Bug ID 2218341: Blobd process crashes when Fatal Signal Handler re-enters recursively

Last Modified: Oct 27, 2026

Affected Product(s):
BIG-IP NEXT CNF None(all modules)

Known Affected Versions:
1.4.0, 1.4.1

Fixed In:
1.4.2

Opened: Jan 29, 2026

Severity: 3-Major

Symptoms

Blobd crashes unexpectedly, generating a core file that shows infinite recursive frames in blobd_fatal_signal_handler; the recursion never terminates, eventually exhausting the stack and triggering a SIGSEGV.

Impact

Blobd crashes and fails to produce a usable core dump, preventing proper post-mortem diagnosis. The crash disrupts traffic processing that depends on blobd until you restart the process.

Conditions

When blobd receives a fatal signal (such as SIGSEGV) at runtime, it runs a handler configured with SA_NODEFER and with SIGSEGV, SIGFPE, and SIGILL removed from the blocked mask. Because the handler then calls raise(signum) to re-raise the signal for a default core dump, the kernel immediately re-delivers the signal to the active handler, which re-enters recursively until it exhausts the stack.

Workaround

N/A

Fix Information

Fixed the signal handler disposition to SIG_DFL before calling raise(signum). This ensures that when the signal fires again, the OS takes the default action (terminate and generate a core dump) instead of re-entering the custom handler, eliminating the infinite recursion.

Behavior Change

Guides & references

K10134038: F5 Bug Tracker Filter Names and Tips