I'm pretty sure there's a bug in the predecoder for x86 on this line:
http://repo.gem5.org/gem5/file/330f8109b199/src/arch/x86/predecoder.cc#l189
The condition of the if should be
else if(emi.opcode.num == 2 && (nextByte == 0x38 || nextByte == 0x3A))
and not:
else if(emi.opcode.num == 2 && (nextByte == 0x38 || nextByte == 0x3F))
The 0x3F at the end should be 0x3A. I'll fix this in the repository
soon, but in the mean time you can fix your copy locally.
Gabe
Quoting Mahmood Naderan <[email protected]>:
I debugged to find the faulty instruction. So I paste the logs:
the exec debug shows:
111733468000: system.cpu T0 : @memset+30.2 : IMUL_R_R : muleh t0,
t18, : IntAlu : D=0x0000000000000000
111733468000: system.cpu T0 : @memset+34.0 : CMP_R_I : limm t1,
0x90 : IntAlu : D=0x0000000000000090
111733468000: system.cpu T0 : @memset+34.1 : CMP_R_I : sub t0,
r8, t1 : IntAlu : D=0x00000000000000a5
111733468000: system.cpu T0 : @memset+41.0 : JNLE_I : rdip t1,
%ctrl153, : IntAlu : D=0x000000000065f14f
111733468000: system.cpu T0 : @memset+41.1 : JNLE_I : limm t2,
0x2d1 : IntAlu : D=0x00000000000002d1
111733468000: system.cpu T0 : @memset+41.2 : JNLE_I : wrip , t1,
t2 : IntAlu :
panic: X86 decoder internal error: 'Three byte opcode shouldn't be
handled by two_byte_opcodes.isa!'
and the O3 debug shows:
111733481000: global: DynInst: [sn:365868543] Instruction destroyed.
Instcount for system.cpu = 96
111733481000: global: DynInst: [sn:365868523] Instruction destroyed.
Instcount for system.cpu = 95
111733481000: global: DynInst: [sn:365868518] Instruction destroyed.
Instcount for system.cpu = 94
111733481000: system.cpu: Activity: 9
111733481000: system.cpu: Scheduling next tick!
111733482000: system.cpu.fetch-iport: Received timing
111733482000: system.cpu.fetch: [tid:0] Waking up from cache miss.
111733482000: system.cpu: CPU already running.
111733482000: system.cpu.fetch: [tid:0] Activating fetch due to
cache completion
111733482000: system.cpu.fetch: Activating stage.
111733482000: system.cpu: Activity: 10
111733482000: system.cpu:
FullO3CPU: Ticking main, FullO3CPU.
111733482000: system.cpu.fetch: Running stage.
111733482000: system.cpu.fetch: Attempting to fetch from [tid:0]
111733482000: system.cpu.fetch: [tid:0]: Icache miss is complete.
111733482000: system.cpu.fetch: [tid:0]: Adding instructions to
queue to decode.
panic: X86 decoder internal error: 'Three byte opcode shouldn't be
handled by two_byte_opcodes.isa!'
The complete error message is:
panic: X86 decoder internal error: 'Three byte opcode shouldn't be
handled by two_byte_opcodes.isa!'
{
leg = 0x10,
rex = 0,
op = {
num = 2,
op = 0x3a,
prefixA = 0xf,
prefixB = 0
},
modRM = 0,
sib = 0,
immediate = 0,
displacement = 0
dispSize = 0}
@ cycle 111733482000
[decodeInst:build/X86_SE/arch/x86/decoder.cc, line 108799]
Thanks for any idea.
Regards
On 11/13/11, Mahmood Naderan <[email protected]> wrote:
I found many some spek2k6 (astar, gamess, gcc, mcf, omnetpp, ....) faced
this error.
I really appreciate any comment or patch from developers.
thanks
--
// Naderan *Mahmood;
On Wed, Nov 9, 2011 at 10:10 AM, Mahmood Naderan
<[email protected]>wrote:
Hi,
With '-d' option in X86_SE, I get this error:
info: Entering event queue @ 339774000. Starting simulation...
panic: X86 decoder internal error: 'Three byte opcode shouldn't be
handled by two_byte_opcodes.isa!'
{
leg = 0x10,
rex = 0,
op = {
num = 2,
op = 0x3a,
prefixA = 0xf,
prefixB = 0
},
modRM = 0,
sib = 0,
immediate = 0,
displacement = 0
dispSize = 0}
@ cycle 340216000
[decodeInst:build/X86_SE/arch/x86/decoder.cc, line 108799]
It is stated that decoder.cc is generated automatically. So is there
any idea bout that?
thanks
--
// Naderan *Mahmood;
--
--
// Naderan *Mahmood;
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users
