1# Note: This will actually execute /apex/com.android.tethering/bin/netbpfload 2# by virtue of 'service bpfloader' being overridden by the apex shipped .rc 3# Warning: most of the below settings are irrelevant unless the apex is missing. 4service bpfloader /system/bin/false 5 # netbpfload will do network bpf loading, then execute /system/bin/bpfloader 6 #! capabilities CHOWN SYS_ADMIN NET_ADMIN 7 # The following group memberships are a workaround for lack of DAC_OVERRIDE 8 # and allow us to open (among other things) files that we created and are 9 # no longer root owned (due to CHOWN) but still have group read access to 10 # one of the following groups. This is not perfect, but a more correct 11 # solution requires significantly more effort to implement. 12 #! group root graphics network_stack net_admin net_bw_acct net_bw_stats net_raw system 13 user root 14 # 15 # Set RLIMIT_MEMLOCK to 1GiB for bpfloader 16 # 17 # Actually only 8MiB would be needed if bpfloader ran as its own uid. 18 # 19 # However, while the rlimit is per-thread, the accounting is system wide. 20 # So, for example, if the graphics stack has already allocated 10MiB of 21 # memlock data before bpfloader even gets a chance to run, it would fail 22 # if its memlock rlimit is only 8MiB - since there would be none left for it. 23 # 24 # bpfloader succeeding is critical to system health, since a failure will 25 # cause netd crashloop and thus system server crashloop... and the only 26 # recovery is a full kernel reboot. 27 # 28 # We've had issues where devices would sometimes (rarely) boot into 29 # a crashloop because bpfloader would occasionally lose a boot time 30 # race against the graphics stack's boot time locked memory allocation. 31 # 32 # Thus bpfloader's memlock has to be 8MB higher then the locked memory 33 # consumption of the root uid anywhere else in the system... 34 # But we don't know what that is for all possible devices... 35 # 36 # Ideally, we'd simply grant bpfloader the IPC_LOCK capability and it 37 # would simply ignore it's memlock rlimit... but it turns that this 38 # capability is not even checked by the kernel's bpf system call. 39 # 40 # As such we simply use 1GiB as a reasonable approximation of infinity. 41 # 42 #! rlimit memlock 1073741824 1073741824 43 oneshot 44 # 45 # How to debug bootloops caused by 'bpfloader-failed'. 46 # 47 # 1. On some lower RAM devices (like wembley) you may need to first enable developer mode 48 # (from the Settings app UI), and change the developer option "Logger buffer sizes" 49 # from the default (wembley: 64kB) to the maximum (1M) per log buffer. 50 # Otherwise buffer will overflow before you manage to dump it and you'll get useless logs. 51 # 52 # 2. comment out 'reboot_on_failure reboot,bpfloader-failed' below 53 # 3. rebuild/reflash/reboot 54 # 4. as the device is booting up capture bpfloader logs via: 55 # adb logcat -s 'bpfloader:*' 'LibBpfLoader:*' 'NetBpfLoad:*' 'NetBpfLoader:*' 56 # 57 # something like: 58 # $ adb reboot; sleep 1; adb wait-for-device; adb root; sleep 1; adb wait-for-device; adb logcat -s 'bpfloader:*' 'LibBpfLoader:*' 'NetBpfLoad:*' 'NetBpfLoader:*' 59 # will take care of capturing logs as early as possible 60 # 61 # 5. look through the logs from the kernel's bpf verifier that bpfloader dumps out, 62 # it usually makes sense to search back from the end and find the particular 63 # bpf verifier failure that caused bpfloader to terminate early with an error code. 64 # This will probably be something along the lines of 'too many jumps' or 65 # 'cannot prove return value is 0 or 1' or 'unsupported / unknown operation / helper', 66 # 'invalid bpf_context access', etc. 67 # 68 reboot_on_failure reboot,netbpfload-missing 69 updatable 70