From 8646b5d6e6019e95063a49fb3105833b07da5ebc Mon Sep 17 00:00:00 2001
From: Daan De Meyer <daan.j.demeyer@gmail.com>
Date: Fri, 14 Jan 2022 16:41:28 +0000
Subject: [PATCH] shared: Copy holes in sparse files in copy_bytes_full()

Previously, all holes in sparse files copied with copy_bytes_full()
would be expanded in the target file. Now, we correctly detect holes
in the input file and we replicate them in the target file.
---
 src/shared/copy.c | 41 +++++++++++++++++++++++++++++++++++++++++
 src/shared/copy.h |  1 +
 2 files changed, 42 insertions(+)

diff --git a/src/shared/copy.c b/src/shared/copy.c
index 1ace40424e5..6877b040a3a 100644
--- a/src/shared/copy.c
+++ b/src/shared/copy.c
@@ -202,6 +202,47 @@ int copy_bytes_full(
                 if (max_bytes != UINT64_MAX && m > max_bytes)
                         m = max_bytes;
 
+                if (copy_flags & COPY_HOLES) {
+                        off_t c, e;
+
+                        c = lseek(fdf, 0, SEEK_CUR);
+                        if (c < 0)
+                                return -errno;
+
+                        /* To see if we're in a hole, we search for the next data offset. */
+                        e = lseek(fdf, c, SEEK_DATA);
+                        if (e < 0 && errno == ENXIO) {
+                                /* If errno == ENXIO, that means we've reached the final hole of the file and
+                                * that hole isn't followed by more data. */
+                                e = lseek(fdf, 0, SEEK_END);
+                                if (e < 0)
+                                        return -errno;
+                        } else if (e < 0)
+                                return -errno;
+
+                        /* If we're in a hole (current offset is not a data offset), create a hole of the same size
+                        * in the target file. */
+                        if (e > c && lseek(fdt, e - c, SEEK_CUR) < 0)
+                                return -errno;
+
+                        c = e; /* Set c to the start of the data segment. */
+
+                        /* After copying a potential hole, find the end of the data segment by looking for the next
+                        * hole. If we get ENXIO, we're at EOF. */
+                        e = lseek(fdf, c, SEEK_HOLE);
+                        if (e < 0 && errno == ENXIO)
+                                break;
+                        else if (e < 0)
+                                return -errno;
+
+                        /* SEEK_HOLE modifies the file offset so we need to move back to the initial offset. */
+                        if (lseek(fdf, c, SEEK_SET) < 0)
+                                return -errno;
+
+                        /* Make sure we're not copying more than the current data segment. */
+                        m = MIN(m, (size_t) e - c);
+                }
+
                 /* First try copy_file_range(), unless we already tried */
                 if (try_cfr) {
                         n = try_copy_file_range(fdf, NULL, fdt, NULL, m, 0u);
diff --git a/src/shared/copy.h b/src/shared/copy.h
index a7b45b4fbf4..d755916bd98 100644
--- a/src/shared/copy.h
+++ b/src/shared/copy.h
@@ -24,6 +24,7 @@ typedef enum CopyFlags {
         COPY_FSYNC_FULL  = 1 << 11, /* fsync_full() after we are done */
         COPY_SYNCFS      = 1 << 12, /* syncfs() the *top-level* dir after we are done */
         COPY_ALL_XATTRS  = 1 << 13, /* Preserve all xattrs when copying, not just those in the user namespace */
+        COPY_HOLES       = 1 << 14, /* Copy holes */
 } CopyFlags;
 
 typedef int (*copy_progress_bytes_t)(uint64_t n_bytes, void *userdata);