windbg vs softice

回应的第 1-10 项留言
后
Message 1 in thread
寄件者：Gaurav Sareen (albelakela@hotmail.com)
主旨：WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-11 12:51:32 PST
 


Sorry for the newbie question.

What are the advantages of each, and is one definitely better? I would
assume that softice is better, but is it so much better to  be a significant
advantage?

I know that WinDbg is free, softice isnt. Softice can debug a live system,
WinDbg ( without livekd from sysinternals) cant.
Message 2 in thread
寄件者：Maxim S. Shatskih (maxim@storagecraft.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-11 13:21:05 PST
 

> What are the advantages of each, and is one definitely better? I would
> assume that softice is better, but is it so much better to  be a significant
> advantage?

Looks like the only advantage of SoftICE is that it requires a single machine only. Looking inside your own kernel while working is
a great thing sometimes.
Its great feature is also transparent debugging of both user and kernel mode code. Yes, you steps into int 2eh and find yourself in
the kernel.

WinDbg has a bit more functions (all those under \"!\" and \".\" prefixes, for instance, it has memory leak detector - !poolused), has
Windows GUI which is able of saving or copy/pasting the stack traces and debug prints to files/email messages. A very useful feature
for team work.
It also has absolutely no problems debugging Boot start type drivers, which is non-trivial with SI.
It can also debug crash dumps - on single machine, without the null modem. It is also a good and very small (~5MB on disk) user-mode
debugger which you can install on the test machine without bothering with full MSVC IDE.
SI also sometimes have problems with newer kernel builds (never XP betas etc).

> I know that WinDbg is free, softice isnt. Softice can debug a live system,
> WinDbg ( without livekd from sysinternals) cant.

What do you mean under \"live\"? WinDbg requires a second machine and null modem cable.

    Max
Message 3 in thread
寄件者：Nathan Nesbit (nathann@microsoft.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-15 10:24:37 PST
 

> What do you mean under \"live\"? WinDbg requires a second machine and null
modem cable.

    Sysinternals has a tool called LiveKD.  My guess at how it works is that
it installs a driver.  The driver creates a file on the HD and makes it look
like a crash dump.  One then launchs the debugger against the crash dump
file.  Reads to the file are redirected to become reads for the right RAM
location.  The end result is being able to use the debugger to view memory
on the system.

    Now LiveKd fakes some stuff (like context record and stack) because the
machine is still running and therefore the context and stack info are
constantly changing.  But it can be used to get some good information from
the machine (query HW config & RAM).

    For WinXP we have enhanced the kernel to do something similar.  We call
it Local KD.  Check out the Windbg docs for details.  The biggest difference
between it and LiveKD is that we don\'t bother to fake information like
context & stack that is incorrect.  The HW test guys here at MS love it to
run their PCI debugging extenstions.


\"Maxim S. Shatskih\" <maxim@storagecraft.com> wrote in message
news:9q4uve$2dp5$1@gavrilo.mtu.ru...
> > What are the advantages of each, and is one definitely better? I would
> > assume that softice is better, but is it so much better to  be a significant
> > advantage?
>
> Looks like the only advantage of SoftICE is that it requires a single machine only. Looking inside your own kernel while working is
> a great thing sometimes.
> Its great feature is also transparent debugging of both user and kernel mode code. Yes, you steps into int 2eh and find yourself in
> the kernel.
>
> WinDbg has a bit more functions (all those under \"!\" and \".\" prefixes, for instance, it has memory leak detector - !poolused), has
> Windows GUI which is able of saving or copy/pasting the stack traces and debug prints to files/email messages. A very useful feature
> for team work.
> It also has absolutely no problems debugging Boot start type drivers, whic h is non-trivial with SI.
> It can also debug crash dumps - on single machine, without the null modem. It is also a good and very small (~5MB on disk) user-mode
> debugger which you can install on the test machine without bothering with full MSVC IDE.
> SI also sometimes have problems with newer kernel builds (never XP betas etc).
>
> > I know that WinDbg is free, softice isnt. Softice can debug a live system,
> > WinDbg ( without livekd from sysinternals) cant.
>
> What do you mean under \"live\"? WinDbg requires a second machine and null modem cable.
>
>     Max
>
>
>
Message 4 in thread
寄件者：Sam Saeki (saekisam@hotmail.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-11 13:32:21 PST
 

My Ideas on this subject:

Major advantage of Soft-ICE: Single Machine.

Major disadvantage of Soft-ICE: You must copy the symbols (.pdb) files in
addition to the .sys you are debugging.  

If I were to choose one debugger: WinDbg.

Reasons:

* WinDbg is Free (softice costs about 300-400 per year if you want updates)
* With WinDbg I can capture logging output easily.
* WinDbg now supports Firewire! That will be great for logging massive
quantities of trace data.  
* With WinDbg I can save debugging workspaces.
* WinDbg has lots of extra commands.
* WinDbg supports debugger extensions.
* I don\'t need to copy symbols every time.
* Same WinDbg works for NT4, 2000, XP. It seems to be always be backward
compatible. Softice is notoriously version dependent.

Don\'t get me wrong...There was a time when I preferred SoftIce. Machines were
more expensive. WinDbg was harder to use. And SoftIce worked for 9X (same
interface). Now that I don\'t do 9X much and WinDbg has been upgraded, I will
use SoftIce less and less.

And, I have not used Firewire yet, but, I can\'t wait to try it out.


--
Posted from  [12.44.37.253]
via Mailgate.ORG Server - http://www.Mailgate.ORG
Message 5 in thread
寄件者：Oleksandr Bublyk (sab@mastereye.kiev.ua)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-12 02:45:18 PST
 

\"Sam Saeki\" <saekisam@hotmail.com> wrote in message
news:c5a70b11a062c0afc18bc93012378876.29727@mygate.mailgate.org...
> My Ideas on this subject:
>
> Major advantage of Soft-ICE: Single Machine.
>
> Major disadvantage of Soft-ICE: You must copy the symbols (.pdb) files in
> addition to the .sys you are debugging.
>

No, you have to say
    \"Package source with symbol table\" and the life is not so bad :)

--
Bye,
      SaB
Message 6 in thread
寄件者：Tim Roberts (timr@probo.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-12 21:05:55 PST
 

\"Oleksandr Bublyk\" <sab@mastereye.kiev.ua> wrote:

>\"Sam Saeki\" <saekisam@hotmail.com> wrote in message
>news:c5a70b11a062c0afc18bc93012378876.29727@mygate.mailgate.org...
>> My Ideas on this subject:
>>
>> Major advantage of Soft-ICE: Single Machine.
>>
>> Major disadvantage of Soft-ICE: You must copy the symbols (.pdb) files in
>> addition to the .sys you are debugging.
>>
>
>No, you have to say
>    \"Package source with symbol table\" and the life is not so bad :)

But you still have to copy the .NMS file to the target machine.  I believe
that\'s what Sam was referring to.
--
- Tim Roberts, timr@probo.com
  Providenza & Boekelheide, Inc.
Message 7 in thread
寄件者：Dejan Maksimovic (dejan@alfasp.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-15 12:41:43 PST
 

    What TARGET machine?:-) No, you don\'t have to copy it - you can just access
it via network.

    Regards, Dejan.

Tim Roberts wrote:

> \"Oleksandr Bublyk\" <sab@mastereye.kiev.ua> wrote:
>
> >\"Sam Saeki\" <saekisam@hotmail.com> wrote in message
> >news:c5a70b11a062c0afc18bc93012378876.29727@mygate.mailgate.org...
> >> My Ideas on this subject:
> >>
> >> Major advantage of Soft-ICE: Single Machine.
> >>
> >> Major disadvantage of Soft-ICE: You must copy the symbols (.pdb) files in
> >> addition to the .sys you are debugging.
> >>
> >
> >No, you have to say
> >    \"Package source with symbol table\" and the life is not so bad :)
>
> But you still have to copy the .NMS file to the target machine.  I believe
> that\'s what Sam was referring to.
> --
> - Tim Roberts, timr@probo.com
>   Providenza & Boekelheide, Inc.

--
Kind regards, Dejan M. CEO Alfa Co. www.alfasp.com
E-mail: dejan@alfasp.com
ICQ#: 56570367
Alfa File Monitor - File monitoring system for Win32 developers.
Alfa File Protector - File protection and hiding system for Win32 developers.
Message 8 in thread
寄件者：Walter Oney (waltoney@oneysoft.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-12 05:50:03 PST
 

Gaurav Sareen wrote:
> What are the advantages of each, and is one definitely better? I would
> assume that softice is better, but is it so much better to  be a significant
> advantage?

I use Soft-Ice exclusively, and I teach my seminars students how to use
it too. Here is the way I\'d stack the two up against each other:

Soft-Ice runs on a single machine *or* over a null-modem cable to any
Windows machine *or* over the Internet if you have the right kind of NIC
in the target machine. WinDbg requires a NULL modem cable connected to a
machine running either Win2K/Xp or Win/Me. I prefer the single-machine
approach because it means I only have to carry *one* laptop with me and
that I can more easily find suitable seminar classrooms all over the
world.

Soft-Ice never \"misses\" its connection with the target computer, whereas
WinDbg sometimes does.

It is very easy to understand which symbol files Soft-Ice will use
because you must specify them explicitly. WinDbg uses so many default
search rules that I\'ve never been able to predict where it would get
symbols and source from. This is a huge problem unless your host and
target machines are identical.

Soft-Ice keeps separate namespaces for each thing you\'re debugging,
wheareas WinDbg either didn\'t or still doesn\'t. Because of this
limitation in WinDbg, people fell into the habit of writing drivers
where every function has a globally unique name -- i.e.,
ToasterAddDevice, OvenAddDevice, etc., instead of just \"AddDevice\". This
in turn makes it harder to move code from one project to another or to
compare files between projects.

Soft-Ice lets you imbed the source code in the symbol file, which
greatly helps you debug problems in older versions of your code. I don\'t
believe WinDbg could let you do something comparable unless you
religiously maintain all your old source trees.

Soft-Ice lets you set \"virtual\" breakpoints in code that hasn\'t been
loaded into memory. The last time I was trying to use WinDbg, I couldn\'t
figure out how to set a breakpoint in the DriverEntry for the driver I
wanted to test.

The newest versions of Soft-Ice will halt on an assertion failure, even
in the free build of the OS. Ditto at the entry to KeBugCheckEx. I don\'t
know whether WinDbg will do this.

Soft-Ice runs in all versions of Windows. WinDbg runs only in NT/2K/Xp
or Me.
观看文件全部内容 （仍有 51 行）

Message 9 in thread
寄件者：Vijay P (vijay_p@hotmail.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-12 08:10:49 PST
 


By the way, WinDBG supports debugging of 64-bit drivers. SoftIce do not have
any support for debugging
on 64 bit drivers as of today.
To be precise, Windbg supports debugging on WinXP-32 bit and WinXP- 64 bit
O.S unlike Softice, where you have support only for WinXP-32 bit.

\"Walter Oney\" <waltoney@oneysoft.com> wrote in message
news:3BC6E834.493084B2@oneysoft.com...
> Gaurav Sareen wrote:
> > What are the advantages of each, and is one definitely better? I would
> > assume that softice is better, but is it so much better to  be a significant
> > advantage?
>
> I use Soft-Ice exclusively, and I teach my seminars students how to use
> it too. Here is the way I\'d stack the two up against each other:
>
> Soft-Ice runs on a single machine *or* over a null-modem cable to any
> Windows machine *or* over the Internet if you have the right kind of NIC
> in the target machine. WinDbg requires a NULL modem cable connected to a
> machine running either Win2K/Xp or Win/Me. I prefer the single-machine
> approach because it means I only have to carry *one* laptop with me and
> that I can more easily find suitable seminar classrooms all over the
> world.
>
> Soft-Ice never \"misses\" its connection with the target computer, whereas
> WinDbg sometimes does.
>
> It is very easy to understand which symbol files Soft-Ice will use
> because you must specify them explicitly. WinDbg uses so many default
> search rules that I\'ve never been able to predict where it would get
> symbols and source from. This is a huge problem unless your host and
> target machines are identical.
>
> Soft-Ice keeps separate namespaces for each thing you\'re debugging,
> wheareas WinDbg either didn\'t or still doesn\'t. Because of this
> limitation in WinDbg, people fell into the habit of writing drivers
> where every function has a globally unique name -- i.e.,
> ToasterAddDevice, OvenAddDevice, etc., instead of just \"AddDevice\". This
> in turn makes it harder to move code from one project to another or to
> compare files between projects.
>
> Soft-Ice lets you imbed the source code in the symbol file, which
> greatly helps you debug problems in older versions of your code. I don\'t
> believe WinDbg could let you do something comparable unless you
> religiously maintain all your old source trees.
>
> Soft-Ice lets you set \"virtual\" breakpoints in code that hasn\'t been
> loaded into memory. The last time I was trying to use WinDbg, I couldn\'t

观看文件全部内容 （仍有 62 行）

Message 10 in thread
寄件者：Dejan Maksimovic (dejan@alfasp.com)
主旨：Re: WinDbg vs Softice
新闻群组：comp.os.ms-windows.programmer.nt.kernel-mode
View this article only
日期：2001-10-15 10:38:31 PST
 


    It does have Win64 support, IIRC - it\'s not released yet. During the beta
test, whoever had 64-bit hardware could try it.

    Regards, Dejan.

Vijay P wrote:

> By the way, WinDBG supports debugging of 64-bit drivers. SoftIce do not have
> any support for debugging
> on 64 bit drivers as of today.
> To be precise, Windbg supports debugging on WinXP-32 bit and WinXP- 64 bit
> O.S unlike Softice, where you have support only for WinXP-32 bit.
>
> \"Walter Oney\" <waltoney@oneysoft.com> wrote in message
> news:3BC6E834.493084B2@oneysoft.com...
> > Gaurav Sareen wrote:
> > > What are the advantages of each, and is one definitely better? I would
> > > assume that softice is better, but is it so much better to  be a significant
> > > advantage?
> >
> > I use Soft-Ice exclusively, and I teach my seminars students how to use
> > it too. Here is the way I\'d stack the two up against each other:
> >
> > Soft-Ice runs on a single machine *or* over a null-modem cable to any
> > Windows machine *or* over the Internet if you have the right kind of NIC
> > in the target machine. WinDbg requires a NULL modem cable connected to a
> > machine running either Win2K/Xp or Win/Me. I prefer the single-machine
> > approach because it means I only have to carry *one* laptop with me and
> > that I can more easily find suitable seminar classrooms all over the
> > world.
> >
> > Soft-Ice never \"misses\" its connection with the target computer, whereas
> > WinDbg sometimes does.
> >
> > It is very easy to understand which symbol files Soft-Ice will use
> > because you must specify them explicitly. WinDbg uses so many default
> > search rules that I\'ve never been able to predict where it would get
> > symbols and source from. This is a huge problem unless your host and
> > target machines are identical.
> >
> > Soft-Ice keeps separate namespaces for each thing you\'re debugging,
> > wheareas WinDbg either didn\'t or still doesn\'t. Because of this
> > limitation in WinDbg, people fell into the habit of writing drivers
> > where every function has a globally unique name -- i.e.,
> > ToasterAddDevice, OvenAddDevice, etc., instead of just \"AddDevice\". This
> > in turn makes it harder to move code from one project to another or to
> > compare files between projects.

观看文件全部内容 （仍有 77 行）

后
 



?2002 Google