diff --git a/kernel/fs/schfs/vfs.c b/kernel/fs/schfs/vfs.c
index 1139932..e7aaac9 100644
--- a/kernel/fs/schfs/vfs.c
+++ b/kernel/fs/schfs/vfs.c
@@ -142,7 +142,18 @@ schfs_vfs_ifree(struct inode *ip)
 // the tree while a copyout could fault. One page at a time keeps that buffer
 // off the 4096-byte kernel stack -- it is kalloc'd -- and bounds how long the
 // volume lock is held.
-#define BOUNCE 2048
+//
+// One filesystem block, which is also one kalloc page: this said 2048 while
+// the comment above it said one page, so half of every bounce buffer went
+// unused and every read did twice the work it needed to.
+//
+// The chunk size is not a copy-size tuning knob. Each trip through the loop
+// takes the volume sleeplock and builds a fresh B-tree cursor to find the
+// extent for that offset, so this is really "how many tree descents per byte
+// read". Measured on an 834 KB file: 417 descents at 2048, 209 at 4096.
+//
+// It must not exceed a page, because the buffer is a single kalloc().
+#define BOUNCE SCHFS_BLKSIZE
 
 static int
 schfs_vfs_read(struct inode *ip, int user_dst, uint64 dst, uint64 off, uint n)