ERROR_NOT_READY (21) in ne_module.c

[Felix Nawothnig]

Since bug 2131 was closed I'll paste it here again:

ne_module.c explicitly sets errorcode 21 (ERROR_NOT_READY) when
(
   LoadLibraryA()ing the owner of a 16bit dll failed
or
   the search for the 16bit dll returned a real (.so)
   dll and not a symlink to the owner
) and trying to load a native version of the 16bit dll failed with
ERROR_FILE_NOT_FOUND.

(dlls/kernel/ne_module.c:1218)

Anyone knows the reason for that behavior?

-flx

[Dustin Navea]

Felix Nawothnig wrote:
> Since bug 2131 was closed I'll paste it here again:
>
> ne_module.c explicitly sets errorcode 21 (ERROR_NOT_READY) when
> (
>   LoadLibraryA()ing the owner of a 16bit dll failed
> or
>   the search for the 16bit dll returned a real (.so)
>   dll and not a symlink to the owner
> ) and trying to load a native version of the 16bit dll failed with
> ERROR_FILE_NOT_FOUND.
>
> (dlls/kernel/ne_module.c:1218)
>
> Anyone knows the reason for that behavior?
>
> -flx

You can still comment to the bug even though it is closed.  Since that 
bug was 2 issues in 1, I went ahead and closed it since the first issue 
cant/wont be fixed, but even though this issue has been posted to this 
list, it will be easier to keep track of if you file a bug for it as 
well..  Of course it could be related to the error in 2131.  Some of the 
devs commented that 16-bit dll's use a PE header, while 32-bit ones use 
a NE header, and that is why it wouldnt be possible to wrap the 16-bit 
routines within the same dll as the 32-bit ones.  This appears to 
honestly be for the same reason, if I understood what they were saying. 
  Let's let them comment though, and if a bug should be filed, they will 
tell you to do so..

Dustin

[Felix Nawothnig]

Dustin Navea wrote:
> You can still comment to the bug even though it is closed.

I know, but my question was neigher related to bug 2131 (it's just that 
it triggers the code in question) nor am I sure that the behaviour is 
wrong - the person who wrote that line must have had something in his 
mind. That's why I ask here.

-flx

[Felix Nawothnig]

(seems WD doesn't like being CCed? Resending...)

Dustin Navea wrote:
> You can still comment to the bug even though it is closed.

I know, but my question was neigher related to bug 2131 (it's just that 
it triggers the code in question) nor am I sure that the behaviour is 
wrong - the person who wrote that line must have had something in his 
mind. That's why I ask here.

-flx

[Alexandre Julliard]

Felix Nawothnig  writes:
> Since bug 2131 was closed I'll paste it here again:
>
> ne_module.c explicitly sets errorcode 21 (ERROR_NOT_READY) when
> (
>    LoadLibraryA()ing the owner of a 16bit dll failed
> or
>    the search for the 16bit dll returned a real (.so)
>    dll and not a symlink to the owner
> ) and trying to load a native version of the 16bit dll failed with
> ERROR_FILE_NOT_FOUND.
>
> (dlls/kernel/ne_module.c:1218)
>
> Anyone knows the reason for that behavior?

Error 21 in LoadModule16 means that the file exists but is 32-bit,
which seems to be the appropriate error in that case.

[Felix Nawothnig]

Alexandre Julliard wrote:
> Error 21 in LoadModule16 means that the file exists but is 32-bit,
> which seems to be the appropriate error in that case.

I dunno about LoadModule16() (no means to test it since it's a Win16 
function...) but for LoadLibrary16() (which is just a one-line wrapper 
around the former in wine) at least WinXP returns 0 and sets the 
errorcode to 87 (ERROR_INVALID_PARAMETER).

-flx