This series consists of a refactoring/correctness of updating the metadata
of tail pages, a couple of fixups for the refcounting part and a fixup for
the stack_start() function.
From this series on, instead of counting the stacks, we count the outstanding
nr_base_pages each stack has, which gives us a much better memory overview.
The other fixup is for the migration part.
A more detailed explanation can be found in the changelog of the respective
patches.
v3 -> v4:
- Fix some typos remarked by Vlastimil
- Add Reviewed-by tag from Vlastimil and Tested-by tag from Alexandre Ghiti
(closed a syzbot report for RISC)
Oscar Salvador (4):
mm,page_owner: Update metadata for tail pages
mm,page_owner: Fix refcount imbalance
mm,page_owner: Fix accounting of pages when migrating
mm,page_owner: Fix printing of stack records
Documentation/mm/page_owner.rst | 73 +++++++------
mm/page_owner.c | 188 ++++++++++++++++++--------------
2 files changed, 147 insertions(+), 114 deletions(-)
--
2.44.0
When seq_* code sees that its buffer overflowed, it re-allocates a bigger
onecand calls seq_operations->start() callback again.
stack_start() naively though that if it got called again, it meant that the
old record got already printed so it returned the next object, but that is
not true.
The consequence of that is that every time stack_stop() -> stack_start()
get called because we needed a bigger buffer, stack_start() will skip
entries, and those will not be printed.
Fix it by not advancing to the next object in stack_start().
Fixes: 765973a09803 ("mm,page_owner: display all stacks and their count")
Signed-off-by: Oscar Salvador <[email protected]>
---
mm/page_owner.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/mm/page_owner.c b/mm/page_owner.c
index b4476f45b376..9bef0b442863 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -872,13 +872,11 @@ static void *stack_start(struct seq_file *m, loff_t *ppos)
* value of stack_list.
*/
stack = smp_load_acquire(&stack_list);
+ m->private = stack;
} else {
stack = m->private;
- stack = stack->next;
}
- m->private = stack;
-
return stack;
}
--
2.44.0
__set_page_owner_handle() and __reset_page_owner() update the metadata
of all pages when the page is of a higher-order, but we miss to do the
same when the pages are migrated.
__folio_copy_owner() only updates the metadata of the head page, meaning
that the information stored in the first page and the tail pages will not
match.
Strictly speaking that is not a big problem because 1) we do not print
tail pages and 2) upon splitting all tail pages will inherit the
metadata of the head page, but it is better to have all metadata in check
should there be any problem, so it can ease debugging.
For that purpose, a couple of helpers are created
__update_page_owner_handle() which updates the metadata on allocation,
and __update_page_owner_free_handle() which does the same when the page
is freed.
__folio_copy_owner() will make use of both as it needs to entirely replace
the page_owner metadata for the new page.
Signed-off-by: Oscar Salvador <[email protected]>
Reviewed-by: Vlastimil Babka <[email protected]>
---
mm/page_owner.c | 137 ++++++++++++++++++++++++++----------------------
1 file changed, 74 insertions(+), 63 deletions(-)
diff --git a/mm/page_owner.c b/mm/page_owner.c
index d17d1351ec84..52d1ced0b57f 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -228,9 +228,58 @@ static void dec_stack_record_count(depot_stack_handle_t handle)
refcount_dec(&stack_record->count);
}
-void __reset_page_owner(struct page *page, unsigned short order)
+static inline void __update_page_owner_handle(struct page_ext *page_ext,
+ depot_stack_handle_t handle,
+ unsigned short order,
+ gfp_t gfp_mask,
+ short last_migrate_reason, u64 ts_nsec,
+ pid_t pid, pid_t tgid, char *comm)
{
int i;
+ struct page_owner *page_owner;
+
+ for (i = 0; i < (1 << order); i++) {
+ page_owner = get_page_owner(page_ext);
+ page_owner->handle = handle;
+ page_owner->order = order;
+ page_owner->gfp_mask = gfp_mask;
+ page_owner->last_migrate_reason = last_migrate_reason;
+ page_owner->pid = pid;
+ page_owner->tgid = tgid;
+ page_owner->ts_nsec = ts_nsec;
+ strscpy(page_owner->comm, comm,
+ sizeof(page_owner->comm));
+ __set_bit(PAGE_EXT_OWNER, &page_ext->flags);
+ __set_bit(PAGE_EXT_OWNER_ALLOCATED, &page_ext->flags);
+ page_ext = page_ext_next(page_ext);
+ }
+}
+
+static inline void __update_page_owner_free_handle(struct page_ext *page_ext,
+ depot_stack_handle_t handle,
+ unsigned short order,
+ pid_t pid, pid_t tgid,
+ u64 free_ts_nsec)
+{
+ int i;
+ struct page_owner *page_owner;
+
+ for (i = 0; i < (1 << order); i++) {
+ page_owner = get_page_owner(page_ext);
+ /* Only __reset_page_owner() wants to clear the bit */
+ if (handle) {
+ __clear_bit(PAGE_EXT_OWNER_ALLOCATED, &page_ext->flags);
+ page_owner->free_handle = handle;
+ }
+ page_owner->free_ts_nsec = free_ts_nsec;
+ page_owner->free_pid = current->pid;
+ page_owner->free_tgid = current->tgid;
+ page_ext = page_ext_next(page_ext);
+ }
+}
+
+void __reset_page_owner(struct page *page, unsigned short order)
+{
struct page_ext *page_ext;
depot_stack_handle_t handle;
depot_stack_handle_t alloc_handle;
@@ -245,16 +294,10 @@ void __reset_page_owner(struct page *page, unsigned short order)
alloc_handle = page_owner->handle;
handle = save_stack(GFP_NOWAIT | __GFP_NOWARN);
- for (i = 0; i < (1 << order); i++) {
- __clear_bit(PAGE_EXT_OWNER_ALLOCATED, &page_ext->flags);
- page_owner->free_handle = handle;
- page_owner->free_ts_nsec = free_ts_nsec;
- page_owner->free_pid = current->pid;
- page_owner->free_tgid = current->tgid;
- page_ext = page_ext_next(page_ext);
- page_owner = get_page_owner(page_ext);
- }
+ __update_page_owner_free_handle(page_ext, handle, order, current->pid,
+ current->tgid, free_ts_nsec);
page_ext_put(page_ext);
+
if (alloc_handle != early_handle)
/*
* early_handle is being set as a handle for all those
@@ -266,36 +309,11 @@ void __reset_page_owner(struct page *page, unsigned short order)
dec_stack_record_count(alloc_handle);
}
-static inline void __set_page_owner_handle(struct page_ext *page_ext,
- depot_stack_handle_t handle,
- unsigned short order, gfp_t gfp_mask)
-{
- struct page_owner *page_owner;
- int i;
- u64 ts_nsec = local_clock();
-
- for (i = 0; i < (1 << order); i++) {
- page_owner = get_page_owner(page_ext);
- page_owner->handle = handle;
- page_owner->order = order;
- page_owner->gfp_mask = gfp_mask;
- page_owner->last_migrate_reason = -1;
- page_owner->pid = current->pid;
- page_owner->tgid = current->tgid;
- page_owner->ts_nsec = ts_nsec;
- strscpy(page_owner->comm, current->comm,
- sizeof(page_owner->comm));
- __set_bit(PAGE_EXT_OWNER, &page_ext->flags);
- __set_bit(PAGE_EXT_OWNER_ALLOCATED, &page_ext->flags);
-
- page_ext = page_ext_next(page_ext);
- }
-}
-
noinline void __set_page_owner(struct page *page, unsigned short order,
gfp_t gfp_mask)
{
struct page_ext *page_ext;
+ u64 ts_nsec = local_clock();
depot_stack_handle_t handle;
handle = save_stack(gfp_mask);
@@ -303,7 +321,9 @@ noinline void __set_page_owner(struct page *page, unsigned short order,
page_ext = page_ext_get(page);
if (unlikely(!page_ext))
return;
- __set_page_owner_handle(page_ext, handle, order, gfp_mask);
+ __update_page_owner_handle(page_ext, handle, order, gfp_mask, -1,
+ current->pid, current->tgid, ts_nsec,
+ current->comm);
page_ext_put(page_ext);
inc_stack_record_count(handle, gfp_mask);
}
@@ -342,7 +362,7 @@ void __folio_copy_owner(struct folio *newfolio, struct folio *old)
{
struct page_ext *old_ext;
struct page_ext *new_ext;
- struct page_owner *old_page_owner, *new_page_owner;
+ struct page_owner *old_page_owner;
old_ext = page_ext_get(&old->page);
if (unlikely(!old_ext))
@@ -355,31 +375,21 @@ void __folio_copy_owner(struct folio *newfolio, struct folio *old)
}
old_page_owner = get_page_owner(old_ext);
- new_page_owner = get_page_owner(new_ext);
- new_page_owner->order = old_page_owner->order;
- new_page_owner->gfp_mask = old_page_owner->gfp_mask;
- new_page_owner->last_migrate_reason =
- old_page_owner->last_migrate_reason;
- new_page_owner->handle = old_page_owner->handle;
- new_page_owner->pid = old_page_owner->pid;
- new_page_owner->tgid = old_page_owner->tgid;
- new_page_owner->free_pid = old_page_owner->free_pid;
- new_page_owner->free_tgid = old_page_owner->free_tgid;
- new_page_owner->ts_nsec = old_page_owner->ts_nsec;
- new_page_owner->free_ts_nsec = old_page_owner->ts_nsec;
- strcpy(new_page_owner->comm, old_page_owner->comm);
-
+ __update_page_owner_handle(new_ext, old_page_owner->handle,
+ old_page_owner->order, old_page_owner->gfp_mask,
+ old_page_owner->last_migrate_reason,
+ old_page_owner->ts_nsec, old_page_owner->pid,
+ old_page_owner->tgid, old_page_owner->comm);
/*
- * We don't clear the bit on the old folio as it's going to be freed
- * after migration. Until then, the info can be useful in case of
- * a bug, and the overall stats will be off a bit only temporarily.
- * Also, migrate_misplaced_transhuge_page() can still fail the
- * migration and then we want the old folio to retain the info. But
- * in that case we also don't need to explicitly clear the info from
- * the new page, which will be freed.
+ * Do not proactively clear PAGE_EXT_OWNER{_ALLOCATED} bits as the folio
+ * will be freed after migration. Keep them until then as they may be
+ * useful.
*/
- __set_bit(PAGE_EXT_OWNER, &new_ext->flags);
- __set_bit(PAGE_EXT_OWNER_ALLOCATED, &new_ext->flags);
+ __update_page_owner_free_handle(new_ext, 0, old_page_owner->order,
+ old_page_owner->free_pid,
+ old_page_owner->free_tgid,
+ old_page_owner->free_ts_nsec);
+
page_ext_put(new_ext);
page_ext_put(old_ext);
}
@@ -787,8 +797,9 @@ static void init_pages_in_zone(pg_data_t *pgdat, struct zone *zone)
goto ext_put_continue;
/* Found early allocated page */
- __set_page_owner_handle(page_ext, early_handle,
- 0, 0);
+ __update_page_owner_handle(page_ext, early_handle, 0, 0,
+ -1, local_clock(), current->pid,
+ current->tgid, current->comm);
count++;
ext_put_continue:
page_ext_put(page_ext);
--
2.44.0
Current code does not contemplate scenarios were an allocation and
free operation on the same pages do not handle it in the same amount
at once.
To give an example, page_alloc_exact(), where we will allocate a page
of enough order to stafisfy the size request, but we will free the
remainings right away.
In the above example, we will increment the stack_record refcount
only once, but we will decrease it the same number of times as number
of unused pages we have to free.
This will lead to a warning because of refcount imbalance.
Fix this by recording the number of base pages in the refcount field.
Reported-by: [email protected]
Closes: https://lore.kernel.org/linux-mm/[email protected]
Fixes: 217b2119b9e2 ("mm,page_owner: implement the tracking of the stacks count")
Signed-off-by: Oscar Salvador <[email protected]>
Reviewed-by: Vlastimil Babka <[email protected]>
Tested-by: Alexandre Ghiti <[email protected]>
---
Documentation/mm/page_owner.rst | 73 +++++++++++++++++----------------
mm/page_owner.c | 34 ++++++++-------
2 files changed, 58 insertions(+), 49 deletions(-)
diff --git a/Documentation/mm/page_owner.rst b/Documentation/mm/page_owner.rst
index 0d0334cd5179..3a45a20fc05a 100644
--- a/Documentation/mm/page_owner.rst
+++ b/Documentation/mm/page_owner.rst
@@ -24,10 +24,10 @@ fragmentation statistics can be obtained through gfp flag information of
each page. It is already implemented and activated if page owner is
enabled. Other usages are more than welcome.
-It can also be used to show all the stacks and their outstanding
-allocations, which gives us a quick overview of where the memory is going
-without the need to screen through all the pages and match the allocation
-and free operation.
+It can also be used to show all the stacks and their current number of
+allocated base pages, which gives us a quick overview of where the memory
+is going without the need to screen through all the pages and match the
+allocation and free operation.
page owner is disabled by default. So, if you'd like to use it, you need
to add "page_owner=on" to your boot cmdline. If the kernel is built
@@ -75,42 +75,45 @@ Usage
cat /sys/kernel/debug/page_owner_stacks/show_stacks > stacks.txt
cat stacks.txt
- prep_new_page+0xa9/0x120
- get_page_from_freelist+0x7e6/0x2140
- __alloc_pages+0x18a/0x370
- new_slab+0xc8/0x580
- ___slab_alloc+0x1f2/0xaf0
- __slab_alloc.isra.86+0x22/0x40
- kmem_cache_alloc+0x31b/0x350
- __khugepaged_enter+0x39/0x100
- dup_mmap+0x1c7/0x5ce
- copy_process+0x1afe/0x1c90
- kernel_clone+0x9a/0x3c0
- __do_sys_clone+0x66/0x90
- do_syscall_64+0x7f/0x160
- entry_SYSCALL_64_after_hwframe+0x6c/0x74
- stack_count: 234
+ post_alloc_hook+0x177/0x1a0
+ get_page_from_freelist+0xd01/0xd80
+ __alloc_pages+0x39e/0x7e0
+ allocate_slab+0xbc/0x3f0
+ ___slab_alloc+0x528/0x8a0
+ kmem_cache_alloc+0x224/0x3b0
+ sk_prot_alloc+0x58/0x1a0
+ sk_alloc+0x32/0x4f0
+ inet_create+0x427/0xb50
+ __sock_create+0x2e4/0x650
+ inet_ctl_sock_create+0x30/0x180
+ igmp_net_init+0xc1/0x130
+ ops_init+0x167/0x410
+ setup_net+0x304/0xa60
+ copy_net_ns+0x29b/0x4a0
+ create_new_namespaces+0x4a1/0x820
+ nr_base_pages: 16
...
...
echo 7000 > /sys/kernel/debug/page_owner_stacks/count_threshold
cat /sys/kernel/debug/page_owner_stacks/show_stacks> stacks_7000.txt
cat stacks_7000.txt
- prep_new_page+0xa9/0x120
- get_page_from_freelist+0x7e6/0x2140
- __alloc_pages+0x18a/0x370
- alloc_pages_mpol+0xdf/0x1e0
- folio_alloc+0x14/0x50
- filemap_alloc_folio+0xb0/0x100
- page_cache_ra_unbounded+0x97/0x180
- filemap_fault+0x4b4/0x1200
- __do_fault+0x2d/0x110
- do_pte_missing+0x4b0/0xa30
- __handle_mm_fault+0x7fa/0xb70
- handle_mm_fault+0x125/0x300
- do_user_addr_fault+0x3c9/0x840
- exc_page_fault+0x68/0x150
- asm_exc_page_fault+0x22/0x30
- stack_count: 8248
+ post_alloc_hook+0x177/0x1a0
+ get_page_from_freelist+0xd01/0xd80
+ __alloc_pages+0x39e/0x7e0
+ alloc_pages_mpol+0x22e/0x490
+ folio_alloc+0xd5/0x110
+ filemap_alloc_folio+0x78/0x230
+ page_cache_ra_order+0x287/0x6f0
+ filemap_get_pages+0x517/0x1160
+ filemap_read+0x304/0x9f0
+ xfs_file_buffered_read+0xe6/0x1d0 [xfs]
+ xfs_file_read_iter+0x1f0/0x380 [xfs]
+ __kernel_read+0x3b9/0x730
+ kernel_read_file+0x309/0x4d0
+ __do_sys_finit_module+0x381/0x730
+ do_syscall_64+0x8d/0x150
+ entry_SYSCALL_64_after_hwframe+0x62/0x6a
+ nr_base_pages: 20824
...
cat /sys/kernel/debug/page_owner > page_owner_full.txt
diff --git a/mm/page_owner.c b/mm/page_owner.c
index 52d1ced0b57f..5df0d6892bdc 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -196,7 +196,8 @@ static void add_stack_record_to_list(struct stack_record *stack_record,
spin_unlock_irqrestore(&stack_list_lock, flags);
}
-static void inc_stack_record_count(depot_stack_handle_t handle, gfp_t gfp_mask)
+static void inc_stack_record_count(depot_stack_handle_t handle, gfp_t gfp_mask,
+ int nr_base_pages)
{
struct stack_record *stack_record = __stack_depot_get_stack_record(handle);
@@ -217,15 +218,20 @@ static void inc_stack_record_count(depot_stack_handle_t handle, gfp_t gfp_mask)
/* Add the new stack_record to our list */
add_stack_record_to_list(stack_record, gfp_mask);
}
- refcount_inc(&stack_record->count);
+ refcount_add(nr_base_pages, &stack_record->count);
}
-static void dec_stack_record_count(depot_stack_handle_t handle)
+static void dec_stack_record_count(depot_stack_handle_t handle,
+ int nr_base_pages)
{
struct stack_record *stack_record = __stack_depot_get_stack_record(handle);
- if (stack_record)
- refcount_dec(&stack_record->count);
+ if (!stack_record)
+ return;
+
+ if (refcount_sub_and_test(nr_base_pages, &stack_record->count))
+ pr_warn("%s: refcount went to 0 for %u handle\n", __func__,
+ handle);
}
static inline void __update_page_owner_handle(struct page_ext *page_ext,
@@ -306,7 +312,7 @@ void __reset_page_owner(struct page *page, unsigned short order)
* the machinery is not ready yet, we cannot decrement
* their refcount either.
*/
- dec_stack_record_count(alloc_handle);
+ dec_stack_record_count(alloc_handle, 1 << order);
}
noinline void __set_page_owner(struct page *page, unsigned short order,
@@ -325,7 +331,7 @@ noinline void __set_page_owner(struct page *page, unsigned short order,
current->pid, current->tgid, ts_nsec,
current->comm);
page_ext_put(page_ext);
- inc_stack_record_count(handle, gfp_mask);
+ inc_stack_record_count(handle, gfp_mask, 1 << order);
}
void __set_page_owner_migrate_reason(struct page *page, int reason)
@@ -872,11 +878,11 @@ static void *stack_next(struct seq_file *m, void *v, loff_t *ppos)
return stack;
}
-static unsigned long page_owner_stack_threshold;
+static unsigned long page_owner_pages_threshold;
static int stack_print(struct seq_file *m, void *v)
{
- int i, stack_count;
+ int i, nr_base_pages;
struct stack *stack = v;
unsigned long *entries;
unsigned long nr_entries;
@@ -887,14 +893,14 @@ static int stack_print(struct seq_file *m, void *v)
nr_entries = stack_record->size;
entries = stack_record->entries;
- stack_count = refcount_read(&stack_record->count) - 1;
+ nr_base_pages = refcount_read(&stack_record->count) - 1;
- if (stack_count < 1 || stack_count < page_owner_stack_threshold)
+ if (nr_base_pages < 1 || nr_base_pages < page_owner_pages_threshold)
return 0;
for (i = 0; i < nr_entries; i++)
seq_printf(m, " %pS\n", (void *)entries[i]);
- seq_printf(m, "stack_count: %d\n\n", stack_count);
+ seq_printf(m, "nr_base_pages: %d\n\n", nr_base_pages);
return 0;
}
@@ -924,13 +930,13 @@ static const struct file_operations page_owner_stack_operations = {
static int page_owner_threshold_get(void *data, u64 *val)
{
- *val = READ_ONCE(page_owner_stack_threshold);
+ *val = READ_ONCE(page_owner_pages_threshold);
return 0;
}
static int page_owner_threshold_set(void *data, u64 val)
{
- WRITE_ONCE(page_owner_stack_threshold, val);
+ WRITE_ONCE(page_owner_pages_threshold, val);
return 0;
}
--
2.44.0
Upon migration, new allocated pages are being given the handle of the old
pages. This is problematic because it means that for the stack which
allocated the old page, we will be substracting the old page + the new one
when that page is freed, creating an accounting imbalance.
There is an interest in keeping it that way, as otherwise the output will
biased towards migration stacks should those operations occur often, but
that is not really helpful.
The link from the new page to the old stack is being performed by calling
__update_page_owner_handle() in __folio_copy_owner().
The only thing that is left is to link the migrate stack to the old
page, so the old page will be subtracted from the migrate stack,
avoiding by doing so any possible imbalance.
Fixes: 217b2119b9e2 ("mm,page_owner: implement the tracking of the stacks count")
Signed-off-by: Oscar Salvador <[email protected]>
Reviewed-by: Vlastimil Babka <[email protected]>
---
mm/page_owner.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/mm/page_owner.c b/mm/page_owner.c
index 5df0d6892bdc..b4476f45b376 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -366,9 +366,12 @@ void __split_page_owner(struct page *page, int old_order, int new_order)
void __folio_copy_owner(struct folio *newfolio, struct folio *old)
{
+ int i;
struct page_ext *old_ext;
struct page_ext *new_ext;
struct page_owner *old_page_owner;
+ struct page_owner *new_page_owner;
+ depot_stack_handle_t migrate_handle;
old_ext = page_ext_get(&old->page);
if (unlikely(!old_ext))
@@ -381,6 +384,8 @@ void __folio_copy_owner(struct folio *newfolio, struct folio *old)
}
old_page_owner = get_page_owner(old_ext);
+ new_page_owner = get_page_owner(new_ext);
+ migrate_handle = new_page_owner->handle;
__update_page_owner_handle(new_ext, old_page_owner->handle,
old_page_owner->order, old_page_owner->gfp_mask,
old_page_owner->last_migrate_reason,
@@ -395,6 +400,16 @@ void __folio_copy_owner(struct folio *newfolio, struct folio *old)
old_page_owner->free_pid,
old_page_owner->free_tgid,
old_page_owner->free_ts_nsec);
+ /*
+ * We linked the original stack to the new folio, we need to do the same
+ * for the new one and the old folio otherwise there will be an imbalance
+ * when subtracting those pages from the stack.
+ */
+ for (i = 0; i < (1 << new_page_owner->order); i++) {
+ old_page_owner->handle = migrate_handle;
+ old_ext = page_ext_next(old_ext);
+ old_page_owner = get_page_owner(old_ext);
+ }
page_ext_put(new_ext);
page_ext_put(old_ext);
--
2.44.0
On 4/4/24 9:06 AM, Oscar Salvador wrote:
> This series consists of a refactoring/correctness of updating the metadata
> of tail pages, a couple of fixups for the refcounting part and a fixup for
> the stack_start() function.
>
> From this series on, instead of counting the stacks, we count the outstanding
> nr_base_pages each stack has, which gives us a much better memory overview.
> The other fixup is for the migration part.
>
> A more detailed explanation can be found in the changelog of the respective
> patches.
It's best to be explicit that this fixes issues in 6.9-rc1 and thus should
be in mm-hotfixes(-unstable). Also you can use e.g. [PATCH mm-hotfixes]
prefix to that effect next time.
> v3 -> v4:
> - Fix some typos remarked by Vlastimil
> - Add Reviewed-by tag from Vlastimil and Tested-by tag from Alexandre Ghiti
> (closed a syzbot report for RISC)
>
> Oscar Salvador (4):
> mm,page_owner: Update metadata for tail pages
> mm,page_owner: Fix refcount imbalance
> mm,page_owner: Fix accounting of pages when migrating
> mm,page_owner: Fix printing of stack records
>
> Documentation/mm/page_owner.rst | 73 +++++++------
> mm/page_owner.c | 188 ++++++++++++++++++--------------
> 2 files changed, 147 insertions(+), 114 deletions(-)
>
On 4/4/24 9:07 AM, Oscar Salvador wrote:
> When seq_* code sees that its buffer overflowed, it re-allocates a bigger
> onecand calls seq_operations->start() callback again.
> stack_start() naively though that if it got called again, it meant that the
> old record got already printed so it returned the next object, but that is
> not true.
>
> The consequence of that is that every time stack_stop() -> stack_start()
> get called because we needed a bigger buffer, stack_start() will skip
> entries, and those will not be printed.
>
> Fix it by not advancing to the next object in stack_start().
>
> Fixes: 765973a09803 ("mm,page_owner: display all stacks and their count")
> Signed-off-by: Oscar Salvador <[email protected]>
Reviewed-by: Vlastimil Babka <[email protected]>
> ---
> mm/page_owner.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/mm/page_owner.c b/mm/page_owner.c
> index b4476f45b376..9bef0b442863 100644
> --- a/mm/page_owner.c
> +++ b/mm/page_owner.c
> @@ -872,13 +872,11 @@ static void *stack_start(struct seq_file *m, loff_t *ppos)
> * value of stack_list.
> */
> stack = smp_load_acquire(&stack_list);
> + m->private = stack;
> } else {
> stack = m->private;
> - stack = stack->next;
> }
>
> - m->private = stack;
> -
> return stack;
> }
>
On 2024/4/4 15:06, Oscar Salvador wrote:
> This series consists of a refactoring/correctness of updating the metadata
> of tail pages, a couple of fixups for the refcounting part and a fixup for
> the stack_start() function.
>
>>From this series on, instead of counting the stacks, we count the outstanding
> nr_base_pages each stack has, which gives us a much better memory overview.
> The other fixup is for the migration part.
>
> A more detailed explanation can be found in the changelog of the respective
> patches.
I think this also should be merged into 6.9-rc1 asap, it is easy to
occur when migration.
Tested-by: Kefeng Wang <[email protected]>
>
> v3 -> v4:
> - Fix some typos remarked by Vlastimil
> - Add Reviewed-by tag from Vlastimil and Tested-by tag from Alexandre Ghiti
> (closed a syzbot report for RISC)
>
> Oscar Salvador (4):
> mm,page_owner: Update metadata for tail pages
> mm,page_owner: Fix refcount imbalance
> mm,page_owner: Fix accounting of pages when migrating
> mm,page_owner: Fix printing of stack records
>
> Documentation/mm/page_owner.rst | 73 +++++++------
> mm/page_owner.c | 188 ++++++++++++++++++--------------
> 2 files changed, 147 insertions(+), 114 deletions(-)
>
On Thu, Apr 04, 2024 at 09:07:00AM +0200, Oscar Salvador wrote:
> Current code does not contemplate scenarios were an allocation and
> free operation on the same pages do not handle it in the same amount
> at once.
> To give an example, page_alloc_exact(), where we will allocate a page
> of enough order to stafisfy the size request, but we will free the
> remainings right away.
>
> In the above example, we will increment the stack_record refcount
> only once, but we will decrease it the same number of times as number
> of unused pages we have to free.
> This will lead to a warning because of refcount imbalance.
>
> Fix this by recording the number of base pages in the refcount field.
>
> Reported-by: [email protected]
> Closes: https://lore.kernel.org/linux-mm/[email protected]
> Fixes: 217b2119b9e2 ("mm,page_owner: implement the tracking of the stacks count")
Does this also fix this?
https://lore.kernel.org/all/[email protected]/
This is a report of the backtrace changing, but the warning was
pre-existing.
> [...]
> -static void dec_stack_record_count(depot_stack_handle_t handle)
> +static void dec_stack_record_count(depot_stack_handle_t handle,
> + int nr_base_pages)
> {
> struct stack_record *stack_record = __stack_depot_get_stack_record(handle);
>
> - if (stack_record)
> - refcount_dec(&stack_record->count);
> + if (!stack_record)
> + return;
> +
> + if (refcount_sub_and_test(nr_base_pages, &stack_record->count))
> + pr_warn("%s: refcount went to 0 for %u handle\n", __func__,
> + handle);
This pr_warn() isn't needed: refcount will very loudly say the same
thing. :)
--
Kees Cook
On Mon, May 06, 2024 at 05:41:28PM +0200, Oscar Salvador wrote:
> On Mon, May 06, 2024 at 07:59:11AM -0700, Kees Cook wrote:
> > This pr_warn() isn't needed: refcount will very loudly say the same
> > thing. :)
>
> Yes, but I wanted to get the handle so I can match it with the
> backtrace.
Although on a second though, I could just switch it to pr_info(),
otherwise the warnings from both refcount and page_owner might get
tangled.
I will check and send a patch later.
--
Oscar Salvador
SUSE Labs
On Mon, May 06, 2024 at 07:59:11AM -0700, Kees Cook wrote:
> Does this also fix this?
> https://lore.kernel.org/all/[email protected]/
Hi Kess,
yes, it does.
>
> This is a report of the backtrace changing, but the warning was
> pre-existing.
>
> > [...]
> > -static void dec_stack_record_count(depot_stack_handle_t handle)
> > +static void dec_stack_record_count(depot_stack_handle_t handle,
> > + int nr_base_pages)
> > {
> > struct stack_record *stack_record = __stack_depot_get_stack_record(handle);
> >
> > - if (stack_record)
> > - refcount_dec(&stack_record->count);
> > + if (!stack_record)
> > + return;
> > +
> > + if (refcount_sub_and_test(nr_base_pages, &stack_record->count))
> > + pr_warn("%s: refcount went to 0 for %u handle\n", __func__,
> > + handle);
>
> This pr_warn() isn't needed: refcount will very loudly say the same
> thing. :)
Yes, but I wanted to get the handle so I can match it with the
backtrace.
Thanks
--
Oscar Salvador
SUSE Labs
On Mon, May 06, 2024 at 05:41:28PM +0200, Oscar Salvador wrote:
> Yes, but I wanted to get the handle so I can match it with the
> backtrace.
Ah! Yes, that makes sense. Thanks!
--
Kees Cook