7 deletions(-)
>
Reviewed-by: Hannes Reinecke <[email protected]>
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
On 5/29/24 15:45, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <[email protected]>
>
> Usually the page cache does not extend beyond the size of the inode,
> therefore, no PTEs are created for folios that extend beyond the size.
>
> But with LBS support, we might extend page cache beyond the size of the
> inode as we need to guarantee folios of minimum order. Cap the PTE range
> to be created for the page cache up to the max allowed zero-fill file
> end, which is aligned to the PAGE_SIZE.
>
> An fstests test has been created to trigger this edge case [0].
>
> [0] https://lore.kernel.org/fstests/[email protected]/
>
> Signed-off-by: Luis Chamberlain <[email protected]>
> Signed-off-by: Pankaj Raghav <[email protected]>
> ---
> mm/filemap.c | 6 +++++-
> 1 file changed, 5 insertions(+), 1 deletion(-)
>
Reviewed-by: Hannes Reinecke <[email protected]>
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
On 5/29/24 15:45, Pankaj Raghav (Samsung) wrote:
> From: Pankaj Raghav <[email protected]>
>
> iomap_dio_zero() will pad a fs block with zeroes if the direct IO size
> < fs block size. iomap_dio_zero() has an implicit assumption that fs block
> size < page_size. This is true for most filesystems at the moment.
>
> If the block size > page size, this will send the contents of the page
> next to zero page(as len > PAGE_SIZE) to the underlying block device,
> causing FS corruption.
>
> iomap is a generic infrastructure and it should not make any assumptions
> about the fs block size and the page size of the system.
>
> Signed-off-by: Pankaj Raghav <[email protected]>
> ---
>
> After disucssing a bit in LSFMM about this, it was clear that using a
> PMD sized zero folio might not be a good idea[0], especially in platforms
> with 64k base page size, the huge zero folio can be as high as
> 512M just for zeroing small block sizes in the direct IO path.
>
> The idea to use iomap_init to allocate 64k zero buffer was suggested by
> Dave Chinner as it gives decent tradeoff between memory usage and efficiency.
>
> This is a good enough solution for now as moving beyond 64k block size
> in XFS might take a while. We can work on a more generic solution in the
> future to offer different sized zero folio that can go beyond 64k.
>
> [0] https://lore.kernel.org/linux-fsdevel/[email protected]/
>
Reviewed-by: Hannes Reinecke <[email protected]>
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
On Wed, May 29, 2024 at 03:45:03PM +0200, Pankaj Raghav (Samsung) wrote:
> @@ -3572,14 +3600,19 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
>
> for (index = off_start; index < off_end; index += nr_pages) {
> struct folio *folio = filemap_get_folio(mapping, index);
> + unsigned int min_order, target_order = new_order;
>
> nr_pages = 1;
> if (IS_ERR(folio))
> continue;
>
> - if (!folio_test_large(folio))
> + if (!folio->mapping || !folio_test_large(folio))
> goto next;
This check is useless. folio->mapping is set to NULL on truncate,
but you haven't done anything to prevent truncate yet. That happens
later when you lock the folio.
> + min_order = mapping_min_folio_order(mapping);
You should hoist this out of the loop.
> + if (new_order < min_order)
> + target_order = min_order;
> +
> total++;
> nr_pages = folio_nr_pages(folio);
>
> @@ -3589,7 +3622,18 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
> if (!folio_trylock(folio))
> goto next;
>
> - if (!split_folio_to_order(folio, new_order))
> + if (!folio_test_anon(folio)) {
Please explain how a folio _in a file_ can be anon?
> + unsigned int min_order;
> +
> + if (!folio->mapping)
> + goto next;
> +
> + min_order = mapping_min_folio_order(folio->mapping);
> + if (new_order < target_order)
> + target_order = min_order;
Why is this being repeated?
> + }
> +
> + if (!split_folio_to_order(folio, target_order))
> split++;
>
> folio_unlock(folio);
> --
> 2.34.1
>
On Wed, May 29, 2024 at 03:45:01PM +0200, Pankaj Raghav (Samsung) wrote:
> @@ -1919,8 +1921,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
> folio_wait_stable(folio);
> no_page:
> if (!folio && (fgp_flags & FGP_CREAT)) {
> - unsigned order = FGF_GET_ORDER(fgp_flags);
> + unsigned int min_order = mapping_min_folio_order(mapping);
> + unsigned int order = max(min_order, FGF_GET_ORDER(fgp_flags));
> int err;
> + index = mapping_align_start_index(mapping, index);
>
> if ((fgp_flags & FGP_WRITE) && mapping_can_writeback(mapping))
> gfp |= __GFP_WRITE;
> @@ -1958,7 +1962,7 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
> break;
> folio_put(folio);
> folio = NULL;
> - } while (order-- > 0);
> + } while (order-- > min_order);
I'd argue you also need to change:
- if (order > 0)
+ if (order > min_order)
alloc_gfp |= __GFP_NORETRY | __GFP_NOWARN;
since that is the last point at which we can fall back. If we can't
immediately allocate a min_order folio, we want to retry, and we
want to warn if we can't get it.
On Mon, Jun 03, 2024 at 01:18:45PM +0100, Matthew Wilcox wrote:
> On Wed, May 29, 2024 at 03:45:01PM +0200, Pankaj Raghav (Samsung) wrote:
> > @@ -1919,8 +1921,10 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
> > folio_wait_stable(folio);
> > no_page:
> > if (!folio && (fgp_flags & FGP_CREAT)) {
> > - unsigned order = FGF_GET_ORDER(fgp_flags);
> > + unsigned int min_order = mapping_min_folio_order(mapping);
> > + unsigned int order = max(min_order, FGF_GET_ORDER(fgp_flags));
> > int err;
> > + index = mapping_align_start_index(mapping, index);
> >
> > if ((fgp_flags & FGP_WRITE) && mapping_can_writeback(mapping))
> > gfp |= __GFP_WRITE;
> > @@ -1958,7 +1962,7 @@ struct folio *__filemap_get_folio(struct address_space *mapping, pgoff_t index,
> > break;
> > folio_put(folio);
> > folio = NULL;
> > - } while (order-- > 0);
> > + } while (order-- > min_order);
>
> I'd argue you also need to change:
>
> - if (order > 0)
> + if (order > min_order)
> alloc_gfp |= __GFP_NORETRY | __GFP_NOWARN;
>
> since that is the last point at which we can fall back. If we can't
> immediately allocate a min_order folio, we want to retry, and we
> want to warn if we can't get it.
That is a good point. It is also a feedback to the admin if they start
using LBS and assess the memory requirement based on memory alloc
warning?
--
Pankaj
> > static int __init iomap_init(void)
> > {
> > + zero_fs_block = alloc_pages(GFP_KERNEL | __GFP_ZERO, ZERO_FSB_ORDER);
> > + if (!zero_fs_block)
> > + return -ENOMEM;
> > +
> > return bioset_init(&iomap_ioend_bioset, 4 * (PAGE_SIZE / SECTOR_SIZE),
> > offsetof(struct iomap_ioend, io_bio),
> > BIOSET_NEED_BVECS);
>
> just create an iomap_dio_init() function in iomap/direct-io.c
> and call that from here. Then everything can be private to
> iomap/direct-io.c...
Sounds good :)
On Mon, Jun 03, 2024 at 01:36:46PM +0100, Matthew Wilcox wrote:
> On Wed, May 29, 2024 at 03:45:03PM +0200, Pankaj Raghav (Samsung) wrote:
> > @@ -3572,14 +3600,19 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
> >
> > for (index = off_start; index < off_end; index += nr_pages) {
> > struct folio *folio = filemap_get_folio(mapping, index);
> > + unsigned int min_order, target_order = new_order;
> >
> > nr_pages = 1;
> > if (IS_ERR(folio))
> > continue;
> >
> > - if (!folio_test_large(folio))
> > + if (!folio->mapping || !folio_test_large(folio))
> > goto next;
>
> This check is useless. folio->mapping is set to NULL on truncate,
> but you haven't done anything to prevent truncate yet. That happens
> later when you lock the folio.
>
> > + min_order = mapping_min_folio_order(mapping);
>
> You should hoist this out of the loop.
>
> > + if (new_order < min_order)
> > + target_order = min_order;
> > +
> > total++;
> > nr_pages = folio_nr_pages(folio);
> >
> > @@ -3589,7 +3622,18 @@ static int split_huge_pages_in_file(const char *file_path, pgoff_t off_start,
> > if (!folio_trylock(folio))
> > goto next;
> >
> > - if (!split_folio_to_order(folio, new_order))
> > + if (!folio_test_anon(folio)) {
>
> Please explain how a folio _in a file_ can be anon?
>
> > + unsigned int min_order;
> > +
> > + if (!folio->mapping)
> > + goto next;
> > +
> > + min_order = mapping_min_folio_order(folio->mapping);
> > + if (new_order < target_order)
> > + target_order = min_order;
>
> Why is this being repeated?
>
> > + }
You are right. There are some repetition and checks that are not needed.
I will clean this function for the next revision.
Thanks.
--
Pankaj