Bad transfer string during exec remote sql command

Hi

This is my SW: Debian GNU/Linux 11 (bullseye) installed ODBC Driver 17 for SQL Server (libmsodbcsql-17.10.so.1.1) and installed sql server 10.5.15-MariaDB and Microsoft SQL Server 2019 (RTM-CU16) (KB5011644) - 15.0.4223.1 (X64) (64-bit).

On MariaDB I have created table (according to the manual):

CREATE TABLE `ExternCommand` (

`cmd` VARCHAR(8000) NOT NULL flag=0,

`number` INT flag=1,

`message` VARCHAR(255) flag=2)

ENGINE=connect

TABLE_TYPE=odbc

CONNECTION='DSN=MSSQL;UID=xxx;PWD=xxx'

CHARSET = utf8

BLOCK_SIZE = 1

OPTION_LIST='Execsrc=1

And in MS SQL I have:

CREATE OR ALTER  PROCEDURE  [mydb].[dbo].[Test]

    @Text nvarchar(max) , @ID INT

AS

BEGIN

    SET NOCOUNT ON;

    INSERT INTO [mydb].[dbo].Tab(ID, [Text]) SELECT @ID, @Text

END

CREATE TABLE [mydb].[dbo].[Tab](

    [ID] [varchar](20) NULL,

    [Text] NVARCHAR(max) NULL

)

If run those:

SELECT  *

FROM `ExternCommand`

WHERE `cmd` = "EXEC [mydb].[dbo].[Test] N'šččřžýýá', 1";

SELECT  *

FROM `ExternCommand`

WHERE `cmd` = "INSERT INTO [mydb].[dbo].Tab([ID], [Text]) SELECT 1, N'šččřžýýá'";

In the table Test in MS SQL string 'šččřžýýá' is stored as 'Å¡Ä Ä Å Å¾Ã½Ã½Ã¡'.

How to set the character set encoding so that the string is transferred correctly?

Inserting data direct into tables mapped to MariaDB via engine connect is correct. However, I need to pass a parameter to the procedure. Isn't it a problem that the data is stored with a different order of BE and LE bytes in both systems? Or just when passing data to ODBC, ODBC evaluates the data as being in a different encoding?

Thanks for any idea how to fix this

Jaroslav