core.test.sh ends up in LDB on arm with --fancy

[3b]

when running test suite, core.test.sh consistently ends up in LDB:
( tested on 1.2.2.1-1c95d65 built with --fancy on armv7l linux (tegra k1),
model name : ARMv7 Processor rev 3 (v7l)
Features : swp half thumb fastmult vfp edsp neon vfpv3 tls vfpv4 idiva idivt )

* test Basic SAVE-LISP-AND-DIE ok (saved core ran)
...
[saving current Lisp image into core-test.core:
writing 1152 bytes from the read-only space at 0x0x4000000
writing 1544 bytes from the static space at 0x0x8000000
writing 45035520 bytes from the dynamic space at 0x0x4f000000
done]
This is SBCL 1.2.2.1-1c95d65, an implementation of ANSI Common Lisp.
...
CORRUPTION WARNING in SBCL pid 2679:
Memory fault at 16 (pc=0x16, sp=(nil))
The integrity of this image is possibly compromised.
Continuing with fingers crossed.
internal error #35 (invalid array index)
    SC: 5, Offset: 2 $1= 0x4f072007: other pointer
    SC: 4, Offset: 0 0x00000000: even fixnum: 0
    SC: 4, Offset: 24 0x00001000: even fixnum: 1024
fatal error encountered in SBCL pid 2679:
internal error too early in init, can't recover

Welcome to LDB, a low-level debugger for the Lisp runtime environment.
ldb>

Backtrace:
<Frame 0xb5e12628 [interrupted], CODE: 0x4F87E007, FIND-SYMBOL*, <no LRA>, PC: 0xcfc>
<Frame 0xb5e1260c, CODE: 0x4FA3B687, FIND-SYMBOL, LRA: 0x4fa3b8ef, PC: 0x240>
<Frame 0xb5e125e8, CODE: 0x4FE42967, OUTPUT-SYMBOL, LRA: 0x4fe42b5f, PC: 0x1b0>
<Frame 0xb5e12530, CODE: 0x4F18C007, OUTPUT-OBJECT, (Not other pointer???), (Not other pointer???), (Not other pointer???), LRA: 0x4f18c44f, PC: 0x380>
<Frame 0xb5e1251c, CODE: 0x4F05D3AF, PRIN1, LRA: 0x4f05d4bf, PC: 0xe8>
<Frame 0xb5e124e4, CODE: 0x503C5007, S-FORMAT-DIRECTIVE-INTERPRETER, LRA: 0x503c524f, PC: 0x1f0>
<Frame 0xb5e124bc, CODE: 0x4F841007, INTERPRET-DIRECTIVE-LIST, LRA: 0x4f84124f, PC: 0x1f8>
<Frame 0xb5e12418, CODE: 0x50AA1007, INTERPRET-FORMAT-LOGICAL-BLOCK, (Not other pointer???), (Not other pointer???), LRA: 0x50aa2207, PC: 0x1150>
<Frame 0xb5e12374, CODE: 0x50AA1007, INTERPRET-FORMAT-LOGICAL-BLOCK, (Not other pointer???), (Not other pointer???), LRA: 0x50aa15f7, PC: 0x540>
<Frame 0xb5e12324, CODE: 0x503B5007, <-FORMAT-DIRECTIVE-INTERPRETER, LRA: 0x503b57ff, PC: 0x798>
<Frame 0xb5e122fc, CODE: 0x4F841007, INTERPRET-DIRECTIVE-LIST, LRA: 0x4f84124f, PC: 0x1f8>
<Frame 0xb5e122b0, CODE: 0x4F18044F, %FORMAT, LRA: 0x4f18065f, PC: 0x1d8>
<Frame 0xb5e12220, CODE: 0x4F05922F, FORMAT, LRA: 0x4f0595ff, PC: 0x390>
<Frame 0xb5e120a0, CODE: 0x4F7FF007, DEBUGGER-DISABLED-HOOK, (Not other pointer???), (Not other pointer???), (Not other pointer???), (Not other pointer???), (Not other pointer???), LRA: 0x4f7ff617, PC: 0x590>
<Frame 0xb5e12044, CODE: 0x4F7E999F, RUN-HOOK, LRA: 0x4f7e9bc7, PC: 0x208>
<Frame 0xb5e1202c, CODE: 0x4F15304F, INVOKE-DEBUGGER, LRA: 0x4f153297, PC: 0x1e0>
<Frame 0xb5e1200c, CODE: 0x4F047007, ERROR, LRA: 0x4f04733f, PC: 0x2f8>
<Frame 0xb5e12000, CODE: 0x4F045007, MEMORY-FAULT-ERROR, LRA: 0x4f045107, PC: 0xe8>

[3b (00003b)]

with further testing, it seems to be specific to --fancy.
The test passes on 1.2.2.4-0d0927a without --fancy, and fails on 1.2.1.4-e8af16d with it.

[Stas Boukarev (stassats)]

In 3e000f4bd82edfae16670c41f113f8c955188a4b.