WarpProcessor.add_path() crashes #8626

Version and Platform (required):

• Binary Ninja Version: 6.1.10601
• Edition: Ultimate
• OS: MacOS
• OS Version: 27
• CPU Architecture: M3 Pro

Bug Description:
Running add_path() crashes script.

import sys, binaryninja
from binaryninja.warp import WarpProcessor

print("start", flush=True)
proc = WarpProcessor(worker_count=1)
proc.add_path('/bin/ls')
print("added path, starting", flush=True)
wf = proc.start()
print("RESULT WarpFile:", wf, flush=True)
sys.exit(0)

Run as python3 script.py > out.log 2>&1; echo $?.

$ python3 warp_test2.py > warp_test2.out 2>&1
$ echo $?
139

Exit code 139 = SIGSEGV (128 + 11). Output file contains only:

start
added path, starting

• i.e. it crashes somewhere inside proc.start() (or possibly add_path itself lazily deferring work into start()), with no Python traceback at all, no exception, just a hard segfault of the whole interpreter.

The equivalent flow using add_binary_view instead of add_path, i.e. doing the load/analyze step manually in Python and only handing WarpProcessor an already-analyzed BinaryView, works fine and was verified multiple times, including batching multiple views into one processor (confirmed dedup: 3 binaries in, 1 merged 2-chunk .warp out):

import binaryninja
from binaryninja.warp import WarpProcessor

proc = WarpProcessor(worker_count=1)
bvs = []
for path in ['/bin/ls', '/bin/cat', '/bin/echo']:
    bv = binaryninja.load(path)
    bv.update_analysis_and_wait()
    proc.add_binary_view(bv)
    bvs.append(bv)

wf = proc.start()          # <-- WarpFile, no crash
data = bytes(wf.to_data_buffer())
for bv in bvs:
    bv.file.close()

This exits 0, prints a valid WarpFile repr, and produces real .warp bytes.