pageserver: do vectored read on each dio-aligned section once (#8763)

Part of #8130, closes #8719.

## Problem

Currently, vectored blob io only coalesce blocks if they are immediately
adjacent to each other. When we switch to Direct IO, we need a way to
coalesce blobs that are within the dio-aligned boundary but has gap
between them.

## Summary of changes

- Introduces a `VectoredReadCoalesceMode` for `VectoredReadPlanner` and
`StreamingVectoredReadPlanner` which has two modes:
  - `AdjacentOnly` (current implementation)
  - `Chunked(<alignment requirement>)`
- New `ChunkedVectorBuilder` that considers batching `dio-align`-sized
read, the start and end of the vectored read will respect
`stx_dio_offset_align` / `stx_dio_mem_align` (`vectored_read.start` and
`vectored_read.blobs_at.first().start_offset` will be two different
value).
- Since we break the assumption that blobs within single `VectoredRead`
are next to each other (implicit end offset), we start to store blob end
offsets in the `VectoredRead`.
- Adapted existing tests to run in both `VectoredReadCoalesceMode`.
- The io alignment can also be live configured at runtime.

Signed-off-by: Yuchen Liang <yuchen@neon.tech>

pageserver/src/virtual_file.rs

@@ -10,6 +10,7 @@
 //! This is similar to PostgreSQL's virtual file descriptor facility in
 //! src/backend/storage/file/fd.c
 //!
+use crate::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT;
 use crate::context::RequestContext;
 use crate::metrics::{StorageIoOperation, STORAGE_IO_SIZE, STORAGE_IO_TIME_METRIC};
 
@@ -1140,10 +1141,13 @@ impl OpenFiles {
 /// server startup.
 ///
 #[cfg(not(test))]
-pub fn init(num_slots: usize, engine: IoEngineKind) {
+pub fn init(num_slots: usize, engine: IoEngineKind, io_buffer_alignment: usize) {
     if OPEN_FILES.set(OpenFiles::new(num_slots)).is_err() {
         panic!("virtual_file::init called twice");
     }
+    if set_io_buffer_alignment(io_buffer_alignment).is_err() {
+        panic!("IO buffer alignment ({io_buffer_alignment}) is not a power of two");
+    }
     io_engine::init(engine);
     crate::metrics::virtual_file_descriptor_cache::SIZE_MAX.set(num_slots as u64);
 }
@@ -1167,6 +1171,61 @@ fn get_open_files() -> &'static OpenFiles {
     }
 }
 
+static IO_BUFFER_ALIGNMENT: AtomicUsize = AtomicUsize::new(DEFAULT_IO_BUFFER_ALIGNMENT);
+
+/// Returns true if `x` is zero or a power of two.
+fn is_zero_or_power_of_two(x: usize) -> bool {
+    (x == 0) || ((x & (x - 1)) == 0)
+}
+
+#[allow(unused)]
+pub(crate) fn set_io_buffer_alignment(align: usize) -> Result<(), usize> {
+    if is_zero_or_power_of_two(align) {
+        IO_BUFFER_ALIGNMENT.store(align, std::sync::atomic::Ordering::Relaxed);
+        Ok(())
+    } else {
+        Err(align)
+    }
+}
+
+/// Gets the io buffer alignment requirement. Returns 0 if there is no requirement specified.
+///
+/// This function should be used to check the raw config value.
+pub(crate) fn get_io_buffer_alignment_raw() -> usize {
+    let align = IO_BUFFER_ALIGNMENT.load(std::sync::atomic::Ordering::Relaxed);
+
+    if cfg!(test) {
+        let env_var_name = "NEON_PAGESERVER_UNIT_TEST_IO_BUFFER_ALIGNMENT";
+        if align == DEFAULT_IO_BUFFER_ALIGNMENT {
+            if let Some(test_align) = utils::env::var(env_var_name) {
+                if is_zero_or_power_of_two(test_align) {
+                    test_align
+                } else {
+                    panic!("IO buffer alignment ({test_align}) is not a power of two");
+                }
+            } else {
+                crate::config::defaults::DEFAULT_IO_BUFFER_ALIGNMENT
+            }
+        } else {
+            align
+        }
+    } else {
+        align
+    }
+}
+
+/// Gets the io buffer alignment requirement. Returns 1 if the alignment config is set to zero.
+///
+/// This function should be used for getting the actual alignment value to use.
+pub(crate) fn get_io_buffer_alignment() -> usize {
+    let align = get_io_buffer_alignment_raw();
+    if align == DEFAULT_IO_BUFFER_ALIGNMENT {
+        1
+    } else {
+        align
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use crate::context::DownloadBehavior;