diff options
Diffstat (limited to 'results/classifier/zero-shot/108/other/2488')
| -rw-r--r-- | results/classifier/zero-shot/108/other/2488 | 80 |
1 files changed, 80 insertions, 0 deletions
diff --git a/results/classifier/zero-shot/108/other/2488 b/results/classifier/zero-shot/108/other/2488 new file mode 100644 index 00000000..8baf04e4 --- /dev/null +++ b/results/classifier/zero-shot/108/other/2488 @@ -0,0 +1,80 @@ +other: 0.817 +permissions: 0.773 +device: 0.749 +socket: 0.723 +network: 0.718 +performance: 0.686 +debug: 0.679 +semantic: 0.669 +PID: 0.644 +graphic: 0.593 +files: 0.550 +KVM: 0.491 +vnc: 0.440 +boot: 0.432 + +m68k: 68030 (?): fmove.p doesn't work (6888[1|2] emulation isn't implemented??) +Description of problem: +The following code should be executing a move to the fpu and then a move from it and then branching. + +``` + ff813590 f2 10 4f 00 fmove.p (A0),FP6 + ff813594 f2 11 6f 7f fmove.p FP6,(A1) {#0x7f} + ff813598 61 00 fe 52 bsr.w SUB_ff8133ec +``` + +However, hitting the instruction at `0xff813590` causes the `PC` to go off into the weeds and then the emulation gets stuck and never proceeds. + +Before executing the instruction the CPU state looks like this + +``` +(qemu) info registers + +CPU#0 +D0 = ffffffff A0 = ff813584 F0 = c004 cc00000000000000 ( -51) +D1 = 0000ffff A1 = 0000335e F1 = c00d a866000000000000 ( -21555) +D2 = 00000002 A2 = ff8138a2 F2 = 401b 91a2b3c000000000 ( 3.0542e+08) +D3 = 00000003 A3 = ff824008 F3 = 3fb4 ab3c4d0000000000 ( 3.54107e-23) +D4 = 00000004 A4 = ff81dbb6 F4 = 3d12 919a22ab33bc4000 (3.84141e-226) +D5 = 00000000 A5 = 00000400 F5 = 1020 8060708090a0b0c0 ( 0) +D6 = 0000000c A6 = 00003790 F6 = 7fff ffffffffffffffff ( nan) +D7 = 00000000 A7 = 0000316e F7 = 7fff ffffffffffffffff ( nan) +PC = ff813590 SR = 2708 T:0 I:7 SI -N--- +FPSR = 00000000 ---- + FPCR = 0000 X RN + A7(MSP) = 00000000 A7(USP) = 80000000 ->A7(ISP) = 00003796 +VBR = 0x0000338e +SFC = 3 DFC 0 +SSW 00000000 TCR 00000000 URP 00000000 SRP 00000000 +DTTR0/1: 00000000/00000000 ITTR0/1: 00000000/00000000 +MMUSR 00000000, fault at 00000000 +``` + +After single stepping: + +``` +(qemu) info registers + +CPU#0 +D0 = ffffffff A0 = ff813584 F0 = c004 cc00000000000000 ( -51) +D1 = 0000ffff A1 = 0000335e F1 = c00d a866000000000000 ( -21555) +D2 = 00000002 A2 = ff8138a2 F2 = 401b 91a2b3c000000000 ( 3.0542e+08) +D3 = 00000003 A3 = ff824008 F3 = 3fb4 ab3c4d0000000000 ( 3.54107e-23) +D4 = 00000004 A4 = ff81dbb6 F4 = 3d12 919a22ab33bc4000 (3.84141e-226) +D5 = 00000000 A5 = 00000400 F5 = 1020 8060708090a0b0c0 ( 0) +D6 = 0000000c A6 = 00003790 F6 = 7fff ffffffffffffffff ( nan) +D7 = 00000000 A7 = 00003166 F7 = 7fff ffffffffffffffff ( nan) +PC = ff8138a2 SR = 2708 T:0 I:7 SI -N--- +FPSR = 00000000 ---- + FPCR = 0000 X RN + A7(MSP) = 00000000 A7(USP) = 80000000 ->A7(ISP) = 0000316e +VBR = 0x0000338e +SFC = 3 DFC 0 +SSW 00000000 TCR 00000000 URP 00000000 SRP 00000000 +DTTR0/1: 00000000/00000000 ITTR0/1: 00000000/00000000 +MMUSR 00000000, fault at 00000000 +``` + +With this code the `VBR` doesn't point at an actual vector table from what I can tell and it is pointing at some memory that contains `0xff8138a2` so I guess it hits the instruction, the FPU isn't implemented so it tries to do an `F-line exception` instead but the vector table doesn't actually contain a handler and it's trying to execute garbage that causes the lock up. + +Basically, I guess I need to implement the 6888[1|2] for this code to work. |