[PATCH v3 resend 3/4] ntdll: Allow NtQueryVolumeInformationFile to query volume information from the mountmgr.

[Erich E. Hoover]

This patch allows NtQueryVolumeInformationFile to pass
FileFsVolumeInformation and FileFsAttributeInformation requests to the
mountmgr.  These requests currently do not work for NT device paths,
which results in a bunch of duplication in kernelbase.

Best,
Erich

[Zebediah Figura]

On 9/8/20 3:02 PM, Erich E. Hoover wrote:
> From 10add5d0acd423e2fb3d762579493dc50f6e6c96 Mon Sep 17 00:00:00 2001
> From: "Erich E. Hoover" 
> Date: Fri, 12 Jun 2020 14:53:43 -0600
> Subject: ntdll: Allow NtQueryVolumeInformationFile to query volume information
>  from the mountmgr.

Though really this patch is more general than that, i.e. it's actually
allowing for asynchronous volume information queries from server objects.

For that matter, I don't know what was requested before, but this patch
split seems a bit awkward. Maybe a better split would be:

0001: allow volume information queries to be asynchronous (including
both the changes to server/fd.c and ntdll/unix/file.c, although maybe
even those should be split...)
0002: implement volume information queries for device files
0003: hook up volume information queries for mountmgr

As a further note, some simple tests (in dlls/ntoskrnl.exe/tests) could
help this series.
> Signed-off-by: Erich E. Hoover
>
> dlls/ntdll/unix/file.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
>
> diff --git a/dlls/ntdll/unix/file.c b/dlls/ntdll/unix/file.c
> index 03e92a5c59e..7c16908959d 100644
> --- a/dlls/ntdll/unix/file.c
> +++ b/dlls/ntdll/unix/file.c
> @@ -6186,15 +6186,26 @@ NTSTATUS WINAPI NtQueryVolumeInformationFile( HANDLE handle, IO_STATUS_BLOCK *io
>      io->u.Status = server_get_unix_fd( handle, 0, &fd, &needs_close, NULL, NULL );
>      if (io->u.Status == STATUS_BAD_DEVICE_TYPE)
>      {
> -
>    struct async_irp *async;
> -
>    HANDLE wait_handle;
> -
> -
>    if (!(async = (struct async_irp *)alloc_fileio( sizeof(*async), irp_completion, handle )))
> -
>        return STATUS_NO_MEMORY;
> -
>    async->buffer  = buffer;
> -
>    async->size    = length;
> -
>    SERVER_START_REQ( get_volume_info )
>    {
> -
>        req->async = server_async( handle, &async->io, NULL, NULL, NULL, io );
>        req->handle = wine_server_obj_handle( handle );
>        req->info_class = info_class;
>        wine_server_set_reply( req, buffer, length );
>        io->u.Status = wine_server_call( req );
>        if (!io->u.Status) io->Information = wine_server_reply_size( reply );
> -
>        wait_handle = wine_server_ptr_handle( reply->wait );
>    }
>    SERVER_END_REQ;
> -
>    if (wait_handle) io->u.Status = wait_async( wait_handle, TRUE, io );
>    return io->u.Status;
> }
>    else if (io->u.Status) return io->u.Status;
>
> -- 
> 2.17.1

[Erich E. Hoover]

On Wed, Oct 28, 2020 at 10:33 AM Zebediah Figura  wrote:
> On 9/8/20 3:02 PM, Erich E. Hoover wrote:
> > From 10add5d0acd423e2fb3d762579493dc50f6e6c96 Mon Sep 17 00:00:00 2001
> > From: "Erich E. Hoover" 
> > Date: Fri, 12 Jun 2020 14:53:43 -0600
> > Subject: ntdll: Allow NtQueryVolumeInformationFile to query volume information
> >  from the mountmgr.
>
> Though really this patch is more general than that, i.e. it's actually
> allowing for asynchronous volume information queries from server objects.

I'll update the description.
> For that matter, I don't know what was requested before, but this patch
> split seems a bit awkward. Maybe a better split would be:

The status was just updated to "needs splitting", so I broke the first
patch up into the three you see now.
> 0001: allow volume information queries to be asynchronous (including
> both the changes to server/fd.c and ntdll/unix/file.c, although maybe
> even those should be split...)
> 0002: implement volume information queries for device files
> 0003: hook up volume information queries for mountmgr

Okay, what I think that you're saying is:
0001: server/fd.c changes
0002: ntdll/unix/file.c changes
0003: ntoskrnl.exe/ntoskrnl.c changes
0004: mountmgr.sys/device.c changes
Please let me know if that sounds right to you, I'd be happy to do that.
> As a further note, some simple tests (in dlls/ntoskrnl.exe/tests) could
> help this series.

I'll put that together.

Best,
Erich

[Zebediah Figura]

On 10/28/20 1:32 PM, Erich E. Hoover wrote:
> On Wed, Oct 28, 2020 at 10:33 AM Zebediah Figura  wrote:
> > On 9/8/20 3:02 PM, Erich E. Hoover wrote:
> > > From 10add5d0acd423e2fb3d762579493dc50f6e6c96 Mon Sep 17 00:00:00 2001
> > > From: "Erich E. Hoover" 
> > > Date: Fri, 12 Jun 2020 14:53:43 -0600
> > > Subject: ntdll: Allow NtQueryVolumeInformationFile to query volume information
> > >  from the mountmgr.
> >
> > Though really this patch is more general than that, i.e. it's actually
> > allowing for asynchronous volume information queries from server objects.
>
> I'll update the description.
> > For that matter, I don't know what was requested before, but this patch
> > split seems a bit awkward. Maybe a better split would be:
>
> The status was just updated to "needs splitting", so I broke the first
> patch up into the three you see now.
> > 0001: allow volume information queries to be asynchronous (including
> > both the changes to server/fd.c and ntdll/unix/file.c, although maybe
> > even those should be split...)
> > 0002: implement volume information queries for device files
> > 0003: hook up volume information queries for mountmgr
>
> Okay, what I think that you're saying is:
> 0001: server/fd.c changes
> 0002: ntdll/unix/file.c changes
> 0003: ntoskrnl.exe/ntoskrnl.c changes
> 0004: mountmgr.sys/device.c changes
> Please let me know if that sounds right to you, I'd be happy to do that.

Right, something like that, the diff to server/device.c also being part
of patch 3.
> > As a further note, some simple tests (in dlls/ntoskrnl.exe/tests) could
> > help this series.
>
> I'll put that together.
>
> Best,
> Erich