Re: TAPE 5/5: add implementation of kernel tape functions #2

[Robert Shearman]

Hans Leidekker wrote:
> +static DWORD set_error_from_status( NTSTATUS status )
> +{
> - DWORD error;
> -
> - switch (status)
> - {
> - case STATUS_SUCCESS:            error = ERROR_SUCCESS; break;
> - case STATUS_DISK_FULL:          error = ERROR_END_OF_MEDIA; break;
> - case STATUS_DEVICE_NOT_READY:   error = ERROR_NO_MEDIA_IN_DRIVE; break;
> - case STATUS_ACCESS_DENIED:      error = ERROR_WRITE_PROTECT; break;
> - case STATUS_DEVICE_BUSY:        error = ERROR_BUSY; break;
> - case STATUS_INVALID_PARAMETER:  error = ERROR_INVALID_PARAMETER; break;
> - case STATUS_NOT_SUPPORTED:
> - case STATUS_UNSUCCESSFUL:
> - default:                        error = ERROR_NOT_SUPPORTED; break;
> - }
> -
> - SetLastError( error );
> - return error;
>
> +}

You're duplicating RtlNtStatusToDosError.

TAPE 5/5: add implementation of kernel tape functions #2

[Hans Leidekker]

On Thursday 26 January 2006 20:37, Robert Shearman wrote:
> You're duplicating RtlNtStatusToDosError.

Not completely. STATUS_DISK_FULL (translated from ENOSPC) for
example is mapped to ERROR_DISK_FULL by RtlNtStatusToDosError
whereas I am mapping it to ERROR_END_OF_MEDIA which I think is
more appropriate in the context of tapes.

But the conversion is lossy already (Unix error -> NT status
-> Win32 error), so I don't care too much. I'll send a patch
that uses RtlNtStatusToDosError so Alexandre can choose.

-Hans

TAPE 5/5: add implementation of kernel tape functions #2

[Alexandre Julliard]

Hans Leidekker  writes:
> Not completely. STATUS_DISK_FULL (translated from ENOSPC) for
> example is mapped to ERROR_DISK_FULL by RtlNtStatusToDosError
> whereas I am mapping it to ERROR_END_OF_MEDIA which I think is
> more appropriate in the context of tapes.

Then the ntdll code should have returned STATUS_END_OF_MEDIA. You
don't have to use the standard errno conversion in ntdll if you can
return more appropriate status codes.