[Bug] Kernel GPF in Claude Code process during memory management syscall

Bug Description
kernel GPF

Oops: general protection fault, non-canonical address 0xffff3dba803cc3b8
CPU: 6 PID: 45492 Comm: claude
RIP: __bitmap_and
native_flush_tlb_multi → flush_tlb_mm_range → tlb_finish_mmu
→ do_madvise → __x64_sys_madvise (ORIG_RAX 0x1c = madvise)

• The faulting process was claude (PID 45492) — the Claude Code CLI I run in — calling a totally routine madvise(addr, 16 KiB, MADV_DONTNEED)
• The kernel handled it with a multi-CPU TLB shootdown, and while ANDing CPU bitmasks it dereferenced a corrupted, non-canonical pointer (0xffff3dba… is garbage — real kernel addresses here look like ffff8aba…). That's the

general-protection fault. With panic_on_oops=1, the oops became a panic + reboot.

Environment Info

• Platform: linux
• Terminal: vte-based
• Version: 2.1.170
• Feedback ID: 3aedac33-357b-4d2a-acc6-d13d93d0d858

Errors

[]