* [Qemu-devel] singlestep and sparc32
@ 2010-04-19 14:33 Artyom Tarasenko
2010-04-19 16:07 ` malc
` (2 more replies)
0 siblings, 3 replies; 5+ messages in thread
From: Artyom Tarasenko @ 2010-04-19 14:33 UTC (permalink / raw)
To: qemu-devel, weil, Blue Swirl
is -single-step option supposed to work under qemu-system-sparc? Or is
it only for non-pipelined CPUs?
I get pretty strange results: some addresses seem to be executed
twice, and then qemu hangs on the last nop:
--------------
IN:
0x00000000: b 0xb788
--------------
IN:
0x00000004: rd %psr, %l0
--------------
IN:
0x00000004: rd %psr, %l0
--------------
IN:
0x00000008: nop
--------------
IN:
0x00000008: nop
--------------
IN:
0x0000000c: nop
--------------
IN:
0x0000000c: nop
... skipped ...
IN:
0x7000a58c: bg 0x7000a588
--------------
IN:
0x7000a590: nop
--------------
IN:
0x7000a590: nop
--------------
IN:
0x7000a594: b 0x7000a570
--------------
IN:
0x7000a594: b 0x7000a570
--------------
IN:
0x7000a578: be 0x7000a59c
--------------
IN:
0x7000a580: stba %l1, [ %l0 ] #ASI_M_BYPASS
--------------
IN:
0x7000a588: deccc %l2
--------------
IN:
0x7000a594: b 0x7000a570
--------------
IN:
0x7000a590: nop
^^^^^^^^^^^ That's the end
Without -singlestep it seems to work better:
0x7000a588: deccc %l2
0x7000a58c: bg 0x7000a588
0x7000a590: nop
--------------
IN:
0x7000a594: b 0x7000a570
0x7000a598: inc %l4
--------------
IN:
0x7000a570: lduba [ %l4 ] #ASI_M_KERNELTXT, %l1
0x7000a574: cmp %l1, 0xff
0x7000a578: be 0x7000a59c
... and so on ...
--
Regards,
Artyom Tarasenko
solaris/sparc under qemu blog: http://tyom.blogspot.com/
^ permalink raw reply [flat|nested] 5+ messages in thread* Re: [Qemu-devel] singlestep and sparc32
2010-04-19 14:33 [Qemu-devel] singlestep and sparc32 Artyom Tarasenko
@ 2010-04-19 16:07 ` malc
2010-04-19 16:19 ` [Qemu-devel] " Stefan Weil
2010-04-19 19:55 ` Blue Swirl
2 siblings, 0 replies; 5+ messages in thread
From: malc @ 2010-04-19 16:07 UTC (permalink / raw)
To: Artyom Tarasenko; +Cc: Blue Swirl, qemu-devel
On Mon, 19 Apr 2010, Artyom Tarasenko wrote:
> is -single-step option supposed to work under qemu-system-sparc? Or is
> it only for non-pipelined CPUs?
Eh? Perhaps you meant that it's only for CPUs without delay slots?
>
> I get pretty strange results: some addresses seem to be executed
> twice, and then qemu hangs on the last nop:
>
> --------------
> IN:
> 0x00000000: b 0xb788
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x0000000c: nop
> --------------
> IN:
> 0x0000000c: nop
>
> ... skipped ...
>
> IN:
> 0x7000a58c: bg 0x7000a588
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a578: be 0x7000a59c
> --------------
> IN:
> 0x7000a580: stba %l1, [ %l0 ] #ASI_M_BYPASS
> --------------
> IN:
> 0x7000a588: deccc %l2
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a590: nop
>
> ^^^^^^^^^^^ That's the end
>
> Without -singlestep it seems to work better:
>
>
> 0x7000a588: deccc %l2
> 0x7000a58c: bg 0x7000a588
> 0x7000a590: nop
>
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> 0x7000a598: inc %l4
>
> --------------
> IN:
> 0x7000a570: lduba [ %l4 ] #ASI_M_KERNELTXT, %l1
> 0x7000a574: cmp %l1, 0xff
> 0x7000a578: be 0x7000a59c
> ... and so on ...
>
>
>
--
mailto:av1474@comtv.ru
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Qemu-devel] Re: singlestep and sparc32
2010-04-19 14:33 [Qemu-devel] singlestep and sparc32 Artyom Tarasenko
2010-04-19 16:07 ` malc
@ 2010-04-19 16:19 ` Stefan Weil
2010-04-19 19:55 ` Blue Swirl
2 siblings, 0 replies; 5+ messages in thread
From: Stefan Weil @ 2010-04-19 16:19 UTC (permalink / raw)
To: Artyom Tarasenko; +Cc: Blue Swirl, qemu-devel
Artyom Tarasenko schrieb:
> is -single-step option supposed to work under qemu-system-sparc? Or is
> it only for non-pipelined CPUs?
>
> I get pretty strange results: some addresses seem to be executed
> twice, and then qemu hangs on the last nop:
>
> --------------
> IN:
> 0x00000000: b 0xb788
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x0000000c: nop
> --------------
> IN:
> 0x0000000c: nop
>
> ... skipped ...
>
> IN:
> 0x7000a58c: bg 0x7000a588
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a578: be 0x7000a59c
> --------------
> IN:
> 0x7000a580: stba %l1, [ %l0 ] #ASI_M_BYPASS
> --------------
> IN:
> 0x7000a588: deccc %l2
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a590: nop
>
> ^^^^^^^^^^^ That's the end
>
> Without -singlestep it seems to work better:
>
>
> 0x7000a588: deccc %l2
> 0x7000a58c: bg 0x7000a588
> 0x7000a590: nop
>
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> 0x7000a598: inc %l4
>
> --------------
> IN:
> 0x7000a570: lduba [ %l4 ] #ASI_M_KERNELTXT, %l1
> 0x7000a574: cmp %l1, 0xff
> 0x7000a578: be 0x7000a59c
> ... and so on ...
I use it for x86, mips and powerpc targets.
So pipelined CPUs work at least for some target
architectures, and sparc's singlestep is broken.
Have a look at target-mips/translate.c or
target-ppc/translate.c and compare their handling
of singlestep to the handling in target-sparc/translate.c.
The sparc variant is very simple - too simple, I assume.
Stefan
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Qemu-devel] Re: singlestep and sparc32
2010-04-19 14:33 [Qemu-devel] singlestep and sparc32 Artyom Tarasenko
2010-04-19 16:07 ` malc
2010-04-19 16:19 ` [Qemu-devel] " Stefan Weil
@ 2010-04-19 19:55 ` Blue Swirl
2010-04-19 20:33 ` Artyom Tarasenko
2 siblings, 1 reply; 5+ messages in thread
From: Blue Swirl @ 2010-04-19 19:55 UTC (permalink / raw)
To: Artyom Tarasenko; +Cc: qemu-devel
On 4/19/10, Artyom Tarasenko <atar4qemu@googlemail.com> wrote:
> is -single-step option supposed to work under qemu-system-sparc? Or is
> it only for non-pipelined CPUs?
I've never used it.
> I get pretty strange results: some addresses seem to be executed
> twice, and then qemu hangs on the last nop:
One possibility is that we don't generate any code at all for nops
(sethi x, %g0). If you delete the "if (rd)" check in translate.c:1784,
does it help?
> --------------
> IN:
> 0x00000000: b 0xb788
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000004: rd %psr, %l0
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x00000008: nop
> --------------
> IN:
> 0x0000000c: nop
> --------------
> IN:
> 0x0000000c: nop
>
> ... skipped ...
>
> IN:
> 0x7000a58c: bg 0x7000a588
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a590: nop
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a578: be 0x7000a59c
> --------------
> IN:
> 0x7000a580: stba %l1, [ %l0 ] #ASI_M_BYPASS
> --------------
> IN:
> 0x7000a588: deccc %l2
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> --------------
> IN:
> 0x7000a590: nop
>
> ^^^^^^^^^^^ That's the end
>
> Without -singlestep it seems to work better:
>
>
> 0x7000a588: deccc %l2
> 0x7000a58c: bg 0x7000a588
> 0x7000a590: nop
>
> --------------
> IN:
> 0x7000a594: b 0x7000a570
> 0x7000a598: inc %l4
>
> --------------
> IN:
> 0x7000a570: lduba [ %l4 ] #ASI_M_KERNELTXT, %l1
> 0x7000a574: cmp %l1, 0xff
> 0x7000a578: be 0x7000a59c
> ... and so on ...
>
>
>
> --
> Regards,
> Artyom Tarasenko
>
> solaris/sparc under qemu blog: http://tyom.blogspot.com/
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* [Qemu-devel] Re: singlestep and sparc32
2010-04-19 19:55 ` Blue Swirl
@ 2010-04-19 20:33 ` Artyom Tarasenko
0 siblings, 0 replies; 5+ messages in thread
From: Artyom Tarasenko @ 2010-04-19 20:33 UTC (permalink / raw)
To: Blue Swirl; +Cc: qemu-devel
2010/4/19 Blue Swirl <blauwirbel@gmail.com>:
> On 4/19/10, Artyom Tarasenko <atar4qemu@googlemail.com> wrote:
>> is -single-step option supposed to work under qemu-system-sparc? Or is
>> it only for non-pipelined CPUs?
>
> I've never used it.
is there another way to reduce a translation block? -icount 1 doesn't
make a difference.
>> I get pretty strange results: some addresses seem to be executed
>> twice, and then qemu hangs on the last nop:
>
> One possibility is that we don't generate any code at all for nops
> (sethi x, %g0). If you delete the "if (rd)" check in translate.c:1784,
> does it help?
Seems to make no difference at all.
>> --------------
>> IN:
>> 0x00000000: b 0xb788
>> --------------
>> IN:
>> 0x00000004: rd %psr, %l0
>> --------------
>> IN:
>> 0x00000004: rd %psr, %l0
>> --------------
>> IN:
>> 0x00000008: nop
>> --------------
>> IN:
>> 0x00000008: nop
>> --------------
>> IN:
>> 0x0000000c: nop
>> --------------
>> IN:
>> 0x0000000c: nop
>>
>> ... skipped ...
>>
>> IN:
>> 0x7000a58c: bg 0x7000a588
>> --------------
>> IN:
>> 0x7000a590: nop
>> --------------
>> IN:
>> 0x7000a590: nop
>> --------------
>> IN:
>> 0x7000a594: b 0x7000a570
>> --------------
>> IN:
>> 0x7000a594: b 0x7000a570
>> --------------
>> IN:
>> 0x7000a578: be 0x7000a59c
>> --------------
>> IN:
>> 0x7000a580: stba %l1, [ %l0 ] #ASI_M_BYPASS
>> --------------
>> IN:
>> 0x7000a588: deccc %l2
>> --------------
>> IN:
>> 0x7000a594: b 0x7000a570
>> --------------
>> IN:
>> 0x7000a590: nop
>>
>> ^^^^^^^^^^^ That's the end
>>
>> Without -singlestep it seems to work better:
>>
>>
>> 0x7000a588: deccc %l2
>> 0x7000a58c: bg 0x7000a588
>> 0x7000a590: nop
>>
>> --------------
>> IN:
>> 0x7000a594: b 0x7000a570
>> 0x7000a598: inc %l4
>>
>> --------------
>> IN:
>> 0x7000a570: lduba [ %l4 ] #ASI_M_KERNELTXT, %l1
>> 0x7000a574: cmp %l1, 0xff
>> 0x7000a578: be 0x7000a59c
>> ... and so on ...
--
Regards,
Artyom Tarasenko
solaris/sparc under qemu blog: http://tyom.blogspot.com/
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2010-04-19 20:33 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-04-19 14:33 [Qemu-devel] singlestep and sparc32 Artyom Tarasenko
2010-04-19 16:07 ` malc
2010-04-19 16:19 ` [Qemu-devel] " Stefan Weil
2010-04-19 19:55 ` Blue Swirl
2010-04-19 20:33 ` Artyom Tarasenko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).