cygcheck -svc segfaults on Windows 8.1 with cygwin64

cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Gabriel Marcano]

cygcheck -svc causes a segfault on Windows 8.1 on line 1610 of cygcheck.cc, based on gdb output. I'm including some gdb output below that showcases this issue:


1610                    strcat (osname, products[prod]);
(gdb) list
1605                  if (prod == PRODUCT_UNLICENSED)
1606                    strcat (osname, "Unlicensed");
1607                  else if (prod > PRODUCT_PROFESSIONAL_WMC)
1608                    strcat (osname, "");
1609                  else
1610                    strcat (osname, products[prod]);
1611                }
1612              else
1613                {
1614                }
(gdb) p prod
$1 = 101
(gdb) p sizeof(products)/sizeof(void*)
$2 = 101


'products[prod]' in this case is accessing memory outside of the 'products' array. The odd thing is that a few lines back, the declaration of products includes 104 elements. I do not know why this discrepancy exists-- anyone have a clue?

So far I've tried installing the experimental release of the cygwin base tools (where I think cygcheck is found), some older version of the base tools, and even just by copying cygcheck from the latest snapshot and replacing the one I had in my /cygwin64/bin folder. None of these alternative versions alleviated the issue (and since the snapshot is stripped of debugging symbols and the symbols I have do not match it, I couldn't debug it, other than the fact it crashed with a segfault with the same output as the other cases).

I have been trying to compile cygwin on my computer, but I have not had much success yet. I will continue to try in the mean time. Thanks in advance for any help!


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1922 bytes --]

On Nov 18 21:35, Gabriel Marcano wrote:
> cygcheck -svc causes a segfault on Windows 8.1 on line 1610 of cygcheck.cc, based on gdb output. I'm including some gdb output below that showcases this issue:
> 
> 
> 1610                    strcat (osname, products[prod]);
> (gdb) list
> 1605                  if (prod == PRODUCT_UNLICENSED)
> 1606                    strcat (osname, "Unlicensed");
> 1607                  else if (prod > PRODUCT_PROFESSIONAL_WMC)
> 1608                    strcat (osname, "");
> 1609                  else
> 1610                    strcat (osname, products[prod]);
> 1611                }
> 1612              else
> 1613                {
> 1614                }
> (gdb) p prod
> $1 = 101
> (gdb) p sizeof(products)/sizeof(void*)
> $2 = 101
> 
> 
> 'products[prod]' in this case is accessing memory outside of the
> 'products' array. The odd thing is that a few lines back, the
> declaration of products includes 104 elements. I do not know why this
> discrepancy exists-- anyone have a clue?

Yes.  The size of the array is not what you think it is.  The reason is
that several commas are missing.  Since I created this array, it's all
my fault :}  I didn't notice this because I'm using the Enterprise
version which have a product value 4, which is pretty safe.

I'm going to fix this in CVS today, but at the same time I'm also going
to look for a solution to differ between Windows 8 and 8.1 (also 2012
vs. 2012R2) in the cygcheck output.  Right now cygcheck just prints
Windows 8.


Thanks for the report,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Warren Young]

On 11/19/2013 03:03, Corinna Vinschen wrote:
>
> I'm also going
> to look for a solution to differ between Windows 8 and 8.1 (also 2012
> vs. 2012R2) in the cygcheck output.

GetVersionEx() should do it: http://goo.gl/DbhsRJ

If you follow the link to the OSVERSIONINFO structure, you will find a 
table that tells you how you distinguish those four cases.

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1245 bytes --]

On Nov 19 09:37, Warren Young wrote:
> On 11/19/2013 03:03, Corinna Vinschen wrote:
> >
> >I'm also going
> >to look for a solution to differ between Windows 8 and 8.1 (also 2012
> >vs. 2012R2) in the cygcheck output.
> 
> GetVersionEx() should do it: http://goo.gl/DbhsRJ
> 
> If you follow the link to the OSVERSIONINFO structure, you will find
> a table that tells you how you distinguish those four cases.

Why do they have to make such a mess out of a simple function like
GetVersionEx?  It returns different OS version numbers based on the
existence of a manifest in the executable.  How dense is that?

So we have thousands of executables, none of them has a 8.1 manifest.
As a result, the uname() function returns OS versions 6.2 rather than
6.3.  Aaaaaargh.

In cygcheck I added a patch to check dwBuildNumber this morning.  If
it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
fix the OS version number of course.  Sigh.

I'm going to tweak the OS version number and I'll do the same in
Cygwin's uname function as well.


Thanks,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Charles Wilson]

On 11/19/2013 12:13 PM, Corinna Vinschen wrote:
> Why do they have to make such a mess out of a simple function like
> GetVersionEx?  It returns different OS version numbers based on the
> existence of a manifest in the executable.  How dense is that?
>
> So we have thousands of executables, none of them has a 8.1 manifest.
> As a result, the uname() function returns OS versions 6.2 rather than
> 6.3.  Aaaaaargh.
>
> In cygcheck I added a patch to check dwBuildNumber this morning.  If
> it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
> fix the OS version number of course.  Sigh.
>
> I'm going to tweak the OS version number and I'll do the same in
> Cygwin's uname function as well.

Good grief. I suppose I need to add something similar to 
/usr/lib/csih/winProductName.exe...

--
Chuck



--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1107 bytes --]

On Nov 19 13:21, Charles Wilson wrote:
> On 11/19/2013 12:13 PM, Corinna Vinschen wrote:
> >Why do they have to make such a mess out of a simple function like
> >GetVersionEx?  It returns different OS version numbers based on the
> >existence of a manifest in the executable.  How dense is that?
> >
> >So we have thousands of executables, none of them has a 8.1 manifest.
> >As a result, the uname() function returns OS versions 6.2 rather than
> >6.3.  Aaaaaargh.
> >
> >In cygcheck I added a patch to check dwBuildNumber this morning.  If
> >it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
> >fix the OS version number of course.  Sigh.
> >
> >I'm going to tweak the OS version number and I'll do the same in
> >Cygwin's uname function as well.
> 
> Good grief. I suppose I need to add something similar to
> /usr/lib/csih/winProductName.exe...

Looks like it, yes.  What on earth were they thinking?


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Larry Hall (Cygwin)]

On 11/19/2013 2:03 PM, Corinna Vinschen wrote:
> On Nov 19 13:21, Charles Wilson wrote:
>> On 11/19/2013 12:13 PM, Corinna Vinschen wrote:
>>> Why do they have to make such a mess out of a simple function like
>>> GetVersionEx?  It returns different OS version numbers based on the
>>> existence of a manifest in the executable.  How dense is that?
>>>
>>> So we have thousands of executables, none of them has a 8.1 manifest.
>>> As a result, the uname() function returns OS versions 6.2 rather than
>>> 6.3.  Aaaaaargh.
>>>
>>> In cygcheck I added a patch to check dwBuildNumber this morning.  If
>>> it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
>>> fix the OS version number of course.  Sigh.
>>>
>>> I'm going to tweak the OS version number and I'll do the same in
>>> Cygwin's uname function as well.
>>
>> Good grief. I suppose I need to add something similar to
>> /usr/lib/csih/winProductName.exe...
>
> Looks like it, yes.  What on earth were they thinking?

Who says they were thinking? ;-)

-- 
Larry

_____________________________________________________________________

A: Yes.
 > Q: Are you sure?
 >> A: Because it reverses the logical flow of conversation.
 >>> Q: Why is top posting annoying in email?

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1743 bytes --]

On Nov 19 14:20, Larry Hall (Cygwin) wrote:
> On 11/19/2013 2:03 PM, Corinna Vinschen wrote:
> >On Nov 19 13:21, Charles Wilson wrote:
> >>On 11/19/2013 12:13 PM, Corinna Vinschen wrote:
> >>>Why do they have to make such a mess out of a simple function like
> >>>GetVersionEx?  It returns different OS version numbers based on the
> >>>existence of a manifest in the executable.  How dense is that?
> >>>
> >>>So we have thousands of executables, none of them has a 8.1 manifest.
> >>>As a result, the uname() function returns OS versions 6.2 rather than
> >>>6.3.  Aaaaaargh.
> >>>
> >>>In cygcheck I added a patch to check dwBuildNumber this morning.  If
> >>>it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
> >>>fix the OS version number of course.  Sigh.
> >>>
> >>>I'm going to tweak the OS version number and I'll do the same in
> >>>Cygwin's uname function as well.
> >>
> >>Good grief. I suppose I need to add something similar to
> >>/usr/lib/csih/winProductName.exe...
> >
> >Looks like it, yes.  What on earth were they thinking?
> 
> Who says they were thinking? ;-)

Point.

I found what happened:
http://msdn.microsoft.com/en-us/library/windows/desktop/dn302074%28v=vs.85%29.aspx

Given that we are unable to provide and change manifests on the fly for
thousands of executables, we will have to hack our way along in future
because all upcoming versions of Windows will return a wrong OS version
number.

Why isn't there at least an additional non-manifest way to claim
compatibility with the current OS? :(


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 2131 bytes --]

On Nov 19 21:29, Corinna Vinschen wrote:
> On Nov 19 14:20, Larry Hall (Cygwin) wrote:
> > On 11/19/2013 2:03 PM, Corinna Vinschen wrote:
> > >On Nov 19 13:21, Charles Wilson wrote:
> > >>On 11/19/2013 12:13 PM, Corinna Vinschen wrote:
> > >>>Why do they have to make such a mess out of a simple function like
> > >>>GetVersionEx?  It returns different OS version numbers based on the
> > >>>existence of a manifest in the executable.  How dense is that?
> > >>>
> > >>>So we have thousands of executables, none of them has a 8.1 manifest.
> > >>>As a result, the uname() function returns OS versions 6.2 rather than
> > >>>6.3.  Aaaaaargh.
> > >>>
> > >>>In cygcheck I added a patch to check dwBuildNumber this morning.  If
> > >>>it's >= 9200, it's 8.1/2012R2, otherwise 8/2012.  But that doesn't
> > >>>fix the OS version number of course.  Sigh.
> > >>>
> > >>>I'm going to tweak the OS version number and I'll do the same in
> > >>>Cygwin's uname function as well.
> > >>
> > >>Good grief. I suppose I need to add something similar to
> > >>/usr/lib/csih/winProductName.exe...
> > >
> > >Looks like it, yes.  What on earth were they thinking?
> > 
> > Who says they were thinking? ;-)
> 
> Point.
> 
> I found what happened:
> http://msdn.microsoft.com/en-us/library/windows/desktop/dn302074%28v=vs.85%29.aspx
> 
> Given that we are unable to provide and change manifests on the fly for
> thousands of executables, we will have to hack our way along in future
> because all upcoming versions of Windows will return a wrong OS version
> number.
> 
> Why isn't there at least an additional non-manifest way to claim
> compatibility with the current OS? :(

Oh, and it's worse than I thought.  dwBuildNumber 9200 is the Windows 8
build number, so my workaround doesn't work and the entire version
information will be stuck at Windows 8 for the forseeable future, unless
somebody knows a workaround for this manifest mess.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Andrey Repin]

Greetings, Corinna Vinschen!

> Why isn't there at least an additional non-manifest way to claim
> compatibility with the current OS? :(

Because this "claim" is informational, or at least it should be.


--
WBR,
Andrey Repin (anrdaemon@yandex.ru) 20.11.2013, <00:56>

Sorry for my terrible english...


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 710 bytes --]

On Nov 20 00:59, Andrey Repin wrote:
> Greetings, Corinna Vinschen!
> 
> > Why isn't there at least an additional non-manifest way to claim
> > compatibility with the current OS? :(
> 
> Because this "claim" is informational, or at least it should be.

But apparently it isn't.  It's enforced.

However, I just had an in-shower inspiration and following up on it
I found a way to access the correct version info despite having no
8.1 manifest.  I'm still playing with it, but I think I have a working
patch by tomorrow.


Stay tuned,
Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 1083 bytes --]

On Nov 19 22:51, Corinna Vinschen wrote:
> On Nov 20 00:59, Andrey Repin wrote:
> > Greetings, Corinna Vinschen!
> > 
> > > Why isn't there at least an additional non-manifest way to claim
> > > compatibility with the current OS? :(
> > 
> > Because this "claim" is informational, or at least it should be.
> 
> But apparently it isn't.  It's enforced.
> 
> However, I just had an in-shower inspiration and following up on it
> I found a way to access the correct version info despite having no
> 8.1 manifest.  I'm still playing with it, but I think I have a working
> patch by tomorrow.

Patch checked in.  It was pretty simple, actually.  Rather than
GetVersionEx from kernel32.dll, just call RtlGetVersion from ntdll.dll.
It returns an OSVERSIONINFOEXW structure just as GetVersionExW, but
it's not tweaked according to the executable manifest, so the returned
info reflects the actual OS info.


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]

RE: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Jim Garrison]

> On Nov 19 14:20, Larry Hall (Cygwin) wrote:
> > On 11/19/2013 2:03 PM, Corinna Vinschen wrote:
> > >Looks like it, yes.  What on earth were they thinking?
> >
> > Who says they were thinking? ;-)
> 
> Point.
> 
> I found what happened:
> http://msdn.microsoft.com/en-
> us/library/windows/desktop/dn302074%28v=vs.85%29.aspx

Hmm... from the quoted link:

    We have made some significant changes in how the GetVersion(Ex) APIs 
    work in Windows 8.1 due to undesirable customer behaviors resulting from how 
    the GetVersion(Ex) APIs have been used in the past.

Pesky customers won't behave the way we want them to...  (sinister music)

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Csaba Raduly]

On Wed, Nov 20, 2013 at 12:41 AM, Jim Garrison  wrote:
(snip)
>> On Nov 19 14:20, Larry Hall (Cygwin) wrote:
>>
>> I found what happened:
>> http://msdn.microsoft.com/en-
>> us/library/windows/desktop/dn302074%28v=vs.85%29.aspx
>
> Hmm... from the quoted link:
>
>     We have made some significant changes in how the GetVersion(Ex) APIs
>     work in Windows 8.1 due to undesirable customer behaviors resulting from how
>     the GetVersion(Ex) APIs have been used in the past.
>
> Pesky customers won't behave the way we want them to...  (sinister music)

This is why we can't have nice things.

Sounds like a case of throwing out the baby with the bathwater.

Csaba
-- 
GCS a+ e++ d- C++ ULS$ L+$ !E- W++ P+++$ w++$ tv+ b++ DI D++ 5++
The Tao of math: The numbers you can count are not the real numbers.
Life is complex, with real and imaginary parts.
"Ok, it boots. Which means it must be bug-free and perfect. " -- Linus Torvalds
"People disagree with me. I just ignore them." -- Linus Torvalds

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Warren Young]

On 11/19/2013 10:13, Corinna Vinschen wrote:
>
> Why do they have to make such a mess out of a simple function like
> GetVersionEx?

Backwards compatibility at all costs?

> How dense is that?

Manifest Density.

(American joke.)


--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple

Re: cygcheck -svc segfaults on Windows 8.1 with cygwin64

[Corinna Vinschen]

[-- Attachment #1: Type: text/plain, Size: 578 bytes --]

On Nov 19 16:08, Warren Young wrote:
> On 11/19/2013 10:13, Corinna Vinschen wrote:
> >
> >Why do they have to make such a mess out of a simple function like
> >GetVersionEx?
> 
> Backwards compatibility at all costs?

Right.  I'm just glad the native API is more reliable.

> >How dense is that?
> 
> Manifest Density.
> 
> (American joke.)

...which I don't get.  Care to explain on cygwin-talk?


Corinna

-- 
Corinna Vinschen                  Please, send mails regarding Cygwin to
Cygwin Maintainer                 cygwin AT cygwin DOT com
Red Hat

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]