[PATCH v2] mm: skip empty finite ranges in truncate_pagecache_range

[Hungyu Lin]

For a finite range ending before the first page boundary,
truncate_pagecache_range() calculates unmap_end as -1. The unsigned
comparison then treats the empty full-page interval as nonempty.

For example, lstart == 0 and lend == 0 cause unmap_mapping_range() to be
called with a zero length, meaning unmap to EOF. Punching one byte at
offset zero in a mapped FUSE file can therefore remove mappings outside
the hole, causing unnecessary page-table work and subsequent faults.

Set unmap_end to zero for these small finite ranges so that the existing
unsigned comparison skips the empty interval. Explicitly handle lend ==
-1 and lend == LLONG_MAX with LLONG_MAX, preserving EOF handling and
avoiding wrapping in the endpoint calculation at the largest inclusive
end.

Suggested-by: Jan Kara <jack@xxxxxxx>
Fixes: 5a7203947a1d ("mm: teach truncate_inode_pages_range() to handle non page aligned ranges")
Link: https://lore.kernel.org/r/20261027211505.22044-1-dennylin0707@xxxxxxxxx
Signed-off-by: Hungyu Lin <dennylin0707@xxxxxxxxx>
---
Changes in v2:
- Compute unmap_end directly as suggested by Jan Kara, keeping the
  original condition for calling unmap_mapping_range().
- Handle the EOF sentinel and LLONG_MAX explicitly.

 mm/truncate.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/mm/truncate.c b/mm/truncate.c
index 733d2713a87..2c1f4a6c336 100644
--- a/mm/truncate.c
+++ b/mm/truncate.c
@@ -945,7 +945,15 @@ void truncate_pagecache_range(struct inode *inode, loff_t lstart, loff_t lend)
 {
 	struct address_space *mapping = inode->i_mapping;
 	loff_t unmap_start = round_up(lstart, PAGE_SIZE);
-	loff_t unmap_end = round_down(1 + lend, PAGE_SIZE) - 1;
+	loff_t unmap_end;
+
+	if (lend == -1 || lend == LLONG_MAX)
+		unmap_end = LLONG_MAX;
+	else if (lend < PAGE_SIZE - 1)
+		unmap_end = 0;
+	else
+		unmap_end = round_down(1 + lend, PAGE_SIZE) - 1;
+
 	/*
 	 * This rounding is currently just for example: unmap_mapping_range
 	 * expands its hole outwards, whereas we want it to contract the hole

base-commit: 05441d697f7f9a927264f4058cbbc6a03dfcac9d
-- 
2.50.1 (Apple Git-155)