Re: [Qemu-devel] Next stable release

[TeLeMan <geleman@gmail.com>]

In fact its still overflow gen_opc_buf.


malc-4 wrote:
> 
> On Thu, 9 Apr 2009, TeLeMan wrote:
> 
>> 
>> 
>> malc-4 wrote:
>> > 
>> > On Mon, 6 Apr 2009, TeLeMan wrote:
>> > 
>> >> 
>> >> I hope you can fix the following bugs in the next release:
>> >> 
>> >>
>> http://www.nabble.com/-PATCH--i386-hard-interrupt-generation-bug-fix-p14921171.html
>> >>
>> http://www.nabble.com/MAX_OP_PER_INSTR-should-be-larger-tt22573338.html
>> >> 
>> > 
>> > You should retry with the current QEMU, immediate ro[lr]s should
>> produce a
>> > lot less tcg ops currently.
>> > 
>> > -- 
>> > mailto:av1474@comtv.ru
>> > 
>> > 
>> > 
>> > 
>> 
>> It is not fixed yet.
>> 
> 
> [..snip..]
> 
> Omitting dump before liveness analysis...
> 
>> 
>>  ---- 0x4468c0
>>  ld_i32 loc24,env,$0x18
>>  nopn $0x3,$0x0,$0x3
>>  movi_i32 loc25,$0x0
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  movi_i32 tmp32,$0xe
>>  shl_i32 tmp30,loc25,tmp32
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x21,$0x3
>>  movi_i32 tmp32,$0x12
>>  shr_i32 tmp14,loc24,tmp32
>>  or_i32 tmp14,tmp14,tmp30
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x20,$0x3
>>  movi_i32 tmp32,$0xe
>>  shl_i32 tmp30,loc24,tmp32
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x21,$0x3
>>  nopn $0x3,$0x21,$0x3
>>  mov_i32 loc24,tmp30
>>  or_i32 loc24,loc24,tmp14
>>  nopn $0x3,$0x21,$0x3
>>  st_i32 loc24,env,$0x18
>>  movi_i32 tmp8,$0x0
>>  nopn $0x2,$0x2
>>  st_i32 tmp8,env,$0x1c
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x6,$0x14,$0x1,$0x26,$0x10,$0x6
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x2,$0x3
>>  nopn $0x3,$0x22,$0x3
>>  nopn $0x3,$0x23,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x11,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x11,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x10,$0x3
>>  nopn $0x3,$0x10,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x10,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x2,$0x3
>>  nopn $0x3,$0x3,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x21,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x21,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x20,$0x3
>>  nopn $0x3,$0x20,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x20,$0x3
>>  nopn $0x2,$0x2
>>  nopn $0x3,$0x2,$0x3
>>  nopn $0x3,$0x3,$0x3
>>  discard cc_dst_0
>>  discard cc_dst_1
>>  nopn $0x2,$0x2
> 
> As you can see above most of the stuff was turned into nops by
> la.. So...
> 
> [..snip..]
> 
>> 
>> 65 ops were translated at 0x4468c0 or 0x4468c3. 
>> 71 ops were translated at the last instruction(0x4468eb).
> 
>> Another question: Who can ensure no other instructions like this? just by
>> testing?
>> 
> 
> -- 
> mailto:av1474@comtv.ru
> 
> 
> 
> 


-- 
View this message in context: http://www.nabble.com/Next-stable-release-tp22897895p22982404.html
Sent from the QEMU - Dev mailing list archive at Nabble.com.