Hi Michael,
I'm using the "--skipto" option so i could simulate a bug that i have
so i can debug it.
i know that its not a bug that before Peach skip to the test case it
run the none mutated test case and only then jumps to the wanted test
case.
but this causes issues, i know that the fuzzing is working well thats
why i want to skip to a test case
i need to go through a usesless test case.
i think that there should be another flag that will tell Peach to run
just this test case or from this test case without runing the defualt
none mutated..
The first iteration with no mutators is always required to setup the
fuzzer state. Peach allows for very complex state models who's path is
not always able to be determined until it has executed for the first
time. The first iteration is used so the fuzzer can observe and
discover all the actions that will be fuzzed.
mike
On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
> Hi Michael,
> I'm using the "--skipto" option so i could simulate a bug that i have
> so i can debug it.
> i know that its not a bug that before Peach skip to the test case it
> run the none mutated test case and only then jumps to the wanted test
> case.
> but this causes issues, i know that the fuzzing is working well thats
> why i want to skip to a test case
> i need to go through a usesless test case.
> i think that there should be another flag that will tell Peach to run
> just this test case or from this test case without runing the defualt
> none mutated..
> The first iteration with no mutators is always required to setup the
> fuzzer state. Peach allows for very complex state models who's path is
> not always able to be determined until it has executed for the first
> time. The first iteration is used so the fuzzer can observe and
> discover all the actions that will be fuzzed.
> mike
> On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
> > Hi Michael,
> > I'm using the "--skipto" option so i could simulate a bug that i have
> > so i can debug it.
> > i know that its not a bug that before Peach skip to the test case it
> > run the none mutated test case and only then jumps to the wanted test
> > case.
> > but this causes issues, i know that the fuzzing is working well thats
> > why i want to skip to a test case
> > i need to go through a usesless test case.
> > i think that there should be another flag that will tell Peach to run
> > just this test case or from this test case without runing the defualt
> > none mutated..
Nope, the point here is one cannot determine the state of the state
model with out first running it. If your doing something super simple
like file fuzzing were your state model just writes a file, thats one
thing, but with a more complex network based state model were choices
are made based on data coming back from the server there is no way to
pre-determine what will occur will out running it.
Perhaps you could explain why running the initial test case is causing a
problem for you and we can figure out a workaround for it.
mike
On 11/3/2009 11:18 AM, shai.sarf...@googlemail.com wrote:
> i understand . but must it use the publisher to do it as well ?
> i mean that i can setup the fuzzer state without acctualy sending the
> data ?
> On Nov 3, 8:25 pm, Michael Eddington <medding...@gmail.com> wrote:
>> The first iteration with no mutators is always required to setup the
>> fuzzer state. Peach allows for very complex state models who's path is
>> not always able to be determined until it has executed for the first
>> time. The first iteration is used so the fuzzer can observe and
>> discover all the actions that will be fuzzed.
>> mike
>> On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
>>> Hi Michael,
>>> I'm using the "--skipto" option so i could simulate a bug that i have
>>> so i can debug it.
>>> i know that its not a bug that before Peach skip to the test case it
>>> run the none mutated test case and only then jumps to the wanted test
>>> case.
>>> but this causes issues, i know that the fuzzing is working well thats
>>> why i want to skip to a test case
>>> i need to go through a usesless test case.
>>> i think that there should be another flag that will tell Peach to run
>>> just this test case or from this test case without runing the defualt
>>> none mutated..
I have build a state machine for a network protocol and it is complex
like you saied
but the thing is that i need to debug a bug that i have in a specific
test case
when i want to debug this case i first have to go through all the
break points that i set
with the none motator test case. and then i can start to debug the
"bug" test case it self
Now, i have an idea how to by pass this: "run-pause-run"
we can set a "pause" for eatch test case - meaning, after a test case
has run Peach will pause untill a user press key.
This way i can desable the break points and let peach run the first
test case and then Peach will pause
and i will re-set the break points and press key so peach will
continue and i will be able to debug it properly.
What do you say? can it be a good solution or this could cause isuues
that i'm not aware of ?
Regards,
Shai.
On Nov 3, 11:57 pm, Michael Eddington <medding...@gmail.com> wrote:
> Nope, the point here is one cannot determine the state of the state
> model with out first running it. If your doing something super simple
> like file fuzzing were your state model just writes a file, thats one
> thing, but with a more complex network based state model were choices
> are made based on data coming back from the server there is no way to
> pre-determine what will occur will out running it.
> Perhaps you could explain why running the initial test case is causing a
> problem for you and we can figure out a workaround for it.
> mike
> On 11/3/2009 11:18 AM, shai.sarf...@googlemail.com wrote:
> > i understand . but must it use the publisher to do it as well ?
> > i mean that i can setup the fuzzer state without acctualy sending the
> > data ?
> > On Nov 3, 8:25 pm, Michael Eddington <medding...@gmail.com> wrote:
> >> The first iteration with no mutators is always required to setup the
> >> fuzzer state. Peach allows for very complex state models who's path is
> >> not always able to be determined until it has executed for the first
> >> time. The first iteration is used so the fuzzer can observe and
> >> discover all the actions that will be fuzzed.
> >> mike
> >> On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
> >>> Hi Michael,
> >>> I'm using the "--skipto" option so i could simulate a bug that i have
> >>> so i can debug it.
> >>> i know that its not a bug that before Peach skip to the test case it
> >>> run the none mutated test case and only then jumps to the wanted test
> >>> case.
> >>> but this causes issues, i know that the fuzzing is working well thats
> >>> why i want to skip to a test case
> >>> i need to go through a usesless test case.
> >>> i think that there should be another flag that will tell Peach to run
> >>> just this test case or from this test case without runing the defualt
> >>> none mutated..
Sounds like a good solution. If your using an Agent, an easy way to add
this would be to create a Monitor that does just that when it's
"OnTestStarting" method or "OnTestFinished" method is called. The input
will be asked for in the Agent console.
THis is a good way to add it with out making any code changes to actual
Peach code.
Check out the monitors in the Peach\Agents folder. You will also want
to read the following, both of which can be used as children in <Agent>
definitions.
> I have build a state machine for a network protocol and it is complex
> like you saied
> but the thing is that i need to debug a bug that i have in a specific
> test case
> when i want to debug this case i first have to go through all the
> break points that i set
> with the none motator test case. and then i can start to debug the
> "bug" test case it self
> Now, i have an idea how to by pass this: "run-pause-run"
> we can set a "pause" for eatch test case - meaning, after a test case
> has run Peach will pause untill a user press key.
> This way i can desable the break points and let peach run the first
> test case and then Peach will pause
> and i will re-set the break points and press key so peach will
> continue and i will be able to debug it properly.
> What do you say? can it be a good solution or this could cause isuues
> that i'm not aware of ?
> Regards,
> Shai.
> On Nov 3, 11:57 pm, Michael Eddington <medding...@gmail.com> wrote:
>> Nope, the point here is one cannot determine the state of the state
>> model with out first running it. If your doing something super simple
>> like file fuzzing were your state model just writes a file, thats one
>> thing, but with a more complex network based state model were choices
>> are made based on data coming back from the server there is no way to
>> pre-determine what will occur will out running it.
>> Perhaps you could explain why running the initial test case is causing a
>> problem for you and we can figure out a workaround for it.
>> mike
>> On 11/3/2009 11:18 AM, shai.sarf...@googlemail.com wrote:
>>> i understand . but must it use the publisher to do it as well ?
>>> i mean that i can setup the fuzzer state without acctualy sending the
>>> data ?
>>> On Nov 3, 8:25 pm, Michael Eddington <medding...@gmail.com> wrote:
>>>> The first iteration with no mutators is always required to setup the
>>>> fuzzer state. Peach allows for very complex state models who's path is
>>>> not always able to be determined until it has executed for the first
>>>> time. The first iteration is used so the fuzzer can observe and
>>>> discover all the actions that will be fuzzed.
>>>> mike
>>>> On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
>>>>> Hi Michael,
>>>>> I'm using the "--skipto" option so i could simulate a bug that i have
>>>>> so i can debug it.
>>>>> i know that its not a bug that before Peach skip to the test case it
>>>>> run the none mutated test case and only then jumps to the wanted test
>>>>> case.
>>>>> but this causes issues, i know that the fuzzing is working well thats
>>>>> why i want to skip to a test case
>>>>> i need to go through a usesless test case.
>>>>> i think that there should be another flag that will tell Peach to run
>>>>> just this test case or from this test case without runing the defualt
>>>>> none mutated..
> I have build a state machine for a network protocol and it is complex
> like you saied
> but the thing is that i need to debug a bug that i have in a specific
> test case
> when i want to debug this case i first have to go through all the
> break points that i set
> with the none motator test case. and then i can start to debug the
> "bug" test case it self
> Now, i have an idea how to by pass this: "run-pause-run"
> we can set a "pause" for eatch test case - meaning, after a test case
> has run Peach will pause untill a user press key.
> This way i can desable the break points and let peach run the first
> test case and then Peach will pause
> and i will re-set the break points and press key so peach will
> continue and i will be able to debug it properly.
> What do you say? can it be a good solution or this could cause isuues
> that i'm not aware of ?
> Regards,
> Shai.
> On Nov 3, 11:57 pm, Michael Eddington <medding...@gmail.com> wrote:
>> Nope, the point here is one cannot determine the state of the state
>> model with out first running it. If your doing something super simple
>> like file fuzzing were your state model just writes a file, thats one
>> thing, but with a more complex network based state model were choices
>> are made based on data coming back from the server there is no way to
>> pre-determine what will occur will out running it.
>> Perhaps you could explain why running the initial test case is causing a
>> problem for you and we can figure out a workaround for it.
>> mike
>> On 11/3/2009 11:18 AM, shai.sarf...@googlemail.com wrote:
>> > i understand . but must it use the publisher to do it as well ?
>> > i mean that i can setup the fuzzer state without acctualy sending the
>> > data ?
>> > On Nov 3, 8:25 pm, Michael Eddington <medding...@gmail.com> wrote:
>> >> The first iteration with no mutators is always required to setup the
>> >> fuzzer state. Peach allows for very complex state models who's path is
>> >> not always able to be determined until it has executed for the first
>> >> time. The first iteration is used so the fuzzer can observe and
>> >> discover all the actions that will be fuzzed.
>> >> mike
>> >> On 11/3/2009 3:57 AM, shai.sarf...@googlemail.com wrote:
>> >>> Hi Michael,
>> >>> I'm using the "--skipto" option so i could simulate a bug that i have
>> >>> so i can debug it.
>> >>> i know that its not a bug that before Peach skip to the test case it
>> >>> run the none mutated test case and only then jumps to the wanted test
>> >>> case.
>> >>> but this causes issues, i know that the fuzzing is working well thats
>> >>> why i want to skip to a test case
>> >>> i need to go through a usesless test case.
>> >>> i think that there should be another flag that will tell Peach to run
>> >>> just this test case or from this test case without runing the defualt
>> >>> none mutated..
|
