考虑下面一种场景：
编译刷机包时，是按照2GB的eMMC制作gpt分区格式的镜像，但是最终使用的eMMC的大小是4GB，如何让最后一个分区占满后面的2GB空间，并且保证分区文件系统原有的内容不丢失呢？
下面在本地模拟一下：
先创建两个大小都为100MB的空文件origin.disk和empty.disk，其中origin.disk所模拟的就是上面2GB的eMMC，然后对其进行分区，并且创建文件系统，在最后一个分区里新建一个文件，用于验证内容是否会丢失。接着将origin.disk和empty.disk进行拼接得到new.disk，此时这个new.disk所模拟的就是上面的4GB大小的eMMC。然后对这个new.disk进行处理，使最后一个分区占满新增加的空间，最后会输出扩展后的分区中的文件的内容，看是否跟原有文件相同。
下面是测试脚本：
#!/bin/bash
# clear
rm -rf origin.disk
rm -rf new.disk
###########################################################################################################
# Create origin.disk: 100MB
dd if=/dev/zero of=./origin.disk bs=1024 count=102400
dd if=/dev/zero of=./empty.disk bs=1024 count=102400
# Create GPT and partition
sgdisk -n 0:0:+30M -c 1:"1st" ./origin.disk
sgdisk -n 0:0:+20M -c 2:"2nd" ./origin.disk
sgdisk -n 0:0:0 -c 3:"3rd" ./origin.disk
# losetup
sudo losetup /dev/loop1 ./origin.disk
sudo partprobe /dev/loop1
# Create fs
echo "Create file system ..."
sudo mkfs.ext4 /dev/loop1p1
sudo mkfs.ext4 /dev/loop1p2
sudo mkfs.ext4 /dev/loop1p3
# Create a file
mkdir -p p3
sudo mount -t ext4 /dev/loop1p3 ./p3
sudo chmod 777 p3
sudo echo "Hello world" > ./p3/demo.txt
sync
df -h | tail -n 1
sudo umount ./p3
# Show Partition
sudo sgdisk -p ./origin.disk
# or
sudo sgdisk -p /dev/loop1
# losetup deattach
sudo losetup -d /dev/loop1
###########################################################################################################
# Create New disk
echo "Create new disk ..."
cat origin.disk empty.disk > new.disk
# losetup
echo "losetup ..."
sudo losetup /dev/loop1 ./new.disk
sudo partprobe /dev/loop1
# Move backup gpt to the end
echo "Move back up gpt partition ..."
sudo sgdisk -e /dev/loop1
# Delete last part
echo "Delete last part ..."
sudo sgdisk -d 3 /dev/loop1
# Recreate last part
echo "Recreate last part ..."
sudo sgdisk -n 0:0:0 -c 3:"New 3rd" /dev/loop1
sudo partprobe /dev/loop1
# Check
echo "e2fsck -f /dev/loop1p3 ..."
sudo e2fsck -f /dev/loop1p3
# Resize
echo "resize2fs /dev/loop1p3 ..."
sudo resize2fs /dev/loop1p3
# mount
echo "mount -t ext4 /dev/loop1p3 ./p3 ..."
sudo mount -t ext4 /dev/loop1p3 ./p3
df -h | tail -n 1
cat ./p3/demo.txt
# umount
echo "umount ..."
sudo umount ./p3
echo "losetup detach ..."
sudo losetup -d /dev/loop1
# end
sudo sgdisk -p ./new.disk
echo "End"
下面是输出log：
1 102400+0 records in
2 102400+0 records out
3 104857600 bytes (105 MB, 100 MiB) copied, 0.299038 s, 351 MB/s
4 102400+0 records in
5 102400+0 records out
6 104857600 bytes (105 MB, 100 MiB) copied, 0.304469 s, 344 MB/s
7 Creating new GPT entries.
8 Setting name!
9 partNum is 0
10 REALLY setting name!
11 Warning: The kernel is still using the old partition table.
12 The new table will be used at the next reboot or after you
13 run partprobe(8) or kpartx(8)
14 The operation has completed successfully.
15 Setting name!
16 partNum is 1
17 REALLY setting name!
18 Warning: The kernel is still using the old partition table.
19 The new table will be used at the next reboot or after you
20 run partprobe(8) or kpartx(8)
21 The operation has completed successfully.
22 Setting name!
23 partNum is 2
24 REALLY setting name!
25 Warning: The kernel is still using the old partition table.
26 The new table will be used at the next reboot or after you
27 run partprobe(8) or kpartx(8)
28 The operation has completed successfully.
29 Create file system ...
30 mke2fs 1.42.13 (17-May-2015)
31 Discarding device blocks: done
32 Creating filesystem with 30720 1k blocks and 7680 inodes
33 Filesystem UUID: b71db7b0-1b0b-4b54-a3b3-4d642d8db8c6
34 Superblock backups stored on blocks:
35 8193, 24577
36
37 Allocating group tables: done
38 Writing inode tables: done
39 Creating journal (1024 blocks): done
40 Writing superblocks and filesystem accounting information: done
41
42 mke2fs 1.42.13 (17-May-2015)
43 Discarding device blocks: done
44 Creating filesystem with 20480 1k blocks and 5136 inodes
45 Filesystem UUID: da901017-8edc-4509-a535-3ce31bbf3af4
46 Superblock backups stored on blocks:
47 8193
48
49 Allocating group tables: done
50 Writing inode tables: done
51 Creating journal (1024 blocks): done
52 Writing superblocks and filesystem accounting information: done
53
54 mke2fs 1.42.13 (17-May-2015)
55 Discarding device blocks: done
56 Creating filesystem with 50156 1k blocks and 12544 inodes
57 Filesystem UUID: 149c00b8-8f2d-49ac-8307-a341e541d738
58 Superblock backups stored on blocks:
59 8193, 24577, 40961
60
61 Allocating group tables: done
62 Writing inode tables: done
63 Creating journal (4096 blocks): done
64 Writing superblocks and filesystem accounting information: done
65
66 /dev/loop1p3 44M 795K 40M 2% /home/peng_dl/work/A15/resize/p3
67 Disk ./origin.disk: 204800 sectors, 100.0 MiB
68 Logical sector size: 512 bytes
69 Disk identifier (GUID): 20C58E58-BC5D-4432-9BB8-B007981C614D
70 Partition table holds up to 128 entries
71 First usable sector is 34, last usable sector is 204766
72 Partitions will be aligned on 2048-sector boundaries
73 Total free space is 2014 sectors (1007.0 KiB)
74
75 Number Start (sector) End (sector) Size Code Name
76 1 2048 63487 30.0 MiB 8300 1st
77 2 63488 104447 20.0 MiB 8300 2nd
78 3 104448 204766 49.0 MiB 8300 3rd
79 Disk /dev/loop1: 204800 sectors, 100.0 MiB
80 Logical sector size: 512 bytes
81 Disk identifier (GUID): 20C58E58-BC5D-4432-9BB8-B007981C614D
82 Partition table holds up to 128 entries
83 First usable sector is 34, last usable sector is 204766
84 Partitions will be aligned on 2048-sector boundaries
85 Total free space is 2014 sectors (1007.0 KiB)
86
87 Number Start (sector) End (sector) Size Code Name
88 1 2048 63487 30.0 MiB 8300 1st
89 2 63488 104447 20.0 MiB 8300 2nd
90 3 104448 204766 49.0 MiB 8300 3rd
91 Create new disk ...
92 losetup ...
93 Warning: Not all of the space available to /dev/loop1 appears to be used, you can fix the GPT to use all of the space (an extra 204800 blocks) or continue with the current setting?
94 Move back up gpt partition ...
95 Warning: The kernel is still using the old partition table.
96 The new table will be used at the next reboot or after you
97 run partprobe(8) or kpartx(8)
98 The operation has completed successfully.
99 Delete last part ...
100 Warning: The kernel is still using the old partition table.
101 The new table will be used at the next reboot or after you
102 run partprobe(8) or kpartx(8)
103 The operation has completed successfully.
104 Recreate last part ...
105 Setting name!
106 partNum is 2
107 REALLY setting name!
108 Warning: The kernel is still using the old partition table.
109 The new table will be used at the next reboot or after you
110 run partprobe(8) or kpartx(8)
111 The operation has completed successfully.
112 e2fsck -f /dev/loop1p3 ...
113 e2fsck 1.42.13 (17-May-2015)
114 Pass 1: Checking inodes, blocks, and sizes
115 Pass 2: Checking directory structure
116 Pass 3: Checking directory connectivity
117 Pass 4: Checking reference counts
118 Pass 5: Checking group summary information
119 /dev/loop1p3: 12/12544 files (0.0% non-contiguous), 6482/50156 blocks
120 resize2fs /dev/loop1p3 ...
121 resize2fs 1.42.13 (17-May-2015)
122 Resizing the filesystem on /dev/loop1p3 to 152556 (1k) blocks.
123 The filesystem on /dev/loop1p3 is now 152556 (1k) blocks long.
124
125 mount -t ext4 /dev/loop1p3 ./p3 ...
126 /dev/loop1p3 141M 1.2M 130M 1% /home/peng_dl/work/A15/resize/p3
127 Hello world
128 umount ...
129 losetup detach ...
130 Disk ./new.disk: 409600 sectors, 200.0 MiB
131 Logical sector size: 512 bytes
132 Disk identifier (GUID): 20C58E58-BC5D-4432-9BB8-B007981C614D
133 Partition table holds up to 128 entries
134 First usable sector is 34, last usable sector is 409566
135 Partitions will be aligned on 2048-sector boundaries
136 Total free space is 2014 sectors (1007.0 KiB)
137
138 Number Start (sector) End (sector) Size Code Name
139 1 2048 63487 30.0 MiB 8300 1st
140 2 63488 104447 20.0 MiB 8300 2nd
141 3 104448 409566 149.0 MiB 8300 New 3rd
142 End
完。
本文章为转载内容，我们尊重原作者对文章享有的著作权。如有内容错误或侵权问题，欢迎原作者联系我们进行内容更正或删除文章。
