isisd: fix crash when obtaining the next hop to calculate LFA on LAN …

…links

When a neighbor connection is disconnected, it may trigger LSP re-generation as a timer task, but this process may be delayed. As a result, the list of neighbors in area->adjacency_list may be inconsistent with the neighbors in lsp->tlvs->oldstyle_reach/extended_reach. For example, the area->adjacency_list may lack certain neighbors even though they are present in the LSP. When computing SPF, the call to isis_spf_build_adj_list() generates the spftree->sadj_list, which reflects the real neighbors in the area->adjacency_list. However, in the case of LAN links, spftree->sadj_list may include additional pseudo neighbors.
The pre-loading of tents through the call to isis_spf_preload_tent involves two steps:
1. isis_spf_process_lsp() is called to generate real neighbor vertices based on the root LSP and pseudo LSP.
2. isis_spf_add_local() is called to add corresponding next hops to the vertex->Adj_N list for the real neighbor vertices.
In the case of LAN links, the absence of corresponding real neighbors in the spftree->sadj_list prevents the execution of the second step. Consequently, the vertex->Adj_N list for the real neighbor vertices lacks corresponding next hops. This leads to a null pointer access when isis_lfa_compute() is called to calculate LFA.
As for P2P links, since there are no pseudo neighbors, only the second step is executed, which does not create real neighbor vertices and therefore does not encounter this issue.
The backtrace is as follows:
(gdb) bt
#0  0x00007fd065277fe1 in raise () from /lib/x86_64-linux-gnu/libpthread.so.0
#1  0x00007fd065398972 in core_handler (signo=11, siginfo=0x7ffc5c0636b0, context=0x7ffc5c063580) at ../lib/sigevent.c:261
#2  <signal handler called>
#3  0x00005564d82f8408 in isis_lfa_compute (area=0x5564d8b143f0, circuit=0x5564d8b21d10, spftree=0x5564d8b06bf0, resource=0x7ffc5c064410) at ../isisd/isis_lfa.c:2134
FRRouting#4  0x00005564d82f8d78 in isis_spf_run_lfa (area=0x5564d8b143f0, spftree=0x5564d8b06bf0) at ../isisd/isis_lfa.c:2344
FRRouting#5  0x00005564d8315964 in isis_run_spf_with_protection (area=0x5564d8b143f0, spftree=0x5564d8b06bf0) at ../isisd/isis_spf.c:1827
FRRouting#6  0x00005564d8315c15 in isis_run_spf_cb (thread=0x7ffc5c064590) at ../isisd/isis_spf.c:1889
FRRouting#7  0x00007fd0653b1f04 in thread_call (thread=0x7ffc5c064590) at ../lib/thread.c:1990
FRRouting#8  0x00007fd06534a97b in frr_run (master=0x5564d88103c0) at ../lib/libfrr.c:1198
FRRouting#9  0x00005564d82e7d5d in main (argc=5, argv=0x7ffc5c0647b8, envp=0x7ffc5c0647e8) at ../isisd/isis_main.c:273
(gdb) f 3
#3  0x00005564d82f8408 in isis_lfa_compute (area=0x5564d8b143f0, circuit=0x5564d8b21d10, spftree=0x5564d8b06bf0, resource=0x7ffc5c064410) at ../isisd/isis_lfa.c:2134
2134    ../isisd/isis_lfa.c: No such file or directory.
(gdb) p vadj_primary
$1 = (struct isis_vertex_adj *) 0x0
(gdb) p vertex->Adj_N->head
$2 = (struct listnode *) 0x0
(gdb) p (struct isis_vertex *)spftree->paths->l.list->head->next->next->next->next->data
$8 = (struct isis_vertex *) 0x5564d8b5b240
(gdb) p $8->type
$9 = VTYPE_NONPSEUDO_TE_IS
(gdb) p $8->N.id
$10 = "\000\000\000\000\000\002"
(gdb) p $8->Adj_N->count
$11 = 0
(gdb) p (struct isis_vertex *)spftree->paths->l.list->head->next->next->next->next->next->data
$12 = (struct isis_vertex *) 0x5564d8b73dd0
(gdb) p $12->type
$13 = VTYPE_NONPSEUDO_TE_IS
(gdb) p $12->N.id
$14 = "\000\000\000\000\000\003"
(gdb) p $12->Adj_N->count
$15 = 0
(gdb) p area->adjacency_list->count
$16 = 0
The backtrace provided above pertains to version 8.5.4, but it seems that the same issue exists in the code of the master branch as well.
The scenario where a vertex has no next hop is normal. For example, the "clear isis neighbor" command invokes isis_vertex_adj_del() to delete the next hop of a vertex. Upon reviewing all the instances where the vertex->Adj_N list is used, I found that only isis_lfa_compute() lacks a null check. Therefore, I believe that modifying this part will be sufficient. Additionally, the vertex->parents list for IP vertices is guaranteed not to be empty.
Test scenario:
Setting up LFA for LAN links and executing the "clear isis neighbor" command easily reproduces the issue.

Signed-off-by: zhou-run <zhou.run@h3c.com>
(cherry picked from commit a970bb5)

isisd/isis_lfa.c

@@ -2126,9 +2126,16 @@ void isis_lfa_compute(struct isis_area *area, struct isis_circuit *circuit,
 		}
 
 		vadj_primary = listnode_head(vertex->Adj_N);
+		if (!vadj_primary) {
+			if (IS_DEBUG_LFA)
+				zlog_debug(
+					"ISIS-LFA: skipping computing LFAs due to no adjacencies");
+			continue;
+		}
 		sadj_primary = vadj_primary->sadj;
 
 		parent_vertex = listnode_head(vertex->parents);
+		assert(parent_vertex);
 		prefix_metric = vertex->d_N - parent_vertex->d_N;
 
 		/*