mgmtd: Fixed bug,In daemon mode, this stream points to /dev/null and is read-only, which caused an error. As a result, the buffer vty->buf was closed, leading to the crash.

[huchaogithup]

The initial cause of the crash was that vty->buf was null. The crash occurred after executing the function fgets(vty->buf, VTY_BUFSIZ, fp). The root cause was traced back to mgmtd loading /etc/frr/zebra.conf, during which the cmd_exit function was called and attempted to write something to vty->wfd (standard error stream). In daemon mode, this stream points to /dev/null and is read-only, which caused an error. As a result, the buffer vty->buf was closed, leading to the crash.

clashlog:
MGMTD: Received signal 11 at 1745121262 (si_addr 0x0, PC 0x7b13a09988f9); aborting...
MGMTD: zlog_signal+0xf9 7b13a0cd2ec9 7ffce3846830 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: core_handler+0xb5 7b13a0d13715 7ffce3846970 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: __sigaction+0x50 7b13a0845330 7ffce3846ac0 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: ---- signal ----
MGMTD: __nss_database_lookup+0x10919 7b13a09988f9 7ffce38477c8 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: _IO_getline_info+0x124 7b13a0886ff4 7ffce38477d0 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: fgets+0xa4 7b13a0885bd4 7ffce3847830 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: config_from_file+0x45 7b13a0c9f6d5 7ffce3847860 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: mgmt_vty_read_configs+0x16c 7b13a0d3263c 7ffce38478a0 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: event_call+0xae 7b13a0d269be 7ffce3848910 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: frr_run+0xc8 7b13a0cc9bd8 7ffce3848a40 /usr/lib/frr/libfrr.so.0 (mapped at 0x7b13a0c00000)
MGMTD: main+0x157 597895ef2b57 7ffce3848b60 /usr/lib/frr/mgmtd (mapped at 0x597895edd000)
MGMTD: __libc_init_first+0x8a 7b13a082a1ca 7ffce3848ba0 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: __libc_start_main+0x8b 7b13a082a28b 7ffce3848c40 /lib/x86_64-linux-gnu/libc.so.6 (mapped at 0x7b13a0800000)
MGMTD: _start+0x25 597895ef3045 7ffce3848ca0 /usr/lib/frr/mgmtd (mapped at 0x597895edd000)
MGMTD: in thread mgmt_config_read_in scheduled from mgmtd/mgmt_vty.c:655 mgmt_vty_init()

#18140

[donaldsharp]

Please put the analysis into the commit message itself. Without it being in the commit message this is going no-where

[huchaogithup]

Please put the analysis into the commit message itself. Without it being in the commit message this is going no-where

Is this modification okay?

[donaldsharp]

I do not think that this is the correct approach to fixing this problem. Opening /dev/null as RW is not something I think we should be doing. @eqvinox opinions here?

[huchaogithup]

I do not think that this is the correct approach to fixing this problem. Opening /dev/null as RW is not something I think we should be doing. @eqvinox opinions here?

I think I more or less understand what you mean now. In the case of systemctl start frr or running docker run -d quay.io/frrouting/frr:10.2.1, it's true that only the standard input stream is kept as /dev/null, while the standard output and error streams are typically replaced by pipes.

or

I might have encountered the issue because I manually started /usr/lib/frr/mgmtd -d -F traditional -A 127.0.0.1 or sudo containerlab deploy -t frr.yaml while trying to troubleshoot a different problem.

Because they both use stty.

[github-actions]

This PR is stale because it has been open 180 days with no activity. Comment or remove the autoclose label in order to avoid having this PR closed.