$ NBDKIT_GDB=1 ./nbdkit -fv --filter=tls-fallback memory 1m (gdb) run In a second window: $ nbdinfo nbd://localhost nbdkit crashes: Thread 2 "nbdkit" received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7ffff7584640 (LWP 1129108)] 0x000000000040625d in backend_block_size (c=0x0, minimum=0x7ffff7583944, preferred=0x7ffff7583940, maximum=0x7ffff758393c) at backend.c:428 428 struct backend *b = c->b; It turns out this is because the filter + plugin context chain is incorrectly constructed. There is only a single layer (for the tls-fallback filter), and this has c->c_next == NULL. We end up calling backend_block_size (c=NULL, ...) which causes the crash above. This is because the filter doesn't always call the next open function. Changing the filter so it always calls it fixes the issue. --- filters/tls-fallback/tls-fallback.c | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) diff --git a/filters/tls-fallback/tls-fallback.c b/filters/tls-fallback/tls-fallback.c index fab9e58b..4ea48d09 100644 --- a/filters/tls-fallback/tls-fallback.c +++ b/filters/tls-fallback/tls-fallback.c @@ -106,13 +106,10 @@ tls_fallback_open (nbdkit_next_open *next, nbdkit_context *nxdata, int readonly, const char *exportname, int is_tls) { - /* We do NOT want to call next() when insecure, because we don't - * know how long it will take. - */ - if (!is_tls) - return &message; /* See NOT_TLS for this choice of handle */ if (next (nxdata, readonly, exportname) == -1) return NULL; + if (!is_tls) + return &message; /* See NOT_TLS for this choice of handle */ return NBDKIT_HANDLE_NOT_NEEDED; } -- 2.35.1