Hi,
as a follow-up the proposal, here is a patch for storing console information in the lbtable, see attachment.
I'm not sure if all of these console types are actually useful for the payload (eg. logbuf), and for some of these consoles the payload will need more information about where to find the hardware that is referred.
VGA and EHCI for device resources, logbuf (should it survive) for a memory location where it can be found, etc. Given how different these are, I'd opt for a new record type each (as per the proposal), instead of a "universal device" record type with 20 values of which usually only the 5 (that apply to this specific device) are used.
Regards, Patrick Georgi
* Patrick Georgi patrick@georgi-clan.de [080126 18:30]:
+void add_console(struct lb_header *header, u16 consoletype) +{
- struct lb_record *rec;
- struct lb_console *console;
- rec = lb_new_record(header);
- console = (struct lb_console *)lb_new_record(header);
- console->tag = LB_TAG_CONSOLE;
- console->size = sizeof(*console);
- console->type = LB_TAG_CONSOLE_SERIAL8250;
Is there a reason you are hardcoding the type here, despite passing one in add_console? ;-)
+}
Otherwise I say
Acked-by: Stefan Reinauer stepan@coresystems.de
Will you implement using those table entries (at least vga and serial console) for grub2?
Stefan
Am Samstag, den 26.01.2008, 21:31 +0100 schrieb Stefan Reinauer:
Is there a reason you are hardcoding the type here, despite
passing one in add_console? ;-)
doh, thanks..
what about the types of console - should I keep them all?
Regards, Patrick Georgi
* Patrick Georgi patrick@georgi-clan.de [080126 22:40]:
Am Samstag, den 26.01.2008, 21:31 +0100 schrieb Stefan Reinauer:
Is there a reason you are hardcoding the type here, despite
passing one in add_console? ;-)
doh, thanks..
what about the types of console - should I keep them all?
At least for v2, yes.