Sagittarius dumps on Göran Weinholt's schjig

[Jason Hemann]

When I run Göran Weinholt's schjig schsigs.scm against sagittarius v. 0.9.8_1 installed via homebrew, I get a segfault halfway through. I’ve run the program a couple of times; my most recent stacktrace is the following:

‌

...
(bytevector-s8-set! bytevector number number)
Caught signal 11 (SIGSEGV)
stack trace:
0   sash                                0x00000001094cd773 print_stack_trace + 48
1   sash                                0x00000001094cd73c abort_handler + 108
2   libsystem_platform.dylib            0x00007fff6b2505fd _sigtramp + 29
3   ???                                 0x0000000000000000 0x0 + 0
4   libsagittarius.0.9.8.dylib          0x000000010955dedf core_bytevector_TOu8_list + 89
5   libsagittarius.0.9.8.dylib          0x000000010952ff70 run_loop + 11497
6   libsagittarius.0.9.8.dylib          0x000000010952a043 evaluate_safe + 762
7   libsagittarius.0.9.8.dylib          0x000000010952b12e Sg_Apply3 + 143
8   sash                                0x00000001094ce45c main + 3190
9   libdyld.dylib                       0x00007fff6b057cc9 start + 1
10  ???                                 0x0000000000000001 0x0 + 1

‌

‌

[Takashi Kato]

Confirmed, it’s also happening on developing version.

[Takashi Kato]

Smallest reproducible code

(bytevector->u8-list '(1 1))

Also this returns extremely large bytevector, instead of one

(bytevector->u8-list '(1))

‌

[Takashi Kato]

Basically type checking was missing. bytevector->u8-list mustn’t accept a list, instead it must raise an &assertion.

[Takashi Kato]

Adding type check on bytevector->u8-list (Fixes #286)

→ <<cset cff3eb239573>>