How to patch and make diff in Linux

Production and Application of patches in Linux   Reference [http://blog.csdn.net/qupanpan110/archive/2010/12/17/6082315.aspx] This is because there are several common files to be modified during the U-Boot porting process. It is too troublesome to manually modify them each time. Patch creation can solve this problem. The collection of learning materials is relatively simple. The method 1 is similar to this kind of elementary problem. There are a lot of online materials. Google or Baidu searches for valuable materials and then selects valuable materials. The method 2 is to read man's online documents. To complete the collection, of course, we will eventually conduct an experiment on our own Linux system, make a summary, and digest and absorb your own things. To remove such a wrong idea, you must learn it all. You need to know that it is impossible to learn all at once. You can only learn the most common ones first. In the future, you will gradually enrich your learning experience and eventually reach a relatively high level. The principle is to apply what you have learned on a daily basis to promote learning. First, describe diff and patch. All options in the man online document will not be described here, which is unnecessary. In the 99% period, we only use a few options. Therefore, you must learn these options. 1. Diff -------------------- Name Diff-find differences between two files Synopsis Diff [Options] From-file to-file -------------------- Simply put, the diff function is used to compare the differences between the two files, and then record them, that is, the so-called diff patch. Syntax format: Diff [Option] source file (folder) Destination File (folder), that is, to patch the source file (folder) to make it into the destination file (folder ), the term is "Upgrade ". The following describes the three most common options: -R is a recursive option. With this option set, diff compares all the corresponding files in the source code directories of two different versions, including subdirectory files. -N option ensures that the patch file correctly processes the files that have been created or deleted. -U option creates patch files in a uniform format, which is more compact than the default format. 2. Patch ------------------ Name Patch-apply a diff file to an original Synopsis Patch [Options] [originalfile [patchfile] But usually just Patch-pnum <patchfile> ------------------ In short, patch refers to the use of patches made by diff to realize conversion between source files (CLIPS) and target files (CLIPS. This means you can have the source file (folder)> the target file (folder) or the target file (folder)> The source file (folder ). The following describes the most common options: -P0 option: Find the target file (folder) from the current directory) -The P1 option should ignore the first-level directory and start searching from the current directory. **************************************** ******************** Here is an example: --- Old/modules/pcitable mon Sep 27 11:03:56 1999 ++ New/modules/pcitable Tue Dec 19 20:05:41 2000 If you use the-P0 parameter, it means to find a folder named Old from the current directory, and find the pcitable file under modules under it to execute the patch operation. If the-P1 parameter is used, ignore the first-level directory (regardless of the old directory), find the modules folder from the current directory, and find pcitable under it. The premise is that the current directory must be the directory where modules is located. The diff patch file can be located anywhere, as long as the path of the diff patch file is specified. Of course, you can use relative paths or absolute paths. However, I usually use relative paths. **************************************** ******************** -E indicates that if an empty file is found, delete it. -The "r" option indicates that the "new" and "old" files in the patch file need to be replaced now (in fact, the new version is patched to make it an old version) The following examples are used to analyze and solve the problem. There are two types: patching a single file and patching multiple files in the folder. Environment: log on with an armlinux user under RedHat 9.0. The directory tree is as follows: | -- Bootloader | -- Debug | -- Images | -- Kernel | -- Program | -- Rootfiles | -- Software | -- Source | -- Sysapps | -- TMP '-- Tools Next, create a patch folder under the Program folder as an experiment, and then enter the patch folder. 1. patch a single file 1. Create test files test0 and test1 [Armlinux @ lqm Patch] $ cat> test0 <EOF > 111111 > 111111 > 111111 > EOF [Armlinux @ lqm Patch] $ more test0 111111 111111 111111 [Armlinux @ lqm Patch] $ cat> test1 <EOF > 222222 > 111111 > 222222 > 111111 > EOF [Armlinux @ lqm Patch] $ more test1 222222 111111 222222 111111 2. Use diff to create a patch test1.patch [Armlinux @ lqm Patch] $ diff-UN test0 test1> test1.patch [Note: The-r option is not required because of a single file. There is no relation between the option sequence, that is, it can be-UN or-nu .] [Armlinux @ lqm Patch] $ ls Test0 test1 test1.patch [Armlinux @ lqm Patch] $ more test1.patch **************************************** ******************** Patch file structure Patch Header The patch header contains two lines starting with ---/++ to indicate the files to be patched. --- The beginning indicates the old file, and the beginning of ++ indicates the new file. Multiple patches in a patch file A patch file may contain many sections starting with ---/++, each of which is used for a patch. Therefore, a patch file can contain many patches. Block The block is the part to be modified in the patch. It usually starts and ends with something that does not need to be modified. They are only used to indicate the location to be modified. They usually start with @ and end with another block or a new patch header. Block indent Blocks will be reduced into one column, and this column is used to indicate whether to add or delete this row. First column of the block + Indicates that this line is to be added. -Indicates that this row is to be deleted. There is no plus sign or minus sign, indicating that this is just referenced and does not need to be modified. **************************************** ******************** * ** The diff Command records the first creation time of these two files in the patch file, as shown below *** --- Test0 09:12:01. 000000000 + 0800 ++ Test1 09:13:09. 000000000 + 0800 @-1, 3 + 1, 4 @@ + 222222 111111 -111111 + 222222 111111 [Armlinux @ lqm Patch] $ patch-P0 <test1.patch Patching file test0 [Armlinux @ lqm Patch] $ ls Test0 test1 test1.patch [Armlinux @ lqm Patch] $ cat test0 222222 111111 222222 111111 3. patches can be removed to restore the old version. [Armlinux @ lqm Patch] $ patch-re-P0 <test1.patch Patching file test0 [Armlinux @ lqm Patch] $ ls Test0 test1 test1.patch [Armlinux @ lqm Patch] $ cat test0 111111 111111 111111 2. Patch multiple files 1. Create a test folder [Armlinux @ lqm Patch] $ mkdir prj0 [Armlinux @ lqm Patch] $ CP test0 prj0 [Armlinux @ lqm Patch] $ ls Prj0 test0 test1 test1.patch [Armlinux @ lqm Patch] $ CD prj0/ [Armlinux @ lqm prj0] $ ls Test0 [Armlinux @ lqm prj0] $ cat> prj0name <EOF > -------- > Prj0/prj0name > -------- > EOF [Armlinux @ lqm prj0] $ ls Prj0name test0 [Armlinux @ lqm prj0] $ cat prj0name -------- Prj0/prj0name -------- [Armlinux @ lqm prj0] $ CD .. [Armlinux @ lqm Patch] $ mkdir prj1 [Armlinux @ lqm Patch] $ CP test1 prj1 [Armlinux @ lqm Patch] $ CD prj1 [Armlinux @ lqm prj1] $ cat> prj1name <EOF > --------- > Prj1/prj1name > --------- > EOF [Armlinux @ lqm prj1] $ cat prj1name --------- Prj1/prj1name --------- [Armlinux @ lqm prj1] $ CD .. 2. Create a patch [Armlinux @ lqm Patch] $ diff-unr prj0 prj1> prj1.patch [Armlinux @ lqm Patch] $ more prj1.patch   Diff-unr prj0/prj0name prj1/prj0name --- Prj0/prj0name 09:25:11. 000000000 + 0800 ++ Prj1/prj0name 08:00:00. 000000000 + 0800 @-1, 3 +, 0 @@ --------- -Prj0/prj0name --------- Diff-unr prj0/prj1name prj1/prj1name --- Prj0/prj1name 08:00:00. 000000000 + 0800 ++ Prj1/prj1name 09:26:36. 000000000 + 0800 @-+ @@ + --------- + Prj1/prj1name + --------- Diff-unr prj0/test0 prj1/test0 --- Prj0/test0 09:23:53. 000000000 + 0800 ++ Prj1/test0 08:00:00. 000000000 + 0800 @-1, 3 +, 0 @@ -111111 -111111 -111111 Diff-unr prj0/test1 prj1/test1 --- Prj0/test1 1970-01-01 08:00:00. 000000000 + 0800 ++ Prj1/test1 09:26:00. 000000000 + 0800 @-+ @@ + 222222 + 111111 + 222222 + 111111   [Armlinux @ lqm Patch] $ ls Prj0 prj1 prj1.patch test0 test1 test1.patch [Armlinux @ lqm Patch] $ CP prj1.patch./prj0 [Armlinux @ lqm Patch] $ CD prj0 [Armlinux @ lqm prj0] $ patch-P1 <prj1.patch Patching file prj0name Patching file prj1name Patching file test0 Patching file test1 [Armlinux @ lqm prj0] $ ls Prj1name prj1.patch test1 [Armlinux @ lqm prj0] $ patch-r-P1 <prj1.patch Patching file prj0name Patching file prj1name Patching file test0 Patching file test1 [Armlinux @ lqm prj0] $ ls Prj0name prj1.patch test0 ------------------- Summary: Single file Diff-UN from-file to-File> to-file.patch Patch-P0 <to-file.patch Patch-re-P0 <to-file.patch Multiple files Diff-unr from-docu to-docu> to-docu.patch Patch-P1 <to-docu.patch Patch-r-P1 <to-docu.patch How to Create a patch in one directory source code in another directory: Patch-P1 <to-docu.patch-D source code directory------------------- Which subdirectory does the patch hit? Iii. Application Patch the kernel. When creating a cross-compilation toolchain, one step is to patch the kernel. At that time, I was not very familiar with it, but now it is clear. Refer to the previous article "establishing a development tool chain based on ARM + LINUX embedded development". 1. decompress the package because the released patch files are compressed using gzip. $ Gunzip ../setup-DIR/patch-2.4.21-rmk1.gz 2. Enter your kernel source code directory. $ Linux-2.4.21 CD 3. Patch $ Patch-P1 <.../../setup-DIR/patch-2.4.21-rmk1 After the patch is installed, check whether the file is rejected, that is, check the existence of the. rej file. Run the following command: $ Find.-Name *. rej If it is found, it will be output to the standard output terminal, the default screen. Of course, you can also use redirection to output to a specified file, such as reject. $ Fine.-Name *. rej> reject Then you can view the reject content.