denozord/strongswan-ext
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

forecast: Make BPF statements for metadata reads clearer

Browse Source 
The kernel ignores the width of these metadata loads, but since we read
32-bit values, this makes it a bit clearer and technically more correct.
This commit is contained in:
Tobias Brunner
2026-07-23 10:26:08 +02:00
parent 7e7c2805df
commit e1609d8159
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/libcharon/plugins/forecast/forecast_forwarder.c
+2 -2
View File
@@ -323,10 +323,10 @@ static bool attach_filter(int fd, uint32_t broadcast, uint32_t ifindex)
{
struct sock_filter filter_code[] = {
/* handle any marked packets (from clients) */
BPF_STMT(BPF_LD+BPF_B+BPF_ABS, SKF_AD_OFF+SKF_AD_MARK),
BPF_STMT(BPF_LD+BPF_W+BPF_ABS, SKF_AD_OFF+SKF_AD_MARK),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, 0, 0, 2),
/* and those from the internal interface */
BPF_STMT(BPF_LD+BPF_B+BPF_ABS, SKF_AD_OFF+SKF_AD_IFINDEX),
BPF_STMT(BPF_LD+BPF_W+BPF_ABS, SKF_AD_OFF+SKF_AD_IFINDEX),
BPF_JUMP(BPF_JMP+BPF_JEQ+BPF_K, ifindex, 0, 5),
/* destination address: is ... */
BPF_STMT(BPF_LD+BPF_W+BPF_ABS, offsetof(struct iphdr, daddr)),