From bca981236a29b642e05d60f0f7e376b50a857a63 Mon Sep 17 00:00:00 2001
From: Dawei Shen <shendawei.sdw@antgroup.com>
Date: Thu, 10 Oct 2024 00:33:46 +0800
Subject: [PATCH] sentry: use read(2) host syscall to perform read on
 disk-backed MemoryFiles

The mf.MapInternal()+safemem.CopySeq() approach used right now incurs a lot of
page faults without page population. Page-by-page faults incurs a lot of context
switching. This commits uses read(2) host syscall instead, which makes one context
switch and faults all the pages that are touched during the read.

Signed-off-by: Dawei Shen <shendawei.sdw@antgroup.com>
---
 pkg/sentry/fsimpl/tmpfs/regular_file.go | 31 +++++++++++++++++++------
 1 file changed, 24 insertions(+), 7 deletions(-)

diff --git a/pkg/sentry/fsimpl/tmpfs/regular_file.go b/pkg/sentry/fsimpl/tmpfs/regular_file.go
index 24c46fbaf9..f80551017f 100644
--- a/pkg/sentry/fsimpl/tmpfs/regular_file.go
+++ b/pkg/sentry/fsimpl/tmpfs/regular_file.go
@@ -655,14 +655,8 @@ func (rw *regularFileReadWriter) ReadToBlocks(dsts safemem.BlockSeq) (uint64, er
 		mr := memmap.MappableRange{uint64(rw.off), uint64(end)}
 		switch {
 		case seg.Ok():
-			// Get internal mappings.
-			ims, err := rw.file.inode.fs.mf.MapInternal(seg.FileRangeOf(seg.Range().Intersect(mr)), hostarch.Read)
-			if err != nil {
-				return done, err
-			}
+			n, err := rw.readFromMF(seg.FileRangeOf(seg.Range().Intersect(mr)), dsts)
 
-			// Copy from internal mappings.
-			n, err := safemem.CopySeq(dsts, ims)
 			done += n
 			rw.off += uint64(n)
 			dsts = dsts.DropFirst64(n)
@@ -812,6 +806,29 @@ exitLoop:
 	return done, retErr
 }
 
+func (rw *regularFileReadWriter) readFromMF(fr memmap.FileRange, dsts safemem.BlockSeq) (uint64, error) {
+	if rw.file.inode.fs.mf.IsDiskBacked() {
+		// Disk-backed files are not prepopulated. The safemem.CopySeq() approach
+		// used below incurs a lot of page faults without page prepopulation, which
+		// causes a lot of context switching. Use read(2) host syscall instead,
+		// which makes one context switch and faults all the pages that are touched
+		// during the read.
+		return hostfd.Preadv2(
+			int32(rw.file.inode.fs.mf.FD()), // fd
+			dsts.TakeFirst64(fr.Length()),   // dsts
+			int64(fr.Start),                 // offset
+			0,                               // flags
+		)
+	}
+	// Get internal mappings.
+	ims, err := rw.file.inode.fs.mf.MapInternal(fr, hostarch.Read)
+	if err != nil {
+		return 0, err
+	}
+	// Copy from internal mappings.
+	return safemem.CopySeq(dsts, ims)
+}
+
 func (rw *regularFileReadWriter) writeToMF(fr memmap.FileRange, srcs safemem.BlockSeq) (uint64, error) {
 	if rw.file.inode.fs.mf.IsDiskBacked() {
 		// Disk-backed files are not prepopulated. The safemem.CopySeq() approach