|
24 | 24 | */ |
25 | 25 | #define MDA_GID_UID_FILTER 1 |
26 | 26 |
|
27 | | -/* |
28 | | - * openEuler-23.03 is an innovative version of openEuler, in the early time, we |
29 | | - * developed kmesh based on openEuler-23.03, and the implementation of kmesh |
30 | | - * was related to the openEuler-23.03 kernel. Now, the general implementation |
31 | | - * of kmesh differs from the previous openEuler-23.03 version, so we need to |
32 | | - * use this macro to distinguish these differences. |
33 | | - * The main differences between the general implementation of kmesh and the |
34 | | - * openEuler-23.03 version are as follows: |
35 | | - * 1. Use replylong parameter instead of directly modifying the remote IP and Port; |
36 | | - * 2. Use bpf__strncmp instead of bpf_strncmp for string comparison; |
37 | | - * 3. Fix Port shift bug on openEuler-23.03.In the kernel network protocol |
38 | | - * stack, the port is stored in u16, but in the bpf network module, the port |
39 | | - * is stored in u32. Therefore, after the endian conversion, the 16-bit port |
40 | | - * needs to be obtained from the 32-bit data structure. |
41 | | - * You need to find the position of the valid 16 bits. Generally, after the |
42 | | - * port is extended from 16 bits to 32 bits, the port is in the upper 16 |
43 | | - * bits after the endian conversion. Therefore, you need to offset the port |
44 | | - * before using the u16 RX port. In some specific kernels, the port stored |
45 | | - * in sockops is in the lower 16 bits and does not need to be offset. |
46 | | - */ |
47 | | -#define OE_23_03 0 |
48 | | - |
49 | 27 | /* |
50 | 28 | * in kernel 6.x version, add the new iter type ITER_UBUF, and we need add code |
51 | 29 | * for the corresponding scenarios. |
|
0 commit comments