tests/msi.c fails when run twice on Wine?

[Paul Vriens]

Hi,

The msi test currently fails for me on Wine when run the second time after 
getting a clean environment:

rm -rf ~/.wine
make msi.ok
   msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
rm msi.ok
make msi.ok
   msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
   msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.

Anyone?

Cheers,

Paul.

[James Hawkins]

On 7/27/07, Paul Vriens  wrote:
> Hi,
>
> The msi test currently fails for me on Wine when run the second time after
> getting a clean environment:
>
> rm -rf ~/.wine
> make msi.ok
>    msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
> rm msi.ok
> make msi.ok
>    msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
>    msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.
>
> Anyone?

Works fine for me.

[Paul Vriens]

James Hawkins wrote:
> On 7/27/07, Paul Vriens  wrote:
> > Hi,
> >
> > The msi test currently fails for me on Wine when run the second time after
> > getting a clean environment:
> >
> > rm -rf ~/.wine
> > make msi.ok
> >    msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
> > rm msi.ok
> > make msi.ok
> >    msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
> >    msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.
> >
> > Anyone?
>
> Works fine for me.

It's even stranger then that. Running 'make msi.ok' gives:

msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.

running 'WINEDEBUG=+msi make msi.ok' gives:

<A LOT OF DEBUG STUFF>
msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.

Can anybody (apart from James) do these tests?

What are the things that could influence these difference between my tests 
failing and James's tests succeeding? Compiler, 32-bit vs 64-bit ???

Cheers,

Paul.

[Paul Vriens]

James Hawkins wrote:
> On 7/27/07, Paul Vriens  wrote:
> > Hi,
> >
> > The msi test currently fails for me on Wine when run the second time after
> > getting a clean environment:
> >
> > rm -rf ~/.wine
> > make msi.ok
> >    msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
> > rm msi.ok
> > make msi.ok
> >    msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
> >    msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.
> >
> > Anyone?
>
> Works fine for me.

Hi James,

Would the attached patch be OK? It makes sure that squash_guid always sets the
out parameter and check the results of squash_guid in the callers (from registry.c).

This fixes the msi test for me.

I'm not sure about the returned error code ERROR_FILE_NOT_FOUND but it made
the most sense to me.

Cheers,

Paul.

diff --git a/dlls/msi/registry.c b/dlls/msi/registry.c
index b9c0af4..bda4e28 100644
--- a/dlls/msi/registry.c
+++ b/dlls/msi/registry.c
@@ -233,6 +233,8 @@ BOOL squash_guid(LPCWSTR in, LPWSTR out)
     DWORD i,n=1;
     GUID guid;

+    out[0] = 0;
+
     if (FAILED(CLSIDFromString((LPOLESTR)in, &guid)))
         return FALSE;

@@ -474,7 +476,8 @@ UINT MSIREG_OpenUserProductsKey(LPCWSTR szProduct, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szProduct));
-    squash_guid(szProduct,squished_pc);
+    if (!squash_guid(szProduct,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szUserProduct_fmt,squished_pc);
@@ -493,7 +496,8 @@ UINT MSIREG_DeleteUserProductKey(LPCWSTR szProduct)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szProduct));
-    squash_guid(szProduct,squished_pc);
+    if (!squash_guid(szProduct,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szUserProduct_fmt,squished_pc);
@@ -508,7 +512,8 @@ UINT MSIREG_OpenUserPatchesKey(LPCWSTR szPatch, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szPatch));
-    squash_guid(szPatch,squished_pc);
+    if (!squash_guid(szPatch,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szUserPatch_fmt,squished_pc);
@@ -528,7 +533,8 @@ UINT MSIREG_OpenUserFeaturesKey(LPCWSTR szProduct, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szProduct));
-    squash_guid(szProduct,squished_pc);
+    if (!squash_guid(szProduct,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szUserFeatures_fmt,squished_pc);
@@ -553,7 +559,8 @@ UINT MSIREG_OpenFeaturesKey(LPCWSTR szProduct, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szProduct));
-    squash_guid(szProduct,squished_pc);
+    if (!squash_guid(szProduct,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szInstaller_Features_fmt,squished_pc);
@@ -574,7 +581,8 @@ UINT MSIREG_OpenUserDataFeaturesKey(LPCWSTR szProduct, HKEY *key, BOOL create)
     LPWSTR usersid;

TRACE("%s\n", debugstr_w(szProduct));
-    squash_guid(szProduct, squished_pc);
+    if (!squash_guid(szProduct, squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

rc = get_user_sid(&usersid);
@@ -607,7 +615,8 @@ UINT MSIREG_OpenComponentsKey(LPCWSTR szComponent, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szComponent));
-    squash_guid(szComponent,squished_cc);
+    if (!squash_guid(szComponent,squished_cc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_cc));

sprintfW(keypath,szInstaller_Components_fmt,squished_cc);
@@ -627,7 +636,8 @@ UINT MSIREG_OpenUserComponentsKey(LPCWSTR szComponent, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szComponent));
-    squash_guid(szComponent,squished_cc);
+    if (!squash_guid(szComponent,squished_cc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_cc));

sprintfW(keypath,szUser_Components_fmt,squished_cc);
@@ -648,7 +658,8 @@ UINT MSIREG_OpenUserDataComponentKey(LPCWSTR szComponent, HKEY *key, BOOL create
     LPWSTR usersid;

TRACE("%s\n", debugstr_w(szComponent));
-    squash_guid(szComponent, comp);
+    if (!squash_guid(szComponent, comp))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(comp));

rc = get_user_sid(&usersid);
@@ -677,7 +688,8 @@ UINT MSIREG_OpenUserDataProductKey(LPCWSTR szProduct, HKEY *key, BOOL create)
     LPWSTR usersid;

TRACE("%s\n", debugstr_w(szProduct));
-    squash_guid(szProduct, squished_pc);
+    if (!squash_guid(szProduct, squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

rc = get_user_sid(&usersid);
@@ -706,7 +718,8 @@ UINT MSIREG_OpenInstallPropertiesKey(LPCWSTR szProduct, HKEY *key, BOOL create)
     LPWSTR usersid;

TRACE("%s\n", debugstr_w(szProduct));
-    squash_guid(szProduct, squished_pc);
+    if (!squash_guid(szProduct, squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

rc = get_user_sid(&usersid);
@@ -735,7 +748,8 @@ UINT MSIREG_DeleteUserDataProductKey(LPCWSTR szProduct)
     LPWSTR usersid;

TRACE("%s\n", debugstr_w(szProduct));
-    squash_guid(szProduct, squished_pc);
+    if (!squash_guid(szProduct, squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

rc = get_user_sid(&usersid);
@@ -763,7 +777,8 @@ UINT MSIREG_OpenProductsKey(LPCWSTR szProduct, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szProduct));
-    squash_guid(szProduct,squished_pc);
+    if (!squash_guid(szProduct,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szInstaller_Products_fmt,squished_pc);
@@ -782,7 +797,8 @@ UINT MSIREG_DeleteProductKey(LPCWSTR szProduct)
     WCHAR keypath[0x200];

TRACE("%s\n", debugstr_w(szProduct));
-    squash_guid(szProduct, squished_pc);
+    if (!squash_guid(szProduct, squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath, szInstaller_Products_fmt, squished_pc);
@@ -797,7 +813,8 @@ UINT MSIREG_OpenPatchesKey(LPCWSTR szPatch, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szPatch));
-    squash_guid(szPatch,squished_pc);
+    if (!squash_guid(szPatch,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szInstaller_Patches_fmt,squished_pc);
@@ -817,7 +834,8 @@ UINT MSIREG_OpenUpgradeCodesKey(LPCWSTR szUpgradeCode, HKEY* key, BOOL create)
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szUpgradeCode));
-    squash_guid(szUpgradeCode,squished_pc);
+    if (!squash_guid(szUpgradeCode,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szInstaller_UpgradeCodes_fmt,squished_pc);
@@ -837,7 +855,8 @@ UINT MSIREG_OpenUserUpgradeCodesKey(LPCWSTR szUpgradeCode, HKEY* key, BOOL creat
     WCHAR keypath[0x200];

TRACE("%s\n",debugstr_w(szUpgradeCode));
-    squash_guid(szUpgradeCode,squished_pc);
+    if (!squash_guid(szUpgradeCode,squished_pc))
+        return ERROR_FILE_NOT_FOUND;
     TRACE("squished (%s)\n", debugstr_w(squished_pc));

sprintfW(keypath,szInstaller_UserUpgradeCodes_fmt,squished_pc);

[James Hawkins]

On 7/30/07, Paul Vriens  wrote:
> James Hawkins wrote:
> > On 7/27/07, Paul Vriens  wrote:
> > > Hi,
> > >
> > > The msi test currently fails for me on Wine when run the second time after
> > > getting a clean environment:
> > >
> > > rm -rf ~/.wine
> > > make msi.ok
> > >    msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
> > > rm msi.ok
> > > make msi.ok
> > >    msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
> > >    msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.
> > >
> > > Anyone?
> >
> > Works fine for me.
>
> Hi James,
>
> Would the attached patch be OK? It makes sure that squash_guid always sets the
> out parameter and check the results of squash_guid in the callers (from registry.c).
>
> This fixes the msi test for me.
>
> I'm not sure about the returned error code ERROR_FILE_NOT_FOUND but it made
> the most sense to me.

hmm it should probably be ERROR_FUNCTION_FAILED, but otherwise it's fine.

[Paul Vriens]

James Hawkins wrote:
> On 7/30/07, Paul Vriens  wrote:
> > James Hawkins wrote:
> > > On 7/27/07, Paul Vriens  wrote:
> > > > Hi,
> > > >
> > > > The msi test currently fails for me on Wine when run the second time after
> > > > getting a clean environment:
> > > >
> > > > rm -rf ~/.wine
> > > > make msi.ok
> > > >    msi: 101 tests executed (0 marked as todo, 0 failures), 0 skipped.
> > > > rm msi.ok
> > > > make msi.ok
> > > >    msi.c:366: Test failed: Expected INSTALLSTATE_UNKNOWN, got 1
> > > >    msi: 101 tests executed (0 marked as todo, 1 failure), 0 skipped.
> > > >
> > > > Anyone?
> > >
> > > Works fine for me.
> >
> > Hi James,
> >
> > Would the attached patch be OK? It makes sure that squash_guid always sets the
> > out parameter and check the results of squash_guid in the callers (from registry.c).
> >
> > This fixes the msi test for me.
> >
> > I'm not sure about the returned error code ERROR_FILE_NOT_FOUND but it made
> > the most sense to me.
>
> hmm it should probably be ERROR_FUNCTION_FAILED, but otherwise it's fine.

Sound reasonable. I'll change that. I'll also have a look at the other callers 
of squash_guid.

It's still strange though that it doesn't happen on your box (you are the only 
one that came back, and of course Alexandre cause I guess he runs the tests 
every time before he commits something).

Cheers,

Paul.

[Juan Lang]

> It's still strange though that it doesn't happen on your box (you are the only
> one that came back, and of course Alexandre cause I guess he runs the tests
> every time before he commits something).

It also succeeds when running twice here.  (make install.ok hangs here, though.)
--Juan

[Paul Vriens]

Juan Lang wrote:
> > It's still strange though that it doesn't happen on your box (you are the only
> > one that came back, and of course Alexandre cause I guess he runs the tests
> > every time before he commits something).
>
> It also succeeds when running twice here.  (make install.ok hangs here, though.)
> --Juan

After a 'rm ~/.wine' ?

Could this be a compiler/platform issue (difference). If you run 'WINEDEBUG=+msi 
make msi.ok' do you see garbage in the 'squished ...' output?

Apart from the fact it only fails for me (for now), I guess the patch I've just 
sent makes sense anyhow.

Cheers.

Paul.

[James Hawkins]

On 7/30/07, Paul Vriens  wrote:
> Juan Lang wrote:
> > > It's still strange though that it doesn't happen on your box (you are the only
> > > one that came back, and of course Alexandre cause I guess he runs the tests
> > > every time before he commits something).
> >
> > It also succeeds when running twice here.  (make install.ok hangs here, though.)
> > --Juan
>
> After a 'rm ~/.wine' ?
>
> Could this be a compiler/platform issue (difference). If you run 'WINEDEBUG=+msi
> make msi.ok' do you see garbage in the 'squished ...' output?
>
> Apart from the fact it only fails for me (for now), I guess the patch I've just
> sent makes sense anyhow.

The garbage in the squished guid is not important...the regopen call
will fail like it's supposed to because the squished guid doesn't
match.

[Paul Vriens]

James Hawkins wrote:
> On 7/30/07, Paul Vriens  wrote:
> > Juan Lang wrote:
> > > > It's still strange though that it doesn't happen on your box (you are the only
> > > > one that came back, and of course Alexandre cause I guess he runs the tests
> > > > every time before he commits something).
> > >
> > > It also succeeds when running twice here.  (make install.ok hangs here, though.)
> > > --Juan
> >
> > After a 'rm ~/.wine' ?
> >
> > Could this be a compiler/platform issue (difference). If you run 'WINEDEBUG=+msi
> > make msi.ok' do you see garbage in the 'squished ...' output?
> >
> > Apart from the fact it only fails for me (for now), I guess the patch I've just
> > sent makes sense anyhow.
>
> The garbage in the squished guid is not important...the regopen call
> will fail like it's supposed to because the squished guid doesn't
> match.

Yeah I guess so, but I didn't have garbage all the time (remember I wrote that 
it succeeded when running with WINEDEBUG=+msi). So if it was just returned as an 
empty string (or at least interpreted as such) a RegOpenKey would succeed on the 
parent of the GUID (Software\Microsoft\Installer\Products) which is most of the 
time there.

Cheers,

Paul.

[Juan Lang]

> > It also succeeds when running twice here.  (make install.ok hangs here, though.)
>
> After a 'rm ~/.wine' ?

Yes.
> Could this be a compiler/platform issue (difference).

Perhaps.
> If you run 'WINEDEBUG=+msi make msi.ok' do you see garbage in the 'squished ...' output?

No.
--Juan