Subject: Re: Intel Ether Express 16
To: None <netbsd-bugs@NetBSD.ORG>
From: is <is@comma.rhein.de>
List: netbsd-bugs
Date: 02/21/1996 14:39:57
In <01BAFACB.F36CAA40@johnam.datastorm.com> "John A. Maier"
<root@beta.datastorm.com> writes:

>I understand the issue of information scrolling off the screen too fast, =
>this could easily be solved by displaying the error message then a =
>"Press <space> to continue..." message or something along these lines.  =

No, it could not. There is at least one port, where you are not guaranteed
to have a keyboard connected to the console device, at least not that early
in the boot process. I'm afraid that during auto configuration time
the console must be considered write-only.

>#ifdef NOUSERDIAGMSG
>if(USER_DIAG_MSG)
>	printf("ie: Your card is dead!\n");
>	printf("ie: Press any key to continue.\n");
>	getchar();
>#endif

same comment. We can't assume we have a keyboard until it is probed (which 
might be too late for this purpose), with the exception of panic()ing into 
the kernel debugger (it doesn't matter if the assumption fails, as we have
crashed anyway) and booting with -D (hm, is this valid for i386?), where the
user has explicitly told us that he assumes to be able to use the keyboard
to operate the debugger.

To my knowledge, the existing console drivers don't allow timed reads from
inside the kernel. I propose giving a short, descriptive message (one line!!).

ie0 at isabus0: no memory mapped, skipping.

The user can always get it back, in single- oder multiuser mode, with
dmesg | more (which he should be told about in the INSTALL document if it
doesn't do so).

Regards,
	Ignatios Souvatzis